Exploring unemployment persistence: a probabilistic analysis in 20 OECD countries to understand its social implications

Exploring unemployment persistence: a probabilistic analysis in 20 OECD countries to understand its social implications Carlos David Cardona-Arenas School of Economics and Finance, Universidad de Manizales, Manizales, Colombia Abstract Purpose –This study assesses the probability of an OECD member country exhibiting high persistence in unemployment duration, considering income inequality, productivity, accumulation of human capital and labor income share in Gross Domestic Product (GDP) between the years 2013–2019. Design/methodology/approach –To achieve the purpose of the study, a probabilistic analysis with panel data is employed, focusing on 20 OECD countries segmented into two groups: those with high persistence and low persistence in unemployment duration. Probit and Logit models are estimated, marginal changes are analyzed and the models are evaluated in terms of their classification accuracy. Finally, trends in probabilities over time are examined. Findings –This paper exhibits that countries with higher human capital index, greater labor income share in GDP, and more relevant productivity for well-being reduce their probabilities of experiencing high persistence in unemployment duration. It is observed that Mexico (MEX), Greece (GRC), Italy (ITA), and Turkey (TUR) have elevated probabilities of experiencing high persistence in unemployment duration in the future, while Costa Rica (CRI), Estonia (EST), Slovakia (SVK), Czech Republic (CZE), Lithuania (LTU), Poland (POL), and Israel (ISR) show a marked downward trend in these probabilities. Lastly, countries like the United Kingdom (GBR), Denmark (DNK), Sweden (SWE), Norway (NOR), Netherlands (NLD), Germany (DEU), United States (USA), and Canada (CAN) present minimal risk of experiencing high persistence in unemployment duration in the future. Research limitations/implications –The measurement of the relationship between development outcomes and persistence in unemployment duration has been scarce. Generally, the literature has focused on the analysis of development and unemployment without delving into the duration of unemployment, let alone persistence in duration. Practical implications –This paper provides a solid foundation for the formulation of policies aimed at promoting sustainable employment and inclusive economic growth. Social implications –Based on the findings of the study, two key development policies are proposed. Firstly, the implementation of investment programs in Human Capital to increase productivity is recommended. Resources should be directed towards initiatives that improve the necessary skills and competencies in the labor markets of OECD countries, especially in strategic economic sectors with higher production linkages. Additionally, incentivizing the application of active labor policies is proposed. This entails prioritizing policies aimed at increasing the labor income share in GDP through progressive fiscal reforms that strengthen social safety nets and ensure fair labor standards. Implementing employment programs targeted at vulnerable groups, such as long-term unemployed individuals, youth, female heads of households and marginalized communities, is also recommended to eliminate structural barriers to labor market participation and reduce disparities in unemployment persistence. Adopting these policies can help mitigate the risk of high unemployment duration persistence and foster sustainable and inclusive long-term economic growth. Originality/value –This is the first study to analyze the probabilities of both developing and developed countries experiencing high persistence in unemployment duration. It specifically evaluates these probabilities over a period of time and also estimates potential outcomes if real investments were made to enhance their human capital, productivity and employability. Keywords Unemployment persistence, Human capital, Productivity, Income, OECD, Panel data, Probabilistic models Paper type Research paper International Journal of Sociology and Social Policy JEL Classification — E24, J64, J24, D78, C25, O15 The current issue and full text archive of this journal is available on Emerald Insight at: https://www.emerald.com/insight/0144-333X.htm Received 5 June 2024 Revised 24 July 2024 Accepted 2 August 2024 International Journal of Sociology and Social Policy © Emerald Publishing Limited 0144-333X DOI 10.1108/IJSSP-06-2024-0245 1. Introduction Unemployment is a significant socioeconomic phenomenon with asymmetric implications at both the individual and aggregate levels, profoundly affecting the economic development of nations (Heimberger et al., 2017). Following the financial crisis of 2008, OECD member countries have faced persistent and elevated unemployment rates compared to previous periods (OECD, 2013). According to the OECD (2013), there has been an increase in structural unemployment in this group of countries, suggesting an underlying complexity in the labor market. However, a decade after the crisis, a reduction in inactivity rates has been observed in most OECD countries, with the exception of Chile, Latvia, Costa Rica, and the Czech Republic. Nevertheless, the number of hours worked has experienced a significant decrease compared to the pre-crisis period, particularly notable in Eurozone countries and the United States. Recently, the world has faced a new crisis, random and exogenous, as highlighted by Arenas et al. (2024), emphasizing its impact on the industry with a clear link to the increase in the duration of unemployment. This sector, in the context of OECD countries, drives economic growth (see: Kaldor, 1966,1967,1968 and Romer, 1986). The latest OECD report (2023) warns of an increase in job vacancies for the last quarter of 2022 compared to the preCOVID-19 crisis period. In this historical post-Covid-19 context, the unemployment rate has shown an upward trend in OECD countries between 2022 and 2024. This situation poses significant challenges for the stability of the labor market in these countries, highlighting the need for effective policies to address the persistence of unemployment. In this sense, Le Clech and Guevara-P� erez (2023) support the idea that improving the efficiency of the labor market can increase productivity by reducing unemployment. For their part, Mart� ınez et al. (2001) highlight that differences in the relationship between unemployment and income distribution in OECD countries explain the risk of increasing poverty in most of these countries. At this point, it is necessary to highlight the link between development outcomes and the persistence in unemployment duration. The literature is not extensive enough as it has generally focused on the analysis of development and unemployment without delving into the duration of unemployment and its persistence. For the Colombian context, the study by Cardona-Arenas and Sierra-Suarez (2023) has addressed the determinants of this phenomenon. However, the understanding of the determinants of unemployment duration remains very scattered. Specifically, Webster (2005) indicates that long-term unemployment results in unemployability, generating social costs in terms of development. The advances in the comparability of data collected by the International Labour Organization between 2013 and 2019 have enabled the empirical estimates in this study. Although the time limitation restricts the use of more recent data, this study is the first documented attempt in the literature to use this data in a probabilistic analysis to determine the persistence in the duration of unemployment for a set of OECD countries using a panel approach. Thus, the present study poses the following research question: What is the probability that an OECD country is characterized by high persistence in unemployment duration given its results in terms of income inequality, productivity relevant to well-being, accumulation of human capital, and labor income participation in GDP? The general hypothesis of the study posits that the probabilities of being a country with high persistence in unemployment duration change as the accumulation of human capital (hc), productivity relevant for wellbeing (cwpft), labor income share in GDP (labsh), and income inequality (gini) change. This general hypothesis arises from a general theoretical framework of Becker (1964) and Uzawa (1965), who argue that the improvement in human capital, measured through education and labor training, leads to an increase in worker productivity and the accumulation of human capital. More recently, the work of Focacci and Perez (2022) stands IJSSP out, who demonstrate the relevance of investing in human capital to achieve economic and social success in the context of the United Kingdom, United States, Germany, and Sweden between 1830 and 1970. Based on this premise, the objective of our research is to determine the probability that an OECD country exhibits high persistence in unemployment duration, considering its results during the period 2013–2019 in terms of income inequality, productivity relevant to well-being, accumulation of human capital, and labor income participation in Gross Domestic Product (GDP). These variables have been selected because they are considered to directly influence unemployment persistence and are supported by previous literature, see: Becker (1964),Uzawa (1965),Evans (1989),Van Ours (2015),Dixon et al. (2017). This is one of the first studies that has been conducted to analyze the probabilities of both developing and developed countries experiencing high persistence in unemployment duration. Using Logit and Probit model estimation with a panel approach, this work specifically evaluates these probabilities over a period of time and also estimates potential outcomes if real investments were made to enhance their human capital, productivity, and employability. The main contribution to the literature consists of evaluating the change in the probability of experiencing high or low persistence in unemployment duration, which evidences potential trends. These trends will allow policymakers to undertake actions to reverse them through intervention in the variables of interest, aligning with their respective public policy agendas. This research is structured as follows: the first section provides a detailed literature review related to the two main categories of analysis in the study: the link between economic development and unemployment duration. The second section clearly describes the sources of information, as well as the variables and data analyzed, along with the methods and techniques used to achieve the research objectives and answer the research question posed. The third section presents the results and discusses them in light of the research background, and finally, the last section outlines the general conclusions. 2. Literature review The analysis of the labor market reveals that prolonged work inactivity can deteriorate an individual’s skills and employability. One of the pioneering works in analyzing labor market characteristics and rigidities was conducted by Narendranathan et al. (1985), who presented evidence for OECD countries suggesting that monetary unemployment benefits actually prolong its duration, a finding supported by Nickell (1997). According to the OECD (1993), the duration of unemployment determines the emergence of welfare benefits, and not vice versa, a topic of ongoing debate. The studies by Topel and Welch (1980) investigate how government subsidies can stimulate the transition between employment and unemployment, facilitating reemployment upon the conclusion of such subsidies. Moreover, G€ ocke (2002) argues that extended periods without work activity can diminish work quality, largely due to the lack of continuous training and skill development. Rogers (1997) analyzes the relationship between the spatial structure of labor markets and job search outcomes. The study findings indicate that the duration of unemployment is influenced by residential location. Dawkins et al. (2005) employed probabilistic models of continuous-time panel data in the United States, explaining unemployment duration based on various characteristics such as residential segregation. Carling et al. (1996) examined transitions out of unemployment in Sweden using a semi-parametric unrestricted risk model, revealing a negative relationship between variables such as age, gender, and monetary assistance concerning the duration of unemployment. On the other hand, Fitzgerald (1998) views voluntary unemployment as an opportunity to invest in acquiring and improving skills. Baker (1992) utilizes data from the Current International Journal of Sociology and Social Policy Population Survey of the United States (1979–1988) and notes that the duration of unemployment is countercyclical and is linked to changes in the composition of the entry into unemployment. Blanchard and Portugal (2001) compare the duration of unemployment in Portugal and the USA, concluding that in Portugal it is longer due to slower and lower flows of entry and exit, influenced by pecuniary and non-pecuniary factors. Zamanzadeh et al. (2019) analyze the duration of unemployment as a result of changes in fiscal, monetary, and business cycle variables. The authors employ a Bayesian mixed effects quantile model; their main finding is that in a scenario of a higher cyclically adjusted primary fiscal surplus, there are higher means and quantiles of unemployment duration for older individuals compared to younger ones. The consequences of prolonged unemployment for individuals can be severe, including decreased self-esteem, mental health problems, loss of professional skills, and reduced income. Research on the degradation of human capital during unemployment takes various perspectives. Pissarides (1992) emphasizes that skill loss can contribute to long-term unemployment persistence. Burdett et al. (2011) analyze the accumulation of human capital during employment, while Coles and Masters (2000) argue that subsidies for job creation may be more effective than training in situations of prolonged unemployment. Rotar and Krsnik (2020) focused on the relationship between unemployment benefits and its duration, examining how the availability and generosity of benefits can influence the duration of unemployment. The concept of persistence in unemployment can be attributed to two currents. The first is a condition of stationarity or unit root in the behavior of the unemployment rate series. In this approach, the works of Blanchard and Summers (1986) stand out, up to the most recent contributions by Furuoka (2017),Bahmani-Oskooee et al. (2017),Mathy (2017), and Marques et al. (2017). The second approach considers that persistence is characterized by slow adjustments in the level and behavior of unemployment. Here, the works of Franz (1990) and Webster (2005) stand out. To achieve a more comprehensive understanding of the relationship between persistent unemployment, its duration, and development, it is important to mention the work of Abraham et al. (2019). They investigate the consequences of long-term unemployment using survey data, emphasizing the effects on various aspects such as welfare, health, and financial stability. More specifically, Apergis and Aergis (2020) investigate whether long-term unemployment is due to skill obsolescence or technological shifts, concluding that improving existing skills or acquiring new ones is effective in mitigating long-term unemployment. Regarding Labor Income and Inequality, in the context of OECD countries, Afonso (2016) demonstrates how labor market institutions impact employment and wages, suggesting that labor market flexibility is associated with higher employment and economic growth, while strong collective bargaining is related to higher wages, lower wage inequality, and increased unemployment and inflation. Bhattarai (2016) examines the trade-offs between unemployment and inflation in OECD countries, highlighting the complexity of macroeconomic policies and their influence on economic stability. Sakamoto (2002) and Sturn (2013) explore the effect of government policies and labor market regimes on unemployment, economic growth, and inflation, pointing out how weak governments and inflexible labor policies can negatively impact the economy and increase unemployment levels. On the other hand, Murtin and Robin (2018) analyze the dynamics of unemployment in nine OECD countries using a stochastic search-matching model. They conclude that effective measures to reduce unemployment include increasing investment in employment services, reducing unemployment insurance, and deregulating the product market. Subsequently, Kriaa et al. (2020) investigate the factors determining the duration of unemployment in young IJSSP men and women in Tunisia. They use empirical data to analyze the influence of education, work experience, and other socio-economic factors on the duration of unemployment in this specific group. Furthermore, Focacci and Perez (2022) underscore the importance of investing in human capital for economic and social success. Their research focuses on the United Kingdom, United States, Germany, and Sweden between 1830 and 1970. Complementarily, Lauder and Mayhew (2020) argue, in the context of the United States, India, Germany, and the Netherlands, that the growing competition in the labor market is a direct consequence of the expansion of higher education. They question the role of academic achievement as an expression of the accumulation of human capital or as a signal of capabilities. Regarding income distribution, Scarchilli and Triventi (2023) study for Italy how individual circumstances present from birth influence the distribution of income in the workforce and how this has changed over time. On the other hand, Akdede and G€ ocekli (2019) research the relationship between unionization and labor market participation in income distribution in OECD countries. They conclude that unionization positively affects equity in income distribution. Whereas Arestis et al. (2023) analyze the relationship between labor flexibility and income distribution in Europe, concluding that stricter provisions on the use of fixed-term contracts and temporary agencies have a positive impact on the growth of labor shares. Finally, Cruz (2023) analyzes labor productivity, real wages, and employment in OECD economies. In a panel study for 25 countries, Cruz highlights that an increase in labor productivity boosts aggregate demand and fosters job creation. This study significantly contributes to addressing the identified gaps in the literature on prolonged unemployment. It considers key factors such as income inequality, productivity, accumulation of human capital, and labor income participation in GDP. This research provides a more comprehensive understanding of the determinants of persistence in unemployment duration, an aspect that remains underexplored in the literature to date. 3. Methodology 3.1 Methods and data The present research adopts an explanatory quantitative design, employing a probabilistic panel analysis approach, utilizing a data structure spanning from 2013 to 2019 and focusing on 20 OECD countries. The data is obtained from the Penn World Table of the Groningen Growth and Development Centre (GGDC) (2024), based on the methodology of Summers and Heston (1991). These data facilitate comparisons of living standards adjusting for purchasing power parity, following Feenstra et al. (2015). The vector of explanatory variables includes the Human Capital Index (hc), the share of labor in national income (labsh), levels of Total Factor Productivity relevant for well-being (cwtfp), and the Gini coefficient derived from data collected by Hasell et al. (2024) on economic inequality from the “Our World in Data” database. To classify the countries in the OECD sample as having high or low persistence in unemployment duration, an indicator of persistence is constructed. This indicator is denoted as LAPU (Long-Term Unemployment Measured as a Percentage of Total Unemployment) and is inspired by the work of Webster (2005, p. 99), who developed the indicator to assess persistence in long-term unemployment. It is calculated as the percentage of individuals who have been unemployed for one year or more relative to the total unemployed for more than one year, as follows in equation 1: International Journal of Sociology and Social Policy LAPUit ¼Unemployedt≥52 weeks Total unemployed t−52 (1) This variable has been recently calculated by Cardona-Arenas and Sierra-Suarez (2023), emphasizing that it allows capturing the inertial effect of unemployment duration, a phenomenon associated with the hysteresis condition of unemployment [1].G€ ocke (2002) explains how the term in the field of economics has been applied in the area of the labor market to describe the dependence that unemployment, or rather the natural rate of unemployment, has on its own trajectory over time. 3.2 Identification of the probabilistic model The identification strategy starts by assuming that the vector of explanatory variables is not determined by the gripping variable (countries with high persistence in unemployment duration versus countries with low persistence in unemployment duration). Given this, we follow the recommendation of Blundell and Dias (2009), where they suggest that the popular procedure consists of using a parametric specification in the form of a Logit or Probit model to overcome possible problems of dimensionality, and whose results reveal fairly good performance. In general, the bias is not large and actually very low in the absence of correlated residuals. The maximum likelihood estimation method with robust errors to heteroscedasticity was selected for both the panel Probit and panel Logit models, which is suitable in the context of a binary outcome variable. Thus, for the general model formulation, it is considered that the outcome variable is constructed based on LAPU. To do this, the persistence in unemployment duration is calculated, and by ranking the countries from highest to lowest persistence, they are categorized as countries with high and low persistence. Countries with high persistence are assigned a value of 1, and those with low persistence are assigned a value of 0. The selection criterion is based on the difference in the mean and extreme values of the LAPU variable. Thus, the theoretical model takes the form: E½Yit ¼1ðXÞ� ¼ Pr½Yit ¼1ðXitÞ� (2) where the probability is denotes by equation 3 as follow: Pr½Yit ¼1ðXitÞ� ¼ b ω jþXit;b βj:(3) where Xjt;It is a vector of quantitative explanatory variables for the categorization between countries with high and low persistence ω jis the constant and βjDetermines the effect of variations in Xjt;on the probability of response Pr½Yt¼1ðXÞ�. These non-linear models are characterized by fitting a distribution function as follows: ½0<GðZÞ<1� ¼ ∀Z¼ ω JþXit;βJ. The goodness-of-fit measures considered in this study are the percentage of correct predictions, for which the estimated probability of occurrence is calculated for each observation at the moment as: Yt¼1:b Pt¼b PðYit ¼1��X1t;...:;XJTÞ ¼ Gðb ω jþXit;b βjÞ:If b Pt>0;5our prediction will be that Yt¼1and if b Pt≤0;5our prediction will be Yt50 The percentage of times the observed value of Ytmatches the prediction is the percentage of correct predictions. It is worth noting that in the case of the Logit model, a cumulative logistic function with robust errors will be calculated. The logistic distribution function is of the form: IJSSP Pr½Yit ¼1ðXÞ� ¼ FðXβÞ ¼ eXβ �1þeXβ�(4) While the distribution function in the case of the Probit model corresponds to: Pr½Yit ¼1ðXÞ� ¼ FðXβÞ ¼ ZXβ −∞ θðzÞdz ¼θðXβÞ(5) The distribution function represented in equations 4 and 5 allows the predicted probability to fall within the range [0,1], ensuring that the probability of occurrence is defined through a nonlinear function. Thus, estimates using these specifications will lie within the desired probability range. Considering the definition of the Logit and Probit methodologies, the derivation of sample estimators for these models is carried out through maximum likelihood (ML). The objective is to derive estimators for an unknown parameter vector βthrough functions fðXβÞdefining Xit and enabling the maximum probability for the probability density function. For the Logit and Probit models, we consider a joint function as follows: lðβÞ ¼ Y n it¼1 GðYitjX;βÞ(6) The equation 4 represents the likelihood function. Since it involves the product of probabilities, the likelihood function is always less than 1, and its logarithm is negative. Through some algebraic operations, we can derive ln ðlðβÞÞto find the estimators bβwith the purpose of obtaining minimum variance, unbiased, and consistent estimators. It’s worth noting that the error distribution of both Logit and Probit models requires the use of the Z statistic for individual significance analysis. Additionally, for testing global significance, the likelihood ratio statistic is used. Considering the specification, some aspects need to be clarified. Since probabilistic models do not correct heteroscedasticity issues, in this study, we proceed to estimate with robust errors to heteroscedasticity to improve the efficiency of the estimators. Secondly, even when the models are estimated separately, the estimators will be similar since the cumulative distribution functions are also similar. Thirdly, the interpretation of marginal effects is not equivalent to a marginal in a linear model; in this case, they are interpreted as the marginal change in probability given a change in the vector of independent variables. Additionally, odds ratios (OR) will be calculated, which are derived from the Logit model with the intention of determining if there is an association between the outcomes in terms of human capital accumulation, productivity, inequality, labor income share in GDP, and the outcome of presenting high persistence in unemployment duration versus low persistence. Specifically, the OR compares the probabilities of an event occurring (presenting high persistence) between two groups: one exposed to a certain factor and the other not exposed conditioned on the vector of explanatory variables. (See Table 3 Odds ratio). As mentioned previously, as a measure of goodness of fit, the percentage of correct predictions for zeros and ones [2] will be separately calculated. Correct predictions are those that are equal to the observed data when they coincide: Y¼0;b Y¼0;Y¼1;b Y¼1. Therefore, the percentage of correct predictions results from counting the number of correct predictions and dividing it by the total number of observations [3]. Similarly, the Pseudo R2 (McFadden’s) will be considered, based on the logarithm of the estimated likelihood function: International Journal of Sociology and Social Policy Pseudo −R2¼1−lðb BÞ lðb B0Þ where lðc BÞIt is the logarithm of the likelihood function for the estimated model and lðc B0Þit is for a model with only a constant term, given that jlðc BÞj <jlðc B0Þj, The value of the Pseudo R 2 is between 0 and 1. 3.3 Sample selection In this section, the sample selection method for the analyzed OECD countries is detailed. The LAPU was calculated for all 38 OECD countries during 2013–2019 and ranked from highest to lowest. Ten countries with high LAPU and ten with low LAPU were identified, compared to the group’s average. The remaining 18 countries, whose LAPU levels were close to the average and not statistically different, were excluded. The long-term LAPU average for the 38 countries was 0.35104, with a 95% confidence interval of 0.23457–0.46751. To verify the suitability of the two analysis groups (G1: low persistence in unemployment duration, and G2: high persistence), statistically significant differences in the explained variables were analyzed. This empirical analysis justifies the group selection and outcome variables. Based on LAPU measurements, mean differences in development variables were analyzed. Table 1 (below) shows statistically significant differences between the OECD groups (low vs. high persistence) in the Human Capital Index (HCI), the welfare-relevant TFP at current PPPs (cwtfp), labor income (labsh), and the Gini coefficient, with a 99% confidence level. 4. Results and discussion Based on the preliminary results, the differences in the means of the variables of interest are analyzed. Figure 1 allows us to visualize the mean differences between groups of countries (those with Low Lapu vs. High Lapu), thereby supporting the results presented in Table 1 for the Two-sample mean t-test with equal variances. Notably, for the group of countries with high persistence in unemployment duration, the mean of the variables: Human Capital Index (HCI), levels of Welfare-relevant Total Factor Productivity at current Purchasing Power Parity (cwtfp), the share of labor income of employees and self-employed workers in GDP (labsh) and the Gini coefficient. These explanatory variables are part of the set that determines the probability of a country in the sample being classified in the group with high persistence in unemployment duration or with low persistence. Table 1 shows statistically significant differences (with a 99% confidence level and a 1% significance level) between the OECD country groups (low Group HC_mean CWTFP mean LABSH mean GINI mean Low persistence in unemployment duration 3.5664 0.8323 0.5840 0.3262 High persistence in unemployment duration 3.1579 0.7184 0.5099 0.3682 Difference 0.4085*** 0.0153*** 0.0098*** �0.0420*** H0: diff 50pvalue: 0.0000 pvalue: 0.0000 pvalue: 0.0000 pvalue: 0.0000 Obs 140 140 140 140 Note(s): (***) significant at 1%, (**) significant at 5%, (*) significant at 10% Source(s): Authors’ own work based on data from 20 OECD countries form Penn world table and ILO data base Table 1. Two-sample mean t-test with equal variances IJSSP persistence in unemployment duration vs high persistence in unemployment duration) regarding the Human Capital Index (HCI), the levels of Welfare-relevant Total Factor Productivity at current Purchasing Power Parity (cwtfp), the share of labor income of employees and self-employed workers in GDP (labsh), and the Gini coefficient. In Figures 2 and 3, scatter plots are depicted showing the relative position of countries compared to the historical mean (2013–2019) of long-term unemployment persistence (LAPU), along with the variables (hc), (labsh), (cwtfp), and (gini) respectively. Two contrasting groups are observed. Group 1 (Mexico, Chile, Costa Rica, and Turkey) exhibits a small cluster with high unemployment persistence rates but low levels of human capital accumulation, contrasting with their high inequality, as measured by the (gini) index. On the other hand, Group 2 (Germany, the Netherlands, Canada, the United Kingdom, Denmark, and Figure 1. Mean difference comparison between groups International Journal of Sociology and Social Policy It was found that countries with higher human capital index, greater labor income share in GDP, and more relevant productivity for well-being tend to reduce the probabilities of experiencing high persistence in unemployment duration. These findings support our initial hypothesis and suggest that investing in human capital, improving income distribution, and increasing productivity are effective strategies to combat unemployment duration persistence. It is important to note that although the Gini coefficient did not prove to be statistically significant as an explanatory variable for the probability of high persistence in unemployment duration, its inclusion in the analysis offers valuable insights into the role of inequality in unemployment dynamics. According toempirical literature on OECD countries, economic inequality has little influence on unemployment behavior, and vice versa. Furthermore, the comparative evaluation of goodness of fit between Probit and Logit models revealed that both present a similar ability to correctly classify observations, with a correct classification rate exceeding 87% in both models. This suggests that the choice between these models may depend more on specific methodological or conceptual considerations than on substantial differences in predictive accuracy. Visualizing the estimated probabilities by country and year allows for a better understanding of trends in unemployment persistence over time and provides useful information for both short and long-term policy formulation. Some countries, such as Chile (CHL), Italy (ITA), and Turkey (TUR), show an increasing trend in their probabilities of high persistence in unemployment duration over the years. This evidence suggests a potentially high risk of an increase in persistent unemployment duration in these nations unless effective measures are implemented to improve outcomes in terms of human capital, productivity, and labor income share in GDP. On the other hand, Costa Rica (CRI), Estonia (EST), Slovakia (SVK), Czech Republic (CZE), Lithuania (LTU), Poland (POL), and Israel (ISR) show a marked downward trend in these probabilities. This suggests that the risk of high persistence in unemployment duration could decrease in the coming years if they continue to improve their human capital, productivity, and labor income share in GDP indicators. However, some countries exhibit structurally high probabilities of high persistence in unemployment duration, such as Mexico (MEX), Greece (GRC), Italy (ITA), and Turkey (TUR). These nations face significant challenges in the labor market that require specific and urgent policy measures to address unemployment persistence. Finally, there are countries that present structurally low probabilities of high persistence in unemployment duration, suggesting minimal risk of experiencing this phenomenon. Examples of these countries include the United Kingdom (GBR), Denmark (DNK), Sweden (SWE), Norway (NOR), Netherlands (NLD), Germany (DEU), United States (USA), and Canada (CAN). These nations can serve as models for understanding and adopting effective practices in labor market management and reducing unemployment persistence. Based on the findings of the study, two key development policies are proposed. Firstly, the implementation of investment programs in Human Capital to increase productivity is recommended. Resources should be directed towards initiatives that improve the necessary skills and competencies in the labor markets of OECD countries, especially in strategic economic sectors with higher production linkages. Additionally, incentivizing the application of active labor policies is proposed. This entails prioritizing policies aimed at increasing the labor income share in GDP through progressive fiscal reforms that strengthen social safety nets and ensure fair labor standards. Implementing employment programs targeted at vulnerable groups, such as long-term unemployed individuals, youth, female heads of households, and marginalized communities, is also recommended to eliminate structural barriers to labor market participation and reduce disparities in unemployment persistence. Adopting these policies can help mitigate the risk of high unemployment duration persistence and foster sustainable and inclusive long-term economic growth. IJSSP Finally, this study presents some limitations that should be considered when interpreting the results. Although efforts have been made to provide a comprehensive analysis of unemployment persistence in OECD countries. First, the study is based on the analysis of LAPU levels, which means that many other variables that could affect the results have not been considered. For example, factors such as government policies, general economic conditions, demographic trends, and individual characteristics of the unemployed can influence unemployment levels and have not been considered in this study. Notes 1. A persistent behavior of the unemployment rate occurs due to a phenomenon in unemployment called hysteresis. It implies that the series follows a “path dependence,” or in other words, it depends on its trajectory (see Blanchard and Portugal (2001) and Blanchard (2018)). 2. The percentage of correct predictions is a useful measure to establish whether the estimates are consistent with the observed data. This procedure involves creating a dummy variable against which predicted values can be contrasted with observed ones, considering the predicted probability given the vector of explanatory variables. 3. PPC ¼obs1þobs2 OBS , Where PPC is the percentage of correct predictions, obs1is the number of correct predictions for Y50 and obs4are the correct predictions for Y¼1, and OBS is the total number of observations. References Abraham, K.G., Haltiwanger, J., Sandusky, K.y and Spletzer, J.R. (2019), “The consequences of longterm unemployment: evidence from linked survey and administrative data”, Industrial and Labor Relations Review, Vol. 72 No. 2, pp. 266-299, doi: 10.1177/0019793918797624. Afonso, � O. (2016), “Effects of labour-market institutions on employment, wages, R & D intensity and growth in 27 OECD countries: from theory to practice”, Economic Modelling, Vol. 53, pp. 48-62, doi: 10.1016/j.econmod.2015.11.009. Akdede, S.H. and G€ ocekli, S.G. (2019), “Unionization and labor share of income distribution: an empirical investigation of OECD countries”, International Journal of Public Finance, Vol. 4 No. 1, pp. 39-48, doi: 10.30927/ijpf.588970. Apergis, E. and Aergis, N. (2020), “Long-term unemployment: a question of skill obsolescence (updating existing skills) or technological shift (acquiring new skills)?”, Journal of Economic Studies, Vol. 47 No. 4, pp. 713-727, doi: 10.1108/JES-12-2018-0424. Arenas, C.D.C., Barreto, D.O. and Cabezas, D.M. (2024), “The short-term effects of COVID-19 on the manufacturing industry in Colombia: unveiling insights through a natural and exogenous experiment”, Journal of Knowledge Economics, doi: 10.1007/s13132-024-01927-8. Arestis, P., Ferreiro, J. and G� omez, C. (2023), “Labour market flexibilization and income distribution in Europe”, University of Cambridge, Department of Land Economy, University of the Basque Country UPV/EHU, Department of Public Policies and Economic History, Cambridge, UK, Bilbao, Spain. doi: 10.2298/PAN2102167A. Bahmani-Oskooee, M., Chang, T.y and Ranjbar, O. (2017), “Testing hysteresis effect in U.S. state unemployment: new evidence using a nonlinear quantile unit root test”, Applied Economics Letters, Vol. 2017 No. 4, pp. 1-5, doi: 10.1080/13504851.2017.1316477. Baker, M. (1992), “Unemployment duration: compositional effects and cyclical variability”, The American Economic Review, Vol. 82 No. 1, pp. 313-321, available at: http://www.jstor.org/stable/2117618 Becker, G.S. (1964), Human Capital: A Theoretical and Empirical Analysis with Special Reference to Education, 1st ed., National Bureau of Economic Research (NBER), available at: https://www. nber.org/books-and-chapters/human-capital-theoretical-and-empirical-analysis-specialreference-education-first-edition International Journal of Sociology and Social Policy Bhattarai, K. (2016), “Unemployment–inflation trade-offs in OECD countries”, Economic Modelling, Vol. 58, pp. 93-103, doi: 10.1016/j.econmod.2016.05.007. Blanchard, O. (2018), “Should we reject the natural rate hypothesis?”, The Journal of Economic Perspectives, Vol. 32 No. 1, pp. 97-120, doi: 10.1257/jep.32.1.97. Blanchard, O. and Portugal, P. (2001), “What hides behind an unemployment rate: comparing Portuguese and U.S. Labor markets”, The American Economic Review, Vol. 91 No. 1, pp. 187-207, doi: 10.1257/aer.91.1.187, available at: http://www.jstor.org/stable/2677904 Blanchard, O. and Summers, L. (1986), “Hysteresis and the European unemployment problem”, NBER Macroeconomics Annual, Vol. 1, doi: 10.1086/654013. Blundell, R. and Dias, M.C. (2009), “Alternative approaches to evaluation in empirical microeconomics”, Journal of Human Resources, Vol. 44 No. 3, pp. 565-640, doi: 10.3368/jhr.44.3.565. Burdett, K., Carrillo-Tudela, C. and Coles, M.G. (2011), “Human capital accumulation and labor market equilibrium”, International Economic Review, Vol. 52 No. 3, pp. 657-677, doi: 10.1111/j.14682354.2011.00644.x. Cardona-Arenas, C.D. and Sierra-Suarez, L.P. (2023), “Remittances, nonlabor income as a source of hysteresis in unemployment in Colombia, 2010-2020”, Latin American Research Review, Vol. 59 No. 2, pp. 1-26, doi: 10.1017/lar.2023.57. Carling, K., Edin, P.-A., Harkman, A. and Holmlund, B. (1996), “Unemployment duration, unemployment benefits, and labor market programs in Sweden”, Journal of Public Economics, Vol. 59 No. 3, pp. 313-334, doi: 10.1016/0047-2727(95)01499-301499-3. Coles, M. and Masters, A. (2000), “Retraining and long-term unemployment in a model of unlearning by not doing”, European Economic Review, Vol. 44 No. 9, pp. 1801-1822, doi: 10.1016/S0014-2921(99)00005-7. Cruz, M.D. (2023), “Labor productivity, real wages, and employment in OECD economies”, Structural Change and Economic Dynamics, Vol. 66, pp. 367-382, doi: 10.1016/j.strueco.2023.05.007. Cysne, R.P. and Turchick, D. (2012), “Equilibrium unemployment-inequality correlation”, Journal of Macroeconomics, Vol. 34 No. 2, pp. 454-469, doi: 10.1016/j.jmacro.2011.12.009. Dawkins, C.J., Shen, Q. and Sanchez, T.W. (2005), “Race, space, and unemployment duration”, Journal of Urban Economics, Vol. 58 No. 1, pp. 91-113, doi: 10.1016/j.jue.2005.02.001. Dixon, R., Lim, G.C. and van Ours, J.C. (2017), “Revisiting the Okun relationship”, Applied Economics, Vol. 49 No. 28, pp. 2749-2765, doi: 10.1080/00036846.2016.1245846. Dorofeev, M.L. (2022), “Interrelations between income inequality and sustainable economic growth: contradictions of empirical research and new results”, Economies, Vol. 10 No. 2, p. 44, doi: 10. 3390/economies10020044. Evans, G. (1989), “Output and unemployment dynamics in the United States: 1950-1985”, Journal of Applied Econometrics, Vol. 4 No. 3, pp. 213-237, Jul.-Sept, doi: 10.1002/jae.3950040302. Feenstra, R.C., Inklaar, R. and Timmer, M. (2015), “The next generation of the Penn world table”, The American Economic Review, Vol. 105 No. 10, pp. 3150-3182, doi: 10.1257/aer.20130954. Fitzgerald, T.J. (1998), “An introduction to the search theory of unemployment”, Economic Review, Vol. 34 No. 3, pp. 2-15, available at: https://core.ac.uk/download/pdf/6608833.pdf Focacci, C.N. and Perez, C. (2022), “The importance of education and training policies in supporting technological revolutions: a comparative and historical analysis of UK, US, Germany, and Sweden (1830-1970)”, Technology in Society, Vol. 70, p. 102000, doi: 10.1016/j.techsoc.2022.102000. Franz, W. (1990), “Hysteresis in economic relationships: an overview”, in Hysteresis Effects in Economic Models, Physica-Verlag HD. doi: 10.1007/978-3-642-51543-9_1. Furuoka, F. (2017), “A new test for analysing hysteresis in European unemployment”, Applied Economics Letters, Vol. 24 No. 15, pp. 1102-1106, doi: 10.1080/13504851.2016.1257209. G€ ocke, M. (2002), “Various concepts of hysteresis applied in economics”, Journal of Economic Surveys, Vol. 16 No. 2, pp. 167-188, doi: 10.1111/1467-6419.00163. IJSSP Groningen Growth and Development Centre (GGDC) (2024), “PWT 10.01jPenn world table”, available at: https://www.rug.nl/ggdc/productivity/pwt/?lang5en Halter, D., Oechslin, M. and Zweim€ uller, J. (2014), “Inequality and growth: the neglected time dimension”, Journal of Economic Growth, Vol. 19 No. 1, pp. 81-104, doi: 10.1007/s10887-013-9099-8. Hasell, J., Arriagada, P., Ortiz-Ospina, E.y, Roser, M. and Thornton, T.B. (2024), “Economic inequality”, Our World in Data, pp. 432-461, doi: 10.4324/9781003455752-19, available at: https://ourworldindata.org/economic-inequality Heimberger, P., Kapeller, J. and Sch€ utz, B. (2017), “The NAIRU determinants: what’s structural about unemployment in Europe?”, Journal of Policy Modeling, Vol. 39 No. 5, pp. 883-908, doi: 10.1016/j. jpolmod.2017.04.003. Kaldor, N. (1966), Causes of the Slow Rate of Economic Growth of the United Kingdom: An Inaugural Lecture, Cambridge University Press, Cambridge. Kaldor, N. (1967), Strategic Factors in Economic Development, Cornell University Press, Ithaca. Kaldor, N. (1968), “Productivity and growth in manufacturing industry: a reply”, Economica, Vol. 35 No. 140, pp. 385-391, doi: 10.2307/2552347. Kim, Y.E. and Loayza, N.V. (2019), “Productivity growth: patterns and determinants across the world”, World Bank Policy Research Working Paper No. 8852. Washington, DC: World Bank, Vol. 42 No. 84, pp. 36-93, doi: 10.18800/economia.201902.003, available at: https:// openknowledge.worldbank.org/handle/10986/31710 Kim, Y.E., Loayza, N. V. and Meza-Cuadra, C.M. (2016), “Productivity as the key to economic growth and development”, World Bank Research Policy Brief No. 3, available at: https://ssrn.com/ abstract53249552 Kriaa, F., Bouhari, M. and Mathlouthi, Y. (2020), “Determinants of unemployment duration for young men and women in Tunisia”, Economics, Management and Sustainability, Vol. 5 No. 2, pp. 78-95, doi: 10.14254/jems.2020.5-2.51. Lauder, H. and Mayhew, K. (2020), “Higher education and the labour market: an introduction”, Oxford Review of Education, Vol. 46 No. 1, pp. 1-9, doi: 10.1080/03054985.2019.1699714. Le Clech, N. and Guevara-P� erez, J.C. (2023), “Latin America and the Caribbean’s productivity: the role of pro-market policies, institutions, infrastructure, and natural resource endowments”, Economies, Vol. 11 No. 5, p. 142, doi: 10.3390/economies11050142, available at: https://www. mdpi.com/2227-7099/11/5/142 Li, H. and Zou, H. (1998), “Income inequality is not harmful for growth: theory and evidence”, Review of Development Economics, Vol. 2, pp. 318-334, doi: 10.1111/1467-9361.00045. Marques, A.M., Lima, G.T. and Troster, V. (2017), “Unemployment persistence in OECD countries after the great recession”, Economic Modelling, Vol. 64, pp. 105-116, doi: 10.1016/j.econmod. 2017.03.014. Mart� ınez, R., Ayala, L. and Ruiz-Huerta, J. (2001), “The impact of unemployment on inequality and poverty in OECD countries”, The Economics of Transition, Vol. 9 No. 2, pp. 417-447, doi: 10. 1111/1468-0351.00082. Mathy, G.P. (2017), “Hysteresis and persistent long-term unemployment: the American beveridge curve of the great depression and World War II”, Cliometrica, Vol. 2017, pp. 1-26, doi: 10.1007/ s11698-016-0158-1. Murtin, F. and Robin, J.-M. (2018), “Labor market reforms and unemployment dynamics”, Labour Economics, Vol. 50, pp. 3-19, available at: https://doi:10.1016/j.labeco.2016.05.025 Narendranathan, W., Nickell, S. and Stern, J. (1985), “Unemployment benefits revisited”, The Economic Journal, Vol. 95 No. 378, p. 307, doi: 10.2307/2233212. Nickell, S. (1997), “Unemployment and labor market rigidities: Europe versus North America”, The Journal of Economic Perspectives, Vol. 11 No. 3, pp. 55-74, doi: 10.1257/jep.11.3.55, available at: http://www.jstor.org/stable/2138184 International Journal of Sociology and Social Policy Organization for Economic Co-operation and Development - OECD (1993), Employment Outlook, Paris, doi: 10.1787/19991266. Organization for Economic Co-operation and Development - OECD (2013), OECD Employment Outlook 2013, OECD Publishing, doi: 10.1787/empl_outlook-2013-en. Paserman, M.D. (2004), “Job search and hyperbolic discounting: structural estimation and policy evaluation”, IZA Discussion Papers, No. 997, Institute for the Study of Labor (IZA), available at: https://docs.iza.org/dp997.pdf Pissarides, C.A. (1992), “Loss of skill during unemployment and the persistence of employment shocks”, Quarterly Journal of Economics, Vol. 107 No. 4, pp. 1371-1391, doi: 10.2307/2118392. Rogers, C.L. (1997), “Job search and unemployment duration: implications for the spatial mismatch hypothesis”, Journal of Urban Economics, Vol. 42 No. 1, pp. 109-132, doi: 10.1006/juec.1996.2017. Romer, P. (1986), “Increasing returns and long-run growth”, Journal of Political Economy, Vol. 94 No. 5, pp. 1002-1137, doi: 10.1086/261420. Rotar, L.J. and Krsnik, S. (2020), “Analyzing the relationship between unemployment benefits and unemployment duration”, Society and Economy, Vol. 42 No. 3, pp. 280-297, doi: 10.1556/204. 2020.0000945. Sakamoto, T. (2002), “Economic performance of ‘weak’ governments and their interaction with central banks and labour: deficits, economic growth, unemployment and inflation, 1961-1998”, European Journal of Political Economy, Vol. 18 No. 3, pp. 445-476, available at: https://papers.ssrn.com/sol3/ Delivery.cfm/SSRN_ID4021626_code2720780.pdf?abstractid54021626andmirid51 Scarchilli, G. and Triventi, M. (2023), “Individual circumstances and income distribution in a changing labour market: Italy, 2005-2019”, available at: https://www.ecineq.org/wp-content/ uploads/2023/06/EcineqAMSE-345.pdf Sturn, S. (2013), “Are corporatist labour markets different? Labour market regimes and unemployment in OECD countries”, International Labour Review, Vol. 152 No. 2, pp. 237-254, doi: 10.1111/j.1564-913X.2013.00179.x5. Summers, R. and Heston, A. (1991), “The Penn World Table (Mark 5): an expanded set of international comparisons, 1950–1988”, Quarterly Journal of Economics, Vol. 106 No. 2, pp. 327-68. Topel, R. and Welch, F. (1980), “Unemployment insurance: survey and extensions”, Economica, Vol. 47 No. 187, pp. 351-379, doi: 10.2307/2553156. Uzawa, H. (1965), “Optimum technical change in an aggregative model of economic growth”, International Economic Review, Vol. 6 No. 1, pp. 18-31, doi: 10.2307/2525621. Van Ours, J.C. (2015), “The great recession was not so great”, CentER Discussion Paper Series No. 2015-006, Vol. 34 pp. 1-12, doi: 10.2139/ssrn.2555484. Webster, D. (2005), “Long-term unemployment, the invention of ‘hysteresis’ and the misdiagnosis of structural unemployment in the UK”, Cambridge Journal of Economics, Vol. 29 No. 6, pp. 975995, doi: 10.1093/cje/bei082. Zamanzadeh, A., Chan, M.K., Ali Ehsani, M. and Ganjali, M. (2019), “Unemployment duration, Fiscal and monetary policies, and the output gap: how do the quantile relationships look like?”, Economic Modelling, Vol. 91, pp. 613-632, doi: 10.1016/j.econmod.2019.12.003. Corresponding author Carlos David Cardona-Arenas can be contacted at: [email protected] For instructions on how to order reprints of this article, please visit our website: www.emeraldgrouppublishing.com/licensing/reprints.htm Or contact us for further details: [email protected] IJSSP