Real options and demographic decisions: empirical evidence from East and West Germany

Abstract

Iyer and Velu (2006) have convincingly argued that contemporary analyses of fertility behaviour fail to explain why a woman (or a couple) will choose to postpone childbirth, and in particular to consider the role of uncertainty in this regard. They have addressed this lacuna in the literature by using a real options approach to model fertility decisions by relating uncertainty experienced by individuals to the likelihood of childbirth. However, they did not present empirical evidence. Since the theory implies the existence of two offsetting effects of uncertainty on fertility decisions, a positive insurance effect and a negative option value effect, it is not easy to reject the theory on the basis of empirical analysis, when one of these effects offsets the other. We construct such a test for East (and also West) Germany during that country's reunification, which takes advantage of the fact that because of the country's strong welfare system, the insurance effect should be dominated by the option value effect, thereby suggesting that the net relationship should be negative. The results provide rather strong support for the real options link, especially for Eastern Germany.

Acknowledgements

The authors would like to thank Gert Wagner, Christian Schimitt, Michaela Kreyenfeld, Jennifer Johnson-Hanks, T. Paul Schultz, David Canning, seminar participants at the Max Planck Institute for Demographic Research and an anonymous referee for helpful comments. They are also grateful to DIW (Berlin) for the use of the GSOEP data. They remain responsible for all remaining errors.

Notes

¹ It is common in the demographic literature to argue that the postponement of childbirth can have a significant impact on the total fertility rate of a woman. Surprisingly, however, most studies confine their attention to the reasons for postponing birth of the first child alone (Happel et al., 1984; Chen and Morgan, 1991).

² Kalemli-Ozcan (2003) uses an overlapping generations’ framework to demonstrate that, if mortality rates are high, there is a precautionary demand for children, and when this uncertainty about the life expectancy of the child is reduced, there is a decline in fertility as well.

³ The equivocal nature of this proposed relationship contrasts with that of Ranjan (1999) who concludes that uncertainty necessarily reduces the likelihood of having children.

⁴ Note that there are no gains to be made by extending the analysis to smaller geographical units like the federal states. Hank (2001) demonstrates that while there is some variation in TFR across West German districts, the variations are small. In 224 of the 338 districts, TFR was between 1.28 and 1.55. By contrast, in the aftermath of the unification, the difference between East German (below 1) and West German districts (about 1.5) is large. Also, low fertility districts within West Germany, with TFR of less than 1.28, which could be of interest to us, are spread throughout the country (Hank, 2002; Fig. 1), thereby making broad north–south type comparisons infeasible.

⁵ See German Democratic Republic (1990).

⁶ Note that this does not mean that the past does not have any impact on this likelihood, since the existing number of children would influence the insurance value of having additional child and thereby the likelihood of childbirth (or its postponement).

⁷ Fixed effects probit estimators for short panels of eight periods (such as used in this study) are deemed to be biased (Greene, 2004).

⁸ The willingness of a woman to bear a child in period t is usually specified to depend on the number of children she has had prior to that time instead of the number of children in the household itself. However, since the time and budget constraints effectively facing the woman would depend on all the children in the household and given our purpose of analysing the incidence of childbirth period by period and the extremely high correlation between the total number of children in the household and the number of her own children, we control for all children in each household.

⁹ By 1995, 36% of the East German women who were employed prior to reunification lost employment, the corresponding figure for East German men being 23%.

¹⁰ In Table 2, we report the correlation matrices for 1998, a year that is approximately the mid-point of the post-unification period covered in our analysis. The correlation matrices for other years are quite similar and are available upon request.

¹¹ This result may be an artefact of the limited (3-point) scale on which this worry is measured and the relatively small variation in this measure across households.

¹² According to Bonin and Euwals (2001), prior to reunification, the labour force participation rate of East German women was nearly 80%. This rate of participation was comparable only to those in Scandinavian countries, and much higher than the labour force participation rate of women in West Germany where the negative impact of laws governing maternity leave and income taxation of couples is well established (Strøm and Wagenhals, 1991; Spahn et al., 1994). This trend persisted after the reunification, despite a significant decline in employment opportunities facing East German woman in the aftermath of reunification. In 2000, 72% of the women in East Germany were labour force participants, the corresponding figure for West Germany being 62%.

¹³ This does not necessarily imply that the probability of unemployment decreased.