Estudios sobre la Economía Española - 2020/17 The Gender Gap in Involuntary Part-time Employment: The Case of Spain Alfonsa Denia (University of Alicante) María Dolores Guillú (University of Alicante)

fedea

Las opiniones recogidas en este documento son las de sus autores y no coinciden necesariamente con las de FEDEA.

The Gender Gap in Involuntary Parttime Employment: The Case of Spain

Alfonsa Denia alfon@ua.es

MarÌa Dolores GuillÛ! guillo@ua.es

Department of Economics and IUDESP University of Alicante

Abstract

The high incidence of non-desired part-time jobs and temporary contracts after the Great Recession has become one of the most important drivers of the outstanding rise in income inequality in Spain during the last decade. We explore the determinants of involuntary part-time work in Spain over the period 2006-2014 and ilqd that gender has a large, signlilcant and robust positive e§ect on having that employment status, even after controlling for the type and duration of contracts, type of activity or occupation. A female worker is about 7.4 - 8.3 percent more likely to have a non-desired part-time job than a male worker with the same characteristics. Moreover, working in the Public Administration or having a temporary contract increases this probability over 10 percentage points. The results highlight the per-sistent precauriousness of the employment recovery in Spain and the need of a careful reáection on the next labor market reform.

JEL classiÖcation: C10, C25, J10, J20, J70

Key words: Gender, Involuntary part-time, Temporary contracts, non-standard employment, Great Recession.

! Correspondence author. Postal Address: Departament of Economics, University ofăDolfdqwh/ăFdpsxvăVdqăYlfhqwh/ăDs1ăFruuhrvă<</ăH0363;3/ăVsdlq1ăIlqdqfldoăvxssruwăiurp wkhă Vsdqlvkă Plqlvwhulră ghă Hfrqrpðdă |ă Frpshwlwlylgdgă wkurxjkă wkhă uhvhdufkă surmhfwv HFR534607644<0S/ăHFR53480:3730SădqgăHFRă53490::5330Sălvădfnqrzohgjhg

SxeolvkhgălqăLqwhuqdwlrqdoăMrxuqdoăriăExvlqhvvădqgăVrfldoăVflhqfh/ăGhfhpehuă534<

1 Introduction

The number of part-time employees that would work full-time given the opportunity to do so has been rising in Spain at the average rate of 7.7 percent per year since 2007. The rise of this type of employment is a widespread e§ect of the recent crisis across developed countries, but it has been especially severe in Spain, where by the end of 2017 still represented 61 percent of total part-time employment, more than double of the European average.¹ According to Horemans and Marx (2013) involuntary part-time work carries the highest poverty risk across the EU-15 countries, showing that most of the cross country variation comes from di§erences in demand side risk factors like low pay and temporary contracts. In the case of Spain, Felgueroso (2018) and Felgueroso et al. (2017) Önd a high incidence of nondesired part-time jobs and temporary contracts on the risk of employment poverty in 2016 and 2017, and Goerlich et al. (2016) show that the rise in part-time jobs and temporary contracts in Spain are among the most important drivers of the outstanding rise in income inequality since 2009. In this study we assess the determinants of involuntary part-time work in Spain over the period 2006-2014 paying special attention to the type and duration of contracts.

Existing literature on the characteristics and behavior of part-time workers is scarce and limited in scope. Buddelmeyer et al. (2004) analyze the determinants of part-time employment in Europe over the 1980s and 1990s, and conclude that although the majority of employees worked part-time voluntarily during the economic downturn of the 1990s, some policies designed to promote part-time work by lowering its labor costs relative to full-time are likely to have the perverse e§ect of increasing further the proportion of involuntary part-time employment. Sandor (2011) alerts about the dramatic increase of part-time employment during the last recession and the need to improve the quality of part-time jobs, highlighting the importance of this type of employment as a means to increase the áexibility of labor markets in Europe. Valletta et al. (2019) explore the determinants of involuntary part-time employement in the U.S. accounting for business cycle e§ects and structural factors and Önd that shifts in the industry composition of employment have held the incidence of involuntary part-time employment slightly more than one percentage point above its pre-recession level. Green and Livanos (2015b) emphasize the need to look at the risk of what they call involuntary non-standard (involuntary part-time and involuntary temporary) employment, especially during economic recessions. Using microdata from the European Union Labour Force Survey (EU LFS) they estimate the risk of involuntary non-standard employment over the period 2006-2010 and Önd that Spain stands out as the country with the highest rate (24.8%) followed by Poland (20.9%), Portugal (20.6%) and Italy (15%). The analysis reveals that young workers, older workers, women, non-nationals, those with low education and those who were unemployed a year ago are at greatest risk of involuntary non-standard employment, but the estimated average marginal gender e§ect (3.8%) is about the same order of magnitude as the old age or education e§ects. In this paper we use microdata from the Spanish Labor Force Survey and Önd that, once we control for the type of contract, the age and education e§ects are non-signiÖcant or minor determinants of involuntary part-time employment, whereas the gender e§ect (8.3%) appears among the largest e§ects together with the temporary contract e§ect (9.4%) and the elementary occupation e§ect (10%).²

¹ The involuntary part-time employment as percentage of total part-time employment in the EU-28 countries and in the Euro-area are 26.4 percent and 29.2 percent in 2017, respectively (Eurostat 2018).

Despite the limited research literature on involuntary partime employment, there is a rising concern about its e§ect on labor productivity and income inequality. According to the OECD Employment Outlook 2018, the persistence of high levels of involuntary employment in Spain despite the continuous improvements in employment rates since 2013 is one of the key factors behind the decline in the Spanish real wage, which in addition to the high incidence of short-term contracts puts the degree of labor security of Spanish workers among the lowest across OECD countries. Goerlich et al. (2016) show that about 75 percent of the increase in income inequality since 2009 is driven by the drop in householdsí working hours, which are implied by the poor performance of unemployment and the rise of temporary contracts and part-time jobs. The concern about the high income inequalty observed in Spain compared with other European countries is also the focus of Gradin (2016), who Önds that this fact is associated largely with the inequality among households that participate in the labour market.

² The deÖnition of involuntary part-time workers used in EUROSTAT and in the Spanish Labour Force Survey refers to workers that declare themselves working part-time because they could not Önd a full-time job. In Felgueroso (2018) and Felgueroso et al. (2017) the deÖnition is extended to include other non-economic reasons like care and other personal and family obligations. In the U.S. Bureau of Labor Statistics data used by Valletta et al. (2019), involuntary part-time workers are those that have a part-time job for ìeconomic reasonsî as opposed to ìnon-economic reasonsî.

The rest of the paper is organized as follows. Section 2 puts part-time work in context and highlights some of its characteristics in Spain, Section 3 presents the econometric analysis and the main estimation results, Section 4 concludes. Finally, some descriptive statistics and other econometric results are relegated to the Appendix.

2 Putting part-time work in context

The development of part-time employment is a feature of a large number of industrial countries since the mid 1980s and shows considerable variation by gender, age, economic activity and occupation³. Buddelmeyer et al. (2008) Önd that institutions and other structural factors like changes in legislation are the main drivers of this development in Europe during the 1980s and 1990s, and conclude that the negative and signiÖcative e§ect of the economic cycle explains at most 17% of the total increase in part-time employment over the period 1992-1999. In a previous work Buddelmeyer et al. (2004) show that over that period part-time work is predominantly voluntary and that the economic cycle a§ects mostly young and male primeage workers (those aged 25-49), being the e§ect unclear for women and older workers. With respect to more recent periods, to our knowledge, there is no a comparable macro-perspective study.

Table 1 reports involuntary part-time rates for some European countries over the period 2006-2016. Clearly, the rise of involuntary part-time employment is a feature of Mediterranean countries, where the rates in 2016 stand well above their pre-crisis levels, specially in Spain where the rate practically doubles the pre-crisis level. Although in 2008, there were already very high rates in some countries, it is very worrying the dramatic increase in cases like Ireland, Italy, Spain, or even the Netherlands, which contrasts with the decreasing levels of Belgium or Germany. Here the involuntary part-time employment refers to the employment status of part-time workers that declare themselfs working part-time because they could not Önd a full-time job, which is the standard deÖnition used in Eurostat and in the Spanish labor Force Survey. For illustration purposes, other reasons for which workers declare working part-time are shown in the Örst column of Table 2, where columns 2-3 and 3-4 illustrate, respectively, the weight of each reason across genders in Spain for the years 2008 and 2014.

³ See for example OECD (2010).

Table 1: Involuntary part-time employment as percentage of total part-time employment in Europe, Eurostat 2016

It is clear from Table 2 that reasons for working part-time other than ëfull-time not foundí (involuntary part-time) have, in general, lost weight (except for a slight increase in menís caring and family obligations) and that involuntary part-time has experienced a sharp increase over the period, specially for men. The strong association between involuntary part-time employment and unemployment rates over the recession period can be observed in Figure 1 for men and women, separately. Note also that the weak employment recovery starting in 2012 does not translate into an improvement of involuntary part-time employment rates, what probably reáects part of the precarious employment creation taking place in Spain over these years. This feature seems again more severe for men than for women.

Table 2: Main Reasons for working part-time, aggregate shares

Figure 1: Involuntary Part-time and Unemployment

Another feature of the Spanish labor market that characterizes the employment recovery is the increasing temporality of contracts, after some years of contention.⁴ There is a strong association between involuntary parttime employment and temporary contracts, a§ecting specially male workers. Comparing the male and female cases, Table 3 shows not only higher rates among the male population, but also a sharper and steady increase since 2006 that puts the male rate about 20 percentage points higher than the female rate in 2014:

⁴ See, for instance, Malo (2015).

Table 3: Part-time employment and temporary contracts, Involuntary Parttime employment and temporary contracts, Spain, EPA

From a micro-perspective, Green and Livanos (2015b) estimate the risk of involuntary non-standard employment in 10 European countries over the period 2006-2010, where non-standard employment includes part-time jobs and temporary jobs. The analysis reveals that young workers, older workers, women, non-nationals, those with low education and those who were unemployed a year ago are at greatest risk of involuntary non-standard employment. In this study Spain stands out as the country with the highest involuntary non-standard employment rate (24.8%) followed by Poland (20.9%), Portugal (20.6%) and Italy (15%).

Recent studies on the Spanish economy reveal the close connection between the rise in non-desired part-time jobs and the higher risk of su§ering poverty from employment (Felgueroso, Mill·n and Torres, 2017; Felgueroso, 2018), or the dramatic increase in income inequality (Goerlich et al. 2018), putting again part-time employment in the policy debate. Despite the advances in the legislation to eliminate discrimination of part-time relative to full-time jobs, it is clear that they are not enough. The great recession has accentuated the prevalence of the negative aspects that still characterize part-time jobs like low wages, low quality, temporary contracts and limited beneÖts that make this type of employment undesirable from the workersí point of view.⁵

3 Econometric Analysis

The aim is to analyze the socioeconomic proÖle of an involuntary part-time worker in Spain over the period 2006-2014, trying to identify and quantify possible gaps in terms of gender, age or other personal characteristics, controlling for the type and duration of contracts and other employment characteristics. With this aim, we also introduce year-time variables to capture possible time speciÖc e§ects and regional dummy variables to capture possible speciÖc territorial and institutional e§ects.⁶

We use microdata from the Spanish labor Force Survey (EPA for the initials in Spanish) for the years 2006, 2008, 2010, 2012 and 2014. The EPA is a quarterly household sample survey that collects data from more than sixty thousand households. We use the standard two-step modelling estimation method based on Heckman (1979), as in Green and Livanos (2015a, 2015b). This method is adopted when the endogenous variable of interest (i.e., involuntary part-time employment) is only observable for a selected sample (having an involuntary part time job requires Örst that the individual decides to be part of the labor market). In order to present a more easily interpretable measure of the estimation results, we report the marginal effects of the two-step equation. In the Örst stage (participation regression), the control variables are six school-age intervals of dependent children and Öve individualís education categories. The estimation results are shown in Table A5 in the Appendix. In the second stage, the focus is on salaried workers aged between 16 and 64.

The dependent variable in the second stage is the involuntary part-time employment status as described in the previous section. The classiÖcation of jobs by type of economic activity or by type of occupation cannot be considered simultaneously in the same regression due to multicollinearity. We consider them separately in Tables 4 and 5, respectively.

⁵ See for example OECD (2010) for an overview of the positive and negative aspects of part-time employment, and Ramos at al. (2015) for an anlaysis of the wage di§erential betwen part-time and full-time jobs in Spain.

⁶ We have also estimated the models controlling for the regional unemployment rate instead of dummy regional variables and found very similar results.

The individual and family controls are the standard in the literature (the presence of children has a negligible e§ect on the rest of variables in this stage and so it is not considered), they include marital status, sex, person of reference in the household, four age categories (more than sixty years old is the reference), and Öve education levels (university degree is the reference). The market variables are the type of employer (public or private), the type of contract (temporary or permanent), and the duration of temporary or Öxed-term contracts (more than a month/temporary1 or less than a month/temporary2).

The economic activity is classiÖed in Öve categories (the primary sector is the reference, see Table 4) and the occupations are classiÖed in six groups (managers are the reference, see Table 5). Comparing the estimated coe¢cients of the rest of variables in Tables 4 and 5, it is clear that the estimations yield very close results.

First of all, the individual characteristics show that women, young people and low education level individuals are more likely to have an involuntary part-time employment. In particular, the gender e§ect is robust to all model speciÖcations. It varies between 7.4 percent and 8.6 percent, being stronger when jobs are classiÖed according to occupations (Table 5). The results also conÖrm that the probability of having an involuntary part-time job decreases with the individualís age and education level, but also that these characteristics lose ináuence when introducing the type of contract. Moreover, comparing Models 2-3 (Table 4) with Models 5-6 (Table 5), it follows that the youth e§ect (individualís age between 16 and 24) is the only age e§ect that persists when jobs are classiÖed according to occupations. Having a temporary or Öxed-term contract increases the probability of involuntary part-time employment around 10 percent (compared to permanent or openended contracts) in all model especiÖcations. However, when we consider the duration of temporary contracts, those that last for less than a month (temporary 2) increase this probability between 4 and 2 percentage points. This feature, in addition to the involuntary workday, points to the precariousness of this type of employment.

Table 4: Involuntary part-time employment, jobs classiÖed by activity

Standard errors in parentheses. *p-v<0.001; **p-v<0.1

Table 5: Involuntary part-time employment, jobs classiÖed by occupation

Standard errors in parentheses. *p-v<0.001; **p-v<0.1

Working in the public sector (compared to the private sector) is always statistically signiÖcant, it decreases the probability of involuntary part-time employment between 8 (Table 4) and 5 (Table 5) percentage points. In contrast, working in the Public Administration (compared to agriculture) becomes signiÖcant only when controlling by the type and duration of contracts (Models 2 and 3 in Table 4). In those cases, the probability of having an involuntary part-time job increases around 11 percentage points, the same as in Services. The fact that all economic activities have positive and statistically signiÖcant coe¢cients shows that involuntary part-time employment is not speciÖc to a given activity.

In contrast, when we classify jobs according to occupations and control by the type and duration of contracts, there is a polarization of results (Models 5 and 6 in Table 5). The intermediate skilled (white collar and blue collar) occupations are not statistically signiÖcant, the highest positive incidence relays on elementary occupations (10 percent), followed by low skilled (3.8 percent) and high skilled (1.5 percent) occupations. In other words, elementary and low skilled occupations have a lower chance to be voluntary (full-time or part-time).

Finally, all the coe¢cients of the year-time variables in all models considered are statistically signiÖcant, positive and increasing over time (Tables 4 and 5). That is, compared to 2006, everything else the same, workers have incresing chances of having an involuntary part-time employment over time. In particular, the probability of having an involuntary part-time employment in 2014 is 7.5 percentage points higher than in 2006, whereas in 2010 it was only about 4 points higher. This is consistent with recent studies that point to the precarious employment recovery as one of the drivers of the increasing labor insecurity, poverty, and inequality in Spain (OECD Employment Outlook 2018; Felgueroso et al. 2017; Felgueroso, 2018; Goerlich et al. 2018). In these studies the increasing precariousness of jobs is also linked to the rise of Öxed-term contracts since 2012. In order to capture this positive trend in the use of temporary contracts, we extend the analysis by introducing the interactions of the year-time and contract type variables (see Table 6) and Önd that the above results are robust.

Table 6: Involuntary part-time employment, Temporality and year interactions. Models (2bis)-(3bis) Activity, Models (5bis)-(6bis) Occupation

Standard errors in parentheses. *p-v<0.001. **p-v<0.1

The results in Tables 4 and 5 show that the individual e§ects of the year and contract type variables are statiscally signiÖcant separatedly, without interactions. The extended models of Table 6 show that these results still hold and that, in addition, the marginal e§ects of the combined variables are also statistically signiÖcant. Moreover, the coe¢cients of the year variables and the combined variables are also increasing over time in this case. For instance, having a temporary contract increases the probability of involuntary part-time employment in 5.4 percentage points in 2006 (Model 2bis), this e§ect becomes 7.5 percentage points in 2008, and more than 20 percent in 2014. But if the duration of the contract is less than a month (temporary 2 contract), these e§ects become 6.9, 9.2 and 31.3 percentage points, respectively!

Furthermore, comparing the overall e§ect of having a temporary contract with and without interaction variables, uncovers a very interesting feature of the data. The overall e§ects of having a temporary contract in 2008 and 2010 are lower in the models with interaction variables (Table 6) than without interactions (Tables 4 and 5). In contrast, after the Great Recession, for the years 2012 and 2014, these e§ects become stronger in the models with the interaction variables. For instance, if the duration of the contract is less than a month, the contract type e§ect is ampliÖed about 10 points after 2012. The estimates for the rest of variables considered in the study are not shown because they are very similar to those reported in Tables 4 and 5, respectively.

To close this section, we consider an alternative measure of involuntary employment that captures the willingness of employees to work more hours as described in Table A4 in the Appendix. The new dependent variable can be seen as an indicator of a non-desired workday for all types of workers, part-time and full time workers. Both measures, involuntary part-time employment and willingness to work more hours, capture the preference of workers to have longer workdays. SpeciÖcally, we repeat the estimations of models in Tables 4 and 5 but with a new dependent variable, willingness to work more hours. In this case, the female gender e§ect falls to 4 percent and the youth e§ect increases until 5 percent. Nonetheless, the probability of involuntary employment for a young female worker remains the same as in previous models. For the rest of variables the estimation results are very similar to those obtained before, but with much larger coe¢cients, except for the economic activity categories that now are less ináuential. The presentation of these estimation results are relegated to the Appendix (see Tables A6 and A7).

4 Conclusion

The main objective of this study has been to explore the determinants of the rise in involuntary part-time employment in Spain over the period 2006- 2014, before and after the Great Recession. We have used microdata from the Spanish Labor Force Survey and employed the two-step modelling estimation of Heckman (1979), as in Green and Livanos (2015a). We have provided evidence that gender had a large, signiÖcant and robust positive e§ect on involuntary part-time employment in Spain over the period, whereas the workerís age and other personal charactersitics were much less ináuential. Our empirical strategy has controlled for regional and year e§ects, and paid special attention to the type and duration of contracts.

We have shown that over the period 2006-2014 a female worker was about eight percent more likely than a male worker to have a non-desired part-time job, whereas a young worker (aged 16-24) was only about one percent more likely than an old one (aged more than 60). Other age groups were not statistically signiÖcant once the estimation controlled for temporary contracts. With respect to the type of activity, working in the Public Administration increased the probability of involuntary part-time employment by more than ten percent óthe same amount as working in Servicesó, what alerts about the role of governments (specially local governments) in the creation of non-standard employment. With respect to the type of occupation/skill, elementary and low skilled jobs had the lowest chances of being voluntary, followed by high skilled jobs, whereas intermediate levels as blue collar and white collar occupations (with negative and positive coe¢cients, respectively) were not statistically signiÖcant.

We have also shown that temporary contracts increased the probability of involuntary employment betwen 10-14 percentage points, and that the year e§ect was statiscally signiÖcant, positive and increasing over the period. Moreover, taken into account the combined e§ect of the year and contract type variables, we have found that these individual e§ects were reinforced after 2012, year of the last labor market reform and of the employment recovery in Spain. In particular, having a very short-term contract makes involuntary part-time employment between 22 - 12.8 percent more likely in 2014 than in 2008 (compared to 2006), depending on whether jobs are classiÖed by type of activity or by occupation. We conclude that the gender e§ect appears among the largest determinats of involuntary part-time employment together with the temporary contract e§ect and the elementary occupation e§ect.

Given the dramatic e§ects of non-standard employment on poverty and income inequality in Spain (e.g., Felgueroso, 2018; Felgueroso et al. 2017; Goerlich et al. 2016), our results suggest that the design of a new labor market reform that eradicates the indiscriminate use of short-term contracts and prevents the abuse of (non-desired) part-time work-weeks among female workers are a priority. Moreover, the relevance of involuntary part-time employment in the Public Administration and the spread of temporary employment require a careful analysis of the regulation and provision of vacancies in the public sector.⁷

References

1. Buddelmeyer, H., G. Mourre and M. Ward (2004), ëRecent Developments in Part-Time Work in EU-15 Countries: Trends and Policyí, IZA DP No. 1415.
2. Buddelmeyer, H., G. Mourre and M. Ward (2008), ëWhy Do Europeans Work Part-time?í, European Central Bank Working Paper Series No. 872, February.
3. Felgueroso, F. (2018), ëPoblaciÛn especialmente vulnerable ante el empleo

⁷ See Malo (2015) for a comprehensive analysis of the Spanish labour market over the period 2008-2013 and a discussion of the measures and timing of di§erent labour markets reforms.

1. en EspaÒa en el aÒo 2018í, Estudios sobre la EconomÌa EspaÒola - 2018/11, FEDEA.
2. Felgueroso, F., A. Mill·n, and M. Torres (2017), ëPoblaciÛn especialmente vulnerable ante el empleo en EspaÒa. CaracterizaciÛn y cuantiÖ- caciÛní, Estudios sobre la EconomÌa EspaÒola -2017/07, FEDEA.
3. Goerlich, F.J., V. Cucarella, L. Hern·ndez, H. GarcÌa and I. Zaera (2016), ëDistribuciÛn de la Renta, Crisis EconÛmica y PolÌticas Redistributivasí, FundaciÛn BBVA.
4. Green, A. E.and I. Livanos (2015a), ëInvoluntary Non-Standard Employment and the Economic Crisis: Regional Insights from the UKí, Regional Studies, Vol. 49, No. 7.
5. Green, A. E.and I. Livanos (2015b), ëInvoluntary Non-Standard Employment in Europeí, European Urban and Regional Studies, pp 1-18.
6. Heckman, J. (1979), ëSample selection bias as a speciÖcation errorí, Econometrica, 47, 153-161.
7. Horemans, J. and I. Marx (2013), ëIn-work poverty in times of crisis: Do part-timers fare worse?í, ImPRovE Discussion Paper No. 13/14, Antwerp.
8. Malo, M. A. (2015), ëLabour market measures in Spain 2008-2013: the crisis and beyondí, International Labour O¢ce, Research Department. Geneva: ILO.
9. OECD (2010), Employment Outlook 2010, ch. 4 How good is part-time work?, OECD Publishing, Paris.
10. OECD (2018), Employment Outlook 2018, ch. 1 Still out of pocket: Recent labour market performance and wage developments, OECD Publishing, Paris.
11. Ramos, R., E. Sanrom· and H. SimÛn (2015), ëAn Analysis of Wage Di§erentials between Full- and Part-Time Workers in Spainí, IZA DP No. 9257, August.
12. Sandor, E. (2011). Part-time work in Europe, Eurofound 2011.
13. Valletta, Robert G., L. Bengali and C. van der List (2019), ëCyclical and market determinants of involuntary part-time employmentí, Journal of Labor Economics 0, no. 0 (-not available-): 000. https//doi.org/10.1086/704496.

Appendix

Table A1. Decriptive Statistics. EPA (2006-2014)

Table A1 (Cont.). Decriptive Statistics. EPA (2006-2014)

Table A2. Total employment by type of workday: Part-Time vs Full Time (%)

Table A3. Total employment by type of contract: Temporary vs Permanent (%)

Table A4. Total employment by type of workday: Willingness to work more hours (%)

Table A5. Participation equation

Standard errors in parentheses. *p-v<0.001. **p-v<0.1

Table A6. Willingness to work more hours, by economic activity

Standard errors in parentheses. *p-v<0.001; **p-v<0.1

Table A7. Willingness to work more hours, by occupation

Standard errors in parentheses. *p-v<0.001; **p-v<0.1