Employment Status and Health in Spain Before and After the Great Recession

Introduction: Employment Status, Recession, and Health

The most-discussed dimension of the labor market in the field of sociology of health is unquestionably unemployment and involuntary job loss. The in-depth report by Jahoda, Lazarsfeld, and Zeisel at the beginning of the 1930s, on the social and psychological effects of closing a textile factory around which life had revolved in the Austrian community of Marienthal, is a classic point of reference. It gained particular influence in the 1970s (when it was published in the United States), and this influence increased significantly during the 1980s. At that time, there was renewed interest in the study by Jahoda and colleagues as a consequence of a crisis giving rise to the highest levels of unemployment seen in industrialized economies since the Second World War. Understandably, there were new studies of factory closures (Kasl, Gore, and Cobb 1975), which established a causal link between loss of employment and deterioration in health. The empirical evidence gathered since that time links unemployment with various indicators of poor health, illness, psychological distress, and other mental illness conditions (Bugard, Brand, and House 2007; D’Arcy and Siddique 1985; Turner 1995; Warr, Jackson, and Banks 1988).

Thus, the classical theories that have dominated research about labor market and health since the 1970s and 1980s, particularly Jahoda’s (1982) deprivation theory, have focused on the impact of unemployment, and have clearly delineated and influenced the present-day research agenda (Selenko, Batinic, and Paul 2011). This emphasis on unemployment as an etiological factor in the deterioration of health is characteristically accompanied by a biased understanding of the labor market, which is described in terms of a fundamental separation between employment (any employment) and unemployment (any unemployment). Such a conception represents an idyllic view of paid work (Cole 2007) and research has been carried out on this idea for decades. Brendan Burchell (1992) argues that the socio-psychological literature suffers from an incomplete understanding of the nature of the labor market, the fundamental feature of which is not and has never been a homogeneous social employment norm that would clearly divide the labor market into employed and unemployed. Rather, the labor market involves the co-existence of varied and heterogeneous states of employment and unemployment (Dooley 2003).

This limitation is a characteristic of the literature that addresses the impact of economic recessions on health. A large group of studies seek to use added variables to explain the effect of recessions on mortality rates. Studies by Brenner (1979) attempted to demonstrate lagged effects of recessions on mortality, arguing that the increases observed in mortality during prosperous years were in fact produced by recessions occurring years earlier. This hypothesis, commonly accepted until the 1990s, has recently been questioned by Ruhm (2000), whose analyses conclude that in reality mortality is procyclical, that is, health improves and mortality falls when the economy deteriorates, while on the contrary, mortality rises in good economic times. It is important to note that this counterintuitive relationship between recession and health is only documented for the short term. ¹ It would be explained by the impact of the phases of recession on particular intermediate variables related to lifestyle (obesity, smoking, diet, and exercise) and by the loss of importance of processes that are characteristic of times of economic growth (e.g., stress and work-related accidents or the relationship between economic well-being and vehicle fatalities). The most relevant point for the present work is that in the studies articulating this debate (see, among others, Gerdtham and Ruhm 2006; Neumayer 2004; Tapia-Granados 2005), unemployment rates are the proxy for economic conditions, such that the existence of the economic situation (i.e., the existence of a recession) is confirmed through the unemployment rates of the countries that are studied, without any kind of indicator relating to the structure of the labor market. In a similar manner, the development of the recent economic crisis, the Great Recession, has led to studies attempting to evaluate the change in health levels of populations subject to the impact of that crisis by using microdata. The results show deterioration in health and psychological well-being for the population during the period of recession. This deterioration is associated with unemployment (Bartoll et al. 2014; Drydakis 2015), especially long-term unemployment (Urbanos-Garrido and López-Valcárcel 2015). Again, in all of these studies, the labor market is characterized as being separated into employment and unemployment. As such, the debate around the relationship between economic recession and health occurs on the basis of a division between employment and unemployment that, as previously stated, involves an incomplete understanding of the labor market and of the concept of employment status (ES). In other words, the debate is framed implicitly assuming that the employment link between economic crisis and health arises through unemployment, ignoring the complexity that is characteristic of contemporary labor markets.

This study seeks to incorporate this complex nature of ES into the debate on the relationship of recessions with health, in this case considering the economic crisis that began in 2008—the Great Recession. The contribution, therefore, is not a (re)evaluation of the impact of ES on health indicators, but rather the analysis of possible variations in the association between ES and health according to the labor context and economic cycle. This objective is based on two fundamental elements. First, research into health implies an analysis of the differences between situations of employment and unemployment, but also (and necessarily) a consideration of the complexity of ES and its impact on well-being indicators. This is seen in numerous studies from the 1990s onward concerning the effect of deleterious characteristics of atypical forms of employment on health. In particular, a negative impact has been found on physical and mental health resulting from situations of job insecurity (Ferrie et al. 1995) and especially temporary and fixed-term employment (Benach et al. 2004; P. Virtanen, Vahtera, et al. 2005; Vives et al. 2013), the latter even being linked with greater mortality rates for general and specific reasons (Khlat et al. 2014). Taken as a whole, the available evidence suggests the existence of notable differences among different types of employment, which are specifically related with health outcomes.

Second, this view of ES is especially important to properly understand the consequences of labour inequality for health distribution patterns within a context of analysis of historical-social contingencies. It could be argued that the relationship between ES and health will not be immutable over time, but rather specifically reflects the prevailing features in a particular labor market at a specific point in the economic cycle. The surrounding hypothesis is that this relationship will fluctuate according to the changing meaning of the different ESs at different points in that cycle. Certainly, previous studies pay attention to contextual influences for the association between ES and health. For example, unemployment appears to have less impact on health in contexts of high unemployment (Dooley 2003; Martikainen and Valkonen 1996), while high morbidity tends to be found among temporary workers in the context of labor markets characterized by low levels of temporary work and unemployment (M. Virtanen, Kivimäki, et al. 2005; P. Virtanen, Vahtera, et al. 2005). However, these studies were not designed to compare two clearly different economic points between which significant changes have occurred in the labor market, and they are only examples of the association between ES and health in specific macroeconomic and labor contexts. One strategy to overcome these limitations consists of comparing the pattern and structure of the association between ES and health at two (or more) clearly differentiated economic points, whether as a consequence of long-term processes of change or radical shifts in the dynamics of an economy.

The last decade has been dominated by the impact of the Great Recession, which shook the foundations of the economic system worldwide from 2008, creating just such a context of change. Given that the consequences of the Great Recession for economic systems and labor markets are specific in nature, varying notably at a national level (O’Reilly et al. 2011), one can hypothesize that the manner in which the crisis brings about an adaptation of the labor markets to the pressures and/or needs arising out of economic processes will determine the intensity and features of the impact different ES situations have on health and well-being for the various groups comprising the labor market. To address this topic, this research analyzes and compares data from representative samples taken at two particularly significant moments in time and in a specific social context, that of Spain in 2006 (at the zenith of the period of economic expansion begun in the 1990s) and in 2012 (at the peak of the consequences of the 2008 global crisis). It is important to note that Spain was one of the countries most negatively affected by the crisis, particularly in employment terms. The following paragraphs are devoted to analyzing the specific economic and labor context that frames this research. This description, together with the evidence and arguments analyzed in this section, will allow the hypothesis to be developed for the study.

Present Research: Hypotheses on the Relationship Between Employment Status and Health During the Great Recession in Spain

Various writers have linked the characteristics of Spanish economy with the variant of capitalism referred to as the Mediterranean model (see Amable 2003; Lallement 2011). This model is characterized by the existence of specific capacities for nonmarket coordination in the sphere of corporate finance, but more liberal arrangements; relatively weak education and training systems, related to difficulties in developing a high value-added production process; labor-intensive growth during growth cycles; and a key role for the banks in corporate financing. In this context, the main characteristic of the labor markets in the Mediterranean model is the existence of strong labor market segmentation (LMS) based on a dual flexibility that generates strong protection for the core labor force and precarity for a segment affecting a growing number of workers. In Spain, LMS can be understood as a result of the impact of a two-tier deregulation policy, originating from the labor reform of 1984. This reform was based on expanding the use of temporary contracts and noncasual fixed-term contracts as a measure to boost employment. It eliminated existing barriers to the use of temporary contracts, which had previously only been valid for seasonal economic activities. Temporary contracts increased vertiginously, establishing a flexible peripheral segment equivalent to 30 percent of employed persons. This segment began to account for the majority of shifts between employment and unemployment. The process was one of growing differentiation, so that “as the flexible segment of temporary work increased in size, it also increased in internal instability, while the opposite phenomenon is true for the core of permanent employment, which became smaller, more secure and more impenetrable” (Polavieja 2003:510). This period of deregulation and segmentation appears to have reached a high-segmentation equilibrium during the early 1990s, described as the prevalence of flexibility at the margin (Davia and Hernanz 2004), that is, the existence of a clearly differentiated peripheral segment within which the volatility of the labor market was concentrated. The importance of temporariness as a segmentation mechanism has been consolidated as the defining feature of the Spanish labor market, ² showing a marked resistance to change. In this regard, the reforms of 1997, 2001, and 2006 had the explicit objective of reducing the share of fixed-term contracts through the introduction of a permanent employment promotion contract. Yet these reforms did not lead to significant changes in the incidence of temporariness in the medium or long term, suggesting the existence of a steady-state share of fixed-term contracts in Spain (Sala, Silva, and Toledo 2009), to the extent that in 2007 a third of employees had temporary contracts. Furthermore, this process took place within the context of relatively high unemployment and was associated with significant salary discrimination affecting workers in the peripheral segment.

To this must be added the growth in self-employed workers, ³ including a high but difficult-to-quantify proportion of bogus and/or dependent self-employed ⁴ workers. Carrasco and Ejrnaes (2012) showed that in Spain, individuals in the lowest part of the wage distribution spectrum and those with lower education/qualifications are often involved in transitions from employment toward self-employment. Similarly, there are a notable number of movements out of unemployment, especially in the case of individuals not receiving unemployment benefits (between 1994 and 1999, the transition rate from unemployment was around 17 percent). It is significant that, for those entering self-employment from a situation of unemployment, the probability of the experience ending in a return to unemployment is higher than the probability of obtaining wage employment. All the foregoing suggests that in Spain, self-employment has constituted an important alternative to wage employment for individuals who may be considered as disadvantaged or vulnerable in the labor market since the middle of the 1980s (Carrasco 1999). This process intensified during the precrisis years of economic growth, generating a stock of marginal own-account workers that may become too large during recessionary phases of the economic cycle (Congregado, Golpe, and Carmona 2010).

In general, one may affirm that in the years prior to the crisis, Spain had a peripheral labor segment that was far more developed than in the majority of the other European economies. Within this context, Lallement (2011:632) argues that the adjustment in the employment environment in Spain during the first years of the 2008 recession took the form of “increased labour market segmentation,” that is, a deepening of flexibility at the margin, emphasizing both the protection of the core labor force and the increasing vulnerability of the peripheral segment. The data suggest that this interpretation is correct. From 2008, there was a vertiginous increase in the unemployment rate, which reached historic heights by rising beyond 25 percent in 2012 (8.5 percent in 2008). In the context of the European Union, “the net destruction of fixed-term employment was the result of developments in two sectors (manufacturing and construction) and one country (Spain). If we exclude Spain, the category of fixed-term employment added nearly 250,000 jobs across the EU during the crisis” (Hurley, Storrie, and Jungblut 2011:47). In fact, of the jobs lost in 2008 and 2009, 90 percent were temporary. This demonstrates that the speed of change in the employment rate in Spain was far higher than in the majority of European countries, reflecting the high flexibility characterizing the labor market (Lallement, 2011). In addition, there was a significant reduction in the number of self-employed workers and a change in their composition. On this point, previous research (Román, Congregado, and Millán 2011) has shown that there is a recession-push process in Spain pursuant to which economic crises generate a recomposition of the self-employed group, with an increase in transitions from unemployment and paid employment to bogus and dependent self-employment. Despite all this, the proportion of temporary workers in 2012 remained at 23.6 percent and practically all of the few new jobs created have been temporary. During the same period, the long-term unemployment rate (people unemployed for a year or longer) increased sixfold, from 1.8 to 11.1 percent. These figures, in summary, show that the costs and consequences of the crisis were concentrated in the most vulnerable labor market segment.

This analysis of the Spanish labor market and of the impact of the Great Recession on its dynamics—considered within the previously discussed framework of the research as related in the preceding section—permits the formulation of an interrelated set of hypotheses. As previously stated, the recent literature suggests a procyclical effect of unemployment on short-term health. If we also take into account that the availability of the greatest unemployment benefits in Spain is concentrated within the first year of unemployment, a reduced deterioration in health associated with short-term unemployment is expected when comparing pre- and postcrisis samples (Hypothesis 1a [H1a]). In addition, the long duration of the crisis in Spain and its profound impact on the labor market have meant that the longer the situation of unemployment, the smaller the possibility of returning to employment. The deterioration in prevailing material/economic conditions resulting from the progressive loss of unemployment benefits must be added to this. Therefore, as conditions of material and psychological deprivation accumulate, one may expect that the association between long-term unemployment and (worse) health will intensify (Hypothesis 1b [H1b]).

The self-employed became clearly peripheral through the slump in economic activity during the crisis, this activity being a fundamental requirement for the businesses of self-employed workers and professionals. During the impact of economic crises in Spain, the vulnerability characteristic of a large proportion of self-employed individuals must be added to the existence of a recession-push process implying an increase in the proportion of dependent self-employment. As has been noted, both processes affect the composition of the self-employed workforce, notably increasing the insecurity and uncertainty associated with this labor position and adversely affecting their working conditions and income. We may hypothesize that in general, self-employed workers would become one of the most vulnerable labor segments during the crisis, which would have resulted in a fall in health in the 2012 sample (Hypothesis 2 [H2]).

Finally, expected outcomes for persons in temporary employment include a change in the relationship between temporary contracts and health during the crisis, such that the association between these variables is diluted with the impact of the Great Recession (Hypothesis 3 [H3]). This hypothesis derives from the rapidity of adjustments facilitated by LMS in Spain, which in the context of high flexibility at the margin generated changes in the meaning of the different ESs and their interrelationships—a repositioning of the ES, if the expression may be permitted. For this reason, the conditions associated with persons in temporary employment, at a time when the dominant trends are (1) the shift toward unemployment within that group, (2) a drastic reduction in the probability of transitions into employment from unemployment, and (3) a drop in status and increased uncertainty affecting the self-employed workforce, suggest a position of relative advantage for temporary workers during the crisis. In other words, after years of deep crisis with an enormous impact on the labor market, a temporary job would mean—at least in a transitory fashion—having access to the material and psychosocial benefits that characterize permanent employment.

Method

Results and Discussion

Summary statistics for the study variables are presented in Table 1. Figures reflect the changes in workforce composition in Spain due to the impact of the economic crisis.

OPEN IN VIEWER

Table 1.

Summary Statistics Spanish National Health Survey, 2006/2012.

Variables	2006 (n = 15,128)	2012 (n = 11,124)	Observed range
Health outcomes
Self-rated health	3.86 (0.80)	4.00 (0.75)	1–5
Psychological distress (GHQ)	1.29 (2.25)	1.34 (2.37)	0–12
Employment status (vs. permanent)
Temporary	.19	.14
Self-employed	.17	.13
Unemployed −1 year	.07	.11
Unemployed +1 year	.04	.12
Female	.52	.46
Age	40.67 (11.06)	42.12 (10.91)	16–70
Urbanicity (vs. + 500,000)
100,000–500,000	.29	.29
10,000–100,000	.36	.36
≤10,000	.24	.21
Married	.58	.55
Spanish	.92	.91
Social class (vs. Class I. Managers)
Class II. Intermediate professions	.25	.20
Class III. Supervisors/skilled	.19	.14
Class IV. Semiskilled workers	.18	.30
Class V. Nonskilled workers	.16	.13
Education (vs. less than secondary)
Secondary level	.14	.17
High school diploma	.16	.13
Vocational qualifications	.17	.20
University degree	.23	.22

Note. Blank cells represent dummy variables. GHQ = General Health Questionnaire. Mean scores (standard deviations) and proportions.

The main results are set out in Tables 2 and 3. Table 2 presents means and standard deviations for both measures of health used in this study across ESs. Table 3 presents the results from our series of regression models. First, the 5-point self-rated scale is examined for both 2006 and 2012 data, and second, the psychological distress measure (GHQ) is considered. The first model examines the ES effect adjusting for sex, age, urbanicity, and marital status. The second and third models add controls for social class and education, respectively.

OPEN IN VIEWER

Table 2.

Means and Standard Deviations for Self-Rated Health and Psychological Distress (GHQ) by Employment Status.

Year/Variables	Permanent	Temporary	Self-employed	Unemployed	Unemployed	F	Range
				1 year−	1 year+
2006
Self-rated health	3.91a,b,c	3.86 c	3.85c,d	3.79c,d	3.53a,b d,e	38.24***	1–5
	(0.78)	(0.79)	(0.81)	(0.85)	(0.86)
Psychologic aldistress (GHQ)	1,13b,c,e	1.43a-d	1.17b,c,e	1.90a,d,e	2.06a,d,e	52.70***	0–12
	(2.11)	(2.30)	(2.01)	(2.87)	(2.87)
2012
Self-rated health	4.04 c	4.04 c	4.00 c	4.01 c	3.74a,b,d,e,	44.24***	1–5
	(0.72)	(0.72)	(0.74)	(0.77)	(0.87)
Psychologic aldistress (GHQ)	1.00a-c	1.17b,c	1.33b-d	1.84a,c-e	2.43a,b,d,e	116.99***	0–12
	(1.94)	(2.15)	(2.90)	(2.82)	(3.28)

Note. All descriptive statistics are based on weighted data. GHQ = General Health Questionnaire.

a

Different from self-employed.

b

Different from individuals unemployed 1 year−.

c

Different from individuals unemployed 1 year+.

d

Different (p ≤ .05) from individuals in a permanent job.

e

Different from individuals in a temporary job.

***

p ≤ .001.

OPEN IN VIEWER

Table 3.

Unstandardized Coefficients From Ordinary Least Squares Regression Models of Self-Rated Health or Psychological Distress (GHQ) on Employment Status—Spanish National Health Survey 2006 and 2012.

Variables	Self-rated health						Psychological distress (GHQ)
	2006			2012			2006			2012
	1	2	3	1	2	3	1	2	3	1	2	3
Employment status (reference: permanent)
Temporary	−.110*** (.018)	−.063*** (.018)	−.048** (.018)	−.071*** (.021)	−.046* (.021)	−.033 (.021)	.218*** (.051)	.173*** (.052)	.154*** (.052)	.180** (.068)	.148* (.068)	.138* (.069)
Self-employed	−.006 (.018)	−.039* (.018)	−.003 (.019)	−.013 (.022)	−.030 (.022)	−.006 (.022)	.064 (.052)	.094 (.053)	.036 (.054)	.381*** (.068)	.407*** (.069)	.383*** (.070)
Unemployed 1 year−	−.170*** (.027)	−.120*** (.027)	−.104*** (.027)	−.100*** (.023)	−.066** (.024)	−.046 (.024)	.739*** (.076)	.695*** (.077)	.674*** (.077)	.912*** (.074)	.875*** (.075)	.857*** (.076)
Unemployed 1 year+	−.300*** (.034)	−.252*** (.033)	−.225*** (.033)	−.283*** (.023)	−.247*** (.023)	−.218*** (.023)	.787*** (.095)	.744*** (.096)	.708*** (.096)	1.417*** (.072)	1.382*** (.073)	1.360*** (.074)
Female	−.121*** (.013)	−.126*** (.014)	−.139*** (.014)	−.088*** (.014)	−.094*** (.014)	−.105*** (.014)	.538*** (.037)	.529*** (.039)	.548*** (.039)	.403*** (.045)	.401*** (.045)	.411*** (.045)
Age (years)	−.015*** (.001)	−.014*** (.001)	−.013*** (.001)	−.015*** (.001)	−.015*** (.001)	−.014*** (.001)	.010*** (.002)	.009*** (.002)	.008 (.002)	.014*** (.002)	.013*** (.002)	.013*** (.002)
Urbanicity (reference: 500,000+)
100,000–500,000	−.078*** (.023)	−.067** (.023)	−.059** (.023)	−.048* (.024)	−.031 (.024)	−.021 (.024)	−.085 (.066)	−.091 (.066)	−.104 (.066)	−.050 (.075)	−.064 (.075)	−.074 (.075)
10,000–100,000	−.113*** (.023)	−.086*** (.023)	−.066** (.023)	−.069** (.023)	−.042 (.023)	−.025 (.023)	−.013 (.065)	−.032 (.065)	−.061 (.065)	−.023 (.073)	−.044 (.074)	−.056 (.074)
≤10,000	−.108*** (.024)	−.066** (.024)	−.036 (.024)	−.032 (.025)	−.006 (.025)	−.029 (.025)	−.137* (.069)	−.170* (.069)	−.213** (.070)	−.159* (.080)	−.190* (.081)	−.206* (.081)
Married	.003 (.014)	.005 (.014)	.009 (.014)	.038** (.015)	.031* (.015)	.033* (.015)	−.216*** (.039)	−.215*** (.039)	−.222*** (.039)	−.258*** (.047)	−.244*** (.047)	−.250*** (.047)
Spanish	−.010 (.026)	−.049(.026)	−.025(.026)	−.040(.025)	−.035(.026)	−.023(.026)	−.385*** (.074)	−.325*** (.075)	−.354*** (.076)	−.023 (.081)	−.029 (.082)	−0.20 (.083)
Social class (reference: Class I. Managers)
Social Class II. Intermediate professions		−.096 *** (.019)	−.020 (.021)		−.067** (.021)	−.023 (.023)		.125* (.055)	.008 (.060)		−.036 (.068)	−.081 (.072)
Social Class III. Supervisors and skilled workers		−.212*** (.021)	−.076** (.025)		−.139*** (.024)	−.069** (.026)		.133* (.061)	.067 (.070)		.017 (.077)	−.055 (.084)
Social Class IV. Semiskilled workers		−.205*** (.021)	−.072** (.025)		−.161*** (.020)	−.075*** (.023)		.153* (.060)	.045 (.071)		.076** (.064)	.007 (.074)
Social Class V. Nonskilled workers		−.278*** (.023)	−.121*** (.027)		−.199*** (.025)	−.096*** (.028)		.320*** (.065)	.093 (.076)		.247** (.080)	.157 (.090)
Education (reference: less than secondary)
Secondary level			.067*** (.022)			.109*** (.025)			−.014 (.062)			.137 (.082)
High school diploma			.165*** (.021)			.185*** (.030)			−.167** (.061)			.017 (.096)
Vocational qualifications			.091*** (.021)			.168*** (.028)			−.122* (.059)			−.002 (.091)
University degree			.263*** (.023)			.253*** (.031)			−.352*** (.067)			−.069 (.099)
Constant	4.656*** (.040)	4.816*** (.042)	4.508*** (.050)	4.768*** (.042)	4.881*** (.038)	4.597*** (.056)	1.015*** (.113)	.877*** (.119)	1.267*** (.143)	.456*** (.132)	.397** (.140)	.456** (.178)
R2	.055	.068	.076	.064	.071	.078	.034	.036	.038	.055	.056	.057

Note. GHQ = General Health Questionnaire.

*

p ≤ .05. **p ≤ .01. ***p ≤ .001.

Focusing on self-rated health, variations in the regression models reported in Table 3 may be interpreted as a structural change compared with 2006, as the results of the Chow test reflect, F(20, 25773) = 11,899; p < .01. The coefficients (after introducing the social class and education variables, Model 3) for both those unemployed for less than one year and temporary workers are significant for 2006, but not for 2012. These results support H1a (suggesting a reduced deterioration in health associated with short-term unemployment after the beginning of the Great Recession) and H3 (according to which a change in economic conditions suggests a position of relative advantage for temporary workers during the crisis, which in turn would result in the reduction of the negative impact of temporary employment on health). Moreover, in contrast with the results for 2006, only the coefficients for the long-term unemployed are significantly different to the reference category (permanent contract) for 2012, supporting the hypothesis that the association between long-term unemployment and health will intensify during the crisis (H1b). Finally, the coefficients for self-employed show no significant differences for either of the two samples, which is not consistent with H2, which predicted that the increased vulnerability of the labor position of self-employed workers would lead to a fall in health. In fact, average self-rated health scores for permanent workers, temporary workers, the self-employed, and those unemployed for less than one year converge significantly, eroding the 2006 comparative advantage for permanent workers (see Table 2) and reflecting the increasing importance of the duration of unemployment in terms of its impact on self-rated health during the Great Recession.

Turning now to psychological distress, the results shown in Table 3 reflect structural changes for this variable between 2006 and 2012, as the Chow test suggests, F(20, 25217) = 3.838; p < .01. The differences in results for the two years analyzed show that there was a significant increase in psychological distress among those in long-term unemployment, whose gap with respect to the other employment categories widens strikingly in 2012, with average GHQ scores becoming 289 percent greater (103 percent in 2006) than those for the reference category (permanent employment). These results support the statement of an intensified association between long-term unemployment and psychological distress (H1b). In contrast, the fact that the coefficient for short-term employees in 2012 remains significant (moreover, the gap increasing between said group and those in permanent employment) does not support H1a and would suggest that, in the case of mental health, there is no short-term procyclical effect of the recession with relation to unemployment. However, it is necessary to note that there were differences that did not exist in 2006 but were found in 2012 among the unemployed depending on the duration of their unemployment (see Table 2). Meanwhile, the volatile status of the self-employed was reflected in the regression results for mental health following the impact of the crisis and the subsequent rapid response by the Spanish labor market, supporting H2, according to which there would be an increase in deterioration for self-employed workers as a consequence of their vulnerable position in the labor market emerging from the crisis, as previously described. In this regard, the regression coefficient (Table 3) for self-employed persons in 2012 shows that the average GHQ scores became 47 percent greater (no significant differences in 2006) than those for persons in permanent employment. In a similar vein, according to H3 the rapidity of the LMS adjustments in Spain would result in a change in the relationship between temporary employment and mental health during the crisis, such that the relation between these variables would be diluted. However, in 2012 the disadvantage in terms of wellbeing implied by holding a temporary job as against more stable working situations remains statistically significant, which appears not to support H3. However, it is noteworthy that this is the sole category of ES for which the gap is reduced in comparison with permanent employees (β = .154; p < .001, for 2006; β = .138; p < .05, for 2012). Moreover, the association between temporary employment and self-employment, on one hand, and psychological well-being, on the other, became inverted during the crisis (see average scores in Table 2). Taken as a whole, therefore, the results allow us to affirm that our hypothesis (H3) is partially confirmed.

Taken as a whole, the results of this study are especially relevant for the debate about contextual effects on health indicators across ESs. In contrast to the previously discussed studies, the results in our research show that the situation of persons in long-term unemployment is associated with worse self-rated health indicators and an increase in psychological distress within a specific context of high unemployment (2012)—the opposite of that suggested by Martikainen and Valkonen (1996) and Dooley (2003), respectively. Focusing now on temporariness, within a context of high unemployment and high temporariness (2012 sample), temporary employment resulted in lower overall distress and better self-rated health—in contrast to what occurs in a context of low unemployment and high temporariness (2006 sample). These findings partially differ from those obtained by M. Virtanen, Kivimäki, et al. (2005) and P. Virtanen, Vahtera, et al. (2005), which showed greater distress in temporary workers within labor markets featuring a low proportion of temporary employment and low unemployment rates.

Two reasons can explain these divergences. First, the studies mentioned do not compare data from the same society at two points in time that exhibit such a radical economic change as the one here analyzed. ⁵ In this regard, our results fit with those obtained in studies specifically addressing the relationship between economic recession and health/mortality (Ruhm 2000), finding a short-term procyclical effect between unemployment and health. This interpretation may be extended to (1) the regression coefficients on self-rated health for the short-term unemployed, which are not significantly different to the reference category for 2012, and (2) the appearance of statistically significant differences in psychological deterioration among short-term and long-term employees in the 2012 sample (Table 2). In addition, the sharp increase in inequalities in psychological deterioration related to long-term unemployment, reflected in the regression coefficients in Table 3, suggests the existence of a long-term countercyclical effect specifically for this variable.

Second, the present article claims that the study of the relationship between economic cycle and health should be founded on a comprehensive analysis of the specific labor market under study—beyond the employment/unemployment dichotomy. In the case of the Great Recession and Spain, we argue that the changes in the distribution of health according to ES reflect the particular effect of the crisis in redefining the meanings of the various categories of this variable in the labor market. In line with the description of the Mediterranean variety of capitalism, the Spanish economic system is characterized by LMS that specifically revolves around flexibility in the margin based on a high level of unemployment, and particularly the role of temporary employment and the existence of a high proportion of vulnerable self-employed persons. It could be argued that this flexibility in the margin defined the response of the labor market to the Great Recession (Lallement 2011), which in turn shapes the health differences emerging with the outbreak of crisis in 2008 and is reflected in the results of our study. In Table 3, this argument is seen in the changes relating to temporary employees and the self-employed (in addition to the previously discussed changes for the unemployed). For the former, the disadvantage existing in 2006 in comparison to permanent employees (1) disappears in 2012 for self-rated health and (2) is the sole disadvantage that is reduced in the case of psychological distress. This result is consistent with the foundation of H3, according to which the impact of the Great Recession on the labor market would result in a redefinition of the meaning and relative uncertainty attached to temporary employment. For this reason, the results for temporary employees must be understood in the context of both the previous discussion about unemployment and the decline in status of the self-employed, for whom the recession-push effect described by previous research increased the proportion of at-risk labor positions within a center-periphery schema. This argument constitutes a viable explanation of the striking impact of the crisis on psychological deterioration among the self-employed, shown by the corresponding regression coefficient in the 2012 sample (β = .383; p < .001; not significant for 2006).

Conclusion

Given the striking impact of the recessions on the labor-related processes, it is necessary to incorporate an analysis of the complexity of contemporary labor markets into the study of the relationship between economic cycle and health, with the objective of overcoming the employment/unemployment dichotomy characteristic of previous research. This article seeks to contribute to covering this gap. In fact, our results show that even the association among economic cycle, unemployment, and health can only be properly understood if the employment/unemployment dichotomy is considered with relation to the other relevant ES categories in a particular labor and economic context. In this regard, the association between ES and health is not structured as an atemporal reality. Rather, to the contrary, the different categories of ES would take their implications for health from the interdependent meaning that they are given by the specific (historical) economic context under consideration. In a related vein, there is reasonable theoretical and empirical support suggesting that the economic and labor processes described for Spain may be classified within the Mediterranean capitalism model. Specifically, LMS processes represent a central element in the description of this model. Our results suggest that these processes may make a remarkable contribution to explaining the changes in the pattern of association between ES and health during the impact of the Great Recession in Spain, such that peripheral segments situated at the margins of the labor market appear to have differing implications for health depending on the moment in the economic cycle. It may be possible to use the results of this study to analyze dynamics in other countries whose recent development fits the Mediterranean variety of capitalism. Likewise, our results suggest that it may be useful to carry out research analyzing the impact of the Great Recession on the distribution of health in economic contexts with characteristics distinct from those described for Spain in particular and for the Mediterranean model in general. This would allow the theoretical and empirical debate to be broadened, analyzing the hypothesis that understanding the impact of economic crises on health depends to a large extent on the specific characteristics of the labor market in the economic model under consideration—a topic that has rarely been explored in previous research. Along these same lines, classical approaches such as deprivation theory would benefit from incorporating the complexity of labor markets in a specific economic context into their analyses. It would hence be possible to explore in detail the role that psychosocial processes play in explaining the impact of economic cycles (especially recessions) on health. This would derive from consideration of the impact on the experience of employment (in its various forms) arising out of the interrelated meaning of the various labor market positions, beyond the employment/unemployment dichotomy.

This discussion must be qualified by the limitations of our study, which in turn indicate the relevance of future lines of research. As regards quantification of variables, the ES measure used reflects some of the main labor market strands, but particularly focuses on those of greatest importance in Spain during the period under study. Important situations such as underemployment and work in the hidden economy are not taken into account. In both cases, qualitative research would be particularly useful at times of social change and/or economic crisis. This is due to the sensitivity of techniques such as in-depth interview in gaining access to the social significance of processes that remain hidden due to their irregular nature, but become clear and express in the daily practices of social interaction such as “black market” work. Moreover, this article does not include data on the job conditions experienced by full-time permanent employees. Some of these features (i.e., job insecurity and competition, time pressures) may help to explain the impact of recessions on health and suggest an appropriate line of research. Furthermore, the two samples used are representative and allow for an efficacious evaluation of the changes in the pattern of association between ES and health in the context of economic crisis. Notwithstanding this, the two samples are cross-sectional, meaning it is not possible to separate the role played by the sample composition in explaining the differences that exist between 2006 and 2012. In particular, being unable to control the results according to a baseline/previous measure of health may mask the impact of the Great Recession in the appearance or intensification of an effect of social selection across ESs by an amount that is difficult to quantify, especially in the case of mental health. In the same manner, our data do not permit the identification of transitions into roles outside the labor force (students, homemakers, retired workers) as a response to unemployment (especially long-term unemployment). These transitions have the potential to be highly threatening for health and psychological well-being. Along these lines, using longitudinal data to examine the transition process between different ESs as a consequence of the Great Recession constitutes a priority area for research.