Did the Great Depression affect Educational Attainment in the US?
ABSTRACT Great Depression is an example of a macroeconomic crisis that produced adverse economic and social effects in all spheres of life. Theoretical arguments about the real effects of the Great Depression on education vary. First is economic hardships, which might force individuals eligible to go to school to work instead. Second is that high unemployment would make going to school the best other viable alternative. Following these theoretical notions, this paper explores the impact of the Great Depression on education, on race (whites and blacks) and gender (males and females), during the period 1930-1940. Furthermore, I test the effects of state employment indices on education. The results (using 1960 census data) show some evidence that education of whites born between 1911 and 1915 was affected. However, there is no evidence that the variation in state employment indices affected the decision of schooling on the average (mean).
- SourceAvailable from: SSRN[show abstract] [hide abstract]
ABSTRACT: The authors establish that season of birth is related to educational attainment because of school start age policy and compulsory school attendance laws. Individuals born in the beginning of the year start school at an older age, and can therefore drop out after completing less schooling than individuals born near the end of the year. Roughly 25 percent of potential dropouts remain in school because of compulsory schooling laws. They estimate the impact of compulsory schooling on earnings by using quarter of birth as an instrument for education. The instrumental valuables estimate of the return to education is close to the ordinary least squares estimate, suggesting that there is little bias in conventional estimates. Copyright 1991, the President and Fellows of Harvard College and the Massachusetts Institute of Technology.The Quarterly Journal of Economics. 02/1991; 106(4):979-1014.
- Brookings Papers on Economic Activity. 02/1973; 4(1):67-132.
Electronic copy available at: http://ssrn.com/abstract=1907942
Did the Great Depression affect Educational Attainment in
Great Depression is an example of a macroeconomic crisis that produced adverse economic
and social effects in all spheres of life. Theoretical arguments about the real effects of the
Great Depression on education vary. First is economic hardships, which might force
individuals eligible to go to school to work instead. Second is that high unemployment would
make going to school the best other viable alternative. Following these theoretical notions,
this paper explores the impact of the Great Depression on education, on race (whites and
blacks) and gender (males and females), during the period 1930-1940. Furthermore, I test the
effects of state employment indices on education. The results (using 1960 census data) show
some evidence that education of whites born between 1911 and 1915 was affected. However,
there is no evidence that the variation in state employment indices affected the decision of
schooling on the average (mean).
The author would like to thank the anonymous referees of this journal for constructive comments on previous version of this
paper; however, the author is solely responsible for any shortcomings.
Citation: kisswani, khalid, (2008) "Did the Great Depression affect Educational Attainment in the US?." Economics
Bulletin, Vol. 9, No. 30 pp. 1-10
Submitted: October 25, 2008. Accepted: December 19, 2008.
Electronic copy available at: http://ssrn.com/abstract=1907942
Macroeconomic crises have adverse effects on various aspects of social and economic
life, generating and profoundly impacting macroeconomic outcomes such as production
and jobs. The Great Depression is a prime example of a macroeconomic crisis that
produced adverse economic and social effects in all spheres of life. Kirkwood (1972)
argues that during the Great Depression the fermenting of a macroeconomic crisis was
characterized by many factors. Gross National Product (GNP) declined by 30% in real
terms, farm income fell by 50 %, which further increased poverty, businesses closed,
factories shut down, and banks ultimately failed. Kirkwood (1972) further shows that
employment was affected severely. In 1929 only 3.2% of the population was
unemployed, whereas in 1933 the unemployment reached as high as 25%. Moreover, net
investment during the Great Depression era was negative, because private investment was
far below depreciation. Thus, most effects of the Great Depression are well known and
fairly cited. However, an important variable that has been ignored and has a bearing on
our understanding the impact of macroeconomic crises is educational attainment.
The theoretical arguments about the real effects of the Great Depression on education
vary. The first is that of economic hardships, which invariably causes the cost of
education to become unaffordable. This might force individuals eligible to go to school
to work for their sustenance. As a result, high school students would drop out, or would
not seek further education. The second argument is that high unemployment would
reduce the opportunity cost of going to school, making going to school the best other
viable alternative. These two theoretical arguments represent the relation between
income and substitution effects, and how one argument dominates the other. As we will
see, these arguments are crucial in understanding the impact of macroeconomic crises on
Following these theoretical notions, this paper intends to explore the impact of the Great
Depression on education, contributing to the literature by providing the evidence on these
effects on race (whites and blacks) and gender (males and females), during the period
from 1930 to 1940. Furthermore, this paper examines the effects of state employment
indices on the average education (at the mean). More importantly, it tests the effects of
state employment indices on the entire distribution of education (quantile regression).
The results (using individual census data from 1960) show some evidence that the Great
Depression affected education of whites born between 1911 and 1915. However, the
results show no evidence that the variation in state employment indices affected the
decision of schooling on the average (mean), but it affected the education of white males
at the top of the distribution (90% percentile).
This paper is organized into five sections. Section 2, reviews the literature. Section 3,
explores the theory behind demand for education. Section 4, describes the data. Section 5,
outlines the effects of the Great Depression on the educational attainment, and section 6
The literature of human capital and earnings has grown out of the work of Mincer (1974)
and Becker (1975). The literature shows that an individual chooses the level of education
that maximizes the present value of earnings compared to the cost of seeking this level
(direct or indirect), given the borrowing rate. Altonji (1993) expanded Mincer’s (1974)
and Becker’s work (1975) by treating education as a sequential choice made under
uncertainty. He examined how variables influencing tastes for schooling, ability to do
college work, and the payoffs to college affect the expected return to schooling.
Due to certain institutional features, like compulsory laws and constraints on borrowing,
the actual demand for education may be different from optimal choices. Such
institutional constraints may affect the educational attainment in two ways. First,
compulsory attendance and child labor laws may constrain education demand for bottom
deciles or percentiles. People may stay in school longer than what they would choose to
in the absence of these laws. Second, borrowing constraints may affect the education
choice for top deciles or percentiles. In the absence of student loans people may be
unable to attend school, even though the return on their schooling could exceed the
interest rate on loans.
A number of studies find large effects of compulsory education laws on the probability of
high school completion between 1920 and 1934. Lang and Kropp (1986) find that
compulsory education laws affect enrollment, even for groups not targeted by the laws.
Angrist and Krueger (1991), through the use of micro-data from the US Census data, find
that children born earlier in the year will obtain fewer years of schooling. In their study
they show a student’s enrollment will depend on his age, which is determined by the
quarter of birth. Lleras-Muney (2002), using the 1960 Census for students in secondary
schooling from 1915 to 1939, also finds that increasing the legal age attendance
requirement by one more year leads to a 5% increase in educational attainment. This
effect was significant for whites but not for blacks.
Extreme circumstances, such as war, natural disasters, or political upheavals might also
affect educational attainment. Meng and Gregory (2002) studied the impact of
educational interruptions during the Chinese Cultural Revolution (1966-1977) when most
schools in urban China ceased operations for 6 years. They find that the school
interruption had a substantial impact on educational attainment, especially when the
interruption occurs at high school level. They also find that children whose parents had
lower educational achievement and lower occupational status were affected the most.
Ichino and Winter-Ember (2004) tested the effects of World War II on educational
attainment. They find that World War II led to a significant drop in the educational
attainment of individuals who were of elementary school age during or immediately after
the conflict, as opposed to those born in the previous or subsequent decades. Comparing
the evidence for four countries, of which two were directly involved in the conflict
(Austria and Germany) and two were not (Sweden and Switzerland), the magnitude of
this educational loss is approximately 20% in years of education in Austria and Germany.
These individuals also experienced a sizable loss of earnings some 40 years after the war,
which may be attributed to the educational loss caused by the conflict.
On the other hand, other research shows that macroeconomic crises do not affect
education. Goldin (1998) finds that secondary-school enrollment and graduation rates
increased spectacularly in much of the US from 1910 to 1940; the advance was
particularly rapid from 1920 to 1935, during the early years of the Great Depression, in
the industrial area of the Northeast and Midwest. De Ferranti et al (2000) suggest that
large enrollment decisions are unaffected by macroeconomic crises, especially moderate
In 2002, Argentina’s economy suffered the culmination of an economic decline that
began in late 1998. The crisis was all the more unusual because it occurred when the rest
of the world was experiencing slow growth, but not recession. Real GDP fell 28% from
peak (1998) to trough (2002). Argentina’s currency, the peso, was devalued in January 2002.
Inflation was 41% in 2002 this caused real wages to fall 23.7% in 2002 and unemployment
rose from 12.4% in 1998 to 18.3% in 2001 and 23.6% in 20021. However, more recent work
on Argentina confirms that the crisis of the late 1990s and early 2000s did not change
overall enrollment level, but it may have negative effects on the quality of schooling
because of high associated rates of teacher absenteeism (Espana et al, 2002).
In Indonesia the deep financial crisis of 1998 appears to have little effect on schooling
outcomes (Thomas et al, 2004). Schady (2004), in his analysis of the impacts of the
macroeconomic crisis (1988-1992 ) on education in Peru, finds no effect on attendance
rates but a significant decline in the fraction of children who are both employed and
attend school. Therefore, given all previous findings, a careful empirical work is needed
in order to understand the ambiguous effect of macroeconomic crises on schooling.
3. Demand for Education.
In the economic model of human capital, individuals acquire schooling until the marginal
benefit of an additional year of education equals the marginal cost in order to maximize
the present value of lifetime income (Willis, 1986; Altonji, 1993). The marginal benefit
of one year of schooling is the resulting increase in the discounted expected stream of
earnings, and the marginal cost is the forgone income and direct private costs like tuition
and books. Given this decision rule, the reason some individuals invest in more
education than others must arise from factors that either raise the rate of return they
receive or lower the cost they must pay for funds.
The effect of a macroeconomic crisis on schooling is ambiguous in theory. In general, an
adverse macroeconomic shock will depress current employment and wage prospects, thus
the opportunity cost of attending school will fall. Holding everything else the same, this
should increase investments in human capital. A shock could also make borrowing
constraints more binding and thus reduce the total amount of schooling attainment.
When macroeconomic shocks are persistent, they may also depress expected lifetime
earnings, thus affecting the marginal benefit from schooling. If the lifetime earnings of
all individuals are reduced by the same percentage, regardless of their schooling, then the
marginal benefit associated with an additional year of schooling will be lower. Crises
need not, however, have a uniform effect across the board on expected earnings.
Additionally, the effect of a crisis on the wages and employment prospects of adults in a
1 United States Congress report (June 2003).
household may also have an effect on the schooling and employment decisions of
children (Schady, 2004).
Thus, because of the ambiguous effect on schooling, students or their parents may choose
more or less schooling; they may anticipate or postpone further schooling. The total
effect of a crisis on schooling will depend on the relative magnitude of the changes in the
marginal costs and benefits from education.
In this paper, I use the 1960 Census which is provided in the Integrated Public Use
Microdata Series (IPUMS)2. The IPUMS consists of twenty-seven high-precision
samples of the American population drawn from fourteen federal censuses. The IPUMS
assigns uniform codes across all the samples and brings relevant documentation into a
coherent form to facilitate analysis of social and economic change. Since the IPUMS
includes nearly all the detail originally recorded by the census enumerations, users can
construct a great variety of tabulations interrelating any desired set of variables.
My sample is drawn from the 1% state sample. I limit the sample to whites and blacks
(men and women) born in the US between 1896 and 1925. The sample contains 240,018
white males, 251,677 white females, 24,360 black males, and 27,588 black females.
Birth year is derived from reported age and quarter of birth.3 Individuals born in Hawaii
and Alaska are excluded from the sample. People reported to be born in the District of
Columbia and people who did not report their state of birth are excluded as well. This is
done because this paper utilizes the state employment indices for the period of 1930-
1940, and these indices do not include the District of Columbia. Depending on the state
of birth, 48 dummy variables for states are included where each equals one if the person
is born in the state, zero otherwise. Years of education are the number of years of
schooling completed, and the number is derived from the highest grades completed (years
of education range between 0-18 years of education). In this paper I add to the data the
states employment indices from 1930 to 1940, which I use from Wallis (1989).
Following Angrist and Krueger (1991), I constructed four dummy variables on the basis
of quarter of birth. This is done because most school districts require students to attain
age six by January 1 of the academic year in which they plan to attend and to turn 16 (in
most of the states) to drop out of school. These requirements cause quarter of birth and
years of education to be correlated (Angrist and Krueger, 1991). A student who is born
in the third quarter will wait until the subsequent year to enroll in school, while another
student born in the first quarter will be able to enroll that same year. This means that the
variation in education is related to quarter of birth. Also, according to the quarter of
birth, some students turn 16 before others, which enable them to drop out of school
2 Ruggless et al, 2004.
3 Census reports data in April, therefore, year of birth is equal to “1960-age”, if born in first quarter, and
year of birth is equal to “1960-age-1”, if not born in first quarter.
Two additional variables are also constructed. The first is employment index at age 18,
and the second is employment index at age 16. They are derived by matching the state
employment indices to the year of birth when the person turned 18 and 16, between 1930
and 1940 (because the indices are reported for this period), and to his state of birth. This
is done because this paper tests the relation between the state employment indices and
years of education, with respect to the fact that students earn their high school diploma at
age 18 and they can drop out of school at age 16. The data is analyzed for each group of
white males, white females, black males, and black females separately.
5. The Effects of the Great Depression on the Educational Attainment
A) Educational Attainment.
The major characteristics of the sample and the main variables are described in this part.
Table 1 report the mean and standard deviation of the variables used in the analysis
according to the person’s weight. Furthermore, figure 1 describes the educational
attainment by year of birth for whites, blacks, males, and females separately. The
measure of educational attainment is calculated by averaging the number of years of
schooling completed by individuals born in each year between 1896 and 1925.
Table 1 and figure 1 are here.
Table 1 show that the mean of the years of education for white females is the largest,
10.42 years, whereas for white males the mean is 10.29 years. Black males and females
show lower means, 6.94 years and 7.74 years respectively. Figure 1 shows that all the
groups experienced an increase in the mean of educational attainment (for people born
between 1896 and 1925). This is consistent with Goldin (1998) that the availability and
expansion in schools, especially secondary schools, increased the enrollments in high
schools through the period of 1910-1940, known as the “high school movement”.
However, panels’ c and d in figure 1 show that the average years of education decreased
for black males born in 1910, 1912, 1915, and 1917. As for black females who were
born in 1912, 1917, and 1919, we see the same pattern of reduction in the average years
of education. This is not the case for whites (males and females). Now, those people
born between 1911 and 1920 have turned 18 during the Great Depression era;
accordingly, a question could be raised if these limps in the educational attainment are a
result of the Great Depression or are caused by other anomaly.
To further explore the cross-cohort differences in educational attainment, I detrend the
secular educational trend. This is done by regressing years of education on the year of
birth, normalized to 1896, and quarter of birth dummies.4 The secular educational trend
was captured by a linear trend for white males and black females, and by a quadratic
trend for black males and white females.5 The residuals from these trends capture the
4 See Angrist and Krueger (1991) for the importance of using the quarter of birth in determining the years
5 For each group detrending was done by regressing years of education on polynomial of years of birth
from linear up to sixth polynomial and on quarter of birth. Then by calculating Schwarz Information
Criteria (SIC), the best detrending model for each group was the one that has the minimum SIC.
effects of the unobservable factors on years of education such as income and borrowing
constraint. Furthermore, these unobservable factors were also influenced by the Great
Depression. Income declined due to higher unemployment rates, and this lower income
affected the borrowing constraint. Therefore, we can conclude that the residuals may be
used to show the impact of the Great Depression on education. The residuals from these
trends are displayed in figure 2.
Figure 2 is here.
The residuals plotted in figure 2 show some evidence that educational attainment for
people born in the early 1910s was lower than the cohorts before and after, especially for
white males and females. The negative deviations from the secular trend for white males
and females characterize the period between 1910 and 1920, while the evidence of
similar negative deviations for blacks (males and females) during the same period is not
To support the evidence from the plotted residuals, I estimate the relation between these
residuals and the birth cohort. The model regresses the residuals, plotted in figure 2, on
dummies for five birth cohorts and quarter of birth dummies with a constant.6 The results
are reported in table 2. In table 2, the difference between the coefficients of two
consecutive cohorts can be interpreted as the difference between the numbers of years of
schooling completed, on average.
Table 2 is here.
Looking at the white male’s column of table 2, the average educational loss of the white
male cohort born between 1911 and 1915 amounts to approximately 6.1% of a year of
schooling (22.3 days) with respect to the previous cohort and to 6.4% (23.4 days) with
respect to the following cohort, and both differences are statistically significant (different
from zero). The corresponding losses for the white female cohort born between 1911 and
1915 are 7.2% (26.3 days) and 4.2% (15.3 days) and are also statistically significant.
This could present some evidence that the Great Depression might have affected the
educational attainment of white students born between 1911 and 1915, as compared to
the cohort before and after.
For blacks (males and females) no statistically significant difference can be found in table
2. The regression shows that the educational attainment for blacks (males and females)
was not affected, since the coefficients are not statistically different from zero. This
suggests that the Great Depression did not affect the schooling decision for blacks (males
One of the explanations for this could be that African Americans were not still fully
integrated in the society during the period 1929-1936. Discrimination still existed, there
were no equal opportunities in schools or jobs, plus their wages were much lower than
6 The regression was done for different models for all groups using each time different cohort periods, and
the results were similar in significance and effects.
their white counterparts, which might have forced blacks into job market instead of
schooling (Freeman et al, 1973; Smith and Welch, 1977). Furthermore, human capital
theory suggests that blacks may have faced a borrowing constraint during the Great
Depression where it could have been difficult to get loans to finance schooling. Becker
called these differences in opportunity (Kaufman and Hotchkiss, 2003). Opportunity, in
this sense, relates to the availability and cost of funds for investment in schooling.
Inequality of opportunity leads to inequality in years of education. This could be the case
for blacks, where lack of available funds and cost of funds led to lower years of
One more issue here could be that blacks had lower demand for human capital investment
(years of education). This may be due to discrimination in the labor market. The rate of
return from a given expenditure on schooling depends on the market earnings from work
after the schooling is completed. Discrimination, whether in the form of wage
discrimination or occupational discrimination, would result in a lower rate of return for a
given expenditure on schooling, causing a lower demand for schooling by blacks.
Another potential explanation could be the child labor laws (Ensign, 1969; Trattner,
1970; Krueger and Tjornhom, 2002). In most states it was illegal to hire students who
had not turned 14. This may have constrained blacks to stay in school until they were 14
and became legally eligible to work, which could be accounted for by the lower mean of
years of education in table 1 (7 years). Therefore, the Great Depression did not show
strong evidence of affecting the education for blacks, because already their years of
education were affected by both low demand for human capital and child labor laws.
As for white males and females, one of the explanations for the effects on the cohort of
1911-1915 could be the added worker effect (Kaufman and Hotchkiss, 2003). A lot of
people lost their jobs during the Great Depression period. This resulted in a great
decrease in their entire families’ incomes, causing a pure negative income effect for other
family members, such as students. Assuming that leisure time is a normal good, the
reduction in family income will lead to a decreased demand for leisure and a greater
supply of labor, as the student children seek work to supplement family income. From
here one can argue that white students found themselves in a situation where they had to
drop out of school and participate in the labor market by looking for a job to support their
families. This could explain why the educational loss was captured by the regression.
However, the cohort between 1911 and 1913 turned 16 between 1927 and 1929, when the
economy was booming. This could mean that these cohorts may have dropped out of
school at age 16, and that’s why the regression is capturing the negative deviations in
years of education.
The residuals analysis shows that whites (males and females) who were born between
1911 and 1915 suffered a loss in their education, while there is no evidence of the same
effect for whites born before or after this cohort. As for the blacks (males and females),
the results show no evidence that the Great Depression had any effect on their education.
To clearly explore if the educational loss, that whites born between 1911 and 1915
suffered, could be attributed to the Great Depression, I use an exogenous variable that
may have affected the years of education and was, in turn, affected by the Great
Depression. The variable is the state employment indices. This is done to overcome the
identification problem. The next part discusses the identification problem and tests the
relation between the employment levels and the years of education.
B) The Impact of Employment Levels on Years of Education
B-I Identification Problem and Employment Indices:
The impact on years of education of whites born between 1911 and 1915 could be
attributed to the Great Depression, or may be a result of factors other than the Great
Depression. Hence, I will use the state employment indices to clarify this issue. The
employment levels demonstrate severity of the Great Depression at the state level. In the
economic theory, the decision of schooling is affected by the employment levels, since it
represents the opportunity cost. Therefore, using the employment indices to test for any
effect on education will help to explore further if the Great Depression have affected the
years of education.
The severe impact of the Great Depression on the labor market is evident in the
unemployment rates. Between 1929 and 1933, the unemployment rate increased by over
20 points (Margo, 1993). For the remainder of the decade, the unemployment stayed in
the double digits. In addition to high levels of unemployment, the 1930s witnessed the
emergence of widespread and persistent long-term unemployment (unemployment
durations longer than one year) as a serious policy problem. Margo (1991) reports that in
Massachusetts State, in 1934, fully 63% of unemployed persons had been unemployed
for a year or more, and similar numbers were observed in Philadelphia in 1936 and 1937.
Margo (1991) also shows that between 1931 and 1940, the aggregate unemployment rate,
including persons holding work relief jobs, never dipped below 14%. The unemployed
were older and had completed, on average, 1.5 fewer years of schooling than the
employed. Therefore, with family incomes dropping, pursuing education would have
been more difficult; however, with insufficient employment opportunities not pursing
education would not yield any positive benefits.
The state employment indices I use in this paper are from Wallis (1989). Wallis (1989)
estimated the state employment indices during the Great Depression period, 1930 to1940.
The Great Depression affected different states differently, where some had severe
unemployment rates while others had only moderate rates. For example, Arizona had
employment indices of 81%, 68.9%, and 72.6% for the years 1931, 1932, and 1933
respectively; while Florida had 96.6%, 101.1%, and 91.8% for the same years. The
indices also show that unemployment declined steadily between 1929 and 1932-1933,
rose through 1937, fell in 1938, and again rose by the end of the decade. Manufacturing
industries were harder hit by the depression than non-manufacturing industries; they fell
farther in the 1929-1933 downturns and again in 1937-1938, ending the decade with low
levels of employment relative to non-manufacturing industries.
The decision to invest in more schooling is affected by the opportunity cost of labor
market (forgone income of not working). This cost was low during the Great Depression
period because of the high rates of unemployment. Therefore, the next section tests if
high employment indices led to more schooling. For this purpose, I derive the variable,
state employment index at age18, which matches the state employment index for each
person, when he turned 18, and the state he was born in. Because Wallis (1989) reports
state employment indices between 1930 and 1940, the state employment index at age 18
is derived for people born between 1912 and 1922, who turned 18 between 1930 and
1940. This is done to keep the variables consistent.
B-II Did Employment Indices Affect Schooling?
To test the relation between years of education and employment levels, I regress years of
education on state employment indices when the person is 18, year of birth dummies for
years between 1912 and 1922 (because students turn 18 between 1930 and 1940 if were
born between 1912 and 1922), state of birth dummies, and quarter of birth dummies, with
Table 3 is here.
The results of the regression (reported in table 3) show that the variation in the state
employment levels did not affect the education of white males and blacks (males and
females). However, the results show a small and negligible effect on white females. The
average educational loss for white females is 0.4% of a year of schooling (1.5 days) for
every 1% increase in employment indices, and significantly different than zero, with
respect to who was born in 1912.
One explanation for this small educational loss could be that white females preferred
work over education when employment indices were high. The Leisure/Labor model
suggests that when the opportunity cost increases, people will demand less leisure and
choose to work more. Therefore, white females preferred to reduce education (leisure)
and increase their participation in the labor market when employment indices increased.
This means that the substitution effect dominates the income effect. This is also
consistent with the findings of Goldin and Katz (1995). They find that females high
school graduates earned more than those with lower education during the Great
Depression period in clerk jobs. This indicates that when unemployment increased white
females believed that the return on education was high, and therefore, they preferred to
go to school.
For blacks (males and females), I find, as expected, that variations in state employment
levels did not affect the years of education. This is consistent with the residuals analysis,
which shows no impact of the Great depression on blacks’ education.
For white males, the variations in the state employment indices did not show any effect
on the years of education, because income effect and substitution effect cancel each other
out at the mean. This indicates that the income effect, demanding more leisure
(education) and less hours of work, is equal to the substitution effect, demanding more
hours of work and less education, causing no effect on the average education.
The previous analysis shows no evidence that the variations in state employment indices
affected the human capital accumulation decision made by white males and blacks (males
and females); at the time they turned 18. As for white females, there is a significant but
small and negligible effect. All in all, the use of state employment indices showed no
evidence, at the mean, that the Great Depression lowered education.
Also because students in the US are required legally to stay in school until the age of 16,
one more issue we can look at is the impact of state employment indices on the student’s
decision to drop out of school when s/he turns 16 (drop-out age). As before, due to the
reason that state employment indices are for the period 1930 to 1940, I derive a variable
that represents state employment indices at age 16 for each person , between 1930 and
1940, according to his or her state of birth. This is done for people born between 1914
and 1924 to be consistent with the employment indices.
In this model, I regress years of education on the state employment indices when the
person is 16, year of birth dummies for years between 1914 and 1924 (because students
turn 16 between 1930 and 1940 if born between 1914 and 1924), state of birth dummies,
and quarter of birth dummies, with a constant.
Table 4 is here.
The results (reported in table 4), once again, show no evidence that the variations in the
state employment indices affected the years of education at the mean, using the drop-out
age. The coefficients for white males and blacks (males and females) are not significant.
This suggests that we can’t reject the null hypothesis that coefficients equal zero (no
relation). However, for white females the coefficient on the state employment index is
negative and significantly different from zero, but it is also small and negligible (for
every 1 percent point increase in the state employment index the loss of white females is
0.7% of years of education (2.6 days)). All this indicates that the variations in state
employment indices did not show any effect on the students’ decision to drop out of
school for white males and blacks (males and females), whereas it shows little evidence
for white females.
This section shows that the state employment indices did not affect the years of
schooling, whether when the student turned 18 or even 16 years old. Since the residuals
analysis suggests that whites born between 1911 and 1915 were affected, this implies that
the total effect is not shown because the substitution effect and the income effect are
canceling each other out at the mean, which is the OLS regression. Therefore, part D of
this paper will discuss the regression around the 90% quantile, to see if still the two
effects cancel each other out for the top 10% students. But before doing so, I test in the
next part if the Great Depression had any impact on the probability of earning a degree,
whether high school or college, using also the state employment indices as an exogenous
C) The Impact of State employment Indices on the Probability of Earning a Degree
For further exploration, this part tests if the state employment indices had any impact on
the probability of earning a college degree or a high school diploma. It tests the relation
between state employment indices and the probability that a person graduates from
college when s/he has turned 18 between 1930 and 1940. It also tests the relation
between state employment indices and the probability that a person continues schooling
and earns a high school diploma when s/he has turned 16 between 1930 and 1940.
For this purpose, I start by estimating the probability of earning a college degree (college
equals 1 if years of education are 16 or more and equals 0 otherwise) on state
employment indices at age 18, year of birth dummies, state of birth dummies, and quarter
of birth dummies. The regression is done for people with 12 years of education or more.
Table 5 is here.
Table 5 reports the results. The results show that for blacks (males and females) and
white males, there is no evidence of any effect. As for white females, the effect is small
and negligible, despite it is being significant (higher state employment indices led to
increase the probability of earning a college degree by 0.05%). Therefore, the state
employment indices did not affect the probability that a person would earn a college
degree. This suggests that there is weak evidence that the Great Depression affected the
probability of a person earning a college degree.
Now I turn to estimate the probability of graduating from high school (high school equals
1 if years of education are 12 or more and equals 0 otherwise) on state employment
indices at age 16, year of birth dummies, state of birth dummies, and quarter of birth
dummies. The regression is done for people having 8 years of education or more.
Table 6 is here.
Table 6 reports the results. The results show a small significant impact for whites (males
and females) and black females. The higher state employment indices led to a decrease
in the probability of earning a high school degree by 0.1%, 0.1%, and 0.2% for each
group respectively. This indicates that the effect is negligible economically. As for the
coefficient on black males, it was not significant. Hence, the results suggest that the
probability of earning a high school diploma was affected but by very little by variations
in the state employment indices during the Great Depression.
From the previous analysis and regressions, the paper shows weak and negligible
evidence for the effects of the state employment indices on years of education, or the
probability of earning a degree (college or high school). This is interpreted as little effect
of the Great Depression on education. As mentioned before, the reason for this could be
that the income effect and the substitution effect of the Great Depression are canceling
each other out at the mean (average education). This suggests that the low opportunity
cost and restrictions on borrowing may have led to a corner solution rather than to an
interior solution, and that is why we do not see any effect on years of education. For this
reason, the next part will investigate if the Great Depression affected the 90% quantile of
students or not, since the regressions around the mean show no significant impact.
D) Quantile regression of Years of Education and State Employment Indices
That there appears to be little effect of the Great Depression on the average schooling
years does not mean that there were little effects on the overall population. For example,
Lleras-Muney plots years of education for each decile for the 1901-1925 birth cohorts. In
her Figure 2 (p. 421), we see a noticeable dip in educational attainment for the 90-
percentile of the distribution for the birth cohorts born during 1910s. In order to see
whether or not this pattern holds in the data in this paper, I create similar plots from my
sample for 10-, 50- and 90-percentiles of the distribution of years of education.
Figure 3 is here.
Figure 3 presents these plots for four different race/sex categories. The plot for the 90th
percentile of white males shows a noticeable dip of the years of education for the 1913
and 1916 birth cohorts and there is a dip for the 1913 birth cohort for black males. On
the other hand, there is no change for white females and an increase for the 1916 birth
cohort for black females at the 90-percentile of educational attainment.
I formally test whether or not the variation in the state-level employment index had
effects on educational attainment at the top tail of distribution. As before, I regress the
years of education on the employment index at age 18, the dummy variables for 48 states,
ten birth-years, and three quarters of birth for a sub-sample of those who were born
between 1912 and 1922 but this time using the quantile regression. Table 7 presents the
results of the quantile regressions separately for four demographic groups.
Table 7 is here.
The results show that top 10% of white males were affected. The coefficient for white
males is positive and statistically significant (3%). It is possible that the variations in the
state employment indices encouraged white male students to get more education and
graduate from school. On the other hand, for white females and blacks (males and
females) there is no evidence that high employment indices had any effect on the top
From this part, I conclude that the substitution effect and the income effect may be
canceling each other out at the mean, but the income effect is winning at the top tail of
the distribution. One explanation for this could be that the 1960 Census includes WWII
veterans. These veterans greatly benefited from the GI Bill, which provided financial
incentives to return to school. This could be the case for white males.
This paper shows evidence that the Great Depression (through the residuals analysis)
affected the years of education for the cohort of white males and females born between
1911 and 1915, but did not affect blacks (males and females). However, when looking at
the impact of the employment indices on schooling, the results show that the state
employment indices, which were affected by the Great Depression and were used as an
exogenous variable, did not affect the educational attainment of students. This suggests
that the income effect and the substitution effect are canceling each other out at the mean,
which is why the variations in state employment indices show no effect on education at
the mean. Also, it could be that the residuals analysis captures the 16 year old students
who preferred to drop out of school when the economy was booming between 1927 and
1929, rather than showing the result of the Great depression.
When looking at the impact of the variations in the state employment indices on the 90th
percentile of students’ education, white males appear affected, which suggests that the
income effect was larger than the substitution effect for the top 10%. The insignificant
impact of the Great Depression on blacks’ educational attainment could be a good topic
for further research.
Years of education
Born in second quarter/(Q2)
Born in third quarter/(Q3)
Born in fourth quarter/(Q4)
Number of observations
- Standard deviation is in parentheses.
- The calculations were according to the personal weight.