View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Home / Publications / Research / Economic Brief / 2021

Economic Brief
March 2021, No. 21-07

Incarceration's Life-Long Impact on Earnings and
Article by: Grey Gordon and Urvi Neelakantan

We estimate the impact of incarceration on individuals' earnings and
employment prospects using a statistical model that controls for race, gender,
education and other factors. The model reveals that rst-time incarceration for
black men with a high school diploma reduces expected lifetime earnings by 33
percent and employment by 22 percent. For high school educated white men, it
reduces expected lifetime earnings by 43 percent and employment by 27

Incarceration has surged in the United States over the past few decades. As seen in Figure
1, from 1978 through 2019, the number of inmates in state or federal correctional facilities
increased from fewer than 300,000 to more than 1.2 million, and the incarceration rate rose
from 0.13 percent to 0.42 percent of the population. This increase has had disparate e ects
on the U.S. population because incarceration rates di er substantially by gender, race and
education. For example, the incarcerated population is overwhelmingly male (93 percent)
and less educated.1 Moreover, the imprisonment rate for black men (2.2 percent) is nearly
six times higher than for white men.2 Given the high and uneven incidence of incarceration,
it is important to understand the impact of incarceration on earnings and employment over
a person's lifetime.

Yet surprisingly little is known about the impact of incarceration on employment and
earnings, particularly across demographic groups. This is due partly to lack of data: Most
household surveys are limited to the noninstitutionalized civilian population. In their work
on the prison boom in the United States, economists Derek Neal of the University of
Chicago and Armin Rick of Cornell University call for more research on the e ects of
incarceration on "the employment and earnings prospects of less-skilled men, and lessskilled black men in particular."3 New research by the authors of this brief, along with
Richmond Fed colleagues John Bailey Jones and Kartik Athreya, attempts to meet this


We construct a statistical model of earnings that accounts for incarceration risk in addition
to other forms of nonemployment risk, such as unemployment and disability.5 We nd that
the impact of incarceration on earnings is enormous, particularly for men with a high
school diploma or less. One key nding is that rst-time incarceration reduces the expected
lifetime earnings of high school graduates by 33 percent for black men and 43 percent for
white men. (The larger percent decline for white men re ects the fact that white high school
graduates have higher earnings than black high school graduates on average.) This massive
reduction in lifetime earning is largely due to a reduction in average working years — 7.6
fewer for black men and 9.4 fewer for white men.

Our primary data source is the National Longitudinal Survey of Youth that began in 1979,
one of the few household panel data sets that reports incarceration. This data set allows us
to follow a group of individuals who were youths in 1979 over time and observe their
earnings, employment and incarceration status.
Our statistical model describes how likely a person is to be incarcerated, jobless or
employed (and, if employed, how much they earn). This likelihood depends on a number of
factors, such as education, race, gender, current earnings, employment and incarceration
history.6 The model also distinguishes between temporary and persistent joblessness and
earnings shocks. For example, if an individual were unemployed for a year and then found
a job with a lower salary, the model generally would predict a temporary jobless spell
followed by a persistent reduction in earnings. Alternatively, if a person became disabled
later in life, then the model likely would predict persistent joblessness. Additionally, the
model allows for asymmetry, such as an increase in earnings being more persistent than a
decrease in earnings. The model also accounts for the fact that incarcerated people are
more likely to be missing from the data.
We estimate the model, a process that aligns the predictions of the model with actual data
as closely as possible. We use the estimated model to create an arti cial data set that
contains a huge number of observations, has no missing data, and distinguishes between
persistent and transitory earnings and nonemployment shocks. Consequently, if we saw an
individual make $25,000 in 1993, go to jail in 1994 and make $15,000 in 1995, for example,
we could attribute $8,000 of that decrease to a persistent drop in earnings and $2,000 to a
temporary drop. More importantly, we could predict the hypothetical individual's entire
future employment path.



We show that the implications of incarceration are profound. For instance, a typical 25-yearold black high school graduate entering jail for the rst time will su er a lifetime income
loss of $137,000 in 1982–84 dollars, a 33 percent drop.7 A typical 25-year-old white high
school graduate entering jail for the rst time will su er a lifetime income loss of $400,000,
a 43 percent drop. For those without a high school diploma, the losses amount to $129,000
and $261,000, respectively, resulting in 50 percent drops for both black and white men.
These large e ects follow from the fact that the drop in earnings experienced after prison
will last a lifetime.
One way to see the e ects of incarceration is by comparing two groups of individuals who
are exactly the same in the model until a point in time, say 1995, when their paths diverge.
The rst group — the treated group — goes to prison in 1996, while the second group —
the control group — stays free. The di erence between the rst group and the second
group from 1996 on reveals the average e ect of incarceration, as seen in Figure 2.


Figure 2 shows the e ects of rst-time incarceration on black and white men with less than
a high school diploma. The top row illustrates the e ect on earnings, and the bottom row
illustrates the e ect on the jobless rate. The blue lines in each of the panels start at zero,
which indicates that the incarcerated and nonincarcerated are identical (within their race
and education group) up to the point at which the incarceration shock occurs.
The top left panel reveals an initial $6,000 annual earnings loss for a black man from going
to jail or prison. A few years after the shock, this annual loss decreases but not by much.
Even 15 years after the shock, the di erence between the earnings of the incarcerated
group and the control group still exceeds $4,000 a year. Consequently, incarceration
translates into a substantial loss in lifetime earnings.
Turning now to the e ect of incarceration on joblessness, the bottom left panel of Figure 2
shows that among black men, the joblessness rate is more than 25 percent higher for a
man who has been incarcerated, compared with an otherwise identical man. This gap
narrows over the next few years but is statistically signi cant even 10 years later.8
Consequently, incarceration leads not only to lower earnings, but also to persistently lower
employment rates later in life.
Incarceration also has persistently negative e ects on white men. For example, pretax
annual earnings (shown in the top right panel of Figure 2) drop by around $8,000 a year,
seemingly for life. Similarly, jobless rates (in the bottom right panel of the gure) increase
by around 10 percentage points.
While Figure 2 illustrates the e ects of rst-time incarceration, both on impact and over
time, it only considers two groups, black men and white men without high school diplomas
who had never before been incarcerated. We summarize the e ects of incarceration for
more groups of individuals in Table 1, using the same comparison of treatment and control
groups. We report the e ects separately for black and white men (B/W), for those with a
high school degree and those with less than a high school degree (H/L) and for those with
an incarceration record (r) and those without.


For all the groups, the e ects of rst-time incarceration on lifetime earnings are large,
typically ranging from $100,000 (around $267,000 in today's dollars) to $400,000 (around $1
million today). There is also a strong link between becoming incarcerated and joblessness;
those who have experienced incarceration spend an extra 0.8 to 4.2 years without jobs.
Additionally, rst-time incarceration increases the number of years people from each group
will spend in jail or prison (accomplished mostly through jail stints later in life).

The United States has one of the highest rates of incarceration in the world, a rate that has
risen sharply over time. For instance, 15 percent of black men who were born in the late
1940s and did not graduate from high school had been incarcerated by the time they
reached their early 30s. That percentage increased to 68 percent for black men who were
born in the late 1970s and did not graduate from high school. For white men with the same
characteristics, the percentage increased from 4 percent to 28 percent during those 30
years.9 With such a large fraction of the population, particularly less-educated men, having
experienced incarceration, it is important to understand its long-term consequences. Our
research contributes to the understanding of the e ects of incarceration by using a
statistical model of earnings that incorporates the joint movement of earnings, joblessness
and incarceration. The model examines di erences by race, gender and educational
attainment. While it is well known that incarceration disproportionately a ects lesseducated men, our research provides a quantitative measure of the size and persistence of
those e ects. We nd that incarceration is followed by an essentially permanent drop in
earnings that can cause lifetime earnings to fall by up to 50 percent. Incarceration also
leads to an increase in the number of years spent unemployed or out of the labor force by
as much as four years. Clearly, incarceration has large e ects that stay with individuals long
after they serve their time.

Grey Gordon is a senior economist and Urvi Neelakantan is a senior policy economist in the
Research Department at the Federal Reserve Bank of Richmond.


Bureau of Justice Statistics, "Surveys of Inmates in State and Federal Correctional Facilities,


Data are from the Bureau of Justice Statistics, "Number of Sentenced State and Federal
Prisoners per 100,000 U.S. Residents of Corresponding Sex, Race, Hispanic Origin, and Age
Groups, December 31, 2019." We used the corrections statistical analysis tool at

Derek Neal and Armin Rick, "The Prison Boom and the Lack of Black Progress after Smith and
Welch," National Bureau of Economic Research Working Paper No. 20283, July 2014.

Grey Gordon, John Bailey Jones, Urvi Neelakantan and Kartik Athreya, "Incarceration, Earnings,
and Race," Federal Reserve Bank of Richmond, Manuscript, March 2021.

One of the few other papers to estimate the e ect of incarceration on earnings is Elizabeth

Caucutt, Nezih Guner and Christopher Rauh, "Is Marriage for White People? Incarceration,
Unemployment, and the Racial Marriage Divide," Center for Economic and Policy Research
Discussion Paper No. 13275, October 2018. They estimate an earnings process for their structural
model in which they assume that ex-convicts transition into either unemployment or the lowest
possible positive earnings state. Additionally, they assume that the probability that an individual
becomes incarcerated in the future depends only on whether the individual is incarcerated at
present. While the researchers' model is not structural, it allows for more exibility in transitions
and incarceration.

For the sake of brevity, this summary of our research focuses on men, who constitute 93
percent of the prison population.

Unless otherwise indicated, all dollar amounts in this brief are expressed in 1982–84 dollars.


When the dashed lines above and below the blue line do not encompass zero, the e ect is
statistically signi cant. This means that, with a certain degree of con dence, the average e ect is
not zero.

See Table 1 on page 11 of Bruce Western and Becky Pettit, "Incarceration and Social Inequality,"
Daedalus: Journal of the American Academy of Arts and Sciences, Summer 2010, vol. 139, no. 3,
pp. 8–19.

This article may be photocopied or reprinted in its entirety. Please credit the authors,
source, and the Federal Reserve Bank of Richmond and include the italicized statement
Views expressed in this article are those of the authors and not necessarily those of the Federal
Reserve Bank of Richmond or the Federal Reserve System.

Subscribe to Economic Brief
Receive an email noti cation when Economic Brief is posted online:
Email address


Contact Us
Karl Rhodes
(804) 697-8144

© 1997-2021 Federal Reserve Bank of Richmond