Outcomes of the Two 1990s Family Policy Reforms at the Turn of the 2000s in Finland

Finnish family policy faced two rather different reform waves in the 1990s. They were justifi ed by saving public spending but relied on different social policy philosophies. The article assesses the impact of the reforms on public spending and poverty rates at the turn of the 2000s. The method is based on tax-benefi t models and representative micro data, i.e. on static microsimulation. The results suggest that the increased poverty is due more to changes in the socioeconomic structure than changes in the levels of family policy benefi ts, even though the biggest cutbacks in the reforms focused on families with small children and single parents. The reforms also had an impact on gender relations by offering more incentives to reinforce than alleviate the traditional division of child care and paid work between the parents. That in part may have affected the mothers’ labor market position and increased income differences between families.


Introduction
The outcomes of the economic turbulence and welfare state reforms in the 1990s have been studied extensively in Finland. The researchers generally accept that the 1990s was a decade that restructured the economy and labor market in Finland. Economic recession and budget defi cits at the beginning of the decade led governments to cut welfare state provisions. However, it is argued that in spite of the cuts, the Finnish welfare state model is still there, even if less generous and less universal than it used to be (Kiander and Virtanen 2002;Kalela et al. 2002;Hiilamo 2002).
Even though poverty and income inequality in Finland are still among the lowest in the OECD, the increase of poverty among families with children was brought into public focus in the early 2000s. The cuts in family benefi ts were blamed as one reason for this phenomenon (Heikkilä et al. 2003;Sauli 2001). Traditional research based on empirical data cannot distinguish between the impact of policy and other factors. In this article the microsimulation method (Harding 1996) is used to assess a hypothetical question: what would the poverty rates have been at the turn of the 2000s, if no reforms on family benefi t had been introduced in the 1990s. The emphasis is thus on trying to separate the potential impact of family benefi t reforms from other factors which, perhaps, also increased poverty. With the same method it is possible to calculate the cost of the reforms on public spending and assess which part of the increase/decrease in spending can be explained by the reforms and which part by other factors. "Other factors" are, of course, a big category of structural, economic and sociodemographic changes, including the indirect effects of the reforms on the behavior of the parents, which also may have had cumulative effects on the macro level and on politics, etc. (e.g. Klevmarken 1997). Potential behavioral effects, caused by reforms and other factors, are discussed but not separated from other structural changes.
The contents of the paper are as follows: The fi rst chapter presents the background of Finnish family policy, the main features of two quite different reform waves and major changes in the socioeconomic structure of families in the 1990s. The next chapter introduces the main outcomes of the reforms on benefi t spending and poverty rate. The reforms are then assessed in greater detail and the chapter ends in a summary of the reform outcomes from the perspective of families in different situations. The last chapter summarizes and discusses the fi ndings.

Starting point in 1991 and the main characteristics of the two reform waves in the 1990s
During the 1990s there were three different government coalitions, and -it can be said -three different family policy models in Finland. Since 1992 and 1995 the Parliament Act was amended in a way that annual budgets and new laws, except the constitution, needed only a simple majority in the Parliament to be revised, whereas the former practice needed a majority of two thirds. That also gave majority governments more power to follow through reforms. From the late 1960s to the late 1980s most government coalitions were based on cooperation between the Social Democrats and the Center (former Agrarian Party), strengthened often with communists and minor parties from the political center. To put it very simply, in the transfer system the Social Democrats and the right-wing parties have often advocated earnings-related benefi ts, and the Center together with the communist parties universal fl at rate grants. Towards the 1980s the upgrading of social services, e.g. in child day care, was advocated especially by the left-wing parties, where as most other parties favored cash transfers. (Mikkola 1991, 146-147;Ervasti 1996.) The family benefi t models from the years 1991, 1994 and 1996 were chosen for comparison. The year 1991 represents an outcome of generous welfare policies from the 1980s, although the cabinet was a new kind of coalition with the Conservatives and the Social Democrats (1987)(1988)(1989)(1990)(1991). The fi rst symptoms of recession were already apparent in 1991, but no cuts had yet been introduced into the benefi ts. The family benefi t model of 1991 can be classifi ed as representing the Nordic social democratic welfare state model (Esping-Anderssen 1990) or the encompassing social policy model (Korpi and Palme 1998), as work history was rewarded in family benefi ts, but also those with no earnings were entitled to fl at rate benefi ts.
After the parliamentary election in 1991 a government coalition of the Center and the Conservatives formed the cabinet in 1991-1995. The year 1994 represents outcomes of the fi rst reform wave, but also a time of deep economic recession. By 1994 unemployment had increased from an overall 3% in 1990 up to 18% for men and 15% for women. The annual growth in the real GDP had been negative, on average 3. 8 percent, between 1991-1993(Honkapohja and Koskela 2001. The family benefi t model was redesigned. Cuts were focused on earnings-related benefi ts but fl at rate benefi t levels were increased for parents with no or low taxable income. The family policy "package" of 1994, in particular, was justifi ed by the aim of improving the situation of small income families (Gov. Prop. 75/1993).
In 1995 a government coalition of the Social Democrats and the Conservatives formed the cabinet, and this time for two electoral periods. The year 1996 represents the start of the second wave of reforms, which was implemented e.g. by freezing the benefi t levels until the early 2000s. In 1996 the economy had already improved, but the unemployment was still on a high level, at 14% among men and 15% among women. This government cut fl at rate benefi ts and tightened entitlement rights to earnings-related benefi ts but alleviated means testing in some respects. However, entitlement rights to child care services were enlarged. The reforms were justifi ed as a work incentive policy and the aim of decreasing income traps (VNK 1996;Laine 2002).
In 2000 the economy had recovered but it had also undergone changes. Unemployment had decreased but still exceeded the fi gures of the early 1990s, and the effects of the second reform wave impacted fully on the population. Family benefi t levels were at their lowest level for the entire decade, and poverty among families with children had doubled. What part did the reforms play in these outcomes? Would the "old" family benefi t model of 1991, which rewarded work and taxable income more, have lowered poverty more effectively in 2000 than the "generous fl at rate" model of 1994 family benefi ts, or vice versa? Or does the model matter at all, but rather the size of expenditures? These questions will be studied with three major family benefi t systems: 1) Child allowance -tax deduction (in respect of children) system, paid for children under 17 or 18, 2) Parental leave benefi ts, paid for about 10-11 months at the time of the child's birth and 3) Child home care allowance paid for families after the parental leave period until the youngest child is 3, if they do not place their child in public child day care services. The economic importance of chosen benefi ts is presented as their shares in all family benefi ts in Figure 1. Their share in the total expenditure on family benefi ts and services has varied between 60-70 percent and in total family cash transfers around 90 percent in the 1990s. The total expenditure on family support, including child care and other services, was about 4.2 billion € in 2000 compared with 4.7 billion € in 1991 and 4.5 billion in 1994 (at 2003 prices). The expenditure decreased from 4.2 percent to 3.1 percent of the GDP in Finland in 1991 In order to illustrate changes in benefi t levels among individuals, a simple example has been selected to represent the reforms. Figure 2 illustrates gross earnings and benefi t levels (at nominal prices) of a single parent with one child. The parent receives either average female earnings, an earnings-related parental leave benefi t adjusted with this salary, minimum parental leave benefi t or maximum child home care allowance with a means tested supplement. Non-taxable child allowance including the value of tax deductions and earnings-related and minimum level unemployment benefi ts are also presented. The minimum level of unemployment benefi t seems interesting especially in relation to the level of child home care allowance (e.g. Sipilä and Korpinen 1998;Forssén et al. in this volume). Figure 2. Gross family benefi ts (nominal value) and average female earnings, euros per month and net benefi ts, % of net earnings (Example: single parent with one child below 3 years of age).
Source: Calculations by the author using the TUJA-model (Salomäki 1996). Women's earnings are based on the average female earnings index, Statistics Finland. Annotations: PLB = Parental leave benefi t, CHCA = Child home care allowance with means tested supplement, Child A+Tax D = Non-taxable child allowance and tax deduction from wage income.  1988 1990 1992 1994 1996 1998 2000 2002 2004 Euro per month Single parents are entitled to at least a public maintenance allowance, if they do not receive alimony from the non-resident parent. The allowance is not presented in the fi gure. It would almost double the non-taxable child supports of the single parent. As can be seen, the fi rst reform wave increased the net income level of minimum parental leave benefi t and child home care allowance, and decreased the compensation of earnings-related parental leave benefi t. The option of having children and taking care of them at home for a longer time was relatively well supported economically. During the latter part of the 1990s, however, taking care of children at home became economically much less advantageous. The benefi ts would be higher, if the parent had more children. If the parent had a spouse, the level of child home care allowance would vary between the level presented and the minimum parental leave benefi t, depending on the partner's income level, due to means testing in the child home care allowance supplement.
The levels of different family benefi ts are linked in different ways to the situation of the benefi ciary. Changes in the number of children, income source, income level, family type and even the district of residence do matter. In order to have some idea about other factors impacting on the end results, an overview of socioeconomic and demographic changes in the 1990s is presented in Table 1. Also some characteristics of the development of the Finnish breadwinner model and family type are included. Table 1 implies a decreased share of the "traditional" two-breadwinner model, in many ways. There were fewer mothers with earned income among the couples and among single parents with small children in 2000 than a decade earlier. The employment and unemployment situation was worse in 2000 than in 1990, but better than in the years of recession. Relatively high long term unemployment compared with other groups can be identifi ed among single parents and mothers whose youngest child was 3-6 years of age, after the child home care allowance period. The increased temporary employment in the 1990s has deteriorated women's status in the labor market more than that of men (Kauhanen 2002;Sutela 1999). It, however, should be noted that the situation has deteriorated among mothers with small children, whereas the situation of mothers with older children has recovered (Haataja and Nyberg 2005;Haataja 2004b).
In the early 1990s there was a small baby boom. The number of families with children increased slightly, but has since then decreased. This is also due to the decreasing population of women of fertile age, and to the structure of families: the number and share of big families with at least three children increased, but the number of families with one or two children decreased. Another feature was the increase of the most vulnerable family group: single parents.

What if no reforms had been made?
The cross sectional micro data of the Income Distribution Surveys (IDS) by Statistics Finland is used for empirical analysis. The IDS samples represent the whole household population and the main information for individual and household income is based on the registers. The sample also uses interview information, which has been valuable in assessing the socioeconomic position of family members. IDS data sets are recast as "model data" sets for simulation. After modifying the data, actual taxes and transfers can be controlled and "manipulated" with algorithmic models containing fi ctitious or real tax-transfer rules of different years. This study uses model data and the static microsimulation model SOMA of the Ministry of Social Affairs and Health (Haataja 1998;Parpo 2004). The quality of models is dependent on the level of details available from the data to simulate taxes and transfers. The reforms often change the information that register authorities collect concerning benefi ts. That weakens the transferability of models from one period's data set to another, which has to be taken account in interpreting the results (for more about the microsimulation model, see e.g. Harding 1996; Sutherland 2001).
The income concept of this study is narrower than the concept of disposable income in the IDS data. The income components are 1) gross taxable income, 2) minus taxes and social security contributions, 3) plus non-taxable child allowances and single parent's maintenance allowance or alimony from the non-resident parent and 4) equals net income. This income concept describes the economic situation and disposable income of families before the impact of means tested housing allowance and living support. That is also why the poverty rates of the study are higher than e.g. in published statistics, but the trend of doubled poverty rates is the same.
First we assess which part of the decrease in the public spending on family benefi ts refers potentially only to reforms, and which part to other factors, such as the economy and socioeconomic and demographic changes, called structural changes. The results are presented in Appendix 1. First the actual expenditure of parental leave benefi ts, child home care allowance and child allowance with tax deductions for children in 1991, 1994 and 2000 are summarized from the IDS data (in 2000 prices). The total changes of the spending in 1991-2000 were minus 23 percent (631 million €) and minus 20 percent (553 million €) in 1994-2000. Secondly, the simulated expenditures and their changes in the respective periods are calculated. Simulated expenditures represent hypothetical situations resulting if no reforms had been made in 1994 or since 1996 and the benefi t model of 1991 or 1994 had been in effect. The changes in spending occur now only if the sociodemographic structure changes, as it did (Table  1). If no reforms had been made, structural changes would have decreased total spending about 7 percent, which is 32 percent of the total decrease of 23 percent in spending in 1991-2000. Correspondingly, the impact of structural changes on total spending in 1994-2000 was 6 percent, amounting to 29 percent of the total decrease of 20 percent. (Appendix 1.) The rest of the decrease in family benefi ts presents the impact of policy and reforms. In 2000 the families were paid 431 million € (20%) less than they would have been paid in the benefi t model of 1991, and about 394 million (18%) less than they would have been paid in the benefi t model of 1994. The impact of the reforms on the decrease in spending is calculated as the difference between the total and the simulated changes.
The reforms in the 1991 model decreased the total costs by almost 16 percent. This means that the reforms explained a majority, about 68 percent, of the total decrease of 23 percent in spending on family benefi ts in 1991-2000. Correspondingly the reforms in the 1994 model decreased total spending on family benefi ts almost 15 percent, meaning 71 percent of the total change of 20 percent in 1994-2000. (Appendix 1.) To summarize the results above, it can be said that the reforms had a major impact of 68 and 71 percent on the decrease of spending on family benefi ts. Correspondingly, the impact of structural changes varied from 32 to 29 percent. More detailed benefi t analyses in the next chapter, however, will demonstrate that some of the structural changes may be partially caused by policy changes, as well.
Final savings in public spending were not as much as cross savings, because of losses in tax revenues from lower taxable benefi ts. The net decrease of disposable income in 2000 was 394 million € compared with the model of 1991 and 341 million € compared with the model of 1994. Furthermore, the total net income of families in 2000 was on average about 2 percent less than the models of 1991 and 1994 would have produced (Haataja 2003). How these macro changes predict changes in poverty rates will be seen in Table 2.
Actual poverty rates of market income increased dramatically among families with children during the recession years but decreased almost down to the same level since then, as is seen in Table 2. However, the poverty rates of net income did not increase in 1991-1994. The impact of the poverty-alleviating effects of the tax-transfer system was at its highest at that time. By the year 2000 the poverty-alleviating effects of the tax-transfer system had become weaker than ever, and the poverty rates of net income were doubled compared with the early 1990s.  1991, 1994, 1996 and 2000 in reality and in potential situations, if reforms had not been made, %.
Replacing existing family benefi ts with the benefi t models of 1991 and 1994 does not seem to change the end results as much as was perhaps expected ( Table 2)   3.9 7.4 1 Poverty rates are calculated for families with income less than 50% of the adjusted net median income of the whole population each year. Net income is family income after taxes added with nontaxable child allowances, i.e. income before means tested housing allowance and living support. Adjustment is done per consumption unit of the "old" OECD scale, where one adult gets the value 1, the next adults a value of 0.7 and children under 18 a value of 0.5.

Reforms and outcomes in more detail
Universal child allowance -tax deduction in respect to children The child allowance system was developed in the late 1940s as a universal non-taxable allowance, and is normally paid to the mother. Since the 1960s and the 1980s the payments per child for the second and further child up to the fi fth child have increased. This feature of pronatalism holds also through the two reform waves in the 1990s, even though the principles changed a little (Hiilamo 2002). Families have been entitled to child allowances for every child below the age of 16 or, since 1994, below 17.
Individual taxation was introduced in 1976, but with a large variety of tax deductions in respect to dependent children and single parenthood. Some of the deductions were aimed at the spouse with the higher income and some at the spouse with the smaller income, and covered children less than 18 years of age. In practice there were specifi c deductions for both mothers and fathers. The tax reform in 1989 simplifi ed the deduction system and liberated the rules regulating who in the family was entitled to child deductions. In practice, spouses with higher income, i.e. mostly fathers, took advantage of the new system (Haataja 2002).
Tax deductions can be fully utilized only by those with enough taxable income. That was also the argument when tax deductions were abolished by the fi rst reform wave in 1994. The loss was partially compensated by increasing the level of non-taxable child allowances. A special single parent tax deduction was also replaced with a single parent supplement of the child allowance, the same amount per child, in the reform. The second reform wave introduced cuts for child allowances in late 1995. The cuts were greater the more children there were in the family, but the level of allowances remained much higher than in the early 1990s. The nominal value of the child allowances and the single parent supplement stayed at the same level until the year 2002. Individualizing fi scal rights, i.e. abolishing child deductions in taxation, tightened up fathers' and single parents' taxation. That potentially alleviated income distribution between spouses, but measured with the combined net income of two-parent families, they either gained or lost from the reform. The immediate impact of the reform in 1994 compared with the situation before the reform in 1993 was that the majority of single parents (58%) were losers compared with less than 30 percent of the couples. In 2000, the model of 1991 would have benefi ted the majority of families. However, about a quarter of single parents and about three percent of couples would have gained under the reformed system in 2000 compared with the 1991 model (Haataja 2002(Haataja , 2003. The gainers could be found among families with low or no taxable income and more children than in the average family, and they totalled more in 2000 than in 1991.

Parental leave benefi t
The maternity leave benefi t was adopted in 1964 as a part of statutory sickness insurance. From the beginning the benefi t was aimed at all mothers, gainfully employed or not, with earnings-related or minimum benefi t. Since 1978 fathers have had the right to take two weeks of paternity leave simultaneously with the mother, and since 1980 fathers have had right the to share maternity (since 1985 parental) leave. Paternity leave, however, shortened the total parental leave period until the early 1990, and the father could use his entitlements to both leaves only with the mother's consent. Parents with a work contract have the right to parental leave, which accumulates work-related benefi ts and paid vacation days. In the insurance reform of 1981 the parental leave benefi t became taxable income and the ceiling was abolished. The compensation rates have income brackets so that the compensation rate is higher for lower earnings and decreases in the higher income brackets. The parent with a higher income and single parents were entitled to provider's supplements for siblings of the newborn.
In 1991 the mother's consent was abolished and the parents were equal in negotiations about the leaves. Paternity leave was extended from two to three weeks, of which the third week did not decrease the parental leave period. Also parental leave was lengthened by two weeks. In 1991 maternity leave was 105 working days (about four months), of which 30-50 days were meant to be taken before the child's birth. Parental leave was about seven months, with the child almost 10 months of age by the end of the leave period.
The fi rst wave of reforms increased the level of minimum benefi t and decreased the earnings-related part. It also abolished the provider's supplements in 1994. The fathers' independent paternity leave was lengthened from one week up to a total of three weeks in 1993. Since then none of the paternity leave weeks decrease parental leave, but two weeks of the total parental leave were cut, justifi ed by "from now on parental leave is cut from all families independent of the fathers' use of paternity leave "(Gov. Prob. 313/1992). The reforms increased the use of paternity leave. The extension of parental leave period slightly increased also fathers' share on parental leave, but later cuts in the leave period decreased participation, which all in all has been very low the entire time (Table 3).
The second reform wave from 1996 onwards cut back the level of the minimum benefi t. After that its nominal value stayed untouched until 2001. The reform of 1996 also changed the adjustments of the earnings-related part of the benefi t. In 1991 all earned income increased the level of the earnings-related benefi t, but since 1994 there has been a low income bracket to be eligible. Since 1996 entitlement rights to the earnings-related benefi t were tightened by increasing the bracket from about 2,500 € in 1994 to 4,500 € in 2000 per year. Compared with the other Nordic countries, Finland went its own way in the 1990s (Haataja 2004a). Instead of developing the parental leave period and the 'daddy quota' inside the social security system and alleviating the division of care and paid work between the spouses, the focus was on child home care allowance and leave.

Child home care allowance and child day care
Child day care services and fees are not included in this study, but an outline of the developments in public child care services is given as necessary background information. In Finland there have been two different systems of public child day care provision since 1990: public child care services and child home care allowance with child care leave. The 1974 Act on Public Child Day Care committed the municipalities to organizing child care services on demand, but the supply fell far behind still in the 1980s. The child day care compromise between the Social Democrats and the center parties was adopted in 1985 (Bergqvist et al. 1999). It guaranteed all children under the age of three a legal subjective right to a place in municipal child day care, if their parents applied for one. Otherwise they were entitled to a child home care allowance. The quality of child care has been high and appreciated by Finns as well as others (Kamerman and Kahn 1991). In the early 1990s, however, the fees for public child care services could be freely decided inside the lowest and highest income limits at the local level. In practice that meant that families in similar situations might face a 4,000 € difference per year in child care costs from one municipality to another (Forssén 1998). The demand side of child care services was, however, for the most part met in the 1990s.
The economic recession in the 1990s pushed local authorities into cutting spending on child day care. This weakened the level of services by increasing group sizes and by decreasing permanent posts for trained personnel (Ikola-Norrbacka 2004). In 1997, however, the second reform wave standardized the level of municipal day care fees throughout the country by reference to family size and income, and in 1996 enlarged subjective rights for public child care services from children under 3 to children below 7 years of age (the school commencement age). In 1997 a system for buying private child care for children under school age was introduced. The private care allowance is paid directly to the carer or the care facility to avoid the former double taxation.
Child home care allowance consists of the basic benefi t and supplements for other siblings below the obligatory school age of seven, if they are not in public child care. The benefi t is taxable for the receiver but does not assume that the receiver should take care of the child her/himself. Allowance was originally a compensation for not using public child care services (Sipilä and Korpinen 1998). If the children were taken care of at home, as they mainly were, and family income was small, the carer was entitled to a means tested supplement. Employed parents who choose the child home care allowance and take care of their children at home are entitled to a leave from work with the right to return to the former job. The child care leave, however, does not accumulate work-related benefi ts as the parental leave period does.
The use of the child home care allowance was very high in the 1990s. About 60 percent of families with children under 3 used the allowance annually. That is partially why the share of children in public child care was one of the lowest in the Nordic countries towards the 2000s (NOSOSCO 2002). The child home care statistics do not tell by whom or how the children are taken care of if they are not in public child care. Unemployed parents could receive an unemployment benefi t and a basic child home care allowance, but not a means tested supplement until 1993. Since then the child home care allowance has cut the level of unemployment benefi t if paid to the same person. Many two-parent families, if one of the spouses was employed and the other unemployed, then changed the allowance so that the recipient was the employed parent. This increased the number of employed men (as well as women) as recipients of child home care allowance. It explains the relatively high share of men as recipients in 1993-1995 of child home care allowance, but not their activity in child care (Table  4). Finally in 1995 this possibility was abolished, and the child home care allowance, paid to either of the parents, if the other was unemployed, started to cut unemployment benefi ts. Only if the other parent was at home taking care of the child with the allowance, was the unemployed spouse's benefi t not cut. The reform altered the child home care allowance from an alternative compensation of not using public child care services towards typical social transfer systems where benefi ts are adjusted to each other to avoid double transfer payments. Table 4. Child care arrangements in the 1990s: the share of children in public child day care, and the share of families with children under the age of 3 receiving a statutory child home care allowance and municipal supplements.
The child home care allowance system belongs to the municipal child day care provisions and costs, which entitle the municipalities to a state grant. It is not a state transfer as e.g. the Norwegian cash-for-care system, and it cannot be used to compensate municipal child care fees, if used less than 30 hours per week, as in Norway (Ellingsaeter 2003). The parliament decides the level of statutory allowances and income limits for the means tested supplement, but municipalities can pay local supplements, if they want e.g. to reduce the demand for child care services. The frequency of local supplements has varied according to the employment situation and according to the statutory level of the allowance. In the 1990s, it could be seen that the higher the statutory basic benefi t and the higher unemployment, the fewer municipalities paid supplements. On the other hand, the higher employment (and potential demand for child day care places) and the lower the statutory part of allowance, the more common local supplements have been ( The share of spending on child home care allowances in total spending on family policy was 14 percent in 1991, 16 percent in 1993 and nearly 9 percent in 2000 ( Figure 1). The total cost decreased 22 percent in 1991-2000 and 39 percent in 1994-2000. In contrast to parental leave benefi ts the structural changes only accounted for about 30 percent of the total decrease in 1991-2000, and about 32 percent of the total decrease in 1994-2000. Reforms thus explain the majority (68-70%) of reduced spending on child home care allowance (Appendix 1).
The fi rst wave of reforms in the early 1990s increased the basic level of the child home care allowance, and the second wave of reforms dropped the levels again in 1996. The nominal level of the basic allowance has been untouched since then until the early 2000s. On the other hand, in order to alleviate income traps, the income limits for the means tested part of the child home care allowance were broadened, especially for larger families. The share of recipients of means tested child home care allowance increased from about 40 percent in 1991 to 68 percent in 2000. The high share can be explained both by the increase of low income families and by alleviated means testing. Alleviating means testing in child home care allowance as a part of work incentive policy may have increased support for dividing, not sharing, child care at home between spouses.

Outcomes by family types
Doubled poverty rates can be identifi ed in all family groups in 1991-2000. In particular, families with children under 3 have been affected by the reforms, because these families faced changes from several reforms simultaneously. However, the greatest poverty risks were met by single parents with children under school age, when their primary income was dependent on family or unemployment benefi ts. The generous fl at rate model of 1994 would have decreased their poverty rates remarkably, but the model of 1991 would sometimes even have increased poverty in 2000. That implies that among single parents there are both mothers who either could have benefi ted from a mixed tax deduction -child allowance system or the new system of child allowances. Two-parent-families and single parents with children already at school age depended less on the reforms and the benefi ts (Table 5.) Table 5. Actual and hypothetical poverty rates in 1991-2000 by family type and age of the youngest child (before the impact of means tested housing allowance and living support).
The impact of the reforms was different for partnered mothers and fathers. Compared with the 1991 model, fathers were losers, on average, and mothers were winners in 2000, because mothers would have benefi ted less frequently than fathers from tax deductions and they gained from increased child allowances. Compared with the 1994 model, reforms were fairly neutral in 2000 for fathers and mothers were losers, because since 1994 fathers were less frequently recipients of family benefi ts than earlier. (Table 6.) Table 6. Average net income of families with children in 2000 and difference of income compared with alternative benefi t models of 1994 and 1991, %.
The impact of the reforms was much higher than the average two percent on net income for other families entitled to parental leave benefi t or child home care allowance, compared with the models of both 1991 and 1994. The net income of these couples was 4-5 percent less in 2000 than it would have been from the benefi ts of 1994 or 1991. For single parents the losses were almost double compared with the losses for couples, if they received parental leave benefi t or child home care allowance.

Conclusions and discussion
This paper has studied the impact of two waves of family policy reforms in the 1990s in Finland. The fi rst reform wave in 1992-1994 changed the family policy principles towards a generous fl at rate model and reallocated family benefi ts for low income parents by increasing fl at rate benefi ts at the cost of cutting earnings-related benefi ts. The second wave since 1996 was justifi ed by work incentives and a poverty-trap alleviating policy in cutting fl at rate benefi ts, alleviating means testing and investing in public child day care.
The impact of reforms was measured as the volume of cuts on public spending and as changes in poverty rates in 2000. About one third of the total decrease in spending on family benefi ts was due to changes in the economic and sociodemographic structure and two thirds due to reforms since 1991 or 1994, but the impact of structural changes varied greatly between different benefi t systems. The magnitude of the structural changes appeared in the doubled poverty rates among families with children. The more generous family benefi t models of 1991 or 1994 could not have reduced the poverty rates in 2000 anywhere near the level of the early 1990s. An average loss for families with children was about 2 percent of yearly net income, but losses varied a lot among families and were greatest for the families dependent on child care benefi ts and for single parents.
The changes in the labor market have especially affected mothers of small children. Nowadays mothers on parental leave less frequently have their own income from employment, less frequently have a recent work history entitling them to earnings-related parental leave benefi t, less frequently have a permanent work contract making it possible to return to work after their leaves, or before the leave periods end in order to let the father share these leaves, and less frequently even have a spouse to bring earnings into the family, because the number and proportion of single parents has increased. By increasing benefi ts for mothers and decreasing support for fathers or working mothers, the policy principle encouraged the division of care and paid work between spouses. It is diffi cult, however, to say whether high fl at rate benefi ts actually decreased the incentive of mothers to look for work, because in the early 1990s there was practically no demand for employment. There was, however, a rather strong linkage between the mother's potential earnings and the choice of child care at home at the turn of 1990s: the lower the potential wage, the higher the probability of choosing child home care (Ilmakunnas 1997). Another study indicated that the level of child home care allowance did not matter as much as the availability of child day care places in the early 1990s, when take up rates of child home care allowance were compared between different municipalities (Kuismanen 1992). A follow-up survey on the work incentive policy in the latter part of the 1990s implied that the impact of the reforms concerning mothers of children between 1-3 years of age was negative on their labor supply (Laine 2002). No clear impact could be found on the fathers' labor supply. Among other groups investigated, employment increased at least a little in 1996-1999. On the other hand, it is unknown whether the demand for female labor had increased so much that the mothers would have had real choices between work and home in the latter part of the 1990s. The state fi rst increased the dependency of mothers on welfare and then decreased it, and the dependency of mothers on their spouse's income increased at the same time (Haataja 2004b). Some other comparative studies have recently implied that the socioeconomic structure and the level and distribution of market income sometimes explain poverty and income distribution between different welfare states better than the benefi t models (Fritzell and Ritakallio 2004;Cantillion et al. 2002;Kangas and Ritakallio 2000).
Concerning the methodological part, this study demonstrated some shortcomings in doing ex-post evaluations properly. The static microsimulation method offers, however, a new kind of information about ex-post materialized gross and net cost of reforms, potential effects on overall income differences, as well as about changes in individual income sources among mothers and fathers. Appendix 1. Total spending on family benefi ts divided into structural and reform based shares, in 1991-2000 and in 1994-2000 1 1991-2000 or in 1994-2000, and the sociodemographic structure of the population would have developed as it actually did. The impact of policy is the difference between actual and simulated spending. 1) Source: Income Distribution Data (IDS) and SOMA-model.