Unemployment Benefits, Active Labor Market Policies, and Labor Market Outcomes Evidence from New Global Data

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


Introduction
Unemployment benefits (UB) and active labor market policies (ALMP) have gained widespread recognition in both developed and developing economies to improve labor market conditions and address social challenges. 1 Despite their effectiveness in cushioning vulnerable groups from shocks, their role in significantly bolstering overall employment and productivity remains ambiguous.Theoretically, UB and ALMP have the potential to increase both employment and productivity by extending job search time through income support, offering active job search support, and enhancing skills through training programs, among other benefits.However, these policies can also strain government budgets, displace or substitute productive jobs, and result in deadweight loss.Therefore, their collective impact on labor market outcomes hinges on which effects dominate at the macro level (Gautier et al. 2018).
Current research investigating the interlinkages between UB and labor market outcomes generally suggests that longer UB periods lead to prolonged unemployment duration and lower formal employment in both developing and developed economies (e.g., Filiz 2017;Hagedorn et al. 2015).The evidence regarding the impact of UB on job quality and earnings primarily concentrates on developed countries and generally indicates a positive association (Caliendo et al. 2013;Tatsiramos 2009).Only two studies have evaluated the relationship between UB and productivity and both indicate a positive relationship between them (Soltes 2017;Acemoglu and Shimer 2000).
Most studies analyzing ALMP use micro-level data to evaluate the effects of a specific program on participants' employment and earnings.Findings indicate that the effects of ALMP on labor market outcomes vary based on program type and design.For instance, evidence from both developed and developing countries suggests that job search assistance, placement, and training programs generally increase participant's employment rates.However, these effects tend to be contingent on program design and differ across participant groups (e.g., Levy Yeyati et al. 2019;Autor et al. 2017;Card et al. 2018;Groh et al. 2012).Research provides limited support for the notion that ALMP enhance earnings (Betcherman et al. 2005;Levinsohn and Leibbrandt 2014).Among the rare macro-level studies, Caicedo et al. (2022) indicate an overall positive impact of ALMP on output and welfare, but also highlight their significant distortionary effects.
Drawing on new global data from the World Bank that covers 191 countries in 2019 and 2020, this paper aims to provide an analysis of the patterns of UB and ALMP across different regions and income groups and investigate their association with labor market outcomes. 2It also utilizes the comprehensive labor market and social assistance policy inventory of the World Bank to present the patterns of UB and ALMP reforms implemented in 223 economies throughout the Covid-19 pandemic.Econometric analysis of the linkages between UB, ALMP and labor market outcomes is conducted using new data from 2019 and the ordinary least squares (OLS) regression model that incorporates interaction terms between income groups and UB and ALMP and includes a wide range of control variables.
The paper contributes to the literature in three ways: first, it is the first study that covers a large number of developing as well as developed countries to analyze the interlinkages between UB and ALMP and labor market outcomes; second, it utilizes rich, novel data from the World Bank covering 191 economies in 2019 and 2020; third, it is the first study assessing the association of both UB and ALMP with employment, unemployment and productivity using macro data and distinguishing among different income groups.There is no previous study analyzing the relationship between ALMP and productivity, and there are only few studies assessing the association of UB with productivity, including Solter (2017) and Acemoglu and Shimer (2000).All other studies focus on the interlinkages between UB or ALMP with employment, unemployment, matching quality, or earnings, and the overwhelming majority of these studies focus on developed economies.
To the best of our knowledge, Pignatti and Van Belle (2018) is the only other study that uses data from both developed and developing economies-covering 36 developed, 64 emerging and 21 developing economies-to analyze the impact of ALMP and UB on unemployment and employment. 3The authors find that in developing countries ALMP spending decreases unemployment and increases employment, while the spending on passive labor market policies (PLMP), like UB, lacks a significant impact on either.They also point out that ALMP and PLMP in developing countries are substitutes, rather than complements, in promoting employment and reducing unemployment.In contrast, in developed economies, both ALMP and PLMP increase unemployment and decrease employment.However, there is a complementary relationship between them, in that when ALMP spending is sufficiently high, higher PLMP spending reduces unemployment and boosts employment.
Different from the work of Pignatti and Van Belle (2018), this study covers a larger number of developing and developed countries.In addition to employment and unemployment, it also investigates the linkages of UB and ALMP with productivity growth and incorporates an interaction term between income categories of countries and UB and ALMP.Moreover, it assesses the association of different sub indicators of UB and ALMP with labor market outcomes and controls for a wide range of relevant indicators put forward in the literature.
The key findings of the descriptive statistical analyses can be summarized as follows: • Of the 191 countries, 47.6 percent have UB schemes and 85.3 percent have ALMP.
• Among ALMP, public employment services (PES) are most prevalent (employed in 81.7 percent of countries), followed by national skills funds (52.9 percent) and government incentives offered for training (53.9 percent).• As income level decreases, the use of ALMP vis-à-vis UB increases substantially.
• On average, employees need to contribute at least 11.6 months to become eligible for UB, and the average maximum UB period is 37.3 weeks.• Among the six eligibility conditions for UB covered by the data, all of which are also ALMP, periodic reporting on active job search, participation in a training program, and registration with PES are most commonly used.Regarding the six services offered by PES, training programs, access to the vacancy database and job placement are most frequently used, followed by counseling and internship/apprenticeship programs.Support for geographic mobility is the least frequently used.• The eligibility conditions most likely to be used together are periodic reporting on active job search and participating in training offered by PES, followed by periodic reporting and registration with PES.As for the services offered by PES, job placement and access to the vacancy database are most likely to be offered together, followed by job placement and counseling services.• According to the World Bank's Social Protection and Jobs Policy Inventory that gathered data from March 2020 to January 2022, 223 economies implemented 3856 policies to safeguard workers, vulnerable groups, and economic activities from the Covid-19 crisis.Among these reforms, 159 were related to UB, 865 were related to ALMP and the rest was on social assistance and other labor market policies.• High-income economies led 62 percent of UB reforms, with upper middle-income and lower middle-income countries contributing 26 and 11 percent, respectively.• A larger proportion of low and lower middle-income economies implemented ALMP reforms during this period, with 8 percent and 17 percent, respectively, compared to 47 percent in highincome countries and 27 percent in upper middle-income countries.• Across all income groups, Covid-19 related UB reforms primarily targeted increasing UB amounts, followed by expanding coverage and extending payment durations.• Labor regulation adjustments were the most prevalent ALMP reform in all income groups.In upper-middle and high-income economies, wage subsidy programs and job search/training initiatives were the second and third most common reforms, respectively.Conversely, public works, while the least common overall, emerged as the second most popular reform in low and lower middle-income economies.
The econometric analysis of the relationship between labor market outcomes and UB/ALMP reveals distinct patterns across income groups: • In low-and lower-middle-income countries, ALMP show a stronger correlation with labor market outcomes compared to upper middle and high-income countries.Specifically, they exhibit positive associations with productivity growth and employment rate in the agriculture sector, and a negative association with employment rate in the services sector.• In upper middle-income countries, both ALMP and UB are positively linked to productivity growth.
UB are also positively associated with industrial employment rate but does not show any direct association with employment rates in agriculture or services sectors.Additionally, both UB and ALMP have a negative association with the aggregate employment rate.• High-income countries exhibit fewer statistically significant findings.Among these, UB are positively associated with agricultural employment rates, but negatively associated with industrial employment rates and productivity growth.• Across all income groups, ALMP consistently display a negative relationship with the selfemployment rate.
The remainder of the paper is organized as follows: Section 2 provides a review of the theoretical and empirical literature; Section 3 explains the data and analyzes the descriptive statistics and patterns of UB and ALMP indicators across regions and income groups; Section 4 presents and discusses the findings of the regression analyses, and Section 5 concludes the paper.

Review of the Literature on the Linkages of UB and ALMP with Labor Market Outcomes
This section provides an overview of theoretical and empirical literature investigating the effects of UB and ALMP on labor market outcomes in developing and developed countries.

UB and labor market outcomes
Conceptually, UB can affect unemployment and reemployment wages through two channels: the substitution effect and the liquidity effect.The substitution effect of UB arises from the distortion it creates in the prices of leisure and consumption, reducing the marginal incentive to search for a job and potentially leading to moral hazard that in turn promotes unemployment and reduces reemployment wage.The potential for moral hazard is expected to be higher in less developed economies due to a larger informal sector and limited institutional capacity to monitor the job search activities of UB recipients.The liquidity effect stems from the liquidity provided by UB, allowing unemployed individuals to effectively manage and sustain their consumption over an extended period.This prompts UB recipients to devote more time to seeking employment opportunities that align with their skills and offer higher wages.Unlike moral hazard, the liquidity effect proves socially advantageous, fostering better employment opportunities and productivity, even though it may temporarily elevate the unemployment rate in the short term.
Empirical studies of UB typically concentrate on Europe and North America and show that longer UB periods or higher amounts of UB result in prolonged unemployment and reduced formal employment, suggesting, to some degree, the presence of moral hazard (Hagedorn et al. 2015;Tatsiramos and Van Ours 2014;Caliendo et al. 2013;Farber and Valletta 2013;Rothstein 2011;Tatsiramos 2009;Van Ours and Vodopivec 2006).Using macro level panel data, Pignatti and Van Belle (2018) showed that in developed countries, when coupled with sufficiently high spending on ALMP, higher public spending on PLMP can cut unemployment and boost employment rates.However, in developing countries, increased public spending on PLMP did not significantly affect unemployment or employment rates.
Country-level research on developing economies suggests that while generous UB tend to increase unemployment and reduce their formal employment more prominently compared to developed economies, this is not necessarily due to moral hazard.For instance, in Turkey, a two-month extension of UB during 2002-2012 decreased the likelihood of job finding by 15 percent, surpassing effects observed in the United States, Germany, and Austria (Filiz 2017).However, individuals entitled to longer UB periods showed reduced tendency to exploit the system, indicating lower moral hazard.Similarly, an increase in UB duration in 2002-2010 led to a lower probability of returning to a formal job.Yet, its efficiency cost was lower than observed in the United States (Gerard and Gonzaga 2021).In Mauritius, Liepmann and Pignatti (2021) found positive and significant welfare effects of UB during 2012-2108 in a highly informal context, with smaller efficiency costs than in developed economies.In addition, the shift to informal employment was more influenced by layoffs than by UB generosity.Therefore, contrary to studies emphasizing moral hazard effects of UB in developing economies (e.g., Van Doornik et al. 2018), these studies underscore the scarcity of formal jobs in developing countries as a key factor reducing the likelihood of reemployment in the formal sector.
Another argument concerning the efficiency of UB in developing countries is the weaker institutional capacity of these countries in administrating and monitoring UB system.In the majority of developing countries, UB covers a small percentage of population and social security database is not well developed.Therefore, in these countries, protecting jobs -through stricter labor market regulations, such as higher severance pay or restrictions in hiring and firing policies -is more common than protecting workersthrough social protection or unemployment benefits (Vodopivec 2013;Holzmann et al. 2012;Asenjo and Pignatti 2019;and Ozkan 2020).High economic cost of strict labor regulations has been well established in the literature, leading to the adoption of "flex-security" policies by many countries that combines relaxing labor regulations and increasing social security and unemployment benefits.
For example, in Brazil, Britto (2022) found that severance pay, which is a one-time lump sum payment to unemployed individuals has more detrimental effect on formal employment and reemployment earnings than extension in UB period, as the latter provides more bargaining power to the job seeker and involves in training and job matching support.Gerard and Naritomi (2021) also showed that upon receiving lump sum severance payment at layoff, workers increased their consumption by 35 percent despite experiencing a 14 percent long-term loss, indicating bias for present time and the importance of the disbursement timing of unemployment and social protection.
In an effort to increase the efficiency of UB system in an environment with limited administrative capacity, Chile introduced a two-tiered UB program in 2002, which provided benefits through a combination of mandatory individual insurance accounts fully funded by individual contributions and solidarity funding.
The first tier would minimize disincentive and moral hazard and the second tier, funded by general tax, would protect vulnerable population without impeding formal jobs (Duval and Loungani, 2019).Other developing countries, such as Brazil, Colombia, Jordan, and Mauritius followed the suit.In Chile, the evaluation of the program revealed that an extension of UB period led to a longer duration of unemployment for beneficiaries utilizing the solidarity fund.However, for those dependent on selffunded individual insurance, there was no change in the duration of unemployment, suggesting the absence of moral hazard in the latter group (Hartley et al. 2010;Sehnbruch et al. 2022).
The impact of UB scheme on productivity and re-employment wages is mixed and the analyses focus on developed countries.On one hand a longer UB duration raises reservation wage as it gives UB recipients more time to search for better jobs and to receive training, on the other hand it lowers wages offered by employers.Research by Soltes (2017) using data from 16 developed countries suggests that a one-indexpoint increase in the UB generosity index is associated with about a one-percentage-point increase in total factor productivity (TFP) growth.In the United States, longer UB duration encouraged workers to seek higher productivity jobs, particularly for those with lower education levels, and promoted job creation (Acemoglu and Shimer 2000;Centeno and Novo 2006).In Austria, there was a 0.5 percent increase in reemployment wage (Nekoei and Weber 2017), while in Germany there was a 0.8 percent reduction in wage offers (Schmieder et al. 2016) and in France it had no discernible impact on subsequent match quality or wages (Le Barbanchon 2016).4

ALMP and labor market outcomes
The theoretical underpinnings of ALMP are multifaceted, as they encompass a wide range of programs and their effects cut through several economic and social areas.The main objective of ALMP is to increase labor supply and demand and improve labor market matching; however, they also have effects on productivity, output, poverty, inequality, and social cohesion, among others.PES programs are the most commonly used ALMP globally, which include the provision of vacancy database and labor market information, training programs, job search assistance, job placement, internship programs, wage subsidies, public works, and national training funds (Andersen et al. 2015).In countries that have UB schemes, ALMP are also used as a condition to receive UB to prevent moral hazard.Therefore, both the design and implementation of ALMP play a critical role in their success, as well as the institutional, economic, and socioeconomic environments of their respective countries.
Channels through which ALMP affect economic and social outcomes are covered widely in the literature.They help unemployed individuals find work by identifying their individual needs and goals through counseling; increasing their skill sets through training and on the job learning; facilitating job matching, and increasing competition.By enabling job seekers to develop human and social capital, they can also increase social inclusion and the productivity, output, and competitiveness of the labor force (Boone and Van Ours 2004;Calmfors 1995).However, ALMP can also have a negative impact on labor markets and the overall welfare of the economy if the jobs created by one program replace other jobs (displacement effect) or if they lead to the substitution of jobs in other sectors by changing the cost of wages through subsidies.Moreover, they may result in deadweight loss when the jobs they generate would have occurred naturally without any intervention and can lead to a lock-in effect, exacerbating unemployment (Calmfors 1995;Brown and Koettl 2015).
The vast majority of studies evaluating the impact of ALMP focus on micro-level analysis.They primarily concentrate on assessing how a specific ALMP program influences the employment opportunities of its participants within a defined timeframe.However, these studies often overlook the broader effects on non-participants and different sectors of the economy, as well as the associated costs of these programs.
In a limited number of macro studies, Graversen and Van Ours (2008) revealed that in Denmark frequent interactions between the unemployed and PES reduced the duration of unemployment among participants and increased their likelihood of securing employment.Additionally, Pignatti and Van Belle (2018) demonstrated contrasting patterns between developed and developing economies: in developing countries, increased public expenditure on ALMP was associated with lower unemployment and higher employment rates, whereas the trend reversed in developed economies when public spending on UB was low.
A comprehensive cross-country meta-analysis of 207 micro level studies of ALMP -comprising 155 OECD, 33 non-OECD, and 19 LAC studies, conducted by Card et al. (2018), produced several interesting outcomes.It demonstrated that, overall, job search programs consistently yielded a positive effect on employment, while the impact of human capital programs became notably significant only after two to three years following program completion.Moreover, it highlighted that public sector employment programs exhibited either negligible or adverse effects across all time frames.Importantly, the effects of the ALMP programs tended to be more pronounced for female participants and individuals experiencing long-term unemployment, and the efficacy of the ALMP interventions was higher during economic downturns.
Another meta-analysis by Betcherman et al. (2004) indicated that training programs for unemployed adults contribute to a higher employment rate, but do not necessarily translate into increased earnings, especially in developing nations.Additionally, the study revealed that retraining initiatives for workers affected by mass layoffs do not yield substantial positive effects on employment.Similarly, training programs tailored for youth in the least developed economies lack significant impact on enhancing employment opportunities.
Country-level studies of ALMP in developing economies typically reveal a positive impact of job search and placement support on labor market outcomes.For instance, Novella and Valencia (2022) showed that a comprehensive employee support program in Bolivia led to 9 and 4 percentage points increase in the probability of overall employment and formal employment, respectively, and a 10 percent rise in earnings.The program's effects were more pronounced among adults, individuals with tertiary education, and women.Abel et al. (2022) studied a temporary wage incentive for secondary school graduates in Mexico and demonstrated a notable increase in permanent formal employment, earnings, and human capital accumulation among program participants.In South Africa, an employment voucher initiative led to a 15-25 percent increase in youth employment probabilities, sustained even two years after the voucher program concluded (Levinsohn and Leibbrandt 2014).Conversely, in Jordan, a six-month voucher supporting wages for female graduates initially resulted in a significant employment rate surge, which dwindled after the voucher period ended (Groh et al. 2012).
Likewise, most research on training programs in developing countries demonstrate evidence for their positive effect on the labor market outcomes of participants.For example, in Brazil a large-scale apprenticeship program in 2000 increased the probability of obtaining permanent jobs in the short term only for former apprentices in high skill occupations, but the effects were similar for all participants in the medium term (Corseuil et al. 2019).In addition, in the medium-term, turnover rates in formal employment decreased and participants who had their first job in large firms or were former apprentices benefited the most from the program.Similarly, in Tunisia, national vocational training in 2001 increased the probability of participants to find a job and their earnings (Bellakhal and Mahjoub 2015).In Dominican Republic, training program promoted the personal skills of the participants, however, only female participants had better labor market outcomes (Acevedo et al. 2020).
One of the few studies investigating the full macro level impacts of ALMP, Caicedo et al. (2022), assessed the effects of a Colombian reform program implemented in 2003.The program required firms to train apprentices by quotas set by firm size and lowered the minimum wage of apprentices.They found that firms in sectors requiring high skills inclined to avoid training apprentices due to high cost of training in these sectors, while firms in low-skill sectors sought apprentices.They also showed that in general equilibrium the reform increased the overall output of the economy and the welfare of apprentices in all sectors, but it reduced the welfare of firms in high skilled sectors and workers in low skilled sectors, indicating a replacement of a higher wage employee with a lower wage apprentice.
Country level studies of ALMP in developed countries reveal mixed results.For instance, PES participation in France led to increased employment rates (Fougère et al. 2005), while a reemployment program in the US state of Nevada raised participant earnings (Manoli et al. 2018).However, in the US city of Detroit, job placement programs increased earnings solely through direct-hire placements, whereas temporary help placements had zero or negative effects on earnings (Autor et al. 2017).Moreover, in France and Denmark, although job search assistance increased the employment of participants, it also decreased the employment of non-participants (Crepon et al. 2012;Gautier et al. 2018).As for training programs in developed countries, Kluve (2010) discovered that integrated training programs in Europe increased employment rates by 40-50 percent compared to traditional methods.Yet, youth-focused programs had a 40-60 percent lower success rate in enhancing employment compared to adult programs.

Description of UB and ALMP Data and Stylized Facts
This section details the novel UB and ALMP data utilized in the paper.It includes a statistical analysis to reveal patterns and correlations with other economic indicators.In addition, drawing on World Bank's Social Protection and Jobs Policy Inventory data, it offers a descriptive analysis of the UB and ALMP reforms implemented during the Covid-19 pandemic.

Description of the data
UB and ALMP data were collected through questionnaires distributed to 900 experts in 191 economies in 2019 and 2020. 5To ensure the comparability of data across economies and over time, the questionnaires are based on a standard assumption that the UB recipient is a 19-year-old former full-time cashier in a supermarket or a grocery store in the largest business city, who has been made redundant after one year of work due to economic reasons.S/he is single with no dependents, not a member of a labor union, unless membership is mandatory, and is not granted more benefits than those mandated by law, regulations, or (if applicable) collective bargaining agreements.
The questions on the UB scheme include the availability and maximum duration6 of UB after one, five, and ten years of employment, the minimum contribution period to be eligible for UB, and eligibility requirements as stipulated in the labor laws of the country.The questions on ALMP include the availability of PES and the types of PES offered; the availability of a national skills/training fund, and government incentives for firms to provide training to their employees.
It is important to note that there is no notable change in UB and ALMP indicators from 2019 to 2020.This can be attributed to two factors.First, data is based on labor laws, which typically do not undergo frequent changes.Second, data collection cycle spanned from May of the preceding year to May of the subsequent year for both years.Consequently, the 2020 data does not adequately capture the majority of the UB and ALMP reforms related to the Covid-19 crisis, which commenced in March 2020.Moreover, the bulk of the UB and ALMP reforms implemented between March and May 2020 fall outside the scope of this dataset, including changes in UB amounts, wage subsidies, public works initiatives, and unemployment assistance, among others.The data collected in 2020 captured reforms in 12 economies between May 2, 2019 and May 1, 2020.Ten of these economies undertook reforms in response to the Covid-19 pandemic, but they did not lead to a significant change in the overall data.7 Therefore, to examine the UB and ALMP reforms carried out during the Covid-19 pandemic, the paper utilizes the comprehensive Social Protection and Jobs Policy Inventory of the World Bank collected during March 2020-January 2022.to incentivize firms to provide training to their employees (67 percent of economies), followed by subsidies (53.4 percent), and cost reimbursement (46.6 percent).

Patterns of UB and ALMP across income groups and regions
Among the 91 economies with UB schemes, only 8 percent, all of which are HIC, require unemployed persons to meet all six criteria, and 5 percent require just one criterion.The most commonly used number of eligibility conditions is four, followed by three, five, and two (30, 22, 19, and 16 percent, respectively).
According to the Jaccard index of similarity,9 periodic reporting on active job search and participating in training offered by PES are most likely to be used together (with a Jaccard similarity index of 81 percent), followed by registration with PES and periodic reporting on active job search (68 percent), and registration with PES and participation in training offered by PES being used together (61 percent).10 Regarding the six services provided by PES reported in Table 1, training, vacancy database, and job placement are the most commonly used services.Among the 156 economies that have PES, 29 percent offer all six services, 27 percent offer five services, and about an equal percentage of countries, at about 13 percent, offer 4, 3, and 2 employment services.Only 0.4 percent of economies offer just one employment service.The highest percentage of countries offering all six services are from HIC, at 42 percent of HIC, compared to 26, 22, and 13 percent of UMIC, lower-middle-income countries (LMIC), and low-income countries (LIC), respectively.Interestingly, the largest share of economies offering five services are from LMIC (at 31 percent of LMIC) followed by HIC (27 percent) and UMIC (26 percent), and the majority of countries with only two services are from LIC (at 38 percent).As for the pairing of these services, job placement services and a vacancy database are most likely to be offered together (with a Jaccard similarity index of 81 percent), followed by job placement and counseling services (78 percent), and counseling services and training (77 percent).11 Out of the three types of training incentives offered by the governments of 95 economies (tax reductions, subsidies, and reimbursement of training costs), 24 percent use all three incentives (38 percent of HIC, 19 percent of UMIC, and 9 percent of LMIC), 33 percent use two incentives (5 percent of LIC, 33 percent of HIC, 25 percent of UMIC, and 36 percent of LMIC) and 43 percent use one incentive (56 percent of UMIC, 55 percent of LMIC, 29 percent of HIC, and 5 percent of LIC).Subsidies and reimbursement of training costs are more likely to be offered together (with a Jaccard similarity index of 51 percent), followed by subsidies and tax reductions (41 percent).
As expected, both the availability of UB schemes and the average duration of UB increase with income levels, while the minimum duration of contribution tends to decrease (Figure 1a).Out of 30 LIC in the study, only Tanzania, Nepal, and Tajikistan 12 have UB schemes (10 percent of LIC), while in other income groups, the share of countries with UB scheme is much higher, (30, 47, and 81 percent in LMIC, UMIC, and HIC, respectively).Similarly, the average maximum UB duration is the lowest in LIC (13 weeks, although this is based only on three countries), followed by LMIC (25) and UMIC (29), and is the highest in HIC (47).
The minimum duration of contribution to be eligible for UB is the highest in LMIC (at an average of 18 months), followed by LIC and UMIC (12 months), and the lowest in HIC (9 months).Among the regions, OECD high income countries (OECD HIC) and Europe and Central Asia (ECA) have the highest shares of economies with UB schemes (with 100 and 88 percent, respectively) and Sub-Saharan Africa (SSA) and South Asia (SAR) have the lowest share (at 8 and 25 percent, respectively), though SAR only has two countries with UB schemes: India and Nepal (Figure 1b).OECD HIC also have the longest maximum UB duration (52 weeks) followed by East Asia and Pacific (EAP) and Middle East and North Africa (MENA) (39 and 30 weeks, respectively), while Latin America and Caribbean (LAC), SAR, SSA, and ECA have the shortest maximum UB durations (the former two have 24 weeks and the latter two have 27 weeks).
Regarding the minimum contribution period for UB, MENA and EAP require the longest period of minimum contribution, with 19 and 16 months, respectively.SSA has the lowest minimum contribution period to be eligible for UB (8 months, though the sample is very small, with only 4 out of 48 countries in the region having UB schemes), followed by OECD HIC (10 months).Predictably, in the majority of the income groups and regions, UB durations increase with the length of the contribution.However, in LIC, as well as in LAC, SSA, and SAR, the UB period is the same for both five and ten years of contributions, with the caveat that LIC have three and SAR have only two countries with UB schemes (Figure 2). Figure 3 outlines eligibility criteria for UB across income groups and regions.In LIC, proof of no other income source is the primary criterion across the three countries with UB.Registering with PES is the most prevalent criterion in LMIC and UMIC (86 and 85 percent, respectively), while in high-income countries (HIC), active job search reporting and participation in PES training program are most popular (Figure 3a).In South Asia (SAR), India and Nepal are the only two countries with UB, and they have five of six eligibility conditions, excluding mandatory acceptance of a suitable job.Leaving aside SAR, registering with PES is widely used in MENA (90 percent), followed by ECA (86 percent) and EAP (80 percent).Job loss proof without personal fault is seen in 60 percent of EAP and MENA countries, while proof of no other income source is less common, mostly found in EAP and HIC (40 and 39 percent respectively).OECD HIC leads in requiring periodical reporting on active job search (97 percent) and PES training participation (85 percent), while SSA stands out for mandating suitable job acceptance in 75 percent of its countries, followed by OECD HIC (70 percent) (Figure 3b).As anticipated, ALMP-measured by the availability and type of PES, national skills fund, and government incentives for training-are also closely linked to income levels of countries (Figure 4a), but it is not as strong as the correlation of UB with income levels.About 57 percent of LIC have PES, compared to 70 percent in LMIC, and 93 percent in UMIC and HIC.The probability of having a national skills fund and government incentives is the highest in HIC, at 66 and 75 percent, respectively, and lowest in LIC, at 27 and 20 percent, respectively.
As for the regions, all countries in ECA, OECD-HIC, and SAR have PES, compared to 84 percent of EAP, 82 percent of LAC, 75 percent of MENA, and 58 percent of SSA (Figure 4b).OECD-HIC also have the highest share of economies with national skills funds (76 percent), followed by LAC (70 percent), as well as the highest share of economies with government incentives for training (97 percent), followed by ECA (79 percent).Interestingly, ECA has the lowest share of economies with national skills funds (21 percent), while LAC has the lowest use of government incentives for training (33 percent), suggesting that these two policies may be used as substitutes in different regions.
Among the types of services offered by PES, across most income groups, training programs and access to a vacancy database are the most widely offered services, followed by job placement, counseling, and internship/apprenticeship, while mobility support is the least commonly used service (Figure 5a).Across regions, ECA and OECD-HIC stand out with the highest share of countries offering training programs, counseling, job placement, and vacancy databases, ranging from 94 percent to 100 percent (Figure 5b).
Internship/apprenticeship is most prevalent in ECA (92 percent of ECA), followed by SAR (88 percent), OECD HIC (76 percent), and SSA (75 percent), while it is least common in LAC and MENA (44 and 47 percent, respectively).A7 in the appendix for the sample size of each category.
In addition to PES and national skills/training fund, governments also offer fiscal incentives to promote labor market outcomes, using primarily tax deduction, subsidy, and cost reimbursement.Tax deductions are used most frequently across all income groups but HIC, where subsidies are used slightly more often than tax deductions (Figure 6a).The LIC use tax deduction the most compared to other income groups, but do not use cost reimbursement, which is used most commonly by HIC, followed by UMIC and LMIC.Across regions, tax deduction is used more frequently as an incentive for training in SAR and LAC, followed by ECA and SSA, while subsidies are more prevalent in OECD HIC, ECA, and MENA (Figure 6b).Use of reimbursing training costs is the lowest in all regions compared to other two incentives, except EAP and SSA where it is the second after tax deduction.A8 in the appendix for the sample size of each category.

UB and ALMP reforms implemented during the Covid-19 pandemic
The Across all income groups, UB reforms primarily targeted increasing UB amounts, followed by expanding coverage and extending payment durations (Figure 7a).As expected, reform numbers are highly correlated with countries' income levels.Across regions, OECD-HIC implemented the most UB reforms (58 out of 159), followed by ECA (30), LAC (29), EAP (24), MENA, and SSA (each with 9 reforms) (Figure 7b).
Nearly half of all ALMP reforms were focused on adjusting labor regulations (415 out of 865; Figure 8a). 14 Wage subsidies emerged as the second most frequently implemented reform (233), followed by job 13 See Figure A1 in the appendix for the full coverage of all non-UB and non-ALMP labor market and social protection reforms. 14These labor regulation adjustments encompassed: (i) increasing remuneration associated with wage subsidies, income, and fiscal support to safeguard worker incomes and preserve jobs, especially among vulnerable groups; (ii) modifying compensation and severance guidelines for dismissals and employer obligations upon termination; (iii) allowing contract or work permit extensions; (iv) introducing new termination justifications, altering notice periods for dismissal, and changing dismissal search assistance and training (140).Conversely, public works reforms were the least common among all ALMP reforms, but they were the second-most popular reform in low middle-and low-income economies (Figure 8a).In terms of regions, OECD-HIC exhibited the highest number of ALMP reforms, totaling 233 (Figure 8b).Unlike UB reforms, where ECA had the second highest and SSA had the lowest number of reforms, in ALMP, ECA carried out the fourth and SSA had the third highest number of reforms, with 132 and 148, respectively.LAC accounted for the second highest count of ALMP reforms (168), whereas MENA and SAR procedures during the pandemic; (v) adapting working conditions to promote telecommuting, flexible hours, locations and rest periods; (vi) modifying paid vacation, sick leave, or other leave entitlements to grant employers greater flexibility in mandating leave; (vii) changing labor inspection policies to enhance guidelines on sanitation, social distancing, personal protective equipment use, and reducing transmission risks; and (viii) reducing administrative burdens on employers through changes in bureaucratic processes, reporting requirements related to labor regulations, and other relevant regulatory modifications.implemented the fewest reforms, with 49 and 25, respectively.Despite having no UB reforms and the fewest ALMP reforms, SAR implemented the highest number of social assistance programs (Gentilini et al. 2022).These initiatives constituted 78 percent of SAR's and low-income countries' social protection and labor market policies.
Overall, most of UB reforms during the Covid-19 pandemic were implemented by high income countries (at 62 percent), while upper middle-income countries accounted for 26 percent and lower middle-income countries for 11 percent of these reforms.Low-income countries had no UB reforms.A larger proportion of low and lower middle-income economies implemented ALMP reforms during this period, with 8 percent and 17 percent, respectively, compared to 47 percent in high-income countries and 27 percent in upper middle-income countries.

Correlations among UB, ALMP, and labor market outcomes
This section presents simple correlations between UB and ALMP indicators and labor market outcomes based on the novel data collected in 2019 and 2020, without controlling for other indicators.As seen from Table 2, the unemployment rate is significantly associated with only two ALMP, namely training and internship programs, the former with a negative and the latter with a positive association.The employment rate, on the other hand, is associated significantly and negatively with four indicators of UB and ALMP: availability of UB scheme, length of minimum contribution period for UB, and two training incentives (subsidy and reimbursement of training costs).The employment rate in agriculture and the selfemployment rates have a significant and negative correlation with the majority of UB and ALMP indicators, while employment rates in industry and services have a positive association.The only exception is that the employment rate in services is negatively related to the length of the minimum contribution period for UB.
These findings imply that, ceteris paribus, countries with UB and ALMP tend to have lower agricultural employment and self-employment rates and higher employment rates in services and industry.Although having a UB scheme is positively associated with both log and growth of labor productivity, a longer UB period is negatively associated with the latter.Moreover, among the ALMP indicators, four have a significant and positive relationship with labor productivity growth: counseling services, support for geographical mobility, internship programs, and training incentives.The log of labor productivity is significantly associated with most indicators of UB and ALMP, and all of these associations are positive.Figure 9 below reports the scatter plots of the maximum length of UB against selected key labor market indicators, including employment rate, log of labor productivity, collective bargaining coverage and labor tax and contribution rates, all of which are expected to be closely linked to UB and ALMP.The first scatter plot of Figure 9 reports employment rate against the maximum length of UB.The figure reveals a negative association between these indicators.Large numbers of countries concentrate in the lower-right and lower-left quadrant of the figure.HIC dominate the lower-right quadrant with higher employment rates and shorter maximum UB periods, followed by UMIC and LMIC.Conversely, LMIC and UMIC prevail in the lower-left quadrant with shorter maximum UB periods and lower employment rates.
As seen in the second scatter plot of Figure 9, there is a significant positive association between labor productivity and the maximum length of UB.All countries in the far end of the upper-right quadrant are HIC, with both longer UB periods and higher productivity, however, most HIC and UMIC have higher labor productivity and shorter UB periods (lower-right quadrant).The majority of LMIC have both lower productivity and shorter UB periods (lower-left quadrant).There are only three countries (Kyrgyzstan, Ukraine, and China) in the upper-left quadrant with longer UB durations and lower productivity.
Predictably, the maximum length of UB period has a strong positive correlation with collective bargaining coverage, as demonstrated in the third scatter plot of Figure 9.Most industrialized European countries, except the United Kingdom and Ireland, have both higher collective bargaining coverage and longer maximum UB periods, while most of UMIC and high-income Eastern European countries, as well as the United States and Canada, have lower collective bargaining coverage and shorter maximum UB periods.Finally, the last scatter plot of Figure 9 confirms that labor tax and contribution has a strong positive association with the maximum length of UB at all income levels, with few exceptions.

Econometric Analysis of Labor Market Outcomes with Respect to UB and ALMP
This section provides an econometric analysis of labor market outcomes with respect to UB and ALMP indicators.As pointed out in the literature review section, in theory, both UB and ALMP should increase formal employment and productivity and decrease unemployment.UB serve as a means to partially compensate for the income lost by unemployed individuals, enabling them to concentrate on their job search to find positions that align well with their skill set.ALMP provide counseling, training, and job matching services, among others, to facilitate their employment search.However, due to their unintended effects, such as moral hazard, substitution and displacement effects, and dead weight loss, their overall impact will depend on which of these effects are dominant (Calmfors and Skedinger 1994;Calmfors 1995;Card et al. 2007;Van Doornik et al. 2018).For example, if UB or ALMP lead to moral hazard, substitution of productive jobs with inferior jobs, or crowd out job seekers who are not part of these programs, and if these negative effects dominate their positive effects, net impact of UB and ALMP on labor market outcomes will be negative.
Therefore, the hypothesis of this analysis is that UB and ALMP are positively associated with employment and productivity and negatively associated with unemployment, assuming that their unintended negative effects are weaker than their positive effects.In addition, given that the countries with higher per capita income have more resources and stronger institutional capacity to design more comprehensive UB and ALMP as well as to implement and monitor them more effectively, impact of UB and ALMP on labor market outcomes might be stronger in these countries.To account for this impact, we incorporated interaction terms between income levels of countries and UB and ALMP indicators in the regression model.In addition, although our analysis does not disentangle the different types of unintended effects of UB and ALMP, since it employs macro level data, these effects are factored in the values of the indicators, allowing us to obtain the overall association of UB and ALMP indicators with labor market outcomes. 15The crosscountry OLS estimations using data from 191 economies in 2019 are based on the following linear regression equation:16 where i denotes country i and Y is the dependent variable referring to labor market outcomes (i.e., aggregate unemployment rate; aggregate and sectoral employment rates, and self-employment rate; and growth rate of labor productivity).The variable X1 represents UB and ALMP indicators used in the analysis: the availability of UB schemes and UB after one year of contribution; maximum length of UB duration; minimum length of contribution period; the availability of PES and types of PES; and the availability of incentives for training and national skills funds. 17The X2 variable refers to control variables informed by the literature: real GDP per capita, labor force participation, income share of labor, investment rate, openness to international trade, world governance indicator, life expectancy, innovation, employeeemployer cooperation, severance pay, labor tax and social contribution rates, inflation rate, gross government debt as a share of GDP, and budget balance as a share of GDP. 18The X3 variable includes income groups; X1*X3 refers to the interaction term between income groups and UB and ALMP; and ei denotes robust white noise residuals.The analysis also controls for fixed effects across regions.
Outliers were removed using a standard method. 19To increase the number of observations, control variables were averaged over 2017-2019 (real GDP per capita, income share of labor, labor force participation, world governance indicators, life expectancy, severance pay, and labor tax and contribution rates) or over 2010-2019 (investment rate, openness to trade, gross government debt, budget balance, inflation rate, rankings of innovation index, and employee-employer cooperation index).
Different from the existing studies on the subject, our econometric analysis incorporates the interaction between income categories of countries and UB and ALMP indicators to obtain more accurate results.Separate OLS analyses of 8 labor market outcomes with respect to 16 UB and ALMP indicators and their interaction terms with income categories resulted in a large number of significant coefficients on UB and ALMP indicators and their interaction terms. 20To keep the focus of the analysis on the main variables of interest, Table 3 only reports statistically significant total association of UB and ALMP indicators with labor market outcomes.These total associations were computed by taking the first derivative of each interaction model shown in equation 1 with respect to UB and ALMP indicators for each income group. 21s seen in Table 3, econometric analysis reveals no significant association between any of the UB indicators and the overall unemployment rate across different income groups.This contrasts with the majority of studies, such as Hagedorn et al. (2015), Caliendo et al. (2013), andTatsiramos (2009), which typically identify a positive relationship between unemployment and UB.This discrepancy in findings could be attributed to three key differences between this study and previous research: Firstly, prior studies predominantly utilize micro or regional data within developed economies to evaluate the impact of UB and ALMP on policy beneficiaries, without assessing their broader influence on macro indicators.Secondly, earlier cross-country studies often overlook the influence of countries' per capita income levels on the relationship between UB, ALMP, and labor market outcomes.Lastly, previous research incorporates smaller sample of developing economies in their assessments.
Similarly, except for upper middle-income countries, neither ALMP nor UB are significantly associated with the aggregate employment rate.In upper middle-income countries, both UB and ALMP are negatively associated with aggregate employment rate.The earlier result is in line with the UB literature in developing countries (e.g., Filiz 2017) and the later result can be driven by the following factors: (i) the presence of the lock-in effect of training programs, which leads to longer durations of unemployment given that unpaid internship/apprenticeship is not counted as employment; (ii) poor design of internship/apprenticeship programs, particularly in developing countries; or (iii) an increase in participation in these programs during periods of high unemployment.
Across all income groups, ALMP consistently display a negative relationship with the self-employment rate.The relationship between the self-employment rate and vocational training has been investigated extensively in the literature, which shows either a modest or no significant relationship between them (Grierson 1997;Premand et al. 2012;Doerr and Novella 2020).These studies point out the low quality of vocational programs, which do not account for the wide range of complexities related to self-employment, decreasing returns to extensive training programs, and the reduced survival of microenterprises in periods of high unemployment.This result can also imply that on-the-job learning programs, such as internship/apprenticeship, may focus on wage employment, given the complexity of training programs for entrepreneurship, or that the supply of these programs increases during periods of low employment.In low-and lower-middle-income countries, ALMP show a stronger correlation with labor market outcomes compared to upper middle and high-income countries, highlighting the fact that they extensively rely on ALMP.Specifically, they exhibit positive associations with productivity and employment rates in the agriculture sector, and a negative association with employment rate in the services and industry.This could be driven by recent trends in the training programs in agriculture in lowincome countries.In fact, an investigation of these programs shows that they promoted agricultural employment in many countries, including Egypt, Ghana, India, Nepal, Nigeria, Somalia, and Uganda (Maïga et al. 2020).
In upper middle-income countries, both ALMP and UB are positively associated with productivity growth and negatively associated with the aggregate employment rate, although UB are also positively associated with the industrial employment rate.The latter findings suggest that, all else being equal, generous UB in these countries may promote industrial employment at the cost of employment in other sectors, potentially due to displacement and substitution effects.
High-income countries exhibit fewer statistically significant findings.Among these, UB are positively related to agricultural employment rate, but negatively associated with industrial employment rate and productivity growth.Tapp (2011) demonstrates this effect in Canada, where an increase in UB led to a decrease in employment in the manufacturing sector and an increase in the resources sector, because the latter was able to create new jobs while the former could not.
As mentioned previously, theoretically, both UB and ALMP are expected to increase productivity growth by promoting quality of job match, skills, and geographical mobility, among others, provided that their indirect negative effects (such as moral hazard, displacement, substitution, and deadweight loss), do not dominate their intended positive effects.In terms of empirical evidence, there are only a few studies investigating the relationship between UB and productivity, and they focus on developed economies.Among these, Soltes (2017) finds a positive association between the UB generosity index and TFP growth in 16 developed economies.Acemoglu and Shimer (2000) show that in the United States, an increase in UB receipts and expenditures by 17.6 percent leads to a 0.7 percent increase in output and 0.6 percent increase in welfare.The remaining studies examine the association of UB or ALMP with post-employment earnings or match quality in developed economies and do not provide conclusive evidence, though there is more evidence for a positive association.
Taken together, the findings that UB and/or ALMP are positively associated with productivity growth in developing countries are in line with the theoretical prediction and the findings of limited number of empirical studies, such as Acemoglu and Shimer (2000) and Soltes (2017).The negative link between UB and productivity growth in high-income countries suggests diminishing returns to UB.Also, the finding that longer UB duration is negatively related to industrial employment but positively related to agricultural employment may offer an explanation for the negative association between UB and productivity growth in high-income countries, as agriculture typically experiences slower productivity growth compared to industry.

Conclusion
The objective of this paper is to analyze the patterns of UB and ALMP indicators across different income groups and regions and investigate the interlinkages between labor market outcomes and UB and ALMP, drawing on novel UB and ALMP data collected from 191 economies in 2019-2020.The paper also aims to assess the trends of global labor market reforms implemented during the Covid-19 pandemic utilizing the World Bank's comprehensive inventory of these reforms.
This investigation reveals new insights on the patterns of UB and ALMP and their linkages with labor market outcomes in low-income, upper-middle-income and high-income economies, especially given the gap in the literature in developing countries.Also, different from past studies, the econometric analysis in this paper incorporates a wide range of UB and ALMP indicators, as well as many categories of labor market outcomes, and controls for the interlinkages between per capita income and UB and ALMP indicators.Finally, unlike most studies that use micro level data, this analysis employs aggregate-level indicators to assess the overall association between UB and ALMP and labor market outcomes. 22he key findings of the descriptive statistical analyses can be summarized as follows: • Among the 191 countries, 47.6 percent have UB schemes and 85.3 percent have ALMP.
• Across ALMP, public employment services (PES) are most prevalent (employed in 81.7 percent of countries), followed by national skills funds (52.9 percent) and government incentives offered for training (53.9 percent).• The use of ALMP vis-à-vis UB increases substantially, as income level decreases, • On average, employees need to contribute at least 11.6 months to become eligible for UB, and the average maximum UB period is 37.3 weeks.• Among the six eligibility conditions for UB covered by the data, all of which are also ALMP, periodic reporting on active job search, participation in a training program, and registration with PES are most commonly used.Regarding the six services offered by PES, training programs, access to the vacancy database and job placement are most frequently used, followed by counseling and internship/apprenticeship programs.Support for geographic mobility is the least frequently used.• The eligibility conditions most likely to be used together are periodic reporting on active job search and participating in training offered by PES, followed by periodic reporting and registration with PES.As for the services offered by PES, job placement and access to the vacancy database are most likely to be offered together, followed by job placement and counseling services.• According to the World Bank's Social Protection and Jobs Policy Inventory that gathered data from March 2020 to January 2022, 223 economies implemented 3856 policies to safeguard workers, vulnerable groups, and economic activities from the Covid-19 crisis.Among these reforms, 159 were related to UB, 865 were related to ALMP and the rest was on social assistance and other labor market policies.• High-income economies led 62 percent of UB reforms, with upper middle-income and lower middle-income countries contributing 26 and 11 percent, respectively.
• A larger proportion of low and lower middle-income economies implemented ALMP reforms during this period, with 8 percent and 17 percent, respectively, compared to 47 percent in highincome countries and 27 percent in upper middle-income countries.• Across all income groups, Covid-19 related UB reforms primarily targeted increasing UB amounts, followed by expanding coverage and extending payment durations.• Labor regulation adjustments were the most prevalent ALMP reform in all income groups.In upper-middle and high-income economies, wage subsidy programs and job search/training initiatives were the second and third most common reforms, respectively.Conversely, public works, while the least common overall, emerged as the second most popular reform in low and lower middle-income economies.
The econometric analysis reveals distinct patterns across income groups: • In low-and lower-middle-income countries, ALMP show a stronger correlation with labor market outcomes compared to upper middle and high-income countries.Specifically, they exhibit positive associations with productivity growth and employment rate in the agriculture sector, and a negative association with the employment rate in the services sector.• In upper middle-income countries, both ALMP and UB are positively linked to productivity growth and negatively linked to the aggregate employment rate.UB are also positively associated with the industrial employment rate but does not show any direct association with the employment rates in agriculture or services sectors.• High-income countries exhibit fewer statistically significant findings.Among these, UB are positively related to agricultural employment rates, but negatively associated with industrial employment rate and productivity growth.• Across all income groups, ALMP consistently display a negative relationship with the selfemployment rate.
While this analysis does not determine the causality between UB and ALMP indicators and labor market outcomes, its key findings offer valuable insights for policymakers and researchers.The significance of these findings stems from the extensive data coverage across both developing and developed countries and carefully designed regression model, which incorporates numerous control variables informed by the literature and the interactions of income categories with UB and ALMP indicators.Additionally, the regression model accounts for fixed effects across income groups and regions, and eliminates outliers.
As emphasized in the literature, although both UB and ALMP are shown to have many benefits, they are costly programs with many unintended distortionary effects in the economy.Therefore, it is crucial to employ a meticulous design informed by robust empirical evidence, along with pilot programs, to assess their impact before considering widespread implementation.
Some limitations of the analyses should also be noted.Although the data used cover the largest number of countries among existing databases, they are only from 2019, limiting the maximum sample size to 191.In addition, most indicators of UB and ALMP are binary with limited information, and the regression analyses do not fully address the endogeneity issues.

Box 1. UB and ALMP reforms captured by the new data during May 2, 2019 -May 1, 2020
Twelve economies implemented reforms in the UB and ALMP indicators captured by new data until May 1, 2020.
Ten of these economies undertook reforms in response to Covid-19 pandemic.Among these, Finland and Chile decreased the minimum contribution period for unemployment benefit for all job tenures to three months and Cabo Verde decreased it to two months.Finland also removed the eligibility condition requiring unemployed persons to prove job loss through no own fault and Sweden removed the condition for periodical reporting on active job search.
Thailand and the United States increased the duration of unemployment benefit for all lengths of contribution to 29 and 39 weeks, respectively.The Gambia opened a job center in 2020, Cambodia set-up a skills development training center, Indonesia introduced an up-skilling program and France made employment services online and by phone.
Regarding the reforms prior to the pandemic, France increased the minimum contribution period for unemployment benefit from 4 to 6 months for all job tenures. 23Kazakhstan increased the unemployment benefit duration to 22 and 26 weeks for those individuals with 5 and 10 years of contribution, respectively.In contrast, the Arab Republic of Egypt increased the minimum duration to be eligible for unemployment benefit to 12 months for all employment tenures and decreased the unemployment benefit period by 4 weeks to 12 weeks for employees with one year of contribution.
A larger range of UB and ALMP reforms implemented in response to the Covid-19 pandemic, which goes beyond the scope of the new data used in this study, is recorded in the World Bank Covid-19 Social Protection and Jobs Policy (SPJ) Inventory.Types of UB and ALMP reforms captured by this data and their patterns across income groups and regions are presented in section 3.3 of this paper.Unemployment rate has 176 observations, like all other unemployment and employment rates used in the analysis.See Table A8 for the sample size of each UB and ALMP indicator, which is the same as one of the four indicators used above.

Figure 1 .
Figure 1.Availability of UB, minimum contribution period for UB, and UB duration, 2019 a.By income group b.By region

Figure 2 .
Figure 2. Length of UB for 1, 5, and 10 years of contributions (weeks), 2019 a.By income group b.By region

Figure 3 .
Figure 3. Types of eligibility condition for UB (percent of countries), 2019 a.By income group b.By region

Figure 4 .
Figure 4. ALMP: PES, national training fund and incentives for training (percent of countries), 2019 a.By income group b.By region

Figure 5 .
Figure 5. Types of ALMP offered by PES (percent of countries), 2019 a.By income group b.By region

Figure 6 .
Figure 6.Types of incentives for training (percent of countries), 2019 a.By income group b.By region ALMP and UB data provided above are collected until May 1, 2020, and do not capture all Covid-19 related reforms.Therefore, this section utilizes comprehensive data from the World Bank's Social Protection and Jobs Policy Inventory (SPJPI) collected in 223 economies between March 2020 and January 2022 to showcase the trends in UB and ALMP reforms.During this timeframe, SPJPI documented 3,856 reforms in labor market and social protection policies.Among these, 159 were UB and 865 were ALMP reforms, comprising over a quarter of all reforms.Within the first 10 months of the Covid-19 pandemic, labor market policies represented 55 percent of emergency response measures, underscoring their significance in safeguarding workers, vulnerable populations, and fostering economic activities in the face of economic crisis(World Bank 2021, Gentilini et al. 2022).Of the remaining 2,833 reforms implemented during March 2020 -January 2022, cash transfers accounted for the highest at 952, followed by utility and financial obligation support (764), and food/voucher support (393).13

Figure 7 .
Figure 7. Number of UB policies during the Covid-19 pandemic a.By policy and income b.By region

Figure 8 .
Figure 8. Number of ALMP programs during the Covid-19 pandemic a.By program and income group b.By region

Figure 9 .
Figure 9. Scatter plots of UB duration against selected labor market indicators

Figure A1 .
Figure A1.Other labor market policies and social assistance during the Covid-19 pandemic

Table 1
provides the descriptive statistics of UB and ALMP indicators in 2019.8Asseen from the table, 47.6 percent of 191 economies have UB schemes and 38.7 percent offer UB after one year of contribution.As expected, eligibility conditions for UB schemes vary across countries.Average minimum period of contribution to become eligible for UB is 11.6 months and about 11 percent of countries with UB schemes do not require a contribution.Periodic reporting on active job search is the most commonly required eligibility condition (80.2 percent of economies), followed by participating in training sessions assigned by PES (74.7%) and registering with PES (73.6 percent).While not as widespread as the above three eligibility conditions, many countries also require accepting a suitable job offered (50.5 percent), proof job loss through no fault of the employee (48.4 percent), and proof of no other income source (37.4 percent).
Source: World Bank Group.Data set is available at https://www.worldbank.org/en/research/employing-workers/dataNote: The following economies do not require a contribution to be eligible for UB: AUS, BLR, HKG, KAZ, NPL, NOR, NZL, RUS, UZB, and ZAF, and the following economies have the highest minimum duration of contribution (at 36 months): AZE, DZA, JOR, MMR,

Table 2 . Simple correlations between UB and ALMP indicators and labor market outcomes, 2019 Ɨ
Source: See TableA9in the appendix for the definition and source of the indicators.* Indicates p values below 10 percent.

Table 3 . Regression analyses with interaction terms: Total association of UB and ALMP indicators with labor market outcomes at different income levels, 2019
See TableA9in the appendix for the definition and source of the indicators used in the regression analyses.Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1.Ϯ Total association of UB and ALMP indicators with labor market outcomes are computed using the first derivative of the regression equation with respect to the UB and ALMP indicators for each country group.Notes: Internship refers to both internship and apprenticeship; incent.: incentives; max: maximum; reimburs.: reimbursement; train.: training; unemp.: unemployment; w/: with.Table repots only the statistically significant coefficients of UB and ALMP indicators obtained through the first derivation of each interaction regression equation, described in equation 1, with respect to UB and ALMP indicators for each income group using lincom command of STATA.Total of 128 full regression analyses are carried out including all control variables (i.e., regression of 8 labor market outcome indicators with respect to 16 UB and ALMP indicators).To ensure robustness, control variables were incrementally introduced into the model.Individual regression tables are not reported in the appendix due to their sheer numbers, but they are available upon request.