The Great Depression and the rise of female employment: A new hypothesis

https://doi.org/10.1016/j.eeh.2020.101383Get rights and content

Abstract

The life-cycle labor supply of women born at the turn of the 20th century diverged sharply from previous cohorts. Although they had similar participation rates in early adulthood, younger cohorts were significantly more likely to work at middle age. This paper documents a link between these changing patterns of female labor supply and the Great Depression. We find that the onset of the Great Depression led to an increase in young women's employment in 1930 via an added-worker effect. Cohorts induced into the workforce in the early 1930s had significantly higher employment rates through the 1940s and 1950s of up to 3 percentage points, suggesting a permanent impact of the Great Depression on women's lifecycle labor supply.

Introduction

The Great Depression was one of the most dramatic events in American economic history. It lasted nearly a decade, witnessed unemployment rates greater than 20% and a decline in GDP of over 25% (Margo, 1993). Economists and economic historians have extensively studied both the causes of the Great Depression and its impact on various socioeconomic outcomes, such as economic activity, fertility, mortality, and marriage.1

This paper provides the first evaluation of the long-run impacts of the Great Depression on female employment. A priori, this relationship is ambiguous. On the one hand, persistent unemployment of husbands, significant asset losses, and high levels of accumulated debt might have led married women to enter the labor market as secondary workers (an added worker effect). On the other hand, reduced labor demand and increased enforcement of marriage bars might have reduced their participation.2

Our empirical analysis exploits variation in the severity of the economic downturn in the early 1930s across different states. We pool individual data on employment status as well as other characteristics from several Censuses between 1900 and 1960. Our primary measure of the severity of the Great Depression is the increase in the ratio of industrial and commercial failures to business concerns. Business failures increase in response to large and persistent shocks rather than to transitory shocks. They are also more akin to labor demand shifts that lead to layoffs than to labor supply shifts. Our main approach consists of comparing employment outcomes of women in different age brackets in states that were more vs less severely affected by the economic downturn during the Great Depression.

We document two main results. First, we show that women of working age in 1930 (16 to 64 years old in 1930) significantly increased their employment immediately after the Great Crash in states that were more severely impacted. Women who turned working age in the 1930s (aged 20 to 24 in 1940) also increased work in 1940 in response to the shock. The fact that married women increased their employment as well is suggestive of an added worker effect. Second, we find that these same cohorts had persistently higher employment rates throughout their lifecycle, decades after the Depression. Our estimates point to cohort-specific effects that are quantitatively important. We calculate that a 16 to 24-year-old woman in 1930 in a state that experienced a sharp increase in the business failure rate during the Great Depression will increase her work propensity by 16% when 25 to 34 years old in 1940 and by 22% when 45 to 54 years old in 1960. Calculations for a 25 to 34-year-old woman in 1930 suggest effects of similar magnitude. Among white women, the cohorts born between 1886 and 1920 – whom we refer to as the D-cohort– are the first to break the long-standing pattern of permanent labor market exit after marriage (Goldin, 1990; Costa, 2000).3

Our identifying assumption is that in the absence of Depression-related business failures, the employment of women would have trended similarly across states. We conduct a variety of checks to ensure the validity of this assumption. First, in addition to standard state, year, and division-year fixed effects, we control for a host of pre-Depression local covariates that could potentially confound the Depression-related effects (such as the size of manufacturing sector, migration, share of farms, etc.). Second, we show that our baseline measure of the Great Depression does not predict higher female labor force participation or employment in pre-Depression years. We further assess whether other unobservable state factors confound our estimates: significant economic downturns preceding the Great Depression, changes in contemporaneous economic conditions, WWII mobilization, the expansion of white-collar employment, or the expansion of the manufacturing and trade sectors (Goldin, 1990, 2000; Acemoglu et al., 2004). In the same spirit, we also employ county-level variation in per capita retail sales between 1929 and 1933 (Fishback et al., 2005) along with state-year fixed effects and verify that our main findings in the short-run go through. In sum, all exercises that we have conducted support the identifying assumption and strengthen our confidence that the link we document is likely causal.

Our findings suggest that the Great Depression could have induced an added worker effect in the short-term. What could explain, however, the persistence of the effects in the long-term? To answer this question, we first study the impact of the Great Depression on wages. We find that in states more severely affected by the shock, the D-cohort received lower wages in 1960, when 45 to 64 years old, relative to women of the same age in 1940. These effects persist after accounting for possible self-selection in the labor market. Habit formation and/or a labor supply shift due to declines in permanent income linked to the shock are plausible explanations of these findings. We obtain similar results for the wages of men in this age group, who could have been the likely spouses of the D-cohort. Employment gaps, absence of opportunities, and human capital depreciation may have led them to shift into lower paying occupations.

An alternative explanation for both the long-term increase in women's work and the decline in their wages is that the Great Depression had a long-term negative impact on the economies more severely affected. To address this possibility, we control for the states’ differential growth of real per-capita GDP in the decades after 1930. Our results, however, remain unaffected. Finally, as an external validity check, we use an alternative source, a 1978 survey (Ridley, 2007) which asked a sample of 1,049 ever married women born between 1901 and 1910 questions relating to their experience surrounding the Great Depression. We find a strong positive link between the total number of years these women worked and the initial impact of the Great Depression on their family incomes.

Our findings complement previous explanations for the rise in women's participation including increased education, the expansion of the white-collar sector, the diffusion of labor-saving technologies, and WWII mobilization (Goldin, 1990, 1991a, 1998, 2000, 2006; Acemoglu et al., 2004; Greenwood et al., 2005; Lewis, 2018; Bailey and Collins, 2011). The paper also adds to the growing research aimed at understanding how households respond to negative shocks and in particular the role of spousal insurance in smoothing shocks due to husbands’ job loss. Our analysis suggests that substantial income shocks, such as that induced by the Great Depression, which alter households’ permanent income and entail long unemployment spells, can lead to persistent increases in women's labor supply.4

The paper proceeds as follows. A historical background is discussed in the rest of this introduction. Section 2 describes the data. Section 3 tests our hypothesis. Section 4 provides identification and robustness checks. Section 5 discusses various channels and examines the impact of the Great Depression on wages. Section 6 presents results from a survey on women born between 1901 and 1910. Section 7 concludes.

The “Roaring Twenties” was a period of prosperity and economic growth, during which the construction and real estate sectors boomed and consumer debt to purchase durable goods, from home appliances to cars, sharply increased. Olney (1999) documents the dramatic expansion of installment payments in the 1920s and argues that “societal attitudes toward borrowers changed radically between 1900 and 1920; by the mid-1920s, buying on credit was considered normal, not sinful.” She shows that this led to unmanageable household debt and to an increase in default rates in the 1930s. Bolin (1978) reports that middle-income households cut back on their expenditures but also tried to maintain installment purchases by “placing additional workers in the labor force”. This meant that also wives of relatively more affluent husbands had to accept paid employment in order to maintain their living standards.5

In addition to the high consumer debt, the credit market expansion fueled a real estate bubble. After the stock market crash, falling prices combined with high unemployment and a sharp decline in incomes increased the real burden of nominal debt. Between 1926 and 1933, the foreclosure rate increased from 3.6 per 1000 home mortgages to 13.3. In 1933, on average 1000 homes were foreclosed daily (Wheelock, 2008). Bolin (1978) calculates that the number of married women in the labor force increased by approximately 50% between 1930 and 1940. Finegan and Margo (1994) report that among women whose husbands were unemployed in 1940 (but not on relief), 24% were in the workforce as opposed to 16% among women whose husbands were regularly employed. These calculations are suggestive of an added worker effect. Due to the hostility married women faced when entering the labor market during the Depression years, some acquired jobs by keeping their marital status secret (Vosko, 2000). Ware (1982) and Kennedy (1999) report that women took low-paid jobs in industries less hard hit such as in nursing, teaching, sewing or domestic services. Klein (2018) argues that after the 1929 crash, many African Americans were fired from positions that were taken over by white workers in need of employment.6 These jobs, predominantly low-skill and low-pay, were also less likely to be restricted by marriage bars.

Fig. 1 corroborates the findings of Bolin (1978) and Finegan and Margo (1994) documenting increased participation of women in the 1930s, which is sharpest for younger and married women. Between 1940 and 1960 the increase in female employment is strikingly cohort-specific, with small increases for young cohorts and large increases for older cohorts, and particularly for those born between 1896 and 1915 (Goldin, 1990, 2000; Smith and Ward, 1985). The latter increased their participation dramatically and persistently relative to women born a decade earlier. Young married women may have been more affected than young single women as they may have bought their homes at peak real estate prices (Bellou and Cardia, 2016a) and/or have installment payments to make on newly purchased durable goods (Olney, 1999).

Section snippets

Data

Our main data source is the US Census obtained from IPUMS (Ruggles et al., 2020).7 The baseline analysis relies on the full count files for 1910, 1930, 1940, the 1% sample for 1950 and the 5% sample for 1960. This data provides micro-level information on employment and other individual characteristics throughout this period as well as on wages from 1940 onwards. We primarily focus on

Main results

The results from estimation of (1) are reported in Table 3. As can be seen, in states with worse economic conditions during the Great Depression, there is an immediate increase in women's work in 1930 across all age groups. Interestingly, these effects carry over in the short-, medium-, and long-run and have a clear cohort-specific pattern: more women, who were 16 to 34 years old in 1930, work when they are older in 1940, in 1950, and in 1960 (see solid black lines). Moreover, women younger

Identification

As discussed in Section 2.3, our analysis relies on the parallel trends assumption. To understand whether the latter is likely to be satisfied, we check whether – conditional on all covariates specified in (1) – our measure of the severity of the Great Depression predicts female labor supply in years prior to the Great Depression. We pool together the 1900, 1910, 1920 and 1930 panels and use a labor force participation indicator as our dependent variable since employment status is not recorded

Female employment over the short-run

Our findings suggest that the Great Depression was an important determinant of employment decisions for the cohorts born in the late 1800 and early 1900. All women increased work in response to the shock, including the married, a group with traditionally very low market participation. Moreover, female employment increased significantly more relative to male employment in states experiencing larger increases in their business failure rate. These results are consistent with an added worker

Results from a survey

A special survey on a subset of women in our cohort provides an additional external validity check on the hypothesis that the Great Depression had a long-term impact on the D-cohort.36 This survey includes ever married women born between 1901 and 1910 who are part of our focal cohort. In 1978, these women were asked questions pertinent to the Great

Conclusion

We have documented a positive link between the severity of the Great Depression and the work behavior of cohorts that were of working age during the depression years. In states where economic conditions deteriorated the most, these women increased employment in the short-run likely to compensate for income or asset losses; many of them however continued working till their retirement. The entire lifetime labor supply profile of these cohorts is persistently linked to the economic conditions of

References (50)

  • P.V. Fishback et al.

    The impact of New Deal expenditures on mobility during the Great Depression

    Explor. Econ. Hist.

    (2006)
  • D. Acemoglu et al.

    Women, war and wages: the effect of female labor supply on the wage structure at mid-century

    J. Polit. Econ.

    (2004)
  • M.J. Bailey et al.

    Did improvements in household technology cause the baby boom? Evidence from electrification, appliance diffusion, and the Amish

    Am. Econ. J.: Macroecon.

    (2011)
  • A. Bellou et al.

    The Great Depression and the Lasting Effects of Mortgage Debt on Women's Work

    (2016)
  • A. Bellou et al.

    Baby Boom, Baby Bust and the Great Depression

    (2016)
  • A. Bellou et al.

    Great Depression and the Rise of Female Employment: A New Hypothesis

    (2020)
  • R. Blundell et al.

    Consumption inequality and family labor supply

    Am. Econ. Rev.

    (2016)
  • W.D.W. Bolin

    The economics of middle-income family life: working women during the Great Depression

    J. Am. Hist.

    (1978)
  • M.D. Bordo et al.

    The Defining Moment: The Great Depression and the American Economy in the Twentieth Century

    (1998)
  • J. Bredtmann et al.

    Husband's unemployment and wife's labor supply: the added worker effect across Europe

    ILR Rev.

    (2018)
  • D.L. Costa

    From Mill Town to Board Room: the rise of women's paid labor

    J. Econ. Perspect.

    (2000)
  • J.B. Cullen et al.

    Does unemployment insurance crowd out spousal labor supply?

    J. Labor Econ.

    (2000)
  • A.T. Finegan et al.

    Added and Discouraged Workers in the Late 1930s: A Re-Examination

    (1993)
  • A.T. Finegan et al.

    Work relief and the labor force participation of married women in 1940

    J. Econ. Hist.

    (1994)
  • P.V. Fishback et al.

    The impact of new deal expenditures on local economic activity: an examination on retail sales, 1929-1939

    J. Econ. Hist.

    (2005)
  • P.V. Fishback et al.

    Births, deaths, and New Deal Relief during the Great Depression

    Rev. Econ. Stat.

    (2007)
  • P.V. Fishback et al.

    Hard times in the land of plenty: the effect on income and disability later in life for people born during the Great Depression

    Explor. Econ. Hist.

    (2014)
  • P.V. Fishback et al.

    The multiplier for Federal spending in the States during the Great Depression

    J. Econ. Hist.

    (2015)
  • C. Goldin

    Understanding the Gender Gap: An Economic History of American Women

    (1990)
  • C. Goldin

    The role of World War II in the rise of women's employment

    Am. Econ. Rev.

    (1991)
  • C. Goldin

    Marriage bars: discrimination against married women workers from the 1920s to the 1950s

  • C. Goldin

    America's graduation from high school: the evolution and spread of secondary schooling in the twentieth century

    J. Econ. Hist.

    (1998)
  • Goldin, C., 2000. Labor markets in the twentieth century. In Engerman, S., Gallman, R. (Eds.) Chapter 10: The Cambridge...
  • C. Goldin

    The quiet revolution that transformed women's employment, education, and family

    Am. Econ. Rev.

    (2006)
  • X. Gong

    The Added Worker Effect and the Discouraged Worker Effect for Married Women in Australia. IZA Discussion Papers No. 4816

    (2010)
  • The authors kindly acknowledge financial support from the SSHRC. We would like to thank Joshua Lewis and participants at the CNEH and SOLE Conferences for their useful comments.

    View full text