Presentation on theme: "The Female Labor Force in Historical Perspective Economics 2333 Topic #9 Spring 2014."— Presentation transcript:
The Female Labor Force in Historical Perspective Economics 2333 Topic #9 Spring 2014
Outline Background Goldin, Bailey: The Quiet Revolution and the Pill Salisbury: Marriage and female income in late C19.
Background Major topic area in economic history Key issues explored: evolution of FLP participation over course of economic development; relationship between fertility change and FLP; male-female differences in earnings and occupations Key work: Goldin (1990), Understanding the Gender Gap. Focus today mostly is C20 (except Salisbury). Key C19 issues: m/f earnings differences and industrialization; U-shaped pattern in female LFP (see Olivetti extension).
Goldin: The Quiet Revolution 4 phases of change in women’s “involvement” in economy. First three are evolutionary, last is revolutionary. Distinction between evolution and revolution involves three aspects of choice. Horizon: at the time of human capital investment is expect LFP long and continuous or brief and intermittent? Identity: job or career Decision-making: are labor supply decisions made jointly with spouse/partner or does the woman optimize taking the other’s decision as given? Revolution: “passive” economic agents to “active” economic agents. Delay marriage until after (well) human capital investments are made. Phase I: late C19-1920s. Phase II: 1930-50. Phase III: 1950-70s. “Quiet” Revolution: 1970s-present.
Basic Argument Focus is on married women, since it is their LFP that changes so much in the C20. Begin with labor supply function, Slutsky equation. Income effect: negative, substitution effect: positive. Shifts in female outcomes (participation, hours, education, occupations, etc.) reflect shifts in demand/supply AND in the parameters of the supply curve (NOTE: not much in Goldin about shifts in male/female substitution on the demand side). Change in labor supply parameters are partly period (e.g. WW2) but MOSTLY cohort. Interesting sidebar on history of thought: evolution of female labor force affects evolution of labor economics as an intellectual discipline.
Phase I 1890s through 1920s. Vast majority of married women did not work. Those who did were either from very poor households or very privileged. Young single women worked, but typical job was in manufacturing paid by the piece. Low return to human capital and little or no gains to experience beyond initial. Women exited the LF upon marriage (not upon first birth). Social norm was that women stayed home, men worked. Reinforced by marriage bars in teaching. Upshot is very small substitution effect and very large income effect. Combination is very inelastic LS curve. Implies that female LP will increase primarily because of outward shifts in LS curve. Must have been some shift because LFP of married women does increase during Phase I. Possible shifters: increase in share of “nice” (clerical) jobs, rising education.
Phase II LFP of married women increased by 15.5 percentage points for ages 35-44 between 1930 and 1950. Approximately half of employed women in 1950 were married. Demand side: growth in clerical sector. Supply side: high school movement, emergence of scheduled part-time work in the 1950s. Diffusion of the “electric” household (Greenwood et al). Income effect declined, substitution effect increased. LS curve became very elastic by the 1950s.
Phase III Large increases in LFP of married women, labor supply curve becomes even more elastic. BUT married women were mostly secondary earners. Human capital investment occurs before marriage, little on the job training, and not much training for “careers”. Relevant cohorts very substantially underestimated the extent to which they would be in the labor force later in life → different behavior with hindsight, passed on to daughters.
Phase IV: revolution Young women adjust expectations of later LFP. Invest more heavily in schooling and careers (occupations and college majors). Identity shifts: women develop their labor market reputations before marriage and retain their names upon marriage. Aspirations while in college become more like men.
Bailey on the Pill Birth control pill is introduced in 1960. Did it matter economically? Much research suggests “very little”. Fertility in decline long before 1960, along with rise in female LFP. Even if it did reduce fertility, various papers suggest that decreases in fertility did not have much to do with rise in female FLP. Goldin and Katz (JPE) argue that access to the pill leads to a later age at first marriage and is also associated with an increased incidence of “careers” among women. Bailey uses variation in state access to examine the effects on younger women. Finds that ELA (early legal access) significantly reduces the likelihood of a first birth before age 22 and raised female LFP. Identification strategy: state level changes in the 1960s-70s that expanded the legal rights of individuals between the ages of 18-21. Indirect effect is to enable younger women to have access to medical care without permission of parents. 1976 is the cutoff date because of relevant Supreme Court decision giving ELA throughout the US. Figure 1: dip in fertility of women in the early 20s about the time the pill becomes available. Figure 2: age at first birth shifts to the right in the affected cohorts Figure 3: fertility dip in LFP disappears for affected cohorts
Why the Pill Might Have Mattered? Pill reduces cost of unwanted births and also enables women to time births. Net effect on completed fertility is small. Timing matters: by reducing likelihood of early births women spend more time in school and more time in the labor market in their 20s. These are the “quiet revolution” women.
Identification strategy ELA varied across states. Obtaining the pill required a prescription. If said prescription was illegal, heavy costs were borne by the physician. Different types of laws. MM = mature minor statutes (originate in Ohio in 1950s). AOM = age of majority statutes, consequence of Vietnam era draft. CFP= comprehensive family planning statute that permitted treatment or did not explicitly restrict minors. 1976: Supreme Court decision in Planned Parenthood of Missouri vs. Danforth: states do not have a compelling access in using age to restrict access to contraceptives. History of laws suggest that “time to liberalization” (date of passage of relevant state law – 1960) uncorrelated with state characteristics. No effects EXCEPT for percent Catholic which delays liberalization. Cannot control directly for this so includes state fixed effects and state time trends.
Empirical Analysis: June CPS Uses March and June CPS Supplements as these have retrospective information on fertility and age at first birth. Results likely attenuated if respondents travelled across state lines, also CPS does not have information on state of residence at age 21. Analysis using census suggests slight attenuation. Focus is on women ages 36-44 at the time of the survey. Key results in Table III: ELA reduces the likelihood of a first birth before age 22. However, no effect before age 19 and no effect on completed fertility. Uses June CPS.
Empirical Analysis: March CPS March CPS is better for labor force analysis but state of residence data are grouped, requiring a (complicated) probabilistic measures of ELA. ELA is associated with high likelihood of labor force participation among women ages 26-30. Mechanism in the argument requires that women who were working ages 26-30 delayed childbearing because of ELA. So if a women had ELA but still had a child anyway, should not be working. Last column of Table IV, which uses June CPS, supports this interpretation. Table V: similar effects on the intensive margin.
Selection Issues Variation in processing times exogenous? Postal system, unpredictable attorney behavior, clerical errors. BUT processing time may depend on effort expended on the application, which varies with expected benefit. Econometric solution: mixed proportional hazards model (structural) with parametric assumptions, two marriage types and two pension types. Estimation using EM algorithm. Results look good.