Married women’s added worker effect during the 2008 economic crisis—The case of Turkey

Abstract

This paper analyzes the labor supply response of married women as a result of their husbands’ job losses (‘added worker effect’). The study uses panel data from Turkey to test the presence of an added worker effect during the global economic crisis of 2008. Identification is achieved by an instrumental variable approach. In particular, an exogenous variation in the output of male-dominated sectors induced by the crisis is used as an instrument for the husbands’ unemployment. Results show that the probability of a woman participating in the labor force increases by 15–28% in response to her husband’s unemployment. However, the effect is not contemporaneous; rather, it appears with a lag of one quarter and only operates for two quarters. The effect is mainly driven by financially-constrained (less-educated and young) couples, which suggests the prevalence of an income effect in spousal labor supply decisions.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3

Notes

  1. 1.

    Characterizing the U.S. recession of 2007–2009 as a ‘mancession’, Starr (2014) gives reference to the substantially larger effects on the male unemployment rate. She documents that the gender-segregated characteristic of this crisis exists in many countries, primarily in the U.S., as well as during other recessionary periods.

  2. 2.

    The distinction between unemployment and non-employment is especially vague during recessionary periods (Starr 2014). The current analysis does not aim to distinguish between the two phenomena, although the focus lies on the non-employment of husbands (regardless of their job search behavior). The two terms are used interchangeably throughout the discussion.

  3. 3.

    The leading studies are Blundell et al. (2012), Cullen and Gruber (2000), Goux et al. (2014), Heckman and MaCurdy (1980), Maloney (1991) and Starr (2014).

  4. 4.

    See studies from the U.S. by Cullen and Gruber (2000), Heckman and MaCurdy (1980), Lundberg (1985).

  5. 5.

    See studies from France by Goux et al. (2014); from the U.K. by Barton et al. (1980); and from the U.S. by Juhn and Murphy (1997), Maloney (1987, 1991).

  6. 6.

    See studies from Japan by Kohara (2010); from Turkey by Baslevent and Onaran (2003), Degirmenci and Ilkkaracan (2013), Karaoglan and Okten (2012); and from the U.S. by Blundell et al. (2012), Mattingly and Smith (2010), Spletzer (1997), Starr (2014), Stephens (2002).

  7. 7.

    Household credits in Turkey are procyclical—positively correlated with output growth—and highly volatile over business cycles (Alp et al. 2012). The recurrent financial crises experienced by the country in the last two decades have deteriorated the functioning of credit markets and constrained the ability to borrow to a large extent. The increase in credit constraints during the 2008 crisis is thus considered one of the potential mechanisms behind the evidence of AWE (Lundberg 1985).

  8. 8.

    One may raise the point of time aggregation bias. If individuals change their labor market status more than once within a quarter or more than twice in two consecutive quarters, the aggregated (quarterly) transitions would be biased as the monthly transitions across states are suppressed. Returning to the previous example given in the text, if the individual becomes unemployed in April and switches back to employment in May and June, then she/he is assigned as being employed for the second quarter. Therefore, no transition would be recorded between the first and second quarters, despite three monthly transitions having occurred across the period. This could have caused a problem if monthly transitions had been overweighed in the sample. However, given the small difference between the number of monthly and quarterly basis transitions (i.e., 9761 quarterly transitions and 10,382 monthly transitions over the 4-year-panel period), time aggregation bias is not expected to be a concern for the current analysis.

  9. 9.

    The seventeen economic activities covered in SNA are Agriculture, Hunting and Forestry; Fishing; Mining and Quarrying; Manufacturing; Electricity, Gas and Water Supply; Construction; Wholesale and Retail Trade; Hotels and Restaurants; Transport, Storage and Communication; Financial Intermediation, Ownership and Dwelling; Real Estate, Renting and Business Activities; Public Administration and Defense; Compulsory Social Security; Education; Health and Social Work; Other Community, Social and Personal Service Activities; Private Housekeeping Services.

  10. 10.

    Agriculture is excluded from the sample because half of agricultural employment is comprised of unpaid family workers, the majority of whom are female (78%). Moreover, the crisis did not hit the agricultural sector, meaning that rural men did not face an unemployment shock due to the crisis. Therefore, AWE is rather an urban phenomenon, at least for the Turkish context.

  11. 11.

    As stated earlier, the crisis causes exogenous variation in the production level of some specific sectors, whereby this variation is exploited as an instrument for the husband’s unemployment. The inclusion of job losses that occurred before the crisis, arguably including voluntary separations, weakens the association between the proposed instrument and the unemployment. The results of the falsification test are available from the author upon request.

  12. 12.

    Since marital break-up is a possible result of a job loss, those who divorce, become widowed or change their partners are dropped from the dataset to rule out this potential outcome. They account for approximately 0.1% of the working-age population sample.

  13. 13.

    The results of the falsification test are not presented here but available from the author upon request.

  14. 14.

    A relevant example is the households in which both wives and husbands are employed. In Turkey, women’s labor force participation (and thus their employment) is strongly correlated with the educational attainment, and well-educated women are more likely to be married to well-educated men. Since well-educated people were less adversely affected by the unemployment shock induced by the current crisis, the sample selection—excluding the employed couples—is considered as a potential source of upward bias in estimation results. However, the fact that the educational profile and other observable characteristics of the selected sample are quite comparable to the population sample alleviates the concerns regarding the selectivity issue. Besides, the controls for the observed as well as unobserved heterogeneity used in the regression analysis help capture the permanent characteristics that might be related to the self-selectivity.

  15. 15.

    In practice, couple fixed effects function as individual fixed effects, given the way of constructing the couple’s sample on which the regression analysis relies. Specifically, the analysis sample is generated by merging the wife’s data with the husband’s data based on the spouse and household identifiers.

  16. 16.

    A similar pattern is observed for employment outcomes. While manufacturing and construction—followed by trade and transportation services—saw the severest decline in the employment rate between 2008 and 2009, the non-male sectors did not see a considerable change during the period.

  17. 17.

    The textile sector represents a counter-example in the manufacturing industry as the female labor force is relatively prominent in this sector (28%). Due to the data limitation on two-digit sectoral output information, the robustness of the results cannot be checked by excluding the textile sector from the male-dominated sectors. On the other hand, the employment share of the manufacturing industry in the male-dominated sectors is 18.1%, of which the textile share accounts for 13.1% (Keskin 2010). Following basic algebra, one can easily find that the female labor force in the textile sector accounts for less than 1% of total employment in the male-dominated sectors. The small share of the textile industry in the male-dominated sectors (2.4%), and hence the small share of the female labor force in these sectors avoids the concerns about the violation of the exclusion restriction.

  18. 18.

    One may question the peaks and troughs in the lines in Fig. 3 over the period, aside from the sharp change in the fourth quarter of 2008. Seasonal effects might account for most of those fluctuations as the trend in both lines is smoothed once the quarter dummies are included in Eqs. (3) and (4).

  19. 19.

    One could make a case for clustering standard errors at the quarter level. Given the low number of clusters, a wild-cluster bootstrap-t procedure is implemented, as suggested by Cameron et al. (2008). The p-value calculated barely increases and the effect remains significant at the 5% level. This eliminates the possibility of an inference problem.

  20. 20.

    As the validity of the instrument depends on the output variation induced by the economic shock, the correlation in the first-stage estimation weakens after the fourth quarter of delay, which roughly corresponds to the period when economic recovery starts. Given this, it is implausible to interpret the IV results since then.

  21. 21.

    One may want to know whether any institutional or policy change exists over the sample period that might have had an effect on women’s participation behavior. A relevant policy to consider is the social security premium incentives provided for employers in the concurrent period. The policy aimed to create new employment for women (aged over 17 years) and men (aged 18–29 years) regardless of their marital status (Ayhan 2013). Aside from this policy, there were crisis-based measures, such as short-term work arrangements, targeting all population sub-groups. Considering the broad target groups of either policy, the institutional changes are not expected to be a worrying issue for the current analysis.

  22. 22.

    In Table 4, the IV estimates of the control variables for the number of children (as well as for many others) become statistically insignificant (or less significant) when fixed effects are included in the specification, mostly due to the larger standard errors. Larger standard errors in fixed effects estimation indicate a great variation in the predictor variables across groups despite a minor variation over time within a group. An outcome would be less precise estimates even if the magnitudes of the coefficients are comparable.

  23. 23.

    The lower age bound is set to 25 to elaborate the impact of fertility decisions on women’s labor supply behaviors. The excluded age group (of 15–24) constitutes less than 10% of the total sample of married couples. The upper age bound is set to 44 to exclude people who are (close to being) eligible for retirement, given that those subject to the prior legislation can be retired at around the age of 45 (Law No. 4759 2002). Retirement decision may result in a postponement in labor supply response of women due to either the potential source of income (through the partner’s pension) or the complementarity in leisure between spouses. In addition, elderly women—who possibly had a long inactivity duration—might be discouraged to enter the labor force due to a lower probability of finding jobs, regardless of the existence of financial constraints. It is difficult to disentangle these factors and hence associate the lack of labor supply behavior of elderly women with a less binding liquidity constraint. Given this, it is deemed reasonable to leave the retirement-aged people out of the sample while examining the liquidity constraint channel.

  24. 24.

    Similar results are obtained when the sample is further restricted to couples in which both spouses are high-school graduates.

References

  1. Alp, H., Baskaya, Y. S., Kilinc, M., & Yuksel, C. (2012). Stylized facts for business cycles in Turkey. TCMB Working Paper 1202, Central Bank of the Republic of Turkey.

  2. Aran, M., Capar S., Sanalmis, H. M. D. O., & Uraz, A. (2010). Recent Trends in female labor force participation in Turkey (in Turkish). Working Report No. 2, World Bank and Turkey State Planning Organization, Ankara.

  3. Ayhan, S. H. (2013). Do non-wage cost rigidities slow down employment? Evidence from Turkey, IZA Journal of Labor Policy, 2(20), 1–23.

  4. Barton, M., Layard, R., & Zabalza, A. (1980). Married women’s participation and hours. Economica, 47(185), 51–72.

    Article  Google Scholar 

  5. Baslevent, C., & Onaran, O. (2003). Are married women in Turkey more likely to become added or discouraged workers. Labour, 17(3), 439–458.

    Article  Google Scholar 

  6. Bredtmann, J., Otten, S., & Rulff, C. (2014). Husband’s unemployment and wife’s labor supply- the added worker effect across Europe. Ruhr Economic Papers No. 484.

  7. Brown, M., & Manser, M. (1980). Marriage and household decision-making: A bargaining analysis. International Economic Review, 21, 31–44.

    Article  Google Scholar 

  8. Blundell, R., Pistaferri, L., & Saporta-Ekstein, I. (2012). Consumption inequality and family labor supply. NBER Working Paper 18445.

  9. Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2008). Bootstrap-based improvements for inference with clustered errors. The Review of Economics and Statistics, 90(3), 414–427.

    Article  Google Scholar 

  10. Chiappori, P. (1992). Collective labor supply and welfare. Journal of Political Economy, 100(3), 437–467.

    Article  Google Scholar 

  11. Cullen, J. B., & Gruber, J. (2000). Does unemployment insurance crowd out spousal labor Supply? Journal of Labor Economics, 18(3), 546–574.

    Article  Google Scholar 

  12. Dayioglu, M., & Kirdar, M. (2010). Determinants of and trends in labor force participation of women in Turkey. Working Report No. 5, World Bank and Turkey State Planning Organization, Ankara.

  13. Degirmenci, S., & Ilkkaracan, I. (2013). Economic crises and the added worker effect in the Turkish labor market. Levy Economic Institute Working Paper No. 774.

  14. Doss, C. (2013). Intrahousehold bargaining and resource allocation in developing countries. The World Bank Research Observer, 28.1, 52–78.

    Article  Google Scholar 

  15. Goux, D., Maurin, E., & Petrongolo, B. (2014). Worktime regulations and spousal labor supply. American Economic Review, 104(1), 252–276.

    Article  Google Scholar 

  16. Gunduz-Hosgor, A., & Smits, J. (2008). Variation in labor market participation of married women in Turkey. Women’s Studies International Forum, 31(2), 104–117.

    Article  Google Scholar 

  17. Heckman, J. J., & MaCurdy, T. E. (1980). A life cycle model of female labour supply. Review of Economic Studies, 47(1), 47–74.

    Article  Google Scholar 

  18. Horney, M., & McElroy, M. B. (1981). Nash-bargained household decisions: Toward a generalization of the theory of demand. International Economic Review, 22, 333–349.

    Article  Google Scholar 

  19. Juhn, C., & Murphy, K. M. (1997). Wage inequality and family labor supply. Journal of Labor Economics, 15(1), 72–97.

    Article  Google Scholar 

  20. Karaoglan, D., & Okten, C. (2012). Labor force participation of married women in Turkey: Is there an added or a discouraged worker effect? IZA Discussion Papers 6616.

  21. Keskin, F. (2010). Textile products manufacturing. Turkish Development Bank Report 107, http://www.kalkinma.com.tr/data/file/raporlar/ESA/ga/2012-GA/esamkitap/pdf/tekstil.pdf, Retrieved on 30.09.2015.

  22. Kizilirmak, B., & Sahin, H. (2007). Determinants of duration of unemployment insurance benefits in Turkey. Applied Economics Letter, 14(8), 611–615.

    Article  Google Scholar 

  23. Kohara, M. (2010). The response of Japanese wives’ labor supply to husbands job loss. Journal of Population Economics, 23(4), 1133–1149.

    Article  Google Scholar 

  24. Law No.4759 (2002). Law on Amendments to Social Security Law. Official Gazette numbered 24772 dated 1 June 2002.

  25. Lundberg, S. (1985). The added worker effect. Journal of Labor Economics, 3(1), 11–37.

    Article  Google Scholar 

  26. Lundberg, S. (1988). Labor supply of husbands and wives: A simultaneous equations approach. The Review of Economics and Statistics, 70(2), 224–235.

  27. Maloney, T. (1987). Employment constraints and the labor supply of married women: A reexamination of the added worker effect. Journal of Human Resources, 22(1), 51–61.

    Article  Google Scholar 

  28. Maloney, T. (1991). Unobserved variables and the elusive added worker effect. Economica, 58(230), 173–187.

    Article  Google Scholar 

  29. Mattingly, M. J., & Smith, K. (2010). Changes in wives’ employment when husbands stop working: A recession-prosperity comparison. Family Relations, 59(4), 343–357.

    Article  Google Scholar 

  30. Mincer, J. (1962). Labor force participation of married women: A study of labor supply. In H. G. Lewis (Ed.). Aspects of labor economics (pp. 63–97). Princeton, NJ: National Bureau of Economic Research, Princeton University Press.

    Google Scholar 

  31. Spletzer, J. R. (1997). Reexamining the added worker effect. Economic Inquiry, 35(2), 417–427.

    Article  Google Scholar 

  32. Staiger, D., & Stock, J. (1997). Instrumental variables regressions with weak instruments. Econometrica, 65(3), 557–586.

    Article  Google Scholar 

  33. Starr, M. A. (2014). Gender, added-worker effects, and the 2007-2009 recession: Looking within the household. Review of Economics of the Household, 12(2), 209–235.

    Article  Google Scholar 

  34. Stephens, M. (2002). Worker displacement and the added worker effect. Journal of Labor Economics, 20(3), 504–537.

    Article  Google Scholar 

  35. TurkStat (2013). Turkish Statistical Institute data base, http://tuikapp.tuik.gov.tr/isgucuapp/isgucu.zul.

  36. Woytinsky, W. S. (1942). Three aspects of labor dynamics. Washington, DC: Social Science Research Council.

    Google Scholar 

Download references

Acknowledgements

I am deeply grateful to Margherita Fort and Riccardo Rovelli for their guidance, advice and criticisms throughout my research. Special thanks to Abdurrahman Aydemir, Erich Battistin, Richard Blundell, Arnaud Chevalier, Fabian Lange, Hartmut Lehmann, Silvia Pasqua, Massimiliano Tani, Semih Tumen, Rudolf Winter-Ebmer and Kamil Yilmaz for invaluable comments, as well as Richard Forsythe and Derek Stemple for a thorough language editing. I also thank the seminar participants at the University of Bologna, IZA, Collegio Carlo Alberto, AIEL 2014 Conference in Pisa and EEA 2015 Conference in Mannheim. Last but not least, I wish to express my gratitude to the journal’s editor, Shoshana Grossbard, and three anonymous reviewers for constructive feedback that helped me improve the manuscript. All errors are my own.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Sinem H. Ayhan.

Ethics declarations

Conflict of interest

The authors declare no competing financial interests.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ayhan, S.H. Married women’s added worker effect during the 2008 economic crisis—The case of Turkey. Rev Econ Household 16, 767–790 (2018). https://doi.org/10.1007/s11150-016-9358-5

Download citation

Keywords

  • Spousal labor supply
  • Added worker effect
  • Gender
  • Economic crisis

JEL Classification

  • J22
  • J16
  • D13
  • E32