THE MACROECONOMIC EFFECTS OF STUDENT DEBT

of Bard College

Levy Economics

Institute

THE

MACROECONOMIC

EFFECTS OF

STUDENT DEBT

CANCELLATION

Scott Fullwiler, Stephanie Kelton, Catherine Ruetschlin, and Marshall Steinbaum

February 2018

2 Student Debt Cancellation Report 2018

Levy Economics Institute of Bard College 3

THE MACROECONOMIC EFFECTS OF

STUDENT DEBT CANCELLATION

Scott Fullwiler, Stephanie Kelton, Catherine Ruetschlin, and Marshall Steinbaum

4 Student Debt Cancellation Report 2018

Table of Contents

EXECUTIVE SUMMARY 6

INTRODUCTION 7

SECTION 1: THE ECONOMIC OPPORTUNITY OF STUDENT DEBT CANCELLATION 9

Social Investment in Higher Education 9

The current state of student debt 10

The social costs of student debt 12

The Distributional Consequences of Student Debt, Student Debt Cancellation, and Debt-Free College 13

The distribution of student debt and debt burden in the cross section 14

The evolution of the distribution of student debt burdens over time 15

What does the evolution of student debt tell us about the labor market? 16

How does student debt interact with longstanding economic disparaties? 16

The real distributional impact of student debt cancellation and free or debt-free college 17

SECTION 2: THE MECHANICS OF STUDENT DEBT CANCELLATION 18

The Mechanics of Student Debt Cancellation Carried Out by the Government 18

Current servicing of student loans from a balance sheet perspective 18

Possible methods of government-ﬁnanced student debt cancellation 19

The government cancels the Department of Education’s loans all at once 20

The government cancels the Department of Education’s loans as borrowers’ payments come due 21

Government-led debt cancellation where the government assumes payments on student loans issued by 22

private investors

Government-led debt cancellation where the government simultaneously purchases and then cancels loans owned 24

by private investors

Government-led debt cancellation where the government purchases student loans issued by private investors and 25

cancels principal as payments come due

Concluding remarks on government-led cancellation of privately owned student loans 26

Concluding remarks on government-led debt cancellation 26

The Mechanics of Student Debt Cancellation Carried Out by the Federal Reserve 27

The Federal Reserve purchases the Department of Education’s loans 27

Some fundamentals of the Federal Reserve’s remittances and their relevance to student loan cancellation 28

The Federal Reserve cancels the Department of Education’s loans all at once 28

The Federal Reserve cancels debt service payments for the Department of Education’s loans 29

The Federal Reserve assumes debt service payments for loans owned by private investors 31

The Federal Reserve purchases and cancels loans owned by private investors 32

The Federal Reserve purchases loans owned by private investors and cancels debt service payments 33

Potential options to avoid costs to the federal government of student loan cancellation carried out by the 34

Federal Reserve

Levy Economics Institute of Bard College 5

SECTION 3: SIMULATING STUDENT DEBT CANCELLATION 36

Models and Assumptions Used for Simulating Student Debt Cancellation 36

Introduction to the Moody’s model 36

Introduction to the Fair model 37

Assumptions for the simulated student debt cancellation 37

Baseline values and macroeconometric simulation 38

Simulation Results 39

Conclusions from simulations 45

Omitted Beneﬁts and Costs of Student Debt Cancellation 46

Small business formation 46

College degree attainment 47

Household formation 48

Credit scores 48

Household vulnerability in business cycle downturns 48

Moral hazard 49

CONCLUSION 50

APPENDIX A: SIMULATION DATA SERIES 52

APPENDIX B: DEPARTMENT OF EDUCATION LOANS AND THE BUDGET DEFICIT 55

APPENDIX C: DIGRESSION ON THE FED’S OPERATIONS 58

NOTES 61

REFERENCES 64

6 Student Debt Cancellation Report 2018

Executive Summary

More than 44 million Americans are caught in a student debt

trap. Collectively, they owe nearly $1.4 trillion on outstanding

student loan debt. Research shows that this level of debt hurts

the US economy in a variety of ways, holding back everything

from small business formation to new home buying, and even

marriage and reproduction. It is a problem that policymakers

have attempted to mitigate with programs that offer reﬁnanc-

ing or partial debt cancellation. But what if something far more

ambitious were tried? What if the population were freed from

making any future payments on the current stock of outstand-

ing student loan debt? Could it be done, and if so, how? What

would it mean for the US economy?

This report seeks to answer those very questions. The

analysis proceeds in three sections: the ﬁrst explores the current

US context of increasing college costs and reliance on debt to

ﬁnance higher education; the second section works through the

balance sheet mechanics required to liberate Americans from

student loan debt; and the ﬁnal section simulates the economic

effects of this debt cancellation using two models, Ray Fair’s US

Macroeconomic Model (“the Fair model”) and Moody’s US

Macroeconomic Model.

Several important implications emerge from this analysis.

Student debt cancellation results in positive macroeconomic

feedback effects as average households’ net worth and dispos-

able income increase, driving new consumption and investment

spending. In short, we ﬁnd that debt cancellation lifts GDP,

decreases the average unemployment rate, and results in little

inﬂationary pressure (all over the 10-year horizon of our sim-

ulations), while interest rates increase only modestly. Though

the federal budget deﬁcit does increase, state-level budget posi-

tions improve as a result of the stronger economy. The use of

two models with contrasting long-run theoretical foundations

offers a plausible range for each of these effects and demon-

strates the robustness of our results.

A one-time policy of student debt cancellation, in which

the federal government cancels the loans it holds directly and

takes over the ﬁnancing of privately owned loans on behalf of

borrowers, results in the following macroeconomic effects (all

dollar values are in real, inﬂation-adjusted terms, using 2016 as

the base year):

• The policy of debt cancellation could boost real GDP by

an average of $86 billion to $108 billion per year. Over the

10-year forecast, the policy generates between $861 billion

and $1,083 billion in real GDP (2016 dollars).

• Eliminating student debt reduces the average unemploy-

ment rate by 0.22 to 0.36 percentage points over the 10-year

forecast.

• Peak job creation in the ﬁrst few years following the elimina-

tion of student loan debt adds roughly 1.2 million to 1.5 mil-

lion new jobs per year.

• The inﬂationary effects of cancelling the debt are macro-

economically insigniﬁcant. In the Fair model simulations,

additional inﬂation peaks at about 0.3 percentage points and

turns negative in later years. In the Moody’s model, the effect

is even smaller, with the pickup in inﬂation peaking at a triv-

ial 0.09 percentage points.

• Nominal interest rates rise modestly. In the early years, the

Federal Reserve raises target rates 0.3 to 0.5 percentage points;

in later years, the increase falls to just 0.2 percentage points.

The effect on nominal longer-term interest rates peaks at 0.25

to 0.5 percentage points and declines thereafter, settling at

0.21 to 0.35 percentage points.

• The net budgetary effect for the federal government is modest,

with a likely increase in the deﬁcit-to-GDP ratio of 0.65 to 0.75

percentage points per year. Depending on the federal govern-

ment’s budget position overall, the deﬁcit ratio could rise more

modestly, ranging between 0.59 and 0.61 percentage points.

However, given that the costs of funding the Department of

Education’s student loans have already been incurred (dis-

cussed in detail in Section 2), the more relevant estimates

for the impacts on the government’s budget position relative

to current levels are an annual increase in the deﬁcit ratio of

between 0.29 and 0.37 percentage points. (This is explained in

further detail in Appendix B.)

• State budget deﬁcits as a percentage of GDP improve by about

0.11 percentage points during the entire simulation period.

• Research suggests many other positive spillover effects that are

not accounted for in these simulations, including increases in

small business formation, degree attainment, and household

formation, as well as improved access to credit and reduced

household vulnerability to business cycle downturns. Thus,

our results provide a conservative estimate of the macro

effects of student debt liberation.

Levy Economics Institute of Bard College 7

Introduction

There is mounting evidence that the escalation of student debt

in the United States is an impediment to both household ﬁnan-

cial stability and aggregate consumption and investment. The

increasing demand for college credentials coupled with rising

costs of attendance have led more students than ever before to

take on student loans, with higher average balances. This debt

burden reduces household disposable income and consump-

tion and investment opportunities, with spillover effects across

the economy. At the same time, the social beneﬁts of investment

in higher education—including human capital accumulation,

social mobility, and the greater tax revenues and social contri-

butions that ﬂow from a highly productive population—remain

central to the economic advantages enjoyed by the United States.

In this context, students, educators, and policymakers have

called for a range of solutions to the rising cost of college and

the encumbrance of borrowers. In this report, we examine the

macroeconomic effects of one of the boldest of these propos-

als: a program of outright student debt cancellation ﬁnanced

by the federal government. If student debt is indeed dampen-

ing household economic activity, we expect liberation from

this debt to produce a stimulus effect that will partially offset

the cost of the program. In fact, we ﬁnd that cancelling student

debt would have a meaningful stimulus effect, particularly in

the ﬁrst ﬁve years, characterized by greater economic activity as

measured by GDP and employment, with only moderate effects

on the federal budget deﬁcit, interest rates, and inﬂation over the

forecast horizon. Overall, the macroeconomic consequences of

student debt cancellation demonstrate that a reorientation of US

higher education policy can include ambitious policy proposals

like a total cancellation of all outstanding student loan debt.

Higher education is a valuable social investment, with

research demonstrating social returns up to ﬁve times the dollar

amount of public spending in the United States (OECD 2015).

The diffusion of these beneﬁts across the economy makes them

a classic example of positive externalities, a condition in which

individual cost/beneﬁt calculations that omit social beneﬁts will

result in a market failure. In these cases, public investment is

necessary to avoid chronic underinvestment. Yet in the United

States over the past three decades, public funding of higher edu-

cation has been in decline (SHEEO 2015). At the same time,

the increasing need for a college credential to access key labor

market entry positions provided incentives for more students

to take on debt. This student loan debt imposes a signiﬁcantly

higher burden on household ﬁnances than ever before, as stag-

nant real incomes and higher average balances combine to

divert a larger portion of household resources toward debt ser-

vice and away from consumption and investment.

It is possible for the federal government to reduce or remove

the burden of student loan debt as a means of direct support

to household spending. In this report, we examine the mecha-

nisms that facilitate debt cancellation using T-accounts to map

the transactions associated with the program. In a government-

ﬁnanced cancellation program, the current loan portfolio of the

Department of Education is cancelled and the federal govern-

ment either purchases and cancels or takes over the payments

for privately owned loans. One of the more signiﬁcant take-

aways here is the realization that, because the loans made by the

Department of Education—which make up the vast majority

of student loans outstanding—were already funded when the

loans were originated, the new costs of cancelling these loans are

limited to the interest payments on the securities issued at that

time. An alternative route, which some have advocated, involves

the Federal Reserve buying up student loan debt and warehous-

ing the losses on its own balance sheet. We consider this option

below, noting that this avenue would most likely require autho-

rization from Congress. Importantly, we also show that any pro-

gram led by the Federal Reserve results in the same consequences

for the federal government’s budget position as a government-

led program—that is, there is no “free lunch” that avoids the

budgetary implications of cancelling student debt.

We also simulated the student debt cancellation program

using two macroeconometric models to examine the implica-

tions of cancellation and incorporate feedback effects that go

beyond the balance sheet analysis. The ﬁrst-round effect of stu-

dent debt cancellation is an increase in the wealth and dispos-

able income of student loan borrowers. These effects translate

to higher spending in a variety of consumption and investment

categories, which represent greater economic activity and pro-

duce additional income, jobs, and tax revenue. We relied on two

macroeconomic models to simulate these effects: Ray Fair of

Yale University’s US Macroeconomic Model (“the Fair model”)

and Moody’s US Macroeconomic Model, the forecasting model

used by Moody’s and Economy.com. The Fair model and the

Moody’s model share a Keynesian short-run theoretical foun-

dation. In the long run, however, the assumed relationships dif-

fer, as Moody’s takes on a “Classical core” while the Fair model

8 Student Debt Cancellation Report 2018

remains fundamentally Keynesian. In addition to two models

with distinct foundational assumptions, we also implemented

two alternative assumptions about the Federal Reserve’s inter-

est rate response to the debt cancellation stimulus. The use of

models with contrasting long-run theoretical foundations and

alternative scenarios demonstrates the robustness of the results

in this report, and also allows us to present a plausible range for

each of the estimated effects of a federally ﬁnanced student debt

cancellation.

A program to cancel student debt executed in 2017 results

in an increase in real GDP, a decrease in the average unemploy-

ment rate, and little to no inﬂationary pressure over the 10-year

horizon of our simulations, while interest rates increase only

modestly. Our results show that the positive feedback effects of

student debt cancellation could add on average between $86 bil-

lion and $108 billion per year to the economy. Associated with

this new economic activity, job creation rises and the unem-

ployment rate declines.

The macroeconomic models used in these simulations

assume an essentially mechanical Federal Reserve response to

lower unemployment. Suppressing this response—in other

words, assuming the Fed does not raise its interest rate target—

provides an upper bound for the range of possible outcomes

associated with more nuanced central bank policy. In fact, both

models forecast little to no additional inﬂation resulting from

the cancellation of student debt. In the Fair model, inﬂation

peaks at an additional 0.3 percent and turns negative after 2020,

meaning that debt cancellation reduces inﬂation in later years.

In the Moody’s model, the inﬂationary effects are never higher

than 0.09 percent throughout the period. These forecasts sug-

gest that there is room for ﬂexibility in the assumptions made

about Federal Reserve tactics as a response to debt cancellation.

Since even the largest effect on inﬂation in a single year is of

little macroeconomic signiﬁcance, it is arguable that the Fed

would not react to the student debt cancellation program by

raising its target interest rate.

Student debt cancellation is a large-scale program in which

the government must repay privately held loans and forego

interest rate payments on the loan portfolio of the Department

of Education. It is reasonable to expect such a program to add to

the federal government’s budget deﬁcit, absent extraordinarily

strong feedback effects from the program’s macroeconomic

stimulus. Our simulations show that student debt cancella-

tion raises the federal budget deﬁcit moderately. The average

impacts on the federal deﬁcit in the simulations are between

0.65 and 0.75 percent of GDP per year. However, the more rel-

evant ﬁgures for the annual impact on the federal deﬁcit fall in a

range between 0.29 and 0.37 percent of GDP—this accounts for

the fact that, for the Department of Education loans, only debt

service on the securities originally issued will add to current

deﬁcits and the national debt. The simulations, by their nature,

assume the full costs of the foregone principal and interest on

the Department of Education loans are incurred in the cancella-

tion. In Section 3 and Appendix B, we explain the reasons for this

assumption embedded in the simulations (which generates esti-

mates of budget impacts relative to a no-cancellation baseline

scenario) and how the lower, more relevant ﬁgures (estimates

of budget impacts relative to current deﬁcit and debt levels) are

arrived at. Only the Fair model enables forecasts of state-level

budget positions, and we ﬁnd improvements in states’ budget

positions as a result of the stimulus effects of the debt cancella-

tion. These improvements will reduce the need for states to raise

taxes or cut spending in the event of future recessions.

It is important to note that the macroeconomic models

used in this report cannot capture all of the positive socioeco-

nomic effects associated with cancelling student loan debt. New

research from academics and experts has demonstrated the

relationships between student debt and business formation, col-

lege completion, household formation, and credit scores. These

correlations suggest that student debt cancellation could gener-

ate substantial stimulus effects in addition to those that emerge

from our simulations, while improving the ﬁnancial positions

of households.

Our analysis proceeds in three sections. Section 1, “The

Economic Opportunity of Student Debt Cancellation,” explores

the US context of student borrowing, including reductions in

public investment in higher education and the rising cost of

a college degree, the social costs of rising debt, and the distri-

butional implications of debt and debt cancellation. Section

2, “The Mechanics of Student Debt Cancellation,” explains

the instruments of debt relief, whether enacted by the fed-

eral government or its central bank (the Federal Reserve), and

demonstrates the balance sheet effects of debt cancellation on

the government, the Federal Reserve, banks, borrowers, and

private lenders. Finally, Section 3, “Simulating Student Debt

Cancellation,” measures the effects of the program on key mac-

roeconomic variables using simulations in two models—the

Fair model and Moody’s model—under alternative assump-

tions, and examines the costs and beneﬁts of student debt relief

that are omitted from the models.

Levy Economics Institute of Bard College 9

Section 1: The Economic

Opportunity of Student Debt

Cancellation

In the United States, attaining a college degree has long been

viewed as a safe investment for individuals and for the nation

as a whole. The funding of higher education from both public

and private sources exempliﬁes the joint character of this invest-

ment, and a strong system of public education has conferred

broad social and economic beneﬁts. Yet over the past three

decades our common commitment to education has broken

down. American households increasingly shoulder the burden

of ﬁnancing higher education. This private ﬁnancing of higher

education requires a growing share of household consump-

tion and investment spending, drawing resources away from

other sectors such as housing and other markets for consumer

ﬁnance. Most students today meet the growing cost by tak-

ing on debt. As a result of the shifting ﬁnancial responsibility

for higher education, student debt is at record highs. College

graduates begin their careers with debt payments that absorb

income and supplant other important early-adulthood invest-

ment opportunities. Moreover, the increasingly private respon-

sibility for ﬁnancing higher education diverts attention from

the important social beneﬁts of an educated population as eco-

nomic decisions focus more and more on individual returns.

Thus, the debt-based system of higher education ﬁnance comes

at a larger cost: student debt limits the economic opportunities

of today’s young people, depletes other forms of consumer and

investment spending in the economy, and undermines the com-

monly shared gains that derive from an educated workforce and

citizenry.

Social Investment in Higher Education

The individual beneﬁts of a college degree are widely acknowl-

edged, but the increasing focus on individual ﬁnancing has

largely neglected similar calculations on the social scale. The

returns to individuals accrue in terms of employment oppor-

tunities and lifetime earnings. Comparing workers with a

bachelor’s degree to those whose education ended with a high

school diploma shows that a postsecondary credential is more

and more valuable, leading college graduates to higher life-

time incomes and lower unemployment rates relative to those

without a degree (Vandenbroucke 2015). However, mounting

economic evidence suggests the labor market is increasingly

credentialized, and hence persistent higher education wage gaps

reﬂect worsening outcomes for those without degrees—thus the

growing imperative of obtaining a higher education credential,

even as the costs are shifted to individual students.

These beneﬁts, combined with wider opportunities in both

the labor market and the higher education system for women

and people of color, have driven rates of college attainment

among adults in the United States from less than one in ten 50

years ago to a record high of approximately one in three today

(US Census Bureau, Table A-2). Similarly, higher education is a

valuable social investment, with positive spillover effects from

generating new knowledge and expanding skills at both local

and national levels. The higher incomes associated with a col-

lege degree represent greater productive capacity and a higher

value of human capital stock economy-wide. These direct and

often clearly monetized gains accrue in terms of skills, income,

and increased productivity that contribute to GDP growth and

rising living standards for the entire economy.

Recent research from the Organisation for Economic

Co-operation and Development (OECD 2015) shows that the

social beneﬁts of public spending on higher education far out-

weigh the public costs. In the United States, lower unemploy-

ment rates, higher tax revenues, and other social contributions

associated with educated workers result in net social beneﬁts

worth between two and ﬁve times the dollar amount of public

spending on higher education. Yet even these impressive ﬁgures

capture only part of the gains. Research shows that better-edu-

cated people live longer, healthier lives, commit fewer crimes,

and are more civically engaged (OECD 2015). Higher educa-

tion plays a key role in our nation’s socioeconomic mobility

and, as a result, access to higher education is a crucial facet of

equality of opportunity for young people and families hoping

to achieve a better life. Finally, an increasingly productive and

highly educated society yields intergenerational advantages, as

the associated institutions, networks, and aptitudes are passed

down over time. Accounting for each of these beneﬁts would

raise estimates of the reward for society beyond the OECD ﬁg-

ures by reducing public expenditures on health care and crime,

improving quality of life, and contributing to equality of oppor-

tunity and political stability.

The social beneﬁts of increasing educational attainment

are dispersed, generating returns even for those individuals who

choose to forego a college degree. These dispersed beneﬁts in

10 Student Debt Cancellation Report 2018

the market for higher education are a classic example of posi-

tive externalities—beneﬁts accruing as a result of exchange that

are not taken into account by private buyers and sellers. Like

other markets where positive externalities exist, the omission of

social beneﬁts in individual cost/beneﬁt calculations results in a

market failure. A higher education market composed of purely

private exchange would lead to conditions of chronic underin-

vestment. The key to avoiding a market failure and capturing

the social returns of an educated population is public support.

The United States has a history of ﬁnancing higher education in

partnership with students and their families, with the majority

of college students attending public institutions supported by

state and federal spending (NCES 2015, Table 303.7). Yet over

recent decades that partnership has devolved and individuals

are taking on a growing share of the cost of higher education.

The rising individual cost burden of attaining a college

degree also has spillover effects on the rest of the economy.

With declining state support and a persistent social and eco-

nomic demand for college credentials, postsecondary educa-

tion is increasingly ﬁnanced through debt. This debt weighs on

household ﬁnances, affecting the consumption and investment

opportunities of borrowers, with ripple effects across other con-

sumer debt markets and beyond. Debt service payments reduce

disposable income and consumption spending. College gradu-

ates focused on paying down debt are putting off other invest-

ments, like buying a home or starting a family—or taking on

yet more debt to obtain graduate degrees that are increasingly

necessary as the labor market credentializes. And as the indi-

vidual investment perspective drives a greater share of the mar-

ket, society risks losing valuable beneﬁts to a higher education

market failure.

The current state of student debt

More than ever before, Americans recognize higher education

as an important milestone on the pathway toward prosperity

and ﬁnancial stability. As a result, waning public support for

higher education and rising individual costs have prompted

the growth of student debt to record levels. According to the

Federal Reserve (2016), outstanding student loan debt totaled

$1.35 trillion as of the ﬁrst quarter of 2016—an amount 28 per-

cent greater than all motor vehicle loans and 40 percent greater

than the value of outstanding student loan debt just ﬁve years

ago. The vast majority of this debt originates from federal lend-

ing, with the private student loan market accounting for just

7.6 percent ($99.7 million) of all student debt (MeasureOne

2015). The growth in borrowing occurred as more college stu-

dents turned to loans to ﬁnance their education and the typi-

cal loan amount per borrower increased (see Figure 1.1). In the

1989–90 academic year, 50.5 percent of undergraduate seniors

ages 18–24 relied on student loans for some portion of their

college costs. The average loan amount among those borrow-

ing was $15,200. At the end of the 2011–12 academic year, 68

Average Loan Balance $26,300

0 10 20 30 40 50 60 70 80 90

100

1989–90

1999–2000

2011–12

Students Who Received Student Loans

Students Who Never Received Student Loans

Figure 1.1 Percentage of Undergraduate Seniors Who

ceived Student Loans

Source: NCES 2014 Digest of Education Statistics, Ta ble 331.95

Average Loan Balance $22,100

Average Loan Balance $15,200

Percent

Share Cumulative

Borrowing Borrowing

Total 69% $29,384

Race/Ethnicity (with multiple)

White 67.5% $29,065

Black or African American 84.2% $33,015

Hispanic or Latino 71.9% $29,517

Asian American 47.2% $23,135

American Indian or Alaska Native 61.5% ‡

Native Hawaiian/Other Paciﬁc Islander ‡ ‡

Other 81.0% $28,052

More Than One Race ‡ ‡

Institution Type

Public Four-Year 64.1% $25,458

Private Not-for-Proﬁt Four-Year 73.1% $32,388

Public Two-Year ‡ ‡

Private For-Proﬁt 87.2% $40,025

Others/Attended More Than One School 71.8% $29,444

Table 1.1 Share Borrowing and Cumulative Amount

Borrowed for Undergraduate Education among Those

with Debt, by Race/Ethnicity and by Institution Type for

Graduating Seniors, 2011–12

‡ Reporting standards not met.

Source: NCES 2011–12 National Postsecondary Student Aid Study (NPSAS:12)

Levy Economics Institute of Bard College 11

percent of graduating seniors left college with some student

debt and the average balance rose to over $26,000 (NCES 2014,

Table 331.95). The growing reliance on student loans marks an

important generational shift as today’s graduates begin their

working lives hampered by debt payments that preclude other

economic opportunities.

Today the majority of college students incur student debt,

but the degree of exposure differs by critical demographic fac-

tors such as race and ethnicity and the type of academic institu-

tion attended (Table 1.1). Students at public institutions were

less likely to borrow than those at private for-proﬁt or not-for-

proﬁt institutions, but even at public schools most students

relied on loans to some extent (NCES 2013). Black and Latino

students are the most likely to take out loans and borrow greater

amounts. In 2011–12, 84 percent of black graduating seniors

had borrowed for college, as had 72 percent of Latinos, 68 per-

cent of whites, and 48 percent of Asian Americans (NCES 2013).

Black and Latino graduates, whose household ﬁnances are

already affected by racial gaps in wealth, income, and employ-

ment—even with a college degree—encounter a dispropor-

tionate burden as debt payments after graduation constitute a

larger portion of household budgets.

Recent research from the

Mapping Student Debt project (Steinbaum and Vaghul 2015)

shows that even below-average student loan balances can be

problematic for low-income borrowers choosing between mak-

ing on-time payments and other ﬁnancial demands. And zip

codes with high minority populations are signiﬁcantly more

likely to be burdened by their student debt payments (as a per-

centage of their income), and thus to go delinquent on their

loans.

The growth in student loan debt is driven by several impor-

tant factors, including a rapid rise in the cost of attaining a

degree (Figure 1.2) at the same time a college education became

increasingly associated with individual advancement and eco-

nomic success. After remaining relatively stable for decades,

average tuition and fees at public, undergraduate, four-year

institutions rose 156 percent between the 1990–91 and 2014–15

academic years. The total price tag, including tuition, fees, and

room and board, doubled over the period to reach $18,632 in

2014–15 (NCES 2015, Table 330.10).

The impact of this growth in the cost of college was magni-

ﬁed by a protracted period of earnings stagnation for the typical

American household (Figure 1.3). According to the US Census

Bureau, real median household income was just 2 percent higher

in 2014 than it was in 1990 (US Census Bureau, Table H-9). The

rising cost of attendance over this period required college stu-

dents and their families to devote a growing portion of house-

hold budgets to higher education (Figure 1.4). In 1990, before

the rapid escalation of the college price tag, average tuition and

fees amounted to 6.3 percent of median household income,

17.6 percent when room and board are included. In 2014, aver-

age tuition and fees for one year of college would require 15.9

$2,000

$4,000

$6,000

$8,000

$10,000

$12,000

$14,000

$16,000

$18,000

$20,000

1963–64

1965–66

1967–68

1969–70

1971–72

1973–74

1975–76

1977–78

1979–80

1981–82

1983–84

1985–86

1987–88

1989–90

1991–92

1993–94

1995–96

1997–98

1999–2000

2001–02

2003–04

2005–06

2007–08

2009–10

2011–12

2013–14

Tuition, Fees, Room and Board

Tuition and Fees

Sour

ce: NCES 2015 Digest of Education Statistics, Table 330.10

Figure 1.2

Cost of Full-Time Attendance at Four-Year Public Institutions (2014–15 dollars)

12 Student Debt Cancellation Report 2018

percent of the median household income, or 34.7 percent with

the inclusion of room and board (NCES 2015, Table 330.10; US

Census Bureau, Table H-9; authors’ calculation). For families

across the income distribution seeking a college credential, the

much faster growth of college costs than household incomes

over this period made debt an essential resource.

One major cause of the increases in the price of college and

incidence and amount of student debt over this period was the

decline in public funding as a share of the cost of education.

Although not the sole source of rising prices, analysis from the

Federal Reserve Bank of New York (among others) ﬁnds that

decreases in state and local higher education appropriations are

associated with net tuition increases, especially since the 2008–

09 recession (Chakrabarti, Mabutas, and Zafar 2012). According

to the State Higher Education Executive Ofﬁcers Association

(SHEEO), which tracks the contributions of public funding,

tuition payments, and other revenue sources in higher educa-

tion, real per-student funding last year was 15.3 percent lower

than in 2008 and 20 percent below its 1990 level (SHEEO 2015).

These decreases in public funding represent a signiﬁcant shift in

higher education ﬁnance over recent decades. As public funds

receded in importance, tuition payments made up an increasing

share of revenues. In 1990, tuition accounted for 25 percent of

revenues at public institutions. In 2015, the share of revenues

drawn from tuition payments was 46.5 percent (SHEEO 2015).

These changes have made a college degree less affordable for

American families, contributing to the rise in student debt, and

have transformed public higher education from a social invest-

ment to an increasingly private one.

The social costs of student debt

Student loan debt today places a signiﬁcantly higher burden on

household ﬁnances than ever before, with implications for the

entire economy. Many borrowers struggle to make payments

due to unemployment, low incomes, and competing ﬁnancial

demands. The US Department of Education publishes default

rates for student loan borrowers who have failed to make pay-

ments for at least 270 days within the ﬁrst three years after leav-

ing college. Among those whose loans entered repayment in

2012, 11.8 percent of borrowers have failed to meet their obli-

gations for at least nine months (US Department of Education

2017). The consequences of default can be severe. Once a loan

is delinquent for 90 days it is reported to credit rating agencies;

at the 270-day default threshold it is assigned to a collection

agency (US Department of Education 2016). For households in

debt, delinquency and damaged credit can make it impossible

to purchase a house, a car, or even a cell phone plan. For the gov-

ernment lender, default raises the administrative cost of student

loan programs and heightens risk.

Even borrowers who are current in their payments face

additional constraints as a result of the growing reliance on debt

to ﬁnance higher education. Households with student loan bal-

ances are less likely to own a home and, consequentially, exhibit

lower net worth than comparable households with no student

debt. According to recent research by the Board of Governors

of the Federal Reserve System (Mezza et al. 2016), every 10 per-

cent rise in student loan debt reduces the homeownership rate

of borrowers by 1 to 2 percentage points in the ﬁrst ﬁve years

after leaving school. When student loan borrowers do purchase

Figure 1.4 Average Tuition and Fees as a Share of Median

Household Income

Source: NCES 2015 Digest of Education Statistics, Table 330.10; US Census

Historical Income Tables H-9

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

Tuition and Fees as a Share of Median Income

Tuition, Fees, Room and Board as a Share of Median Income

Percent

Figure 1.3 Growth in Cost of College Attendance and

dian Household Income (1990=1)

0.5

1.5

2.5

1990 1995 2000 2005 2010

Tuition and Fees, Indexed to 1990

Tuition, Fees, Room and Board, Indexed to 1990

Median Income, Indexed to 1990

Sour

ce: NCES 2015 Digest of Education Statistics, Table 330.10; US Census

istorical Income Ta bles H-9

Levy Economics Institute of Bard College 13

homes, they build equity more slowly. Researchers at the

Washington University Center for Social Development (Elliot,

Grintstein-Weiss, and Nam 2013) ﬁnd that the home equity of

student loan debtors amounts to just half that of nonborrow-

ers. Since home equity is a central wealth-building vehicle for

American households, the interaction between student debt

and homeownership in the short run can have long-run conse-

quences for borrowers and feedback effects on other markets as

consumption patterns respond to lower wealth.

The broader economic effects of high student debt burdens

also manifest across markets in the form of reductions in busi-

ness formation, greater household vulnerability to economic

shocks, and reduced consumption spending due to lower dis-

posable income. Research from the Federal Reserve Bank of

Philadelphia (Ambrose, Cordell, and Ma 2015) demonstrates

that relatively high student loan debt dampens business forma-

tion, with the strongest effects among the smallest ﬁrms. Small

business owners in particular tend to rely on personal credit as

a substantial portion of business ﬁnancing, and student loan

debt reduces access to alternative forms of credit. The authors of

the Philadelphia Fed study deduce a direct negative relationship

between student loan debt levels and small business formation,

amounting to a 14 percent decline in new businesses in coun-

ties where student loan debts are highest. Student loan debt also

creates disproportionate exposure to economic downturns. A

study from the St. Louis Federal Reserve (Elliot and Nam 2013)

shows that during the most recent recession households with

student debt experienced greater reductions in net worth than

households with no student debt. According to the authors,

every $1 increase in student loan debt in 2007 was associated

with $0.87 less net worth in 2009. Finally, given that rising loan

payments drain households’ disposable incomes, student debt

potentially presents new headwinds in the economy. While

increasing household debt drove higher spending in the years

leading up to the 2008–09 recession, households deleveraged in

the wake of the crisis, resulting in a drag on the economy as

consumer spending lagged (Albuquerque and Krustev 2015).

During the period that households paid down other forms of

debt, student loans continued to grow (Brown et al. 2014). Since

consumer spending depends on both wealth and income—each

of which is affected by the presence of debt and debt service

obligations—soaring student loan debt can result in slower

growth in the economy overall.

The rising burden of college costs on household balance

sheets has been accompanied by new risks, including slow

growth, deepening vulnerability to economic shocks, and the

potential for a higher education market failure. Complete can-

cellation of outstanding student loans could undo many of

these negative effects. By reversing the drag imposed by $1.35

trillion in outstanding student loans, we expect a net stimu-

lus to the economy through housing markets, small business

formation, growth in consumer spending, and the feedback

effects these changes create. These directly measurable effects of

student debt cancellation would be complemented by unmea-

sured social beneﬁts like greater social mobility and quality of

life. Based on this research, student debt cancellation presents a

signiﬁcant economic opportunity not only for the households

burdened by debt but for the entire US economy.

The Distributional Consequences of Student Debt,

Student Debt Cancellation, and Debt-Free College

This report ﬁnds that cancelling all outstanding student loan

debt would modestly improve output and employment.

However, the main controversy over student debt generally and

debt cancellation in particular has not been its macroeconomic

impact, but rather the implications for people in different

income and wealth quantiles and the impact on inequality. The

controversy arises from the factual observation that among bor-

rowers, those with the largest amount of debt outstanding tend

to have the highest incomes, and those who spend the most on

college (and who therefore—so the story goes—have the most

to gain from the option of free college) come from the highest-

earning families.

These observations have been widely interpreted as discred-

iting the sorts of policies we model here. For example, Sandy

Baum (2016) of the Urban Institute writes “forgiving all student

debt is such a misguided idea that it is hard to know where to

begin…. Because households in the top quartile of the income

distribution owe a disproportionate amount of student debt,

they would reap a disproportionate amount of the beneﬁts if

all education debt were forgiven.” Jordan Weissmann (2016) of

Slate was even more strident: “An especially half-baked idea for

dealing with America’s student debt burden has been bubbling

up from the far reaches of the political left lately: Washington, a

few well-meaning souls say, should just forgive all of the loans—

wipe the slate clean.” He continues: “The most important thing

to realize about student loans is that most borrowers don’t have

too much trouble handling it.”

14 Student Debt Cancellation Report 2018

The widespread criticism of ambitious policies to address

the student debt crisis, based on their supposedly regressive

impact, is overdrawn. In some cases, it misinterprets the evi-

dence about who is most burdened by student debt and who

would beneﬁt most from relief. In this section, we consider the

evidence about the distribution of debt and debt burdens in

the population and the evolution of those distributions over

time. Our main point is that, while the largest loan balances

are indeed held by comparatively high-earning households, the

extent to which student debt is held by the rich has diminished

signiﬁcantly. Moreover, the argument that the distribution of

the burden of student debt has not, in fact, changed very much,

even as the total amount of debt outstanding has increased dra-

matically, fails to consider the signiﬁcant changes in the popu-

lation of people with any student debt at all. These issues of

interpretation extend beyond accurately assessing the distribu-

tional impact of the policies we model—they point to larger

problems with the assumptions behind existing higher educa-

tion, student debt, and labor market policies. The student debt

crisis is one of several linked manifestations of those problems.

Others are wage stagnation, underemployment, and increasing

inequality of household wealth.

Student debt was once disproportionately associated with

graduate school and with relatively well-off households, in part

because it was possible to graduate from community college or

a four-year public institution with little or no debt, and in even

larger part because many people did not need to obtain any

higher education credentials in order to access the labor market.

What has happened in recent decades, and especially since the

mid-2000s, is a vast expansion of student borrowing, such that

the preponderant share of younger cohorts newly entering the

labor market carry student debt. This expansion is due in part

to much higher tuition, mostly thanks to state-level cutbacks in

funding for higher education, and in part because it is simply

far more difﬁcult to access the labor market now without higher

education credentials. And that “credentialization,” in turn, is

due to the underperformance of the labor market since 2000

and especially since the ﬁnancial crisis and the Great Recession

that began in 2008. Since 2000, the most important federal labor

market policy has been the extension of student debt and the

encouragement of a larger share of the population to obtain

debt-ﬁnanced higher education credentials, on the theory that

underemployment and stagnant wages were caused by a “skills

gap” that could be remedied through debt-ﬁnanced higher edu-

cation. The most obvious and acute effect of that policy was

the growth of the high-priced for-proﬁt higher education sec-

tor, but it was also evident in rising enrollment across all types

of institutions, even as tuition rose. The “skills gap” was a false

diagnosis of the labor market’s problems, and hence the pre-

scription of more debt-ﬁnanced credentials not only failed to

solve the problem, it also created its own problem in the form

of unsustainable debt.

The distribution of student debt and debt burden in the

cross section

The total amount of student debt outstanding is dispropor-

tionately held by those in the top income quintile. Looney and

Yannelis (2015) report that just over a third of all outstanding

debt is held by the top 20 percent, as deﬁned either by labor

market earnings from the Current Population Survey or by total

taxable income as reported in tax return data.

By contrast, only

13 percent of debt is held by the lowest quintile as determined

by labor market earnings, or 15 percent as determined by tax-

able income. Thus, debt outstanding is more skewed toward

the rich than is the distribution of the number of borrowers:

25 percent of federal borrowers were in the top quintile. The

reason for that is simply that the borrowers with the largest bal-

ances have a disproportionate share of the debt: a total of 62

percent of all outstanding debt is held by the top quintile of

the distribution of borrowers (as opposed to the distribution

of income). The borrowers with the largest balances have the

majority of total debt, and they are largely also drawn from the

richest households. These ﬁndings are the basis for the claim

that student debt cancellation would be regressive.

One thing that is not directly reported in the Looney and

Yannelis data is the distribution of income among quintiles of

the student debt distribution. According to the authors’ calcula-

tions, it appears that the top quintile of indebted households by

either labor market earnings or total taxable income earns at

minimum 50 percent of the total earnings/taxable income

—

which is greater than their share of debt (between 33 and 36

percent). This means that the richest households have lower

total debt-to-income ratios than households in the middle or

bottom of the distribution, even if they do have the highest debt

loads.

Looney and Yannelis do not report the distribution of

“debt burdens”—meaning the student-debt-service-to-income

ratio—across households by income. Instead, they report

on the distribution of debt burdens for borrowers with debt,

regardless of their income. They ﬁnd that the median burden

Levy Economics Institute of Bard College 15

for borrowers two years after entering repayment in 2010 was 6

percent, a number that has not changed much over the length

of their sample (which, for the purpose of calculating debt

service, begins in 1999). Akers and Chingos (2014) similarly

compute the distribution of debt burdens from a different data

source, the Survey of Consumer Finances (SCF). According to

their analysis of the 2013 SCF, the median debt burden among

households headed by individuals aged 20–40 with any student

debt and positive labor market income was 4 percent, with a

16 percent debt burden being the cutoff for the 90th percentile.

In an unpublished analysis of credit reporting data linked

to American Community Survey outcomes at the zip code level,

Steinbaum and Vaghul (forthcoming) found that the burden of

student debt (meaning the ratio of debt service to income) is,

in general, inversely correlated with the average income of a zip

code. But that pattern obscures a slightly more nuanced geogra-

phy. The debt burden is highest in two types of zip codes: those

in historically disadvantaged areas, urban and rural, where both

incomes and absolute loan amounts are low; and those in rela-

tively high-earning urban areas disproportionately populated

by young cohorts with a large amount of student debt accumu-

lated and as-yet relatively low earnings given their credentials.

That pattern suggests that student debt is a problem of those

with low incomes relative to their credentials. It does not sug-

gest that those with credentials are necessarily ﬁnancially secure.

The evolution of the distribution of student debt burdens

over time

One of the key ﬁndings motivating the argument that there is no

student debt crisis is that the distribution of student debt bur-

dens, conditional on having any student debt, has not changed

very much over time. The aforementioned study by Akers and

Chingos shows that the median payment-to-income ratio of

4 percent for 2013 has barely moved over successive waves of

the SCF. The right tail of that distribution has lengthened over

time: as reported above, the threshold for the 90th percentile of

households in their sample in 2013 is a debt burden of 16 per-

cent of income, whereas that threshold was 13 percent in 2007.

But it was much higher than that longer ago: 22 percent in 1998

and 20 percent in 1992. Akers and Chingos interpret these ﬁnd-

ings to mean that borrowers are, on the whole, not signiﬁcantly

more burdened by student debt now than they were in the past,

and hence there is little to worry about in the way of a student

debt crisis.

That is, quite simply, a misinterpretation of the data. The

set of households that has any student debt at all is quite differ-

ent now than it was a decade or two ago. Households that would

have appeared as “zeroes”—that is, not included—in the com-

putation of the student debt burden distributions in the 1990s

or the mid-2000s now enter those distributions with positive

values for their debt burdens. Akers and Chingos condition on

positive student debt to include households in their sample,

but if they had instead conditioned on a given level or range

of income or on a given educational attainment, they would

have found that the distribution of debt burdens had shifted

substantially to the right. In other words, to reach a given rung

on the “job ladder” that characterizes the labor market requires

more education credentials and therefore more student debt

now than it did in the past, while the earnings available on any

given rung have, for the most part, either stagnated or declined.

It may be the case that the median debt burden has not

changed very much even as the total amount of debt outstand-

ing has increased, but the median borrower has changed a great

deal. Whereas once debtors were likely to come from the ranks

of “traditional students”—that is, those attending either gradu-

ate school or private four-year institutions right after high

school, often with a family history of higher education and with

the family wealth to accompany it—recent student cohorts are

much more likely to be nontraditional students, often begin-

ning later in life and without a family background of college

attendance. This is exacerbated by rising costs associated with

the withdrawal of state support for higher education, which is

itself often regressive in the sense that the ﬁnancially worst-off

institutions, which tend to serve nontraditional populations,

face the steepest budget cuts relative to enrollment and lack the

cushion of strong alumni support to fall back on. For all of those

reasons, student debt increasingly burdens a (growing) share of

the population that is ill-situated to carry and repay that debt.

An additional ﬁnding that highlights the growing burden

of student debt is the deteriorating repayment rates across

cohorts, especially following the Great Recession. Cohorts

entering repayment since 2011 are carrying more debt now than

they were in 2011 (Looney and Yannelis 2015). That means they

have made negative progress in absolute terms, and are nowhere

close to the benchmark of steady and full repayment over 10

years—thanks to low earnings, high delinquency rates, and a

labor market that demands ever-more credentials just to get

a job, necessitating more and more degrees and more debt. A

recent report by the National Center for Education Statistics

16 Student Debt Cancellation Report 2018

(Woo et al. 2017) found that repayment rates were worse and

delinquency rates much higher for the cohort of students who

entered college in 2003–04 compared to those who entered in

1995–96, and these disparities were particularly true of minor-

ity students, even those who completed their degrees. That dire

picture—which would in all likelihood appear even worse if it

included data on students who entered college during or after

the Great Recession—makes it hard to tell the old story that

going into debt to ﬁnance higher education is a route to the

middle class, and that the debt is easily affordable thanks to the

increased earnings graduates can expect.

What does the evolution of student debt tell us about the

labor market?

The country’s most signiﬁcant labor market policy of the past

several decades is the growth of the federal student loan pro-

gram to ﬁnance higher education for a rising share of the pop-

ulation of entering cohorts across many types of institutions:

for-proﬁt, two-year public, four-year public and private tradi-

tional undergraduate, and graduate school. The theory behind

that ambitious policy was that the labor market is characterized

by a “skills gap,” whereby workers lack the skills necessary to

succeed in the global economy, and increasing higher education

attainment would close that gap.

Moreover, thanks to the higher

wages students could expect when they joined the labor market,

the beneﬁciaries of that expansion would be able to ﬁnance it

with student loans. The skills gap offered a conveniently uniﬁed

explanation both for earnings and wage stagnation overall and

for rising inequality in earnings among workers: those enjoying

large increases, so the story goes, were the ones with the skills

that positioned them to succeed in a changing and competitive

global economy.

The theory does not look very good in light of the student

debt crisis, as well as other labor market indicators: wage and

earnings stagnation even for workers with higher education

qualiﬁcations and the increasing credentialization that sees

workers with degrees take jobs that did not previously require

them (Clark, Joubert, and Maurel 2014). If the skills gap had

indeed been the problem, as student debt increased enormously

alongside degree attainment, so would have “human capital,”

and thus both aggregate earnings and the earnings of the newly

educated (who would not have gotten degrees had they been

part of earlier cohorts) would have increased. But, in fact, as

the share of the population with each level of credential has

increased—at the expense of the share with no credentials at

all—the effect has been to degrade the value of each credential

in terms of what jobs and earnings its holders can expect. The

result is a classic “rat race,” in the sense that the only thing worse

than taking on burdensome debt in order to ﬁnance an increas-

ingly expensive credential that leads to worse outcomes than it

did previously is not doing so. That rat race only strengthens the

argument that the current policy of encouraging the expansion

of debt-ﬁnanced higher education has been a failure, and there-

fore a radical departure is in order.

Many researchers who have investigated the student debt

crisis, and in particular the issue of delinquency on student

loans, agree that the weak labor market is the prime culprit

(Dynarski 2016; Muller and Yannelis 2017). But that does not

mean that higher education or student debt policy is not to

blame, because they are premised on the labor market value

of debt-ﬁnanced higher education. Rising delinquency is one

manifestation of the failure of that premise, and the fact that

credentialed workers continue to take jobs at the expense of

those without credentials (who increasingly lack access to

employment opportunities at all) does not imply that continu-

ing or escalating the rat race is a sound policy. Yet that is what

is implied by continued insistence on the stability of the college

wage premium—thanks to deteriorating outcomes for those

without credentials—as justiﬁcation for the status quo. On the

contrary: it is time to undertake a real labor market policy, and

to overcome squeamishness about acknowledging the failures

of the status quo. This includes acknowledging that student

debt accumulated to date might not be economically feasible

for debtors to carry and, eventually, pay off.

How does student debt interact with longstanding economic

disparities?

One inescapable conclusion from studying the incidence and

impact of student debt is that black and Hispanic borrowers

suffer disproportionately from its effect (Steinbaum and Vaghul

2016; Huelsman 2015a). Delinquency is highest in minority

neighborhoods and, for a given level of educational credential,

minority borrowers take on more debt and have more trouble

paying it off. The causes of that disparate impact are longstand-

ing racial disparities in both the credit and labor markets, house-

hold and family wealth far lower than white counterparts (even

for those making similar incomes), and segregation in higher

education itself, which clusters minority students in poorly

resourced institutions that are the least likely to lead to the

most valuable employment networks. The for-proﬁt education

Levy Economics Institute of Bard College 17

sector, where the most acute victims of the student debt crisis

are to be found, itself exists because more economic options are

often effectively closed off to nontraditional students, who are

likely to be minorities (McMillan Cottom 2017).

The pattern of racial disparity in student debt is especially

injurious because, as the engine of social mobility, higher edu-

cation is supposed to be the solution to disparities in back-

ground and family wealth. But what we are seeing now with the

student debt crisis repeats the themes that came to light during

the housing crisis and the Great Recession. Financing the pur-

chase of an asset with debt as a supposed mechanism of social

mobility is a facile policy for closing racial gaps when it fails to

account for disparities in access and quality. As with housing, so

with higher education: a mythology associated with its value as

an agent of social mobility gives rise to policies that encourage

borrowing to ﬁnance its acquisition. When the asset turns out

to be worth much less than promised, it is those who began at

the greatest disadvantage who are left holding the most burden-

some debts and who struggle to pay them off.

It is for this reason that a “public option” for higher educa-

tion would be worth the most to the most disadvantaged stu-

dents (Huelsman 2015b). The supposedly egalitarian argument

against free or debt-free college is that those who currently

spend the most on higher education are those from the most

well-off backgrounds, and they would have the most to gain by

switching to the cheaper option. But that ﬂies in the face of evi-

dence about the impact of universal public services: the rich do

not, generally, switch to them. Instead, it is those who lacked all

access before or who were forced to use substandard providers

who will beneﬁt most from an affordable and universally acces-

sible option and the competitive pressure that puts on incum-

bent providers who had previously been able to beneﬁt from a

captive, segmented market.

The real distributional impact of student debt cancellation

and free or debt-free college

The arguments against student debt cancellation and free or

debt-free college arising from their ostensibly inegalitarian

impact fail to account for the reality of the student debt crisis

and those who are its greatest victims. In this report, we show

that a radical policy for ending the student debt crisis could be

undertaken without doing economic harm—in fact, to moder-

ate macroeconomic beneﬁt. That does not mean it is the only

possible solution, but there is no reason to constrain the policy

space based on outdated assumptions about the distribution

of student debt and its implications for both the underper-

formance of the labor market and for longstanding economic

disparities.

18 Student Debt Cancellation Report 2018

Section 2: The Mechanics of

Student Debt Cancellation

This section will describe the mechanics of student debt cancel-

lation carried out by the federal government and the Federal

Reserve. As a basis for understanding and explaining the instru-

ments of debt relief, we ﬁrst examine the current mechan-

ics of student debt and demonstrate balance sheet effects via

T-accounts throughout. T-accounts are tools that track changes

in assets (labeled A) and liabilities or equity (L/E) that result

from transactions or debt cancellations. It is important to keep

in mind, however, that the ﬁnancial statement effects shown

and discussed here are not the same as the ultimate macroeco-

nomic effects that emerge from the policy. That is, the ﬁnancial

statement effects shown in the T-accounts are the immediate

ones for the government, private investors, and student loan

borrowers. Additional effects—multiplier effects, as they are

often labeled—and feedback effects inﬂuencing the federal gov-

ernment’s budget position (as in so-called “dynamic scoring”)

are by necessity omitted. These important but omitted dynamic

effects will be explicitly considered in Section 3, using large

macroeconometric models to simulate the debt cancellation.

There are two rationales for presenting these mechanics.

First, they present the direct effects of student debt cancellation

on government spending and revenue, household income and

net ﬁnancial wealth, and private investors; these are the transac-

tions incorporated into the macroeconometric models in the

following section. Second, and most fundamentally, because

every transaction in the economy affects the ﬁnancial state-

ments of those involved, knowing how ﬁnancial statements are

affected is a basic prerequisite to fully comprehending the trans-

action itself (which is the key unit of analysis in the ﬁeld of eco-

nomics). This ﬁnal point also relates to the core concern of this

report, which is the macroeconomics of student debt cancella-

tion: while there are competing approaches to “accounting” for

the government’s cost of student loan programs (such as “fair

value”), for the purposes of understanding the macroeconomic

effects it is the impacts on ﬁnancial statements of the various

sectors that are of interest.

The following analysis results in two key takeaways:

• The cancellation of the Department of Education’s loans will

result in an absolute increase in the national debt equal to the

debt service due on the securities previously issued to fund

these loans, not the amount of the loans themselves.

• There is no “free lunch” if the Federal Reserve carries out stu-

dent debt cancellation instead of the federal government. That

is, the direct effects upon the federal government’s immedi-

ate and future budget position would be essentially the same

whether the federal government or Federal Reserve carries

out the cancellation. As a later part of this section explains,

however, there are a few potential caveats to this conclusion,

involving what some might consider creative accounting on

the part of the Federal Reserve—though for several reasons

these accounting maneuvers would probably require changes

to current laws and are likely to be politically contentious.

The Mechanics of Student Debt Cancellation Carried

Out by the Government

Current servicing of student loans from a balance sheet

perspective

It is useful to ﬁrst consider the effects of current student loan

repayment on the balance sheets of the borrowers, govern-

ment, and private owners of student loans. Table 2.1 shows the

ﬁnancial statement effects of debt service on loans held by the

Department of Education (hereafter, ED) using T-accounts.

Table 2.1, as with all T-account tables in this section, shows

T-accounts for the federal government, Federal Reserve, banks,

student loan borrowers, and private investors. When borrowers

make a student loan payment, it reduces their deposits as the

payment is settled. Borrowers are assumed to be households, so

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Tsy Acct (p,i) -RBs (p,i) -RBs (p,i) -D

(p,i) -D

(p,i) -Loans (p)

-Loans (p) +Eq

Govt

(i) +Tsy Acct (p,i) -Eq

(i)

Table 2.1 Balance Sheet Effects of Servicing of Student Loans Owned by Department of Education

Borrowers

Service ED

Loans

Levy Economics Institute of Bard College 19

the reduction in the deposits of households is represented -D

in the table. The payments also reduce the amount of outstand-

ing student loans (-Loans). In addition, because the payments

include an interest component, the payments and thus reduc-

tion in deposits are greater than the reduction in student loans,

reducing the borrowers’ equity (-Eq

). On the banks’ balance

sheet, the banks settle the borrowers’ payment to the ED by

reducing the borrowers’ deposits (-D

, which reduces bank

liabilities), while their own reserve balances held at the Fed are

debited (-RB, an offsetting reduction in assets). On the Federal

Reserve’s balance sheet, banks’ reserve balances (RBs) are deb-

ited while the Treasury’s account at the Fed is credited (+Tsy

Acct). On the federal government’s balance sheet, the credit to

the Treasury’s account (+Tsy Acct)—which would internally be

credited to the ED—reduces outstanding student loans (-Loans)

while the interest portion of the debt service payment raises the

net worth or equity of the federal government (+Eq

Govt

). Finally,

(p) refers to a payment of loan principal, or an entry of the same

size; (i) refers to the same for the interest portion of debt service

payments; and (p,i) refers to a payment of principal and inter-

est, or an entry of the same combined size. While the mechanics

are a bit complex due to the necessary inclusion of the banks

and the Fed, the net effect is to reduce student loans outstanding

while transferring deposits and net worth to the federal govern-

ment via the Treasury’s account at the Fed.

Changes in balance sheets for the servicing of student

loans held by the private sector are shown in Table 2.2. Here

the entries for the borrowers are the same as in Table 2.1. The

private investors will be credited with deposits (+D

) by their

banks, which, as in Table 2.1, will be larger than the reduction

in loan principal, thereby raising equity (+Eq

). Banks in the

aggregate simply debit the accounts of the borrowers and credit

the accounts of investors. (To avoid unnecessary complexity,

Table 2.2 abstracts from transfers of reserve balances between

the borrowers’ banks and the investors’ banks.) It is worth men-

tioning that while the servicing of student loans results in both

equity and deposits effectively being transferred from debtor to

creditor in both Tables 2.1 and 2.2, in the latter the transfers

are within the private sector and thus alter the distribution of

deposits and net worth within the sector, whereas in the for-

mer there is a reduction in the private sector’s total net worth as

well as a reduction in the total quantity of deposits (that is, the

“money supply”).

Possible methods of government-ﬁnanced student debt

cancellation

Though there may be many details, the mechanics of debt can-

cellation are simple in general. The primary actions would be

the following:

• The current portfolio of student loans held by the ED would

be cancelled or, equivalently, borrowers would simply be

allowed to stop making payments and any principal due on

a given date would be cancelled at that time (that is, the loan

would effectively be cancelled in stages as payments come

due). As of the second quarter of 2016, the ED’s outstanding

loans totaled $986.19 billion.

• The federal government would either purchase and then can-

cel, or, equivalently, take over the payments on student debt

currently held by the private sector. As with the ED’s loans,

if the government purchases the privately held loans it can

choose to cancel them immediately or as borrowers’ payments

come due. The government-guaranteed loans are $266.69 bil-

lion, while nonguaranteed privately issued loans are $101.58

billion, both as of the second quarter of 2016. Having the gov-

ernment assume these payments or purchase and cancel the

loans is preferable to cancellation by private investors. The

latter would require the private sector to write down nearly

$370 billion in both assets and equity, which could be highly

destabilizing (or worse) for the affected sectors.

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-D

(p,i) -D

(p,i) -Loans (p) +D

(p,i) +Eq

(i)

(p,i) -Eq

(i) -Loans (p)

Table 2.2 Balance Sheet Effects of Servicing of Student Loans Owned by Private Investors

Borrowers

Service Private

Loans

20 Student Debt Cancellation Report 2018

The government cancels the Department of Education’s loans

all at once

The ED’s loans were originally funded via sale of Treasury

securities.

The loan payments from student borrowers then in

theory enable the retirement of the securities, while the govern-

ment proﬁts from the difference between the interest earned on

the ED’s loans less interest on the Treasury securities (though

there are some other considerations such as administrative

costs of the program and whether or not interest on the loans

is tax deductible for the borrowers). The balance sheet effects

are shown in Table 2.3. In the ﬁrst transaction, the govern-

ment increases its liabilities (the Treasury securities, or +Tsys)

and receives payment for them into its account at the Fed when

private investors (not necessarily the same investors as in Table

2.2; at the margin these are Treasury dealers) have their bank

accounts debited while their banks have their reserve balances

debited. The loan reduces the Treasury’s account and adds back

reserve balances and deposits. The deposits are listed as belong-

ing to the academic institution (+D

UNI

for “university” in this

case). The loan simultaneously raises the borrowers’ liabilities

and reduces net worth. Not shown here due to space constraints

is that the deposit would add to the net wealth of the academic

institution (ceteris paribus). While it is often the case that the

borrower receives some of the deposits—given that student loans

regularly cover expenses related to college, not just direct costs of

college itself (tuition, room and board, books, and so forth)—as

the borrower spends these funds, the end result will be a reduc-

tion in net wealth as the complementary entry to the loan. The

ﬁnal entries present totals or net effects of the two transactions.

The cancellation of debt held by the ED is a concurrent

reduction in the federal government’s assets and its equity,

while it is simultaneously a reduction in debt and an increase

in equity for the debtors. This is shown in Table 2.4. Also, while

it reduces the government’s equity, the cancellation of publicly

owned student debt on its own does not increase the debt (i.e.,

liabilities) of the federal government. An increase in the federal

government’s debt can only result from issuing debt directly via

Treasury securities or similar means. For the student loan bor-

rowers, the cancellation reduces debt (the student loans) and

raises net worth, exactly offsetting the totals from Table 2.3.

When a loan is cancelled and equity reduced, an accom-

panying entry charges this reduction in equity to the income

statement, which thereby reduces proﬁts. This reduction in

proﬁts is a noncash charge that brings internal consistency for

the income statement and balance sheet. However, for the fed-

eral government, the income statement position is most often

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Tsy Acct(p) +Tsys(p) -RBs(p) -RBs(p) -D

(p) -D

(p)

+Tsy Acct(p) +Tsys(p)

-Tsy Acct(p) +RBs(p) +RBs(p) +D

UNI

(p) +Loans(p)

+Loans(p) -Tsy Acct(p) -Eq

(p)

+Loans(p) +Tsys(p) -D

(p) +Loans(p) -D

(p)

UNI

(p) -Eq

(p) +Tsys(p)

Table 2.3 Treasury Issues Securities to Private Investors and Department of Education Lends to Student Borrowers

Government

Issues

Securities

ED Lends

to Student

Borrowers

Totals

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Loans (p) -Eq

Govt

(p) -Loans (p)

+Eq

(p)

Table 2.4 Cancellation of Student Loans Owned by Federal Government

ED Loans

Cancelled

Levy Economics Institute of Bard College 21

presented in the context of the government’s budget surplus or

deﬁcit, which is more of a cash ﬂow measure that largely omits

such noncash charges in order to have consistency with the gov-

ernment’s issuing, repurchasing, or retiring of Treasury securi-

ties (that is, changes in the government’s liabilities). In the case

of cancelling the ED’s loans, the budgetary effects (and thus the

effects on the government’s outstanding liabilities) would be

incurred as debt service payments from borrowers that are not

received, reducing the government’s revenues relative to what

would have been projected and relative to expenditures.

However, this neglects a point that has signiﬁcant implica-

tions: the absence of payments from borrowers is not in fact

a transaction but a counterfactual. Considering Tables 2.3 and

2.4 together in order to clarify, recall that the loans themselves

were originally funded by issuing Treasury securities and that

the borrowers’ payments were to have enabled the payment of

interest and retirement of principal on these securities. That is,

absent debt cancellation, the timing of the borrowers’ debt ser-

vice payments and the government’s payments to the security

holders were to have lined up for the most part. Therefore, the

deﬁcits incurred each period that the borrowers’ debt service

is not received do not result in additional government liabili-

ties. Instead, the existing liabilities will be rolled over (that is,

a new security is issued to pay for the maturing one) and the

liabilities incurred when the loans were created simply become

permanent. In the sense that the government’s liabilities were

projected to be reduced via student loans being paid down, the

cancellation results in additional government debt relative to

that baseline, but in terms of the absolute size of the govern-

ment’s outstanding liabilities, the cancellation of the ED’s loans

will result in no addition to the government’s liabilities, rela-

tive to a precancellation baseline (except for the amount of debt

service due).

The government cancels the Department of Education’s loans

as borrowers’ payments come due

The conclusions reached regarding the effect of the cancella-

tion of the ED’s loans on the federal government’s outstanding

liabilities are more clearly shown by assuming the government

instead allows borrowers to cease payments and cancels the

loan principal due each period. In this scenario, it is easier to

illustrate the ﬁnancial statement changes that would occur each

period and to then explain which of them are actual transac-

tions that occur as a result of cancellation and which result from

a comparison to projections without debt cancellation (that is,

are counterfactuals).

Table 2.5 presents the cancellation of principal owed on the

ED’s student loans each period. As the label indicates, the ﬁrst

entry in Table 2.5 simply reverses the transaction in Table 2.1.

This is the counterfactual, which is not an actual transaction

but shows ﬁnancial statement changes relative to the no-cancel-

lation scenario. The second entry is a transaction—the govern-

ment cancels the portion of the loan principal that would have

otherwise been due. This is much like the debt cancellation in

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Tsy Acct(p,i) -Eq

Govt

(i) +RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Loans(p)

+Loans(p) -Tsy Acct(p,i) +Eq

(i)

-Loans(p) -Eq

Govt

(p) -Loans(p)

+Eq

(p)

+Tsy Acct(p,i) +Tsys(p,i) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i)

+Tsy Acct(p,i) +Tsys(p,i)

+Tsys(p,i) +D

(p,i) +D

(p,i) +Eq

(p,i) -D

(p,i)

-Eq

Govt

(p,i) -D

(p,i) +Tsys(p,i)

Table 2.5 Department of Education Loans Cancelled as Principal Comes Due—Relative to No Cancellation

REVERSE

Borrowers

Service

ED Loans

Government

Cancels

Principal Now

Due

Totals

Government

Issues

Securities

22 Student Debt Cancellation Report 2018

Table 2.3, but in this case refers only to the principal due on

the speciﬁc date. The third entry shows the government issu-

ing new securities to offset the reduction in the government’s

budget position. The ﬁnal row of entries presents the totals

or net of the two entries, which shows that the net effect is for

student loan borrowers to have more deposits and higher net

worth than is the case when they make debt service payments,

and for the federal government to have both reduced equity and

increased liabilities equal in size to the foregone principal and

interest payment. Another way of stating this is that student

loan borrowers now have an extra, say, $300 per month on aver-

age not debited from their bank accounts that becomes available

to spend, save, borrow against, or some combination of these,

relative to the no-cancellation scenario.

Because the ﬁrst entry in Table 2.5 is a counterfactual, the

actual transactions are presented in Table 2.6. The ﬁrst two trans-

actions are the same as the second and third entries in Table 2.5.

The signiﬁcance of considering actual transactions is seen in the

combination of transactions two and three in Table 2.6. What

happens is government securities that would have been retired

by debt service payments are coming due. Absent the debt ser-

vice payments, the government instead issues new Treasury

securities to replace or roll over the previously issued securi-

ties. The payment due on Treasury securities is partly principal

and partly interest, so as the government issues new securities

to cover this payment, its liabilities increase by the amount of

interest due. (Note that the payment of interest on the Treasury

securities reduces the government’s equity and raises that of the

investors.) Therefore, the new securities issued will be in this

amount—that is, the national debt does in fact increase mod-

estly each year by the amount of interest due on the securities

issued when the loans were originated. Importantly, however,

the cancellation of the ED’s loans does not increase the national

debt by the value of the loans cancelled, but rather only by the

size of the interest payment due annually on the Treasury secu-

rities that ﬁnanced the loans.

Government-led debt cancellation where the government

assumes payments on student loans issued by private investors

Unlike the ED’s loans, the government does not yet own these

private loans and thus would issue new Treasury securities

to purchase them. Financing the purchase and cancelling the

privately owned loans would be essentially simultaneous, and

therefore the changes to both the budget and liabilities of the

government would also be essentially simultaneous. As with

the ED’s loans, the government could also choose to cancel

debt service as it becomes due. A third option is for the govern-

ment to take over the borrowers’ debt service payments on these

loans, in which case the changes to the government’s budget and

liabilities would be incurred over the rest of the lives of the loans

themselves. The discussion here begins with the latter option,

then turns to the former cases involving the government’s pur-

chase of these loans.

Table 2.7 presents entries for the federal government’s

assumption of debt service on student loans owned by private

investors. In the ﬁrst entry, as in Table 2.5’s counterfactual,

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Loans(p) -Eq

Govt

(p) -Loans(p)

+Eq

(p)

+Tsy Acct(p,i) +Tsys(p,i) -RBs(p,i) -RBs(p,i) -D

(p) -D

(p)

+Tsy Acct(p,i) +Tsys(p,i)

-Tsy Acct(p,i) -Tsys(p) +RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Eq

(i)

-Eq

Govt

(i) -Tsy Acct(p,i) -Tsys(p)

-Loans(p) +Tsys(i) -Loans(p) +Tsys(i) +Eq

(i)

-Eq

Govt

(p,i) +Eq

(p)

Table 2.6 Department of Education Loans Cancelled as Principal Comes Due—Actual Transactions

Government

Cancels

Principal Now

Due

Government

Issues

Securities

Government

Retires Securities

and Pays Interest

Totals

Levy Economics Institute of Bard College 23

borrowers do not service their loans—this simply reverses all

transactions in Table 2.2. In the second entry, the government

makes the debt service payments, which debit its account at

the Fed and its equity by the principal and interest. This raises

reserve balances held by the Fed by the same amount as banks

then credit the deposits of the private investors. The loans are

assumed to still be liabilities of the borrowers (since the fed-

eral government did not purchase these loans), so the govern-

ment’s debt service payment reduces the principal of the loan

and raises the net worth of the borrowers by the same amount.

For this entry, the government’s net worth has fallen by the

value of the principal and interest; the borrowers’ net worth has

increased by the value of the principal payment; and the private

investors’ net worth has increased by the value of the interest

payments. In the third entry, the government issues securities

to cover the shortfall, just as in Tables 2.5 and 2.6. The sum or

net of these entries is an increase in household deposits and net

worth, both by the value of principal and interest; a decrease

in the government’s net worth of the same size, offset by an

increase in its liabilities; and the net for the private investors is

a purchase of the new Treasury securities (note that the private

investors holding the student loans are not assumed to be the

same investors purchasing the new Treasury securities). Table

2.7 also shows that, as a result of the cancellation, the borrowers

now have deposits and net worth they would not otherwise have

had, much like in Table 2.5.

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

(p,i) +D

(p,i) +Loans(p) -D

(p,i) -Eq

(i)

-D

(p,i) +Eq

(i) +Loans(p)

-Tsy Acct (p,i) -Eq

Govt

(p,i) +RBs(p,i) +RBs(p,i) +DPI(p,i) -Loans(p) +D

(p,i) +Eq

(i)

-Tsy Acct (p,i) +Eq

(p) -Loans(p)

+Tsy Acct (p,i) +Tsys (p,i) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i)

+Tsy Acct(p,i) +Tsys(p,i)

+Tsys (p,i) +D

(p,i) +D

(p,i) +Eq

(p,i) -D

(p,i)

-Eq

Govt

(p,i) -D

(p,i) +Tsys (p,i)

Table 2.7 Federal Government Assumes Debt Service on Privately Owned Student Loans—Relative to No Cancellation

REVERSE

Borrowers

Service

Private Loans

Government

Services Private

Loans

Totals

Government

Issues

Securities

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Tsy Acct (p,i) -Eq

Govt

(p,i) +RBs(p,i) +RBs(p,i) +D

(p,i) -Loans(p) +D

(p,i) +Eq

(i)

-Tsy Acct (p,i) +Eq

(p) -Loans(p)

+Tsy Acct(p,i) +Tsys (p,i) -RBs (p,i) -RBs (p,i) -D

(p,i) -D

(p,i)

+Tsy Acct (p,i) +Tsys (p,i)

+Tsys (p,i) -Loans(p) -Loans(p) +Eq

(i)

-Eq

Govt

(p,i) +Eq

(p) +Tsys(p,i)

Table 2.8 Federal Government Assumes Debt Service on Privately Owned Student Loans—Actual Transactions

Government

Services Private

Loans

Government

Issues

Securities

Totals

24 Student Debt Cancellation Report 2018

The actual transactions for the government’s assumption

of debt service payments for privately owned loans are in Table

2.8, which omits the reversal of borrowers’ debt service (from

Table 2.7). The totals row shows that the government has issued

liabilities as its net worth has fallen; student loan borrowers

have seen their loan principal fall and net worth rise in kind;

and private investors have had some loan principal paid off,

increased net worth by the amount of interest due in the period,

and purchased Treasury securities in the amount of principal

and interest (again, the investors holding the student loans are

not necessarily the same as those investing in Treasury securi-

ties). Note that the totals for the government are the same for

Tables 2.7 and 2.8. In other words, the counterfactual is not rel-

evant to the government’s position: the government’s liabilities

are increasing by the combined principal and interest payment,

both relative to no cancellation and in absolute terms. This is

also the core difference from the ED’s loans, which only raise

the government’s liabilities in absolute terms by the amount of

interest due annually.

Government-led debt cancellation where the government

simultaneously purchases and then cancels loans owned by

private investors

As noted above, an alternative to assuming the debt service to

private investors is to have the federal government purchase

these loans from the private investors at the outset. Table 2.9

presents the T-accounts for transactions involved in this sce-

nario. The ﬁrst transaction is the government issuing Treasury

securities that will replenish the Treasury’s account at the Fed

and enable the purchase of the private securities. The second

transaction is the purchase of the privately owned student

loans, which essentially reverses the entries from the ﬁrst trans-

action, except that private investors have effectively swapped

student loans for Treasury securities. Table 2.9 assumes, for

simplicity, that the government acquires the loans from private

investors at their book value with no interest accrued since the

most recent payments from the borrowers (alternative assump-

tions for either of these would likely not have macroeconomic

signiﬁcance and would not change the nature of the analysis

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Tsy Acct(p) +Tsys(p) -RBs(p) -RBs(p) -D

(p) -D

(p)

+Tsy Acct(p) +Tsys(p)

-Tsy Acct(p) +RBs(p) +RBs(p) +D

(p) +D

(p)

+Loans(p) -Tsy Acct(p) -Loans(p)

+Loans(p) +Tsys(p) +Tsys(p)

-Loans(p)

Table 2.9 Federal Government Purchases Privately Owned Student Loans

Government

Issues

Securities

Government

Purchases

Private Loans

Totals

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Loans(p) +Tsys(p) +Tsys(p)

-Loans(p)

-Loans(p) -Eq

Govt

(p) -Loans(p)

+Eq

(p)

+Tsys(p) -Loans(p) +Tsys(p)

-Eq

Govt

(p) +Eq

(p) -Loans(p)

Table 2.10 Federal Government Purchases and Cancels Privately Owned Student Loans

Totals

from

Table 2.9

Government

Cancels Private

Loans

Totals

Levy Economics Institute of Bard College 25

here). The totals or net changes for Table 2.9 are simply swaps

of Treasury securities for privately owned student loans among

the federal government and private investors.

Table 2.10 adds the government’s cancellation of the stu-

dent loans, which reduces the government’s equity while also

raising the equity of the student loan borrowers. An account-

ing difference between the government taking over the pay-

ments, as in the discussion above, and purchasing/cancelling

the loans here is that only the principal amounts show up in

the transactions. This is because the purchase and cancellation

occur essentially simultaneously, rather than over the remaining

life of the loans. However, this is simply a difference in timing

of the transactions, not in gross size of total transactions (of

macroeconomic signiﬁcance). Having sold the loans, investors

can reinvest the proceeds from the sale to earn the interest they

would have earned on the loans (of course, interest rates on

the new investments could be different from those previously

earned on the student loans). Similarly, and as in the scenario in

which the government takes over debt service payments to pri-

vate investors, borrowers that have had their loans cancelled will

effectively have the foregone principal and interest payments as

additional income at the time debt service would have been due,

and thus additional deposits and equity relative to no cancella-

tion. From the government’s perspective, it is simply incurring

the deﬁcits related to principal payments to private inves-

tors at the beginning rather than over the course of the loans’

maturities, and thus it is simply a difference between the present

value and future value of the loan cancellation. As with the ED’s

loans, the cancellation has eliminated its source of revenue to

service the Treasury securities issued in Table 2.9. The govern-

ment therefore will be issuing Treasury securities in the future

as it rolls over the portion of the outstanding liabilities that is

maturing and the liabilities effectively become permanent. As it

rolls these liabilities over, it will also increase its liabilities by an

amount equal to the debt service due on them.

Government-led debt cancellation where the government

purchases student loans issued by private investors and

cancels principal as payments come due

As with the ED’s loans, the process of rolling over principal from

the securities originally issued and then issuing new securities

to cover interest payments due is more straightforward in the

case in which the government purchases the privately owned

student loans and cancels principal as payments come due. In

Table 2.11, the ﬁrst entry is the reversal of the debt service pay-

ments, or the counterfactual. The second entry is the cancel-

lation of the principal due. The third entry is the issuance of

Treasury securities, since the cancelled borrowers’ debt service is

not forthcoming to service the Treasury securities issued when

the government purchased the privately issued loans in Table

2.9. The total or net effect relative to the no-cancellation sce-

nario is that borrowers have additional deposits reﬂecting the

Table 2.11 Federal Government Purchases and Cancels Privately Owned Student Loans as Payments Come Due—Relative to

No Cancellation

REVERSE

Borrowers

Service

Private Loans

Owned by

Government

Cancels

Principal Now

Due

Totals

Government

Issues

Securities

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Tsy Acct (p,i) -Eq

Govt

(i) +RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Loans(p)

+Loans(p) -Tsy Acct (p,i) +Eq

(i)

-Loans(p) -Eq

Govt

(p) -Loans(p)

+Eq

(p)

+Tsy Acct(p,i) +Tsys (p,i) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i)

+Tsy Acct(p,i) +Tsys(p,i)

+Tsys (p,i) +D

(p,i) +D

(p,i) +Eq

(p,i) -D

(p,i)

-Eq

Govt

(p,i) -D

(p,i) +Tsys (p,i)

26 Student Debt Cancellation Report 2018

increased net worth each period, private investors (which are

Treasury dealers, at the margin) have purchased Treasury secu-

rities, and the government’s reduced net worth and offsetting

additional liabilities are equal in amount to the foregone princi-

pal and interest for the period.

The actual, direct transactions are shown in Table 2.12. As

previously, the reversal of the student loan payment in Table

2.11 is not an actual transaction, even though the borrowers

do have more income due to reduced ﬁnancial obligations for

the period relative to no cancellation. In Table 2.12, the govern-

ment cancels principal currently due, then issues securities for

the combined principal and interest foregone in order to retire

and pay interest on securities sold when the privately held loans

were purchased. As with the ED’s loans, the securities matur-

ing are equal in size to the principal payment, so the net issu-

ance of securities is equal to the interest portion. However, by

contrast with the cancellation of the ED’s loans—which were

already funded with the issuance of Treasuries when the loans

were created—the securities issued by the government to pur-

chase privately owned loans, as illustrated in Table 2.9, are a new

budgetary cost, relative to current deﬁcit and debt levels.

Concluding remarks on government-led cancellation of

privately owned student loans

As with the ED’s loans, the difference between Tables 2.11 and

2.12 depends on whether one considers the cancellation in com-

parison to the no-cancellation scenario or in terms of actual

transactions. For the borrowers, the former is more useful, since

they will have additional income available that would otherwise

have been committed to the principal and interest payment.

For the government, the latter is more useful, since it purchased

the loans with the purpose of cancelling them. Therefore, it is

most appropriate to view the government’s purchase of pri-

vately owned student loans as raising its liabilities by the same

amount, while the foregone debt service after the principal is

cancelled (either at once or as it comes due) raises the govern-

ment’s liabilities annually by the amount of interest due on the

securities issued to purchase the loans.

Overall, aside from the timing of the government’s transac-

tions, there is no difference of macroeconomic signiﬁcance for

those involved if the government purchases and then cancels the

loans, as Tables 2.9, 2.10, 2.11, and 2.12 present, or takes over

the responsibility for repaying the loans, as assumed earlier in

Table 2.8. The only issue of much note is whether to cancel the

loans immediately after purchasing them or as the borrowers’

payments come due, but this is simply an issue of present value

of the loans compared to the future value. And in either case,

the government will essentially roll over the Treasury securities

issued to purchase the loans as the borrowers’ payments come

due, while adding to them new securities issued in the amount

of interest due annually.

Concluding remarks on government-led debt cancellation

The net ﬁnancial effects of student debt cancellation for student

loan borrowers would be (1) an increase in net ﬁnancial wealth,

and (2) an increase in after-tax income available to spend, save,

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Loans(p) -Eq

Govt

(p) -Loans(p)

+Eq

(p)

+Tsy Acct(p,i) +Tsys (p,i) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i)

+Tsy Acct(p,i) +Tsys(p,i)

-Tsy Acct(p,i) -Tsys(p) +RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Eq

(i)

-Eq

Govt

(i) -Tsy Acct(p,i) -Tsys(p)

-Loans(p) +Tsys(i) -Loans(p) +Tsys(i) +Eq

(i)

-Eq

Govt

(p,i) +Eq

(p)

Table 2.12 Federal Government Purchases and Cancels Privately Owned Student Loans as Payments Come Due—Actual

Transactions

Government

Cancels

Principal Now

Due

Government

Issues

Securities

Government

Retires Securities

and Pays Interest

Totals

Levy Economics Institute of Bard College 27

or borrow against rather than commit to debt service. Further,

even if all of the loans remained liabilities for borrowers and

were only cancelled in small steps as each payment came due, it

would be reasonable for households to behave as if these loans

had been cancelled all at once—since, regardless of the timing

of the cancellation, households will ﬁnd themselves with the

additional income that would have otherwise been used for

debt service. It is the same from the government’s perspective—

the timing of the effects on its budget position and rolling over

of liabilities is largely unaffected by the choice of whether to

cancel the loans outright or as payments come due. Finally, it

is also largely irrelevant for the government’s budget position

and liabilities whether it assumes debt service payments or pur-

chases the loans.

There is, however, a signiﬁcant difference with respect to

the impact on the government’s liabilities between the cancel-

lation of the ED’s loans and the loans owned by private inves-

tors. For the former, because the liabilities have already been

issued in the past to make these loans, in absolute terms cancel-

lation results in new liabilities only in the amount of interest

due on the securities issued earlier; for the latter, however, new

liabilities will be incurred in order to purchase the loans. The

point of presenting tables for both the counterfactuals and the

actual transactions is that there is a large difference, which has

substantial political signiﬁcance, between the actual increase

in the government’s liabilities that results from cancellation of

the ED’s loans and the increase relative to the no-cancellation

scenario (which scenario/table is more relevant depends on the

question being asked). Particularly in an environment of very

low interest rates, the interest due on $1.1 trillion in govern-

ment securities may be as low as $10 billion to $30 billion per

year (depending on maturity of securities issued and time of

issuance), which is obviously much different than incurring the

entire $1.1 trillion in cancelled loans (plus interest) as new gov-

ernment liabilities.

The Mechanics of Student Debt Cancellation Carried

Out by the Federal Reserve

This section will discuss debt cancellation carried out by the

Fed, rather than by the government, since some advocates have

argued in favor of this approach. The primary rationale for sug-

gesting the Fed carry out the cancellation appears to be a belief

that there may be a “free lunch” in the sense that the govern-

ment’s budget position does not need to be affected. As this

section will demonstrate, that is probably not the case. The dis-

cussion here will consider the same scenarios as those in which

the federal government carries out the debt cancellation. The

core takeaway from this section is that there is no macroeco-

nomically signiﬁcant difference for the federal government’s

budget position and liabilities issued (that is, the national debt)

if the Fed carries out the debt cancellation rather than the fed-

eral government.

The Federal Reserve purchases the Department of

Education’s loans

Since the ED loans are not marketable, an act of Congress would

appear necessary if it is not legally permissible for the ED to

sell the loans to the Fed. Even if it is permissible, congressional

action could be necessary if the Fed were unwilling to take such

action on its own (particularly given the ultimate purpose of

cancelling the loans). Once the Fed acquires the loans, it credits

the Treasury’s account at the Fed for the same amount. A rea-

sonable assumption—though not necessary in terms of macro-

economic signiﬁcance—is that the Fed would acquire the loans

at their current value. In this case, the transaction is a simple

transfer of loans and balances between the federal government

and its central bank. This is shown in Table 2.13. As the current

value of student loans held by the ED is near $1 trillion, the

Treasury now has this much added to its account. The federal

government is therefore essentially $1 trillion further under the

national debt ceiling than it was prior to the Fed’s acquisition of

the student loans.

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Tsy Acct (p) +Loans (p)

+Tsy Acct (p)

-Loans

(p)

Table 2.13 Federal Reserve Acquires Loans Owned by Department of Education

Fed Buys Loans

from

28 Student Debt Cancellation Report 2018

Some fundamentals of the Federal Reserve’s remittances and

their relevance to student loan cancellation

To understand the mechanics of the Fed carrying out student

loan cancellation, it is useful to consider ﬁrst what happens if

the Fed owns loans but instead does not cancel them and simply

collects payments as they become due. This is shown in Table

2.14. As borrowers pay debt service, their debt and deposits fall,

but so does their net worth, due to the additional interest cost of

the debt service. The borrowers’ banks simply debit the deposit

accounts of the borrowers while their own reserve accounts are

debited in kind. The loans owned by the Fed are debited by the

amount of principal paid down. Due to the interest portion of

the debt service payment, the quantity of reserve balances deb-

ited is greater than the reduction in the student loan principal,

thereby raising the Fed’s equity.

The portion of the Fed’s equity that has increased is called

“surplus capital,” to distinguish it from the “paid-in capital”

contributed by member banks. The Fed’s additions to surplus

capital—proﬁts after costs and dividend payments to member

banks on their paid-in capital—were capped by law at $10 bil-

lion by the 2015 Fixing America’s Surface Transportation Act

(FAST Act), which amended the Federal Reserve Act to require

that any capital surplus greater than this amount be transferred

to the Treasury.

The Fed’s “surplus capital” account is much

like a corporation’s “retained earnings” account on its balance

sheet; therefore, the Fed is essentially limited to cumulative

retained earnings of $10 billion. In January 2016, for instance,

the Fed transferred $98 billion of what would otherwise have

been added to the capital surplus, or “retained earnings,” to the

Treasury’s account, as well as another $19 billion that had been

accumulated beyond the $10 billion statutory limit prior to the

imposition of the new cap.

The transfer from the Fed’s surplus capital to the Treasury’s

account is shown in Table 2.15. The ﬁrst entry repeats Table

2.14. In the second entry, the Treasury’s account is credited for

the interest payments as a result of the Fed’s transfer of its addi-

tional capital surplus to the Treasury’s account. The total or net

effect is that the proﬁts from the loan are now added to the gov-

ernment’s budget position, while the Fed is returned to a posi-

tion in which its equity is unchanged.

The Federal Reserve cancels the Department of Education’s

loans all at once

Turning to the cancellation, now assume all of the ED loans

are cancelled. In Table 2.16, the ﬁrst entry is the Fed’s purchase

of the loans from the ED. The second entry is the cancellation.

This is not the end though, because the reduced Fed equity

position will be transferred to the Treasury over time as the Fed

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Loans(p) -RBs(p,i) -RBs (p,i) -D

(p,i) -D

(p,i) -Loans(p)

+Eq

Fed

(i) -Eq

(i)

Table 2.14 Borrowers Service Student Loans Owned by Federal Reserve

Borrowers

Service Loans

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Loans(p) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i) -Loans(p)

+Eq

Fed

(i) -Eq

(i)

+Tsy Acct(i) +Eq

Govt

(i) +Tsy Acct(i)

-Eq

Fed

(i)

+Tsy Acct(i) +Eq

Govt

(i) -Loans(p) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i) -Loans(p)

+Tsy Acct(i) -Eq

(i)

Table 2.15 Transfer from Federal Reserve’s Surplus Capital Account to Treasury’s Account

Borrowers

Service Loans

Fed

Remittances to

Government

Totals

Levy Economics Institute of Bard College 29

waits until it accumulates enough proﬁts—which would likely

require several years, if not a decade or more—for its surplus

capital to return to $10 billion. Thereafter, the Fed’s proﬁts

send its surplus capital above $10 billion, at which time the Fed

ﬁnally has proﬁts to remit to the Treasury as in Table 2.16.

The core point here is that there is no budgetary “free

lunch” available to the federal government if the Fed carries out

the loan cancellation. Whether the Fed or the government can-

cels the ED’s loans, the loss will be incurred by the government.

A minor difference would be that the losses would follow the

pattern of the Fed’s proﬁts. In the case of the government car-

rying out the cancellation, it is the path of principal and interest

payments on the Treasuries issued (in order for the ED to lend

to student borrowers) that determines the annual effects on the

government’s budget position and the securities that are then

issued to roll over the principal and interest payments.

The Federal Reserve cancels debt service payments for the

Department of Education’s loans

As with the government carrying out the cancellation, interac-

tions between the Fed and the Treasury’s account are easier to

present in the case where student debt is cancelled as payments

come due. Table 2.17 presents this scenario relative to no can-

cellation. In the ﬁrst row of Table 2.17, households’ deposits

increase by the principal and interest, while their loan liability

and equity increase by principal and interest due. For the Fed,

this raises the loan principal and reserve balances but reduces

surplus capital by the accrued interest. When borrowers’ loan

principal coming due is cancelled in the second entry, their

equity or net worth rises in kind. For the Fed, loan principal

falls and reduces equity in kind. The totals show that household

deposits and equity have risen by the amount of the principal

and interest due, the Fed’s equity has fallen by the same amount,

and bank reserve balances have risen.

The fall in the Fed’s equity, booked as a reduction in sur-

plus capital, then reduces the Fed’s remittances to the federal

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Tsy Acct (p) +Loans (p) +Tsy Acct (p)

-Loans (p)

-Loans(p) -Eq

Fed

(p) -Loans(p)

+Eq

(p)

Table 2.16 Federal Reserve Cancels Loans Purchased from Department of Education

Fed Buys

Loans from ED

Fed Cancels

Loans

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Loans(p) +RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Loans(p)

-Eq

Fed

(i) +Eq

(i)

-Loans (p) -Eq

Fed

(p) -Loans(p)

+Eq

(p)

+RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Eq

(p,i)

-Eq

Fed

(p,i)

Table 2.17 Federal Reserve Cancels Debt Service Payments for Department of Education Loans—Relative to No Cancellation

REVERSE

Borrowers

Service Loans

Fed Cancels

Loan Principal

Now Due

Totals

30 Student Debt Cancellation Report 2018

government by the same amount. Therefore, the cancellation

of the student loan payments raises the government’s budget

deﬁcit by the same amount relative to no cancellation, which is

shown in Table 2.18. The ﬁrst entry is the totals row from Table

2.17. The second is the reduction in remittances. The totals for

Table 2.18 show that the Treasury’s account and equity—and

thus the federal government’s budget position—have been

reduced by the foregone principal and interest.

Whether the fall in the Treasury’s account in Table 2.18

results in the Treasury issuing new securities depends on what

the government does with credits to its account from Table 2.16.

While the additional balances might appear to enable the gov-

ernment to “print money” rather than spend via tax revenues or

issuing Treasuries, the balances credited in Table 2.16 are instead

already “spoken for,” since they will offset reduced remittances

from the Fed cancelling student debt service payments.

Nonetheless, the government can choose to use the bal-

ances credited to its account in Table 2.16 to either offset

subsequent remittance reductions or retire previously issued

securities. Again, one might consider the latter choice equiva-

lent to “monetizing the debt,” but the budgetary consequences

of the two choices are essentially the same. Table 2.19 shows the

government retiring Treasury securities after the Fed purchases

the ED’s loans. The ﬁrst entry is the same as in Table 2.16. In the

second entry, the retirement of securities raises the Fed’s reserve

balances in kind. The totals show that while the government has

used the loan proceeds to retire liabilities incurred when the ED

made the loans, the reserve balances remain.

As explained in Appendix C in the digression on the Fed’s

daily operations in the federal funds market, for the Fed to

achieve its interest rate target in the presence of such a large

increase in the supply of reserve balances requires it to pay inter-

est on the reserve balances (IOR) at its target rate to banks (see

Figure C.3 in Appendix C). But IOR payments to banks reduce

Fed proﬁts and then reduce Fed remittances, leaving the govern-

ment bearing the cost of IOR at roughly the same magnitude as

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Tsy Acct(p) +Loans(p) +Tsy Acct(p)

-Loans(p)

-Tsy Acct(p) -Tsys(p) +RBs(p) +RBs(p) +D

(p) +D

(p)

-Tsy Acct(p) -Tsys(p)

-Loans(p) -Tsys(p) +Loans(p) +RBs(p) +RBs(p) +D

(p) +D

(p)

-Tsys(p)

Table 2.19 Federal Government Retires Securities Using Funds from Sale of Loans to Fed

Fed Buys

Loans

from ED

Treasury

Retires

Securities

Totals

Table 2.18 Federal Reserve Reduces Remittances to Treasury—Relative to No Cancellation

Totals

from

Table 2.17

Fed

Reduces

Remittances

Totals

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Eq

(p,i)

-Eq

Fed

(p,i)

-Tsy Acct(p,i) -Eq

Govt

(p,i) -Tsy Acct(p,i)

+Eq

Fed

(p,i)

-Tsy Acct(p,i) -Eq

Govt

(p,i) +RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Eq

(p,i)

-Tsy Acct(p,i)

Levy Economics Institute of Bard College 31

it would bear the cost of interest on the securities if it did not

retire them. In other words, whether or not the government uses

the credits in Table 2.19 to retire securities, its budget position

will be impacted as if it did not retire them.

Overall, as the Fed passes the cost of cancelled student debt

payments to the government via reduced remittances, the gov-

ernment will either draw down its account at the Fed, as in Table

2.18, or issue new securities that effectively replace those retired

in Table 2.19. The net impact on the government’s budget posi-

tion relative to no cancellation is thus unaltered by whether the

Fed or the government cancels the ED’s loans.

Table 2.20 shows the actual transactions for the Fed cancel-

ling student debt service payments. The ﬁrst two transactions

are the second entries for Tables 2.17 and 2.18, respectively. For

simplicity, Table 2.20 assumes that the government keeps the

balances credited to its account in Table 2.16 instead of retiring

securities. In the third and fourth transactions, therefore, the

government rolls over the securities issued when the ED made

its loans, resulting in a net issuance of new securities that covers

interest due. In the totals, credits to the Treasury’s account in

Table 2.16 fall by the cancelled principal payments. This repeats

each period until all principal payments are cancelled and all

balances credited in Table 2.16 are debited.

Instead of issuing new securities to make the interest pay-

ments, in Table 2.20 the government of course could debit the

balances credited in Table 2.16. But this would add reserve

balances, resulting in IOR payments to banks ultimately borne

by the government via reduced remittances.

Overall, in Table 2.20 there is again no ﬁnancial beneﬁt to the

government if the Fed purchases the ED’s loans and cancels them.

The Federal Reserve assumes debt service payments for loans

owned by private investors

If the Fed assumes debt service payments on loans owned by

private investors, it will be making direct principal and inter-

est payments to the investors on behalf of a third party. This

will directly reduce the Fed’s proﬁts and thus its equity. As in

the case of the ED’s loans, it is expected that an act of Congress

would be required for the Fed to take over the debt service pay-

ments on privately owned loans or purchase the loans and then

cancel them.

The transactions are shown relative to no cancellation for

this scenario in Table 2.21. The ﬁrst entry is the counterfac-

tual in which borrowers do not service the loans. In the second

entry, the Fed services the loans by creating reserve balances that

then result in the private investors’ banks crediting their deposit

accounts. In the third entry, the Fed reduces remittances by the

amount of the fall in its equity for the period, which reduces the

government’s budget position by the same amount. The fourth

entry is the government issuing securities in response to the

budget shortfall. The totals here are the same as when the govern-

ment assumed these debt service payments in Table 2.7; while

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Loans (p) -Eq

Fed

(p) -Loans(p)

+Eq

(p)

-Tsy Acct(p) -Eq

Govt

(p) -Tsy Acct(p)

+Eq

Fed

(p)

+Tsy Acct(p,i) +Tsys(p,i) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i)

+Tsy Acct(p,i) +Tsys(p,i)

-Tsy Acct(p,i) -Tsys(p) +RBs(p,i) +RBs(p,i) +D

(p,i) +D

(p,i) +Eq

(i)

-Eq

Govt

(i) -Tsy Acct(p,i) -Tsys(p)

-Tsy Acct(p) +Tsys(i) -Loans(p) -Tsy Acct(p) -Loans(p) +Tsys(i) +Eq

(i)

-Eq

Govt

(p,i) +Eq

(p)

Table 2.20 Federal Reserve Cancels Debt Service Payments for Department of Education Loans—Actual Transactions

Fed Cancels

Loan Principal

Now Due

Fed Reduces

Remittances

Government

Issues

Securities

Govt Retires

Securities and

Pays Interest

Totals

32 Student Debt Cancellation Report 2018

the Fed is making the payments, the ultimate effects impact the

government’s budget position and its liabilities, not the Fed’s.

The actual transactions are in Table 2.22. As in Table 2.8

where the government assumed the payments, the totals for the

government are the same in Tables 2.21 and 2.22. Table 2.22

simply shows that the private investors will receive their pay-

ments, while the households will have their principal cancelled.

Table 2.21, by contrast, shows that, relative to no cancellation,

there is no difference for the private investors (they receive their

payments regardless), while the households have both principal

and interest available as additional disposable income.

The Federal Reserve purchases and cancels loans owned by

private investors

Table 2.23 shows the T-accounts for the Fed’s purchase and

cancellation of the privately owned student loans. Here again,

as a simplifying assumption, the Fed is purchasing the student

loans at the current value, thereby not affecting the equity of the

private investors.

Because the cancellation reduces the Fed’s

equity by the amount of the loans, the Fed will reduce its annual

remittances until they sum to this amount. This will reduce the

government’s budget position annually, requiring the govern-

ment to issue and pay interest on new securities of the same

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

(p,i) +D

(p,i) +Loans(p) -D

(p,i) -Eq

(i)

-D

(p,i) +Eq

(i) +Loans(p)

+RBs(p,i) +RBs(p,i) +D

(p,i) -Loans(p) +D

(p,i) +Eq

(i)

-Eq

Fed

(p,i) +Eq

(p) -Loans(p)

-Tsy Acct(p,i) -Eq

Govt

(p,i) -Tsy Acct(p,i)

+Eq

Fed

(p,i)

+Tsy Acct(p,i) +Tsys(p,i) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i)

+Tsy Acct(p,i) +Tsys(p,i)

+Tsys(p,i) +D

(p,i) +D

(p,i) +Eq

(p,i) -D

(p,i)

-Eq

Govt

(p,i) -D

(p,i) +Tsys(p,i)

Table 2.21 Federal Reserve Assumes Debt Service on Privately Owned Student Loans—Relative to No Cancellation

REVERSE

Borrowers

Service Private

Loans

Fed Services

Private Loans

Fed Reduces

Remittances

Government

Issues

Securities

Totals

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+RBs(p,i) +RBs(p,i) +D

(p,i) -Loans(p) +D

(p,i) +Eq

(i)

-Eq

Fed

(p,i) +Eq

(p) -Loans(p)

-Tsy Acct(p,i) -Eq

Govt

(p,i) -Tsy Acct(p,i)

+Eq

Fed

(p,i)

+Tsy Acct(p,i) +Tsys(p,i) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i)

+Tsy Acct(p,i) +Tsys(p,i)

+Tsys(p,i) -Loans(p) -Loans(p) +Eq

(i)

-Eq

Govt

(p,i) +Eq

(p) +Tsys(p,i)

Table 2.22 Federal Reserve Assumes Debt Service on Privately Owned Student Loans—Actual Transactions

Fed Services

Private Loans

Fed Reduces

Remittances

Government

Issues

Securities

Totals

Levy Economics Institute of Bard College 33

annual amount to cover the reduced remittances. As with the

Fed cancelling the ED’s loans all at once, the timing of the bud-

getary impacts will match the remittance shortfall rather than

the path of debt service payments foregone.

Because the reserve balances the Fed used to purchase the

private loans are left circulating, the Fed will have to pay IOR

on them at its target rate. These IOR costs of the Fed will be

borne by the government as the Fed’s remittances fall. Then,

while the Fed’s remittances fall until the total value of the loans

has been accounted for, the government will be issuing addi-

tional Treasury securities that will, over time, effectively replace

the reserve balances. Thus, whether the Fed purchases the loans

with new reserve balances or the government does it by issuing

new securities, the government will bear the interest costs of

ﬁnancing the purchase.

The Federal Reserve purchases loans owned by private

investors and cancels debt service payments

As with the government carrying out the cancellation, the

results discussed for the Fed purchasing and then cancelling the

privately owned loans all at once are more easily seen when it is

assumed that debt service payments are cancelled as they come

due. Nonetheless, this is a more complex case to illustrate than

the case of the government cancelling these loans. Therefore,

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

+Loans(p) +RBs(p) +RBs(p) +D

(p) +D

(p)

-Loans(p)

-Loans(p) -Eq

Fed

(p) -Loans(p)

+Eq

(p)

+RBs(p) +RBs(p) +D

(p) -Loans(p) +D

(p)

-Eq

Fed

(p) +Eq

(p) -Loans(p)

Table 2.23 Federal Reserve Purchases and Cancels Privately Owned Student Loans

Fed Purchases

Private Loans

Fed Cancels

Private Loans

Totals

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Loans(p) -Eq

Fed

(p) -Loans(p)

+Eq

(p)

+RBs(i) +RBs(i) +Eq

Banks

(i)

-Eq

Fed

(i)

-Tsy Acct(p,i) -Eq

Govt

(p,i) -Tsy Acct(p,i)

+Eq

Fed

(p,i)

+Tsy Acct(p,i) +Tsys (p,i) -RBs(p,i) -RBs(p,i) -D

(p,i) -D

(p,i)

+Tsy Acct(p,i) +Tsys(p,i)

+Tsys(p,i) -Loans(p) -RBs(p) -RBs(p) -D

(p,i) -Loans(p) -D

(p,i)

-Eq

Govt

(p,i) +Eq

Banks

(i) +Eq

(p) +Tsys(p,i)

Table 2.24 Federal Reserve Purchases and Cancels Privately Owned Student Loans as Payments Come Due—Actual

Transactions

Fed Cancels

Principal Now

Due

Fed Pays

IOR

Fed Reduces

Remittances

Government

Issues

Securities

Totals

34 Student Debt Cancellation Report 2018

only the T-accounts for the actual transactions are shown—the

counterfactual or “relative to no cancellation” scenario is omit-

ted. Table 2.24 shows the actual transactions for this scenario.

The ﬁrst transaction is the Fed cancelling principal now due.

The second transaction is the Fed paying interest on reserves

on the reserve balances created when it purchased the privately

owned loans in Table 2.23. Together, the cancelled principal

and IOR payment (which raises banks’ equity) reduce the Fed’s

remittances by the combined amount in the third transaction.

This reduces the government’s budget position by the same

amount and leads the government to issue new securities in

kind. From the totals, the government has issued new securities

in the amount of principal plus interest. This amount is equal

to the Fed’s IOR payment from its purchase of the loans and the

reduction in the Fed’s reserve balances; in other words, the Fed’s

liabilities created by the purchase of the loans—the reserve bal-

ances—are reduced by the amount of the principal due, while

the government’s liabilities increase to offset.

Over the life of the loans, the Fed’s reserve balances created

by the purchase will fall each year, eventually to zero, while the

Treasury’s liabilities will rise eventually to the amount of the

Fed’s original purchase plus all IOR payments made through-

out. Thus, unlike Table 2.12, in which the government issued

securities up front to purchase the loans and thereafter new

government liabilities were only equal to the annual debt

service on these securities, here the government’s liabilities

increase each period by the amount of principal and interest

due. Nevertheless, there is once again ultimately no “free lunch”

from having the Fed carry out the cancellation, as the end result

is that the full amount of the original purchase plus interest is

eventually added to the government’s liabilities.

Potential options to avoid costs to the federal government of

student loan cancellation carried out by the Federal Reserve

The keys to understanding the argument to this point—that

there is no “free lunch” from having the Fed carry out the can-

cellation rather than the federal government—are as follows:

(1) the Fed’s surplus capital above $10 billion each year is remit-

ted to the Treasury, and (2) the Fed necessarily provides interest

on reserve balances (or some alternative to it, like reverse repur-

chase agreements) at its target rate in the event that the quantity

of reserve balances supplied is beyond the quantity demanded

at the target rate. Nevertheless, this section explores two options

(both highly theoretical) that could potentially enable the Fed

to carry out a student loan cancellation while also obtaining

a “free lunch”—that is, where the costs would not ultimately

affect the federal government’s budget.

As suggested previ-

ously, it is highly unlikely that the Fed would desire to carry out

student loan cancellation. Moreover, it is questionable whether

the Fed has the legal authority to do so—and even if it techni-

cally does have such authority, exercising it may result in legal

and political repercussions for the Fed’s prized independence.

Consequently, it is presumed that an act of Congress and presi-

dential signature would be required to allow (or force, as the

case may be) the Fed to engage in any such actions (Carrillo et

al. forthcoming).

The ﬁrst option recognizes that the Fed funds itself mostly

via interest on its portfolio of securities and loans. As several

rounds of quantitative easing since the ﬁnancial crisis have

shown, the Fed’s proﬁts can rise substantially as its balance sheet

expands. In theory at least, as the Fed cancels the student loans

that it owns, it could begin purchasing other ﬁnancial assets,

and the earnings on the latter could offset the losses, thereby

not affecting remittances to the Treasury. Again, this is all quite

theoretical. The Fed’s losses from student loan cancellation

could rival or even signiﬁcantly exceed the proﬁts it currently

earns on a balance sheet of around $3 trillion in a given year.

Any proposal that the Fed purchase perhaps another $3 trillion

in ﬁnancial assets—or a lot more than this—would quickly run

into many difﬁculties, such as (most obviously) the potential

effect on interest rates and asset prices.

A second option would be to expand on the Fed’s existing

accounting practices regarding operating losses. Currently, if

the Fed experiences an operating loss in any week, it capitalizes

the losses as a “deferred asset” that is a negative liability rather

than reducing surplus capital; remittances to the Treasury are

not paid until an operating proﬁt is run large enough to offset

any previous operating losses. If the Fed were instead able to

isolate losses from cancelling student loans from the rest of its

operating proﬁts—perhaps by deferring them permanently—it

could in theory result in the losses not negatively affecting remit-

tances to the Treasury (Carrillo et al. forthcoming). In this case,

the Fed would create a separate, deferred account on its asset

side, rather than its liability side. In terms of balance sheets, the

T-account transactions would look like those in Table 2.25. An

alternative could be for the Fed to not write down the loans and

its equity but instead capitalize the losses or write down both

as an asset and as a “deferred asset” on the liability side (similar

to loan-loss provisioning by banks), which Table 2.26 presents

Levy Economics Institute of Bard College 35

(“deferred loan loss” is used here rather than “deferred asset” to

avoid potential confusion from crediting an asset as a liability).

In either case (from Table 2.25 or 2.26), under the appro-

priate legal circumstances the Fed’s operating proﬁts—and thus

its surplus capital and subsequent remittances to the Treasury—

could be shielded from being affected by the losses it would oth-

erwise incur from carrying out student loan cancellation.

In conclusion, while both options discussed here are not

“usual” approaches to managing the Fed’s operating proﬁts,

they are potential avenues for obtaining a “free lunch” in car-

rying out student loan cancellation. Obviously either option

would be quite controversial, and would likely be opposed by

the Fed as stridently as anyone else. In the absence of either of

these two options being instituted, there is no “free lunch” in

having the Fed carry out the debt cancellation, and it is erro-

neous to consider such actions by the Fed as analogous to its

earlier quantitative easing operations.

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

-Loans No Change

-Loans

Fed

+Capitalized +Eq

Loan

Writedow n

Table 2.25 Federal Reserve Capitalizes Losses from Cancelling Student Loan Debt Service

Fed Cancels

Loan

Payments

Federal Government Federal Reserve Banks Student Loan Borrowers Private Investors

A L/E A L/E A L/E A L/E A L/E

No Change No Change

-Loans

Loan Values Eq

Fed

+Capitalized +Deferred +Eq

Loan

Losses Losses

Table 2.26 Federal Reserve Capitalizes Losses from Cancelling Student Loan Debt Service without Writing Down Loans

Fed Cancels

Loan

Payments

36 Student Debt Cancellation Report 2018

Section 3: Simulating Student Debt

Cancellation

We simulated the student debt cancellation using two macro-

econometric models in order to examine the implications of

student debt cancellation and incorporate feedback effects that

go beyond ﬁnancial statement analysis. This section presents the

results from simulations of a debt cancellation carried out within

the Moody’s model and the US version of the Fair model.

Feedback effects necessarily omitted from ﬁnancial statement

analysis of the mechanics of student debt cancellation are explic-

itly accounted for in these simulations. At the same time, it is very

important to understand that many potential effects are out-

side the scope of macroeconometric models. For instance, nei-

ther model explicitly integrates different income or wealth tiers

within the private sector, business startup decisions, and so forth.

To the degree that beneﬁts beyond the scope of these simulations

are present (discussed later in this section), the results presented

here represent a subset of the beneﬁts (or costs, as the case might

be) of student debt cancellation. Both models are well regarded

among economists. The Moody’s model is well-known among

professional macroeconomic forecasters and is frequently refer-

enced in the press, particularly in the context of policy debates.

For its part, simulations of and econometric investigations into

the Fair model have been the subject of several dozen academic

publications over the past 40 years or more.

Models and Assumptions Used for Simulating

Student Debt Cancellation

Introduction to the Moody’s model

The Moody’s model is the macroeconomic forecasting model

used by Moody’s and Economy.com. It has around 1,800 vari-

ables and “can best be described as ‘Keynesian in the short run,

and classical in the long run’” (Zandi and Hoyt 2015). The

model is “structural,” taking a middle ground between pure

time-series-based models and those founded on strict micro-

economic theoretical foundations, which means it attempts to

balance theoretical assumptions with econometric evidence

(Zandi and Hoyt 2015). The methodological documentation of

Moody’s model by Zandi and Hoyt explains that,

By taking a middle ground between theory and data,

this approach attains neither the theoretical elegance of

the DSGE [Dynamic Stochastic General Equilibrium]

approach or the empirical ﬂexibility of a VAR [Vector

Autoregression]. At the same time, however, it manages

to avoid the shortcomings of either one; imposing theory

to restrict the ﬂexibility of econometric speciﬁcations

allows more efﬁcient estimation and greater explanatory

power than a VAR can achieve. However, structural mac-

roeconomic models do not require some of the extreme

and somewhat unrealistic assumptions that render

DSGEs susceptible to misspeciﬁcation. (2015, 2)

A structural model is built via numerous aggregative equa-

tions that have a grounding in macroeconomic theory; these

“stochastic” equations are then estimated econometrically

either individually or simultaneously. The econometric regres-

sions determine key variables that are then inserted into identity

equations (such as those from National Income and Product

Accounts or Flow of Funds) in order to determine the rest of the

model’s endogenous variables through an iterative method that

solves all of the identities simultaneously. As Zandi and Hoyt

(2015, 2) put it, the strength of structural models “is the great

detail they can provide.”

Though VARs and DSGEs can incorporate no more

than a few variables of interest such as aggregate GDP,

a benchmark bond yield, and CPI inﬂation, structural

macroeconomic models are able to specify and gener-

ate forecasts for a rich array of macroeconomic data,

detailing the composition of both spending and indus-

trial activity, the entire maturity yield curve and many

other interest rates, and prices for goods, services, and

assets throughout the economy.

The paper by Zandi and Hoyt provides the speciﬁcation

and estimation results for several of the model’s stochastic

equations, including consumer services spending, vehicle sales,

investment in industrial equipment, private inventories, goods

exports, vehicle and parts imports, hourly compensation, the

federal funds rate, rates on the 10-year Treasury note, corporate

proﬁts, aggregate hours worked, industry employment, labor

force, wages and salaries, single-family permits, and a housing

price index. Several of these are discussed below within the con-

text of the simulations of student debt cancellation.

Levy Economics Institute of Bard College 37

Introduction to the Fair model

The Fair model is a large-scale, structural, macroeconometric

model designed by Ray Fair at Yale University. Created in the

1970s, the model’s structure has changed little across more than

40 years of business cycles and macroeconomic events. The “US

model” version of the Fair model has about 225 variables within

25 stochastic equations and 100 identity equations that com-

pletely integrate National Income and Product Accounts and

Flow of Funds data. The simulations presented in this report

are via the US model because it is updated quarterly; Fair’s

“Multicountry” version of the model, which includes the US

model’s 125 equations but then adds roughly 1,550 additional

equations for around 50 other countries, has not been updated

since 2013. Several books, published papers, and a website are

the primary sources for the Fair model.

The approach to build-

ing the model’s structural equations is derived from the Cowles

Commission, which, like the design of the Moody’s model,

uses macroeconomic theory to guide the design of stochastic

equations that are then estimated (in the Fair model’s case, via

two-stage least squares techniques) and repeatedly tested for

statistical signiﬁcance, misspeciﬁcation, the structure of error

terms, stability, predictive ability, and so forth.

The entire

models are then rigorously tested for consistency with empiri-

cal evidence and predictive ability (Fair 2013, 2004, 1992). Fair

presents updated estimates for structural equation coefﬁcients

and single equation tests each quarter in an “Appendix” posted

to his website. His published research argues that the Fair model

dominates VAR, New Keynesian, and Real Business Cycle models

in predicting real GDP both four and eight quarters ahead (Fair

2007, 2004). Overall, like Zandi and Hoyt, Fair ﬁnds the DSGE

models to be less useful than the larger structural econometric

models for real-world policy analysis.

Whereas the Moody’s model has a self-described “Classical

core”—or long-run assumptions based upon estimated supply-

side fundamentals—to complement a more “Keynesian” short

run, the Fair model is more traditionally Keynesian through-

out. In contrast to the Moody’s model, Fair’s econometric stud-

ies have rejected the so-called nonaccelerating inﬂation rate of

unemployment (NAIRU) dynamics. To be more precise, Fair

ﬁnds that the more accurate relationship between inﬂation

and unemployment is nonlinear, whereby inﬂation is not very

responsive to unemployment rates across a wide range, but

could become much steeper at very low unemployment rates,

perhaps around 2 percent (that is, at very low unemployment

rates, inﬂation could rise substantially). To be clear, though, with

so few real-world observations, Fair’s econometric studies were

not able to entirely conﬁrm or reject the likelihood of a steep rise

in inﬂation at very low unemployment rates (Fair 2013, 147–60).

Both models have been used to project events in the mac-

roeconomy. The economic proposals for both 2016 presidential

candidates were simulated in the Moody’s model. The model is

updated monthly for incoming macroeconomic data in order

to provide new macroeconomic forecasts and other simula-

tions presented in various Moody’s publications (Zandi et al.

2016; Zandi, Lafakis, and Ozimek 2016). Fair’s books from 2004

and 2013 provide summaries of numerous earlier published

papers (often in leading macroeconomic journals) forecasting

the effects of the Obama stimulus (the 2009 American Recovery

and Reinvestment Act); effects of monetary and ﬁscal policies;

econometric analyses of unemployment, inﬂation, and produc-

tion; ﬁnancial crises; and so forth. Finally, Fair predicted the

1990s stock market bubble after structural stability tests of the

Fair model’s stochastic equations showed that only the variable

associated with capital gains in equities had evidence of a struc-

tural break.

Assumptions for the simulated student debt cancellation

The version of student debt cancellation simulated in the mod-

els is based upon the mechanics provided in the previous sec-

tions: the federal government cancels student loans owned by

the Department of Education (ED) and takes over payments

owed on privately owned student loans. Assumptions for the

simulations are the following:

1) The student debt cancellation occurs at the beginning of

2017. For the ﬁrst quarter of 2017, we assume debt in each

of the three categories of student loans to be at the follow-

ing levels:

• Debt owned by the ED = $1,024 billion.

• Debt owned by private investors and government-

guaranteed = $277 billion.

• Debt owned by private investors and not

government-guaranteed = $105 billion.

2) From the perspective of those borrowers servicing student

loans prior to the debt cancellation, the effect of cancelling

the ED-owned loans and taking over payments of loans

owned by private investors is to essentially cancel all of their

student loan debt at the beginning of 2017. In other words,

they will act as if their student loans are cancelled. The sum

of the three categories of student loans is $1.406 trillion;

38 Student Debt Cancellation Report 2018

borrowers currently servicing student loans will feel as if

their net wealth has increased by this amount. Further, the

income that households would have devoted to servicing

their loans is now available for households to spend, save,

or borrow against. These two effects—the rise in net ﬁnan-

cial wealth from the debt cancellation and additional dis-

posable income previously devoted to debt service—are the

avenues through which the debt cancellation stimulates the

macroeconomy in the simulations.

3) From the perspective of the private investors owning stu-

dent loans, there is no change, as they continue to receive

the payments they would receive in the absence of the can-

cellation. The difference is that the borrowers are now in

possession of greater discretionary income, since they are

no longer the ones making these payments.

4) For simplicity of comparison, in the absence of the cancel-

lation, all student loans owned by the ED and all privately

issued, not government-guaranteed loans are assumed to

have otherwise been paid down in equal principal install-

ments over 10 years, in accordance with the ED’s standard

repayment plan. All privately issued loans that are guar-

anteed by the government are assumed to have otherwise

been paid off by the end of 2020, also in equal principal

installments. This scenario also assumes a 10-year repay-

ment plan for privately issued, government-guaranteed

student loans, since the federally guaranteed student loan

program ended in 2010.

5) Interest rates on the three types of student loans are

assumed to be the following:

• Debt owned by the ED = 4.6 percent (consistent

with current law).

• Debt owned by private investors and government-

guaranteed = the previous quarter’s short-term

interest rate (determined within the models’

simulations) plus 2.3 percent (consistent with

current law).

• Debt owned by private investors and not government-

guaranteed = 10 percent (as a proxy average for

interest rates on current outstanding loans).

Baseline values and macroeconometric simulation

Simulations require a baseline level for comparison and evalua-

tion of the simulated changes. That is, the simulation of a policy

change itself is only meaningful relative to a model’s baseline,

since the latter represents the model’s simulated values without

the policy in place. This is particularly the case for simulations

involving model forecasts for more than a few years, as struc-

tural models thereafter converge to a “trend” or longer-run

path that may have little to offer in terms of a useful forecast

without careful adjustments to many exogenous variables. For

the Fair model, the baseline is the forecast for the model from

2017 to 2026. Fair currently provides forecasts on his website

through 2022 only. To simulate 2023 to 2026, we extrapolated

Fair’s assumptions for changes in certain exogenous variables—

all of which are reported on his website—forward to 2026.

Additionally, Fair’s empirical research ﬁnds macroeconomically

signiﬁcant age-related effects, such as the ratio of working-age

to non-working-age people. To incorporate these, US Census

population projections were used to design the variables Fair

incorporates into stochastic equations. The Fair model simula-

tions were carried out using the Fair-Parke program, available

on Fair’s website. For Moody’s model, the baseline is deﬁned as

the model’s projection for 2017–26.

We demonstrate the relevance of one additional policy

assumption in the simulations by presenting results under two

scenarios for each model. First, we simulate the models with the

Fed’s interest rate reaction function included—that is, the Fed’s

interest rate target adjusts to “lean against” macroeconomic

changes associated with the cancellation. Next, we simulate an

alternative scenario in which the Fed’s interest rate reaction

function is “turned off.” In other words, the behavior of the

short-term interest rate in these latter simulations will be iden-

tical to that in the respective baseline simulations. The rationale

for turning off the Fed is that the student debt cancellation pro-

duces little to no inﬂationary impact in either model (shown

and discussed below). Despite the Federal Reserve’s essentially

mechanical responses to lower unemployment assumed within

the models as the positive macroeconomic effects of the debt

cancellation appear, there might be little actual rationale for

real-world central bankers to raise interest rates and subse-

quently dampen these positive effects.

It is important to recall that in both models there is no

breakdown of the private sector into borrowers and nonbor-

rowers of student loans. In the Moody’s model, there is some

capacity to break down the consumption function into groups

related to income or wealth in order to account for differing

propensities to consume, but our simulation does not account

for which income or wealth cohort is beneﬁting from the debt

cancellation. Instead, both models have a “household sec-

tor” that supplies labor, purchases consumption goods (also

Levy Economics Institute of Bard College 39

broken down into multiple categories in both models), and

buys houses. In the Fair model, there are relative-sized age

cohorts, and also gender effects for consumption and labor sup-

ply, but the model does not enable directly tying age and gender

cohorts to the distribution of debt relief from the cancellation

within the simulations. This all means that the simulated mac-

roeconomic impacts for both models are “general” or “average”

in the sense that they assume that the increase in net wealth and

reduced debt service beneﬁts the entire household sector, not

speciﬁc components of the household sector. Depending upon

how the beneﬁts would actually be distributed—for instance, to

younger individuals that might then be more likely to purchase

a home—the macroeconomic impacts could be more signiﬁ-

cant than reported here.

Finally, as is explained below in the discussion of the

impacts of the student debt cancellation on the government’s

budget position (and elaborated on in Appendix B), the presen-

tation in Section 2 regarding the ﬁnancial statement effects for

the federal government of cancelling the ED’s loans needs to be

reconciled with the approach taken in the simulations. The base-

line case for both models—inherent in the “assumptions for the

simulated student cancellation” presented above—is a scenario

in which no cancellation of student loans occurs and all the

loans are paid back over the 10-year period. Consequently, the

net budgetary impacts for the federal government in the simu-

lations are presented as a change in comparison to this base-

line, not in comparison to the government’s current budgetary

position. As explained in Section 2, the costs of funding the

ED’s loans have already been incurred; cancelling these loans

merely requires continued servicing of the securities issued at

that time. Although the foregone interest and principal pay-

ments on the ED’s loans both affect the government’s budget

position in comparison to the counterfactual assumed in the

simulations (a baseline of no cancellation and continued repay-

ment), the impact on the actual government budget position

and outstanding liabilities (that is, the national debt) would be

far smaller compared to its current levels.

Simulation Results

Turning to the results of the simulations, the discussion here

ﬁrst covers the main macroeconomic variables (real GDP,

employment, and inﬂation). Thereafter, other results of inter-

est from the simulation are presented. All, or nearly all, of the

related ﬁgures show simulated effects relative to the baseline

levels noted above for the respective models, since those are the

actual impacts of the debt cancellation within the simulations.

Figure 3.1 shows the total contribution of the cancellation

to real GDP (in 2016 $ billions) over 10 years in all four simula-

tions—the Fair model, the Fair model with the Fed’s reaction

function turned off (reported in this and subsequent ﬁgures as

“no Fed” for both models), the Moody’s model, and the Moody’s

model with no Fed reaction function. For the Fair model, the

cancellation creates $943 billion in total inﬂation-adjusted GDP

(or $94 billion per year, on average) “with the Fed,” which rises

to about $1,083 billion total (or $108 billion per year, on aver-

age) when the Fed is turned off. For the Moody’s model, with

the Fed the cancellation results in an additional $252 billion

total inﬂation-adjusted GDP (just under $25 billion per year, on

average), which rises to $861 billion total when the Fed is turned

off (just under $86 billion per year, on average).

20, 21

As can be seen in Figure 3.1, three of the simulation results

are very similar, while the Moody’s model with the Fed presents

a signiﬁcantly smaller total effect on real GDP. Figure 3.2 shows

the models’ results for each year of the simulation. Particularly

interesting is that all four simulations are similar in magnitude

and pattern through 2020, though the Fair model results with-

out the Fed are about $45 billion above the other three simu-

lations, on average. In all four simulations, the cancellation

creates a signiﬁcant increase in real GDP during 2017–20 that

peaks in 2018. After 2021, the increases are smaller because (1)

the initial wealth effect of the cancellation is decaying over time,

and (2) the government-guaranteed student loans owned by the

200

400

600

800

1,000

1,200

Fair Model Fair Model No Fed Moody’s Moody’s No Fed

Figure 3.1 Total Additional Real GDP Resulting from

Student Loan Cancellation for 2017–26

Source: Authors’ calculations

$ Billions (2016)

40 Student Debt Cancellation Report 2018

private sector are assumed to be paid off in 2020 in the absence

of the cancellation (that is, the stimulus of additional income

to borrowers ends in 2020, since without the cancellation the

loans would have been paid off by then). For both models, the

effect of turning off the Fed’s reaction function is that the GDP

increase is higher and tails off a bit more slowly, as the Fed’s

interest rate target does not rise above the baseline forecasts of

the models.

Starting in 2021, the Moody’s simulations with the Fed

show a substantially smaller rise in GDP relative to the other

three simulations, and even show a subtraction from real GDP

during 2022–26. While the other three simulations stabilize in

the range of an additional $40–60 billion in real GDP during

2022–26, clearly the Moody’s model is very sensitive to changes

in the Fed’s interest rate policy. This is unlike the Fair model

for these years, in which the simulation results with and with-

out the Fed turn out quite similar to one another in pattern and

magnitude. An important consideration for understanding and

evaluating the simulation results and their implications for the

cancellation, therefore, is whether or not such assumed changes

in the Fed’s target rate are warranted as a response to the mac-

roeconomic effects of the cancellation policy. This relates spe-

ciﬁcally to the cancellation’s effect on inﬂation in the ﬁgures and

analysis below, since the Fed’s stated long-run goal is low inﬂa-

tion rates that enable maximum sustainable growth in real GDP.

Figure 3.3 shows the average unemployment rates in the

two models for the entire 10-year period. Recall from above that

the baseline values for the models are not on their own signiﬁ-

cant—of actual interest are the differences between baseline val-

ues and simulation values. The simulations for the entire period

show unemployment rates 0.22 percentage points and 0.25 per-

centage points below the baseline values for the Fair model sim-

ulations with and without the Fed, respectively. For the Moody’s

simulations, the cancellation reduces the unemployment rate by

0.13 percentage points with the Fed and 0.36 percentage points

without the Fed.

Figure 3.4 shows the reduction in unemployment rates

from baseline values for each year of all four debt cancellation

simulations. As with real GDP in Figure 3.2, the reduction in

unemployment rates is greatest during 2017–20, with an aver-

age reduction of about 0.4 percentage points for the Fair model

simulations (peaking at 0.48 percentage points and 0.55 per-

centage points for the with-Fed and without-Fed simulations,

respectively) and 0.6 percentage points for the Moody’s simula-

tions (peaking at 0.87 and 0.88 percentage points, respectively).

During 2022–26, the Moody’s simulation with the Fed’s target

rate rule in place is once again an outlier, with unemployment

Fair Model

Fair Model No Fed

Moody’s

Moody’s No Fed

-100

-50

100

150

200

250

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

Sour

ce: Authors’ calculations

Figure 3.2 Additional Real GDP Resulting from Student

Loan Canc

ellation for Each Year of the Simulation

$ Billions (2016)

4.00

4.25

4.50

4.75

5.00

5.25

5.50

Figure 3.3 Average Unemployment Rate for Each Student

Debt Cancellation Simulation

Sour

ce: Authors’ calculations

Fair Model Baseline

Fair Model Baseline + Cancellation

Fair Model Baseline + Cancellation with No Fed

Moody’s Baseline

Moody’s Baseline + Cancellation

Moody’s Baseline + Cancellation with No Fed

Percent

Levy Economics Institute of Bard College 41

rates actually modestly higher (0.19 percentage points higher,

on average) than the baseline levels. The other three simula-

tions, on the other hand, average fairly stable reductions during

2022–26 of 0.13 percentage points (Fair model with the Fed),

0.19 percentage points (Fair model without the Fed), and 0.21

percentage points (Moody’s without the Fed).

Figure 3.5 presents the number of additional jobs created as

a result of the cancellation in each year. The job creation results

in the Fair model peak in 2018–20 at 1.18 million additional

jobs per year with the Fed’s interest rate rule in place and 1.44

million with the Fed’s rule turned off. For the Moody’s model,

the job creation effects peak at about 1.53 million jobs with the

Fed’s reaction function, and 1.55 million without it. In all four

simulations, job creation relative to the baseline tails off thereaf-

ter, but again the Moody’s simulation with the Fed is an outlier.

While the other three simulations show job creation stabiliz-

ing at 400,000 jobs above the respective baseline levels during

2022–26, the Moody’s simulation with the Fed shows a worsen-

ing economy and thus job reductions during this period.

By way of comparison, job creation in the United States

from 2010 to 2015 averaged 2.23 million per year. The simu-

lations thus suggest that two years after inception, student

debt cancellation alone might create 50 percent to 70 percent

as many jobs in its peak year as the current economic expan-

sion creates in an average year, and could continue to sustain

about one-third of the job creation seen in the cancellation’s

peak years throughout the duration of the cancellation. As with

the unemployment rates, the simulations also suggest that the

Moody’s model job creation projections are more sensitive to

the state of the economy than are the Fair model’s, as a greater

increase in real GDP in the latter leads to about the same peak

job creation as in the former. And yet again, there is signiﬁcantly

greater sensitivity to the Fed’s interest rate reaction function in

the Moody’s model. Although the results for the Moody’s simu-

lation without the Fed are similar to those for both Fair model

simulations, after the ﬁrst four years the Moody’s simulation

with the Fed’s reaction function produces very different results

from the other three simulations.

Figure 3.6 presents results for inﬂation. In the Moody’s

model, the inﬂation measure is the Consumer Price Index (here-

after, CPI); for the Fair model, the measure is the model’s own

index for ﬁrm-sector pricing, which has historically been highly

correlated with standard measures of consumer price inﬂa-

tion like the CPI and the Personal Consumption Expenditures

Figure 3.5 Additional Private-Sector Jobs Resulting from

Student Loan Cancellation

250,000

500,000

750,000

1,000,000

1,250,000

1,500,000

1,750,000

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

-500,000

-250,000

Fair Model

Fair Model No Fed

Moody’s

Moody’s No Fed

Source: Authors’ calculations

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

Figure 3.4 Reduction in Unemployment Rate for Each Year

of the Simulation

Percent

Fair Model

Fair Model No Fed

Moody’s

Moody’s No Fed

Source: Authors’ calculations

42 Student Debt Cancellation Report 2018

Price Index (PCEPI). In the Fair model, the inﬂationary effect

of student debt cancellation is modest, peaking at just below

0.3 percentage points of additional inﬂation (that is, compared

to the baseline) with the Fed’s interest rate rule in place, and a

bit above this level with the Fed’s rule turned off. After 2020,

consistent with the tailing off of real GDP contributions noted

above, the inﬂationary impact is actually negative—that is, the

cancellation reduces inﬂation in these years. For the Moody’s

model, the inﬂationary effects are even smaller—essentially at

zero, given that they never rise above 0.09 percentage points.

Overall, because even the largest effect on inﬂation in a sin-

gle year in either model (0.32 percentage points in 2018 in the

Fair model simulation with no reaction from the Fed’s interest

rate rule) is of little macroeconomic signiﬁcance, it is at least

arguable that the Fed would not respond to the student debt

cancellation by raising its interest rate target. Recall that this is

the rationale for including simulations for both models in which

the Fed’s interest rate rule is turned off.

Stated differently, a

case can be made that the Fed would not, or at least should not,

react to the cancellation by raising interest rates given its stated

goal of keeping inﬂation from rising above its target. If so, the

Moody’s simulation with the Fed’s target rate reaction function

included—the clear outlier in these simulations—is not as use-

ful a guide to the cancellation’s impacts as the other three sim-

ulations. Even the Fair model simulation with the Fed’s target

rate rule turned on might be of less interest, although (1) the

simulations with and without the Fed’s rule are not very dis-

similar (that is, the Fair model is less sensitive to interest rate

target changes than the Moody’s model), and (2) the inﬂation-

ary impacts of the cancellation, though small in terms of mac-

roeconomic signiﬁcance, were signiﬁcantly greater in the early

years of the Fair model simulation than in the Moody’s model.

Figures 3.7 and 3.8 show the simulated effects of debt can-

cellation on nominal interest rates. Figure 3.7 presents the Fed’s

response to the cancellation in both models when its reaction

function is turned on. In the Fair model, the Fed responds by

increasing its target rate by around 0.5 percentage points above

the baseline in 2018 and 2019; thereafter, the increases slowly

taper off until the Fed’s target rate settles at around 0.2 percent-

age points above the 2023–26 baseline.

In the Moody’s model, the Fed’s reaction is very simi-

lar during the ﬁrst six years of the simulation—the Fed raises

its interest rate target by 0.4 to 0.46 percentage points during

2019–20, after which the effect of the cancellation tails off to

almost nothing by the end of 2026. In fact, by 2024 the Fed is

setting the interest rate slightly lower than the baseline level.

In other words, the Fed’s early interest rate hikes that peak at

0.46 percentage points in 2020 have a sizeable enough impact in

slowing the economy that by the end of the simulation the Fed

is reversing course to attempt mild stimulus. This is consistent

Figure 3.6 Inflation Impacts of Student Debt Cancellation

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

Percent

Fair Model

Fair Model No Fed

Moody’s

Moody’s No Fed

Source: Authors’ calculations

Figure 3.7 Effect of Student Debt Cancellation on the

Federal Reserve’s Interest Rate Target—Differences from

Baseline Va lues

Sour

ce: Authors’ calculations

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

Percent

Fair Model

Moody’s

Levy Economics Institute of Bard College 43

with the results for the Moody’s model presented in the previous

ﬁgures. Most importantly, this again suggests that the Moody’s

results with the Fed’s target rule turned on are an outlier here.

From Figure 3.6, there is little evidence that the Fed should raise

interest rates, while from Figure 3.7, the Fed is only modestly

raising interest rates. The Moody’s model is clearly very sensi-

tive to changes in the Fed’s interest rate target, as a very modest

0.46 percentage point interest rate hike is sufﬁcient to bring the

stimulus effect of the student debt cancellation (an immediate

$1.4 trillion increase in net ﬁnancial wealth and a reduction in

debt service on this amount over 10 years) to an effective stand-

still (or worse) by the end of 2021.

The effects on the 10-year Treasury rate in Figure 3.8 are

also not large. The peaks in both models are not much differ-

ent—0.34 percentage points in the Fair model and 0.54 per-

centage points in the Moody’s model, both including the Fed’s

interest rate reaction function. The Moody’s model peaks in

2022, two years after the Fair model’s peak in 2020, following

the pattern of the impacts on the federal funds rate in Figure

3.7. The difference between the student debt cancellation’s aver-

age effects on the 10-year Treasury rate in the two models is

even smaller—the rate averages being 0.25 percentage points

higher than the baseline in the Fair model and 0.38 percentage

points higher in the Moody’s model.

Figure 3.8 only presents results without the Fed’s reac-

tion function in the case of the Moody’s model. Unlike in the

Moody’s model, the long-term interest rate in the Fair model

showed a negligible change from its baseline value when the

Fed’s reaction was turned off. The reason for this difference

is that while the Moody’s model incorporates the government

debt-to-GDP ratio as an explanatory variable for the 10-year

Treasury note, the Fair model does not. This is one of the sig-

niﬁcant differences in the two models. The Fair model takes a

more traditional Keynesian approach to explaining the 10-year

rate’s spread above the short-term rate: through lagged values

of the spread and changes in both the short-term rate and in

lagged values of the spread. The Moody’s model incorporation

of the national debt-to-GDP ratio, on the other hand, is con-

sistent with its self-described “Classical” long-run structure.

23, 24

In the Moody’s model simulations, the effects on the 10-year

Treasury rate when the Fed’s reaction function is turned off

average a 0.27 percentage point increase above the baseline. In

other words, the impact of changes in the government’s budget

position induced by the cancellation amount to a 0.11 percent-

age point increase in the 10-year Treasury rate (a 0.38 percent-

age point increase on average with the Fed’s reaction function,

minus 0.27 percentage points on average without the Fed’s reac-

tion function). If one considers the normal variation in longer-

term Treasury rates, this 0.11 percentage point difference is

Source: Authors’ calculations

0.0

0.2

0.4

0.6

0.8

1.0

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

Figure 3.8 Effect of Student Debt Cancellation on the

10-Year Treasury Rate—Differences from Baseline Values

Fair Model

Moody’s

Moody’s No Fed

Percent

Figure 3.9 Average Annual Deficit Impacts of Student Debt

Cancellation (percent of GDP)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Fair Model Fair Model No Fed Moody’s Moody’s No Fed

Percent

Sour

ce: Authors’ calculations

Deficit Impact

Deficit Impact Less Interest

44 Student Debt Cancellation Report 2018

within the realm of “statistical noise”; indeed, the increases in

the longer-term rates shown here resulting from the debt can-

cellation are not of macroeconomic signiﬁcance relative to his-

torical changes.

A large-scale debt cancellation can be expected to worsen

the federal government’s budget position absent extraordinarily

strong feedback effects from the program’s macroeconomic

stimulus. Figure 3.9 presents two separate views for each of the

simulations. First, the annual deﬁcit impacts—that is, after the

cancellation affects macroeconomic performance, which feeds

back to the government’s budget position—are between 0.65

percent and 0.75 percent of GDP in the Fair model simulations

and 0.75 percent of GDP in the Moody’s simulation without

the Fed’s rule in effect. This is consistent with the smaller mac-

roeconomic impact in the latter scenario, which would thereby

have a smaller positive feedback effect on the budget. In the

Moody’s simulation with the Fed’s rule in effect, however, the

net budget deﬁcit increase (0.89 percent of GDP on average)

is larger than the direct costs of the cancellation. This is due

to the poorer macroeconomic performance of the economy in

that scenario, which negatively affects the government’s budget.

Second, the simulations here assumed both that the govern-

ment was already running a deﬁcit (that is, the baseline for both

models is that the federal government is in a deﬁcit position at

least through 2026) and that there were no budgetary offsets

to the cancellation via spending cuts or revenue increases. As

a result, the deﬁcit effects include increased government debt

service. However, in the event that the government begins in

a surplus position or offsets the costs of the cancellation, this

debt service would not exist or be less than estimated here. This

would also be true if interest rates turn out to be signiﬁcantly

lower than is assumed in the models’ baseline cases (these base-

lines assume interest rates in the later years of the simulations

rise higher than what the Fed’s policymakers have been fore-

casting). The columns in Figure 3.9 labeled “Deﬁcit Impact Less

Interest” show that the average noninterest deﬁcit effects of the

cancellation are about 0.6 percent of GDP in both Fair model

simulations and also in the Moody’s simulation without the

Fed’s rule in effect. With the Fed’s rule in effect, the noninterest

deﬁcit effect is 0.69 percent of GDP for the Moody’s model.

From the earlier discussion in this section of the two mod-

els’ baseline forecasts, the deﬁcit impacts presented in Figure

3.9 signiﬁcantly overstate how the government’s actual budget

position and outstanding liabilities would be affected, relative

to their current levels. Appendix B explains this in detail as it

progresses through to an estimate of the more relevant average

annual deﬁcit impacts of the student debt cancellation: between

0.29 and 0.37 percent of GDP for the two Fair model simula-

tions and the Moody’s simulation without the Fed’s rule in

effect. These ﬁgures are not adjusted for debt service, and thus

are adjustments to the larger amounts in Figure 3.9. Adjusting

again for debt service would reduce these estimates still further.

Figure 3.10 presents the average multiplier effects of student

debt cancellation. The multiplier effect here is the total increase

in nominal GDP during the full simulation period divided by

the sum of the government’s total revenue loss from foregone

debt service and the spending increases on debt service paid

to private investors during the full simulation period (in other

words, the total direct costs of the cancellation). For the Fair

model, the multiplier of 1.32 (with the Fed’s reaction function

in place) and 1.5 (without the Fed’s reaction function) are at the

higher end of—though still in line with—those found in other

empirical studies, though less so when one considers that part

of the stimulus is from the wealth effect of private debt cancel-

lation (Eichengreen and O’Rourke 2012; Blanchard and Leigh

2013; Zandi 2008). The multipliers in the Moody’s model are

smaller. With the Fed’s reaction function in place, the multiplier

is 0.56, which is at the low end for most studies, particularly if

one considers again the wealth effect at work in the simulation.

Without the Fed’s interest rate reaction function, the multiplier

for the Moody’s model is 1.04, which is in a typical range. Again,

if the economy is less sensitive to the Fed’s interest rate increases

Sour

ce: Authors’ calculations

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Fair Model Fair Model No Fed Moody’s Moody’s No Fed

Figure 3.10 Average Multiplier Effects of Student Debt

Cancellation

Levy Economics Institute of Bard College 45

than the Moody’s model suggests, the weight of the evidence

falls in the 1.04 to 1.50 range for the multiplier.

Ignoring for the moment the wealth effect of the debt can-

cellation, it is worth noting that the multiplier’s path in these

simulations is through one of the traditionally smaller mul-

tipliers, at least from a Keynesian perspective. The cancella-

tion effectively raises borrowers’ incomes, and therefore has a

smaller multiplier effect than government stimulus via direct

purchases in these models. In other words, those that believe

direct increases in household income—as in a tax cut—will

have larger multipliers than traditional Keynesian models sug-

gest should also believe that multiplier effects of the debt can-

cellation will be at least as large as those reported here (if not

larger).

Recalling that the analysis in Appendix B of the costs of

student debt cancellation is more relevant for understanding

the costs relative to the current level of the national debt, one

can consider the multiplier effects of the cancellation from that

perspective. In this case, the multiplier effects rise to 2.79 and

3.33 for the Fair model simulations with and without the Fed’s

rule in effect, respectively. For the Moody’s model, the multipli-

ers are 1.03 with the Fed’s rule in place and 1.94 with the Fed’s

rule turned off.

Finally, in the Fair model the debt cancellation leads budget

positions to improve at the state level as a result of the stronger

macroeconomy, as shown in Figure 3.11 (in nominal terms).

This is important because state budgets tend to be procyclical

and exacerbate macroeconomic swings; improved state budget

positions would reduce the need to raise taxes or cut spending

in a recession. State budget position results were not available

from the Moody’s model. The dollar amounts shown in Figure

3.11 represent, on average, 0.11 percent of GDP for the simula-

tion with the Fed’s rule in place and 0.12 percent of GDP with

the Fed’s rule turned off. From a consolidated federal-plus-state

budget position perspective, the improvement in state budgets

offset by about one-ﬁfth the net budgetary effects reported in

Figure 3.9. If one uses the Fair model’s estimates of effects on

state budgets to proxy for the Moody’s model simulations, then

the net budgetary effect, on a consolidated basis, is an increase in

the deﬁcit-to-GDP ratio of 0.45–0.76 percentage points across

all four simulations, with the most likely range being 0.45–0.6

percentage points.

For the lower bound estimate without debt

service, the most likely range for the increase in the deﬁcit-to-

GDP ratio falls to 0.39–0.46 percentage points. From the analy-

sis in Appendix B of the budgetary impacts relative to current

levels, on a consolidated basis the deﬁcit-to-GDP ratio increases

by even less: within a range of 0.17 to 0.25 percentage points

(this includes the federal government’s debt service).

Conclusions from simulations

During the ﬁrst ﬁve years, all four simulations are quite simi-

lar and consistent. The Moody’s model with the Fed’s interest

rate target rule turned on only diverges from the other three

simulations in the latter ﬁve years of the simulated period.

Interestingly, the differences between the two models did not

arise from the major theoretical difference between them—the

“Classical” long run in the Moody’s model that is not present in

the Fair model. The core difference was that the Moody’s model

is signiﬁcantly more sensitive to small changes in interest rates

than the Fair model, as noted above several times. As the student

debt cancellation stimulates the economy, the Fed raises interest

rates by nearly the same amount in both the Fair model and the

Moody’s model; the effect in the Moody’s model is to slow the

economy signiﬁcantly. However, given that both models show

very little inﬂationary impact from the cancellation—both in

absolute terms and in terms of macroeconomic signiﬁcance—

it should not be unreasonable to expect that the Fed would

not react to the cancellation by raising rates, and therefore the

Moody’s simulations without the Fed’s reaction function would

be the more relevant ones. This is consistent with the fact that

the results of the two Fair model simulations are very much in

line with those of the Moody’s model with the Fed’s interest rate

target rule turned off.

Figure 3.11 Average Improvement in State Budget

Positions Resulting from Student Loan Cancellation

2017 2018 2019 2020 2021 2022 2023 2024 2025 2026

Sour

ce: Authors’ calculations

Fair Model

Fair Model No Fed

$ Billions

46 Student Debt Cancellation Report 2018

Given this, the primary takeaways from the Fair model and

Moody’s simulations of the debt cancellation are the following

(these exclude the “Moody’s with Fed” results, except in the ﬁfth

bullet point summarizing interest rate changes):

• The most likely range for the total increase in real GDP (in

2016 dollars) is estimated to be between $861 billion and

$1,083 billion for the entire 10-year period (or $86 billion to

$108 billion per year, on average).

• Unemployment rates could fall by about 0.22 to 0.36 percent-

age points on average over the entire period.

• There could be signiﬁcant macroeconomic improvements,

with real GDP rising (particularly early on), and peak addi-

tional job creation about 50 percent to 70 percent as large as

a typical year’s overall job creation in the 2010–15 expansion.

The Fair model suggests these effects peak about a year earlier

on average than in the Moody’s model.

• Inﬂationary effects appear to be small and macroeconomi-

cally insigniﬁcant.

• Interest rates rise modestly, if at all. The Fed raises rates 0.3–

0.5 percentage points early on in step with the economy’s

improvement, and then the increase relative to the baseline

values falls to 0.13 percentage points by the end of 2026.

Increases in longer-term rates peak in the range of 0.25–0.4

percentage points, mostly in a manner consistent with the

Fed’s approach to shorter-term rates (although the Moody’s

model suggests 0.2 to 0.25 percentage points of this increase

is due to government deﬁcits). However, given that there is

effectively no inﬂationary impact from the cancellation, it is

highly questionable whether the Fed would or at least should

raise interest rates in the ﬁrst place.

• The cancellation’s impact on the federal government’s budget

is, on average, modest, with a deﬁcit impact of 0.65 to 0.75

percent of GDP, which falls to between 0.59 and 0.61 percent

of GDP if the government is not in an overall deﬁcit position

(either because it begins with a surplus when the cancellation

is implemented or the cost of the cancellation is offset) or

interest rates remain very low. Furthermore, the calculations

from Appendix B suggest that the more relevant estimate of

the average annual rise in the deﬁcit ratio, relative to current

levels of the government’s budget position and outstanding

liabilities, is much smaller still: between 0.29 and 0.37 percent

of GDP.

• The cancellation improves state budget positions such that,

from the standpoint of a consolidated state-plus-federal

budget, the net increase in the (consolidated) budget deﬁcit

is 0.10 to 0.12 percent of GDP lower than the increase for the

federal government alone.

Of course, as discussed earlier, there are many potential

beneﬁts of debt cancellation that cannot be simulated in a mac-

roeconometric model. These simulations should therefore be

considered as providing estimates of a “subset” of these poten-

tial beneﬁts. The following section presents and explains some

of the beneﬁts beyond the scope of these simulations.

Omitted Beneﬁts and Costs of Student Debt

Cancellation

The macroeconomic results presented in this report summarize

the income and wealth effects of student debt cancellation but

cannot capture all of the potential advantages and disadvan-

tages of the program. Evidence supports a number of socioeco-

nomic beneﬁts that have been omitted from the models here.

Despite their formal exclusion, this section provides context

for the additional beneﬁts of student debt cancellation, from

increases in business formation, college attainment, household

formation, and credit scores, to reduced economic vulnerabil-

ity for some households. Additionally, the projected costs asso-

ciated with changes in future attitudes toward borrowing are

only partially incorporated into the model, in order to isolate

the effects of the cancellation from other policy changes. The

implications of a one-time student debt cancellation for moral

hazard are also considered below.

Small business formation

Starting a business requires access to capital and an appetite for

risk—two characteristics that may be inhibited by high student

debt levels. Small businesses in particular tend to depend on

ﬁnancing from personal debt. Since student loan debt appears

as negative net worth on household balance sheets, borrowers

with high debt balances and monthly payments ﬁnd it more

difﬁcult to accumulate startup capital through saving or bor-

rowing. As a result, the growth of student loan debt is associated

with reductions in small business formation.

Recent research from the Federal Reserve Bank of

Philadelphia examined this relationship between the growth in

student loan debt and small business formation. Philadelphia

Federal Reserve researchers Ambrose, Cordell, and Ma (2015)

show that student loan debt reduces an individual’s ability to

Levy Economics Institute of Bard College 47

save startup capital or access alternative forms of credit for busi-

ness formation. The authors estimate these trends at the county

level over the period 2000–10, while accounting for county-level

differences in demographic and risk factors. During the period

of study, student debt as a share of total personal debt rose

across all counties, with an average rise of 4.7 percentage points

(from 2.8 percent of total debt in 1999 to 7.5 percent in 2009)

and a standard deviation of 3.3 percent. The authors estimate

a 14.4 percent decline in small business formation associated

with an increase of one standard deviation of relative student

debt (that is, relative to total personal debt). These effects are

strongest among the smallest-sized category of business—those

with one to four employees—where personal credit is presumed

to be a larger portion of total business ﬁnancing.

The negative relationship between student loan debt and

entrepreneurship identiﬁed by Ambrose, Cordell, and Ma

is supported by survey responses solicited by the Consumer

Financial Protection Bureau (CFPB). In 2013, the CFPB pub-

lished an analysis of detailed public comments on student loan

affordability (CFPB 2013). Individuals, small business coali-

tions, and advocacy groups identiﬁed student loan debt as a

barrier to accessing credit and debt payments as a diversion of

business startup and expansion funds. According to the report,

young entrepreneurs in particular are less able and less likely to

form businesses due to the accumulation of student loan debt.

College degree attainment

One principal beneﬁt of pursuing postsecondary education,

and taking on student loans, is the education premium that

accompanies an advanced degree. This premium is only real-

ized if a degree is completed, but mounting debt makes degree

attainment more difﬁcult for students facing ﬁnancial con-

straints. The relationship between debt and college completion

is documented in several papers, with some consensus around

the ﬁnding that high levels of student debt reduce college com-

pletion while access to grant funding promotes enrollment and

persistence, leading to higher degree attainment.

Three articles drawing on the National Longitudinal Survey

of Youth identify a positive correlation between student debt

and college completion at lower levels of debt, turning nega-

tive at an inﬂection point around $10,000—well below the aver-

age debt level of borrowers. In the journal Social Forces, Dwyer,

McCloud, and Hodson (2012) ﬁnd that this relationship holds

true across the income distribution but that the relationship is

stronger for students from the bottom 75 percent of earnings.

The same authors reproduce these ﬁndings in a 2013 article, and

ﬁnd that while high debt reduces the chances of graduating for

both men and women, men are likely to drop out at lower levels

of debt (Dwyer, McCloud, and Hodson 2013). In the Journal of

Sociology and Social Welfare, Min Zhan (2014) shows that stu-

dent loan debt above $10,000 reduces college graduation even

after accounting for family assets.

High student debt as a share of total debt reduces an individual’s ability to access

alternative forms of credit for business formation. Small business formation declines 14.4

percent with an increase of one standard deviation of relative student debt.

Grant funding increases college attendance and reduces college dropout rates. A $1,000

increase in Pell Grants is associated with a 1.2 percent to 8.6 percent decrease in students

leaving college; $1000 in non-need-based grant aid increases college attendance by 3.6

percent.

A $1,000 increase in student debt is associated with a 2 percent decline in the likelihood

of ﬁrst marriage among female degree-holders.

Student loan borrowers have lower credit scores, potentially leading to household credit

constraints and reduced consumption.

Households with student debt experienced greater reductions in net worth than

households with no student debt during the most recent recession. A $1 increase in

student loan debt in 2007 was associated with $0.87 less in net worth in 2009.

Table 3.1 Additional Beneﬁts of Student Debt Cancellation

Predicted Effect

of Student Debt

Cancellation

Description Source

Increased small

business formation

Increased college

degree attainment

Increased household

formation

Higher credit scores

Reduced vulnerability

to economic shocks

Ambrose, Cordell, and Ma (2015)

Bettinger (2004)

Dynarski (2003)

Bozick and Estacion (2014)

Addo (2014)

Edmiston, Brooks, and Shepelwich (2013)

Li (2013)

Elliott and Nam (2013)

48 Student Debt Cancellation Report 2018

For predicting changes in college completion associated

with a student debt cancellation in the context of free or debt-

free college, it may be most useful to look at the relationship

between debt-free education ﬁnancing and degree attainment.

Here two studies support a relationship between grant fund-

ing and college completion. In a 2003 article in the American

Economic Review, Susan Dynarski (2003) demonstrated that

the elimination of grant aid reduced college attendance among

the previously eligible population by more than one-third.

Dynarski found that the availability of grant aid increases both

attendance and completion, with an offer of $1,000 in funding

increasing the chance of attendance by 3.6 percentage points.

In a 2004 National Bureau of Economic Research study, author

Eric Bettinger (2004) performs both panel and cross-sectional

analysis on the relationship between need-based Pell Grants and

college completion. The panel data shows a strong negative cor-

relation between Pell Grant increases and drop-out rates, with

a $1,000 increase in Pell Grants associated with 6.4 percent to

8.6 percent decreases in students leaving college. The cross-

sectional analysis shows similar though smaller results, with 1.2

percent to 4 percent decreases in the likelihood of dropping out

associated with a $1,000 Pell Grant increase.

The cancellation plan would reduce the current debt bur-

den on those enrolled in school, and will likely increase the

rate of completion. Over a longer time horizon, this increase

in degree attainment will extend the positive effects of the can-

cellation by increasing the income and productivity of the US

labor force.

Household formation

Entering marriage and beginning a household is associated with

a range of socioeconomic beneﬁts, including better health and

higher income, while household public goods and risk pool-

ing can be a means out of poverty (Schwartz 2005). Existing

research suggests that debt, including student debt, is associated

with a decreased probability of household formation. Recent

studies isolating the relationship between student debt and

marriage support this claim.

The impact of educational debt on decisions to marry is

observed by Bozick and Estacion (2014) using data from the

1993 Baccalaureate and Beyond Longitudinal Study. Using

a discrete-time hazard model, they ﬁnd that the odds of ﬁrst

marriage decline by 2 percent with an increase of $1,000 in

student loan borrowing among females in the ﬁrst four years

after attaining a college degree. In Demography, Addo (2014)

produces similar results using the National Longitudinal Survey

of Youth.

Credit scores

Student loan borrowers in general exhibit lower credit scores

than the population overall (Edmiston, Brooks, and Shepelwich

2013). The New York Federal Reserve’s Brown and Caldwell

(2013) demonstrate that student loan borrowers ages 25 to 30

had Equifax risk scores 15 to 24 points below those of nonbor-

rowers in the years between 2008 and 2013. Researchers disagree

whether this divergence in credit ratings occurs due to higher

delinquency and default rates associated with unmanageable

student debt levels or due to increased lending to student bor-

rowers who already had low credit scores after the reform of

bankruptcy discharge laws that made it increasingly difﬁcult

for borrowers who cannot make their student loan payments to

discharge the debt (Edmiston, Brooks, and Shepelwich 2013; Li

2013; Darolia and Ritter 2015). In either case, low credit scores

may reduce access to other forms of credit despite the higher

earning potential of college graduates.

Household vulnerability in business cycle downturns

Credit constraints, delayed household formation, lower net

worth, and debt service obligations can all be sources of eco-

nomic fragility associated with student debt. For these reasons,

student loan debt can have negative ﬁnancial consequences for

individuals and families even when all payments are on time

and up to date. In a 2013 publication from the Federal Reserve

Bank of St. Louis, authors Elliot and Nam (2013) investigate the

relationship between student debt and household economic

security. They use the Survey of Consumer Finances to mea-

sure the relationship between student debt and net worth dur-

ing periods of economic instability. Using the years 2007 and

2009 as reference points for the Great Recession, the authors

ﬁnd that households carrying student debt faced greater losses

in net worth during the recession compared to similar house-

holds with no student debt. According to their research, each $1

increase in student loans for the median household in 2007 was

associated with lower net worth of $0.87 in 2009. The negative

relationship between student debt and net worth appeared con-

sistently, regardless of net worth quintile, but the largest relative

losses occurred among households at the bottom of the income

distribution.

Levy Economics Institute of Bard College 49

Moral hazard

The best context for student debt cancellation is one where a

high-quality college education is available to all students who

seek it without the need for debt ﬁnancing. Without a change

to our current system of increasingly private responsibility for

funding higher education, households will continue to meet

the growing cost of a college degree by taking on debt, divert-

ing household resources from other types of investment and

consumption. The primary theoretical criticism of debt can-

cellation plans focuses on the reaccumulation of debt follow-

ing the cancellation, in particular the potential for problems of

moral hazard to arise. From this perspective, debt relief today

could change the incentives of future student debtors who may

increase borrowing with the expectation that the loans will

be forgiven, causing an even faster accumulation of debt and

increasing the negative consequences at the household, local,

and macroeconomic levels. The perverse incentives for unsus-

tainable borrowing in this scenario are the result of inappro-

priate policy institutions that absolve borrowers of their debts

while perpetuating the necessity of increasing debt. In order to

avoid problems of moral hazard, any restructuring of student

debt—including our debt cancellation proposal—should be

accompanied by strong and appropriate policies that enforce

the consequences of borrowing and address the market failures

that lead to undesirable social costs. In combination with debt

cancellation, publicly funded free or debt-free college would

provide the institutional reform necessary to avert the problem

of moral hazard.

Although complementary reform of higher education

ﬁnancing should accompany a student debt cancellation, this

research is focused on the speciﬁc question of the impact of

total cancellation of current debts. It is not an attempt to study

the institutions necessary to frame a debt cancellation. Each

model in this report isolates the effects of debt cancellation. In

Moody’s structural macroeconomic model, the cancellation is

evaluated in the context of the Clinton Compact, a policy mak-

ing debt-free public college attainable for more than 80 per-

cent of households and largely eliminating the need for future

debt associated with a four-year college degree. The difference

between the modeled effects of debt-free college alone and

debt-free college in conjunction with a program of student debt

cancellation is the positive impact of debt cancellation in the

absence of moral hazard. In the Fair model scenario, student

debt reaccumulates beginning in the ﬁrst quarter following the

cancellation. No complementary policy is incorporated into

the simulation. Each model imposes an institutional context in

which moral hazard problems do not arise in order to focus the

analysis on student debt cancellation.

50 Student Debt Cancellation Report 2018

Conclusion

This report examines the context, implementation, and out-

comes of a program of complete student debt cancellation. We

ﬁnd that student debt cancellation produces positive feedback

effects that improve several macroeconomic variables, including

GDP and job growth, while imposing only moderate increases

on the federal deﬁcit and interest rates and no signiﬁcant inﬂa-

tionary pressure. These results support the continued inclusion

of bold proposals such as student debt cancellation in public

policy deliberations surrounding the future of higher education

in the United States. Our ﬁndings offer an essential contribu-

tion to this debate.

In Section 1, we review the trends in higher education costs,

public ﬁnancing, and student debt. The increasing need for a

college degree to attain ﬁnancial security drew more students

into higher education at the same time that public support for

education declined, prompting the growth of student debt to

record levels. Today, student loan debt presents a signiﬁcantly

higher burden on household ﬁnances than ever before, with

implications for the entire economy. Many borrowers struggle

to make payments, while others forego important investment

opportunities such as homeownership and business forma-

tion. These limitations translate into lower consumption and

investment spending in the aggregate, leading to slower growth,

greater vulnerability to economic shocks, and the potential for

a higher education market failure. Complete cancellation of

outstanding student loans could reverse many of these negative

effects.

The possibility of enacting student debt cancellation is the

subject of Section 2, where we examine the current mechan-

ics of student lending and the balance sheet effects of a pro-

gram of debt cancellation. There are two key takeaways. First,

the new budgetary costs of cancelling the loans issued by the

Department of Education (ED) do not come in the form of for-

gone principal and interest payments on those loans—the sole

costs would be the continuation of debt service on the securities

originally issued by the Treasury to fund the ED loans. Second,

whether the cancellation is carried out by the federal govern-

ment or the Fed, the outcome is the same in terms of the ﬁnan-

cial positions of borrowers and the federal budget. There is no

budgetary “free lunch” in having the Fed carry out the cancella-

tion. Under the Fed-initiated debt cancellation, however, there

are two possible options to reduce the effects on the govern-

ment balance sheet. One option is for the Fed to accommodate

the losses by purchasing new ﬁnancial assets; the other option is

for the Fed to isolate losses from cancelling student loans from

the rest of its operating proﬁts. Both options are quite contro-

versial, but would shield the federal government from the bal-

ance sheet effects of enacting the cancellation.

Finally, in Section 3 we forecast the effects of debt cancella-

tion over a 10-year horizon using two macroeconomic models,

the Fair model and Moody’s model. The results of these simula-

tions take into account the feedback effects of greater household

consumption and investment that are not captured in the bal-

ance sheet analysis in Section 2.

Our simulations show that student debt cancellation

results in an increase in GDP, a decrease in the average unem-

ployment rate, and little to no inﬂationary pressure over the

10-year horizon, while interest rates increase only modestly.

(Results reported here are from the two Fair model simula-

tions and the Moody’s simulation with the Fed’s interest rate

reaction function turned off.) Estimates for new GDP range

from $861 billion to $1,083 billion over the entire period, or

on average between $86 billion and $108 billion per year. This

increase is accompanied by new job creation that peaks at 1.18

to 1.55 million additional new jobs per year, or 50 to 70 percent

of the entire job creation for a typical year in the 2010–15 eco-

nomic expansion. Average unemployment rates over the period

are reduced by between 0.22 and 0.36 percentage points. The

predicted effects of the cancellation on inﬂation are negligible,

with a peak of 0.3 percentage points of additional inﬂation in

the Fair model and negative pressure on inﬂation in later years,

and no more than 0.09 percentage points of additional inﬂation

in the Moody’s model over the entire period. The simulations

suggest that the Federal Reserve raises target rates modestly in

the early years of the cancellation, adding 0.3 to 0.5 percentage

points to the rate, with lower effects in later years. The effect on

longer-term interest rates peaks at 0.25 to 0.4 percent. Finally,

government spending to repay privately held loans and the loss

of interest income from ED loans results in a larger budget deﬁ-

cit for the federal government. The average effect of the cancel-

lation on the federal government’s net budget position ranges

between –0.65 and –0.75 percent of GDP per annum. However,

those ﬁgures assume all the foregone revenues from cancelling

the Department of Education’s loans are incurred anew (see

Appendix B). The more relevant estimate of these impacts—

relative to current levels of deﬁcits and the national debt—is a

Levy Economics Institute of Bard College 51

range of –0.29 to –0.37 percent of GDP. Finally, the Fair model

shows an improvement in state budget positions (these effects

were not available in the Moody’s simulations).

Our analysis suggests that debt cancellation is a feasible

program that would increase economic activity in the short

run with moderate consequences on the federal deﬁcit. These

consequences should be balanced against the important social

gains available from greater investment in higher education and

the relief of debt as educators, advocates, borrowers, and poli-

cymakers continue to debate the path forward for US higher

education.

52 Student Debt Cancellation Report 2018

Appendix A: Simulation Data Series

The macroeconomic simulations generate a $1.406 trillion

one-time increase in net wealth of the household sector in the

ﬁrst quarter of 2017. The 10-year horizon for the government’s

reduction in revenues, payments to private investors, and debt

service payments for the Department of Education (ED) and

privately owned loans are presented in Appendix Tables A.1,

A.2, A.3, and A.4. Table A.1 shows the estimated principal

reduction of the different types of loans in the absence of debt

cancellation for the purposes of determining (1) the amount of

revenue the government does not receive but would have with-

out the cancellation and (2) the debt service payments to pri-

vate investors that the federal government subsequently bears.

Table A.2 presents the estimated debt service for loans owned by

the ED. These are calculated as the pay down in principal each

quarter (the change in the respective rows in Table A.1) plus

interest (interest rate multiplied by principal owed at the end of

Table A.1 Assumed Pay Down of Student Loan Debt in the Absence of Student Debt Cancellation ($ billions)

Owned by ED Privately Owned, Government-Guaranteed Privately Owned, Not Government-Guaranteed

Counterfactual for Debt Owed Counterfactual for Debt Owed Counterfactual for Debt Owed

at End of Quarter at End of Quarter at End of Quarter

End of 2016 1024.61 276.95 105.44

2017Q1 998.64 260.66 102.87

2017Q2 973.68 244.37 100.30

2017Q3 948.71 228.08 97.72

2017Q4 923.74 211.79 95.15

2018Q1 898.78 195.50 92.58

2018Q2 873.81 179.20 90.01

2018Q3 848.85 162.91 87.44

2018Q4 823.88 146.62 84.87

2019Q1 798.91 130.33 82.29

2019Q2 773.95 114.04 79.72

2019Q3 748.98 97.75 77.15

2019Q4 724.02 81.46 74.58

2020Q1 699.05 65.17 72.01

2020Q2 674.08 48.87 69.44

2020Q3 649.12 32.58 66.86

2020Q4 624.15 16.29 64.29

2021Q1 599.18 0.00 61.72

2021Q2 574.22 0.00 59.15

2021Q3 549.25 0.00 56.58

2021Q4 524.29 0.00 54.01

2022Q1 499.32 0.00 51.43

2022Q2 474.35 0.00 48.86

2022Q3 449.39 0.00 46.29

2022Q4 424.42 0.00 43.72

2023Q1 399.46 0.00 41.15

2023Q2 374.49 0.00 38.58

2023Q3 349.52 0.00 36.00

2023Q4 324.56 0.00 33.43

2024Q1 299.59 0.00 30.86

2024Q2 274.63 0.00 28.29

2024Q3 249.66 0.00 25.72

2024Q4 224.69 0.00 23.15

2025Q1 199.73 0.00 20.57

2025Q2 174.76 0.00 18.00

2025Q3 149.80 0.00 15.43

2025Q4 124.83 0.00 12.86

2026Q1 99.86 0.00 10.29

2026Q2 74.90 0.00 7.72

2026Q3 49.93 0.00 5.14

2026Q4 24.97 0.00 2.57

Source: Authors’ calculations

Levy Economics Institute of Bard College 53

the previous quarter). The debt service for the ED-owned loans

is the estimated reduction in government revenue as a result of

the cancellation. Table A.3 is similar to Table A.2, but instead

shows debt service for student loans owned by the private sector

without government insurance; this is an assumed outlay in the

simulations, as the government takes on these payments. Table

A.4 is the debt service for the privately owned, government-

guaranteed loans, which are also borne by the government in

the cancellation but are assumed to be paid down completely

by the end of 2020. Interest on these loans is calculated as the

outstanding principal from the previous quarter multiplied by

the short-term rate from the previous quarter plus the markup

(currently 2.3 percent), and then divided by four for quarterly

compounding. As noted above, the short-term interest rate is

determined within the simulations. As an example only, Table

A.4 uses the baseline level of the T-Bill rate from the Fair model

forecasts (that is, the simulated model without the debt cancel-

lation incorporated).

Government Owned

Interest Rate = 4.6%

Counterfactual Debt Service ($ billions)

2017Q1 36.74

2017Q2 36.16

2017Q3 35.88

2017Q4 35.59

2018Q1 35.30

2018Q2 35.01

2018Q3 34.73

2018Q4 34.44

2019Q1 34.15

2019Q2 33.87

2019Q3 33.58

2019Q4 33.29

2020Q1 33.01

2020Q2 32.72

2020Q3 32.43

2020Q4 32.14

2021Q1 31.86

2021Q2 31.57

2021Q3 31.28

2021Q4 31.00

2022Q1 30.71

2022Q2 30.42

2022Q3 30.13

2022Q4 29.85

2023Q1 29.56

2023Q2 29.27

2023Q3 28.99

2023Q4 28.70

2024Q1 28.41

2024Q2 28.12

2024Q3 27.84

2024Q4 27.55

2025Q1 27.26

2025Q2 26.98

2025Q3 26.69

2025Q4 26.40

2026Q1 26.11

2026Q2 25.83

2026Q3 25.54

2026Q4 25.25

Table A.2 Assumed Debt Service Payments Not Received

after Cancellation of Student Loans Owned by the

Department of Education ($ billions)

Source: Authors’ calculations

Table A.3 Assumed Debt Service Payment Outlays for the

Government after Cancellation of Student Loans Owned by

the Private Sector and Not Government-Guaranteed

($ billions)

Privately Owned, No Government Insurance

Interest Rate = 10%

Counterfactual Debt Service ($ billions)

2017Q1 5.21

2017Q2 5.14

2017Q3 5.08

2017Q4 5.01

2018Q1 4.95

2018Q2 4.89

2018Q3 4.82

2018Q4 4.76

2019Q1 4.69

2019Q2 4.63

2019Q3 4.56

2019Q4 4.50

2020Q1 4.44

2020Q2 4.37

2020Q3 4.31

2020Q4 4.24

2021Q1 4.18

2021Q2 4.11

2021Q3 4.05

2021Q4 3.99

2022Q1 3.92

2022Q2 3.86

2022Q3 3.79

2022Q4 3.73

2023Q1 3.66

2023Q2 3.60

2023Q3 3.54

2023Q4 3.47

2024Q1 3.41

2024Q2 3.34

2024Q3 3.28

2024Q4 3.21

2025Q1 3.15

2025Q2 3.09

2025Q3 3.02

2025Q4 2.96

2026Q1 2.89

2026Q2 2.83

2026Q3 2.76

2026Q4 2.70

Source: Authors’ calculations

54 Student Debt Cancellation Report 2018

Table A.4 Assumed Debt Service Payment Outlays for the Government after Cancellation of Government-Guaranteed

Student Loans Owned by the Private Sector ($ billions)

Privately Owned, Government-Guaranteed

Interest Rate = Previous Quarter’s Fed Funds Rate + 2.3%

Principal Remaining at (Example) Fair Model Base Interest Rate Counterfactual

End of Previous Quarter T-Bill Rate from Markup (%) Debt Service

($ billions) Previous Quarter (%) ($ billions)

2017Q1 260.66 0.31 2.30 18.10

2017Q2 244.37 0.25 2.30 17.95

2017Q3 228.08 0.27 2.30 17.86

2017Q4 211.79 0.33 2.30 17.79

2018Q1 195.50 0.40 2.30 17.72

2018Q2 179.20 0.48 2.30 17.65

2018Q3 162.91 0.59 2.30 17.59

2018Q4 146.62 0.72 2.30 17.52

2019Q1 130.33 0.85 2.30 17.45

2019Q2 114.04 0.98 2.30 17.36

2019Q3 97.75 1.12 2.30 17.27

2019Q4 81.46 1.25 2.30 17.16

2020Q1 65.17 1.39 2.30 17.04

2020Q2 48.87 1.53 2.30 16.92

2020Q3 32.58 1.67 2.30 16.78

2020Q4 16.29 1.81 2.30 16.63

2021Q1 0.00 1.95 2.30 24.33

Source: Authors’ calculations

Levy Economics Institute of Bard College 55

Appendix B: Department of

Education Loans and the Budget

Deﬁcit

Macroeconometric simulations by nature compare a policy or

change in one or more variables to a “baseline case.” The simula-

tions presented in this report likewise compare the student debt

cancellation to the baseline case of no student debt cancellation.

However, in Section 2 of the report, it is shown that for the can-

cellation of the Department of Education’s (ED’s) loans there

is a difference between comparing the change in government

liabilities relative to no cancellation and the actual increase in

liabilities that would occur. This is because the ED’s loans were

previously funded via issuance of government securities; as a

result, only the interest due on these securities is ﬁnanced by

new increases in government securities outstanding. There is no

increase in government liabilities from cancellation of the princi-

pal on the ED’s loans. Rather, the principal amount of the loans,

funded originally by previously issued securities, is rolled over.

The fundamental difference is that the baseline case in

the simulations assumes the ED’s loans are paid down, which

would in theory enable retiring the securities previously issued

to fund the ED’s loans. Relative to this assumed baseline, student

debt cancellation raises the government deﬁcit and increases

the amount of government securities issued by the combined

amount of the principal and interest on the ED’s loans. This

larger deﬁcit is then compounded as it further raises future debt

service on the national debt, raising future deﬁcits as well. While

the simulation results reported are “correct” in the sense that

they report changes from the baseline, they also signiﬁcantly

overstate the actual, absolute increases in the national debt—

and thus also the actual, annual deﬁcits—that would result

from student debt cancellation relative to their current levels.

The purpose of this section is to provide some understanding

of the size of this overstatement.

To generate an estimate of the actual deﬁcit impact of stu-

dent debt cancellation relative to current levels, the ﬁrst step

is to estimate the effect of the cancellation of the ED’s loans.

Table B.1 presents the Federal Reserve’s interest rate targets

in the three core simulations: the Fair model, the Fair model

Fair Model Fair Model No Fed Moody’s No Fed

2017 0.49% 0.31% 2.03%

2018 1.15% 0.66% 3.69%

2019 1.71% 1.18% 3.87%

2020 2.17% 1.74% 3.75%

2021 2.61% 2.29% 3.78%

2022 3.07% 2.85% 3.85%

2023 3.60% 3.40% 3.85%

2024 4.17% 3.97% 3.81%

2025 4.76% 4.57% 3.78%

2026 5.33% 5.14% 3.78%

Table B.1 Federal Reserve’s Interest Rate Target in the

Student Debt Cancellation Simulations

Source: Authors’ calculations

Fair Model Fair Model No Fed Moody’s No Fed

2017 5.05 3.20 20.77

2018 11.84 6.79 38.52

2019 17.75 12.24 41.87

2020 22.94 18.17 42.18

2021 28.21 24.38 44.08

2022 34.06 30.98 46.67

2023 41.15 38.06 48.40

2024 49.36 45.95 49.72

2025 58.74 54.94 51.27

2026 68.88 64.70 53.18

Total 337.99 299.40 436.66

Table B.2 Annual Interest Cost for Securities Issued to

Fund Department of Education Loans in the Student Debt

Cancellation Simulations ($ billions)

Source: Authors’ calculations

2017 146.79 243.70 60.24%

2018 141.79 236.90 59.85%

2019 137.08 229.38 59.76%

2020 132.37 221.05 59.88%

2021 127.66 144.14 88.57%

2022 122.96 138.38 88.86%

2023 118.25 132.61 89.17%

2024 113.54 126.85 89.51%

2025 108.83 121.09 89.88%

2026 104.12 115.32 90.29%

Totals 1253.45 1709.48 73.32%

Table B.3 Data Series Estimates from Simulations for

Department of Education Loans and Total Debt Cancellation

($ billions)

Source: Authors’ calculations

Total Direct Spending

Increases/Revenue Losses

Due to Cancellation of All

Student Loans

(B)

Foregone Principal

and Interest on

ED’s Loans

(A) (A) / (B)

56 Student Debt Cancellation Report 2018

If one makes the (admittedly fairly crude but not altogether

unreasonable) assumption that the relative size of the direct

revenue/spending effect from cancelling the ED’s loans with

respect to the total for cancelling all student loans is equal to

its contribution to the annual deﬁcit, then an alternative deﬁcit

estimate can be made that is more appropriate as a change from

current levels. Note that the more similar the multiplier effects

from cancelling the different types of student loans are, the more

reasonable this assumption is. Since all of the cancellations are

ultimately affecting the economy via the household sector, mul-

tiplier effects, at least within these models, would seem fairly

consistent across different loan types. Table B.4 presents the

increases in annual budget deﬁcits from baseline forecast results

in each simulation. Table B.5 multiplies the entries in Table B.4

by the third column of Table B.3. In other words, Table B.5 can

serve as an estimate of how much the assumed loss of revenues

from cancelling the ED’s loans affected the government’s budget

in the simulations.

To calculate an estimate of the actual effect of cancelling the

ED’s loans on the deﬁcit, relative to current levels, subtract the

entries in Table B.5 from the entries in Table B.4, then add the

entries in Table B.2. This simply replaces the effect of the ED’s

loans on the deﬁcit in Table B.4 with the estimates of debt ser-

vice on the securities issued to fund the ED’s loans in Table B.2

in calculating the deﬁcit. If anything, this would be an overes-

timate, because there are no feedback effects assumed from the

spending in Table B.2 that might reduce the deﬁcit. Though the

multiplier effects of government debt service on the economy

are likely small, there are some offsets to the spending’s effect on

Fair Model Fair Model No Fed Moody’s No Fed

2017 212.84 210.63 227.67

2018 189.19 175.16 169.88

2019 198.77 171.47 187.42

2020 210.54 176.97 212.11

2021 156.39 123.08 158.74

2022 162.08 132.56 165.56

2023 158.91 132.66 160.26

2024 155.14 129.53 157.13

2025 153.11 126.37 158.88

2026 152.71 124.32 164.16

Totals 1749.69 1502.76 1761.86

Average 174.97 150.28 176.19

Table B.4 Increases in Annual Budget Deﬁcits from

Baseline Forecasts in the Student Debt Cancellation

Simulations ($ billions)

Source: Authors’ calculations

Fair Model Fair Model No Fed Moody’s No Fed

2017 128.20 126.87 137.13

2018 113.23 104.83 101.67

2019 118.78 102.47 112.01

2020 126.08 105.97 127.02

2021 138.51 109.01 140.59

2022 144.01 117.78 147.11

2023 141.69 118.28 142.90

2024 138.86 115.93 140.64

2025 137.61 113.58 142.80

2026 137.87 112.24 148.21

Totals 1324.89 1127.02 1340.14

Table B.5 Estimated Deﬁcit Impact of Cancellation of

Department of Education Loans in the Student Debt

Cancellation Simulations ($ billions)

Source: Authors’ calculations

without the Fed’s rule in effect, and the Moody’s model without

the Fed’s rule in effect. These interest rates would be applied

to the rolling over of the securities previously issued to fund

the ED’s loans. As Section 2 of this report explains, this debt

service cost is the actual cost to the government of cancelling

the ED’s loans relative to current spending levels. To estimate

this cost, the interest rates in Table B.1 can be applied to the

value of the outstanding ED loans at the beginning of the simu-

lations ($1.024 trillion) that will subsequently be cancelled. As

the debt is rolled over each period and the interest rates also rise

throughout the simulation’s duration, the cost rises. Table B.2

presents these annual costs for each simulation, which would

be estimates of the direct costs to the government of the cancel-

lation of the ED’s loans, relative to current levels. The ﬁnal row

sums up each column.

Because the simulations instead presented a counterfac-

tual, the full revenue loss from foregone interest and principal

payments were assumed to be the direct costs of cancelling the

ED’s loans. From the data series estimates of the debt cancella-

tion used in the simulations, the principal and interest foregone

on the ED loans make up roughly 60 percent of the total direct

spending/direct revenue loss from the cancellation during the

2017–20 period (i.e., the ﬁrst four years), and roughly 90 percent

during the 2021–26 period. The ﬁgures used in the simulation

are shown in Table B.3. The ﬁrst column is estimated foregone

principal and interest from cancelling the ED’s loans. The sec-

ond column is the total estimated direct spending increases and

revenue losses from the cancellation of all student loans. The

third column is the ﬁrst column divided by the second column.

Levy Economics Institute of Bard College 57

the deﬁcit, at the very least through taxation of interest in cer-

tain cases. Table B.6 presents the new estimates of government

deﬁcits from student debt cancellation.

The differences between Table B.4 (from the simulation)

and Table B.6 (the new, adjusted deﬁcits) are signiﬁcant. From

the totals for each, the average annual deﬁcit in the simulations

relative to baseline levels in Table B.4 ranges from $150 billion

to $176 billion. For the adjusted deﬁcits in Table B.6, the annual

average ranges from $68 billion to $86 billion, or about one-

half as large. Tables B.7 and B.8 present the same data as a per-

cent of nominal GDP generated in the respective simulations.

Here again, the difference is signiﬁcant, with simulation deﬁcit

impacts ranging between 0.65 and 0.75 percent of GDP, and

deﬁcit impacts relative to current levels ranging between 0.29

and 0.37 percent of GDP.

Which of the two sets of estimates is “correct”? They both

are. For the purpose of a counterfactual, as is standard for mac-

roeconometric simulations, the larger deﬁcit impacts in Tables

B.5 and B.7 are the “correct” estimates of differences from

“baseline levels” that assume no cancellation. But for the pur-

pose of estimating actual deﬁcit and national debt impacts rela-

tive to their current levels, the smaller estimates in Tables B.6

and B.8 are “correct.” In other words, in the event of a student

debt cancellation similar to the one simulated in this report, if

one assumes the results from these three simulations are reason-

able estimates, the range of increases in total government deﬁ-

cits over the next 10 years should be expected to be $675 billion

to $858 billion, not $1,127 billion to $1,340 billion.

A ﬁnal reason for considering the smaller estimates in

Tables B.6 and B.8 as the more relevant ones is that the baseline

case in the simulations assumes that the loans are paid down

in a timely manner in the absence of the cancellation. But it

is well-known that student loans have a high rate of repay-

ment difﬁculties relative to many other types of loans, and that

there are signiﬁcant costs to recouping past-due payments. For

the purposes here, the degree to which student loans become

increasingly problematic for borrowers to service in the future

reduces the relative cost of student loan cancellation now. That

is, as more difﬁculties are encountered in the future collecting

student loan payments, the baseline assumption that loans are

repaid should be revised to reﬂect this, and the cost of cancella-

tion relative to this baseline would thereby decline.

Fair Model Fair Model No Fed Moody’s No Fed

2017 89.69 86.95 111.30

2018 87.80 77.11 106.73

2019 97.73 81.24 117.29

2020 107.40 89.16 127.27

2021 46.08 38.44 62.22

2022 52.12 45.75 65.12

2023 58.36 52.43 65.76

2024 65.64 59.54 66.21

2025 74.24 67.74 67.35

2026 83.71 76.78 69.13

Totals 762.78 675.13 858.37

Average 76.28 67.51 85.84

Table B.6 Estimated Deﬁcit Impact of Student Debt

Cancellation Relative to Current Levels ($ billions)

Source: Authors’ calculations

Table B.7 Increases in Annual Budget Deﬁcits from Baseline

Forecasts in the Student Debt Cancellation Simulations as a

Percent of GDP

Fair Model Fair Model No Fed Moody’s No Fed

2017 1.11% 1.09% 1.16%

2018 0.93% 0.86% 0.82%

2019 0.94% 0.80% 0.86%

2020 0.95% 0.79% 0.94%

2021 0.67% 0.53% 0.68%

2022 0.66% 0.54% 0.68%

2023 0.62% 0.52% 0.63%

2024 0.57% 0.48% 0.60%

2025 0.53% 0.44% 0.58%

2026 0.50% 0.41% 0.58%

Average 0.75% 0.65% 0.75%

Source: Authors’ calculations

Table B.8 Estimated Deﬁcit Impact of Student Debt

Cancellation Relative to Current Levels as a Percent of GDP

Fair Model Fair Model No Fed Moody’s No Fed

2017 0.47% 0.45% 0.57%

2018 0.43% 0.38% 0.51%

2019 0.46% 0.38% 0.54%

2020 0.48% 0.40% 0.57%

2021 0.20% 0.16% 0.27%

2022 0.21% 0.19% 0.27%

2023 0.23% 0.20% 0.26%

2024 0.24% 0.22% 0.25%

2025 0.26% 0.24% 0.25%

2026 0.28% 0.25% 0.24%

Average 0.33% 0.29% 0.37%

Source: Authors’ calculations

58 Student Debt Cancellation Report 2018

Appendix C: Digression on the

Fed’s Operations

The Fed targets the federal funds rate in the federal funds mar-

ket. In the federal funds market, (mostly) banks borrow and lend

reserve balances held in their accounts at the Fed. Individual

banks use reserve balances to settle payments and to meet reserve

requirements, and their demand for reserve balances is known

to be quite interest-insensitive—banks short of reserve balances

need them and banks with unwanted excess have little use for

them. Because reserve balances in the aggregate are a liability of

the Fed, banks in the aggregate cannot affect the quantity cir-

culating with their trading in the federal funds market, which

simply moves balances from one bank to another. In order to

achieve its target rate, the Fed traditionally adjusted reserve bal-

ances on a daily basis—to offset a deﬁciency or surplus relative

to banks’ demand for them at the Fed’s target rate—through

repurchase agreements (short-term, securitized loans to securi-

ties dealers, where the collateral was usually a Treasury-issued

security). The Fed’s operations were thus “accommodative” to

banks’ demand for reserve balances.

A simple model of the federal funds market under “nor-

mal” conditions is built in the next few ﬁgures.

Figure C.1

shows a standard demand curve for reserve balances in the

federal funds market. The vertical axis is the federal funds rate

and the horizontal axis is for reserve balances. There are three

interest rates—the federal funds rate target the Fed attempts to

fedfunds

penalty

fedfunds

IOR

Figure C.1 Demand for Reserve Balances in the Federal

unds Market

Figure C.2 Supply and Demand for Reserve Balances in the

“N

ormal” Federal Funds Market

RB*

fedfunds

penalty

fedfunds

IOR

achieve in the federal funds market (i

fedfunds

*), the penalty rate the

Fed charges banks to borrow from their regional Fed bank rather

than from other banks in the federal funds market (i

penalty

), and

the rate the Fed pays banks for reserve balances held overnight

IOR

). The demand curve for reserve balances (D

) is nearly

vertical, with a bit of a downward slope, but becomes horizontal

at i

IOR

. The nearly vertical portion of D

is because banks have

a particular amount of reserve balances they desire to hold in

order to settle payments and meet reserve requirements, and

they have little interest in holding much more or less than this

amount.

The horizontal portion is because i

IOR

becomes a

price ﬂoor for the federal funds rate if the Fed provides reserve

balances beyond the amount banks reasonably would hold to

meet reserve requirements and settle payments.

Figure C.2 shows a general model of the federal funds

market with both D

from Figure C.1 and a supply curve for

reserve balances (S

). The S

schedule is vertical and then

kinks to become horizontal at i

penalty

. The vertical region repre-

sents the fact that it is the Fed that adjusts the aggregate quantity

of reserve balances—while banks can borrow/lend reserve bal-

ances among themselves, or send/receive payments in reserve

balances, these shift reserve balances around but do not change

the aggregate quantity. As all reserve accounts lie on the Fed’s

balance sheet, only an offsetting change to the Fed’s balance

sheet (such as a purchase of securities or a loan to a bank) can

alter the aggregate quantity of reserve balances. The horizontal

region of S

represents the Fed’s standing facility, also called

the “discount window” at the regional Federal Reserve Banks.

At the discount window, the Fed will lend to banks at i

penalty

Levy Economics Institute of Bard College 59

accommodate shifts in D

at the target federal funds rate) and

“offsetting” (offsetting changes to the Fed’s balance sheet incon-

sistent with the quantity of reserve balances banks desire at the

Fed’s target federal funds rate).

Since Fall 2008, the Fed has left the quantity of reserve bal-

ances well beyond any level that would be consistent with the

downward sloping portion of D

, rising from around $20 billion

prior to the ﬁnancial crisis to about $800 billion later in 2008,

and then rising again through successive rounds of quantitative

easing to its current level of around $2.5 trillion. Essentially, this

leaves the quantity of reserve balances in the lower horizontal

portion of D

consistent with i

IOR

. Consequently, the Fed has set

its target rate roughly equal to i

IOR

since late 2008, as discussed

above. Figure C.3 provides a simple illustration of the post-2008

federal funds market, where the Fed has pushed reserve balances

well beyond RB* (from Figure C.2) out to RB

, thereby set-

ting its target rate equal to i

IOR

, as noted above. Again, this is

basic supply and demand analysis—pushing a supply curve well

beyond a demand curve reduces the price either to zero or to a

price ﬂoor set to keep the price from falling to zero.

It is well-known that the Fed’s target rate since late 2008

has actually been below i

IOR

because some nonbank entities

with reserve accounts do not earn interest—such as the Federal

Home Loan Banks, for instance—and must search for an inter-

est-bearing opportunity to invest their reserve balances. Given

the large quantity of excess reserve balances circulating, and

thus the dearth of competition among banks to borrow more

of them, these nonbank entities regularly ﬁnd opportunities to

invest their reserve balances at banks only at rates below i

IOR

Consequently, prior to December 2015, while the Fed had i

IOR

Figure C.4 Corridor of Interest on Reserve Balances and

verse Repurchase Agreement Rates

IOR

penalty

fedfunds

RRP

fedfunds

which effectively places a ceiling on how high i

fedfunds

can rise.

From Figure C.2, the combination of D

and S

presents a

general picture of the federal funds market where i

penalty

and

IOR

set a ceiling and a ﬂoor, respectively, on how high or low

fedfunds

can move. The difference between these two rates is fre-

quently referred to as a “corridor” within which i

fedfunds

* would

be expected to settle (in “normal” times) as the Fed shifted the

vertical portion of S

to intersect the nearly vertical portion of

at its target rate.

It is clear from Figure C.2 that if the Fed grows reserve

balances much beyond RB*, the federal funds rate will fall to

IOR

. This would mean that the Fed had set a de facto interest

rate target at i

IOR

. Similarly, if the Fed were to reduce reserve bal-

ances, the federal funds rate would rather quickly rise to i

penalty

at which point the Fed would provide the reserve balances

through its standing facilities at the regional Fed banks until

the total quantity of reserve balances circulating equaled the

quantity banks desired to hold at i

penalty

. The Fed would then

have set a de facto target rate equal to i

penalty

The importance

of this reality—which is essentially basic supply and demand

analysis—comes from coupling it with an understanding of the

Fed’s balance sheet. Changes to the Fed’s balance sheet that are

not directly under the Fed’s control—changes in the balance of

the Treasury’s account or banks purchasing vault cash from the

Fed using balances in their reserve accounts, for instance—alter

the quantity of reserve balances circulating. Given the fairly

vertical slope of D

between i

penalty

and i

IOR

, the Fed must off-

set these changes in order to achieve its target rate. In other

words, the Fed’s operations under “normal” circumstances are

“accommodative” (altering the quantity of reserve balances to

Figure C.3 Simple Model of the Federal Funds Market with

a Large Ex

cess of Reserve Balances as a Result of

Quantitativ

e Easing

fedfunds

* = i

IOR

penalty

fedfunds

60 Student Debt Cancellation Report 2018

set at 0.25 percent, the actual level of i

fedfunds

and thus i

fedfunds

ﬂuctuated between zero and 0.25 percent, usually settling near

the midpoint between the two. When the Fed raised i

IOR

to 0.5

percent in late 2015 (from the earlier level of 0.25 percent), it

also instituted a reverse repurchase agreement (RRP) opportu-

nity. The opportunity was made available to a number of non-

bank ﬁnancial institutions at 0.25 percent—including, but not

limited to, those with reserve accounts not earning interest on

reserve balances—in order to ensure that the federal funds rate

did not fall below that level. In other words, there is another cor-

ridor, which exists between i

IOR

and i

RRP

(the rate the Fed pays to

nonbanks investing in its RRPs). Figure C.4 shows the federal

funds market with the lower corridor between i

IOR

and i

RRP

, the

range (instead of a speciﬁc rate) the Fed has targeted for i

fedfunds

The horizontal region of D

has dipped below i

IOR

(since banks

have so many excess reserve balances they will not offer i

IOR

nonbank entities investing their reserve balances) but remains

above i

RRP

(since nonbank entities can now earn i

RRP

from the

Fed if banks do not offer at least that much).

As the Fed con-

tinues to raise the federal funds rate target thereafter, it does so

by simply announcing increases in both i

IOR

and i

RRP

. Overall, the

earlier analysis applies: if the Fed is going to oversupply the bank-

ing system with large quantities of excess reserve balances beyond

the levels banks would desire to hold at i

fedfunds

*, it will have to pay

interest on reserve balances, interest on RRPs, or interest on some

other liability it might issue (such as issuing its own securities or

time deposits) in order to achieve a positive i

fedfunds

Levy Economics Institute of Bard College 61

Notes

1. The authors wish to express their appreciation to Joseph

Ballegeer for research assistance and to Mark Zandi, Chris

Lafakis, and Moody’s for generously sharing their time and

expertise with the Moody’s US Macroeconomic Model. We

also wish to thank Mary Green Swig and Steven L. Swig for

generous research support.

2. The simulation results summarized here incorporate two

Fair model simulations (with and without the Federal

Reserve’s interest rate target rule in effect) and one Moody’s

model simulation (without the Federal Reserve’s interest

rate target rule in effect). The exception is in the discussion

of the effect on interest rates, in which the largest effects

reported are from the Moody’s simulation with the Federal

Reserve’s interest rate target in effect. The simulation sec-

tion of this report (Section 3) discusses the rationale for

viewing these three as the most representative of the simu-

lation results.

3. For racial wealth and income gaps, see Emmons and Neoth

(2015).

4. The authors adjust household income and labor earnings

quintiles for the age distribution of student loan borrowers.

5. This calculation is based on the reported thresholds for

the quintiles. The greater the degree of inequality within

the top quintile, the more the 50 percent estimate from

thresholds underestimates the true income share of the top

quintile. Note that the World Wealth and Incomes Database

(WWID) computes that the income share of the top decile

was approximately 50 percent in 2015. The universe of all

tax units reported in WWID is richer than just the age-

adjusted sample used by Looney and Yannelis, since those

with student debt tend to be younger than the population

as a whole, but nonetheless, it is likely that the top quintile

of households even in the Looney and Yannelis data earns

more than 50 percent of total income.

6 For an overview, see Goldin and Katz (2010).

7. Accessed on July 20, 2016 from https://studentaid.ed.gov/

sa/about/data-center/student/portfolio

8. See, for instance, GAO (2014, 2001).

9. The Fed annually credits its proﬁts to the Treasury’s account

after paying dividends on member banks’ investment

into Federal Reserve capital (“paid-in capital”) and any

investments into its own “surplus” capital account. As of

December 2015, the Fed now pays a dividend of 6 per-

cent to those member institutions with less than $10 bil-

lion in consolidated assets and the lesser of 6 percent or

the high yield from the most recent 10-year Treasury note

auction, all as prescribed in the Fixing America’s Surface

Transportation Act (FAST Act). Prior to the FAST Act, the

Fed paid a 6 percent dividend on invested capital to all

member institutions based on their investment. Member

institutions’ investments are required by the Federal

Reserve Act; a member bank’s investment is based upon the

size of its own capital, and thus is regularly increased or

decreased in kind. Prior to the FAST Act, the Fed’s surplus

capital (that is, retained earnings) was maintained at a level

equal to its paid-in capital from member banks. However,

the FAST Act limited the Fed’s paid-in capital to $10 billion.

This resulted in a transfer from the Fed’s capital account to

the Treasury’s account of $19.3 billion in December 2015

in order to reduce its surplus capital to the legally man-

dated level. See Board of Governors of the Federal Reserve

System (2016). For more on the Fed’s surplus capital, see

Goodfriend (2014).

10. See Board of Governors of the Federal Reserve System

(2016) for more information on the Fed’s remittances to

the Treasury, which, according to the press release, have

totaled $679.4 billion since 2006.

11. As with the earlier analysis of the federal government pur-

chasing the private loans, the analysis changes very little if

it is assumed the Fed pays a higher market price such that

the investors receive a capital gain, but there are additional

steps for explaining that scenario that unnecessarily com-

plicate an already detailed discussion.

12. The analysis in this section relies on the insights of a forth-

coming working paper by Raúl Carrillo, Rohan Grey, Robert

Hockett, and Nathan Tankus. The draft version is entitled

“The Legality of Student Loan Purchase and Forgiveness by

the Federal Reserve.”

13. Simulations using the Moody’s model were facilitated by

Chris Lafakis and Mark Zandi at Moody’s.

14. Ray Fair’s website is https://fairmodel.econ.yale.edu. See

also Fair (2015, 2013, 2004).

15. See Fair (2004, 2013) for a full description of statistical tests

applied to the Fair model.

62 Student Debt Cancellation Report 2018

16. See Fair (2000). A draft version was published earlier on

his website, along with an interactive page that enabled the

user to determine how overvalued the market was.

17. As the sections on the mechanics of the student debt can-

cellation demonstrate using T-accounts, there is no double

counting involved in incorporating both an instantaneous

net wealth effect from cancellation and an income effect

over the following 10 years from cancelled debt service.

Both would actually occur.

18. The 10 percent assumption is an assumed average from ﬁg-

ures shown on pp. 12–14 in CFPB (2012).

19. The baseline used for the Moody’s model is a bit more com-

plicated. As noted above, the Moody’s model was used for

simulating the economic proposals of the two candidates

for US President in 2016. For the simulation of Secretary

Clinton’s proposals, this included her “College Compact”

proposal for reduced college tuition expenses. For the

Moody’s simulations in this report, the baseline reported is

the normal baseline forecast for the Moody’s model during

2017–26 with Secretary Clinton’s “College Compact” pro-

posal added to it. In other words, one might consider the

standard Moody’s baseline forecast as a “baseline 1” and a

“baseline 2,” which is equal to “baseline 1” plus the Clinton

“Campus Compact” proposal. The Moody’s simulations in

this report use “baseline 2” as their baseline. Because the

Moody’s model is structural, there is no difference between

simulating the student debt cancellation starting from

“baseline 1” or “baseline 2.” The differences between either

baseline and the subsequent addition of the student debt

cancellation will be the same. Consequently, the appropri-

ate interpretation of the results from either model is the

macroeconomic impact of an initial increase in household

net ﬁnancial wealth that falls slowly over time, as in Table

A.1, combined with reduced revenue and increased outlays

by the federal government shown in Tables A.2, A.3, and

A.4. (Secretary Clinton’s proposed “College Compact” is

explained here: https://www.hillaryclinton.com/brieﬁng/

factsheets/2015/08/10/college-compact-costs/)

20. Both models currently use 2009 as the base year for inﬂa-

tion calculations—the year in which nominal and real GDP

are equal, and thus the price level using the GDP deﬂator

is set to 1. Using 2016 as the base year is more intuitive

here, particularly given simulations through 2026. The

method for converting a 2009 base year to a 2016 base year

is straightforward: divide the GDP deﬂator in each year of

the simulation period (2017–26) by the respective model’s

GDP deﬂator in 2016. The GDP deﬂator at the end of 2016

is now 1; the GDP deﬂator for years 2017–26 is now com-

puted for a 2016 base year. Then, divide nominal GDP for

all simulations by the respective GDP deﬂator with a 2016

base year.

21. The nominal GDP ﬁgures for the 2017–26 simulation

period are $2,214 billion for the Fair model, $2,516 for the

Fair model without the Fed, $940 billion for the Moody’s

model, and $1,735 billion for the Moody’s model without

the Fed.

22. Both models also have separate indexes for housing prices.

Results for housing price inﬂation were not included in the

discussion because the impact of the student debt cancel-

lation on housing prices was essentially negligible in both

models.

23. Zandi et al. (2016) report that the less-positive macro-

economic results from their simulation of Mr. Trump’s

economic proposals are strongly affected by the fact that

the national debt-to-GDP ratio rises substantially, thus

raising long-term interest rates and slowing the economy.

Interestingly, Peter Navarro, an economics professor at the

University of California-Irvine criticized Moody’s report

as “Keynesian” (see Cox 2016) even though the national

debt-to-GDP ratio effects on interest rates are not part

of traditional Keynesian economics but rather part of the

“Classical” long-run “core” of the Moody’s model.

24. In terms of the econometrics, Moody’s reports an

R-squared of 0.98 for its estimation of the 10-year Treasury

note that includes the national debt-to-GDP ratio (Zandi

et al. 2016, 11), whereas Fair obtains an essentially identical

ﬁt (R-square = 0.97) without including the national debt-

to-GDP ratio as an explanatory variable.

25. The authors offer their thanks to one of the reviewers for

pointing this out.

26. The authors wish to thank a reviewer for pointing out the

offsetting effect of improving state budgets on the consoli-

dated federal- and state-level budget positions.

27. “Normal” here is taken to mean a corridor system as

depicted in Figure C.2, which has become a standard

approach to modeling central bank operations in inter-

bank markets in the literatures published by central banks

and in academic journals. In reality, the Fed did not have

Levy Economics Institute of Bard College 63

the authority to pay interest on reserve balances until

after the failure of Lehman in the fall of 2008. In that case,

the demand for reserve balances in Figure C.2 would not

have a ﬂat portion at the interest on reserve balances and

would instead continue downward to the horizontal axis.

The Fed’s plan, though, is that eventually there will be a

return to “normal” that will look like the graph in Figure

C.2, which is also illustrative of the approach of many other

central banks prior to 2008.

28. Reserve requirements are typically represented via a more

horizontal portion of the demand for reserve balances at

RB*. Because banks can meet reserve requirements over a

two-week maintenance period, they can trade off balances

held across days and on any given day do not necessarily

have a vertical demand curve at the target rate. By the end

of the maintenance period, however, a more vertical curve

like that in Figure C.1 is more applicable, since at that point

banks have far less ability to offset surpluses or excesses in

meeting reserve requirements.

29. In practice, due to a number of factors beyond the scope of

the analysis here, it is possible for i

fedfunds

to rise above i

penalty

before banks turn to the discount window to relieve a deﬁ-

ciency in reserve balances.

30. The Fed’s abilities to reduce reserve balances are limited

by the need to support the payments system and reserve

requirements—consistent with the text, not providing suf-

ﬁcient reserve balances for both sources of demand simply

results in a de facto interest rate target equal to i

penalty

as the

necessary reserve balances are supplied at that rate as banks

borrow from their regional Fed banks.

31. Figure C.4 is consistent with publications by Federal

Reserve researchers. See, for instance, Ihrig, Meade, and

Weinbach (2015).

64 Student Debt Cancellation Report 2018

References

Addo, F. R. 2014. “Debt, Cohabitation, and Marriage in Young

Adulthood.” Demography 51(5): 1677–701.

Akers, E. J., and M. M. Chingos. 2014. “Student Loan Update:

A First Look at the 2013 Survey of Consumer Finances.”

Brookings Institution, Brown Center on Education Policy.

Washington, DC: Brookings Institution.

Albuquerque, B., and G. Krustev. 2015. “Debt Overhang and

Deleveraging in the US Household Sector: Gauging

the Impact on Consumption.” European Central Bank

Working Paper No. 1843. Frankfurt am Main: European

Central Bank.

Ambrose, B. W., L. Cordell, and S. Ma. 2015. “The Impact of

Student Loan Debt on Small Business Formation.” Federal

Reserve Bank of Philadelphia Working Paper No. 15-26,

July. Philadelphia: Federal Reserve Bank of Philadelphia.

Baum, S. 2016. Student Debt: Rhetoric and Realities of

Higher Education Financing. Basingstoke, UK: Palgrave

MacMillan.

Bettinger, E. 2004. “How Financial Aid Affects Persistence.”

In C. M. Hoxby (ed.), College Choices: The Economics of

Where to Go, When to Go, and How to Pay for It. Chicago:

NBER and University of Chicago Press.

Blanchard, O., and D. Leigh. 2013. “Growth Forecast Errors

and Fiscal Multipliers.” International Monetary Fund

Working Paper No. 13/1. Washington, DC: International

Monetary Fund.

Board of Governors of the Federal Reserve System. 2016.

“Federal Reserve Board announces Reserve Bank income

and expense data and transfers to the Treasury for 2015.”

Press release, January 11.

Bozick, R., and A. Estacion. 2014. “Do student loans delay mar-

riage? Debt repayment and family formation in young

adulthood.” Demographic Research 30(69): 1865–91.

Brown, M., and S. Caldwell. 2013. “Student Loan Borrowers

Retreat From Housing and Auto Markets.” The Federal

Reserve Bank of New York, Liberty Street Economics. April 17.

Brown, M., A. Haughwout, D. Lee, J. Scally, and W. van

der Klaauw. 2014. “Measuring Student Debt and Its

Performance.” Federal Reserve Bank of New York Staff

Report No. 668. New York: Federal Reserve Bank of

New York.

Carrillo, R., R. Grey, N. Tankus, and R. Hockett. Forthcoming.

“The Legality of Student Loan Purchase and Forgiveness

by the Federal Reserve.” Clarke Business Law Institute,

Program on the Law and Regulation of Financial

Institutions and Markets. Cornell University.

CFPB (Consumer Financial Protection Bureau). 2013.

“Student Loan Affordability, Analysis of Public Input on

Impact and Solutions.” Online report, May 8.

http://ﬁles.consumerﬁnance.gov/f/201305_cfpb_rﬁ-report_

student-loans.pdf.

———. 2012. “Private Student Loans.” Report to the Senate

Committee on Banking, Housing, and Urban Affairs,

the Senate Committee on Health, Education, Labor, and

Pensions, the House of Representatives Committee on

Financial Services, and the House of Representatives

Committee on Education and the Workforce. August 29.

http://ﬁles.consumerﬁnance.gov/f/201207_cfpb_Reports_

Private-Student-Loans.pdf.

Chakrabarti, R., M. Mabutas, and B. Zafar. 2012. “Soaring

Tuitions: Are Public Funding Cuts to Blame?” Federal

Reserve Bank of New York, Liberty Street Economics.

September 19.

Clark, B., C. Joubert, and A. Maurel. 2014. “The Career

Prospects of Overeducated Americans.” IZA Discussion

Paper No. 8313. Bonn: Institute of Labor Economics.

Cox, J. 2016. “Trump’s Policies Will Not Cause Recession—

Economist.” CNBC.com, June 27.

Darolia, R., and D. Ritter. 2015. “Do Student Loan Borrowers

Opportunistically Default? Evidence from Bankruptcy

Reform.” Federal Reserve Bank of Philadelphia Working

Paper 15-17R. Philadelphia: Federal Reserve Bank of

Philadelphia.

Dwyer, R. E., L. McCloud, and R. Hodson. 2013. “Gender,

Debt, and Dropping Out of College.” Gender & Society

27(1): 30–55.

———. 2012. “Debt and Graduation from American

Universities.” Social Forces 90(4): 1133–55.

Dynarski, S. M. 2016. “The trouble with student loans?

Low earnings, not high debt.” Evidence Speaks Report.

Washington, DC: Brookings Institution.

———. 2003. “Does Aid Matter?” American Economic Review

93(1): 279–88.

Edmiston, K. D., L. Brooks, and S. Shepelwich. 2013. “Student

Loans: Overview and Issues (Update).” Federal Reserve

Bank of Kansas City Research Working Paper 12-05.

Kansas City: Federal Reserve Bank of Kansas City.

Levy Economics Institute of Bard College 65

Eichengreen, B., and K. O’Rourke. 2012. “Gauging the

Multiplier—Lessons from History.” Vox EU, October 23.

Elliot, W., M. Grintstein-Weiss, and I. Nam. 2013. “Is Student

Debt Compromising Homeownership as a Wealth

Building Tool?” Center for Social Development Working

Paper No. 13-33. St. Louis, MO: Washington University in

St. Louis.

Elliott, W., and I. Nam. 2013. “Is Student Debt Jeopardizing the

Short-Term Financial Health of U.S. Households?” Federal

Reserve Bank of St. Louis, Review 95(5): 405–24.

Emmons, W. R., and B. J. Neoth. 2015. “Why didn’t higher

education protect Hispanic and black wealth?” Federal

Reserve Bank of St. Louis, In the Balance, Issue 12-2015.

Fair, R. 2015. “Reﬂections on Macroeconometric Modeling.”

The B. E. Journal of Macroeconomics 15(1): 445–66.

———. 2013. “Macroeconometric Modeling.” Unpublished

manuscript, November 11. Available at: https://fairmodel.

econ.yale.edu/mmm2/mm.pdf.

———. 2007. “Evaluating Inﬂation Targeting Using a

Macroeconometric Model.” Economics: The Open-Access,

Open-Assessment E-Journal 1(2007–8): 1–52.

———. 2004. Estimating How the Macroeconomy Works.

Cambridge, MA: Harvard University Press.

———. 2000. “Fed Policy and the Effects of a Stock Market

Crash on the Economy.” Business Economics 35(2): 7–14.

———. 1992. “The Cowles Commission Approach, Real

Business Cycle Theories, and New Keynesian Economics.”

In M. T. Belongia and M. R. Garﬁnkel (eds.), The Business

Cycle: Theories and Evidence—Proceedings of the 16th

Annual Economic Policy Conference of the Federal Reserve

Bank of St. Louis. Norwell, MA: Kluwer Academic

Publishers.

Federal Reserve. 2016. “Consumer Credit, G.19 Statistical

Release.” Released April 2016.

GAO (Government Accountability Ofﬁce). 2014. “Federal

Student Loans: Borrower Interest Rates Cannot Be

Set in Advance to Precisely and Consistently Balance

Federal Revenues and Costs.” Report to Congressional

Committees. GAO-14-234. Washington, DC. January.

———. 2001. “Department of Education: Key Aspects of the

Federal Direct Loan Program’s Cost Estimates.” Report to

Congressional Requesters. GAO-01-197. Washington, DC.

January.

Goldin, C., and L. F. Katz. 2010. The Race between Education

and Technology. Cambridge, MA: Harvard University

Press.

Goodfriend, M. 2014. “The Relevance of Federal Reserve

Surplus Capital for Current Policy.” Economics 21,

Manhattan Institute for Policy Research, March 17.

Huelsman, M. 2015a. “The Debt Divide: The Racial and Class

Bias Behind the ‘New Normal’ of Student Borrowing.”

Online report. New York: Demos.

———. 2015b. “The Case for Debt-Free College.” Online

report. New York: Demos.

Ihrig, J. E., E. E. Meade, and G. C. Weinbach. 2015. “Monetary

Policy 101—A Primer on the Fed’s Changing Approach

to Policy Implementation.” Finance and Economics

Discussion Series 2015-047. Washington, DC: Board of

Governors of the Federal Reserve System.

Li, W. 2013. “The Economics of Student Loan Borrowing

and Repayment.” Federal Reserve Bank of Philadelphia

Business Review Q3.

Looney, A., and C. Yannelis. 2015. “A crisis in student loans?

How changes in the characteristics of borrowers and

in the institutions they attended contributed to rising

loan defaults.” Brookings Papers on Economic Activity.

Washington, DC: Brookings Institution.

McMillan Cottom, T. 2017. Lower Ed: The Troubling Rise of

For-Proﬁt Colleges in the New Economy. New York, NY: The

New Press.

MeasureOne. 2015. “The MeasureOne Q3 2015 Private Student

Loan Report.” Accessed June 12, 2016.

Mezza, A., D. R. Ringo, S. M. Sherlund, and K. Sommer.

2016. “On the Effect of Student Loans on Access to

Homeownership.” Finance and Economics Discussion

Series 2016-010. Washington, DC: Board of Governors of

the Federal Reserve System.

Mueller, H. M., and C. Yannelis. 2017. “Students in Distress:

Labor Market Shocks, Student Loan Default, and Federal

Insurance Programs.” NBER Working Paper No. 23284.

Cambridge, MA: National Bureau of Economic Research.

NCES (National Center on Educational Statistics). 2015.

“Digest of Education Statistics, 2015.” US Department

of Education. https://nces.ed.gov/programs/

digest/2015menu_tables.asp.

66 Student Debt Cancellation Report 2018

Vandenbroucke, G. 2015. “Lifetime Beneﬁts of an Education

Have Never Been So High.” Federal Reserve Bank of

St. Louis, The Regional Economist July: 10–11.

Weissmann, J. 2016. “Forgiving All Student Loan Debt Would

Be an Awful, Regressive Idea.” Slate, August 15.

Woo, J., A. Bentz, S. Lew, N. Smith, and E. D. Velez. 2017.

“Repayment of Student Loans as of 2015 among 1995–96

and 2003–04 First-Time Beginning Students.” First Look,

NCES 2018-410. Washington, DC: National Center for

Education Statistics.

Zahn, M. 2014. “The Impact of Youth Debt on College

Graduation.” Journal of Sociology and Social Welfare 41(3):

133–56.

Zandi, M. 2008. “Assessing the Macroeconomic Impact of

Fiscal Stimulus 2008.” Moody’s Economy.com.

Zandi, M., and S. Hoyt. 2015. “US Macro Model Methodology.”

Moody’s Analytics—Economic and Consumer Credit

Analytics. April.

Zandi, M., C. Lafakis, and A. Ozimek. 2016. “The

Macroeconomic Consequences of Secretary Clinton’s

Economic Policies.” Moody’s Analytics. July.

Zandi, M., C. Lafakis, D. White, and A. Ozimek. 2016. “The

Macroeconomic Consequences of Mr. Trump’s Economic

Policies.” Moody’s Analytics. June.

———. 2014. “Digest of Education Statistics, 2014.” US

Department of Education. https://nces.ed.gov/programs/

digest/2014menu_tables.asp.

———. 2013. “2011–12 National Postsecondary Student

Aid Study (NPSAS: 12).” US Department of Education.

Computation by NCES QuickStats on April 13, 2016.

OECD (Organisation for Economic Co-operation and

Development). 2015. “Education at a Glance, Tables A7.4a

and A7.4b.” OECD Indicators. Paris: OECD.

Schwartz, J. 2005. “The Socio-Economic Beneﬁts of Marriage:

A Review of Recent Evidence from the United States.”

Economics Affairs 25(3): 45–51.

SHEEO (State Higher Education Executive Ofﬁcers

Association). 2015. “SHEF: FY 2015, State Higher

Education Finance.” Boulder, CO: State Higher Education

Executive Ofﬁcers Association.

Steinbaum, M., and K. Vaghul. 2016. “How the student debt

crisis affects African Americans and Latinos.” Online

report. Washington, DC: Washington Center for Equitable

Growth.

———. 2015. “An introduction to the geography of student

debt.” Washington, DC: Washington Center for Equitable

Growth.

———. Forthcoming. “The Student Loan Crisis is a Labor

Market Crisis.”

US Census Bureau. CPS Historical Time Series Tables, Table

A-2. “Percent of People 25 Years and Over Who Have

Completed High School or College, by Race, Hispanic

Origin and Sex: Selected Years 1940 to 2015.” Accessed on

June 11, 2016. https://www.census.gov/data/tables/

time-series/demo/educational-attainment/cps-historical-

time-series.html.

———. Historical Income Tables: Households, Table H-9.

“Race of Head of Household by Median and Mean

Income.” Accessed on June 11, 2016. https://www.census.

gov/data/tables/time-series/demo/income-poverty/

historical-income-households.html.

US Department of Education. 2017. “Comparison of FY 2014

Ofﬁcial National Cohort Default Rates to Prior Two

Ofﬁcial Cohort Default Rates, Calculated August 5, 2017.”

https://www2.ed.gov/ofﬁces/OSFAP/defaultmanagement/

schooltyperates.pdf.

———. 2016. “Understanding Delinquency and Default.”

Accessed June 12. https://studentaid.ed.gov/sa/

repay-loans/default.

Levy Economics Institute of Bard College 67

of Bard College

Levy Economics

Institute

The Levy Economics Institute of Bard College,

founded in 1986, is a nonproﬁt, nonpartisan,

research organization

devoted to public service.

Through scholarship and research

it generates viable,

effective public policy responses to

important economic problems

that profoundly affect the quality of life

in the United States and abroad.

levy economics institute of bard college

Blithewood

Annandale-on-Hudson, NY 12504-5000

845-758-7700

E-mail: info@lev y.org

levyinstitute.org

68 Student Debt Cancellation Report 2018