Printer Friendly

Gendering insiders and outsiders: labour market status and preferences for job security.

1. Introduction

This paper analyzes the relationship between labour market status and preferences for job security. It starts from the seminal work of Rueda (2005, 2007) who has shown that an individual's labour market status has strong effects on her/his demand for welfare policies. This is particularly the case for attitudes concerning active labour market policies and job security regulations. Most interestingly, Rueda observes a cleavage within the group of dependent employees, between insiders, i.e. dependent employees having a full-time and permanent job, and outsiders, i.e. unemployed people or dependent employees working part-time or only having a temporary contract. Insiders support policies that strengthen their position in the labour market, most notably job security regulations, and disapprove of policies that strengthen their main competitors for jobs, e.g. active labour market policies. In contrast, outsiders support policies that increase their chances of finding employment, e.g. active labour market policies, while they disapprove of policies that reduce labour market turnover and therefore the number of open positions, e.g. job security regulations. The remaining labour market participants - upscale and self-employed respondents - have not much to gain from these policies, but can expect to pay for them. As a result, they tend to oppose job security regulations and, to a lesser extent, active labour market policies.

Rueda's (2005, 2007) theoretical account is very convincing. However, in his re-examination of the link between labour market status and preferences for job security, Emmenegger (2009a) does not find across-the-board empirical support for Rueda's claims. Most notably, the difference between the preferences for job security of labour market insiders and labour market outsiders is smaller than expected and there is considerable within-group variation in the case of labour market outsiders. While part-time and temporarily employed respondents consider job security considerably somewhat less important than labour market insiders--a finding in accordance with the expectations formulated by Rueda (2005, 2007)--unemployed respondents, the archetype of outsiders, consider job security more important than any other group examined, including labour market insiders. These findings are clearly at odds with the insider/outsider theory of employment and unemployment and call for further research.

Emmenegger (2009a: 135-137) discusses three possible explanations for how this result can be explained. Firstly, the classification of labour market participants into insiders, outsiders, self-employed and upscales might be too crude and too static (Hausermann and Schwander 2009); secondly, the exclusive focus on the labour market neglects household relationships and the increasing feminization of labour markets (Pierson 2001); and thirdly, the long-term interest of eventually obtaining an insider position might induce outsiders to support job security regulations despite short-term disadvantages (Tsakalotos 2004).

This paper adds to this on-going debate by examining one of these three potential explanations. In the following, we 'gender' the insider/outsider theory of employment and unemployment by, on the one hand, theoretically conceptualising and empirically testing the relationship between labour market status and gender and, on the other hand, by taking into account the relationship between labour market participation and the household situation (e.g. number of persons in household, presence of small children and division of work within the household).

We hypothesize that the insider/outsider theory of employment and unemployment, as formulated in its original version by Lindbeck and Snower (1988), suffers from a gender bias. Its focus is on fulltime male labour market participation in highly unionized manufacturing sectors and unemployed workers within these sectors. It neither considers the possibility of family-related labour market transitions (e.g. due to pregnancy or in order to tend to family needs), the division of household labour, nor different forms of non-standard employment, most notably part-time and temporary employment. (1) However, if we are to understand the preferences for job security regulations, we have to acknowledge that in the 21st century almost 50 percent of the labour market participants are female and that the share of nonstandard employment has been constantly increasing in recent decades (see Figure 1). In the 21st century, Western Europe's labour markets are largely de-industrialized, feminized and characterized by the presence of numerous nonstandard employment relationships (Pierson 2001). Full-time male labour market participants in the manufacturing sector are all but a small minority group within the labour force. In this paper, we attempt to theoretically and empirically capture the implications of these developments for the insider/outsider theory of employment and unemployment.

[FIGURE 1 OMITTED]

This paper is structured as follows. We start with a discussion of the insider/outsider theory of employment and unemployment. Subsequently, we discuss, on the one hand, the relationship between gender and labour market status and, on the other hand, the role of the household situation for preferences for job security. We develop two hypotheses: Firstly, we hypothesize that gender interacts with labour market status. More concretely, we expect female insiders to be more interested in job security than male insiders while we expect female outsiders to be less interested in job security than male outsiders. As we argue, these differences are the result of the fact that, on average, women experience more labour market transitions than men. Secondly, we hypothesize that the household situation has a profound impact on attitudes concerning the importance of job security. More concretely, we expect labour market participants not living alone, with small children at home and labour market participants who are the main contributors to the household income to be more interested in job security than labour market participants living in single house holds, without small children and who are not the main contributors to the household income.

After discussing the employed data and the operationalization of the variables, we turn to the empirical evidence. We show that, contrary to our expectations, gender does not--neither in interaction with labour market status nor as an independent effect--affect individuals' preferences for job security. In contrast, we find significant effects of variables capturing the household situation on preferences for job security. Most notably, we observe that individuals living together with their partners and main contributors to the household income consider job security to be particularly important, while the presence of small children in the household does not affect preferences for job security. Analyzing these relationships in more detail, we observe that living together only affects preferences for job security in case of main contributors to the household income. Simultaneously, an individuals' role in contributing to the household income only affects preferences for job security in case of respondents living together with their partners.

Finally, turning to the macro level, we explore the cross-national differences with regard to, on the one hand, differences between respondents living together and respondents living alone and, on the other hand, differences between main contributors to the household income and equal contributors/secondary wage earners. Contrary to our expectations, we find no robust relationships between these cross-national differences and policies that are to help reconciling work and family life (childcare and family benefits) as well as policies that are to protect labour market participants' welfare (unemployment benefits, social expenditure, employment protection legislation and active labour market policies). A final section concludes.

2. Insiders, Outsiders and Preferences for Job Security

The insider/outsider theory of employment and unemployment has been developed in order to answer an economic puzzle in the 1980s. While the sharp rise in unemployment in the aftermath of the first oil price shock had been credited to the combination of high inflation and economic stagnation, the persistence of high unemployment confronted economists with a problem. A decline in productivity growth due to the rapid de-industrialization had been observed, but wage setters did not adjust to the new reality of lower productivity growth (Blanchard 2006: 19). Seeking a solution to this puzzle, Lindbeck and Snower (1988) turn to the behaviour of labour market insiders because collective bargaining typically takes place between a union representing the employed workers (insiders) and firms. The unemployed lack representation in such negotiations. Consequently, Lindbeck and Snower (1988) call them outsiders.

Lindbeck and Snower (1988) argue that insiders focus only on their own employment and demand wage increases that are beyond the interests of outsiders. Thereby, they keep unemployment above the natural rate. Firms have the option of replacing their employees with unemployed workers--or can use the threat of hiring currently unemployed workers to negotiate lower wages with trade unions. However, this threat's credibility is a function of labour turnover costs, i.e. the cost of replacing employed workers. These costs are a function of several factors, but the most important sources of labour turnover costs are job security regulations (Lindbeck and Snower 1988: 3).

Labour market insiders gain from low labour turnover and high labour turnover costs respectively. It is crucial that both are strongly influenced by job security regulations, i.e. labour turnover is low and the cost of labour turnover is high when job security regulations are very strict. As a consequence, a distinction between insiders and outsiders is necessary because these two groups have opposing preferences with regard to job security regulations. While job security regulations increase the insiders' market power vis-a-vis their firm, they also constitute a barrier to entry for labour market outsiders (Saint-Paul 2002; Emmenegger 2009a).

The group of insiders does not include all those currently working. There are also other groups such as the self-employed and the so-called 'upscales', who are employees in privileged positions in the labour market. The former are not subject to collective bargaining, while the 'upscales' occasionally are. However, due to their human and social capital the 'upscales' do not need to be afraid of unemployment. Consequently, they cannot be expected to support social policy interventions. Quite the contrary, due to their privileged position in the labour market, they have to expect to carry a disproportionate part of the costs of social policy interventions. As far as the self-employed are concerned, protection against dismissals is for obvious reasons useless.

Similarly, workers with temporary contracts and workers involuntarily confined to part-time jobs cannot be classified as insiders. The majority of these jobs pay poorly and are concentrated in low-skills sectors (Rueda 2005: 63). Consequently, Rueda classifies part-time and temporarily employed as outsiders. Job security regulations are of little use to temporary workers. They cannot expect to work for the same enterprise for the rest of their lives. They have a fixed-term contract and as soon as this contract expires they will either get a new contract or leave, independent of job security regulations. Although they may hope to get a permanent contract one day, the present value of job security regulations, at least in the short term, is very low. Part-time jobs are often less protected by job security regulations than full-time jobs (Samek Lodovici 2000: 38). Per definition, involuntary part-time work implies that these workers would like to switch to a full-time job if possible. However, according to the insider/outsider theory of employment and unemployment, it is exactly the existence of restrictive job security regulations that decreases their chance to do so. As a result, they cannot be expected to support restrictive job security regulations.

The situation is more complicated in the case of currently employed workers with a recent history of unemployment. Due to their recent transition back into employment, they might appreciate flexible labour markets. Moreover, their history of unemployment can be understood as an indicator of a comparatively weak labour market attachment. As a result, their risk of becoming unemployment again might be disproportionally high. However, now that they are employed again, they benefit from restrictive job security regulations. As argued by Esping-Andersen (1999: 136), job security regulations provide a shelter for those already employed, even the low-skilled ones and those with a weak labour market attachment. Moreover, their higher risk of becoming unemployed increases the short-term value of job security regulations (Emmenegger 2009c).

It is important to note that the insider/outsider theory of employment and unemployment does not claim that labour market outsiders support labour market deregulation. Nonetheless, significant differences between labour market insiders and labour market outsiders in terms of preferences for job security regulations can be expected. This is due to the fact that the difference between labour market insiders and labour market outsiders is a relative one. Labour market insiders have a very clear and rational interest in more restrictive job security regulations (they want to keep their current jobs). In contrast, the situation is less clear for labour market outsiders. Thus, we need not assume that outsiders openly support the deregulation of job security regulations in order to observe relative differences between labour market insiders and labour market outsiders. It is sufficient to assume that labour market insiders have a stronger (and more visible) interest in job security regulations.

3. Gendering Insiders and Outsiders

Job security regulations have been identified as an important form of social protection by leading welfare state scholars such as Esping-Andersen (1999), Bonoli (2003) and Kaufmann (2003). As all forms of social protection, job security regulations have distributional effects. According to the insider/outsider theory of employment and unemployment, job security regulations have negative distributional consequences for labour market outsiders. This, however, is not the only distributional consequence. Especially Esping-Andersen (1996, 1999) has forcefully argued that job security regulations also affect the relative position of men and women in the labour market. In the following, we will gender the insider/outsider theory of employment and unemployment by taking into account the additional life-course risks women face and their more precarious labour market positions (Svallfors 1997) as well as by integrating the household situation in order to go beyond the mere distinction between different labour market states.

Women and men are different. Women are more likely to interrupt their working life in order to tend their family's needs (Estevez-Abe 2005: 190). Some of these interruptions are due to biology: women give birth to children while men do not. However, there are also other reasons. Women tend to undertake a bigger part of the household labour because a considerable number of citizens of Western societies consider this to be the 'natural' division of labour. As a result, women tend to spend more time away from paid work than men. Put differently, on average, women experience more labour market transitions than men. The important point here is that the higher number of labour market transitions increases the importance of institutions regulating these transitions, e.g. job security regulations.

Although not all women experience more career breaks than men, employers assume that this is the case. This phenomenon is often referred to as 'statistical discrimination'. Independent of women's actual career intentions, employers tend to adjust their employment policies in order to anticipate that women are on average more likely than men to quit their jobs. Statistical discrimination has an effect on all women independent of their skills. There is no personal characteristic that makes women insulated from statistical discrimination. Rather, it is the mere fact that they are women that weakens their labour market position. As a result, all women are treated as if they experience more labour market transitions.

We hypothesize that gender can be expected to exacerbate the differences between insiders and outsiders. There are three reasons for this. Firstly, due to additional life-course risks, women experience more transitions from unemployment to employment and vice versa. As a result, institutions mediating these transitions, such as job security regulations, become more important. Secondly, the transition from unemployment to employment is more difficult for women than for men due to statistical discrimination. If two equally qualified candidates apply for the same job, employers will pick the male candidate because they can expect male workers to remain with the firm for a longer period than female workers. As a consequence, female outsiders are particularly likely to suffer from the negative labour market effects of job security regulations. Thirdly, job security regulations are particularly beneficial for labour market participants in a weak labour market position (Emmenegger 2009c). As noted above, female insiders are on average in a weaker labour market position than male insiders. As a result, we expect female insiders to be more supportive of job security regulations than male insiders.

Thus, we expect gender to exacerbate insider/outsider divisions: On the one hand, female outsiders can be expected to be even more critical of labour market institutions that complicate their return to an insider position. They experience more often the transition from unemployment to employment than men. Moreover, due to statistical discrimination, this transition is more difficult. On the other hand, the more precarious labour market positions of women makes job security regulations particularly valuable to female insiders as they face a higher risk of unemployment than their male counterparts. As they are exposed to a higher risk of joblessness, they will be more supportive of any institution that protects their currently favourable position. Figure 2 displays the relationship graphically.

[FIGURE 2 OMITTED]

Alternatively, it could be argued that due to more frequent career interruptions women prefer (and are better off in) less regulated labour markets (Estevez-Abe 2005). Since the main sources of these career interruptions are pregnancy and childbirth, this argument ignores that all Western democracies know employment-protected statutory maternity leave arrangements that prohibit the dismissal of women during and immediately after pregnancy (OECD 2007: 122, 126). As a result, currently employed women do not need to worry about being dismissed due to pregnancy. Rather, the existing job security regulations protect their positions as insiders even during these periods. As a result, job security regulations become even more valuable to insider women.

The number of not work-related labour market transitions is a function of the ability to reconcile work and family life. Especially, childcare facilities have been identified as particular important instruments to reconcile work and family life. The availability of free or reasonably priced childcare allows women to work in the labour market while freeing them, at least to a certain extent, from spending the main bulk of their working time on child-rearing. Thus, rather than spending a big part of their time for family work, childcare allows women to pursue a career as full-time employees in the labour market. Not surprisingly, countries characterized by high levels of publicly provided childcare tend to exhibit higher levels of female employment (Stadelmann-Steffen 2008). Thus, the presence of childcare facilities decreases the number of not work-related labour market transitions. As a result, career patterns of men and women would become more alike, and, as a consequence, also their preferences for job security regulations. Consequently, we expect female insiders (outsiders) to be particularly interested (uninterested) in job security in countries with low levels of childcare coverage. In contrast, we expect smaller differences between female and male insiders (outsiders) in countries with high levels of childcare coverage.

Childcare is not equally important for all women. Even if childcare is not provided free of charge by the state, it can still be bought in the private market. Thus, it is not the availability per se that matters but the availability of free or reasonably priced childcare. Better-off women are in a good position to reconcile work and family life because they can buy the necessary services in the labour market in order to regain the ability to pursue a largely uninterrupted working career (Stadelmann-Steffen 2007). Thus, no relationship between the existence of free childcare and the labour market preferences of female upscale respondents can be expected. In contrast, free or reasonably priced childcare is very important for the labour market situation of female labour market insiders and female labour market outsiders. The easier these women can reconcile work and family life, the more their labour market preferences will resemble the preferences of their male counterparts.

The household situation is an often-overlooked determinant of labour market preferences (Lewis 1992). As the division of household labour is not randomly distributed among men and women, the household situation is likely to influence the relative preferences for job security among men and women. In the following, we distinguish two different effects: responsibility and mobility. Respondents with the responsibility to provide for the economic livelihood of their family can be expected to be strongly interested in social policy interventions that increase their job security (Erlinghagen 2008). In case of job loss, not only her or his economic situation is in question, but the whole family's welfare. The situation is quite different in case of respondents who can rely on working partners and who are not the main contributors to the family income. These respondents, if they work at all, can be expected to be less interested in job security since their partners serve as an insurance against welfare loss in case of unemployment. Thus, we expect main contributors to the household income to support job security regulations while respondents with working partners can be expected to be comparatively critical of job security regulations.

In a recent contribution, Neugart (2008) has argued that in countries characterized by high levels of job security regulations and low levels of female employment, family members other than the male breadwinner support high levels of job security regulations because their economic welfare is dependent on the income of the main earner. Put differently, the whole family's economic security is a function of the main earners' job security. As a result, many female secondary wage earners adopt the social policy preferences of their male breadwinners. In a similar vein, Hakim (1995) has argued that some female outsiders may choose their outsider position voluntarily. Formerly full-time workers, they transfer quickly to undemanding part-time jobs as soon as a breadwinning spouse facilitates it. As in Neugart's account, these secondary wage earners can be expected to adopt their male breadwinners' social policy preferences. Other authors have criticized this view. These authors interpret the relationship between marriage partners as the result of bargaining under constraints. Iversen and Rosenbluth (2006) argue that considering contemporary divorce rates, the simplifying assumption of treating the family as a single unit is no longer warranted. These authors argue that there is a considerable conflict of interests between marriage partners, which is likely to affect preferences for social protection.

We follow Iversen and Rosenbluth (2006) and maintain that secondary wage earners living together with a full-time working partner (the main contributor to the household income) can be expected to develop their own social policy preferences. Firstly, labour market participation is likely to affect the social policy preferences of secondary wage earners, as their objective labour market position is different from their full-time working partners. So are the problems associated with their objective labour market position. (2) Secondly, it is not per se clear why secondary wage earners should take over the social policy preferences of their full-time working partners. It is equally possible that the latter adopt the social policy preferences of their not full-time working partners, especially considering the latter's often weaker labour market positions. Consequently, it can be expected that some of these effects cancel each other out. Thirdly, considering contemporary divorce rates secondary wage earners living in a steady relationship can no longer expect to spend the rest of their lives with their partner. As a result, these secondary wage earners need an independent source of economic security.

The household situation is also very likely to affect respondents' mobility preferences. Marital status is strongly related to the probability of establishing residence and home ownership. Once settled, the willingness to move diminishes. As a result, married respondents can be expected to be less mobile (or willing to be mobile) and, consequently, to be more interested in labour market policies that decrease the probability of moving homes such job security regulations (Schmid 2002: 408; Belot 2007). It is important to note that mobility preferences and the responsibility to provide for the economic livelihood of the household are not independent. Respondents not living in steady relationships (i.e. mobile respondents) are also very likely the main contributors to the household income.

The presence of small children in the household can be expected to influence both perceived responsibility and mobility preferences. On the one hand, the presence of children increases the number of individuals that are economically dependent on the income of the main earner. Put differently, the presence of small children increases the economic costs of job loss. On the other hand, the presence of children increases the costs of mobility as more household members would have to be moved and more social networks would have to be severed.

Table 1 summarizes the argument. Respondents can be expected to support high levels of job security regulations if they are in a steady relationship (high mobility costs) and the main income earners (high levels of responsibility) as well as if there are small children in the household (high mobility costs and high levels of responsibility). In contrast, respondents can be expected to be critical of job security regulations if they are not in a steady relationship and not the main contributor to the household income, and if there are no small children in the household.

Social policy institutions are likely to influence perceptions of responsibility. The presence of generous unemployment benefits (or other forms of social insurance) decreases the family's economic dependency on the main income earner's job. In case of job loss, the income of the main earner is replaced by generous unemployment benefits. This argument reflects claims of a trade-off between the generosity of unemployment insurance and job security regulations (Boeri et al., 2003; Iversen 2005: 46-52). Scholars in this tradition argue that the presence of generous unemployment insurance systems decreases the importance of job security regulations as a form of social protection. As a result, support for job security regulations is likely to drop. Similarly, generous family benefits provide a source of income independent of the main income earner's job. This additional income can be expected to at least partially alleviate the financial cost of having children. In the presence of generous family benefits, perceptions of responsibility for the economic welfare of children in the household should be weaker than in the absence of generous family benefits. As a result, support for restrictive job security regulations among respondents with small children should decrease as family benefits become more generous. Finally, policies that help reconciling work and family life simplify the 'sharing of the burden' in the sense of helping the secondary wage earner to increase her/his labour market participation in case of economic need.

With regard to mobility costs, we expect cross-national differences to be affected by the presence of active labour market policies (Schmid 2002: 408). These policies aim at supporting the mobility of labour market participants by providing information on open jobs and training for unemployed labour market participants as well as financially supporting the relocation of labour market participants to places characterized by more favourable labour market developments. Supporting the mobility of labour market participants through active labour market policies (ALMP) has a long tradition in Sweden, where they are a central element of the famous Rehn-Meidner model. As noted by Armingeon (2007: 906), ALMP was originally developed as part of the social democratic strategy of merging competitiveness of an open economy with the security of full employment and social justice. However, ALMP have become accepted by liberal and conservative politicians and economists alike. Nevertheless, considerable cross-national differences in the employment of ALMP can be observed. While Scandinavian countries spend around two percent of GDP on ALMP in the late 1990s, spending on ALMP is almost zero in countries such as Greece or the Great Britain (see Table 8 below).

4 Empirical Part

4.1 Individual level data and variables

The present analysis relies on the Eurobarometer 56.1 (2001) survey. This survey supplies complementary information on preferences for job security and individual attributes. Eurobarometer 56.1 (2001) contains data on 15 countries of interest, corresponding to the EU15. These countries are Austria, Belgium, Denmark, Finland, France, Germany, Great Britain, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain and Sweden. (3) The employed survey contains several variables of interest. Most importantly, respondents have been asked to evaluate the importance of job security. Following Rueda (2005, 2007), we operationalize preferences for job security using the following question: 'For you personally, how important do you think each item is in choosing a job: A secure job?' Respondents were offered five possible answers ranging from 1 ('not important at all') to 5 ('very important'). This question is very general (Kangas 1997), but it has the advantage of not making any references to other policy fields.4 Non-employed respondents have not been asked to evaluate job security. As Table 2 shows, this variable is skewed, that is, over 90 percent of the respondents opted for the highest two categories ('very important' and 'important' respectively).

Labour market status has been operationalized using the question: 'What is your current occupation?' Eurobarometer 56.1 distinguishes between self-employed respondents, managers (employed professionals, general and middle management), other white-collar respondents (working mainly at a desk and/or travelling), manual workers (service jobs, supervisors, skilled and unskilled manual workers), housekeepers, unemployed and retired (retired and/or unable to work through illness) respondents as well as students. Other questions allow distinguishing permanent from temporarily employed and fulltime from part-time employed respondents. We define labour market insiders as full-time employees under permanent contract that do not occupy a position in the general management and are not employed professionals. The latter have been coded as 'upscales'. Outsiders are employees that are working part-time, under a temporary contract or are currently unemployed. Note that there are no part-time and/or temporarily employed respondents that occupy a higher-grade professional or managerial position in the sample. Housekeepers, retired respondents and students have not been coded as outsiders since the large majority of them may be neither interested in a new job nor able to accept one. As such, they are no competition to labour market insiders. Rather, they form an own category (non-employed respondents). The last group are the self-employed. Note that all five groups are exclusive, that is, no respondent has been classified into more than one group.

As noted above, people currently not working have not been asked to evaluate the importance of job security. As a result, the group referred to as 'non-employed' above has been omitted from the sample. Moreover, unemployed respondents are not taken into account in the quantitative analysis below. As a consequence, the group of labour market outsiders only incorporates respondents in non-standard employment relationships (part-time and temporary employment). Finally, there are only very few respondents which have been classified as upscales. This could lead to estimation problems in some countries. Therefore, we merge the groups of 'self-employed' and 'upscale' respondents. Henceforth, we refer to 'upscale' and 'self-employed' respondents both as 'upscales' (see Rueda 2005 for a similar approach).

The operationalization of gender is straightforward. The household situation has been measured using three survey questions. Firstly, respondents who contribute the main part to the household income have been classified as 'main earners' while respondents who contribute the smaller part to the household income have been classified as 'secondary wage earners'. We use respondents that contribute equally with their partners to the household income as reference category. Secondly, family responsibilities have been further operationalized by a dummy variable for respondents with children below the age of five years. Finally, a dummy variable for married respondents and respondents who live in steady relationships has been used to operationalize mobility preferences.

We further introduce a battery of control variables. These variables are: (1) age and its square, (2) education, operationalized as a dummy variable for respondents who were still studying at the age of 20 or more and (3) a dummy variable for respondents that made the experience of unemployment in the last five years. To estimate the models, we use ordered logit regression with country fixed effects. The models are interpreted using SPost by Long and Freese (2006).

4.2 Individual level empirical evidence

Table 3 displays the results of ordered logit regressions of preferences for job security on variables measuring labour market status, gender, the household situation and controls. Model 1 only incorporates variables measuring labour market status and controls. Model 2 further incorporates a gender dummy variable. As can be seen in Table 3, in both models, upscale respondents consider job security to be less important in choosing a job. In contrast, the difference between labour market insiders and outsiders ceases to be significant once a control for gender is included. This result echoes the findings of Emmenegger (2009a) who observes that the preferences for job security of labour market insiders and labour market outsiders are more similar than one is inclined to expect on the basis of the insider/outsider theory of employment and unemployment.

In the present paper, we are particularly interested in investigating whether respondents' gender might help explaining this apparent contradiction between theory and empirical evidence. As displayed in Table 3's Model 2, gender does not have a significant net effect on attitudes concerning the importance of job security in choosing a job. The LR test statistic is 1.60 (df = 1, p-value = 0.206). This relationship remains insignificant even when the dependent variable is dichotomized, thereby distinguishing between respondents considering job security very important and everybody else, and logit regressions are estimated (z-value of -1.78; results not reported). As we argue above, this finding is not surprising as gender can be expected to have diametrically opposed effects on labour market insiders and labour market outsiders. Moreover, we expect the household situation to affect preferences of female and male respondents differently.

As a consequence, we first incorporate the four dummy variables measuring the household situation (Model 3 in Table 3). The incorporation of variables measuring the household situation clearly increases the explanatory power of the model. Comparing Model 3 to Model 2 (Table 3), the LR test statistic is 20.58 (df = 4, p-value = 0.000). More concretely, respondents who contribute the main part to the household income and respondents living together with partners consider job security to be more important in choosing a job than equal contributors to the household income and respondents not living together with their partners. In contrast, secondary wage earners are not less concerned with job security than equal contributors to the household income. Finally, the presence of small children does not affect preferences for job security. Turning back to the dummy variable measuring gender, Table 3's Model 3 shows that gender remains insignificant (z-value of 0.44).

In a next step, we introduce an interaction between gender and labour market status (see Model 4 in Table 3). The interpretation of interaction effects is fundamentally different in generalized linear models (such as ordered logit models) as compared to linear regression models. It is important to note that an insignificant estimate of the interaction coefficient neither indicates an insignificant effect nor does the sign of the coefficient denote the direction of the effect (Ai and Norton 2003). To test whether there is indeed no gender gap, we estimate predicted probabilities based on Table 3's Model 4 only varying the gender variable.

Table 4 presents the predicted probabilities for all three different labour market states and all five possible values of the dependent variable. The displayed values denote the change in predicted probabilities if the gender of the hypothetical respondent is changed (from male to female) and everything else remains equal. The 95 percent confidence interval is shown in brackets below the change in predicted probabilities. The table can be read as follows: In case of labour market insiders (column four), the probability of considering job security very important (y = 5) increases by 0.2 percent if the hypothetical individual turns female. However, since the confidence interval includes both positive and negative values, we cannot reject the null hypothesis that the change in predicted probabilities is not significantly different from zero (at the 95 percent level). In fact, there is not one single significant change in predicted probabilities displayed in Table 4. As a result, Table 4 does not support our hypothesis concerning an interaction effect between gender and labour market status.

Although we do not find a relationship between gender and attitudes concerning the importance of job security in choosing a job, we are able to observe significant effects of household variables on preferences for job security. More concretely, we observe that main contributors to the household income consider job security more important in choosing a job than equal contributors and secondary earners. Table 5 displays the substantive effects, estimated using Table 3's Model 3. The change in predicted probabilities is shown for a hypothetical respondent who changes from an equal contributor to a main contributor to the household income, everything else remaining equal. As can be seen in the last two rows of Table 5, the probability of answering that job security is very important in choosing a job increases by more than seven percent when the hypothetical respondent is a main contributor to the household income. In contrast, the probability of answering that job security is 'only' important decreases by more than five percent.

Similarly, we observe a significant effect of a dummy variable measuring whether a respondent is married or lives in a steady relationship (which we use to operationalize mobility preferences). As can be seen in Table 6, the probability of answering that job security is very important in choosing a job increases by more than six percent when the hypothetical respondent is married or lives in a steady relationship. In contrast, the probability of answering that job security is 'only' important decreases by almost five percent.

However, a word of caution is in order. The distinction between different forms of contribution to the household income is only meaningful for persons living together. Conversely, only persons living together are able to allocate tasks between partners, such as one being the main contributor to the household income and the other one being the secondary wage earner. Put differently, an interaction effect between 'living together' and 'main contributor to the household income' can be expected. As a consequence, we have re-estimated Table 3's Model 3 incorporating an interaction for the dummy variables 'living together' and 'main contributor to the household income' (model not reported). (5) Table 7 displays the change of predicted probabilities for all four possible combinations of 'living together' (0/1) and 'main contributor' (0/1). As can be seen, 'main contributor' only affects preferences for job security in case of respondents living together with their partners, i.e. respondents who are married or in steady relationships (column 2). In this case, the probability of answering that job security is very important increases by over seven percent (column 2, y = 5). In contrast, main contributors are not significantly more interested in job security when they are not living together with their partners (column 3). Similarly, 'living together' only affects preferences for job security if these respondents are also main contributors to the household income (column 4). In this case, the probability of answering that job security is very important increases by over six percent (column 4, y = 5). In contrast, 'living together' does not affect the preferences for job security of respondents who are not the main contributors to the household income (column 5).

4.3 Aggregate Level Data and Variables

In a second step, we explore the determinants of cross-national differences in attitudes concerning the importance of job security in choosing a job. As shown above, we can observe two significant determinants of preferences for job security: 'living together' and 'main contributor to the household income'. In the following, we investigate whether the effect of being the main contributor to the household income and the effect of living together respectively on the evaluation of job security varies systematically across countries. In the theoretical part, we have argued that these cross-national differences might co-vary with social policies and policies that help reconciling work and family life. In the case of cross-national differences with regard to the evaluation of job security by different forms of contribution to the household income, we test whether these differences can be explained by policies that help reconcile work and family life (childcare and family benefits) and social policies that protect the income of households (unemployment benefits). In case of cross-national differences with regard to the effect of living together on the evaluation of job security, we test whether these differences can be explained by policies that help reconcile work and family life (childcare and family benefits) and social policies that reduce mobility costs (active labour market policies). Moreover, we control whether these cross-national differences can be explained by the level of employment protection, i.e. the regulation of hiring and firing, as we hypothesize that preferences for job security reflect preferences for job security regulations. (6) Table 8 displays the aggregate level data and the data sources.

To test these aggregate level relationships, we estimate a hierarchical model in two steps. First, we estimate for each country a separate model analogue to Table 3's Model 3. In a second step, we regress the country specific coefficients for the variable of interest on the explanatory variables. To adjust for uncertainty in the first step, we employ feasible weighted least squares in the aggregate level regressions (Jusko and Shively 2005). We further incorporate a control variable for labour market risks, that is, a variable reflecting the current state of the labour market. This variable is the sum of the z-standardized values of the change of the unemployment rate (multiplied by -1), the change of the employment rate and average GDP growth in the period 1996 to 2001 (Cronbach's alpha = 0.86) in a given country. See Table 8 for the raw data.

As noted above, there is likely an interaction effect between 'main contributor to the household income' and 'living together'. However, there are only between 294 (Luxembourg) and 511 (Sweden) observations per country. Moreover" the number of respondents living together with a partner and at the same time contributing the main part to the household income varies between 1f6 (Finland) and f49 (Sweden). As a result" we present both the models without (Tables 9 and 11) and with the interaction effect (Tables 1l and 1f). In the latter case" we estimate for each country a separate model including an interaction effect between 'main contributor to the household income' and 'living together'. (7) We then use the country specific change in predicted probabilities as the dependent variable in our cross-national regression models" for which we use ordinary least squares regression. Again" we control for labour market risks.

4.4 Aggregate Level Empirical Evidence

Table 9 displays the results of feasible weighted least squares of cross-national differences with regard to the evaluation of job security by main contributors to the household income as opposed to equal contributors on social policies and policies that help reconciling work and family life. We control for labour market risks and further add a dummy variable for Great Britain as regression diagnostics have identified Great Britain as a particularly influential case (identified using Cook's D after OLS regressions" with Great Britain scoring 0.444" which is considerably bigger than 4/n). Given the low number of observations (14 to 1.)" a slightly less stringent significance level could be employed. Nevertheless" as Table 9 shows" no significant effects" apart from the dummy variable for the extreme case Great Britain" can be observed. Moreover" wald tests of each regression model displayed in Table 9 indicate that none adds any explanatory power as compared to an intercept-only model.

For Table 10, we take the interaction effect between 'main contributor' and 'living together' into account. As a dependent variable, we calculate for each country separately the change of predicted probabilities when changing from not being main contributor to being main contributor to the household income in case of respondents living together with their partners. As in case of Table 9, there is only little evidence for a systematic aggregate level relationship. Only in Model 16, we are able to observe an adjusted [R.sup.2] clearly bigger than zero (adj. [R.sup.2] = 0.216). According to Model 16, high levels of employment protection legislation are associated with smaller differences between main contributors to the household income and the remaining respondents. However, this result should be taken with a grain of salt as a similar relationship could not be found in Table 9.

With regard to the cross-national differences in attitudes concerning the importance of job security in choosing a job of respondents living together as opposed to the remaining respondents, Table 11 shows the results of feasible weighted least squares regressions of cross-national differences on social policies and policies that help reconciling work and family life using the model specification without an interaction effect between 'main earner' and 'living together'. As in Tables 9 and 10, we again introduce a control for labour market risks and further add a dummy variable for Greece as regression diagnostics have identified Greece as a particularly influential case (identified using Cook's D after OLS regressions, with Greece scoring 0.451, which is considerably bigger than 4/n). As Table 11 shows, we can observe several significant relationships. Cross-national differences increase as a function of childcare fees while they decrease as a function of childcare enrolment and family benefits. Moreover, there is a significant negative effect of spending on active labour market policies on cross-national differences.

However, as Table 12 shows, we should be careful when interpreting these findings. For Table 12, the interaction effect between 'main contributor' and 'living together' has been taken into account. The dependent variable, calculated for each country separately, corresponds to the change of predicted probabilities when changing from living alone to living together in case of respondents who are the main contributors to the household income. Using this dependent variable, all coefficients, except the coefficients for the variable 'childcare enrolment' and the dummy variable for the extreme case Greece, cease to be significantly different from zero. (8) Thus, we are only able to identify one robust aggregate level relationship: cross national differences with regard to the attitudes concerning the importance of job security in choosing a job of respondents living together as opposed to the remaining respondents decrease as a function of childcare enrolment. However, considering the unclear interpretation of this result and potential problems of endogeneity, that is, policies explaining attitudes towards related policies, we refrain from interpreting this finding in any detail. (9)

5. Conclusions

In the present paper, we have hypothesized that the insider/outsider theory of employment and unemployment suffers from a gender bias. It is based on the comparison of full-time employees with permanent contracts to unemployed individuals. Moreover, the possibility of not work-related labour market transitions is not taken into account. In his seminal work, Rueda (2005, 2007) has advanced the insider/outsider theory by incorporating the role of non-standard employment relationships, most notably part-time and temporary contracts. In this paper, we maintain that this is not enough. Most importantly, the insider/outsider theory in its present form does not allow for gender differences.

We believe that there are very good reasons to expect gender to matter. Firstly, on average, women experience more labour market transitions than men. Some of these labour market transitions are not job-related but the result biological differences (women give birth to children while men do not). Although not all women experience more labour market transitions than men, the phenomenon of statistical discrimination brings about that all women are treated as if they would experience more labour market transitions than men. Because the transition from unemployment to employment is more difficult for women than for men, we hypothesize that female outsiders are even more critical of labour market institutions that complicate their return to an insider position. As a consequence, we expect them to express less interest in job security regulations. Conversely, as job security regulations strengthen the labour market position of employed women, and since women tend to have weaker labour market positions than men, we expect female insiders to express more interest in job security than their male counterparts.

Secondly, the insider/outsider theory of employment and unemployment does not incorporate the household situation. In this paper, we hypothesize that the household situation--theoretically conceptualized as mobility and responsibility preferences--not only affects preferences for job security, but also that it affects women and men differently. The latter is the result of the uneven distribution of paid and unpaid labour between the genders.

In the empirical part, we show that respondents carrying the main responsibility for the economic livelihood of the family, the main contributors to the family income, consider job security more important in choosing a job than equal contributors and secondary wage earners. Moreover, we show that respondents living together with a partner are more interested in job security than respondents living alone. In contrast, we observe that gender does not--neither as an independent variable nor in interaction with labour market status--significantly affect attitudes concerning the importance of job security in choosing a job.

In a second step, we explored whether social policies and policies that help reconciling work and family life can explain cross-national differences with regard to, on the one hand, cross-national differences between respondents living together and respondents living alone and, on the other hand, cross-national differences between main contributors to the household income and equal contributors/ secondary wage earners. The results of the empirical analysis are sobering. We observe no robust cross-national co-variations.

Where do we go from here? We identify two important directions for future research and one important qualification of our results. Firstly, we have shown that the household situation has an important effect on preferences for job security. Theoretically, we conceptualized the household situation as mobility and responsibility preferences. However, the theoretical arguments put forward are still rather simplistic. A more thorough conceptualization should further take the interaction between mobility and responsibility into account. With regard to empirics, better data is needed in order to analyze the link between the household situation and preferences for job security. Although the Eurobarometer 56.1 (2001) survey provides data on both the household situation and preferences for job security, sample size complicates the empirical analysis. For instance, there are only 126 Finnish respondents who simultaneously live together with a partner and contribute the main part to the household income.

Secondly, we still lack an adequate explanation for the 'odd' finding that labour market outsiders are as interested in job security as labour market insiders. We advance two possible explanations for this finding. Firstly, the classification of labour market participants into insiders, outsiders, self-employed and upscales is too crude and too static. While we are sympathetic to this critique, we do not believe that it can explain our surprising results. Most notably, this argument is at odds with our finding that a history of unemployment, that is, the experience of unemployment in the last five years before the survey was conducted, has no significant effect on preferences for job security. Secondly, the long-term interest of eventually obtaining an insider position might induce outsiders (and respondents experiencing an above average number of labour market transitions) to support job security regulations despite short-term disadvantages (long-term rationality). However, with the data at hand, this hypothesis cannot be followed up.

Finally, the presented findings have to be qualified for two important reasons. Firstly, we acknowledge that the operationalization of the dependent variable is problematic. We are first and foremost interested in preferences for job security regulations. However, the dependent variable only captures preferences for secure employment. Although our operationalization follows the literature on the determinants of preferences for job security regulations (Rueda 2005, 2007; Emmenegger 2009ac), we acknowledge that this operationalization diminishes the relevance of our findings. Secondly, due to data availability, we were not able to include the unemployed in our analysis. The unemployed are the archetype of labour market outsiders. Nevertheless, they are as supportive of job security regulations as labour market insiders (Emmenegger 2009a). In the present paper, we were able to shed some light on the reasons for the surprisingly high levels of support for job security regulations among part-time and temporarily employed respondents. Unfortunately, we were not able to do the same for unemployed respondents.

ACKNOWLEDGEMENTS

This manuscript has been previously published as RECWOWE working paper No. 2/2010. We would like to thank in particular Lucas Leemann, all participants in the panel "Multilevel Approaches to the Demand for Welfare Policies" at the 5th ECPR General Conference and the anonymous reviewers for very helpful comments and suggestions. All remaining errors are the authors' responsibility.

(1.) Rueda (2005, 2007) takes the role of non-standard employment contracts into account.

(2.) Note that the model already distinguishes between labour market outsiders (unemployed, temporarily or part-time employed respondents) and non-employed respondents. The latter are expected to be more supportive of job security regulations than labour market outsiders (see Figure 2).

(3.) Due to different institutionalist experiences in the postwar period, significant differences between East and West Germany have to be expected (Svallfors 2006: 193). Thus, East and West Germany are often treated as two different 'countries' in quantitative analyzes of individual level data. In multilevel settings, this implies that two different 'countries', that is East and West Germany, have the same scores on the aggregate level indicators, mostly reflecting formerly West German institutions. We avoid this situation by omitting East Germany from the dataset. In the present analysis, Germany only refers to West Germany.

(4.) Admittedly, this dependent variable is far from perfect. Preferences for job security regulations are often operationalized using a question on attitudes towards governmental support for declining industries to protect jobs (e.g. Algan and Cahuc 2006; Emmenegger 2009ab). This question, however, has the drawback of including aspects of industrial policy, which may not necessarily be supported by the respondents. Emmenegger (2009b) shows using ISSP 1996 and ISSP 1997 surveys that both questions lead to similar results.

(5.) To simplify the model, we have omitted the dummy variable for secondary wage earners, i.e. we use equal contributors and secondary wage earners as reference category.

(6.) We are certainly aware of the fact that there is no straightforward relationship between individual level preferences and aggregate level institutions.

(7.) However, in order to simplify the model, we have again omitted the dummy variable for secondary wage earners, i.e. we use equal contributors and secondary wage earners as reference category.

(8.) Again, we control for labour market risks and incorporate a dummy variable for the extreme case Greece (identified using Cook's D, with Greece scoring 0.306, which is clearly bigger than 4/n).

(9.) Also note that the coefficient of the variable 'childcare fees' ceases to be significantly different from zero in Model 23 in Table 12.

REFERENCES

Ai, C. and Norton, E. C. (2003), "Interaction Terms in Logit and Probit Models," Economic Letters 80(1): 123-129.

Algan, Y. and Cahuc, P. (2006), "Job Protection: The Macho Hypothesis," Oxford Review of Economic Policy 22(3): 390-410.

Armingeon, K. (2007), "Active Labour Market Policy, International Organisations, and Domestic Politics," Journal of European Public Policy 14(6): 905-932.

Armingeon, K., Gerber, M., Leimgruber, P. and Beyeler, M. (2008), "Comparative Political Data Set 1960-2006," Institute of Political Science, University of Bern.

Belot, M. (2007), "Why is Employment Protection Stricter in Europe than in the United States?" Economica 74(295): 397-423.

Blanchard, O. (2006), "European Unemployment: The Evolution of Facts and Ideas," Economic Policy 21(45): 5-59.

Boeri, T., Conde-Ruiz, J. I. and Galasso, V. (2003), "Protection Against Labour Market Risk: Employment Protection or Unemployment Benefits?" IZA DP-834.

Bonoli, G. (2003), "Social Policy through Labor Markets: Understanding National Differences in the Provision of Economic Security to Wage Earners," Comparative Political Studies 36(9): 1007-1030.

Elmeskov, J., Martin, J. P. and Scarpetta, S. (1998), "Key Lessons for Labour Market Reforms: Evidence from OECD Countries' Experiences," Swedish Economic Policy Review 5(2): 205-252.

Emmenegger, P. (2009a), "Barriers to Entry: Insider/Outsider Politics and the Political Determinants of Job Security Regulations," Journal of European Social Policy 19(2): 131-146.

Emmenegger, P. (2009b), Regulatory Social Policy: The Politics of Job Security Regulations. Bern-Stuttgart-Wien: Haupt.

Emmenegger, P. (2009c), "Specificity vs. Replaceability: The Relationship between Skills and Preferences for Job Security Regulations," Socio-Economic Review 7(3): 407-430.

Erlinghagen, M. (2008), "Self-Perceived Job Insecurity and Social Context: A Multi-Level Analysis of 17 European Countries," European Sociological Review 24(2): 183-197.

Esping-Andersen, G. (1996), "Welfare States without Work: The Impasse of Labour Shedding and Familialism in Continental European Social Policy," in Esping-Andersen, G. (ed.), Welfare States in Transition: National Adaptations in Global Economies. London-Thousand Oaks, CA-New Delhi: Sage, 66-87.

Esping-Andersen, G. (1999), Social Foundations of Postindustrial Economies. New York: Oxford University Press.

Estevez-Abe, M. (2005), "Gender Bias in Skills and Social Policies: The Varieties of Capitalism Perspective on Sex Segregation," Social Politics 12(2): 180-215.

Hakim, C. (1995), "Five Feminist Myths about Women's Employment," British Journal of Sociology 46(3): 429-455.

Hausermann, S. and Schwander, H. (2009), "Varieties of Dualization? Identifying Outsiders across Countries," RECWOWE WP- 09.

Iversen, T. (2005), Capitalism, Democracy, and Welfare. Cambridge: Cambridge University Press.

Iversen, T. and Rosenbluth, F. (2006), "The Political Economy of Gender: Explaining Cross-National Variation in the Gender Division of Labor and the Gender Voting Gap," American Journal of Political Science 50(1): 1-19.

Kangas, O. (1997), "Self-Interest and the Common Good: The Impact of Norms, Selfishness and Context in Social Policy Opinions," Journal of Socio-Economics 26(5): 475-494.

Kaufmann, F. (2003), Sozialpolitisches Denken. Frankfurt: Suhrkamp.

Lewis, J. (1992), "Gender and the Development of Welfare Regimes," Journal of European Social Policy 2(3): 159-173.

Lindbeck, A. and Snower, D. J. (1988), The Insider-Outsider Theory of Employment and Unemployment. Cambridge, MA-London: MIT Press.

Long, J. S. and Freese, J. (2006), Regressions Models for Categorial Dependent Variables Using Stata. College Station, TX: Stata Press.

Long Jusko, K. and Shively, W. P. (2005), "Applying a Two-Step Strategy to the Analysis of Cross-National Public Opinion Data," Political Analysis 13(4): 327-344.

Neugart, M. (2008), "The Choice of Insurance in the Labor Market," Public Choice 134(3-4): 445-462.

OECD (2007), Babies and Bosses: Reconciling Work and Family Life. Paris: OECD.

Pierson, P. (2001), "Coping with Permanent Austerity: Welfare State Retrenchment in Affluent Democracies," in Pierson, P. (ed.), The New Politics of the Welfare State. New York: Oxford University Press: 410-456.

Rueda, D. (2005), "Insider-Outsider Politics in Industrialized Democracies: The Challenge to Social Democratic Parties," American Political Science Review 99(1): 61-74.

Rueda, D. (2007), Social Democracy Inside Out: Partisanship and Labor Market Policy in Industrialized Democracies. New York: Oxford University Press.

Samek Lodovici, M. (2000), "The Dynamics of Labour Market Reform in European Countries," in Esping-Andersen, G. and Regini, M. (eds.), Why Deregulate Labour Markets? New York: Oxford University Press, 30-65.

Schmid, G. (2002), Wege in eine neue Vollbeschaftigung. Ubergangsmarkte und aktivierende Arbeitsmarktpolitik. Frankfurt/New York: Campus.

Stadelmann-Steffen, I. (2007), "Der Einfluss der sozialpolitischen Kontexte auf die Frauenerwerbstatigkeit in der Schweiz," Kolner Zeitschrift fUr Soziologie und Sozialpsychologie 59(4): 589-614.

Stadelmann-Steffen, I. (2008), "Woman, Labor, and Public Policy--The Female Labour Market Integration in the OECD Countries: A Comparative Perspective," Journal of Social Policy 37(3): 383-408.

Svallfors, S. (1997), "Worlds of Welfare and Attitudes to Redistribution: A Comparison of Eight Western Nations," European Sociological Review 13(3): 283-304.

Svallfors, S. (2006), The Moral Economy of Class: Class and Attitudes in Comparative Perspective. Stanford CA: Stanford University Press.

Tsakalotos, E. (2004), "Market Constraints, Economic Performances and Political Power: Modernizers versus Leftists," Socio-Economic Review 2(3): 415-424.

PATRICK EMMENEGGER

emm@sam.sdu.dk

University of Southern Denmark
Table 1 Preferences for Job Security as a Function of Household
Situation

                                                 Small children in
                                                 household?
                                                 (Mobility
                                                 and responsibility)

                                                 Yes     No

Steady          Yes    Main earner?       Yes    +++     ++
relationship?          (responsibility)   No     ++      +
(mobility)      No     Main earner?       Yes    ++      +
                       (responsibility)   No     +       0

Table 2 Importance of Job Security in Choosing a Job: Frequencies

Very important                            3346       56.14%

Important                                 2242       37.62%

Neither important nor unimportant          240        4.03%

Not important                              111        1.86%

Not important at all                        21        0.35%

N                                         5960      100.00%

Table 3 Determinants of the Importance of Job Security in Choosing a
Job

Dependent variable:                   Importance of job security

                                       Model 1      Model 2

Upscales                                -0.40 ***    -0.42 ***
                                       (-4.8)       (-5.0)
Insiders                                 0.16 *       0.14
                                        (2.3)        (1.9)

Outsiders                             Reference category

Gender (woman = 1)                        --          -0.07
                                                     (-1.3)
Upscales x gender (woman = 1 )            --

Insiders x gender (woman = 1 )            --           --

Age                                     -0.02        -0.02
                                       (-1.6)       (-1.6)
[Age.sup.2]                              0.00         0.00
                                        (1.4)        (1.4)
History of unemployment                 -0.13        -0.13
                                       (-1.8)       (-1.8)
Highly educated                         -0.42 ***    -0.42 ***
                                       (-7.2)       (-7.2)
Living together

Main contributor to family income         --           --

Secondary wage earner                     --           --

Equal contributors to family income   Reference category

Child below five years                    --           --

FE: Belgium                           Reference category

FE: Denmark                           -0.88 ***    -0.88 ***
                                     (-6.5)       (-6.5)
FE: Germany (West)                    -0.30 *      -0.30 *
                                     (-2.1)       (-2.1)
FE: Greece                             0.63 ***     0.63 ***
                                      (3.9)        (3.9)
FE: Italy                             -0.18        -0.18
                                     (-1.3)       (-1.2)
FE: Spain                              0.12         0.12
                                      (0.8)        (0.8)
FE: France                            -0.2         -0.2
                                     (-1.4)       (-1.4)
FE: Ireland                           -0.66 ***    -0.66 ***
                                     (-4.6)       (-4.6)
FE: Luxembourg                        -0.03        -0.03
                                     (-0.2)       (-0.2)
FE: Netherlands                       -1.10 ***    -1.10 ***
                                     (-7.7)       (-7.7)
FE: Portugal                          -0.77 ***    -0.77 ***
                                     (-5.5)       (-5.4)
FE: Great Britain                     -0.52 ***    -0.51 ***
                                     (-3.6)       (-3.6)
FE: Finland                            0.10         0.11
                                      (0.7)        (0.7)
FE: Sweden                            -0.54 ***    -0.54 ***
                                     (-4.0)       (-4.0)
FE: Austria                           -0.08        -0.08
                                     (-0.6)       (-0.5)
cut1
Constant                              -6.76 ***    -6.81 ***
                                    (-18.5)      (-18.6)
cut2
Constant                               4.89 ***     4.94 ***
                                    (-16.1)      (-16.1)

cut3
Constant                              -3.79 ***    -3.84 ***
                                    (-12.8)      (-12.9)
cut4
Constant                              -1.22 ***    -1.27 ***
                                     (-4.2)       (-4.3)
Pseudo [R.sup.2]                       0.036        0.037
Log likelihood                     -5256.2      -5255.4
No. of observations                   5960         5960

Dependent variable:                   Importance of job security

                                       Model 3      Model 4

Upscales                              -0.43 ***     -0.41 **
                                     (-5.0)        (-3.0)
Insiders                               0.12          0.16
                                      (1.7)         (1.3)

Outsiders

Gender (woman = 1)                     0.03          0.07
                                      (0.4)         (0.5)
Upscales x gender (woman = 1 )                      -0.02
                                                   (-0.1)
Insiders x gender (woman = 1 )            --        -0.06
                                                   (-0.4)
Age                                   -0.04 **      -0.04 **
                                     (-2.7)        (-2.7)
[Age.sup.2]                            0.00 *        0.00 *
                                      (2.3)         (2.3)
History of unemployment               -0.11         -0.11
                                     (-1.5)        (-1.5)
Highly educated                       -0.41 ***     -0.41 ***
                                     (-7.1)        (-7.1)
Living together                        0.25 ***      0.25 ***
                                      (3.7)         (3.7)
Main contributor to family income      0.29 **       0.28 **
                                      (2.9)         (2.9)
Secondary wage earner                  0.11          0.11
                                      (1.1)         (1.1)
Equal contributors to family income

Child below five years                 0.00          0.00
                                      (0.0)         (0.0)
FE: Belgium

FE: Denmark                           -0.89 ***     -0.89 ***
                                     (-6.6)        (-6.6)
FE: Germany (West)                    -0.30 *       -0.30 *
                                     (-2.0)        (-2.1)
FE: Greece                             0.62 ***      0.63 ***
                                      (3.8)         (3.8)
FE: Italy                             -0.14         -0.13
                                     (-0.9)        (-0.9)
FE: Spain                              0.15          0.15
                                      (1.0)         (1.0)
FE: France                            -0.2          -0.2
                                     (-1.4)        (-1.4)
FE: Ireland                           -0.65 ***     -0.65 ***
                                     (-4.5)        (-4.5)
FE: Luxembourg                        -0.04         -0.04
                                     (-0.3)        (-0.3)
FE: Netherlands                       -1.12 ***     -1.12 ***
                                     (-7.8)        (-7.8)
FE: Portugal                          -0.78 ***     -0.78 ***
                                     (-5.5)        (-5.5)
FE: Great Britain                     -0.52 ***     -0.52 ***
                                     (-3.6)        (-3.6)
FE: Finland                            0.13          0.13
                                      (0.9)         (0.9)
FE: Sweden                            -0.54 ***     -0.54 ***
                                     (-4.0)        (-4.0)
FE: Austria                           -0.06         -0.06
                                     (-0.4)        (-0.4)
cut1
Constant                              -6.73 ***     -6.71 ***
                                    (-17.7)       (-17.4)
cut2
Constant                               4.86 ***      4.84 ***
                                    (-15.1)       (-14.7)

cut3
Constant                              -3.76 ***     -3.74 ***
                                    (-12.0)       (-11.7)
cut4
Constant                              -1.19 ***     -1.16 ***
                                     (-3.8)        (-3.7)
Pseudo [R.sup.2]                       0.039         0.039
Log likelihood                     -5245.1       -5245.1
No. of observations                    5960          5960

Note:   Ordered logit regressions with country fixed effects,
z-values in parentheses. Significance: *** p < 0.001, ** p < 0.01,
* p < 0.05.

Table 4 Labour Market Status x Gender: Change in Predicted
Probabilities

                       Upscales

                       [DELTA][PI]

P(y = 1 | x)            -0.0002
                       [-0.0012, 0.0008]
P(y = 2 | x)            -0.0011
                       [-0.0064, 0.0043]

P(y = 3 | x)            -0.0021
                       [-0.0129, 0.0087]
P(y = 4 | x)            -0.0079
                       [-0.0484, 0.0325]
P(y = 5 | x)             0.0113
                       [-0.0463, 0.0689]

                       Outsiders

                       [DELTA][PI]

P(y = 1 | x)            -0.0002
                       [-0.0010, 0.0006]
P(y = 2 | x)            -0.0010
                       [-0.0052, 0.0031]

P(y = 3 | x)            -0.0022
                       [-0.0111, 0.0066]
P(y = 4 | x)            -0.0129
                       [-0.0639, 0.0380]
P(y = 5 | x)             0.0164
                       [-0.0484, 0.0812]

                       Insiders

                       [DELTA][PI]

P(y = 1 | x)            -0.0000
                       [-0.0004, 0.0004]
P(y = 2 | x)            -0.0001
                       [-0.0022, 0.0020]

P(y = 3 | x)            -0.0003
                       [-0.0048. 0.0043]
P(y = 4 | x)            -0.0016
                       [-0.0312, 0.0279]
P(y = 5 | x)             0.0020
                       [-0.0345, 0.0385]

Note: Change of predicted probabilities (from male to female)
estimated on the basis of Table 3's Model 4 using SPost
(Long and Freese 2006).

Table 5 Contribution to Household Income:
Change in Predicted Probabilities

                       Main contributor
                       Reference: equal contributors
                       [DELTA][PI]

P(y = 1 | x)            -0.0009
                       [-0.0017, -0.0001]
P(y = 2 | x)            -0.0049
                       [-0.0086, -0.0012]

P(y = 3 | x)            -0.0102
                       [-0.0178, -0.0026]
P(y = 4 | x)            -0.0548
                       [-0.0911,-0.0185]
P(y = 5 | x)             0.0708
                        [0.0227, 0.1188]

Note: Change of predicted probabilities (from equal contributor to main
contributor) estimated on the basis of Table 3's Model 3 using SPost
(Long and Freese 2006).

Table 6 Living Together: Change in Predicted Probabilities

                       Living together

                       [DELTA][PI]

P(y = 1 | x)            -0.0008
                       [-0.0014, -0.0002]
P(y = 2 | x)            -0.0043
                       [-0.0068, -0.0017]
P(y = 3 | x)            -0.0089
                       [-0.0140, -0.0039]
P(y = 4 | x)            -0.0482
                       [-0.0734, -0.0229]
P(y = 5 | x)             0.0622
                        [0.0292,0.0951]

Note: Change of predicted probabilities (from living alone to living
together) estimated on the basis of Table 3's Model 3 using SPost (Long
and Freese 2006).

Table 7 Living Together x Main Contributor:
Change in Predicted Probabilities

Effect of:        Main contributor       Main contributor
Situation:        Living together        Not living together
                  [DELTA][PI]            [DELTA][PI]

P(y = 1 | x)       -0.0009                -0.0002
                  [-0.0015, -0.0003]     [-0.0011, 0.0007]
P(y = 2 | x)       -0.0048                -0.0012

                  [-0.0075, -0.0020]     [-0.0060, 0.0036]
P(y = 3 | x)       -0.0101                -0.0025
                  [-0.0157, -0.0045]     [-0.0124, 0.0074]
P(y = 4 | x)       -0.0582                -0.0121
                  [-0.0882, -0.0282]     [-0.0590, 0.0347]
P(y = 5 | x)        0.0740                 0.0161
                   [0.0355, 0.1124]      [-0.0463, 0.0785]

Effect of:        Living together        Living together
Situation:        Main contributor       Not main contributor
                  [DELTA][PI]            [DELTA][PI]

P(y = 1 | x)       -0.0007                -0.0004
                  [-0.0013, -0.0002]     [-0.0013, 0.0005]
P(y = 2 | x)       -0.0039                -0.0020

                  [-0.0063, -0.0015]     [-0.0067, 0.0026]
P(y = 3 | x)       -0.0084                -0.0043
                  [-0.0134, -0.0034]     [-0.0140, 0.0055]
P(y = 4 | x)       -0.0493                -0.0211
                  [-0.0776, -0.0209]     [-0.0673, 0.0252]
P(y = 5 | x)        0.0623                 0.0277
                   [0.0264, 0.0982]      [-0.0337, 0.0892]

Note: Change of predicted probabilities (columns 2 and 3: from not main
contributors to main contributors; columns 4 and 5: from not living
together to living together) estimated using SPost (Long and Freese
2006). Model not reported.

Table 8 Aggregate Level Variables

                  Childcare fees per two-          Enrolment rates in
                      year old, attending      childcare for children
                   accredited early-years        under 3 years (2004)
                       care and education
                 services as % of average
                     worker income (2004)

Country

Austria                               9,6                         4,1
Belgium                              19,7                        38,5
Denmark                               8,4                        61,7
Finland                               7,6                        22,4
France                               25,1                        26,0
Germany                               9,1                         9,0
Great Britain                        24,7                        25,8
Greece                                4,5                         7,0
Ireland                              24,8                        15,0
Italy                                                             6,3
Luxembourg                           32,4                        14,0
Netherlands                          17,5                        29,5
Portugal                             27,8                        23,5
Spain                                30,3                        20,7
Sweden                                4,5                        39,5

                  Family benefits as % of    Unemployment benefits as
                  net 100% average worker     % of net average worker
                    income (2001 ), type:        income (2001), type:
                  married couple with two     married couple with two
                     children, spouse not        children, spouse not
                                  working                     working

Country

Austria                              12,6                        65,0
Belgium                              11,7                        49,7
Denmark                              11,2                        72,9
Finland                              12,8                        82,0
France                                5,9                        69,4
Germany                              12,8                        71,3
Great Britain                         6,8                        61,5
Greece                                2,2                        42,3
Ireland                               9,4                        69,4
Italy                                 7,9                        51,3
Luxembourg                           13,2                        87,7
Netherlands                           5,9                        68,7
Portugal                              4,8                        76,4
Spain                                 0,0                        75,7
Sweden                               12,1                        69,1

                            Public social        Employment protection
                         expenditure as a        legislation indicator
                        percentage of GDP                       (2001)
                                  '(2001)

Country

Austria                              26,6                         2,2
Belgium                              25,8                         2,2
Denmark                              26,1                         1,5
Finland                              24,2                         2,0
France                               27,9                         3,1
Germany                              26,3                         2,3
Great Britain                        20,1                         0,7
Greece                               20,6                         3,5
Ireland                              14,4                         0,9
Italy                                23,5                         2,0
Luxembourg                           20,8
Netherlands                          19,7                         2,1
Portugal                             19,9                         3,7
Spain                                20,0                         3,1
Sweden                               28,9                         2,2

                      Average spending on         Average spending on
                     active labour market      active labour policies
                     policies as % of GDP    as % of GDP (1996-2001),
                              (1996-2001)     divided through average
                                                    unemployment rate
                                                          (1996-2001)

Country

Austria                               0,5                        12,6
Belgium                               1,2                        14,5
Denmark                               1,9                        35,8
Finland                               1,2                        10,2
France                                1,2                        10,8
Germany                               1,2                        13,8
Great Britain                         0,4                         6,0
Greece                                0,3                         2,8
Ireland                               1,1                        15,0
Italy                                 0,5                         4,5
Luxembourg                            0,2                         6,5
Netherlands                           1,5                        35,0
Portugal                              0,6                        10,7
Spain                                 0,6                         3,3
Sweden                                2,1                        26,9

                           Improvement of              Improvement of
                          employment rate    unemployment rate change
                              (1996-2001)                 (1996-2001)

Country

Austria                               0,4                         0,6
Belgium                               3,4                         3,3
Denmark                               2,2                         2,7
Finland                               5,6                         5,3
France                                2,7                         3,3
Germany                               1,7                         1,0
Great Britain                         2,7                         3,4
Greece                                0,2                        -0,5
Ireland                               9,9                         8,3
Italy                                 3,5                         2,1
Luxembourg                            3,7                         1,5
Netherlands                           7,1                         3,9
Portugal                              5,7                         3,3
Spain                                 9,6                        11,5
Sweden                                3,8                         4,9

                       Average GDP growth
                              (1996-2001)

Country

Austria                               2,6
Belgium                               2,4
Denmark                               2,5
Finland                               4,4
France                                2,7
Germany                               1,9
Great Britain                         3,3
Greece                                3,6
Ireland                               9,0

Italy                                 1,9
Luxembourg                            5,5
Netherlands                           3,7
Portugal                              3,7
Spain                                 4,0
Sweden                                3,0

Sources: OECD (2007), OECD Tax-Benefit Calculator (accessed 26.6.2009),
OECD Social Expenditure Database (accessed 10.8.2009), OECD Labour
Force Statistics (accessed 10.8.2009), Penn World Table 6.2
(accessed 10.8.2009) and Armingeon et al. (2008).

Note: For the calculation of family and unemployment benefits, family
benefits have not been considered as part of the net income of the
average worker.

Table 9 Main Contributor to Household Income:
Determinants of Cross-National Differences

Dependent variable:    Main contributor to household income:
                       cross-national differences

                        Model 5      Model 6      Model 7

Childcare fees           -0.002         --           --
                        (-0.15)
Childcare enrolment                   -0.008
                                     (-1.21)
Family benefits            --           --          0.017
                                                   (0.61)
Unemployment
benefits

Social expenditure

Employment
protection
legislation

Labour market risks       0.014        0.002        0.010
                         (0.29)       (0.05)       (0.24)
Great Britain (dummy     -0.972(*)    -1.020(*)    -0.991(*)
variable)               (-1.75)      (-1.90)      (-1.83)

Constant                  0.353        0.541 **     0.207
                         (1.50)       (2.94)       (0.76)

Wald test,                0.309        0.156        0.247
[chi.sup.2] (3)

N                          14           15           15

Dependent variable:    Main contributor to household income:
                       cross-national differences

                        Model 8      Model 9      Model 10

Childcare fees             --           --           --

Childcare enrolment

Family benefits            --           --           --

Unemployment             -0.005
benefits                (-0.44)

Social expenditure                    -0.017
                                     (-0.44)
Employment                                         -0.160
protection                                        (-1.02)
legislation

Labour market risks       0.010       -0.016       -0.006
                         (0.23)      (-0.29)      (-0.16)
Great Britain (dummy     -1.060 *     -1.115 *     -1.303 *
variable)               (-1.97)      (-1.98)      (-2.19)

Constant                  0.679        0.766        0.729(*)
                         (0.93)       (0.82)       (1.94)

Wald test,                0.265        0.265        0.186
[chi.sup.2] (3)

N                          15           15           14

Notes: Feasible weighted least squares regressions, z-values in
parentheses. Significance: *** p < 0.001, ** p < 0.01,
* p < 0.05, (*) p < 0.10.

Table 10 Main Contributor to Household Income (when Living
Together): Determinants of Cross-National Differences

Dependent variable:    Main contributor to household income
                       (when living together): cross-national
                       differences

                        Model 11     Model 12     Model 13

Childcare fees            0.000          --           --
                         (0.06)
Childcare enrolment                   -0.001
                                     (-0.75)
Family benefits            --                       0.008
                                                   (1.76)
Unemployment
benefits

Social expenditure         --           --           --

Employment
protection
legislation

Labour market risks      -0.001       -0.002        0.001
                        (-0.11)      (-0.25)       (0.15)

Constant                  0.044        0.075(*)    -0.019
                         (0.89)       (2.03)      (-0.43)

Adj. [R.sup.2]           -0.181       -0.108       0.0792

N                          14           15           15

Dependent variable:    Main contributor to household income
(when living together):
                       cross-national differences

                        Model 14     Model 15     Model 16

Childcare fees             --           --           --

Childcare enrolment

Family benefits            --           --           --

Unemployment              0.001
benefits                 (0.39)

Social expenditure         --         -0.006         --
                                     (-0.82)
Employment                                         -0.046 *
protection                                        (-2.33)
legislation

Labour market risks      -0.004       -0.008       -0.005

                        (-0.41)      (-0.75)      (-0.87)
Constant                  0.002        0.184        0.147 **
                         (0.02)       (1.14)       (3.10)

Adj. [R.sup.2]           -0.145       -0.098       0.216

N                          15           15           14

Notes: Ordinary least squares regressions, t-values in parentheses.
Significance: *** p < 0.001, ** p < 0.01, * p < 0.05, (*) p < 0.10.

Table 11 Living Together: Determinants of Cross-National Differences

Dependent variable:    Living together: cross-national differences

                        Model 17     Model 18     Model 19

Childcare fees            0.021 *         --           --
                         (2.39)
Childcare enrolment                   -0.014 **
                                     (-3.02)

Family benefits                         --         -0.037(*)
                                                  (-1.71)
Employment
protection
legislation

Active labour market
policy

Standardized ALMP

Labour market risks      -0.027       -0.009       -0.023
                        (_0.80)      (-0.31)      (-0.71)
Greece (dummy            -0.808(*)    -1.321 **    -1.347 **
variable)               (-1.66)      (-2.70)      (-2.61)

Constant                 -0.112        0.644 ***    0.614 **
                        (-0.67)       (4.48)       (2.87)
Wald test,                0.0169       0.0031       0.0546
[chi.sup.2] (3)

N                          14           15           15

Dependent variable:    Living together: cross-national differences

                        Model 20     Model 21     Model 22

Childcare fees             --           --           --

Childcare enrolment

Family benefits            --           --           --

Employment                0.073
protection               (0.76)
legislation

Active labour market                  -0.284 *
policy                               (-2.25)

Standardized ALMP                                  -0.018 *
                                                  (-2.45)
Labour market risks      0.001        -0.005       -0.011
                        (0.03)       (-0.15)      (-0.36)
Greece (dummy           -1.110 *      -0.127 **    -1.278 **
variable)              (-2.24)       (-2.56)      (-2.60)

Constant                 0.103         0.575 ***    0.545 ***
                        (0.47)        (3.73)       (4.07)
Wald test,               0.1589        0.0208       0.0136
[chi.sup.2] (3)

N                          14           15           15

Notes: Feasible weighted least squares regressions, z-values in
parentheses.

Significance: *** p < 0.001, ** p < 0.01, * p < 0.05, (*) p < 0.10.

Table 12 Living together (when Main Contributor to Household
Income): Determinants of Cross--National Differences

Dependent variable:    Living together (when main contributor to
                       household income): cross--national differences

                        Model 23     Model 24     Model 25

Childcare fees             0.002           --           --
                          (0.49)
Childcare enrolment                   --0.003 *
                                     (--2.41)
Family benefits               --                     0.001
                                                    (0.14)
Employment
protection
legislation

Active labour market
policies

Standardized ALMP

Labour market risks      --0.012      --0.011      --0.010
                        (--0.97)     (--1.35)     (--0.84)

Greece (country            0.239(*)   --0.323 **   --0.253(*)
dummy)                  (--2.13)     (--3.64)     (--2.04)

Constant                   0.044        0.161 **     0.069
                          (0.68)       (0.040)      (0.93)
Adj. [R.sup.2]             0.175        0.470        0.192

N                             14           15           15

Dependent variable:    Living together (when main contributor to
                       household income): cross--national differences

                        Model 26     Model 27     Model 28

Childcare fees                --           --           --

Childcare enrolment

Family benefits               --           --           --

Employment               --0.034
protection              (--1.00)
legislation

Active labour market                  --0.034
policies                             (--0.74)

Standardized ALMP                                  --0.002
                                                  (--0.63)
Labour market risks      --0.011      --0.010      --0.011
                        (--1.12)     (--1.05)     (--1.08)

Greece (country          --0.214(*)   --0.286 *    --0.283 *
dummy)                  (--1.93)     (--2.65)     (--2.61)

Constant                   0.145(*)     0.112(*)     0.102 *
                          (1.89)       (2.14)       (2.25)

Adj. [R.sup.2]             0.252        0.229        0.219

N                             14           15           15

Notes: Ordinary least squares regressions, t--values in parentheses.
Significance: *** p < 0.001, ** p < 0.01, * p < 0.05, (*) p < 0.10.
COPYRIGHT 2010 Addleton Academic Publishers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2010 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Emmenegger, Patrick
Publication:Economics, Management, and Financial Markets
Date:Sep 1, 2010
Words:12853
Previous Article:Confidence sets for some partially identified parameters.
Next Article:Management strategies and performance: an examination of minority managers in public organizations.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |