Time to Work: A Comparative Analysis of Preferences for Working Hours HAYA STIER NOAH LEWIN-EPSTEIN Tel-Aviv University The article examines the preferences for working time among men and women in 22 countries. The main question was whether working hours reflect workers' preferences and tastes or whether they were a constraint imposed by the organization of the labor market and economic considerations. The study is based on the ISSP survey on Work Orientation conducted in 1997 and employs hierarchical linear modeling (HLM) to determine individual- and country-level factors affecting the preferences for working time. Thefindings suggest that a significant number of workers are still constrained by structural factors in their time allocation. Preferences for work were affected by both individual-level and country-level characteristics. Keywords: working hours; time preferences; country comparison; ISSP; hierarchical linear modeling In recent decades, a major transformation in working time took place in most industrialized countries. The general trend in most countries has been toward a reduction in the number of hours devoted to work (OECD, 1992). Nonetheless, there is considerable variation among countries in the number of weekly working hours and in the trend in working hours over time. For example, working hours in the United States, United Kingdom, and Sweden increased during the 1980s and early 1990s, whereas they continued to decline in Japan, Germany, and other North European countries (OECD, 1998, p. 156). Furthermore, although in most OECD countries the working week amounts to 40 hours on average, this number masks the growing Authors' Note: This article was prep are d for the Work Orientation Workshop at the annual meeting of the International Social Survey Program (ISSP), Lisbon, Portugal, May 5, 2000. The authors wish to thank Anna Ioffefor her tireless effort and her competent assistance in the analysis presented in the article. WORK AND OCCUPATIONS, Vol. 30 No. 3, August 2003 302-326 DOI: 10.1177/0730888403253897 O 2003 Sage Publications 302 Stier, Lewin-Epstein / PREFERENCES FOR WORKING HOURS 303 diversity in labor market activity. In particular, recent trends show a tendency toward polarization in hours of work: One segment of the population works very long hours and another segment has a short working day (Figart & Golden, 1998; Jacobs & Gerson, 1998, 2001; OECD, 1992, 1998). More workers are working irregular hours or at home, and many more can choose their hour arrangements (Horrell, Rubery, & Burchell, 1994). In light of these changes and the rising demand for nonstandard working hours, workers' preferences concerning the amount of time allocated to market work becomes an important issue, both for employers and for employees. These preferences for working time are at the center of the current study. In this article, we examine the preferences for working time among men and women in 22 countries. In the next section, we discuss the main theoretical approaches that deal with time devoted to work, followed by a review of country and gender differences in working time. We then use multilevel analysis to identify factors at the individual level and the country level that affect time preferences. INDIVIDUAL-LEVEL DIFFERENCES IN TIME DEVOTED TO WORK The decline in working hours and the changes in their dispersion had been attributed to structural changes in the characteristics of the working population and the organization of work, on one hand, and to workers' preference for working time, on the other hand. Part of the decline in working hours resulted from the growing participation of women in the labor force as many women, especially married women, are employed part-time. Although part-time employment has become a prominent feature of the workforce, especially in Europe, it is important to note that in many countries (e.g., Germany, Denmark, and Portugal) all or most of the decline in working hours is attributable to the decrease in hours of full-timers (OECD, 1998). Firm and government policies have created more flexible arrangements to promote the competitiveness of firms but also to facilitate workers' preferences. In many countries, flexible time arrangements have led to an increase of part-time work, although in others, a substantial part of the workforce is employed full-time and is engaged in (paid and especially unpaid) overtime. In the United States, for example, a sizeable segment of the workforce is working very long hours as a standard requirement in highly skilled, professional positions (Coleman &Pencavel, 1993a, 1993b; Figart& Golden, 1998; Jacobs & Gerson, 1998). Several theoretical approaches, representing two general lines of argument, attempt to explain variation and changes in working time. From a supply-side point of view, working hours are seen as reflecting workers' 304 WORK AND OCCUPATIONS / August 2003 preferences. Workers, based on their tastes, needs, or alternative activities, choose how much time to allocate to market work. In addressing what she viewed as growth in the number of hours Americans were devoting to work, Schor (1998), for example, argued that it was drivenby a culture of consumption. Competitive materialism constantly requires additional financial resources, which many Americans try to meet by increasing the number of hours they work for pay. Hochschild (1997), on the other hand, focused her explanation on the conflicting time demands produced by the workplace and the family and the changing perceptions of the rewards they offer. She found that many workers, especially women, experience time pressures and insensitive demands from their families and, as a response, seek refuge at work, where they find support and recognition from coworkers and enhanced career opportunities. Thus, the observed long working hours experienced mainly by the highly educated, white-collar professionals may signify their preferences for time allocation to work activities. By the same token, although presenting an entirely different context, Hakim (1997, 1999) contended that the growth in part-time employment reflects lower commitment to work on the part of some segments of the population. Accordingly, she argues that the polarization in working hours, which is evident in the labor force as a whole, is also descriptive of the changes taking place in the female workforce. One segment of women is career-oriented and thus interested in allocating more time to market activity, whereas the other segment is less interested in market work and is more concerned with activities outside of the market (i.e. domestic or leisure activities). Thus, part-time workers can be viewed as less committed to market work because they have alternative interests in life. In most cases, these will be married women, who are less dependent on their own salary. This approach, which views actual working hours as the result of employee preferences, was criticized by some researchers who failed to find support for the "preference" argument. For example, Fagan and Rubery (1996) found that a large proportion of women who work part-time would actually prefer a longer working day. This finding is supported by additional evidence that a substantial portion of part-time employment is involuntary (Cohen, Stier, & Nadiv, 2000; Jacobs & Gerson, 2000; OECD, 1995; Tilly, 1991; Walwei, 1998). Similarly, Maume and Bellas (2001) found no support for the claim that work is more rewarding than home and they concluded that "personal choice is unrelated to work schedules" (p. 1148). At the other extreme are theories that focus on the "demand side," mainly the organization of work by employers. By structuring the workday, employers limit workers' choices regarding the allocation of time to market work. Schor (1991) argued that the increase in working time reflects employers' Stier, Lewin-Epstein / PREFERENCES FOR WORKING HOURS 305 interests in increasing their profits. Their interests are buttressed by an organizational culture that uses job insecurity to impose a long working day. In addition, unions, which in the past were able to influence work schedules, are weakening and are thus less able to affect the length of the working day (Maume & Bellas, 2001). As a consequence, working time is imposed on workers and thus many Americans feel "overworked" (Jacobs & Gerson, 2000). Maume and Bellas (2001), indeed, find support for this argument, which is also reflected in the finding that most workers would prefer a reduction in their working hours (Jacobs & Gerson, 2001). Moreover, Jacobs and Gerson (2001) emphasized that the bifurcation of the labor force is reflected also in bifurcated preferences: Those who work longer hours prefer a reduction in their workload, whereas those who allocate relatively few hours to market activity are more likely to want more hours of work (see also Bell & Freeman, 1995). Thus, ratherthanreflectingworkers'preferencesformarket activity, the actual number of hours worked indicates the nature of employment arrangements in specific segments of the labor market. The two approaches yield opposing predictions, especially regarding the relationship between actual hours of work on time preferences. From a supply-side vantage point, actual working hours are expected to reflect preferences, although from a demand-side point of view, hours of work are expected to be negatively correlated to time preferences. COUNTRY DIFFERENCES IN TIME PREFERENCES Attitudes toward work are formed in specific social contexts. Even among the fairly homogeneous group of industrialized nations, preferences for work differ across countries. Hence, we view individuals' preferences and decisions as embedded in country-level institutional and cultural peculiarities. Consequently, countries may differ both in the total levels of actual employment and preferences and also in the way in which personal attributes affect preferences and decisions at the individual level. In some countries, preferences for part-time work are growing, not only among persons out of the laborforce, but also among those who work full-time (Hakim, 1997). In other countries, a sizeable portion of the adult population would prefer additional working hours. A recent survey of European Community members revealed that one quarter of all full-time employees would accept a reduction in earnings to achieve shorter working hours (an additional one third of both men and women said they would accept shorter hours if their pay would not decrease). The numbers were highest in Denmark and the Netherlands. The 306 WORK AND OCCUPATIONS / August 2003 preferences for shorter hours of work were more prevalent among women who work full-time than among men similarly employed (Hakim, 1997). In contrast to the decline in working hours in North and West European countries and the normative preference of workers to spend less time at work, Schor (1991,1998) found that Americans were working more hours, and she argued that they were motivated to do so to keep pace with their desire for material consumption. In line with the argument of the overworked American, Bell and Freeman (1995) found that as many as one third of all U.S. employees would like to increase their hours of work and only a minority preferred a shorter working day. Similar findings are reported from Canada (Drolet & Morissette, 1997). Jacobs and Gerson (2001), however, have criticized the studies that found Americans to be working more than in the past on methodological grounds, and they reported that only 17% of U.S. workers prefer to increase their workload compared to almost half who would like to reduce their working hours (Jacobs & Gerson, 2000). In reporting the results of a comprehensive cross-national study of women's part-time employment, Blossfeld (1997) noted that after the breakdown of socialism in Central and Eastern Europe there was a large reduction of employment of men and women. This was due largely to the declining economic situation and the disintegration of old employment arrangements. Nonetheless, the employment patterns of women from post-socialist states remained quite distinct from those prevalent in Western Europe. Blossfeld and Hakim (1997) concluded that "socialist policies over several decades obviously have changed the attitudes of women toward being a full-time homemaker" (Blossfeld & Hakim, 1997, p. 318). A common difficulty with such cross-national comparisons is that they often generalize information over entire working populations, although the composition of jobs and employment situations may differ substantially from one country to another. Hence, additional insight can be gleaned from more controlled comparisons thatfocus on workers in similarjobs. Perlow (2001), for example, found substantial differences in the extent of working hours among software engineers performing essentially the same work in China, India, and Hungary. She concluded that the country-specific organization of work was the most important factor accounting for these differences. Such cross-national differences are not limited to actual working hours and are evident in employees' work preferences as well. In a study of financial professionals from the United States, Great Britain, and Hong Kong who held similarjobs in one division of a multinational company, Wharton and Blair-Loy (2002) found that Hong Kong employees were significantly more likely to express a preference for part-time employment. After rejecting several alternative explanations, the authors adopt a cultural explanation, concluding that Stier, Lewin-Epstein / PREFERENCES FOR WORKING HOURS 307 Hong Kong employees "are more likely to resent work as an intrusion onto their lives outside of work" (Wharton & Blair-Loy, 2002, p. 55). In discussing the fact that there are large cross-national differences in the amount of time devoted to work, Bell and Freeman (1995) pointed out that countries differ significantly in the general level of preferences for working time. This is because they differ in their socioeconomic characteristics and with regard to culture and institutional arrangements that affect not only the patterns of employment but attitudes toward work as well. Among the most important characteristics at the macro level is the extent to which members of the society are dependent on the market for subsistence. It is reasonable to hypothesize that, in countries with extensive decommodification, people would be free to choose to allocate less of their time to work activity because their economic well-being is less dependent on their market wage. High rates of income tax, high wages, and low wage inequality may affect workers' preferences in a similar direction (Bell & Freeman, 1995). On the other hand, we might expect that a relatively low standard of living (or low levels of GNP per capita), high unemployment rates, and low levels of decommodification will increase the demand for long working hours. Thus, preference for working hours will differ, on average, in countries that vary along these dimensions. GENDER DIFFERENCES IN TIME PREFERENCES Although women have been part of the cash economy in industrialized countries for decades and their share of the labor force increased substantially in the past half century, their participation rates remain lower than those of men and employed women work fewer hours than men on average (OECD, 1998; Spain & Bianchi, 1996). In most industrialized countries, women's labor force participation increased substantially, whereas men's participation declined over time. Yet the ratio of women to men participation rate ranges between 0.52 (in Italy) to 0.87 (in Sweden) (Spain & Bianchi, 1996, p. 101). Similarly, men are working longer hours than women in all countries (OECD, 1998). In the United States, for example, where part-time employment is relatively uncommon, women's paid work averages fewer than 40 hours a week compared to 46 hours for men (Gerson & Jacobs, 2000, p. 83). Time differences between men and women are substantially higher in countries with a high proportion of women working part-time such as Sweden, Germany, the United Kingdom, and The Netherlands (Smith, Fagan, & Rubery, 1998). Men are more likely than women, in all countries, to work long hours, whereas women are more likely to work fewer than 20 hours a week (OECD, 1998). 308 WORK AND OCCUPATIONS / August 2003 During the past 30 years, part-time employment has grown substantially in most European countries (although not in all), alongside the rise in female laborforceparticipation(Smithetal. 1998). Indeed, part-time work is mainly women's work, attributed largely to women's inability to allocate more time to market work because of their domestic responsibilities (Duffy & Pupo 1992; Fagan& O'reilly, 1998; Hochschild, 1997) orto their lack of interest in market activity (Hakim 1997, 1999; Pfau-Effmger, 1998). In the normative sphere, it is still the case that men are supposed to consider work as their primary time priority, whereas women are expected to make the family their first priority. A recent report based on a large-scale survey of working conditions in Europe concluded that gender roles—women's position as secondary earners and holders of prime responsibility for family-related activities—are the major explanatory factors of women's part-time work (Tijdens, 2002). Consequently, family and work responsibilities are seen as a source of conflict for women but not for men (Epstein & Kalleberg, 2000). These normative expectations have significant implications forunder-standing gender differences in work-time preferences. Even among professional workers, women are more likely than men to express a preference for part-time work and the gender difference is especially large among married workers (Wharton & Blair-Loy, 2002). Both the normative expectations and the actual employment patterns of men and women require that their preferences regarding hours of work be discussed and evaluated separately. First, the preference for adding or reducing working hours takes on a different meaning for men as compared to women because they work longer hours on average. Second, the factors that affect work time preferences may be quite different for the two gender groups. For example, women might be expected to prefer a reduction in hours of employment to cope with household responsibilities and child care. Likewise, women, especially if married, might be less reluctant than men to give up part of their salaries in exchange for a shorter working day because the earnings of most women provide a smaller contribution to the household economy. Although gender differences in the likelihood of employment and amount of time devoted to market work are found in most countries, women's employment is more likely to be affected by the country's specific cultural and institutional context. To date, however, only a few studies have attempted to systematically address the social and cultural embeddedness of women's (and men's) employment decisions (for recent exceptions, see Charles, Buchmann, Halebsky, Powers, & Smith, 2001; Diprete & McManus, 2000). The present study joins this line of research and hopes to expand our understanding of the individual- and macro-level factors that are associated with work preferences. Stier, Lewin-Epstein / PREFERENCES FOR WORKING HOURS 309 DATA AND MEASUREMENTS The study is based on the International Social Survey Program (ISSP) Survey on Work Orientation conducted in 1997 in 27 countries. A unified data file was prepared by JD Systems, Madrid, and the Zentralarchiv fuer Empirische Sozialforschung (ISSP, 1999), with a total of about 30,000 cases (adults age 18 to 65). To this survey we added information at the country level that was collected from World Development Indicators (World Bank, 1998) and International Labour Statistics (ILO, 1999). The analysis focuses on two aspects of preferences for working time. First, we refer to the general question that was posed to all respondents: "Would you like to spend more time in a paid job?" Second, we analyze a question posed to workers only: "Which of the following choices would you prefer: 1. work longer hours and earn more money; 2. work the same number of hours and earn the same money; 3. work fewer hours and earn less money." The two questions pertain to different aspects of the preferences for working time. The former is more general in nature and makes no direct reference to the economic consequences of one's choice to increase or reduce working hours. It taps a general orientation to work. The latter establishes a direct relationship between working hours and remuneration and is directed to respondents who already hold a paid job. For the purposes of the present study, we analyze a pooled file that includes data from most ISSP countries. Included in our analyses are Canada, the United States, United Kingdom, Germany (East and West), France, Switzerland, The Netherlands, Sweden, Norway, Denmark, Italy, Spain, Portugal, Israel, New Zealand, Japan, Russia, the Czech Republic, Poland, Bulgaria, Hungary, and Slovenia.1 All analyses are conducted separately for men and women. Our general expectation is that women will show a weaker interest than men do in increasing their market time and that their time preferences are affected by different factors due to their distinct market prospects and domestic responsibilities. We introduce two types of explanatory variables: individual-level variables and macro-level country characteristics. The variables at the individual level include marital status (1 = married, 0 = otherwise), age (ranges between 18 and 65), education (an eight-category ordinal measure of the highest education attained), employment status (full-time, part-time, unemployed, not in the labor force), and a measure of family income. To achieve comparability and overcome the differences in the unit of measurement, we standardized the income using a measure of relative family income. Accordingly, we calculated the relative distance of the respondent's family income from the maximum income level reported in the country-specific sample (for a discussion 310 WORK AND OCCUPATIONS / August 2003 of this procedure, see Gornick & Jacobs, 1996).2 For the analyses that pertain to the working population, we also include measures of satisfaction with present job (an answer to the question, "How satisfied are you in your job?" Answers range from 1 {high job satisfaction) to 7 {low job satisfaction) and concern of losing the job (answers to the question, "To what extent do you worry about losing your job?" ranging from 1 {high insecurity) to 4 {low insecurity). The effect of employment status on preferences for working hours is not straightforward. Based on Schor (1991) and Jacobs and Gerson (2001), full-time workers (who may feel overworked) are expected to have the lowest preferences for an increase in working hours. The unemployed are expected to have the highest preferences. It is not clear to what extent part-time workers would prefer longer or shorter hours of work. According to Jacobs and Gerson, they will show higher preferences for working hours than full-time workers. According to Hakim's contention, however, part-time workers, especially women, are less committed to market work and thus will be less likely to prefer an increase in market time. Marital status is expected to affect differently the preferences of men and women. Being married is expected to increase market-time preferences for men and to affect negatively the preferences of women due to the differences in their domestic responsibilities. We expect education to be negatively correlated with preferences for more work (that is, the more educated are expected to prefer to invest less of their time in market activity). This is mainly because their jobs more often entail higher levels of earnings and security (see also Drolet & Morissette, 1997). Similarly, we expect a negative correlation between family income and preferences for working hours. This effect should be particularly strong in our second measure, which pertains to preferences for hours in exchange for money. Those with lower levels of income may have high preferences for leisure but cannot afford cutting their working hours and thus their salaries. Older respondents are expected to prefer a reduction in working time. Workers with higher job satisfaction and those with higher levels of job insecurity are expected to prefer an increase in working hours. At the macro level, we introduce measures of economic conditions and inequality (GNP per capita, Gini coefficient, inflation rate) for which data were taken from the World Development Indicators (World Bank, 1998); characteristics of the labor force taken from the International Labor Statistics (ILO, 1999) (percentage unemployed, percentage of women in the labor force, the gender wage gap3); country's average level of education; the level of decommodification (level of expenditures for public assistance as a percentage of total public expenditures); and a measure of the country's level of Stier, Lewin-Epstein / PREFERENCES FOR WORKING HOURS 311 orientation to work. The latter variable was derived from the ISSP file, based on the country's mean response to the statement, "work is a person's most important activity." Responses ranged from 1 (strongly agree) to 5 (strongly disagree). We expect a higher level of preference for working time in countries with a less stable economy and lower rate of economic development. We also expect a greater preference for work in countries with higher rates of inequality, as dependency on work to maintain an acceptable standard of living is high (Bell & Freeman, 1995, 2000). The level of decommodification should be inversely related to preference of working hours. It is not clear how female labor force participation rates will affect the preference for work. On one hand, higher rates of female participation increase the economic well-being of families and may thus reduce the preference for work. On the other hand, high female participation rates indicate greater similarity of women and men's roles, in which case women's preferences for work would be higher. The gender wage gap is expected to affect women's preferences for work—in countries where women are treated more equally to men, women will prefer to invest more of their time in market activity. This indicator, however, is not expected to affect men's preferences for market work. Aside from the effects of country characteristics on the level of preferences for work, we also expect certain macro-level factors to affect the relationship between education (at the individual level) and time preferences within countries. We hypothesize that as the level of inequality increases, education will have a stronger effect on the preferences for working hours. That is, although in general we expect a negative relationship between the preferences for working time and education, the magnitude of this relationship will increase when inequality is high. One reason for this is that the rates of return for education are high and highly educated people may more willingly substitute work for leisure. A second reason is that, in situations of high inequality, the weaker segments of the population may feel greater deprivation and will view additional work as a means of attaining a higher standard of living. We also expect that, in societies with a strong normative orientation to work, the relationship between education (at the individual level) and time preferences will be weaker. METHOD OF ANALYSIS Because we are interested in the distribution of preferences within countries as well as among them, we employ hierarchical linear modeling (HLM), in which the dependent variable is the preferences for working time and both individual and structural variables serve as independent variables. Using HLM, we are able to model the two components simultaneously and to 312 WORK AND OCCUPATIONS / August 2003 differentiate between the effects of individual characteristics and country-level characteristics on time preferences. This method allows us to identify the factors that affect country differences intime preference for work and also to test whether important correlates of time preferences such as employment status, education, or family status have a similar effect on time preferences across countries. The two-level model can be represented by a set of equations. The first is a within-country equation that models time preference as a function of the independent variables described earlier. The general form of this equation is illustrated by the following example: (Time preferences!. = ß0, + ßj.(Education Y. + e„ (1) Another set of equations models the between-country variation: ß«, =1«, + Yoi(inflation rate),+v„, (2) ßi, = Yio + Yn(GINI Coefficient) + v,, (3) Here, the ß coefficients derived from Equation 1 constitute the dependent variables in Equations 2 and 3. In this example, the equations respectively model the variation in the average level of time preference across countries (i.e., effects on the intercept) by the inflation rate and country differences in the effect of education on time preferences by the level of inequality within the country. The other variables that we include in the model are interpreted in a similar way. FINDINGS PREFERENCES FOR WORKING TIME Figures la through 2b present the distribution of the two preference questions across all countries separately for men and women. Turning first to the general preferences for working time (Figure la for men and lb for women), we see substantial country differences. A clear preference for increasing time spent on work is evident in former socialist countries (especially Russia and Bulgaria), and Mediterranean countries such as Spain and Israel. The proportion preferring to devote more time to work activity is lowest in developed countries such as Sweden, Japan, the United Kingdom, and France. As a general pattern, we find that in the majority of countries, the proportion of those who would prefer to decrease their market time is higher than the proportion of those who want an increase it. Stier, Lewin-Epstein / PREFERENCES FOR WORKING HOURS 313 Figure 1a: Preferences for Working Time, Males Age 18 to 65 Figure 1b: Preferences for Working Time, Females Age 18 to 65 314 WORK AND OCCUPATIONS / August 2003 Denmark Sweden Norway New Zealand Great Britain Canada USA Switzerland France The Netherlands Germany-West Hungary Russia Bulgaria Poland Slovenia Czech Republic Germany-East Japan Spain Portugal Israel Italy ■work longer hours and earn more money pwork fewer hours and earn less money | Figure 2a: Preferences for Working Hours in Exchange for Money, Males Age 18 to 65 Denmark Sweden Norway New Zealand Great Britain Canada USA Switzerland France The Netherlands Germany-West Hungary Russia Bulgaria Poland Slovenia Czech Republic Germ any-East Japan Spain Portugal Israel Italy -20 0 20 40 60 80 ■work longer hours and earn more money Pwork fewer hours and earn less money | Figure 2b: Preferences for Working Hours in Exchange for Money, Females Age 18 to 65 Stier, Lewin-Epstein / PREFERENCES FOR WORKING HOURS 315 Women have a higher preference for market time than men in several countries (Italy, Portugal, Spain, New Zealand, the United Kingdom, Switzerland, France, The Netherlands, and West Germany) and a lower preference in market time in the postcommunist countries, Scandinavia, Canada, and the United States. One general conclusion is that women more than men are dissatisfied with their current time-allocation situation. It is important to note that this question was addressed to those who participate as well as those who do not participate in the laborforce. Many more women than men are out of the labor force or work part-time, and would like to increase their involvement in the market. It is not clear, however, whether among those who work, women still have higher preferences than men for increasing working hours. The distribution of responses discussed above can be viewed as representing a general orientation to work activity in the society. The next question addresses more specifically the relationship of work and leisure in view of income constraints. The question was posed to the working population and underscored the trade-off between income and leisure. The distribution of preferences for more hours (in exchange for more money) and for fewer hours (for less money) is presented in Figures 2a and 2b, for men and women, respectively. Among men and women alike, there are two clearly defined groups of countries: those in which workers want to increase their workload (to earn more) and those in which workers prefer a decrease in working hours even at the expense of lower earnings. In the former group, most noticeable are the postcommunist countries in which almost all men and women prefer longer hours of work, accompanied with higher income. Only a minority in these countries would prefer to keep the current situation as it is or to reduce their workload for a reduction in money. A similar picture can be seen among the Mediterranean countries. Quite surprisingly, English-speaking countries who generally enjoy the highest standard of living exhibit the same pattern as in postcommunist countries; namely, a substantial proportion of workers would prefer to increase the time devoted to work. We will address this issue in the concluding section. At the other extreme, several countries exhibit a pattern whereby workers prefer a reduction in their workload and are willing to incur lower earnings. This is clearly the case in the Scandinavian countries and, for men, also in Switzerland. The preference for a reduced workload in these countries is more pronounced among men than among women, probably reflecting men's higher workload to begin with. On balance, these figures suggest that in countries with higher levels of decommodification (Esping-Andersen, 1999), workers are willing to settle on lower earnings in exchange for a lighter workload. A better understanding of these patterns would require 316 WORK AND OCCUPATIONS / August 2003 considerations of other country- and individual-level characteristics that might affect the divergent patterns we observed. INDIVIDUAL AND COUNTRY DETERMINANTS OF WORK PREFERENCES The second stage of the analysis focuses on the individual- and country-level factors that affect the preferences for work. The analyses are based on HLM models for ordinal dependent variables. The ordered logit model assumes an ordinal outcome variable with categories that can be ranked, but the distance between them cannot be determined (Long, 1997). The model calculates the probability that an estimated linear function, plus an error, is within a range of cut-off points estimated for that outcome, and can be expressed as follows: Pr(outcomej. =/) = Pr(Aj_1 < ßlXl, +ß2*2j. + ...+$kxk] + Vj