1. DO FINANCIAL INCENTIVES AFFECT FERTILITY? Alma Cohen Rajeev Dehejia Dmitri Romanov Israeli National Council for Economy and Society Feb. 12, 2008

2. Subject Paper investigates empirically whether financial incentives, and in particular governmental child subsidies, affect fertility. Takes advantage of a comprehensive individual- level panel dataset of fertility histories and detailed individual controls for married Israeli women.  Data covers 1999-2005 – a period with substantial variation in level of child subsidies but no changes in eligibility.

3. Main Findings Child allowances have a significant positive effect on fertility. This effect:  Is concentrated in the bottom half of the income distribution.  Is present across all religious groups, and is especially strong for ultra-Orthodox Jews and Arab Muslims.  Is stronger for new immigrants.

4. Motivations for Study (1) Ever since the now-canonical fertility model of Becker (1960), there has been significant interest among researchers in whether and to what extent fertility responds to financial incentives. Some believe that fertility decisions are largely shaped by social, religious, and cultural forces (Gauthier 1996), and that financial incentives – at least incentives of the magnitude used by governments – have little effect.

5. Motivations for Study (2) Whether fertility is responsive to financial incentives has significant policy implications. Facing a low birthrate below the replacement level, many developed economies have adopted policies that “encourage procreation:” Many countries have adopted different kids of policies. The main policies: –Child subsidies –Tax Benefits –Child care subsidies. –Maternity leaves Despite the prevalence of these polices, the evidence of their impact on fertility has been thus far inconclusive.

6. Fertility Rate around the World

7. Prior Literature – Theory Becker (1960): fertility decisions can be analyzed as an economic decision – the demand for children is a demand for a special type of capital good and is sensitive to their costs. Model extended in various directions (e.g., Cigno 1986, Manski and Mayshar 2003).

8. Prior Empirical Work (1) Cross-Country Studies: [e.g., Demeny 1986, Gauthier and Hatzius 1997, Whittington, Alm, and Peters 1990] Problems with Cross-Countries Studies: Hard to measure the incentive effects of transfer in a very heterogeneous population. Policies across countries are not uniform

9. Prior Empirical Work (2) Individual Data studies: [Milligan 2005, Laroque and Salanié 2005, Whittington 1992, Blau and Robins 1989] Relative to them, our dataset has three advantages: –Our rich CBS dataset enables us to control for a rich set of individual/family characteristics. –Our six-year period enables examining multiple changes in allowance level in both directions. – The unanticipated and large 2003 reduction provides us with a good setting for a diff-in-diff specification. None of the earlier studies using Israeli data (Manski and Mayshar 2003, Frish 2004, Schllekens 2006) is able to use such a rich dataset or uses the same variation in child allowances.

10. Institutional Background: The Child Allowance in Israel (1) Was first introduced in 1959 and since has undergone many changes. From 1975 the benefits became universal (the so-called Ben-Shahar reform) – and it was given to families for each child under age 18. With the one requirement that at least one member of the family served in the army (which excluded all Arab Muslims, and some of the Bedouim and the Druze). This “veteran” requirement was phased out between 1994-1996.

11. Institutional Background: The Child Allowance in Israel (2) The period we study (1999-2005) has many changes in the level of child allowances but not in eligibility and coverage. During this period there were two significant policy reforms: 1)The Halpert Law (implemented in the beginning of 2001 and lasted for a year and a half): Increased the benefit for fifth and higher-parity births by 33 to 47 percent. 2)The “Netanyahu” reform (in June 2003): The largest and most unanticipated change in child allowance.

12. Institutional Background: The Child Allowance in Israel (3) The 2003 Reform: Under the 2003 reform bill, mothers of children born after June 2003 receive an allowance equivalent to that of the first two children in the family regardless of their birth parity. Established a transition regime for children born prior to the reform – a gradual decrease over the subsequent seven years so that by 2009 every child will receive a uniform allowance irrespective of birth parity. Although child allowances were reduced across the board, the biggest reduction in benefits was for families with many children.

13. Institutional Background: The Child Allowance in Israel (4)

14. Institutional Background: The Child Allowance in Israel (5) For example: a family with 4 children received: These are meaningful changes for many of the affected families – especially bearing in mind that the highest fertility groups (the ultra-Orthodox and Arab Muslims) are also the poorest. For existing children For new born In 20021,259 NIS782 NIS After the 2003 reform937 NIS144 NIS

15. Institutional Background: The Child Allowance in Israel (4) Comparison to Allowance Levels in 22 OECD Countries: In 2003, among 22 OECD countries: –14 countries paid child cash allowance to all families regardless of their income –3 countries paid allowances that decreased with income. –3 countries paid allowances only to low-income families –2 countries did not make such payments. In most of the countries that do pay child allowances, the marginal per child allowance increases with the number of kids.

16. The Data: Includes all married women under 45 with at least 2 kids. We merged number of datasets, each maintained by the Israeli CBS Fertility History and Basic Demographic Characteristics Education Religion Income

17. The Data: Fertility History and Basic Demographic Characteristics From the Population Register’s data –information for both mother and husband on: date of birth country of origin year of immigration for individuals not born in Israel country of origin and year of immigration for parents of Israeli-born women and husbands. number of children and their birth dates locality identifier

18. The Data: Education Data on education of mothers and husbands was complied from various data sets: (1) From the administrative records of Israel's higher- education institutions – information on both mother's and husband's most recent academic degrees (if graduated in Israel). (2) From the school registry record (created when parents register their children in public schools and public kindergartens) – years of schooling for parents who had children already enrolled in primary school or public kindergartens and reported information about their years of schooling. (3) From the immigration registry – years of schooling. Combine the different sources by generating a categorical variable (1 for primary school; 2 for high school graduate; 3 for college, and 4 for post-graduate education).

19. The Data: Religion We infer the degree of religiosity for the Jewish population by using information on the kind of kindergarten and school that their children attend (secular, national religious, or ultra-orthodox).

20. The Data: Income Obtained from the matched employer-employee database, based on income tax files. For both the mother and the husband, we have the following: Employment status (self-employed or wage earners) Number of jobs held Number of months worked Gross income Income tax We use these data to create socio-economic controls.

21. Descriptive Statistics (1) Summary Statistics

22. Descriptive Statistics (2) Distribution of number of kids for married women age 40-45 by Religious Group for year 2005

23. Fertility Trends, 1999-2005

24. Empirical Strategy We examine the relationship between the fertility decision and child allowance for a marginal child. Our dependent variable is an indicator which is equal to 1 if the mother became pregnant in that year and 0 otherwise. Our main variable of interest is the child allowance for the incremental child. Include a broad set of household controls: education, income, work status, religion dummies for the number of previous children, macro control, and time trend.

25. Results (1) Results (1) Average effect of Child Allowance on the Probability of Pregnancy Probit regression Dependent variable: becoming pregnant dummy

26. Results (2) Results (2) Average effect of Child Allowance on the Probability of Pregnancy by Religious Group

27. Results (3) Results (3) Average effect of Child Allowance on the Probability of Pregnancy by Socioeconomic Group

28. Results (4) Results (4) Average effect of Child Allowance on the Probability of Pregnancy by Religious and Socioeconomic Group

29. Results (5) The Differential Effect of Child Allowance on New Immigrants

30. Diff-in-Diff: The 2003 Changes Look at changes in fertility around the policy change in 2003. We compare the fertility of low- income women in a three- or four-month window before and after the policy change using top income quartile women as a comparison group:

31. Results (diffs-in-diffs)

32. Simulation (1) No. of Births in Alternative Scenarios

33. Simulation (2) No. of Births in Alternative Scenarios

34. Conclusions Child allowances have a significant positive effect on fertility, This effect:  Is concentrated in the bottom half of the income distribution.  Is present across all religious groups, and is especially strong for ultra-Orthodox Jews and Arab Muslims.  Is stronger for new immigrants (who may have less resources to rely on). Future work: In several years, it would be worthwhile to do a follow-up study to investigate the long-term effects of the 2003 reform.

35. Description of Variables

36. Year Dummies (Secular)

37. Year Dummies (Religious)

38. Year Dummies (Ultra-Orthodox)

39. Year Dummies (Arab-Muslim)