Introduction
The 2008–2009Footnote 1 economic crisis had substantial impacts on the labour market conditions of not only the developed countries but also many developing countries which are closely linked with the developed world in terms of their trade and financial structure. These impacts included higher unemployment rates, lower employment-to-population ratios, and increasing informality of labour markets. American labour markets experienced the deepest downturn of the post-war era and most European Union (EU) countries faced high unemployment rates and lower participation rates, especially for the younger segment of the labour market (Reference FarberFarber, 2010; International Labour Organization (ILO), 2012). In less than a year, the impacts of the crisis were transmitted to the developing world and across the globe, and the army of unemployed had risen to over 200 million with a job deficit of around 50 million, in comparison to the pre-crisis situation of 2007–2009 (ILO, 2012).
Turkey was hit hard by the 2008–2009 crisis, even though it had been a fast-growing developing country during the 2000s. Prior to the crisis, during the 2000s, the economy experienced relatively high growth rates of real gross domestic product (GDP), but the period has been described as one of ‘jobless growth’, with stable double-digit levels of unemployment rates and lower labour force participation rates (Figure 1(a) and (b)). Over the last two decades, the Turkish economy has experienced three other crises, in 1994, 1998 and in 2001, yet it was only in the 2008–2009 crisis that the average unemployment rate reached unprecedented figures. The impact was more severe for women (Turkish Statistical Institute (TurkStat), 2010).Footnote 2
Figure 1. (a) Unemployment rates, 1988–2011 and (b) labour force participation rates, 1988–2011.
Source: Turkish Statistical Institute (TurkStat) Household Labour Force Survey, from 1988 to 2011.
Research on the employment impacts of the crisis tends to focus on changes in two conventional indicators – employment and labour force participation rates. Even within these standard measures, experience from earlier crises suggests that impacts differ according to gender. This study questions the validity of the official indicators used in crisis analyses and aims to uncover the real extent of joblessnessFootnote 3 by redefining the concept ‘unemployed’ and bringing an uncounted group of unemployed into the picture through incorporating the marginally attached.
It is argued that the standard definition of the unemployment rate, while helpful for tracking general trends in labour markets, falls short of providing a thorough picture of the impacts of the crisis on employment – particularly in developing or emerging market economies. Turkey is a case in point. Incorporating the marginally attached when analysing the labour market impacts of the crisis not only provides additional information about experiences during the 2008–2009 economic crisis, it also has important implications for research on the added worker effect (that is the increasing incidence of labour force participation) and the discouraged worker effect (that is workers’ withdrawal from labour market due to the failed search for work) that occur during times of crisis.
So, what is missed out by the official definition of unemployed? Current job search criteria might serve well in identifying the unemployed in developed countries, where the majority of the population engages in regular paid employment and where information about available jobs is more easily obtained. However, this may not be the case in many developing countries, where labour market information channels are weak or do not exist. Owing to the significance of the rural sector and agricultural employment in these countries, both high seasonality and a large amount of unpaid family work during times of unemployment are commonly observed (Reference Bulutay and TaştıBulutay and Taştı, 2004). Agriculture is a female-dominated sector in much of the developing world, and most women workers in agriculture are unpaid family workers. In Turkey, in 2011, 42% of women workers were in agriculture and 77% of them worked as unpaid family workers (TurkStat, 2012). The sociological literature on Turkey indicates that many women classed as ‘economically inactive’, in fact, wish to work in the market – although they do not qualify as active job seekers. Thus, the actual unemployment rates for women based on this argument may significantly differ from the officially reported low rates (Reference Ecevit and ÖzbayEcevit, 1998; Reference Özbay and TekeliÖzbay, 1990).
Second, the existing unemployment insurance system is highly inadequate in reaching out to all individuals, owing to the relatively weak flows of labour market information in Turkey. In 2009, only 7.4% of those out of work were covered by unemployment insurance, with the amount of the payment corresponding to around 80% of the minimum wage. The majority of women in Turkey – given their very low official employment (26.2%) and participation rates (29.6%), were excluded from the insurance system. Because unemployment insurance is intended to benefit individuals who have been temporarily removed from the labour force, a prerequisite for the receipt of benefits is a demonstrated attachment to the labour force.
Low employment and participation rates in the Turkish labour market imply a tendency to stay out of the formal labour market (Figure 1(a) and (b)). This again indicates the inadequacy of the official definition of unemployment. In times of crisis, the unemployment rate does not capture effective job search, which increases following a rise in the fraction of the marginally attached, thus making transition to employment much more difficult for the unemployed (Reference Centeno, Maria and NovoCenteno et al., 2010). Quantifying the uncounted labour force would contribute to an understanding of cyclical variations in the labour force over the business cycle (Reference Gray, Heath and HunterGray et al., 2005).
We begin by discussing how we define the marginally attached and explaining why they should be included in measurements of unemployment. The section ‘Alternative definitions of unemployment’ also provides a descriptive analysis of the broadly defined unemployment rates adjusted by the inclusion of the marginally attached. The section ‘Data and methodology’ outlines the data and methodology used in the empirical analysis, along with a descriptive analysis of the marginally attached. In the section ‘Empirical analysis’, we explore flows into and out of unemployment and evaluate the transition dynamics of the marginally attached, examining their distribution by employment status in two consecutive periods. This section also summarises the estimated results concerning the determining factors for being marginally attached relative to other labour force states for women and men separately. The ‘Conclusion’ summarises the results obtained and suggests some implications of these findings.
The study makes two contributions. First, it presents additional information on the impacts of the crisis on unemployment, based on evidence from Turkey. Second, while there are some studies analysing labour force attachment in developing and emerging market economies, they do not discuss the issue in the context of an economic crisis (Reference Byrne and StroblByrne and Strobl, 2004; Reference Kingdon and KnightKingdon and Knight, 2000). Conversely, there are other studies that provide analysis of the flows between different labour force states during times of crisis, but these studies do not consider the marginally attached as a distinct labour force status (Reference Tansel and KanTansel and Kan, 2012; Reference Tansel and TaşçıTansel and Taşçı, 2005). This study thus breaks new ground by systematically incorporating marginal labour force attachment into an analysis of economic crises.
Alternative definitions of unemployment
According to 1954 ILO guidelines, a person is unemployed if he or she is (a) not working, (b) currently available for work and (c) seeking work. ILO broadened the definition of unemployment in 1982, allowing for partial or full relaxation of the active job search requirement in situations
where the conventional means of seeking work are of limited relevance, where the labour market is largely unorganised or of limited scope, where labour absorption is, at the time, inadequate, or where the labour force is largely self-employed … (ILO, 1982)
This definition of ‘unemployed’, which is currently used in almost all countries as the official definition, refers to individuals aged 15–65 years who did not work during the 7 days prior to the interview, who want to work and are available to start working within a week after the interview, and who have taken steps to either look for work or create some form of self-employment in the 4 weeks prior to the interview. The expanded definition excludes the last criterion, allowing for the passive search for work. There are several developing countriesFootnote 4 that report the rate either by fully or partially relaxing job search criteria and expanding the narrow definition of unemployment. However, this does not necessarily indicate that the expanded definition includes all the marginally attached who are not currently searching for work in either an active or passive way.
Based on ILO guidelines, the TurkStat defines the ‘unemployed’ as all persons 15 years of age and over who had not been employed during the reference week and who had used at least one channelFootnote 5 for seeking a job during the prior 3 months and were available to start work within 2 weeks. Persons who have already found a job and who will start working within 3 months, or who have established their own job but are waiting to complete necessary documents to start work are also considered to be unemployed.
Persons employed are defined as those who are at work and who were economically active during the reference week for at least 1 hour. The employed group includes all regular employees, casual employees, employers, self-employed or unpaid family workers who worked during the reference week for at least 1 hour. All self-employed and employers who had a job but were not at work in the reference week for various reasons are also considered as employed. Regular employees with a job who did not work during the reference period for various reasons are considered as employed only if they have an assurance of return to work within a period of 3 months, or if they receive at least 50% of their wage or salary from their employer during their absence.
Those outside the labour force include all persons 15 years of age and over who are neither unemployed nor employed. This group includes discouraged workers, who are those available to start a job but who do not seek one, either because they had looked for one before but were not able to find one, or because they believe that they cannot find a job matching their qualifications. Persons who are not seeking a job for reasons such as being a seasonal worker, student, retired, disabled or property income earner, or because they are occupied with household chores, are also counted in this group.
The distinction between unemployed and inactive depends on the search criteria; that is, a non-employed person who wants to work and is eligible for work is counted as unemployed if he or she is currently looking for a job and inactive otherwise. However, the inactive population consists of a highly heterogeneous group of individuals, including those who want to work but who are not actively searching for a job – such as discouraged workers – together with those who have caregiving responsibilities for household and community members (the ‘forced’ into inactivity), and those who prefer to stay outside the labour market (the truly inactive). In other words, among the inactive, the degree of attachment to the labour market varies significantly. The search criterion, therefore, becomes problematic: a person who wants to work and is eligible for work should be taken as closer to being unemployed than inactive, although he or she has stopped looking for a job. Accordingly, when we reclassify the labour market states based on the Household Labour Force Survey (HLFS) questionnaires and identify the persons marginally attached who are neither employed nor unemployed but would like to work if a job opportunity exists, we obtain a fourfold classification of the labour force. Figure 2 summarises how we construct the new labour force classification as (1) employed, (2) unemployed, (3) marginally attached and (4) out of the labour force. Table 1 presents the figures for the redefined unemployed and redefined labour force vis-a-vis the official ones. When compared to the official figures, women’s shares in the redefined unemployed and labour force are much higher. We also observe that changes in unemployment and the labour force over the period of analysis are also underestimated by conventional indicators based on the official definitions.
Figure 2. Construction of the four-state classification of the labour force.
Table 1. Official and redefined unemployed and labour force, Turkey.
Source: Authors’ calculations.
LF: labour force.
Without taking into account the marginally attached identified above, the TurkStat reports only the narrowly defined unemployment rate. Statistics on discouraged workers are also published in monthly reports, but they are not counted as ‘unemployed’ in the official definition and no alternative unemployment rates are provided. The literature, though very limited, provides a few examples of countries that publish alternative definitions and measures of unemployment. For example, for the United States, the Bureau of Labour Statistics (BLS) reports five alternative measures of labour underutilisation in addition to the official unemployment rate. Among these, U1 corresponds to persons unemployed 15 weeks or longer, as a percent of the civilian labour force, U2 includes job losers and persons who completed temporary jobs, and U3 is the official unemployment rate, which refers to the total unemployed as a fraction of the labour force. The three others correspond to widened definitions; U4 includes discouraged workers with the other unemployed, U5 incorporates the other marginally attached, as well, and finally U6 includes all persons covered by U5 plus persons employed part-time for economic reasons and those who want and are available for full-time work but have had to settle for a part-time schedule. According to the BLS (2012) definition, persons who are marginally attached are those who have looked for a job sometime in the prior 12 months (or since the end of their last job if they held one within the past 12 months), but were not counted as unemployed because they had not searched for work in the 4 weeks preceding the survey.
It is not possible to derive from HLFS data whether the respondent has looked for work or not over the year, although he or she is currently not searching. In fact, the HLFS questionnaire asks about the respondent’s labour force status in the same month 1 year before the survey, but does not include the appropriate questions to capture information on their status over the year. Similarly there is no certain time period specified in TurkStat’s definition of discouraged worker. Unlike in the US case, in Turkey discouraged workers are officially defined as persons who have given a job market-related reason for not currently looking for work, yet would like to work if the opportunity exists within 2 weeks. When choosing the definition of marginally attached used here, we took into account TurkStat’s definition of the discouraged. Thus, unlike the BLS definition, we used the broadest definition of the marginally attachedFootnote 6 without a time constraint with respect to the time elapsed after the last search activity.
The previous empirical literature on the ‘marginally attached’ includes studies with different definitions. In their research on the Canadian labour market, Reference Jones and RiddellJones and Riddell (1999, 2002) classify as marginally attached, those who are not searching for work but state that they want to work. In their Australian study, Reference Gray, Heath and HunterGray et al. (2005) use a similar definition. This is the definition we use here. On the other hand, there are studies that use the definition with a time constraint. For example, Reference Brandolini, Cipollone and VivianoBrandolini et al. (2006) in their study of EU countries, define all job seekers whose last search action occurred more than 4 weeks before the interview as members of a potential labour force. On the other hand, Reference Byrne and StroblByrne and Strobl (2004) define as marginally employed anyone who worked some time in the 3 months prior to the interview, and is willing and able to work, although not looking for a job at the time of the interview. Discouraged workers are, by definition, a subset of the marginally attached in the above-mentioned studies. With this in mind, in order to keep compatibility with official discouraged worker statistics, we, like Reference Jones and RiddellJones and Riddell (1999, 2002) and Reference Gray, Heath and HunterGray et al. (2005), used the broadest definition.
Taking insights from earlier research, here we provide three alternative unemployment rates for Turkey besides the official rate: U4, U5 and U6. U4 here corresponds to the U4 in the BLS case. U5 is defined as the unemployed plus discouraged workers as a percent of the sum of the official labour force and the discouraged. U6 is the unemployed plus the marginally attached as a percent of the sum of the official labour force and the marginally attached. Table 2 and Figures 3 and 4 present unemployment rates for Turkey, by these alternative definitions.
Table 2. Alternative definitions of unemployment, Turkey.
Source: Authors’ calculations.
U3-Turkey is the official unemployment rate. The total unemployed (official) as a ratio of the labour force (by official definition) (the figures by official definitions of employed, unemployed and labour force are presented in Table 1).
U4-Turkey is the total unemployed plus the discouraged workers as a ratio of the official labour force plus the discouraged.
U5-Turkey is the total unemployed plus the marginally attached (which includes the discouraged workers) as a ratio of the official labour force plus the marginally attached.
U6-Turkey is the total unemployed plus the marginally attached and part-time workers as a ratio of the official labour force plus the marginally attached.
Figure 3. Alternative unemployment rates: Turkey, 2004–2010.
Figure 4. Alternative unemployment rates by sex, Turkey 2004–2010.
As expected, Figure 3 shows that the redefined unemployment rates are higher than the official rate; U4 is approximately two percentage points higher than the official rate (U) throughout the period. Incorporating the marginally attached widens the difference remarkably: between U5 and U it is around six percentage points, on average, over the period. Disaggregating by sex enables us to show that the expanded rates substantially exceed the official rate in the case of women; U4 is about three percentage points higher than U, while the difference between these two rates is two percentage points, on average, for men. Conversely, inclusion of the marginally attached leads to a drastic divergence of U5 from the official rate for women: U5 is approximately 11 percentage points higher than U, whereas this difference is just around four percentage points in the case of men. During the years when the crisis impacts were more severely observed in the Turkish economy (2008–2009), the difference between U and U5 was as high as 12 percentage points for women. We also provide our calculations for U6. We observe a constant rise in U6 particularly for women over the whole period of analysis, but also during the crisis period, when the difference between U6 and the others expands significantly.
When only the official rates are considered, the difference between women and men is not very significant (less than one percentage point). However, as the broader measure (U5) shows, the gender gap widens drastically and the difference becomes nine times more than the officially identified gap. These findings reveal once again how official unemployment rates underestimate the actual numbers, particularly when it comes to women’s joblessness. If we look at the ratio of the marginally attached to the working age population in order to see the extent of the underestimation of the official rates, we see that approximately 4% of women of working age are marginally attached. For men, the figure is around 3%. As a ratio to the unemployed, the figures are much higher. Throughout the period, the marginally attached corresponds to around 41% for men, on average, while for women, in every year, it is more than 100% (marginally attached numbers are, on average, 1.16 times higher than the number of unemployed women).Footnote 7 These figures point to a major structural issue in women’s participation and joblessness in Turkey. Despite their willingness to work, women in Turkey cannot engage in job searches and cannot participate in the market as employed. Thus, neither the labour force participation rates nor the unemployment rates reflect women’s actual state regarding the labour market.
Beyond the structural pattern summarised, the size of the marginally attached also presents a cyclical trend over time,Footnote 8 which calls for special attention to the crisis period in our case. When we focus only on the period from 2007 to 2009,Footnote 9 we observe that the increase in the marginally attached for women was five times more than that for men (225,000 vs 44,228, respectively) (Table 3). In all, 84% of the total increase in the number of the marginally attached were women, which indicates the disproportionate impacts of the crisis on women in Turkey. Conversely, we observe that 319,000 women (29% of the total increase) and 775,000 men became unemployed in official terms in the same period. The growth rates of the unemployed and the marginally attached by sex display these differences more evidently (Table 3).
Table 3. Crisis in numbers.
Source: Authors’ calculations.
Data and methodology
We use the annual data derived from HLFS between 2004 and 2010 in this study. HLFS, which has regularly been conducted since 1988, is the main data source on the labour market situation of the country from the supply side. It gives information on economic activity, occupation, status in employment and hours worked for employed persons, as well as information on several specifications – such as the duration of unemployment and occupations sought by the unemployed. All geographic regions in Turkey are covered and roughly 13,000 persons aged 15 years and over are interviewed every month.
From 2004 to 2010, the figure accounting for the marginally attached has almost doubled, reaching nearly 2 million people in 2010. Figures 5 and 6, when taken together, show how remarkable the increase is in the number of the marginally attached, particularly for women. The share of women in the total number of marginally attached is considerably higher than that of men for the whole period, and the gap widens throughout the period.
Figure 5. Distribution of the marginally attached by sex.
Figure 6. The marginally attached, 2004–2010.
Distribution of the marginally attached by key individual characteristics shows that the majority of the marginally attached women (57%) report doing care workFootnote 10 as the reason for not looking for a job (Figure 7). The corresponding figure is only 5% for their male counterparts. There is, in fact, a large body of literature that explores the impact of unpaid care work responsibilities on the labour force status of individuals,Footnote 11 and these statistics support earlier findings. Conversely, for men, being discouraged is the main reason for not looking for work.
Figure 7. Distribution of the marginally attached by reason for not looking for job (a) women and (b) men.
By levels of education and age, the majority of the marginally attached has lower levels of formal education and is younger in Turkey (Figures 8 and 9). This evidence supports the argument that education increases the probability of being in the labour force.Footnote 12 As can be seen by Figure 10, the highest proportion of the marginally attached corresponds to women living in urban areas (reaching almost 40% after 2007). Throughout the period, their share increased regardless of residential location, while the share of men, both in rural and urban areas, declined. Similar to many other developing countries, the recent economic crisis first affected the industry most, and thus the urban population experienced the crisis more severely in Turkey.
Figure 8. Distribution of the marginally attached by sex and education.
Figure 9. Distribution of the marginally attached by sex and age.
Figure 10. Distribution of the marginally attached by sex and location.
For the empirical analysis, we pooled the data collected in different years, adjusting their sampling weights. There have been some changes in the coding used in the survey along the period of analysis. In order to ensure compatibility, variables including the marital status, educational attainment and reasons for not looking for a job are checked and recoded. The methodology applied in empirical analysis has two parts. First, in order to observe the flows into and out of the pool of joblessness, we construct a transition matrix that consists of unconditional transition probabilities between different labour force states given by
Here, we assume that at each period, individuals are in one of the four different labour force states: employed (E), unemployed (U), marginally attached (M) and not in the labour force (N). Pij is the probability of moving from state i at the initial period to state j in the following period, defined as the ratio of the number of people who are in state j at time t + 1 while they were in state i at time t to the total number of people in state i at time t. Accordingly, P EU represents the probability of being unemployed at time t + 1 while being employed at time t. There are supposed to be 16 transition probabilities in total as we assume four different labour force states. However, data we use do not allow us to observe whether the individual was marginally attached or not in the previous year.Footnote 13 This is partly because in the HLFS questionnaire, the question regarding the previous year’s status does not provide being marginally attached as a possible answer among the others. Additionally, the yearly data we pooled do not have a panel data structure, which would allow us to track the individual in previous years. Owing to these data limitations, the third row in matrix (1) cannot be calculated. Thus, we have a 3 × 4 matrix.
Second, aiming to understand the determining factors behind being in different labour force states, we estimate the conditional probabilities of being marginally attached relative to other states. For our purpose, we use a logistic regression, as shown by the following equation
Accordingly, the right-hand side of the equation shows the probability of transiting to state j at time t + 1 of the kth individual who is in state i at time t. X is a vector of variables including key individual characteristics (age, sex, education and marital status) and household characteristics (location, number of children and elderly). We interpret the results calculating the marginal effects based on the estimates. Our estimations are done separately for women and men, as we aim to understand the relative impact of the crisis on unemployment, particularly by sex.
Empirical analysis
Following the first step of the methodology described in ‘Data and methodology’ section, we first calculated the transition probabilities of individuals between different labour force states. We identified the current year status for each individual based on our fourfold classification of the labour force as (1) employed, (2) unemployed, (3) marginally attached and (4) not in labour force. However, for the previous year status, due to data limitations,Footnote 14 we stick to the conventional three-group classification as (1) employed, (2) unemployed and (3) not in labour force.
Our results show that the probability of transition from being employed to being marginally attached is higher for women than for men throughout the period of analysis. The probability of keeping an employed status decreases over the crisis period, and the decline is more significant for women than for men. This result indicates that women in Turkey are at a higher risk for losing their jobs when compared to their male counterparts, a pattern also observed during the previous crisis in Turkey.Footnote 15 Additionally, these reflect that the discouraged worker effect of the crisis is more pronounced for women in Turkey. Furthermore, the difference between women and men in terms of their transition probability from being employed to being marginally attached and to out-of-labour-force status displays an increasing trend over the crisis years (Table 4).
Table 4. Transition probabilities.
LF: labour force.
Conversely, although among the unemployed the ratio of those moving to a marginally attached status is higher for men compared to women, they have a higher likelihood of finding a job (changes from 38.1% to 45%). For women, the probability of staying unemployed is much higher and rises over the crisis period, 2008–2009 (changes from 44.3% to 50.2%) (in Table 4). These findings highlight the fact that existing gender-based inequalities in the labour market were deepened by the asymmetric impacts of the crisis.
Next, we analyse the probability of being marginally attached, conditional on factors including individual and household characteristics. Age is included to capture possible life-course influences on individuals’ attachments to the labour market. Dummies for educational attainment (primary, secondary and tertiary) are expected to reflect the close association between the levels of education and employment status. The indicators for household structure (number of children less than 5 years old, number of children aged between 5 and 14 years, and the number of elderly aged over 64 years) and marital status are significant in determining any household member’s labour market attachment given the fact that these variables are closely linked with household care needs. In a country like Turkey, where the prevalent division of labour between women and men in the household is strictly traditional and public provisioning of care services is very limited, these needs are covered, in general, by women in the household, which directly affects their degree of labour market attachment.
Additionally, we include a regional dummy as rural/urban (1/0) to control for the effect of regional disparities in labour market attachment. This would help capture the differences in the labour market conditions that arise more from the demand side. In order to observe possible impacts of the crisis, we also include the year dummies as explanatory variables.
We use the probability of being marginally attached relative to unemployed and out-of-labour force as the dependent variable, and estimate logistic regressions for women and men separately as we look for the relative impact of the crisis particularly by sex. Results are reported in Table 5. Coefficients display the marginal effects of the independent variables, calculated as the effect of a one-unit change in an explanatory variable from the average level on the probability estimated, holding all other variables at their average value. Marginal effects of the binary variables reflect the effect of having the characteristic given all other variables again at their average value.
Table 5. Determinants of being marginally attached relative to other employment statuses.
The dependent variable takes the value 1 if the individual is marginally attached and 0 if in the comparison group. Sampling weights are adjusted when pooling multiple years of Labour Force Survey (LFS) data. Coefficients display the marginal effects of independent variables. Robust standard errors are in parentheses.
*** p < 0.01;
** p < 0.05.
The younger the individual, the more likely he or she was to be marginally attached relative to being out-of-labour force. This is also supported by the results of the descriptive analysis in ‘Data and methodology’ section. Living in rural areas increased the probability of being marginally attached, which indicates a stronger attachment of the rural population to the labour market in Turkey. Labour force participation in rural areas was much higher for women and men, when compared to the urban areas.Footnote 16
Higher levels of education increased an individual’s attachment to the labour market. This was observed for both women and men. However, compared with single women, married women were less likely to be marginally attached and more likely to be out-of-labour force, while the reverse was true for men. A similar influence is also observed for the number of children aged less than 5 years, which indicates that critical life transitions, such as marriage and having children, had asymmetric gender impacts.Footnote 17
When we look at the results for being marginally attached versus being unemployed, we observe this time that the older the individual is, the more likely he or she is to be marginally attached rather than unemployed. Higher levels of education increase the likelihood of entering into the labour force and looking for a job actively. Marriage, again, has asymmetric impacts for women and men; married women are more likely to be marginally attached, whereas the opposite is observed for their male counterparts. For the population living in rural areas, there is also a higher incidence of being marginally attached relative to being unemployed. Further to these, the higher the number of children as well as the higher the number of elderly, being marginally attached compared to being unemployed is more likely for both women and men. As we have shown in a study based on time use data, performing unpaid care work taking care of children and/or the elderly exerts significant influence on the decision to enter into the labour force and look for a job.Footnote 18
Regarding the possible impacts of the crisis, we find that before the crisis (except for 2006), women were more likely to stay out-of-labour force than to being marginally attached, but the reverse has been true for the years after 2008. This suggests that during the crisis, women in Turkey entered into the labour market and started actively looking for jobs. Although we observe similar impacts for both women and men, the likelihood of being marginally attached relative to being out-of-labour force was much higher for women over the period.
Conclusion
A more complete understanding of the impacts of the 2008–2009 economic crisis on joblessness in Turkey is gained by including marginally attached persons, who are not counted as unemployed in official statistics. The marginally attached would like to work if the opportunity existed, even though they are not actively engaged in job search. It has been argued that standard definitions of unemployment rates are questionable from a developing country perspective. The search criteria used to identify who is unemployed and who is inactive could be very problematic in a country where agricultural employment is relatively significant for the economy (a large portion of which consists of unpaid family workers) and where there is a lack of adequate unemployment insurance systems.
Calculation of broader measures of the unemployment rates has revealed that the extent of joblessness is significantly underestimated by the official definitions of unemployment, and this effect is more marked for women than for men. The recent economic crisis has sharpened these existing gender-based differences in employment. The article has shown that women are more likely to be marginally attached than men in Turkey, and over the crisis period, the number of marginally attached grew significantly faster for women, and women had a higher risk of losing their jobs. These findings emphasise the importance of marginal attachment in analysing the extent of joblessness in Turkey.
The study presents results that may help address gender-based inequalities, by contributing to analyses of the factors underlying the low employment and participation rates of women in Turkey. It suggests a major obstacle that will need to be overcome in order to achieve the current policy targets in the National Employment Strategy (2012–2023) – of decreasing women’s unemployment rate by half, while increasing their labour force participation rate from 27.6% to 35% by 2023. Care work is the main reason behind women’s ‘weaker’ attachment to the labour market, so any policy that promotes women’s employment in the market should develop strategies that enable the sharing of this burden, as well as providing employment opportunities in the market. Using conventional criteria, we believe it is not possible to identify the degree of motivation to work or search for work, in the case of women in Turkey, where the possibilities of shifting the care burden to someone else or an institution are very limited, as a result of the inadequate provisioning of public care services.
Acknowledgements
The authors would like to thank the anonymous reviewers for their comments and suggestions on an earlier version of this study. They also wish to acknowledge valuable inputs of Rania Antonopoulos, Nilufer Cagatay and Tamar Khitarishvili throughout the conduct of the research. The usual disclaimer of errors and omissions applies.
Funding
The authors acknowledge the generous support provided by the Scientific and Technological Research Council of Turkey (TUBITAK-112K516) for this research.
