SP DISCUSSION PAPER NO.9903 20118 A Bundle of Joy or an Expensive Luxury: A Comparative Analysis of the Economic Environment for Family Formation in Western Europe Pierella Paci January 1999 ronttion AOR MARKETS PENSIONS, SOCIAL ASSISTANCE T H E W O R L D B A N K Table of Contents Page Table of Contents i List of Tables ii List of Figures iii Executive Summary iv 1. INTRODUCTION 2 2. CHANGES IN FAMILY STRUCTURE. 2 3. THE ECONOMIC THEORY OF THE FAMILY 6 3.1 The determinants of fertility rates. 9 3.2 Policy implications. 10 4. A COMPARISON OF THE GENDER PAY GAP AND WOMEN CONTRIBUTION TO 10 FAMILY BUDGET IN EUROPE. 5. CROSS-COUNTRY DIFFERENCES IN STATE POLICIES TOWARDS THE FAMILY. 12 5.1 Exploring the gender and demographic implication of alternative tax 13 systems. 5.2 Financial support for children 16 5.3 Maternity and parental leave benefits 18 5.4 The provision and cost of childcare. 23 6. THE COST OF CHILDBEARING IN SELECTED EUROPEAN COUNTRIES. 26 6.1 Estimates of cross-country variations in the cost of childbearing 27 6.2 State policies and women's labor market participation. 27 6.3 State policies and women's earnings. 32 7. EMPIRICAL EVIDENCE ON THE EFFECTS OF STATE POLICIES TOWARDS THE 33 FAMILY. 7.1 Do welfare benefits increase fertility? 33 7.2 Do welfare benefits increase the probability and the length of lone 37 parenthood? 7.3 Summarizing the empirical findings. 38 8. CONCLUSIONS 39 9. REFERENCES 44 i List of Tables Page Table 1 Activity rates of women 1970-1995 3 Table 2 Total Fertility Rates in the EU-15 5 Table 3 Elements of national systems of earned income tax in the EU 14 Table 4 Comparison of the level of financial support for children, 1993 17 Table 5 Maternity leave benefits duration and pay, paternity leave and 20 take-up rates, and -hild-care leave Table 6 Public provision . hnild-care in EU- 15 countries, 1994 24 Table 7 Welfare benefits and female employment: a review of the 29 econometric analyses Table 8 Welfare benefits and fertility: a review of the econometric studies 35 Table 9 Welfare benefits and family structure: a review of the 37 econometric studies Table 10 International comparison of state policies towards the family 40 ii List of Figures Page Figure 1 Activity rates of women (1970-95) 4 Figure 2 Trends in total fertility rates in the EU (1950-90) 6 Figure 3 Gender pay ratio in selected countries (1970-95) 11 Figure 4 Comparison of the level of financial support for children 18 iii Executive Summary This paper provides a description of the diversity of current policies towards the family across the European Union and an account of the current 'state of the art' on the effects of these policies on demographic and labor market behavior. This remains a relatively unexplored issue despite its obvious importance in the Europe of the late 1990s, where public concern about low levels of fertility is growing and governments are increasingly questioning the current system of state support to the family. Traditionally, when reforms to the tax and benefits system are considered and introduced little attention is paid to their potential effects on gender relations and demographic behavior and when the issue of gender equality in the labor market is raised its potential demographic effects are neglected. The implicit assumption is that the tax and benefit system (and any change to it) is somehow gender- and family-neutral and that the gender relations do not affect family formation. The main aim of this paper is to fill this gap by exploring the extent to which the course of demographic trends within a country depends on State policies towards the family and its commitment to gender equality. The extent of the variation amongst EU countries in this area, coupled with the relatively homogenous attitude towards the family and women's position within the household, makes the comparison worthwhile. The starting point of the analysis is the rather puzzling finding that in the 1990s amongst the EU- 15, the countries with the highest fertility rates are not those where the tradition of large families is longstanding, relations are stable, women's participation in the labor market is low and the tax system is based on a 'strong breadwinner' model. On the contrary, fertility is highest in Scandinavia where marriage is a rare and fragile institution and women have a relatively strong attachment to the labor market. So what is the magic formula for the Scandinavian high fertility/high participation solution? Our cross-country comparison points to two crucial factors: the relatively low- cost of childbearing and the more equal division of labor within the household. The role of the State and of economic policies in these areas is paramount. It is therefore not surprising that women appear to be most able to combine childbearing with employment in countries where the greatest attempt to 'socialize the costs of familyhood' and to adopt family-friendly policies have been made. The evidence from Europe thus suggests that egalitarian work patterns, egalitarian parenthood and higher fertility move in the same direction. iv A bundle of joy or an expensive luxury: a comparative analysis of the economic environment for family formation in Western Europe. by Pierella Paci Department of Economics, City University, London and HDNED, The World Bank ABSTRACT In the Europe of the 1 990s low fertility rates are a prime concern. Contrary to the general view that associates declining fertility with increasing women's participation in paid work, this paper shows that amongst the EC-15 countries, those where women are more active in the labor market are also those where fertility rates are highest. These are not the countries where large families are a tradition. On the contrary, they are those where women's contribution to the household budget is the highest and where the State provides the strongest support to family formation and gender equality. The low cost of childbearing allows women in these countries to be both mothers and earners. The others may be forced to choose. ACKNOWLEDGEMENTS: This report was prepared for the Seminar on "Maternity, Childcare, Social Insurance and the Labor Market" organized by the HDNSP Labor Market Group in June 1998. I thank the discussant and the participants to the Workshop for helpful comments. A special thanks goes to Zafiris Tzannatos for fruitful discussions, helpful comments and financial assistance. 1. Introduction The transformations undergone by the family in the developed world since the 1960s have attracted much attention in both academic and non-academic circles. Amongst the factors involved two have received particular attention and have often been linked in a simplistic way: the increase in the proportion of women in paid employment and the fall in fertility rates. Economic theory suggests the story is more complicated. The economic theory of fertility, in its New Home Economics version, claims that recent trends in fertility can only be fully understood if seen as integral part of the changes in family structure that has been taking place since the 1960s. These include the reduced stability of unions and the development of non-marital cohabitation as well as the decline in fertility and the increase participation of women in the labor force. The first three of these trends have generated widespread concern and the issue of the potential role of the state in this context has gained increasing priority in the political agenda of many industrialized countries. Although this paper is designed to address the whole question of state policies towards the family, its focus is the demographic and labor market effects. It uses the diversity of policies and family structures found in the EU countries to address questions such as the extent to which cross-country differences in patterns of fertility and employment can be related to differences in the tax and benefit system in general, and in the degree of state support for the family in particular. This is an important but often neglected issue. Traditionally, when reforms to the tax and benefits system are considered and introduced little attention is paid to their potential long-run/general equilibrium effects on demographic behavior. The implicit assumption is that the tax and benefit system (and any change to it) is somehow gender and family neutral. The realization that the future course of demographic trends could be dependent on state policies is a relatively new discovery and their potential links remain a relatively underdeveloped area of research. It is hoped that this report will shade some light on the subject by identifying paradigms from which lessons for developing countries (especially those in transition) can be derived. It begins with a review of the changes in family structure occurred in Europe since the 1 960s. It then moves on to summarize briefly the economic theory of the family and its main implications for women's demographic and labor market decision making. A major factor in these decisions is the comparative advantage of men and women in paid work. Section three presents a comparison of the women's contribution to the family's budget and the gender pay gap across Europe. Cross-country differences in state policies towards the family and in the gander bias of the tax system are discussed in section four and their implications for the cost of childbearing, gender and demographic highlighted in the following section. Finally the empirical evidence on the effects of alternative policy is reviewed in section 5. 2. Changes in family structure. Paid work never was the exclusive province of men, but one of the most striking developments in the European labor market during the last 30 years has been the increase in the proportion of women in paid employment. 2 Table 1: Activity rates of women 1970-1995 1970 1980 1990 1995 High activity Sweden 76.1 Denmark 36.9 45.2 77.6 73.3 Finland 70.00 UK 31.3 35.9 66.1 66.0 Medium activity Germany 30.3 31.5 57.6 61.3 France 28.9 32.6 58.0 60.6 Portugal 18.1 35.4 57.1 59.1 Netherlands 22.7 52.4 58.9 Belgium 24.9 30.6 46.1 51.7 Low activity Spain 17.3 20.0 40.6 44.9 Ireland 20.1 41.9 Greece 21.2 42.6 44.3 Luxembourg 26.0 42.4 44.1 Italy 21.9 26.0 43.2 42.5 Source: Eurostat, various years. Table 1 shows that in 1995 the E- 15 countries could be clearly divided into three groups. In Scandinavia and the UK more than half of the women of working age were in paid employment. At the other extreme were the Southern European countries, Ireland and Luxembourg with activity rates lower than 40%. In the remaining countries, including Portugal, women's activity rates were somewhere between 40% and 50%. Figure 1 shows that, in the last 30 years, women's participation in paid work has increased considerably across Europe. In 1970, of the countries for which comparable data for is available, only in Denmark over a third of women were in paid work and indeed in the others activity rates were less than 25%. 3 Fig 1: Activity rate of women, 1970-1995 31 995 Ge~~~~~~~~~~~~~~~~ ~r31 990 M1980 0 20 40 60 80 10 0 The increase in women participation in paid employment has often been indicated as a major determinant of the phenomenon often referred to by popular commentaries as "Europe's falling birth rate". It may therefore come as a surprise to find that, although below the replacement ratio, the average EU fertility rate has changed little since the early 1 970s. Indeed, with few temporary exceptions, the average family size in Europe has been about two children for the last sixty years. However, as Figure 2 points out, the average number hides large cross-country disparities. The total fertility rates (TFR) in some countries have indeed increased substantially between the 1980s and the early 1990s and, although it has somewhat declined since, they remained the highest in Europe (Table 2). The Southern European countries, on the other hand, have behaved differently. Italy and Greece shared consistently low fertility during the 1 960s and 1970s, while during that period, in Spain and Portugal fertility remained high. However, in Italy the decline in fertility started well before that of the other Southern countries and come close to an end by 1985. In Greece, Spain and Portugal the fall started around 1980 and has not yet ended. Spain now has the lowest TFR in the world (1.1) (Table 2). 4 Table 2: Total Fertility Rates in the EU-15 TFR 1996 Scandinavian countries Finland 1.76 Denmark 1.75 Sweden 1.61 Pro-natalist countries Luxembourg 1.76 France 1.72 Belgium 1.55 Italy 1.22 Southern countries Portugal 1.44 Greece 1.31 Spain 1.15 Other countries Ireland 1.91 UK 1.70 Netherlands 1.50 Source: various Even a superficial comparison of Tables 1 and 2 shows that the conventional wisdom of a inverse relationship between women participation in paid employment and fertility a rates in the European context is not supported by empirical evidence. Generally speaking, amongst EC countries, those with relatively high women participation in the labor market also experience comparatively high fertility rates. Moreover, participation has continued to increase even in periods of raising fertility. Of course cultural differences and changes of attitudes over time may accounts for much of the cross-country differences and intra-country trends. Many, for example, have explained the declining fertility of the Southern European countries in terms of a late surge of feminism. However, economic theory suggests that many other factors may be at stake and other trends should be considered. 5 Fig 2 TFR trends, 1950-1990 1990~ E1970 0 1 2 3 4 5 TFR trends Other trends of importance refer to the occurrence and stability of marriage. In particular, marriages are occurring later in people's lives, and in some countries people will forgo marriage entirely. Contrary to popular believe, only a small part of the decline is due to increase in cohabitation without legal marriage, showing that people are waiting longer to enter any cohabiting partnership. For instance it has been shown that less than half of the decline in marriage experienced by British women in their twenties in the I 970s and 1 980s can be accounted for by increased cohabitation (Ermisch, 1994). Divorce is also on the increased and this has made the single contribution to the growth in one-parent famrilies (Ermisch, 1991). Although no European country has reached the levels of the United States, current tr ds suggest that over 60% of the marriages end in divorce in Sweden and over half in the UK. The economic theory of the family suggests that these changes are interrelated and any single one can be better understood in association with the others. It also suggests that they are unlikely to provide the full picture and the diversity of the economic and policy environments in which employment and fertility decisions are taken in also likely to be very important. In 6 order to identify those factors is therefore important to outline briefly the economic perspective on family formation. 3. The economic theory of the family The New Home Economics theory of family formation treats the household as an economic agent whose objective is to maximize the production of 'domestic output' using the optimal combination of its members' time and the purchase of goods and services from the market. In a one-person household the maximization would be subject to the person's time constraint and an income constraint and would entail the allocation of the individual's time between home production and market work. The comparative analysis of this model suggests that if two individuals A and B producing the same domestic output, differ in their wage rate and the non-earned income they will also differ in their allocation of time. In particular, if A earns more but B enjoys higher non-earned income, the former would allocate less of his time to home production and purchase more goods and services then the second. However, this is not the end of the story. If A bought some of her required goods and services from B (i.e., some of his time) each person's domestic output would increase. As A's higher earning power gives her a comparative advantage in goods, there are gains from trade. But if A and B combined into a two-person household, they would do even better because the latter could specialize in home production and the former could supply more of his or her higher paid time to the market. The combined household produces more that the sum of what A and B could have produced separately. The gains from 'merging' are directly related to the difference between their wage rates. Therefore the incentive to 'combine' into a household increases with the earning differential between its members. Higher joint output does not necessarily imply higher individual welfare. Decisions within households must, however, take into account the individual tastes/preferences of each member and there is potential for conflict between the partners. The structure of the AB household therefore emerges as the outcome of a cooperative bargaining process. The welfare that A and B could achieve in the single-person household (or perhaps in forming a household with person C) represents the outside options for the two parties. Anything that alters the payoffs to the 'merge' changes the relative bargaining power of the parties within the household, the welfare gain associated with forming the household and therefore the likelihood of the it being formed. A major determinant of the pay-off to 'combining' is the differential treatment of the one-person and the combined households in the tax and benefit system. Thus taxes and benefits may be instrumental in the creation of the 'household' and in the molding of its structure. Assuming A and B decide to merge in the household AB. How long would they stay together? In the case where A and B had predicted correctly the payoffs from 'combing', the household would survive for as long as these payoffs do not change. But, in a world of uncertainty, A and B may have miscalculated the payoffs. In this case the household will dissolve if the gains from trade were found to be negative. Thus in this context households dissolve because the gains from household formation are found to be less than anticipated or disappear over time because of new developments and information. Therefore the theory suggests that factors that 7 reduce the expected welfare within the household and/or increase the outside option, will increase the probability of the household breaking-up. Allowing for the possibility of household dissolution has important consequences for the conclusions of this theory. A crucial finding of the economic theory of the family is that for the household's welfare to be maximized at least one of its members must specialize fully in either market work or home production (Becker, 1981; Cigno, 1991). In the case in which both persons have the same initial endowment of earning power the optimality of complete specialization arises from the element of 'increasing returns' created by the accumulation of human capital in market work via 'on-the-job' investment. When the household members manifest initial differences in earning power, as in the case of our persons A and B, the superiority of complete specialization as an outcome is even more obvious. However, if employment experience increases earnings, the person specializing in home production forgoes this earnings enhancing investment. By entering the 'combined' household this person therefore accepts lower earnings that he/she would have achieved in the one- person non specialized household. The contrary is true of the other person who, by specializing in market activity, is able to improve his/her earnings faster than it would have been possible under the one-person household. In the simple model with perfect information this does not matter a great deal since once formed the households stay together, unchanged in structure, indefinitely. In this context, the members that have specialized in home production never suffer from the reduction in their earning potential since they never need to go back to market activity. However, if the possibility of household dissolution is allowed for, the likelihood of having to go back to market activity must also be considered. The negative effect on future earnings of specializing in home production must be taken into account when evaluating the long-run welfare of the household members. In this context full specialization may no longer be the optimal outcome. Thus any reduction in the expected gains for household formation may begin a vicious circle where the risk of the household dissolving is increased, specialization is reduced and this reduces the gains from cooperation, the risk of breaking up is then increased and specialization further reduced. In the treatment above we have deliberately maintained a gender-neutral approach and in our example we have refereed to A, the member with higher earning power, as 'she' and to the other member, B, as a 'he'. However, there is ample empirical evidence to suggest that, at least amongst the older generations, women have comparative advantage in home production and men hold higher stocks of earnings enhancing characteristics. Then this theory provides the economic rationale for what is shown as the 'breadwinner' model of division of labor according to which, in a combined household, the man specializes in market work and the woman is responsible for home production. It also explains why, in times of increasing uncertainty about the stability of the household, women have appeared increasingly reluctant to fully specialize. Since the only readily available measure of the propensity for households to break-up, is the divorce rate, the strong empirical relation that this variable appears to have with women employment, confirms the prediction of this theory (Ermisch, 1996). 8 The negative effect of the incidence of divorce in the society does not only discourage household formation and specialization; it also discourages childbearing. The costs of bringing up children are much higher for single-person households then for couples and the benefit from having children are substantially reduced for the absent parent who has limited access to them. But children represent an important source of the benefit of household production. The current trends in these crucial variables, as highlighted above, therefore, reinforce each other. Having fewer children and having them later in life lowers both the incidence of marriage and raises the divorce rate by reducing the expected benefits from household formation. It also increases women's participation in employment. 3.1 The determinants offertility rates. However, the uncertainty surrounding the stability of the household is only one of the many possible factors affecting fertility. The economic theory assumes the households to be utility maximizers with a utility function increasing in their own consumption but also in the number of children and the human capital of each child. In other words, parents are assumed to derive utility from having children and to care about their well being, with well being defined in terms of future earnings. Earning capability is enhanced by investment in human capital. Parents can invest in children's human capital by increasing the proportion of time they devote to children or by purchasing goods and services from the market. A given level of human capital can be obtained with different combinations of time and services. The parents' objective is to maximize utility subject to an intertemporal budget constraint given by the condition that the sum of the expenditure on their own consumption and on the acquisition of children's human capital cannot exceed their wealth. Thus, given the preference, richer households tend to spend more on children but this does not mean that they have larger families. In the assumption that within a household all children are treated the same, the total expenditure in children's human capital depends on the cost per unit of human capital, the number of units acquired per child and the number of children. Becker (1981) points out that this creates an important interaction between family size and amount invested in each child. Indeed, given the way the budget constraint is formulated the 'shadow price' of an additional child is proportional to the sum invested on each child and viceversa. This interaction has some important implications. The first is that family size may be highly responsive to exogenous changes in any of the three components of the budget constraint. So, for example, a small increase in the amount spent per child - say due to increased skill requirements in the labor market - leads to a substantilal reduction in the number of children, since it raises their shadow price. This reduces the shadow price of human capital investment so that expenditure per child increases further and so does the shadow price of children. Similarly, even a small exogenous decrease in fertility - say because of lower contraception costs - can increase per- child human capital investment and depress fertility further. The second is that - although children are normal goods - the income elasticity of fertility may appear to be negative. This is because the shadow price of children (i.e., the human capital investment per child) changes with income and indeed may have higher income elasticity than family size. 9 The cost per unit of human capital, on the other hand, is given by the cost of its components, parental time and purchased goods and services. The latter, in this case, are given predominantly by the cost of childcare and educational activities provided outside the household. If some degree of specialization exists within the household, parental time is primarily the mother's time. So the unit cost of children's human capital is directly related to the cost of childcare services and to the cost of the mother's time: the wage she could have earned in employment and the loss of earning-related benefits, such as pensions. If the cost of children is assessed in terms of cost of parental consumption, the cost of both additional children and additional human capital investment per child increases as the real wage of women, relative to men, increases. 3.2 Policy implications. What does the theory tell us? In short, the economic theory of the family suggests women's demographic and labor market behavior to be affected by a number of common factors. These factors, however, operate in opposite directions: fertility is expected to increase with the individual propensity to create stable households while women employment declines in it. In turn, the individual's propensities to create and maintain a combined household increases in the joint payoffs to specialization (i.e., the extent of the gains from trade) and in the share s/he is likely to obtain. The former increases with the difference in earning power between the individuals involved. The latter rises with the potential individual's welfare outside the household. They are both dependent upon the partners' absolute and relative earnings and on the way in which the tax and benefit system treats single-individual and combined households respectively. In addition, the degree of specialization chosen by women within the combined household depends on the stability of the household, the structure of the labor market and on the degree to which paid employment is compatible with the women's responsibilities as provider of home services. These conclusions are reinforced by the analysis of the fertility that centers on the concept of 'cost of children'. This points to the gender gap and to the cost of childbearing as major determinants of women's demographic and labor market decisions. Thus one would expect cross-country differences in these variables to be reflected in variations in fertility and participation rates. 'What follows explores this issue. 4. A comparison of the gender pay gap and women' contribution to family budget in Europe. In all EU-countries the average woman contributes less than half of the total household income. This is the result of two factors: the shorter hours worked by the typical working and the lower hourly wages she earns. Although large cross-country variations in gender differences in working patterns exist, the average female employee in the EU12 works eight hours less per week than the average man (Rubery et al, 1998). She also receives a lower basic hourly wage and she is less likely to receive overtime or unsociable hours premiums (Rubery et al, 1995). Women's contribution to the household income is relatively high in Germany. Indeed in the former East Germany in 1988 women contribute nearly half (44%) of the household net income of dual earner couples and around 40% of all households budget. For 10 West Germany, however, the values are considerably lower, 38% and 18% respectively. In the UK, women's contribution to the budget of married couples aged 24-55 years rose from 16% in 1979 to nearly 20% in 1991 (Machin and Waldfogel, 1994). Similar trends and similar shares of household income came from women's earnings in Belgium and Luxembourg (Meulders, 1995). However, due to the differential role of unearned income and transfer payments in these countries, while women earnings account for only a quarter of all households earnings of in the UK in Belgium and Luxembourg they constitute over a third. Women contribute a higher share of household income in Spain (25%) but a lower one in Greece (16%) (Molto', 1995; Cavouriaris et al. 1995). Although it is always difficult to draw conclusions from a rough international comparison of this type, a clear trend for an increase in women's contribution to the household income is emerging across Europe. This trend is the result of a combination of factors. One is the high and persistent occurrence of long-term unemployment amongst men and the related increased reliance on unearned income and transfer payments. The others are the increased propensity for younger generations of women to follow a pattern of higher and more continuous involvement in the labor market coupled with the decline in the average gender pay ratio which all European countries have experienced since 1970. The latter point is evident from Figure 3, which shows the trends in men's wages relative to women's in some selected EU countries since 1970. Because of the varying definitions and coverage of the series presented (see note to Figure 3), their relative levels cannot be taken as seriously as the speed of the transition towards more equal rates of pay that they all display. 11 Fig 3: Gender pay ratio Men's to women's wages 2.01 1.81 1.6 1.4 ;>- % X 1.2 E^.<. 1.0 70 75 80 85 90 95 UK Belgium France West Germnany........ Finland + Denmark Sweden . Ireland Netherland Virtually all the European countries for which data is available have seen some reduction in the men's lead over women's pay since the 1 970s. However, the reduction was very limited in West Germany, gradual in Sweden and France and sharp and concentrated in a few years around the time of an Equal Pay initiative in others (e.g. UK and Denmark). Thus, by 1995 a clear distinction had emerged between a group of countries where the men's lead was less than a quarter of women's wages and the others. The former group included the Scandinavian countries, France and the UK. These countries also has the strongest legislative framework for equal pay for men and women (Joshi and Paci, 1998). Taking the example of Sweden, its institutional framework includes: * collective bargaining agreements effectively mandating equal hourly pay for part-time work, * labor and management initiatives, implemented as early as 1960 to phase in the elimination of negotiated pay disparities between men and women, * an equal pay law passed in 1980, instituting an Equal Pay Ombudsman to supervise implementation. The finding that in 1995 these countries also had the highest female activity rates in the EU- 15 is not surprising. What is more of a surprise is that they also experience the highest fertility rates. The explanation for this apparent contradiction may be found in the fact that, with the exception of the UK, these countries share an established tradition of 'family friendly' policy, stronger matemity leave, parental leave, part-time protections, and child care policies than the others. These policies affect the magnitude of the costs of social reproduction and the way they are shared between the state and the family, and between men and women within the 12 family, as reviewed by Folbre (1994). The remainder of this paper looks at these issues in more details. 5. Cross-country differences in state policies towards the family. The government has long intervened in family life, through specific legislation, the provision of services and the differential treatment of the family in the tax and benefit system. A useful starting-point for cross-country comparisons of the state treatment of the family is the distinction of countries into strong, modified and weak breadwinner states. The distinction is based on the extent to which the policies of each country reflect the presumption that households are, or should be, organized according to the male breadwinner model (Lewis, 1992). This is reflected in (i) the degree to which the tax and social security system treats women primarily as mothers, wives or workers (ii) the extent of childcare support offered. A strong breadwinner state treats all women as dependent wives. Married men receive favorable treatment in the assumption that they support a depend spouse and children (e.g., tax relief) and women are given derived benefits on the basis of their marital status (e.g., health care, pension). They role is primarily that of wives and mothers and their employment is discouraged (or not encouraged). In contrast the assumption underpinning the weak breadwinner state is that all fit adults (not in full-time education) are employed or looking for work. Women in couples are treated as individuals rather than as wives or mothers. Employment of mothers is facilitated through a combination of childcare subsides and leave for care responsibilities. The modified breadwinner model sits happily between these two stools with mothers deriving some benefits from household status and others from labor market status. Classifying countries according to the model they adopt is not an easy task and there is some disagreement in the literature about the classification of some countries (Rubery et al, 1998). However, there is general consent that Ireland and Germany represent examples of strong breadwinner states, that France is a modified state and that the Scandinavian countries adopt a weak breadwinner model (Duncan, 1995; Sainsbury, 1994). Recent years have seen contradictory changes in the model adopted by all European countries. On the one hand, growing concern about equal treatment of men and women - voiced in a number of EU Directives - and the increasing number of female lone mothers has moved countries away from the strong breadwinner model. On the other the continuos pressure on government to reduce public expenditure has made it difficult for women to obtain independent benefit entitlements. Thus the reforms implemented in recent years have often been piecemeal and have lead to contradictions in the tax and social security systems of many countries. For example, the tax system may reflect a strong breadwinner logic and the benefit system be based on a weak breadwinner principle. 5.1 Exploring the gender and demographic implication of alternative tax systems. A tax system is characterized by three factors: the selection of the tax unit (i.e., individual or household), the definition of taxable income (earnings minus tax allowances and reduction), and the structure of the tax rates. The potential gender bias of a tax system arises when the household is selected as the tax unit and/or tax allowances and deduction are designed on the 13 basis of marital status. In the individual tax system the tax unit is the individual and each faces the same structure of tax rates independently of his/her gender and marital status. In its simplest form each individual enjoys the same amount of personal allowances and reductions. Some systems provide additional tax relieves for household dependents (e.g., children and elderly parents) and others allow the transferability of any unused allowance from one spouse to the other. However, earnings levels, rather than marital status, are the major determinants of individuals' tax bills. Thus this system is primarily neutral with respect to gender and marital status and as such it minimizes the distortions on the women labor supply and on marital and demographic decisions. As Table 3 shows, this system is the rule in five member states (Denmark, Greece, Italy, Netherlands and the UK), is optional in another five but is not available in France and Luxembourg. 14 Table 3 Elements of national systems of earned income tax in the EU Tax unit Allowances/deductions Additional tax relief for lone parents Same relief married For children For other For childcare, and single people dependents domestic help, etc Individual taxation Denmark Yes, transferable Yes No No No, help with childcare costs Greece Yes Yes, against Yes Yes No, extra tax husband's reductions income Italy No, deduction for Yes Yes If both employed; Yes dependent spouse Netherlands No No No Yes Yes, allowance = marriage allowance UK No, but marriage No No No Yes allowance for couples Aggregation, but can opt for individual taxation Austria No, but marriage Yes No No Yes allowance for couples Belgium No, lower if single Yes, against No No Yes, additional husband's personal allowance income Germany Yes, but spouses No Yes, for second Yes, personal benefit from income child or if one allowance and splitting parent is childcare costs disabled/ill from 1St child Finland No No No No Yes Ireland No Yes No Yes, child allowance via quotient Portugal No, slightly higher Yes Yes No deduction per married couple Spain No, extra deduction Yes Yes Yes, if both No for dependent spouse employed and income below a threshold Sweden No Yes No Family quotient, individual taxation is not an option France No, extra relief via Yes, via No Yes, if both Yes, additional quotient system quotient employed allowance Luxembourg No Yes, if both are Yes employees Source: J Rubery et al. (1998), Table 6.1; H. Sutherland (1998) Tables 2,3,4 15 The simplest form of household taxation, on the other hand, is the aggregation of earnings for married couples. This is the norm in Spain, Ireland and Portugal. In this system the total household tax bill is calculated by adding the individual earnings together and taxing the joint income according to the marginal tax structure. This means that taxation of the second income (generally the wife's) starts at the highest marginal tax rate faced by the first earner, i.e., at a higher rate that if it was taxed on an individual basis. Income splitting and the family quotient system are variants of the joint taxation. In the former the joint income is divided equally in two; the tax is calculated on one half and then doubled. In the latter the aggregation includes children's income and the total is divided by a factor which takes into account the number of children. Income splitting is common in Belgium and Germany. The family quotient is compulsory in France and Luxembourg. These forms of taxation redistribute tax payments away from the first earner in the household at the expenses of the second earner, unless both partners have the same earnings. Thus taxing the married couple rather than the individual reinforces the breadwinner model of household organization in two ways. First, this system discourages labor market participation of married women by increasing the negative substitution effect on labor supply that is associated with taxes on eamed income. Second it accentuates any existing gap in gross wages between the (predominantly male) primary eamer and the (usually female) second eamer. It does so by increasing the tax bill on married women and - in the case of the split income - reducing the burden on their husbands. Both of these outcomes can be modified by providing additional tax relief to married couples in general and dual eamer couples in particular. In reality, however, most countries with join taxation systems provide significant aggregate tax relief for non- employed spouse. Individual based systems, with the exception of Greece, provide more limited tax relief through the ability to transfer not used allowances from one spouse to the other. The UK is the only country where individuals are taxed separately but receive an allowance for being married. Tax relief, however, are unlikely to counteract the negative effect of joint taxation on married women's participation in paid work and on their payoff to marriage. Unless the partners have similar earnings, opting for individual taxation increases the joint tax bill for the married couple especially if the system has much tax brackets (i.e., Germany). Thus the likelihood is that, given a choice, the household will opt for joint taxation. This implies that in seven countries the tax system present a clear gender bias and differential treatment of women on the basis of their marital status. A female-friendly taxation system is one that does not tax women at a higher tax rate just because they are married. Thus, it is one that does not subscribe to the breadwinner principle that presume women to be second earners within the household and, as such, financially dependent upon their husbands. This can be accomplished by using the individual as a tax unit and directing any tax relief for dependents at the costs of raising children rather than marital 'dependency'. A family-friendly taxation system is one that encourages family formation and fertility. It is one that gives tax allowances for children and maintains the financial risks associated with divorce at a minimum. 16 5.2 Financial support for children. Since the immediate post-Second World War, the right of families to state support became enshrined in international treaties. For instance, the United Nations Declaration of Hurnan Rights (1948) acknowledges the right of the family to be protected by the society and the state (Article 16), and the right to social security (Article 22). However the gender and demographic implications of different benefits systems received little attention until the 1970s when the rapidly declining fertility rates begun to fall below- replacement levels, women's participation in paid employment was growing fast and substantial transformation were beginning to question the traditional identify of the family. A sharp division emerged in Europe on the role benefits could and should play in this context. The general view in the 'pro-natalism' countries was that a higher level of state support for families could encourage childbearing, and that indeed it was the government's responsibility to make such pro-natalist intervention. These countries however were a small minority led by France but also including Luxembourg and Belgium. In the other countries the potential effect of pro-natalist policies and their legitimacy was highly questioned. In particular, the concern was with the role benefits could have in encouraging (or discouraging) women in general and young mothers in particulars from joining the labor market. In this context the governments' views varied widely from a high support to working mothers in the Scandinavian countries to a traditionally minimal support in the UK. Before the beginning of the 1 990s in the UK, these issues were not considered to be state responsibility and there was some concern about the potentially perverse effects of benefits in undermining the traditional family and encouraging lone parenthood, marital instability, and welfare dependency. This worry was shared with the Scandinavian countries [see for example Popenoc (1988, 1991) and for a different view Sandqvist and Andersson, 1992]. The policies implemented by the current UK government in the last year, however, represent a sharp U-turn on the traditional Anglo-Saxon tradition. Indeed the Labour government has highlighted the increase in the state support of the family as one of its priorities. 17 Table 4 Comparison of the level of financial support for children, 1993 Level of tax Spending on Family allowances (tax relief plus concessions' family transfer payments) as % of net benefitsb average earnings 1 child 2 children 3 children Luxembourg Extremely high Very high 22 28 40 France High High 1 22 50 Belgium Very High Very high 7 20 38 Germany Extremely High Very high 6 12 2 1 Denmark None Extremely high 6 1 1 15 Netherlands None Average 4 10 1 6 UK None High 5 9 1 3 Portugal High average Low average 4 9 14 Greece Very low Extremely low 4 8 12 Italy Low Low average 3 6 11 Ireland Low average Average 2 4 6 Spain High average Extremely low 2 3 5 Sources: Bradshaw et al. (1993: Table 2.17) for tax classification and spending on family benefits; CEC (1993b: Table 16) forfamily allowances Notes: a Value of tax relieffor dependent children as % of average gross earnings; ranking: extremely high, very high. High, high average, average, low average, low, extremely low, none (for details, see original source) b Universal child benefits plus income-relatedfamily assistance as a % of GDP standardized by % of population who are children Differences in the attitude towards state support of the family are reflected in the different degrees of financial support for children. This can take two forms: tax relief and cash benefits. Table 4 ranks countries according to the combined value of the tax relief and the spending on family benefits (i.e., the family allowance) as a percentage of net average earnings. It shows that nine countries offer tax relief for dependent children. The exceptions are Denmark, Netherlands and the UK while the relief is particularly high in Luxembourg, Germany, Belgium and France. These countries also spend relatively large amounts on cash benefits, with the result that they provide the highest level of family allowance (as a percentage of net average earnings) in Europe. In contrast Denrnark has a high level of state support delivered exclusively via cash benefits, reaching a relative level of family allowance which is very similar to the German one for a family with average income and one or two children (Figure. 18 4). Fig 4: Comparison of the level of financial support for children, 1993 60 C 4 60 - i0 El :-:--iE2 children CO 30 Sn 520 E~~~~~~~~~~~~~~~~r E; 71 I 11aA This ranking is, however, limited on several grounds. Firstly it depends on the reference group. In particular, in countries such as France the ranking is substantially improved when the reference is on three-child family. Secondly the analysis is restricted to family allowances provided by central governments. It excludes other direct (e.g., birth grant) and means-tested benefits and those targeted at specific families (such as the single-parent famnilies) which in some countries are substantial. It also ignores benefits provided by regional governments as in Germany. 5.3 Maternity andparental leave benefits. A different type of state support for the family is represented by the availability of maternity and parental leave. Maternity and parental leave are important because they provides some public recognition of the parental responsibilities of many workers, and enables leave takers to preserve their wage and occupational position while putting time into family care work. Thus they reduce the cost of childbearing in termns of foregone earnings. The effects on women's labor market participation are complex. Long leave entitlements may encourage some women to reduce their labor supply when their children are young. The right to withdraw temporarily from employment or to move to part-time hours without quitting the labor market is likely to encourage women to remain in employment after childbearing. 19 However, their effects are not all positive. The experience of taking a long period of leave may reduce women's labor market attachment and be associated with actual, or perceived, erosion of skills. Partly in recognition of this, France and Spain have introduced legislation granting the right to reinsertion training at the end of a period of leave (Rubery et al, 1998). Furthermore, when there is a shortage of suitable childcare services, parental leave may simply postpone labor market quits. For example, the German statutory parental leave was extended from one to three years in 1993 in the context of the post-unification dismantling of childcare services in the East and a general emphasis upon the harmonization of social policies. This reform offered women an extended right to retain their employment as a substitute for childcare services, rather than a complement. Thus, it had the effect of institutionalizing an interrupted employment pattern as women's ability to return to paid work grew increasingly constrained by the lack of childeare. Parental leave in societies with a relatively well established system of childcare subsidies, on the other hand, may divert resources and shift, rather than expand, the options for reconciling eaming and caring work (See Boje, 1994)'. Finally parental leave may reinforce women's status as second eamer in the strong breadwinner model, which sees them primarily as mothers. The presumption behind parental leave arrangements which are unpaid (e.g. Portugal, Greece) or supported by a low rate of benefit (e.g. Belgium, Germany), for example, is that the leave taker has an employed spouse. The lower average earnings of women create a financial incentive for the leave to be taken by the mother. This reinforces the presumption that women are second earners who interrupt their labor market activity and full-time employment trajectories. The second eamer bias in leave taking is likely to be exacerbated when unpaid leave is explicitly restricted to - -men as an extension of maternity leave (UK, Ireland), or where parental leave is allocated pei Iamily and fathers' involvement is minimal. In contrast, a non-transferable individual entitlement, by providing some incentive for men to take leave discourages their specialization in market activities (e.g. Belgium, Denmark, and the Netherlands). I For example, few Danish women interrupt their employment when they have young children, and even the tendency to switch to part-time work seems to be declining (Boje 1994). This pattem of labor market activity is facilitated by a relatively high level of publicly funded childcare services (see th iollowing sub-section). The January 1994 reform to the Danish leave scheme permits the municipal authorities to supplement the national rate of payment for parental leave, which is due to fall in the coming years. One incentive for this local addition might be to reduce the demand for childcare services, which are subsidized by municipal authorities (Boje 1994: 80). This may assist parents in areas where there is a shortage of childcare places, but the long-term outcome may be a gradual dismantling of service provision in favor of leave arrangements. Increased leave entitlements, potentially accommodated by a decline in childcare places, may therefore enforce the notion of women as second earners unless extended parental leave also becomes a feature of male employment patterns. 20 Table 5: Maternity leave benefits duration and pay, paternity leave and take-up rates, and child-care leave Maternity leave Parental leave Take up rates of Optional leave Reduced parental leave ' hours (in months)" leave Duration in Pay %c Duration in Claimants as % of Paid Unpaid weeks weeksd eligible leave takers Austria 16a 100 112 24 Belgium 15a 80 130 12 Denmark 18a 90 10+ 82% of eligible 12 mothers < 5% fathers Finland 17.5a 80 26-156 36 France 16a 84 0-156 <2% in 1992 36 Germany 14a 100 156 95% of all registered 18 births claim parental leave, <5% are fathers Greece 1 5b 50 6 Ireland 14b 70 3.5 Italy 22a 80 26 12 Luxembourg 16b 100 24 Netherlands 16b 100 27% of women; 11% 6 of men Portugal 13b 100 24 Spain 1 b 75 36 Sweden 65b 75 62 High %men 18 U.K. 14-40a 45 10 Source: a 1994figuresfrom Paci & Joshi (year?), Table 1.3 b 1990figuresfrom A.H. Gauthier (year?) Table 10.4 C l990figuresfiom A.H. Gauthier (year?) Table 10.4 d 1994figures from Paci & Joshi (year?) Table 1.3 e J Rubery (year?) Table 6.5 f 1990 figures A.H. Gauthier Table 9.3 The adoption of the EU Maternity Leave Directive in 1992 gave all women a statutory entitlement to fourteen weeks of paid maternity leave. The effect of this Directive has been and is likely to be different for different member states. In most EU-countries employed women already had a virtually automatic entitlement to between fourteen and eighteen weeks - slightly longer in France and Italy - combined with a high, earnings-related benefit, or a high flat-rate payment in the case of Denmark (EC Childcare Network, 1994; Rubery et al., 1995). 21 However, maternity legislation has being revised in Portugal to increase the statutory leave entitlement (from just over twelve weeks long) and significant changes had to be implemented in Spain to provide the required protection from dismissal (Lopes and Perista, 1995: 90; Molto', 1995: 137). In the UK the major impact was felt by women with very marginal employment records who previously did not fulfill the minimal employment or contributory records to be entitled to maternity leave. Table 5 highlights the large cross- country differences in length of the entitlements and the payment arrangements in the EU countries. The pro-natalist countries and Italy offer longer periods of maternity leave and higher payment than Britain and the Southern European countries. However the longest duration and higher benefits are found in the Scandinavian countries in general and Sweden in particular. Parental leave is a more recent development than maternity leave. A Directive to establish a statutory entitlement to Parental Leave and Leave for Family Reasons was proposed in 1983. However, following continual blocking by the UK, an agreement between fourteen countries has only recently been reached under the Maastricht social protocol. This gives entitlement to three months' parental leave for each parent. The funding, flexibility and availability of parental leave is quite diverse in the member states. Statutory parental leave payments are made in only eight countries. Denrmark provides the highest eamings replacement rate, paid to all members of the workforce. More limited financial support is offered in Belgium and Italy, for employees only. In both Germany and France the parental leave payment may be claimed not only by employed parents but also by the unemployed or inactive. The benefit is income- tested in Germany and restricted to the second and third child in France. In both countries the scheme has been extende. n Germany, a third year of unpaid leave was introduced in 1993, and in France the benefit became payable for the second as well as the third child in 1995. In Belgium, the career break scheme offers the potential for the longest period of leave - up to five years subject to various conditions. Statutory entitlements range from three years per family in France, Germany and Spain to a mere four weeks' extension to maternity leave for the mother in Ireland. Parental leave usually starts immediately after maternity leave and runs continuously, although in four countries there is the option to take the leave in one or more 'fractions' over an extended period (Belgium, Denmark, Germany and Greece). Employees who take leave have the right to return to work in all countries. This job guarantee applied only to the first year of leave in Spain until March 1995, but since then both employment protection and seniority have been extended to cover the full three-year period (OECD, 1 995a: 179). Parents normally remain on the payroll so that welfare entitlements are maintained but leave still does not count for seniority in some countries, such as Portugal and Greece. Reduced hours, rather than full-time leave may have the advantage of enabling parents to retain contact with employment and maintain their skill levels. Part-time leave is the only option in the Netherlands. is a legal entitlement in France, and can also be taken in Belgium and Germany subject to the employer's agreement. Other entitlements permit mothers to reduce their working day by up to two hours in the child's first year in Portugal, Italy and Greece, while in Spain hours can be reduced by up to 50 per cent until the child is 6 years old (Rubery et al, 1995: Table 1.13). Finally, the most generous statutory paid leave arrangements 22 to care for sick children in the twelve member states are found in Germany (EC Childcare Network, 1994). Here, the entitlement is ten days' paid leave per child per year, funded from statutory health insurance. However, few considerations warn us against reading too much into this international comparison. First it ignores inter-country differences in coverage and take-up rates of paternity leave. Strict eligibility criteria, in terms of duration of employment, number of hours of work, and category of employment, mean that in some countries many parents are not eligible for the benefit. But the legislation in this field is complex and, although information is available on the eligibility rules, few data are available on the proportion of parents covered by the schemes. Current survey sources and administrative records are also not designed to provide adequate information on the take-up and form of parental leave (see OECD, 1995a: 183). Indeed the distinction between maternity and parental leave is made in the labor force surveys in only three OECD countries (the Netherlands, Sweden and the US). Table 5 also presents the limited available data on take-up rates from different national sources. It indicates that the incidence of parental leave is rather low, lower than the take-up of other leaves, such as educational and sabbatical reasons in Denmark or partial retirement under the Belgian career break scheme (Rubery et al, 1998). In most countries less that 5% of eligible fathers takes parental leave, although the recently introduced Dutch part-time scheme appears more popular (11%). Sweden is the only country where a significant minority of men takes leave. The low take-up by fathers undermines the equality objective and reinforces the traditional view of the breadwinner male. In all EU countries the majority of leave takers are women. Take-up by eligible mothers is high in Denmark (82 per cent) and in the former West Germany (96 per cent in 1992). In the Netherlands the recently introduced scheme is used by only 27 per cent of eligible women, but a full assessment must wait until the scheme has become more established. In contrast, less than 2 per cent of entitled parents took either paid or unpaid leave in France in 1992. Of then a disproportionate share were women with three or more children who thus received some financial support (Gauvin et al., 1994: 74). The unpaid leave provisions are also estimated to be rarely used in Greece, Portugal and Spain (EC Childcare Network, 1994). The EC Childcare Network (1994) concluded that a number of elements deter parents from taking of full-time parental leave in some countries. These are: * the loss of income; * the inflexibility of some schemes, which do not allow several short leave periods or the option for part-time leave; + inadequate guarantees of employment conditions on the return to work; * employer's reluctance to grant leave, particularly among those with small businesses and general lack of workplace support for those on leave; * and a general lack of information on both sides. Finally, the use of the part-time leave option varies between countries. Less than 1 per cent of 23 mothers on leave work part-time in Germany, while in France part-time leave accounted for 5 per cent of leave taken in the public sector and 13 per cent in the private sector. In contrast, in Belgium just over half (55 per cent) of women taking career breaks opt for reduced hours rather than a full interruption (Meulders et al, 1994: 21). Furthermore, in the Netherlands more than half the mothers remained in part-time work at the end of their leave period, and interestingly 14 per cent of fathers also reduced their hours at the end of the leave period (Plantenga and Sloep, 1995: 81). Where part-time leave is possible but underused, the particular deterrents are + more restrictive conditions than for full-time leave; * the problem of finding childcare (particularly since the two parents cannot take leave at the same time in most countries); * and maternal preferences for full-time leave (OECD, 1995a: 187). The data in Table 5 also ignores the fact that, in some countries, including Belgium, Canada, Denmark, France, Germany, Greece, Ireland, and the Netherlands, ceilings are imposed on maternity leave benefits. As a result, the amount actually received by women may represent a lower proportion of earnings than that shown in table 5. For example, estimates for Denmark suggest that benefits actually received during maternity leave represent less than 70 per cent of the average wages of economically active women, as compared to the theoretical wage replacement rate of 90 per cent (Gauthier, 1994). Moreover, the data presented above do not reflect the fact that some women are eligible, in addition to the state benefits, to additional benefits provided by their employers. Again little is known about the proportion of women eligible for such occupational benefits. The few studies that have looked at this issue have revealed that this proportion can be considerable. For example, in Britain, McRae's (1991) study has revealed that 14 per cent of women received additional occupational benefits. Comparable data for the other countries are unfortunately not readily available. However, it is often the case that parental leave entitlements are enhanced in the public sector. Public sector workers receive their full pay in Denmark, 75 per cent of their pay in the Netherlands and in Italy full wage for the first month of leave and 80 per cent for the next two. In Luxembourg only public sector workers are entitled to parental leave and paid leave for sick children (Rubery et al., 1998). 5.4 The provision and cost of childcare. The third form of state support considered here concerns the provision of child-care. Child- care can be provided directly by the state or by private organizations. State provided child- care is essentially free to parents although some nominal fees is charged in some countries. State intervention in privately provided childcare may be of different degrees ranging from mere quality control to complete funding. State funding of privately provided childcare can take several forms: subsidies to child-care institutions, direct subsidies to parents (in cash or via a voucher scheme), tax relief and benefit disregards. Comprehensive cross-national data on the availability of childcare are limited. Rough estimates of the public provision for child- care are given in table 6 where the percentage of children enrolled in publicly funded day-care 24 institutions is expressed as a percentage of children of pre-school age. Table 6: Public provision of child-care in EU-15 countries, 1994 % childrer under school age enrolled in Cost of care as % publicly tunded day-care institutions female earnings b 3 to school age Under 3 France 99 33 0 Belgium 95+ 18-19 0 Italy 95 6 7 Denmark 79 60 8 Austria 69 3 11 Germany 68a 4 3 UK 66 7 28 Greece (1988) 65 5 Spain (1988) 65 Sweden 63 32 7 Luxembourg (1988) 55 <5 Finland 50 25 7 Netherlands (1988) 50 <5 Portugal (1988) 35 5 Source: 1994figures arefrom Paci and Joshi (1998), table 1.3; 1988figures arefrom A.H. Gauthier, 1994, table 9.4 and Bradshaw et al., 1996 Note: Data for Ireland is not available a Percentages are for the former West Germany; provision in the former East Germany was higher. b Calculated by Bradshaw et al, (1996), this measure is the average child care cost for one child age 2 years and 11 months, divided by the average earnings for a lone mother. In most EU countries compulsory schooling begins at 6 years of age, and there is generally some form c, primary education, or scope for children to enter primary school prior to the compulsory starting age. A notable exception is the UK where compulsory schooling begins earlier (5 or earlier in some regions). Table 6 shows that primary school hours and the way they are organized during the day vary considerably across countries. Opening hours for pre- primary education and childcare is estimated to be even more varied. A comprehensive system of publicly provided pre-school education exists in the pro-natalist countries (France and Belgium) where children begin to enter at the age of 2. In both countries 25 around one-third of 2 year olds has a place, rising to nearly complete coverage by the age of 3 (OECD, 1 995b: 130-131). Indeed, in Belgium, recent initiatives have focused upon developing facilities for sick children, out-of-school care and care for children whose parents have to work outside the normal opening hours of creches and nurseries (Meulders, 1995: 81- 82). Most children aged 3-6 years are also in funded places in Spain (82+ per cent), Italy and Demnark. With the exception of Spain, in these countries pre-school weekday provision is long, at seven to eight hours per day. In Spain there is a three-hour lunch break, although this is supervised in a growing number of urban schools (EC Childcare Network, 1996). In France and Belgium schools are closed on Wednesday afternoon and out-of-school care on this day is not available for all children. As a result, many mothers in France arrange their working hours to be off on this day (Dex et at, 1993; Hantrais, 1994). In Denmark and Italy daily attendance hours are shorter once the child enters primary school, partly compensated in Denmark by a comparatively high coverage of out-of-school hours. In Italy the system of six mornings per week largely coincides with public sector working hours, providing some incentive for women to enter this area of employment (Bettio and Villa, 1994). Pre-school provision is lower in the other countries. Coverage generally increases with the age of the child, but the opening hours of pre-school and primary institutions vary markedly in these countries and publicly funded out-of-school care is generally very limited. By the age of 4 nearly all children in Luxembourg and the Netherlands are in primary school. Children aged 4 years upwards attend for seven hours per day in the Netherlands, but there is a long lunch break (often unsupervised) and mothers have to arrange non-school care on Wednesday aftemoons. In Luxembourg, mothers have to acconmmodate the schools closing for two afternoons per week and sometimes a two-hour lunch break on the other three days, but their children attend school on Saturday mornings. In the UK, 90 per cent of 4 year olds are in primary school but they attend for only half a day, progressing to a longer six-and-a-half hour day for 5-year-olds. West German kindergartens and primary schools operate on a part-time basis, and once children enter primary school in Greece, Portugal and Ireland they too attend on a part-time basis. In the Netherlands the emphasis of policy has shifted over time to state funding of employer- run facilities. This policy has been criticized on the grounds that it does little to establish a permanent national system of provision since firms are sensitive to cyclical economic pressures, and are likely to limit this 'fringe benefit' to higher-level staff (Plantenga and Sloep, 1995: 78). This prediction was largely borne out in the UK when various initiatives in the banking sector were stalled or dismantled once labor shortages gave way to rationalization and rising unemployment. Another alternative to publicly fund childcare places is to have tax concessions or direct cash benefits to parents who purchase childcare. Currently there is not universal tax relief for childcare in five countries. In three of them (Belgium, Denmark and Italy) this is because the governnent has opted for direct provision of the services. In the UK and Ireland, however, this reflected a general lack of government support in the area, which has only been reversed in the 1990. In his 1990 Budget John Major introduced tax relief for childcare provided by the employer. Since 1994 claimants of the Family Benefits (and other in-work means-tested 26 benefits) are allowed to ignore part of their income if it is spent on childcare ('childcare disregard'). A childcare voucher for 4 years old was introduced in 1997. However, this was replaced by the incoming Labour Government by the 'Working Families Tax Credit', announced in the March 1998 budget and expected to be introduced in October 1999. This represents the second of three elements in the Labour Government' national strategy to make childcare accessible, affordable and high quality. The first was the financing of a target 30,000 new out-of-school childcare clubs; the third a green paper on the issue of quality and childcare training. The 'Working Families Tax Credit' is intended to cover 70% of childcare costs, up to a maximum of £70 a week for families with one child and income lower than £14,500 a year. The maximum qualifying and the credit ceiling increase with the number of children and will taper off as income grows. In some countries, tax relief for childcare might target the available places to employed mothers. For example, in the Netherlands around 30 per cent of childcare places are occupied by the children of non employed mothers, while 40 per cent of employed mothers rely upon other forms of childcare (Plantenga and Sloep, 1995: 78). Tax relief may also stimulate additional market provision in some countries, but the effectiveness may be limited in other national contexts, and in particular the market may fail to deliver quality care at a suitable price for those on low income. Furthermore, some low-income groups may not earn sufficient income to take full advantage of the tax relief. Indeed, for many women the main obstacle to the use of formal childcare remains the cost of the service relative to their wages. In countries where the state provides few childcare subsidies, women often rely upon informal care. Increased charges for subsidized services may also lead some women to switch to informal provisions. The last column of Table 6 shows the cost of childcare as a percentage of average female eaming for the countries for which data is available. In the UK, where state provision is very limited, over a quarter of the earning of the average mother goes to cover childcare costs. It is therefore not surprising that the use of informal care in this country is extensive (Ward et al., 1996 and Duncan et al, 1995)). This is also true of Netherlands and Greece, not included in the table (Rubery et al., 1998). The result also appears to hold true for the US where the use of formal childcare increases with mother's earnings and family income and declines in the price of this type of care (Hofferth and Wissoker, 1992). 6. The costs of childbearing in the EU countries. The economic theory of fertility concludes that the total cost of children for a family is given by the product of the cost per unit of human capital, the number of units acquired per child and the number of children. In the assumption that there is little variation across EU countries in the number of units of human capital acquired per child, inter-country variations in the cost of children can only be due to differences in the number of children and/or the unit cost of human capital. Children human capital is acquired by the combination of parental time and purchase of goods and services. According to the breadwinner model parental time is predominantly - if not exclusively - the mother's time. The cost of parental time is the opportunity cost of her time on the labor 27 market. In the most conservative definition this is typified by the earnings women forgo over the lifetime if they follow the typical employment pattern of a mother rather than remain childless. A wider definition encompasses any foregone access to social security and individual pension entitlements (Rubery, et al, 1998 and Owen and Joshi, 1990). Earnings may be forgone for three reasons: absence from the labor market; reduced hours of work; and lower pay per hour (Joshi, 1990). The lower rates of pay are due to loss of experience and seniority around childbearing, occupational downgrading on re-entry into the labor market and potential lower rate in part-time work (Joshi and Paci, 1998). All these factors also lower any pension entitlement. The cost of parental time therefore increases in the woman's earnings, the length of any absence from the labor market, the likely of re-entry at reduced hours of work and the extent of any 'pay penalty' to part-time work. In the basic fertility model childcare is a substitute for parental time in the production of human capital. Its unit cost, therefore, also have a direct effect on the cost of children. For a given level of income, therefore, any policy that reduces the size of the negative impact of childbearing on women's earnings and/or lowers the cost of purchased childcare should increase fertility. 6.1 Estimates of cross-country variations in the cost of childbearing Estimates of the opportunity cost of childbearing are limited. Using a national cross-section of British women in 1980, the Women and Employment Survey, Joshi (1990) confirms evidence that the presence of dependent children causes an interruption in, rather than the termination of, labor market participation and reduces hours of work in subsequent employment. It also suggests that the combination of these factors reduces mothers' pay compared to other women. She estimates the indirect costs of childbearing (in terms of the mother's foregone earnings) to be roughly half of the woman's potential earnings, considerably higher that the direct costs of child-related purchase of good and services. These estimates are over double those found by Calhoum and Espenshade (1986) for the US. However, in a comparison of four European countries, Davies and Joshi (1994) found the cost of childbearing in Germany to be similar to those in the UK and considerably higher that those of France and Sweden. In the light of our previous discussion this is not surprising. 6.2 State policies and women's labor market participation. A major component of the cost of childbearing is represented by the loss of job experience and seniority which is often associated with childbirth. The question of the effects of welfare benefits on mothers' participation in paid work has attracted considerable interest in Europe as part of the general concern about the work disincentive nature of welfare benefits. A large proportion of this literature concentrates on the effects of welfare benefits on the behavior of lone-mothers. The rest can be classified into two groups: the cross-national comparisons and the single country analysis. The former exploit the wide variation in the cost of childbearing across countries to assess its effects on women's employment decisions. This approach has the advantage of acting as a quasi-experiment in an area of social policy were 28 real experiments are rare. However, it fails to adequately pick up intra-country heterogeneity and inter-country differences in variables other than the welfare system, which may affect women's fertility and employment decisions. It is therefore not surprising that most of the empirical studies in this area have concentrated on individual countries and used either aggregate time series or individual life history data. The first type of studies overcomes the problem of unobservable differences between countries, the second that of intra-country heterogeneity. Neither of them, however, is exempt from other methodological shortcomings. In particular, a serious limitation of the studies based on individual data derives from the omission of information on the individuals' and families' entitlement to benefits (and their actual receipt of benefits) from most surveys on family and employment histories. Thus, attempts to assess the effects of welfare benefits on families had to be based on estimated, or imputed, values of welfare benefits, rather than observed ones. Table 7, which summarizes some of these studies, groups them under two heading: those concerned with the employment behavior of married women and those that focus on lone mothers. Table 7 here Among the first group of studies, Heckman (1974) examine the effects of child-care programs on women's work, indirectly, through a derived measure of the cost of child-care. The results confirmed the expected relationship, that higher child-care costs reduce the participation of women in the labor force. By extension, the author concluded that subsidies for child-care should have the opposite effect, and encourage female labor-force participation. This conclusion was supported by subsequent studies. 29 Table 7: Welfare benefits and female employment: a review of the econometric analyses Country Dataa Dependent variable Independent variablesb 1/0 Married women Duncan et a] (1995) UK CS0 Labor-force Cost of childcare and I participation childcare policies Ribar (1992) US LHd Labor-force Cost and quality of I participation childcare Michalapous et al. (1992) US LHd Labor-force Childcare tax credit I participation Leibowitz et al US LH' Probit of employment Cost of childcare, I (1992) Childcare tax credit Gustafsson and Stafford (1992) Sweden LH' Probit of employment Subsidy to and I availability of childcare Robins and Blau (1991) US LH' Probit of employment Cost of childcare. I Childcare tax credit, AFDC benefit Blau and Robins (1989) US LH9 Hazard rates fertility/ Cost of childcare, I employment Childcare tax credit Robins, et al (1980) US LH' Probability of being Negative income tax 0 employed Waldfogel et al. (1998) US, UK, CS' Job retention Matemity pay 0 Japan Joshi et al (1996) UK LHi Probability of being No-break at childbearing 0 employed Macran et al (1996) UK LHI Re-entry employment No-break at childbearing 0 after childbirth McRae (1991) UK LHk Return to work after Maternity pay 0 childbirth McRae (1993) UK LHk Return to work after Maternity pay 0 childbirth Lone mothers Ermisch (1991) UK PI Probability of being Supplementary benefits I employed Ermisch and Wright (1991) UK LHm Entry/exit from Supplementary benefits I . employment Maintenance allowance Jenkins (1992) UK LH" Probability of being Housing benefits I employed Maintenance allowance, 0 Cost of childcare I Walker (1990) UK P0 Probability of being Supplementary benefits, I employed Maintenance allowance, Farnily income 0 supplement & housing benefits I Blank (1985) US CSp Labor-force AFDC benefits 0 participation Housman (1980) US LHW Labor-force Negative Income Tax 0 participation Moffitt (1983) US LHr Labor-force AFDC benefits I participation 30 Source: Expanded on Gauthier (1994), table 9.7 Notes: a Types of data: P: Pooled cross-sectional and time-series data; TS: Time-series data; LH: Life history data. b I/O: Data on welfare benefits either Imputed or Observed. 1991-92 British General Household Survey. d 1984 Survey of Income and Program Participation. eNational Longitudinal Survey of Youth (women aged 14-21 in 1979); data used from 1982 to 1986 surveys. 'Swedish household survey. 5 Employment Opportunity Pilot Projects Survey (married women in 1980). h Seattle-Denver experiment (including both married and lone mothers). 'For the US: 1992-95 March Current Population Survey; For Britain: 1993-95 Labour force Survey (LFS); For Japan: 1992 Employment Status Survey (ESS). National Child Development Study (NCDS). k Nationally representative sample of women who had babies, Dec 1987-Jan 1988. ' 1973-82 British General Household Survey. m 1980 Women in Employment Survey. n 1989 Lone Parents Survey. ° 1979-84 Family Expenditure Survey. P 1979 March Current Population Survey q Gary maintenance income experiment 1971-4 (black female headed households) 'Panel study of income dynamics (female headed households). Using data from the 1984 panel of the American Survey of Income and i-rogram Participation (SIPP), Ribar (1992) found that the cost of market childcare had strong negative effects on the labor supply of married women. Blau and Robins (1989) also found t1-at higher child-care costs have a negative effect on employment, by encouraging exit from paid employment, and by deterring entry into it. They also found that subsidies to childcare, in the form of tax credit, have the opposite effect in reducing departure from paid employment. No significant effect was, however, found with respect to the entry into paid employment. These findings, however, were partially contradicted by later results by the same authors who found that, contrary to expectation, higher child-care costs have no effect on the likelihood of taking up paid employment (Robins and Blau, 1991). Tax credit for child-care was again found to have a positive effect on women's labor supply. These findings are supported by those of Michalopous et al (1992) who also use the 1984 panel of the American Survey of Income and Program Participation (SIPP). Leibowitz et al. (1992), however, using data from the National Longitudinal Survey of Youth (NLS-Y), conclude that the effect of tax credit for child-care encourage early return to work but has little effect on later labor supply. The large availability of high quality publicly provided childcare also appears to encourage labor market activity amongst Swedish women (Gustafsson and Stafford, 1992). In their later paper Robins and Blau also addressed the question of the effect of means-tested cash transfers on employment and, as expected, found them to have a disincentive effect on employment. This result confirmed what had emerged from the Negative Income Tax Experiment some years earlier. Then the availability of a guaranteed income had been found to increase the duration of any career break by 42 per cent for married women, and by 60 per cent for lone mothers (results from the Seattle-Denver experiment, reported by Robins, Tuma, and Yeager, 1980). 31 Duncan et al. (1995) use the 1991-92 British General Household Survey (BGHS) to simulate the effect of introducing different forms of childcare subsidies on the labor supply of British women. They found that childcare subsidies have predominantly positive effects on labor supply, particularly when the subsidy is available to mothers of small children. They also found labor supply effects of different magnitude for alternative forms of subsidies with a universal cash subsidy having the largest effect. The scope of the studies examined above was limited to the effects of direct and tax-related cash transfers. The studies by McRae (1991 and 1993), on the other hand, analyses the potential effect of maternity-related benefits on women's participation in paid work. This study, based on a survey of women who had a baby in 1987-8, examines their employment pattern nine months after childbirth. As expected, the receipt of higher maternity leave benefits seems to be associated with a greater propensity to return to work. Indeed, being in receipt of contractual maternity pay (occupational benefits) made women two and one-half times more likely to return to work after childbearing than women on ordinary public maternity pay (McRae, 1991: 232).2 On the other hand, this study also suggests that ineligibility for public maternity benefits, or eligibility only for reduced maternity benefits, has no significant effect on employment. Although the issue analyzed by Waldfogel et al (1998) is slightly different, the results for the UK are in line with McRae's earlier finding. Using data from large nationally representative panel labor force surveys for the US, the UK and Japan, they analyze the effect of maternity leave on women's job retention with the same employer. They found that maternity leave cover raised the likelihood of a woman returning to her job by 16% in the UK, 23% in the US and 76% in Japan. The higher effect of the policy in the US and, especially, in Japan may result from the rather limited coverage the policy has in those countries. Finally Joshi et al (1996) and Macran (1996), using data from the National Child Development Study, found that for a cohort of British women born in 1958, maintaining continuos employment over childbearing increased the probability of being in employment at age 33 by up to 20%. However, they highlight the presence of large variations in the effects of such policy across women. We turn now to the studies of the employment behavior of lone mothers. The studies by Ermisch (1991), Jenkins (1992), and Walker (1990), based on British data, all found that supplementary benefits, in the form of income support, lowered the probability of employment of lone mothers. On the other hand, Ermisch and Wright (1991), also based on British data, did not find any evidence to support such a conclusion. Other studies on the effect of other types of cash benefits on women's employment have also found no evidence in support of the disincentive hypothesis. By contrast, maintenance allowance (Jenkins, 1992; Walker, 1990), family income supplement, and housing benefit (Walker, 1990) all appear to have a positive effect on employment. So do Child Support payments (Bingley et al., 1995). 2 Although this result is supported by theory, it is worth noting that it may also be a reflection of the fact that the receipt of private benefit is often associated with the obligation to returning to work after childbirth (Gauthier, 1994). 32 The differential effect of income support and other benefits may be partly explained by the different nature and eligibility rules of cash benefits. The three welfare benefit schemes operating in Britain until April 1998 - i.e., the Family Credit, Income Support, and Housing Benefit - were means-tested and were dependent on the work status of the claimant. Family Credit (which replaced the former Family Income Supplement in April 1988) was payable to parents who work at least sixteen hours per week. Income Support (which replaced the former Supplementary Benefits in April 1988) was payable to those who worked less than sixteen hours per week or who, if unemployed, were available for work (this qualification does not apply to lone parents). Families with savings or assets above £8,000 were not eligible for either benefit. For others, the amount of benefits to which they were entitled was reduced by 70 pence for each £1 of income in excess of £3,000 (savings between £3,000 and £8,000 are treated as if each £250 is equivalent to bringing in an income of £1 per week). The rules governing Housing Benefit were similar, but the amount of benefit, and the tax on excess income, varies according to whether claimants are receiving Income Support or not. It should also be noted that receipt of maintenance allowance was taken into account in the calculation of a claimant's disposable income. Several of the American studies also highlight the potential disincentive effect of cash benefits on employment. For instance, Blank (1985) concluded that higher AFDC benefits have a deterrent effect on number of hours of work. Moffitt (1983) reached a similar conclusion, and also pointed out that the disincentive effect of welfare benefits was small in comparison to the effect of other non-benefit variables. Finally, Hausman (1980) also underlined the potential deterrent effect of welfare benefits on employment. His study, based on the Negative Income Tax experiment, suggests that higher marginal tax rate and higher transfer payments both reduce the probability of taking up paid employment. In general, these studies support the hypothesis that higher cash benefits can have a deterrent effect on employment; a finding with important pol'icy implications. However their findings suffer from a variety of limitations. They are primarily based on imputed, rather than observed, data on welfare benefits. With the exception of Moffitt's study, they tend to ignore the problem of partial take-up of benefit. This may introduce a considerable methodological bias. In Britain it is estimated that less than 60 per cent of families eligible to the means-tested Family Credit actually claim it (Department of Social Security, 1993: 22). In the United States, Mofitt finds a non-take-up rate of 25 per cent with regard to AFDC benefits among female-headed households. This is predominantly due to the perceived stigma attached to being welfare dependent. However, once a person has become a recipient of welfare, benefit levels do not appear to have no additional behavioral or stigmatizing effects. 6.3 State policies and women's earnings. Human capital theory and other less competitive theories of how the labor market operates - i.e., internal labor market theory, etc - emphasis the positive relation between continuity of employment and earnings. Thus it is rational to expect that, if family friendly policies encourage continuos employment, they will also lead to higher wages for mothers. Conversely, if extensive parental leaves allow women to take long periods away from the 33 labor market, the effect on future earnings would be negative. The sign of the relation between state policies and women's wage therefore will depend on the particular policy in question. The scope of the empirical literature todate has been limited to the effect of maternity leave on wages. This is exactly the policy for which the direction of the wage effect is unclear a priori. However, the evidence so far has suggested a positive effect of maintaining employment with the same employer over the childbearing period in both the Britain and the US (Joshi, Paci and Waldfogel, 1996; Waldfogel, forthcoming). There is also some recent evidence of a positive effect of maternity leave coverage (Ruhm, 1998). 7. Empirical evidence on the demographic effects of 'family friendly' state policies. The theoretical debate on the gender and demographic implications of different state policies is reflected in the empirical literature in this area that focuses on the effects of state policies on family formation and structure. The view of the 'pro-natalism' countries - that a higher level of state support for families could encourage childbearing - is subject to empirical scrutiny and their potential perverse effects on traditional family values evaluated. The issues of the potential effect of maternity and paternity leaves and of alternative benefits and tax systems on labor market participation of women in general and mothers in particular has received also considerable attention. 7.1 Do welfare benefits increase fertility? The studies of the potential effect of state policies on fertility have follows three main approaches: * the qualitative approach, a the descriptive intuitive approach, and the * econometric approach. The first approach uses qualitative data provided by public opinion surveys. An example of this type of studies is given by the Population Policy Acceptance Survey carried out in nine European countries in 1990-2. This showed that nearly 50% of the respondent agreed that family policy measures would make it easier for them to have the number of children they were intending to have. However only a much small percentage (10-30%) of them would have altered their optimal number of children as result of the policies (Moors and Palomba, 1995). The analysis of fertility trends according to the second approach has led to the conclusion of a positive correlation between fertility and state policy at least in the case of France and Germany (Butterner and Lutz, 1990: Calot and Hecht, 1978; Chesnais, 1987). These studies, however, -fail to account for any fertility determinants other then the state policy. The econometric studies overcome this limitation. They are the focus of this review and are summarized in table 8, sub-divided according to the data they use among the lines discussed above. Among the studies based on aggregate data, the first two are very similar in both the scope of analysis - cash benefits only - and the methodology used, OLS. The main difference is in the 34 number of countries for which the analysis is carried out. The study by Ekert (1986) uses pooled time series and cross-section data from eight EU countries for the period 1971-83, the later paper by Blanchet and Ekert-Jaffe' (1994), extends the analysis to eleven EU and Scandinavian countries and include 1970 in the analysis. The results are also very similar. They both provide mild support for the pronatalist hypothesis: the extension of cash benefits equivalent to those provided by the French system to the other countries would result in each woman having 0.2 to 0.3 more children than under the system in operation at that time. Caudill and Mixon (1993) find similar results for 51 US states in the year 1985-86 using means-tested cash benefit as a dependent variable. The conclusions of these studies, however, should be interpreted with great care for at least three reasons. First, the studies take into account only limited forms of welfare benefits (family cash allowances and some means-tested benefits). Second, being based on aggregate data, they ignore potential inter- and intra-country differences in the effects of the benefits. In particular, they assume that the relationship between fertility rates and benefits is the same for all countries as well as for all sub-groups within each country. Finally, by using static models they fail to keep up the long-run effects of the benefits, which - due to the slow process with which demographic decisions are taken - may be substantially higher that the short-run one (Rubery et al, 1998). The study by Gauthier and Hatzaius (1997) overcomes some of these limitations. It extends the analysis to include maternity benefits and tax allowances as well as cash benefits.3 It also allows for inter-country variation in the effects of welfare benefits on fertility and uses a dynamic rather a static model. In addition it is based on data for a larger number of countries (22) anc for a longer time-series (1970-90). The results, however, are not dissimilar. Benefits 25% higher than average would results in women having 0.07 children more. Tax allowances were dropped from the estimation when they were found to be not significant. 35 Table 8 Welfare benefits and fertility: a review of the econometric studies Country Data Dependent Independent I/Ob typea variable variables Aggregate Data Ekert (1986) 8 EC P Total fertility rate Cash benefit index Blanchet and 11 EC & Nordic P Total fertility rate Cash benefit index Ekert-Jaffe (1988) Caudill and Mixon US (51 states) P Total fertility rate Means-tested cash (1993) benefits Gauthier and 22 industrialized P Total fertility rate Cash, maternity, Hatzaius (1997) childcare' Ermisch (1988) UK TS Conditional birth Child benefit rateg Zhang et al. (1994) Canada TS Total fertility rate Family allowance, tax relief, maternity leave Individual data Barmby and Cigno UK LHW Parity-specific birth Child benefit I (1988) rate Robins and Blau US LHd Probit presence of Cost of childcare I (1991) children aged 0-5 Childcare tax credit AFDC benefit Blau and Robins US LHW Hazard rates Cost of childcare I (1989) fertility/ Childcare tax credit employmenth Whittington (1992) US LHf Conditional fertility Tax exemption logith Child-care tax credit Source: Expanded on Gauthier (1994), table 9.5, Gauthier and Hatzaius (1997), table 1. Notes: a Types of data: P: Pooled cross-sectional and time-series data; TS: Time-series data; LH: Life history data. b I/O: Data on welfare benefits either Imputed or Observed. c 1980 Women and Employment Survey (women aged 16-59). d National Longitudinal Survey of Youth (women aged 14-21 in 1979); data usedfrom 1982 to 1986 surveys. e Employment Opportunity Pilot Projects Survey (married women in 1980). fPanel Study of Income Dynamics (married women aged 15-44); data usedfrom 1977 to 1983 surveys. g Parity-specfifc indices. h These dependent variables are not parity specif c. Indices of cash benefits, maternity leave benefits, and child-care facilities. Like the earlier studies, however, this analysis uses the total period fertility rate as dependent variable. This is an index highly sensitive to changes in the timing of births. Thus, it is not appropriate if one expects benefits to have an effect on both the number of births, and their 36 timing, and to have a differential effect by birth order. In view of the trend experienced since the 1960s towards increasing age at entry into motherhood, the use of this index can give rise to a potentially serious bias. From this point of view, the study by Ermisch (1988) on fertility in Britain is methodologically and substantively more satisfactory as it uses as dependent variable a parity-specific index. The results confirm the importance of the bias overlooked by other studies. The effects of the benefits appear to be predominantly on the timing of births rather than on the size of the completed family. Higher benefits encourage early motherhood. The Canadian study by Zhang et al (1994) considered a more comprehensive set of benefits than those included in previous studies. Like previous studies, Ermisch's analysis was conducted at the aggregate level and ignored potential intra-country differences. The other studies reported in table 8 carried out on the basis of individual data, consequently represent a further step. They are, however, also limited since, in all cases, data on welfare benefits were imputed rather than observed, and since they have failed to acknowledge the potential differential effects of benefits upon various sub- groups of the population. The first of these studies uses life-history data from the British 1980 Women and Employment Survey and imputes data on cash benefits, which are restricted to the universal child benefit (Barmby and Cigno, 1988). This finding confirm those of Ermisch: in Britain higher benefits encourage early motherhood but have not significant effect on completed fertility. Unfortunately, this study does not take into consideration other cash benefits such as tax-related and means-tested benefits, which may also have an effect on fertility. The American studies reported in table 8 include other forms of imputed cash benefits. Among them, the studies of Blau and Robins (1989 and 1992) use different data and different methodology and reach slightly different results. The first paper, based on life-history data of households in twenty geographical areas in the United States, suggested that higher child care subsidies (through tax credit) have a small, but not significant, positive effect on the probability of having a child while employed but no effect on non-employed women. The later study, using nationally representative data from the National Longitudinal Survey of Youth, finds that child-care tax credit and AFDC benefits both have a statistically significant effect on fertility (measured through the presence of a pre-school-age child). On the other hand, and as expected, higher child-care costs have a negative effect on fertility. A similar conclusion emerges from the study of Whittington (1992) which uses the Panel Study of Income Dynamics and which finds an additional effect on the timing of births. These studies consequently appear to give a mild support to the pro-natalist hypothesis. In view of their limitations their findings cannot, however, constitute strong evidence. Not only were data on benefits imputed rather than observed, but also the possibility of differential effects of benefits upon different sub-groups of the population was neglected. In particular these studies ignore the potential effect of housing benefits on fertility. This effect has been tested only indirectly through indices such as housing cost (see Ermisch, 1988) and housing tenure (see Murphy and Sullivan, 1985). The evidence suggests that housing has a determinant effect on both fertility and family structure. 37 7.2 Do welfare benefits increase the probability and the length of lone parenthood? The second question we want to address it that of the effects of welfare benefits on the entry into and exit from lone parenthood. Table 9 summarizes the empirical evidence on this issue. Table 9: Welfare benefits and family structure: a review of the econometric studies Country Data Dependent Independent I/Ob typea variable variables Entry in lone parenthood Ermisch (1991) UK LHC Hazard rates of pre- Supplementary I marital birth benefits Duncan and Hoffman US LHd Probability of birth AFDC benefits I (1991) out of wedlock Bishop (1980) US LHe Probability of Negative 0 divorce Income Tax Exitfrom lone parenthood Ermisch (1991) UK LHc Hazard rates Supplementary I fertility of benefits remarriage Source: Expanded on Gauthier (1994), table 9.6 Notes: a Types of data: P: Pooled cross-sectional and time-series data; TS: Time-series data; LH: Life history data. b l/O: Data on welfare benefits either Imputed or Observed c 1980 Women and Employment Survey (women aged 16-59). d Four income Maintenance Experiments (Seattle-Denver, Gary, New Jersey, Rural) There are two ways of entering lone parenthood: having a child out of wedlock and divorcing after having had a child. The first two studies in Table 9 investigated the probability of having a pre-marital birth. The scope of the study by Duncan and Hoffman (1991) is limited. They use data from the Panel Study of Income Dynamics to analyses the effect of AFDC benefits on the probability of having a child out of wedlock for black teenage girls. The study, which uses imputed data on welfare benefits, finds that the incidence of lone parenthood amongst girls living in States with higher AFDC benefits was higher than average but the effect was smaller once they had controlled for differences in parental income. Ermisch (1991) extends the analysis to all women. He used data from the British Women and Employment Survey with imputed value of the means-tested supplementary benefits received by the mother of one child. Welfare benefits appear to have a positive but small effect on the likelihood of a pre- marital birth: a 10 per cent increase in welfare benefits was estimated to raise the percentage of women having a pre-marital birth by less than two percentage points (Ermisch, 1991: 79). The probability of entering lone parenthood through divorce has also been analyzed in several studies. Among them, the evidence reviewed by Bishop (1980) comes from the Negative 38 Income Tax experiment. The policy tested consisted of extending welfare benefits (previously restricted to lone-parent families) to two parent families, in addition to providing them with a minimum guaranteed income above which all earned and unearned income were subject to taxes. The most striking of the results generated from this experiment came from Seattle-Denver and suggested that the Negative Income Tax increased the likelihood of divorce by as much as 50 per cent (Bishop, 1980; Garfinkel and McLanahan, 1986). However, this and other findings from the experiments were so controversial that the suggested welfare reform was never extended beyond the experimental stage. The studies in the second part of Table 9 address the question of the effects of welfare benefits on the exit from lone parenthood through remarriage for once-married women, or through a first marriage for never-married mothers. They provide no empirical support for the Anglo- Saxon hypothesis that higher welfare benefits discourage remarriage and discourage pregnant single women from marrying (for a review see Ermisch, 1991). By contrast they suggest that the main effect of welfare benefits is on the living arrangement of lone mothers. The availability of benefits allows them to establish an independent residence rather than having to live with parents (see Garfinkel and McLanahan, 1986: 58). Although these studies do not allow for any intra-women heterogeneity4 a clear message emerges. Studies on the effects of welfare benefits should focus not only on the marital status of the mother, but also on her living arrangements. 7.3 Summarizing the empiricalfindings. So does the empirical evidence support the 'pro-natalism' view? Is the concern about the perverse effect of welfare benefits on the traditional family justified? What are the effects of state support for the family on women's labor market participation and their eamings. From the above review, it is possible to make at least four methodological points about ways of assessing the effects of welfare benefits (Gauthier, 1994): * Individual data is to be preferred to aggregate data as it is capable to capture the heterogeneity arnong the population in terms of the entitlement to benefits, and the receipt of them; * all available benefits should be included in the analysis; i.e., universal, means-tested, tax-related, and private (occupational) benefits; * the effects of benefits should be allowed to differ for different sub-groups of the population (for example, low income versus high income and, more importantly, households where women contribute a large share of income v/s others). Failure to incorporate such points may partly account for the inconclusive findings of the empirical literature todate. However, data limitation prevents methodologically more satis- factory analyses. 4 In practice clear differences exist between economically active and economically inactive women, between women who regularly receive maintenance allowance and those who do not, and between women who have easy access to informal (and cheap) child-care arrangements, and those who do not. 39 A further point that needs to be considered is the issue of the time sequence in which decisions about fertility, family structure and employment are taken. In the empirical studies reviewed above, and in much of the theoretical analysis, the effects of welfare benefits on fertility, family structure, and female employment were considered separately. The implicit assumption behind this approach is that the fertility decisions take place in a vacuum, in isolation from considerations related to family structure and female employment. It must be asked whether this is a methodologically sound way of addressing the question of the effects of benefits. Indeed it is the view of some that may be more satisfactory to consider demographic and labor market decisions as part of the combined strategy to maximize the woman - or the household - long -term utility rather than as the result of separate decision making (Joshi, Paci and Waldfogel, 1996). Leaving aside these methodological limitations and considerations, the empirical evidence reviewed above concludes that state support of the family has played only a limited role in the making of demographic and labor market trends in Europe. It suggests that it is improbable that fertility would rise significantly if state support for the family were increased, or that reducing the benefits received by lone mothers could reduce the incidence of single parenthood. Indeed, other factors, such as education and income appear to be more important that welfare benefits in determining demographic behavior. 8. Conclusions Table 10 summarizes the findings of our international comparison of State policies towards the family. The importance of the tax and benefit system in determining both fertility patterns and women's labor force participation is striking. Countries differ in the extent of state support to family and in the role that their tax and benefit system assign to women within the family. Four distinct groups of countries emerge from our analysis. At one extreme are the Scandinavian countries. Here the State support for the family is high. Parents receive generous family related benefits and enjoy long and well- paid entitlements to maternal and paternal leave. The resulting comparatively low costs of childbearing have busted the fertility rates in these countries and brought them well above those of countries where large families have long been a tradition.. At the same time the availability of good quality - and predominantly state financed - childcare combined with a tax system based on a 'weak-breadwinner' model and a strong government commitment to reducing gender inequalities in pay, have allowed the Scandinavian mothers to maintain high labor market participation. 40 Table 10: International comparison of state policies towards the family Type of TFR Activity Family Maternal/Paternal % children in country 1996 Rate 1995 allowances Leave publicly funded day- %* care institutions 3 and over <3 Scandinavian Finland 1.76 70.0 Long/Well paid 50 25 Denmark 1.75 73.3 11 Long/Well paid 79 60 Sweden 1.61 76.1 Long/Well paid 63 32 Pro-natalist Luxembourg 1.76 36.0 28 Medium/Well 55 <5 France 1.72 60.6 22 Long/Well paid 99 33 Belgium 1.55 51.7 20 Long/Well paid 95+ 19 Italy 1.22 42.5 6 Long/Well paid 95 6 Southern Portugal 1.44 59.1 9 Short/Well paid 35 5 Greece 1.31 44.3 8 Short/Badly paid 65 5 Spain 1.15 44.9 3 Medium/Badly 65 Anglo-Saxon Ireland 1.91 40.0 4 Short/Medium UK 1.70 66.0 9 Short/Badly paid 66 7 Netherlands 1.50 58.9 10 Medium/Well 50 5 Source: various Notes: * as % of net average earnings for family with two children. At the other extreme are the Southern European countries (excluded Italy). Here the state support for the family is very limited, the cost of childbearing is high, the gender pay gap is relatively large and the 'strong breadwinner' model of division of labor within the household is entrenched in the tax system as well as in the social network. It is therefore not surprising that in the last decade fertility in these countries - except Portugal - has dropped to the lowest rates worldwide and women's labor market participation remains low. Portugal is rather different since in this country both fertility and participation have always being historically higher then in other Southem European countries, despite the relative high cost of childbc-aring. The other countries sit somewhere between these extremes. The pro-natalist countries mirror Scandinavia in the degree of support for the family. Family related benefits are the European highest, long and well-paid entitlements to both maternal and maternal leave are usually available and childcare provision is extensive. Childbearing costs are low and, as a 41 consequence, fertility rates are generally high. However, contrary to what happens in Scandinavia, the taxation system of these countries is generally based on a 'strong breadwinner model' which discourages participation by the secondary worker and reinforces the traditional division of labor within the household. Moreover the limited availability of part-time work in these labor markets makes mixing family commitments with labor market participation difficult despite the widespread availability of childcare. The result is a relative low rates of women's participation in paid work. Italy is the odd one out within this group of countries. In Italy the cost of childbearing is low but so is the fertility rate, the tax system does not follow the strong breadwinner model but women's labor market participation is the lowest in Europe. This contradicts all predictions of the economic theory of the family. Amongst the possible explanations are the rigidity of the Italian labor market which limits part-time opportunities, and the extent of its underground economy, which employs women in large numbers. However, sociological factors are also important. Amongst those a breadwinner model that goes beyond the tax system, into social tradition, and a late surge of feminism ideology. Finally we have the 'high fertility/high childbearing costs' countries. These are Ireland on the one hand and the Netherlands and the UK on the other. Ireland presents very high fertility and low participation rates despite the high costs of childbearing. This is consistent with a society based on a strong breadwinner model where - with limited exceptions - abortion is still outlawed. The Netherlands and the UK present more of a puzzle to the economic theory of the family. Compared to other European countries they share relatively high fertility and participation rates despite the comparatively high childbearing costs and the limited availability of childcare. The economic theory of the family would explain the high fertility rate of these countries in terms of Dutch and British families having a stronger preference for having children than in the rest of Europe. This must be particularly true in the Netherlands where fertility rates have consistently been considerably higher than the European average. The higher participation rates, on the other hand, may simply reflect the higher availability of part-time work in a context were, despite the 'weak breadwinner' characteristics of the tax systems, housework remains predominantly a woman's responsibility. This brings me to my final point: the role played by fathers in the demographic and labor market decision making of the family. The most egalitarian arrangements in terms of division of labor within the household appear in Scandinavia. There it has been made the greatest attempt to 'socialize the costs of familyhood' (Esping-Anderson, 1990), or at least 'to adopt pro-family policies that were more pro-woman and with fewer restrictions on reproductive choice' (Folbre, 1994). It is not surprising that these appear to be the countries were women are most likely to be able to combine childbearing with long-term well-paid employment. As Heitlinger (1993) also points out, it is possible for policies to promote both the family and women's economic equality. The evidence from Europe suggests that egalitarian work patterns, egalitarian parenthood and higher fertility lie in the same direction. Therefore, leaving aside the question of the legitimacy of government intervention in this area, EU 42 governments may consider directing their effort towards a general family support plan, rather than a strictly defined family policy (Gauthier, 1994). 43 Bibliography Barmby, T and A. Cigno (1988) A sequential probability model of fertility patterns, Research paper no. 160, Department of Economics and Commerce, University of Hull. Becker, GS (1981), A Treatise on the Family, Oxford: Oxford University Press. Bettio, F and P. Villa (1994), Changing patterns of work and working-time for men and women in Italy. Report for the European Commission. Working Paper, EC Network of Experts on the situation of Women in the Labour Market: Manchester School of Management, UMIST, UK. Bingley, P, G. Lanot, E. Symons and I. Walker (1995) Child support reform and the Labor Supply of Lone Mothers in the United Kingdom, Journal of Human Resources, 30(2). Bishop, J (1980), Jobs, cash transfers and marital instability: a review and synthesis of the evidence, Journal of Human Resources 25: 301-335. Blanchet, D and 0. Ekert-Jaffe' (1994), The demographic impact of family benefits: evidence from a micro-model and from macro-data, in Ermisch, J and N Ogawa (eds), The Family, the Market and the State in Aging Societies, Oxford: Oxford University Press. Blank, RM (1985), The impact of state economic differentials on household welfare and labor force behavior, Journal of Political Economy 28: 25-58. Blau, DM and P.K. Robins (1989), Fertility, employment and childcare costs, Demography 26(2): 287-299. Boje, T (1994), Changing patterns of work and working-time for men and women in Denmark: towards the integration of the segmentation or the labour market. Report for the European Commission. Working Paper, EC Network of Experts on the situation of Women in the Labour Market: Manchester School of Management, UMIST, UK. Butterner, T and Women Lutz (1990), Estimating fertility responses to policy measures in the German Democratic republic, Population and Development Review 16(3): 539-55. Calhoum, CA and T.S. Espenshade (1986) Childbearing and wives' foregone earnings, Population Studies, 42(1): 5-38. Calot, G and J Hecht (1978), The control of fertility trends, in Council of Europe, Population Decline in Europe; Implications of a Declining and Stationary Population, London: Edward Arnold: 178-96. Caudill, SB and F.G. Mixon Jr. (1993), A note on the effects of AFDC payments on birth rates, International Review of Economics and Business, 40(4): 379-84. Cavouriaris, M, K. Karamessini and H. Symeonidou (1995), Women and the employment rate in Greece: the causes and effects of variations in patterns of female participation and employment. Report for the European Commission. Working Paper, EC Network of Experts on the situation of Women in the Labour Market: Manchester School of Management, UMIST, UK. 44 Chesnais, JC (1987), Quand un peuple en divient deux:Une Allemagne et l'autre, Population et Societes, 209. Cigno, A (1991), The Economics of the Family, Oxford: Oxford University Press. Davies, H and H. Joshi (1994), The foregone earnings of Europe's mothers, in 0. Ekert-Jaffe' (ed.) (1994), Familles et niveau de vie: observation et analyse, INED: Paris. Department of Social Security (Britain) (1993) Income Related Benefits: Estimates of Take Up in 1989, London: DSS. Duncan, A, C. Giles and S. Webb (1995), The impact of subsidising childcare, Research Discussion Series No 13, Manchester: Equal Opportunity Commission. Duncan, GJ and S.D. Hoffman (1991), Teenage underclass behavior and subsequent poverty: have the rules changed?, in C. Jencks and P.E. Peterson (eds.) The Urban Underclass, Washington, DC: The Brookings Institution. Duncan, S (1995), Theorising European gender systems, Journal of European Social Policy 5(4): 263-284. EC Childcare Network (1994) Leave arrangements for workers with children: a review of leave arrangements in the Member States of the European Community and Austria, Finland, Norway and Sweden, Equal Opportunity Unit, DGV. Brussels: European Commission. Ekert, 0 (1986) Effets et limites des aides financieres aux familles:un experience et un modele, Population 41(2): 327-48. Errmisch, J (1988), Economic influences on birth rates, National Institute Economic Review, Nov: 71-81. Errmisch, J (1991), Lone Parenthood: An Economic Analysis, Cambridge: Cambridge University Press. Ermisch, J (1996), The economic environment for family formation, in D. Coleman (ed.), Europe's Population in the 1990s, Oxford: Oxford University Press. Ermisch, J and R.E. Wright (1991), Employment dynamics amongst British single mothers, Oxford Bulletin of Economics and Statistics 53(2): 99-122. Esping -Andersen, G (1990), Three Worlds of Welfare Capitalism. Cambridge: Polite Press. Folbre, N (1994), Who pays for the kids?, London: Routledge. Garfinkel, I and S. S. McLanahan (1986), Single Mothers and Their Children: A New American Dilemma, Washington, DC: The Urban Institute Press. Gauthier, AH (1994), The measured and unmeasured effects of welfare benefits on families: implications for Europe's demographic trends, in D. Coleman (ed.), Europe's Population in the 1990s, Oxford: Oxford University Press. Gauthier, AlH and J. Hatzaius (1997), Family benefits and fertility: an econometric analysis, Population Studies, 51: 295-306 Gauvin, A, C Graine and R Silvera (1994), The evolution of employment forms and working time among men and women in France: towards the integration or the segmentation of the labour market. Report for the European Commission. 45 Working Paper, EC Network of Experts on the situation of Women in the Labour Market: Manchester School of Management, UMIST, UK: 74 Gustafsson, S and F. Stafford (1992) Child care subsidies and labor supply in Sweden, Journal of Human Resources 27(1): 204-230. Hausman, JA (1980), The effect of wages, taxes and fixed costs on women's labor force participation, Journal of Public Economics 14: 161-194. Heckman, JJ (1974), Effects of child-care programs on women's work effort, Journal of Political Economy 82(2, Part II): S 136-163. Heitlinger (1993) Hofferth, SL and D.A.Wissoker (1992), Price, quality and income in child care choice, Journal of Human Resources 27(1): 70-111. Jenkins, SP (1992), Lone mothers' employment and full-time work probabilities, Economic Journal 102: 310-20. Joshi, H (1990), The cash opportunity cost of childbearing: an approach to estimation using British data. Population Studies 44: 41-60. Joshi, H, S. Macran and S. Dex (1996) Employment after childbearing and women's subsequent labor force participation: evidence from the British 1958 birth cohort, Journal of Population Economics 9: 325-348. Joshi, H and P. Paci (1998), Unequal Pay for Men and Women: Evidence from the British Birth Cohort Studies, Cambridge, Mass: MIT Press. Joshi, H, P. Paci and J. Waldfogel (1996), The wages of motherhood: better or worse?, STICERD Discussion Paper WSP/121. London: London School of Economics. Leibowitz, A, JA Klerman and LJ Waite (1992), Employment of new mothers and child care choice: differences by children's age, Journal of Human Resources 27(1): 112-133. Lewis, J (1992), Gender and the development of welfare regimes, Journal of European Social Policy 2(3): 159-173. Lopes, MC and H. Perista (1995), Women and the employment rate in Portugal: the causes and effects of variations in patterns of female participation and employment. Report for the European Commission. Working Paper, EC Network of Experts on the situation of Women in the Labor Market: Manchester School of Management, UMIST, UK. Machin, S and J. Waldfogel (1994), The decline of the breadwinner: changing shares of husbands' and wives' earnings in family income. Discussion paper WSP/103, Welfare State Programme. Suntory-Toyota International Centre for Economics and Related Disciplines, London School of Economics. Macran, S, H. Joshi and S. Dex (1996) Employment after childbearing: a survival analysis, Work, Employment and Society 10(2): 273-296. McRae, S (1991), Maternity Rights in Britain: The Experience of Women and Employers. London: Policy Studies Institute. McRae, S (1993) Returning to work after childbearing: opportunities and inequalities, European Sociological Review 9,125-138. 46 Meulders, D (in collaboration with C. Hecq and R. Ruz Torres) (1994), Changing patterns of work and working-time for men and women in Belgium: Towards the Integration of the Segmentation of the Labour Market. Report for the European Commission. Working Paper, EC Network of Experts on the situation of Women in the Labour Market: Manchester School of Management, UMIST, UK: 21 Meulders, D (in collaboration with C. Hecq and R. Ruz Torres) (1995), Women and the employment rate in Portugal: the causes and effects of variations in patterns of female participation and employment. Report for the European Commission. Working Paper, EC Network of Experts on the situation of Women in the Labour Market: Manchester School of Management, UMIST, UK. Michalopous, C, P.K. Robins and I. Garfinkel (1992), A structural model of labour supply and child care demand, Journal of Human Resources 27(1): 166-203. Moffitt, R (1983), an economic analysis of welfare stigma, American Economic Review 73(5): 1023-35. Molto', ML (1995), Women and the employment rate in Spain: the causes and effects of variations in patterns of female participation and employment. Report for the European Commission. Working Paper, EC Network of Experts on the situation of Women in the Labour Market: Manchester School of Management, UMIST, UK. Moors, H and R. Palomba (eds.) (1995), Population, family and Welfare. Vol. 1. A Comparative Survey of European Attitudes, Oxford: Clarendon Press. Murphy, MJ and 0. Sullivan (1985), Housing tenure and family formation in contemporary Britain, European Sociological Review 1(3): 230-43. OECD (1995), Employment Outlook: 179 Owen, S and H. Joshi (1990) State equality and the state pension, Fiscal Studies 1(1). Plantenga, J and M Sloep, (1995), Women and the employment rate in the Netherlands: the causes and effects of variations in patterns of female participation and employment. Report for the European Commission. Working Paper, EC Network of Experts on the situation of Women in the Labour Market: Manchester School of Management, UMIST, UK. Popenoc, D (1988), Disturbing the Nest: Family Change and Decline in Modern Societies, New York: Aldine de Gruyter. Popenoc, D (1991) Family decline in the Swedish welfare state, Public Interest, Winter. Ribar, DC (1992), Childcare and the labor supply of married women: reduced form evidence, Journal of Human Resources 27(1): 134-165. Robins, P and D Blau (1991) Childcare demand and labor supply of young mothers over time, Demography 28(3): 33-54. Robins, P, NB Tuma, and KE Yeager (1980) Effects of SIME/DIME on changes in employment status, Journal of Human Resources 15(4): 545-73. Rubery, J, C. Fagan and M. Smith (1995) Changing patterns of work and working- time in the European Union and the impact on gender division. Report for the 47 European Commission, DGV - Equal Opportunity Unit. Brussels: European Commission: Table 1.13 Rubery, J, M. Smith C. Fagan and Discrimination Grimshaw (1998), Women and European Employment, London: Routledge Ruhm, C (1997), The economic consequences of parental leave mandates: lessons from Europe, Quarterly Journal of Economics. Sainsbury, D (1994), Gendering Welfare States, London: Sage. Sandqvist, K and B.-E. Andersson (1992), Thriving families in the Swedish welfare state, Public Interest, Fall. Vinning, DR Jr. (1984) Family salaries and the East German birth rate: a comment, Population and Development Review 10(4): 693-6. Waldfogel, J (forthcoming) Understanding the 'Family Gap' in pay for women with children, Journal of Economic Perspectives. Waldfogel, J, Y. Higuchi and Market Abe (1998) Maternity leave policies and women's employment after childbirth: evidence from the United States, Britain and Japan, STICERD Discussion Paper CASE/3. London: London School of Economics. Walker, 1 (1990), The effects of income support measures on the labour market behaviour of lone mothers, Fiscal Studies 11: 55-755. Ward, C, H Joshi and A Dale (1996), Income dependency within couples, in L. Morris and S Lyon (ed.) Gender Relations in Public and Private Life: New research perspectives, London: Macmillan: 95-120. Whittington, LA (1992) Taxes and the family: the impact of the tax exemption for dependents on marital fertility, Demography 29(2): 215-226. Zhang, J, J. Quan and P Van Meerbergen (1994), The effects of tax-transfer policies on fertility in Canada, 1921-88, Journal of Human Resources 29(1): 181-201. 48 Social Protection Discussion Paper Series No. Title 9801 World Bank Lending for Labor Markets: 1991 to 1996 9802 Export Processing Zones: A Review in Need of Update 9803 Building an Environment for Pension Reform in Developing Countries 9804 Risks in Pensions and Annuities: Efficient Designs 9805 The Hungarian Pension System in Transition 9806 Government Guarantees on Pension Fund Returns 9807 A World Bank Perspective on Pension Reform 9808 Women and Labor Market Changes in the Global Economy: Growth Helps, Inequalitier HIurt and Public Policy Matters 9809 Financing the Transition to Multinillar 9810 Pension Reform in Britain 9811 An Alternative Technical Education System: A Case Study of Mexico 9812 The Role of Choice in the Transition to a Funded Pension System 9813 Unemploymc-. 3enefits 9814 Family Allowances 9815 The Quest for - ion Reform: Poland's Security through Diversity 9816 Getting an Earful: A Review of Beneficiary Assessments of Social Funds 9817 Supervising Mandatory Funded Pension Systems: Issues and Challenges 9901 Active Labor Market Programs: A Review of the Evidence from Evaluations 9902 World Bank Lendin- for Labor Markets: 1991 to 1998 9903 A Bundle of Joy or an Expensive Luxury: A Comparative Analysis of the Economic Environment for Family Formation in Western Europe 9904 Social Protection as Social Risk Management: Conceptual Underpinnings for the Social Protection Sector Strategy Paper 9905 The Effects of Legislative Change on Female Labour Supply: Marriage and Divorce, Child and Spousal Support, Property Division and Pension Splitting Summary Findings In the Europe of the 1 990s low fertility rates are a prime concern. Contrary to the general view that associates declining fertility with increasing women's participation in paid work, this paper shows that amongst the EC-] 5 countries, those where women are more active in the labor market are also those where fertility rates are highest. These are not the countries where large families are a tradition. On the contrary, they are those where women's contribution to the household budget is the highest and where the State provides the strongest support to family formation and gender equality. The low cost of childbearing allows women in these countries to be both mothers and earners. The others may be forced to choose. HUMAN DEVELOPMENT NETWORK