1 / 28

Coping with Prosperity: The Response of Parents’ Child Care Time Use to Rising Earnings

Coping with Prosperity: The Response of Parents’ Child Care Time Use to Rising Earnings. Coping with Prosperity: The Response of Parents’ Child Care Time Use to Rising Earnings. James M. Payne. Friday Afternoon Research Group November 15, 2013. Defining the dependent variables.

marge
Download Presentation

Coping with Prosperity: The Response of Parents’ Child Care Time Use to Rising Earnings

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Coping with Prosperity: The Response of Parents’ Child Care Time Use to Rising Earnings Coping with Prosperity: The Response of Parents’ Child Care Time Use to Rising Earnings James M. Payne Friday Afternoon Research GroupNovember 15, 2013

  2. Defining the dependent variables Defining the dependent variables • FACETIME—direct child care time • Physical & medical care • Reading, playing, sports, arts & crafts • Schooling & homework • Conversing • BEHALFTIME—indirect child care time • Organizing and planning • Attending events, conferences, etc. • Dropping off/picking up/waiting • Obtaining medical care • Using child care services Coping with Prosperity

  3. Coping with Prosperity Propositions • Parents seek to accumulate human capital in their children,with household production function: human capital = f (FACETIME, Market services) where market services = f (Purchased services, BEHALFTIME)An increase in parents’ wage rates raises: • nominal incomes, and thus demand for human capital in children • absolute prices of both own time and market services, and the relative price of own time Propositions

  4. Coping with Prosperity Propositions • Expectation: Rising wages will lead to • an increase in FACETIME (probably) • an increase in market services (unambiguously) and hence in BEHALFTIME • an increase in BEHALFTIME relative to FACETIME Propositions

  5. Coping with Prosperity Challenges • Wages   child care time use  is well established, but . . . • no studies have separated the income and price effects Challenges (1)

  6. Coping with Prosperity Challenges • Zeroes in time use data • A. 4 sources of zeroes1. measurement error • 2. the individual will not participate in the activity3. corner solution4. diary window is shorter than the consumption period • B. Log transformation cannot be used to reduce heteroscedasticity • C. Tobit models common, but generate biased estimatesGould (1992), Keen (1986), Daunfeldt and Hellström (2007), Stewart (2009), Frazis and Stewart (2010) Challenges (2)

  7. Coping with Prosperity Challenges • Wage variable problems--endogeneity--sample selection bias Challenges (3)

  8. Coping with Prosperity Challenges Missing income data in the Current Population Survey (CPS)A. 34% missing wage data (2003) with negative selectionHeckman and LaFontaine (2006); 15.6% in my parents-only ATUS sampleB. Census’ modified hot deck imputation produces match biasHirsch and Macpherson (2004), Bollinger and Hirsch (2006, 2010) Challenges (4)

  9. Coping with Prosperity Data • American Time Use Survey (ATUS) 2003 – 2010 Data • Initiated by Bureau of Labor Statistics (BLS), 2003 • One-day time use diaries • Linked to CPS demographic and labor force data • n = 45,716

  10. Two dependent variables Descriptive statistics(minutes on diary day) Coping with Prosperity

  11. Theoretical model Theoretical model • Two inputs for producing human capital in children: • Own time, which is FACETIME with price pf • Market services (S) • Two components: • Purchases (P) with share γand price pp • BEHALFTIME with shareτand price pb= pf • γ+τ= 1 Coping with Prosperity

  12. Theoretical model • Key point: How will a change in hourly earnings (w) affect input prices? • pf: FACETIME consists only of time, so pf = w, = 1 • ps = γpp + τpb (weighted average of component prices) • = 0 (prices of services are orthogonal to w) • = τ (since pb = pf and = 1) • So = τ < 1: an increase in earnings reduces the price of market services S relative to FACETIME, and . . . • predicts that BEHALFTIME will be substituted for FACETIME Coping with Prosperity

  13. ATUS earnings data • Challenges (4): Estimating earnings from CPS • Use updated ATUS earnings data • If available and valid, use hourly wage rate (TRERNHLY) • If not, use weekly earnings (TRERNWA) divided by usual weekly work hours (TEHRUSLT) • If topcoded, use estimates from Hirsch and Macpherson (2011) • If not available, set to missing and impute later • Adjust to 2003 constant dollars using CPI-U-SL to createrEARNHR Coping with Prosperity

  14. Coping with Prosperity Multiple imputation • Rubin (1987) • Process is intended to . . . • Fill blanks with “neutral” values • Preserve variation in the data • Use EM/MCMC multiple imputation method • Create several (m)different data sets, and • Model each set (imputation) separately • Combine estimates using Rubin’s Rule Challenges (4): Multiple imputation (MI) of missing data

  15. Multiple imputation • Multiple imputation • Impute all missing covariates • Create m = 10 imputed data sets Coping with Prosperity

  16. ATUS earnings data • Challenges (3): Endogeneity and sample selection in real hourly earnings (rEARNHR) • Two problems—rEARNHR is: • Endogenous with child care time (Heckman r = 0.14, p-value < 0.001) • Unobserved for nonworking parents Coping with Prosperity

  17. ATUS earnings data • Two-step solution—Millinet (2001) from Amemiya (1985):Heckman (1979) sample selection model, with • 1st step: probit model—estimate probability of being employed (WORKING = 1) • 2nd step: Include instrument (METRO) in OLS model for estimated real hourly earnings • Result: one estimated earnings variable, rEARNHRhat, accommodating both endogeneity and sample selection bias Coping with Prosperity

  18. ATUS earnings data • Two-step solution Coping with Prosperity double hurdle models with x = rEARNHRhat 1st step probit (Heckman) for WORKING = 1 2nd step OLS (Heckman) for rEARNHRhat Instrument for endogeneity(METRO*)

  19. Inverse hyperbolic sine • Challenges (2): Inverse hyperbolic sine transformation of dependent variablessinh-1y = ln(y + √ y2 + 1 ) Coping with Prosperity 4 3 2 1 0 -1 -2 -3 -4 sinh-1 • Burbidge, et al. (1988) • Defined for all real numbers • Equivalent of a log-linear model • Predicted values can be returned to levels by taking sinh ln -10 0 10 y

  20. Coping with Prosperity Double-hurdle model • Cragg (1971), Lin & Schmidt (1984) • Assumes a corner solution • Zero time use observations arise because some people never do the activity • Two hurdles to engaging in the activity: • 1st hurdle: choosingwhether to do the activity • 2nd hurdle: choosing how much to do the activity Challenges (2): Double hurdle model

  21. Estimation—Double hurdle Estimation—Double hurdle Two-part model—estimate, separately for both FACETIME and BEHALFTIME A. First-hurdle probit models for all observations B. Second-hurdle truncated normal models using OLS for nonzero observations of FACETIME and BEHALFTIME Coping with Prosperity

  22. Estimation—Double hurdle Double hurdle results--FACETIME Coping with Prosperity Boldface effects are significant at α = .05 *Marginal effects at means of regressors

  23. Estimation—Double hurdle Double hurdle results--BEHALFTIME Coping with Prosperity Boldface effects are significant at α = .05 *Marginal effects at means of regressors

  24. Theoretical model • Challenges (1): Testing the hypothesis • Construct a ratio of BEHALFTIME and FACETIME and observe the effect of earnings on this • Due to zeroes in the data, I construct the ratio of the marginal effects (ME) in the probit model Coping with Prosperity

  25. Substitution of BEHALFTIME for FACETIME • Substitution of BEHALFTIME for FACETIME • 1st hurdle probit model—strong results for both variables for women • Fit a bivariate probit model for women for each imputed data set, and combine results using Rubin’s Rule • For both variables, calculate marginal effects at: • P10Q1MedianQ3P90P95of rEARNHRhat • P10Q1MedianQ3P90of CHILDAGE • Medians of WORKHRS (30 hrs.) and SPOUSEHRS (40 hrs.) • Means of other continuous variables • White, female, college graduate, US citizen, married, homeowner, not a student, employed by a private firm, for a Monday in January, 2009 Coping with Prosperity

  26. Substitution of BEHALFTIME for FACETIME Sensitivity analysis—marginal effects from bivariate probit model Coping with Prosperity *Earnings shown are the means of the 10 imputations; models were estimated using imputation-specific means

  27. Substitution of BEHALFTIME for FACETIME • Comparisons from bivariate probit model • Substitution of BEHALFTIME for FACETIME occurs as earnings rise • Effect is similar but: • smaller for blacks • smaller for high school graduates • larger for single parents Coping with Prosperity

  28. Conclusions • Conclusions • Primary hypothesis • Parents substitute BEHALFTIME for FACETIME as earnings rise • Secondary hypotheses • Substitution effect is larger for single parents • For women, schooling positively affects the likelihood of engaging in both FACETIME and BEHALFTIME as well as the amount of time; for men, it affects the likelihoods only Coping with Prosperity

More Related