WINNERS OR LOSERS OF REFORM? GENDER AND UNEMPLOYMENT IN POLAND AND HUNGARY 1

Transcription

1 Review of Sociology Vol. 7 (2001) 2, WINNERS OR LOSERS OF REFORM? GENDER AND UNEMPLOYMENT IN POLAND AND HUNGARY 1 Christy GLASS Janette KAWACHI Department of Sociology, Yale University. 140 Prospect New Haven, CT 06511; christy.glass@yale.edu and janette.kawachi@yale.edu Abstract. In this paper we analyze whether and how unemployment patterns changed during the transition from socialism to capitalism in Hungary and Poland. Specifically, we test two competing theories that make predictions about women s employment outcomes as a result of market reforms. First, Fodor s theory of revalued resources predicts that women will have a significant advantage over men in the labor market as a result of attaining high levels of education and gaining substantial work experience in the servicesector under the socialist regime. Alternatively, the re-traditionalization theory predicts that women s employment will gradually deteriorate as a result of traditional gender ideologies that have re-emerged throughout the transitional period. In order to adjudicate between these competing theories, our analysis draws upon two major cross-national surveys conducted in Central and Eastern Europe in 1993 and By comparing labor market figures in Hungary and Poland in 1988, 1993, and 2001, our analysis maps the trajectory of men and women s unemployment from the period immediately preceding market reforms to the current phase of transition. Overall, we find modest evidence for both theories depending on the country of interest. Furthermore, while women, as compared to men, have sustained a relatively strong position in the labor market after over a decade of reforms in Hungary, the employment situation for women in Poland is becoming increasingly bleak. Keywords: gender, unemployment, market transition, Eastern Europe 1 The authors contributed equally to this article. We wish to thank Hannah Brueckner, Éva Fodor, Roger Gould, Gail Kligman, Eric Kostello, Iván Szelényi, and the participants of the Comparative Research Workshop at Yale University for helpful comments and suggestions. An earlier version of this paper was presented at the 3 rd Annual Graduate Student Retreat of the Society for Comparative Research in May Akadémiai Kiadó, Budapest

2 110 CHRISTY GLASS JANETTE KAWACHI INTRODUCTION The countries of Central and Eastern Europe are now over a decade into the transition from a socialist redistributive system to a market-oriented economy. With a few notable exceptions (Gal and Kligman 2000a, 2000b), and despite a vast literature dedicated to the dramatic politico-economic transformation, studies have failed to explicitly address how men and women are differently affected by these changes. To date, most research concerning issues of gender inequality in transition has been mainly theoretical and based largely upon limited qualitative and anecdotal evidence. 2 In addition, a general tendency throughout the transition literature has been to over-generalize the impacts of transitional processes across countries, with minimal attention to the precise mechanisms and institutions of change. Without a deeper understanding of how changes in the political, economic, and social spheres of post-socialist societies have differently affected social groups, the state remains limited in its ability to develop policies and programs that will lead the transformation down the most equitable path. The goal of this paper is to analyze whether and how unemployment patterns changed during the transition from socialism to capitalism in Hungary and Poland. Specific theories of gender and unemployment in the transitional labor market have often been inconsistent, speculative, and/or contradictory. By analyzing over-time, as well as cross-country survey data, we are in an ideal position to test and adjudicate among many competing claims. Specifically, we will test two sets of competing theories, both, which make predictions about gender-specific labor market outcomes as a result of transition. While one set of hypotheses predict that women will have significant advantages in the labor market as a result of market reform, the other set of hypotheses predict that women will be the losers of reform in terms of employment chances. Unemployment is an ideal way to measure emerging gender inequalities in the restructuring labor markets in Central Eastern Europe. Occupational segregation by sex, sex-based discrimination, and a sex-specific wage gap are all well documented features of advanced capitalist societies (England 1992; Blau 1998; Reskin and Roos 1990). Likewise, gender differences in unemployment rates have been offered as evidence of women s overall disadvantage in capitalist labor markets (Corcoran 1999). However, socialism ostensibly aimed at emancipating women from above through women-friendly labor market policies and practices and full equality for women in economic and political spheres. 3 Indeed, women entered the socialist labor force en masse, and the state enabled their participation through seemingly child-, 2 There are notable exceptions to this trend, including Fodor 1997, 1998; Fong and Paull In reality, although socialist regimes required full employment from women, little was done in the way of transforming the gendered division of labor in the home and family. The state s neglect of the relations in the private sphere was not inconsistent with communist ideology, which assumed that women s emancipation in all spheres would follow from women s equality in the labor force. However, as a consequence, women s participation in the formal labor market only added to the existing responsibilities women faced as caretakers, wives, and mothers. Moreover, women s participation in the socialist labor market was far from equal and was characterized by wage inequalities, legalized occupational sex segregation, and discrimination. For a good discussion of women s double burden under socialism and the consequences during the transition, see Corrin 1992.

3 WINNERS OR LOSERS OF REFORM? 111 family-, and woman-friendly social policies, including subsidized childcare, generous maternity and family leave allowances, job protection, and guaranteed healthcare. As a result, women enjoyed substantial educational and occupational opportunities under socialism. Understanding gendered patterns in unemployment will illuminate potential needs in wide-ranging areas of social policy, from poverty alleviation to labor market regulation, in transitional societies. In addition, if women in post-socialist countries do indeed exhibit advantages over or equal to men in the transitional labor market, identifying the mechanisms by which they attain and maintain those advantages will shed light on the causes of the seeming durability of gender inequality in the labor markets in advanced capitalist countries. Why Hungary and Poland? Because the market and market forces are proposed as the major mechanisms driving the predicted outcomes in the theories of labor market participation and gender inequality, our analysis will hold the level of marketization constant. 4 To this end, we will compare employment trends in two countries, Poland and Hungary, which followed somewhat similar paths of transition and achieved advanced levels of market development to date, relative to other countries in Central and Eastern Europe. Holding market development constant allows us to look at the trajectory of market transition and its effects on employment patterns. Thus, we will analyze unemployment trends at three time points, which cover the entire span of transition to the present. Our three time points include 1988, one year before the transition began, 1993, arguably a midpoint of the transition when many of the most damaging effects of reform, such as rapid economic contraction, skyrocketing unemployment, and increasing levels of poverty, were first being felt, 5 and 2000, more than a decade after the transition began, when the market was fairly well-developed and economic growth, albeit limited, was occurring in both countries. 4 We hold marketization constant to the extent possible. We recognize that even to the degree that Hungary and Poland appear equal or relatively equal on nearly all aggregate measures of development, these two countries pursued rather diverse strategies of transition. However, cross-country comparisons require some degree of simplification on several macro-level measures and, as such, are necessarily imperfect. Moreover, this paper does not aim at comparing market transition per se, but rather, specific dynamics within the overall process of transition labor market restructuring and the effects on men and women s employment. 5 Between 1990 and 1993, for example, unemployment rates in Hungary and Poland exploded. In 1991, official statistics report less than 2% unemployment in Hungary and 3.5% in Poland. By 1993, the equivalent figures were over 12% in Hungary, and nearly 16% in Poland (Fodor 1997).

4 112 CHRISTY GLASS JANETTE KAWACHI THEORIES OF GENDER AND UNEMPLOYMENT IN TRANSITION Revalued Resources Fodor s revalued resources theory predicts that state socialist-specific assets, such as high levels of academic education, experience, and positions in service sector occupations, while devalued under socialism, are being revalued under the new post-socialist regime (1997, 1998). This formulation builds upon job segregation theory, which posits that most jobs are segregated by sex and that such segregation is nearly universally bad for women, producing a gender gap in pay and the concentration of women in the least valued occupations. Fodor amends this theory to argue that whereas job segregation was undoubtedly a major disadvantage to women in terms of pay, prestige, and advancement under socialism, during the transition, women s concentration in certain types of jobs, particularly those in the service sector, may actually protect them from unemployment. Hence, occupational segregation may actually become a valuable resource for women in maintaining employment during times of radical economic restructuring. In addition to the revaluation of service occupations, Fodor points to the revaluation of academic credentials in transition as a major resource for protecting women in the labor market. Under socialism, women were more likely to pursue academic credentials, while men were more likely to pursue vocational training. 6 Not surprisingly, these educational differences corresponded to occupational differences, where women were concentrated in administrative or professional occupations while men were concentrated in manual jobs in industry and agriculture. Thus, although women were more academically educated than men, their educational credentials did not allow them to secure jobs in the most prestigious areas of the socialist planned economy heavy industry and production. During the transition, however, Fodor argues that women s educational credentials may evolve into a highly valued form of cultural capital not only in terms of academic achievement and credentials, but also in terms of skills such as fluency in languages, analytic skills, better self-presentation, and more flexible retraining possibilities (Fodor 1997: 486). These skills, combined with higher levels of educational attainment, may translate into greater security in the labor force for many women relative to their less educated and less skilled male and female counterparts. Retraditionalization of Gender Ideologies Proponents of the retraditionalization thesis point to the increasing prevalence of traditional gender ideologies in transitional societies, and posit that such ideologies are likely to translate into gender inequalities in the labor market. Observers argue 6 Relying on data from 1989 through 1995, Fuszara observes that, despite the increasingly privatization of education, and rising tuition requirements, women continue to outpace men in seeking higher education. However, women also continue to predominate in rather feminized majors, including humanities and pedagogy, while men predominate in more financially lucrative majors, including management, engineering, and computer science (Fuszara 2000: 263).

5 WINNERS OR LOSERS OF REFORM? 113 that such traditional ideologies are, in large part, the result of a backlash against socialism and against socialist policies that required full employment and active participation in politics for women, without concomitant changes in gendered division of labor in the household. In some instances, a return to the home for women has been proposed as a way to erase the damages done to society and to the family by the evils of socialism a process Gal and Kligman refer to as the sacralization of the family and of women s traditional roles within it (2000a). The rather substantial influence of the Catholic Church in society and politics has been proposed as a major factor in the strength, articulation, and penetration of traditional gender ideologies in transitional societies. 7 Evidence for the emergence of traditional gender ideologies is provided by ubiquitous public images of the highstatus stay-at-home wives and mothers, political dialogues centered on the needs and rights of women to stay at home with children, open calls by political and religious leaders for women s return to the home, the virtual disappearance of women as well as the rejection of women s issues from politics at the local and national levels, and the constant pressure for abortion restrictions on parliamentary agendas (Goven 2000; Wolchik 1993). 8 Market Discrimination Though the market discrimination thesis predicts outcomes consistent with the retraditionalization thesis, the argument points to structural changes in the political and economic organization of society rather than to gender ideologies and culture as the key factors limiting women s chances in the transitional labor market. More specifically, proponents of this thesis point to the precise channels through which capitalist reforms of the labor market and retrenchment of the welfare state may negatively affect women s employment chances. Here the market per se causes discrimination against women and, therefore, the development of capitalist markets in post-communist societies will parallel the rise of discrimination against women. 7 The elements of retraditionalization theory follow a similar logic as Swidler s model of culture s direct influence on action during periods of social transformation, or what she refers to as unsettled cultural periods. According to Swidler (1986), during periods of transformation, explicit, articulated, highly organized meaning systems (both political and religious) establish new styles or strategies of action (p. 278). However, Swidler argues that while ideologies, which emerge during unsettled periods, may have strong control over action, the long-term influence of such governing ideologies depends primarily on structural opportunities for their survival. Therefore, the long-term viability of women s exit from the labor force, if it has occurred, will likely depend on the structural ability of women and families to maintain traditional family roles while avoiding downward mobility or even poverty. 8 Though the retraditionalization theory predicts that women will exit the labor force either by choice or by force, practical economics makes the former path increasingly unlikely. Under socialism, the standard of full employment meant that all or most families and couples were essentially dual-earning households. The necessity of two incomes has only increased during the transition. Thus, unless households are willing to lower their standard of living, or the general wage structure shifts so dramatically in these countries as to increase overall male earnings, women are unlikely to leave work willingly. Given these unlikely exigencies, the more probable hypothesis is that women will be forced out involuntarily and kept out of the labor force through discriminatory practices by employers and/or unfair labor market policies.

6 114 CHRISTY GLASS JANETTE KAWACHI The theory of market discrimination argues that macro-level changes brought about by the disappearance of the socialist welfare state have simply made it more difficult for women (especially wives and mothers) to maintain secure and stable employment. These changes include the reduction or elimination of state-sponsored childcare benefits, which during socialism provided low-cost childcare for working women, 9 the elimination or reduction of maternity leave benefits and mandatory flexible schedules, 10 and massive reductions in the number of state-funded nurseries and kindergartens. 11 Furthermore, these trends parallel the disappearance of state protection and enforcement of antidiscrimination laws, which by many accounts has been accompanied by massive increases in sex discrimination in hiring and firing, and a rise in incidences of sexual harassment in the workplace 12 (Goven 2000; Kotowska 1995; Zielinska 2000). In terms of discrimination, observers note that because employers are increasingly responsible for providing maternity leave and childcare benefits, and because the state no longer enforces anti-discrimination laws, employers have real incentives to discriminate in favor of men or in favor of unmarried women without children (Heinen 1995; Fuszara 2000; Kotowska 1995). Therefore, discrimination in hiring and firing need not be the result of re-emerging traditional ideologies, but simply due to the fact that the financial risks of employing women of reproductive age are perceived by employers as prohibitive, particularly for firms struggling to survive in the nascent market economy. 13 Transitional Strategies and Market Discrimination in Poland and Hungary Although the market discrimination theory makes the crucial link between macrostructural change and gender inequality in the labor market, the theory fails to explain how the strategy and timing of reform may affect women s unemployment, in both the short and long-term. The theory of market discrimination is essentially a theory of gender relations in fully developed market economies and cannot, therefore, produce reliable predictions regarding the trajectory of those relations in transitional societies. Thus, what remains missing from this literature is a theoretical bridge which links the structural predictions of the market discrimination theory to the unique institutional processes of countries in transition. We therefore supplement the 9 Even where child-support payments continue to be paid, the payments are not indexed to the cost of living. The result, according to Zajicek (1995), is that such allowances can be as low as $.50 per child per month. 10 In 1996 in Hungary, for example, the parental leave allowance (EPA), which had previously paid new mothers up to 75% of a woman s average earnings from the end of her maternity leave (i.e., when the child was 6 months old) until the child turned 2 year old, was abolished. Also in 1996, a second form of parental allowance, known as FRA, which had previously been a universal benefit, became means-tested. For a more complete review of the changes in parental leave benefits, see Haney 1997, as well as Goven In Poland, for example, between 1989 and 1995, the number of nurseries declined by nearly 60%, while the number of kindergartens declined nearly 25%. (Kotowska 1995) 12 Neither Polish, nor Hungarian law, for example, currently protects women from sexual harassment. 13 Heinen (1995) points out that, at least in Poland, employers perceive women as risky and unreliable workers, with high rates of absenteeism due to family obligations. According to Heinen, this is especially true for young, married women who are at risk for becoming pregnant.

7 WINNERS OR LOSERS OF REFORM? 115 previous theory by arguing that the nature and pace of the reform process undertaken in both countries will, to a large extent, determine how, when, and whether working women will experience the negative effects of market discrimination. As noted above, although Hungary and Poland appear relatively equal across almost all aggregate measures of development by the year 2000, these two countries in fact pursued rather divergent strategies of transition. Thus, our country-specific theory of transition-based market discrimination points to differences between Poland and Hungary in terms of (1) initial transition strategy, (2) rate and form of initial privatization measures, and (3) timing of reforms of the socialist welfare system. Such differences, we predict, are likely to produce varied outcomes in terms of women s labor market chances in Poland versus Hungary. First, it has been widely argued that Hungary and Poland pursued divergent strategies of reform, particularly in the earliest stages of transition. Economists, in particular, have argued that while Poland initially pursued a radical, or shock therapy approach, Hungary followed a more gradualist approach to market reform (Róna- Tas 1996; Duke and Grime 1994; Mizsei 1993; Kolodko 1997). 14 The goal of Polish reforms immediately following 1989 was to privatize quickly despite probable negative short-term social and economic consequences. The shock therapy approach to reform was heavily endorsed by various Western-trained neo-liberal scholars and consultants, who asserted that a rapid and comprehensive transition strategy was the only path to successful and irreversible reform (Sachs 1990a, 1990b). Hungarian reformers, on the other hand, preferred a slower process of privatization in order to build upon existing institutions while maintaining a strong social safety net. 15 In addition to the pace of reform, Hungary and Poland also pursued varying strategies of initial privatization. While Poland prioritized the rapid privatization of the small, feminized tertiary sector, Hungary turned to large, male-dominated industrial enterprises in its early privatization program (Duke and Grime 1994). Finally, while Poland instituted major reforms of the socialist welfare system as early as , Hungary delayed major reforms of the welfare system until after the 1994 elections We use the categories of shock therapy and gradualism with caution. Though often used to contrast Hungary and Poland, these categories tend to over-emphasize the differences between the transition strategies. However, there were differences between the timing and strategies of Hungarian and Polish reform processes, and it is these differences that we wish to explore. 15 Part of the difference in initial strategy can be explained by differences between Hungary and Poland, which pre-date During the 1980s, for instance, Hungary s experiments with market socialism had led to the establishment of many of the institutions and practices of a market economy. In fact, by 1989, Hungary already had a relatively well-developed small business sector, particularly in the service industry. Duke and Grime (1994) argue that this unique legacy allowed Hungary to delay rapid small-scale privatization and instead focus on large-scale privatization. Poland, on the other hand, lacked a comparable small-business sector prior to 1989, therefore motivating rapid privatization of small businesses immediately following the fall of socialism. (See also Mizsei [1993] for a discussion of the political and economic motivations for Hungary s gradual approach). Initial widespread political support for the post-1989 Solidarity government probably also played a role in enabling the institution of radical reforms in Poland versus Hungary (Gomulka 1993; Duke and Grime 1994; Mizsei 1993; Róna-Tas 1996). 16 On the issue of Hungary s delayed reforms, Mizsei (1993) writes that from 1989 through the 1994 elections, the government instituted not one major structural reform of the social welfare system (p. 155). Indeed, until the parliamentary reform programs of , the Hungarian welfare system continued the socialist legacy of universal entitlement and complete inclusion of social welfare in the central budget (Goven 2000).

8 116 CHRISTY GLASS JANETTE KAWACHI Poland s strategy of radical reform led to massive economic recession, hyperinflation, and an unprecedented rise in unemployment and poverty. 17 In addition, as part of the larger reform strategy, Polish reformers revoked many of the protections previously offered by the state to vulnerable workers, including wives, mothers and unskilled, low-educated workers. For instance, while jobs were formerly protected for women on maternity and parental leave, restructuring firms were no longer forced to recognize such protection. 18 Thus, women on maternity leave (i.e., mothers of young children) were the first to be laid off in the early 1990s. 19 In addition, rapid reform led to the early classification and removal of non-productive workers. Not surprisingly, such workers were often young, married women (who had the potential to become pregnant and, thus, exit the labor force), as well as unskilled (especially female) workers (Fuszara 2000). At the same time Poland was instituting relatively radical reforms in the economic sphere, the state was also chipping away elements of the socialist welfare state as part of the larger goal of market reform. For example, already by 1991, childcare allowances a previously universal allowance provided by the state to all families with young children became means tested. In addition, financial responsibility for childcare previously guaranteed and provided by the central government was shifted downwards to local governments, which quickly instituted fees for admission (Kotowska 1995). Indeed, by 1991, average tuition for childcare facilities was equal to one-third of an average month s income in Poland (USAID 1991). Specific groups of women were likely to be disproportionately affected by the initial privatization scheme for several reasons. First, lay-offs and early retirement in the small tertiary sector were much more likely to affect women early on, simply due to the overrepresentation of women in such businesses. Whereas Hungary began small-scale privatization and the development of an institutional structure for policies and practices as early as 1980, Poland plunged into the process during the very initial stages of reform with little or no experience or preparation. Thus, from the outset, Hungary was in a relatively more stable position than all other post-socialist countries to implement a privatization program without the risk of severe labor market dislocations. The rapid shake-up of small businesses in the initial stages of Poland s transition, on the other hand, was likely to push many women out of the labor force through unemployment and early retirement. 17 In 1989, official statistics reported less than 1% unemployment. By 1992, unemployment in Poland affected more than 2.3 million people (Duke and Grime 1994). In addition, in a study conducted by the World Bank, poverty in Poland more than tripled between 1989 and 1993, while during the same period, poverty in Hungary increased only minimally (Milanovic 1994). 18 The 1990 Polish Employment Amendment removed the protection of jobs for women on maternity leave in firms that had liquidated and/or declared bankruptcy. Because such classifications applied to such a vast number of companies in Poland in the early 1990s, the effects on mothers labor market participation was immense (Bryant and Mokrzycki 1994). 19 This trend is supported by Fodor (1997), who found that women in Poland who were on maternity leave in 1988 were much more likely to be unemployed in 1993, compared to men and to women not on maternity leave in Women with young children faced no such penalty in Hungary in See also Fuszara (2000) for a description of the increasing obstacles to sustainable employment for women in Poland.

9 WINNERS OR LOSERS OF REFORM? 117 In addition, small businesses are least likely to be able to afford generous maternity leaves for their employees, given that such businesses operate with very few employees and very little revenue, compared to larger firms. Thus, women in such businesses are likely to be the hardest hit by the state s removal of income support and legal enforcement of mandatory childcare leave benefits and are likely to be the targets of active discrimination by employers. Furthermore, small servicesector businesses are much more difficult to monitor and regulate, especially in transitional societies where a solid legal structure and institutional framework for protecting employees in the private sector are unlikely to be sufficiently established. Small businesses compared to larger enterprises are notoriously bad at providing health care and other employment benefits and at upholding restrictions on discrimination. Therefore, not only was Poland s strategy of privatization likely to push women disproportionately into the ranks of the unemployed, but also was likely to make it difficult for female employees with families and/or children to sustain employment in the long run. Finally, we predict that what makes the effects of the timing and strategy of reform in Poland so harmful for women is that re-entry for the unemployed is extremely difficult (Kotowska 1995; Heinen 1995). Immediately following the institution of shock therapy in Poland, unemployment rates increased dramatically. Radical restructuring led to a shrinking labor market, where education, skills, and perceived reliability and flexibility became paramount to obtaining and maintaining employment in the new economy. Thus, once unemployed particularly in the earliest stages of transition reentry into the labor market was extremely difficult, especially for specific groups of women, namely, wives, mothers, and women who lacked skills and education. Furthermore, given the great difficulty of documenting, charging, and penalizing employers for anything less than blatant discrimination, and the strong disinclination of women in both countries to formally report such violations, employers can easily discriminate without much concern for punishment. It is estimated that in Poland, for example, unemployed women receive an average of seven times fewer job openings than their unemployed male counterparts (Titkow 1994). In addition, a USAID report noted that even as early as 1991, there were an estimated 74 unemployed women for every job offer for women, whereas there were only 20 unemployed men for every job offer for men (USAID, 1991). As a result, not only were women likely to exit the labor market very early on in the transition process, women were also unlikely to reenter even after moderate economic growth was achieved. By 1993, those most dependent on the state for protection in the labor market in Poland faced serious obstacles to sustainable employment. In Hungary, on the other hand, such major reforms were postponed, by and large, until the mid-1990 s, due largely to political strategizing. Until , a year after the second electoral cycle following the transition, the state continued to provide relatively generous and universal social welfare benefits, delayed certain aspects of the privatization program, and postponed massive streamlining of the workforce. Thus, for a while, Hungarian women unlike their Polish counterparts were able to maintain a strong hold in the labor market, due to the continuation of socialist-born policy measures meant to

10 118 CHRISTY GLASS JANETTE KAWACHI ensure job security and protection for all workers. In addition, because Hungary was most successful in the early privatization of large, state-owned enterprises in industry and manufacturing sectors, women were less affected than men by early lay-offs and structural unemployment. Finally, the delayed reforms of the small, femaledominated service sector in Hungary allowed women to maintain employment, to accumulate skills and experience in the new economy and, most importantly, to avoid the fate of early unemployment with little or no hope of re-entry. Summary of Arguments Overall, what theories in the past have failed to address is the confluence of structural patterns, institutional arrangements, and cultural factors that are likely to produce varied labor market outcomes in transitional societies. Fodor s theory of revalued resources offers a valid, yet ephemeral, explanation of women s labor market experiences during the initial stages of transition. However, offering only a narrowly structural and temporally limited account, Fodor s theory remains inadequate to explain gendered patterns of employment as the new social order develops and matures. In contrast, though, the theory of market discrimination presents some insightful propositions as to how working women may fare in a fully developed market economy, it fails to address how the unique institutional arrangements and changes in transitional societies may impinge upon such predictions. Finally, the theory of retraditionalization, though contributing an important cultural perspective to the literature, is remiss to obvious structural forces at play and fails to identify any specific mechanisms that will generate its predicted outcomes. To the existing literature we contribute a theory of institutional adaptation that has thus far been absent from the literature, which emphasizes the importance of transitional reform strategies in predicting labor market outcomes for men and women. Thus, taken together, our goal is to develop a synthesized theory of women s employment patterns in transitional societies, taking into account structural conditions, cultural trends and institutional processes. We hypothesize that in the few short years between 1989 and 1993, the bottom simply dropped out for Polish women, especially wives, mothers, and low-educated or unskilled women workers, in a way that it did not for their Hungarian counterparts. Largely due to the rate and form of the initial stages of Polish reform, women as wives and mothers were pushed out of the labor market and, once unemployed, found it increasingly difficult, even impossible, to re-enter the labor market. In Hungary, on the other hand, women had a few years to adapt to the new economy. These years served as a buffer for women against the harshest effects of market reform, namely, the rapid disappearance of social welfare benefits, job security, and the rapid removal of non-productive workers from the labor force. We further extend our hypotheses by arguing not only that women will predominate among the unemployed, but also that the experience, as well as the consequences, of unemployment will be substantially worse for women than men in Poland. Borrowing elements from both the market discrimination and retraditionalization theories, we predict that by 1993, women in Poland will be more likely to be

11 WINNERS OR LOSERS OF REFORM? 119 unemployed than men, but this will not necessarily be the case for women in Hungary. This cross-country divergence in female employment outcomes is due to both the more powerful position of the Catholic Church in Polish society throughout the transition period, as well as Poland s more aggressive and hasty approach to reform relative to Hungary. Furthermore, due to the immediate withdrawal of women-friendly benefits in Poland, we expect that already by 1993, Polish women will bear significant marriage and child penalties in terms of employment, making wives and mothers particularly those with low levels of education much more likely to be unemployed. By 2000, we expect such penalties to continue to exist in Poland and perhaps even worsen. As mentioned above, the sudden and massive dislocation of women in the labor market during the early stages of reform are likely to have long-term consequences, including feminized, long-term unemployment. In Hungary, on the other hand, we predict that by 2000, neo-liberal reforms although initially delayed will have had time to take hold and, therefore, Hungarian women s employment rates will begin to decline as well, though not to the degree of Polish women. DATA AND METHODS Data Source Our data are drawn from two major cross-country surveys conducted in Central/Eastern Europe in 1993 ( Social Stratification in Eastern Europe after 1989 ) and 2000 ( Poverty and Social Structure in Transitional Countries ). From the 1993 dataset, we use individual-level data taken from a nationally representative sample of 4,221 households in Hungary and 3,520 households in Poland. From the 2000 dataset, we employ individual-level data drawn from a nationally representative sample of 1,002 households, including an over-sample of 447 poor households in Hungary, and a nationally representative sample of 1,015 households, with an over-sample of 501 poor households in Poland. Together, these datasets allow us the unique opportunity to adjudicate among the above competing hypotheses regarding gender inequality in transitional labor markets across countries and over time. Methods and Analytical Strategy Along with various descriptive analyses, we use multivariate logistic regression to examine the factors that lead to unemployment in Hungary and Poland in 1993 and Our sample for the logistic regression models will consist of all those currently in the labor force (i.e., working, on maternity leave, or unemployed) between the ages of 20 and 59. Given these restrictions, we are left with 2,857 observations in 1993 and 617 observations in 2000 for Hungary and 2,380 observations in 1993 and 773 observations in 2000 for Poland.

12 120 CHRISTY GLASS JANETTE KAWACHI Outcome Measure Table 2a and 2b present the distribution of variables included in the models for each country and each year. For both years, our outcome measure is coded one if the respondent reported being unemployed for their current main activity or, additionally, if he/she reported not working but looking for work in a separate question. Respondents were coded zero if they reported either working or being on maternity leave. Independent Measures The relevant independent measures for our analysis are grouped roughly by 1) demographic information, 2) family responsibilities, 3) educational attainment and 4) most recent job characteristics. Among the demographic set of variables, the one pertinent for our analysis is sex, where females are coded as one and males as zero. The analytically significant variables associated with family responsibilities include marital status and the presence of young children in the household. Respondents are coded as one if they are married and zero if they are not. We tested for differential effects between various categories of marital status by including dummy variables for divorced, separated, widowed, single, cohabitating, etc. but found no significant inter-status differences. We therefore keep the more straightforward dichotomous categorization. To test the effects of child dependency, respondents that report having children less than the age of six in the household are coded as one and those that do not are coded as zero. We designate a specific age cutoff since children begin school around the age of six and become less of an encumbrance to mothers seeking employment. 20 To test for differential effects of young children and marital status on unemployment for men and women, we incorporate two interaction terms, one for females and young children and another for females and marital status. Education is coded as a set of dummy variables at five different levels in 1993 and four levels in This difference is due to the dissimilar ways in which respondents were asked about their educational credentials. Respondents with an elementary school education or less serve as the reference group for both years. Comparison categories in 1993 include university, academic secondary, academic technical and vocational training, while those for 2000 include university, academic secondary, and apprenticeship training. Finally, relevant measures regarding respondent s most recent job include employment type, full-time/part-time status and industry. For employment type, respondents are coded as one if they report that in their last/current job they were selfemployed, and coded as zero if they were employed otherwise. For job status, those that report part-time positions for their last/current job are coded as one, and zero if they report full-time employment. We divide industry into a set of 5 dummy variables 20 We use a dummy variable for young children rather than a continuous variable since the latter does not yield any significant effects in the models. This suggests that, although the presence of children has a significant impact on one s probability of unemployment, there are no added effects for each additional child.

13 WINNERS OR LOSERS OF REFORM? 121 that include industry, agriculture, service sector, trade and other. We leave out those who report employment in industry as the reference group in the models. 21 Control Measures Some of the measures included in our models are not of significant interest to our theoretical considerations but are incorporated for purposes of control. First, in the 2000 models, we include a dummy flag variable that acts as a control for those respondents that were included in the poor over-sample. Among the demographic variables, we include age and rural residence. Age is coded by years and also serves as a crude proxy for work experience. We test for non-linear effects by adding an agesquared term in all the models. Various theories argue that in times of unemployment crises, the young and elderly workers tend to be the first groups laid-off. Lastly, for rural residence, respondents are coded as one if they report living in a rural area, and coded as zero if they live in an urban area. RESULTS Overall Unemployment Not surprisingly, with the abolition of the state socialist policy of full employment, the working population has contracted significantly, and unemployment rates have risen sharply since 1988 across all transitional societies. Thus, as shown in Table 1, unemployment rates soared from 1988 to 1993 in Poland ( ) and in Hungary ( ). Table 1 Unemployment Rates in 1988, 1993 and 2000 Hungary Poland From 1993 to 2000, we observe a significant deterioration in the condition of Poland s labor market, as unemployment rates increased from an already high 13.1 percent, to 24.4 percent. Figures for Hungary remain stable, yet moderately high, at around 12 percent for both time periods. Thus, overall unemployment figures in Poland seem to indicate an increasingly deteriorating labor market while those for 21 Though the formulation of these employment variables are far from ideal, we hope to at least capture some idea of how experience in certain industries and occupations help to protect individuals from unemployment. This variable tells us whether or not the characteristics of the respondent s most recent employment position, as of the survey date, have helped protect them from unemployment. This was also the exact variable employed in Éva Fodor s 1997 article on gender and unemployment, which was based upon the same dataset.

14 122 CHRISTY GLASS JANETTE KAWACHI Hungary show potential signs that it may be inching closer to a state of relative labor market equilibrium. What about gender differences in unemployment rates? Results from our logistic regression analyses of the 1993 and 2000 data are given in Tables 3 4 and 5 6, respectively. For both years, Model 1 and 2 are identical across both countries while Model 3 is country-specific. Model 1 represents the full model, which includes all the variables discussed above, Model 2 is the reduced model, including only those variables that are shown to be significant in either Poland or Hungary, and Model 3 is our final model, including only those variables (or groups of variables) that are significant for each country. Overall, though there are a few caveats, it appears that in both Hungary and Poland, women in the labor market benefited significantly from the reforms during the initial stages of the transition. For Hungary in 1993, a statistically significant odds ratio of 0.72 (exp ( 0.330) = 0.72, p<0.05) for females shows that, all else equal, women were less than two-thirds as likely to become unemployed than men. In Poland, however, the same story only holds true for non-married women with no young children. Model 3 for Poland shows that the coefficient for females is negative and statistically significant (-0.461, p=0.07), indicating a sizeable advantage for women in the labor market. However, the positive and statistically significant interaction terms for married women and women with young children suggest that this advantage is only relevant for a very specific population, namely, single, childless females. These findings reveal a significant employment penalty for marriage and children among female workers, lending preliminary support to theories of retraditionalization and market/transitional discrimination. Though these earlier figures seem to bode well for female workers in postsocialist labor markets, this employment advantage seems to be gradually waning in Hungary and has become dramatically overturned in Poland by In Hungary, Model 3 of Table 5, shows a still negative, but slightly weaker (and statistically insignificant) coefficient for females, indicating greater variance in women s chances for unemployment, controlling for all other factors. However, this slight adjustment pales in comparison to the astonishing drop of women s employment conditions in Poland during the past seven years. Results for Poland in the final model of Table 4 show that, controlling for all other factors, women are approximately three times as likely as men to become unemployed. Furthermore, the 2000 models continue to show a strong and significant penalty for motherhood, though the additional marriage penalty for females has somewhat diminished and is no longer statistically significant. Again, these results lend strong and robust support to theories of retraditionalization and market/transitional discrimination in Poland s labor market. Revalued Resources The 1993 figures in Table 2a confirm that, on average, Polish women under socialism did indeed attain higher levels of education relative to Polish men. However, contrary to Fodor s initial assertion, women in Hungary do not demonstrate

15 WINNERS OR LOSERS OF REFORM? 123 Table 2a. Distribution of variables in models Hungary and Poland, 1993 Hungary Poland Total Women Men Total Women Men Unemployed Demographic Characteristics Age (mean) Rural Residence Female Household Responsibilities Married Spouse Employed Young Child in HH Educational Attainment University Academic Secondary Academic Technical Vocational Training Elementary or Less Most Recent Job Part-time Position Self-Employed Most Recent Industry Agriculture Industry Trade Services Other N

16 124 CHRISTY GLASS JANETTE KAWACHI Table 2b. Distribution of variables in models Hungary and Poland, 2000 national representative sample Hungary Poland Total Women Men Total Women Men Unemployed Demographic Characteristics Age (mean) Rural Residence Female Household Responsibilities Married Spouse Employed Young Child in HH Educational Attainment University Academic Secondary Apprenticeship Elementary or Less Most Recent Job Part-time Position Self-Employed Most Recent Industry Agriculture Industry Trade Services Other N

17 WINNERS OR LOSERS OF REFORM? 125 a clear educational advantage over men. Although women have a clear proportional advantage at the academic secondary level, most are actually concentrated at the elementary school level where over one-third are concentrated. In addition, the gender gap is much smaller at the university level where the proportions are about equal in Hungary and only slightly higher for women in Poland. Importantly, just as the returns to education begin to rise in these maturing capitalist labor markets, figures for 2000 show that men are increasingly pursuing higher levels of education and steadily weakening women s former advantage. Polish women continue to retain a slight advantage over men at higher levels of education, but figures in Hungary at these levels are nearly identical. Consistent with Fodor s theory, results from the logistic regression analyses for 1993, displayed in Table 3 and 4, show that on the whole, education has a strong and significantly negative effect on one s odds of unemployment. However, as mentioned earlier, the distributions imply that this advantage was not weighted towards women as Fodor had hypothesized. In fact, interaction terms between females and level of education (university and academic secondary) were tested to assess whether returns to education differed between the sexes, but were excluded from the models due to statistical insignificance. Additionally, it is important to note that in 1993, though the coefficients for academic secondary schooling are negative, they are not statistically significant in either Poland or Hungary across all three models. Given that women had their strongest advantage in this category, these results raise questions about the generalizability and accuracy of the revalued resources theory. More specifically, they suggest that, though education may have protected a significant proportion of women from unemployment, this security was both limited and highly variable. Results from the 2000 data presented in Table 5 and 6 show that over time, the effects of education on unemployment have increased and become fairly monotonic in both Hungary and Poland. The comparison of education effects in 1993 and 2000 suggest that education is indeed becoming a revalued resource in these increasingly marketized societies. However, it appears that full, rational revaluation has come at a point when women s educational advantage is rapidly declining. Moreover, though education may still play a significant role in protecting women in Hungary from unemployment in 2000, this certainly is not the case for Poland, where women suffer disproportionately from unemployment, yet continue to exceed men in educational attainment. This result indicates that, although women will benefit from competitive tendencies of the market that reward educational attainment, their benefits may be significantly offset by discriminatory practices against female employment. In addition to educational attainment, the revalued resources theory predicts that women will enjoy a significant advantage in the newly reformed labor market due to their dominance in service sector occupations during the socialist era. Table 7 illustrates this massive shift out of industry and into the services for both sexes in Poland and Hungary. Whereas in 1988, industry was clearly the dominant sector, accounting for nearly fifty percent of all jobs in both countries, by 2000 the service sector has become the leading industry, comprising the majority of occupations in both labor markets. In addition, industry figures by sex show that women have constituted the majority of service workers since 1988 and continue to do so in 2000, while the same is true of men in industry.

18 126 CHRISTY GLASS JANETTE KAWACHI Table 3. Logistic coefficients from the regression of unemployment on selected independent variables: 1993, Hungary, ages Independent Variables Full Reduced Country-Specific Model 1 Model 2 Model 3 Age (0.045) (0.045) (0.044) Age-sq (0.001) (0.001) (0.001) Rural Residence (0.123) Female * (0.242) (0.241) (0.134) Married (0.210) (0.210) Married*Female (0.278) (0.278) Spouse Employed *** *** *** (0.146) (0.146) (0.122) Young Child in HH (0.196) (0.195) Young Child*Fem ale (0.295) (0.295) Education (Ref = <=Elem) University *** *** *** (0.272) (0.268) (0.268) Academic Secondary (0.197) (0.194) (0.194) Academic Technical ** ** ** (0.191) (0.189) (0.188) Vocational Training * * * (0.147) (0.147) (0.146) Most Recent Job Part-time Position (0.348) Self-Employed ** ** ** (0.231) (0.229) (0.229) Most Recent Industry (Ref=Industry) Agriculture 0.474** 0.471** 0.475** (0.157) (0.155) (0.154) Service * * (0.230) (0.229) (0.229) Service*Female * * ** (0.361) (0.361) (0.360) Trade (0.181) (0.181) (0.180) Other * * * (1.016) (1.016) (1.015) Constant * * * (0.813) (0.811) (0.795) Chi-square *** *** *** Degrees of Freedom 20 Note: Numbers in parentheses are standard errors. Change in chi-square from Model 1 to Model 3 is not significant for all models. * p<0.05 **p<0.01 ***p<0.001

19 WINNERS OR LOSERS OF REFORM? 127 Table 4. Logistic coefficients from the regression of unemployment on selected independent variables: 1993, Poland, ages Independent Variables Full Reduced Country-Specific Model 1 Model 2 Model3 Age (0.050) (0.050) (0.050) Age-sq (0.001) (0.001) (0.001) Rural Residence (0.146) Female (0.269) (0.259) (0.259) Married * (0.239) (0.239) (0.220) Married*Female 0.732* 0.729* 0.710* (0.310) (0.309) (0.308) Spouse Employed (0.175) (0.175) Young Child in HH *** *** *** (0.261) (0.261) (0.260) Young Child*Female 1.003** 1.006** 0.957** (0.327) (0.327) (0.324) Education (Ref = <=Elem) University *** *** *** (0.267) (0.263) (0.262) Academic Secondary (0.249) (0.248) (0.247) Academic Technical ** ** ** (0.230) (0.228) (0.227) Vocational Training (0.192) (0.191) (0.191) Most Recent Job Part-time Position (0.356) Self-Employed *** *** *** (0.257) (0.257) (0.257) Most Recent Industry (Ref=Industry) Agriculture * ** ** (0.280) (0.273) (0.274) Service ** ** (0.271) (0.183) (0.183) Service*Female (0.340) Trade (0.207) (0.205) (0.205) Other (0.617) (0.617) (0.617) Constant (0.869) (0.863) (0.858) Chi-square *** *** *** Degrees of Freedom 20 Note: Numbers in parentheses are standard errors. Change in chi-square from Model 1 to Model 3 is not significant for all models. * p<0.05 **p<0.01 ***p<0.001

20 128 CHRISTY GLASS JANETTE KAWACHI Table 5. Logistic coefficients from the regression of unemployment on selected independent variables: 2000, Hungary, ages Independent Variables Full Reduced Country-Specific Model 1 Model 2 Model3 Poor Sample 0.649* 0.604* 0.624** (0.258) (0.254) (0.252) Age (0.095) Age-sq (0.001) Rural Residence 0.918*** 0.915*** 0.905*** (0.248) (0.246) (0.242) Female (0.499) (0.359) (0.257) Married 0.901* 0.798** 0.786** (0.355) (0.282) (0.278) Married*Female (0.543) Spouse Employed *** *** *** (0.313) (0.306) (0.305) Young Child in HH * * * (0.434) (0.401) (0.340) Young Child*Female (0.748) (0.737) Education (Ref = <=Elem) University * * * (0.780) (0.779) (0.773) Academic Secondary (0.332) (0.330) (0.324) Apprenticeship (0.290) (0.288) (0.285) Most Recent Job Part-time Position (0.254) (0.253) (0.251) Self-Employed ** ** ** (1.042) (1.041) (1.038) Most Recent Industry (Ref=Industry) Agriculture (0.416) (0.412) Service (0.346) (0.344) Service*Female (0.538) (0.534) Trade (0.460) (0.454) Other (0.541) (0.538) Constant *** *** (1.722) (0.391) (0.361) Chi-square *** *** *** Degrees of Freedom Note: Numbers in parentheses are standard errors. Change in chi-square from Model 1 to Model 3 is not significant for all models. * p<0.05 **p<0.01 ***p<0.001