## wpiea2025226-source-pdf

## Source details

**Canonical URL:** [wpiea2025226-source-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025226-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025226-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025226-source-pdf.pdf.json)

---

### Overview and motivation
- Recent censuses (North Macedonia 2021, Albania 2023, Kosovo 2024) produced substantial downward revisions to population figures, reflecting high emigration rates to EU countries.
- Core questions:
  - How do economic factors drive emigration in the Western Balkans?
  - How will further EU integration impact migration, population dynamics, and output?
  - How can policymakers support domestic labor markets and boost output?
- Approach: a quantitative multi-country labor market model of Europe that links inactivity, unemployment, employment, and bilateral migration endogenously.

### Key labor market and migration facts (Introduction & Section 2 preview)
- Labor participation and unemployment:
  - As of 2023, labor participation of the working-age (15-64) population: around 40 percent in Kosovo and between 60 to 75 percent in the other Western Balkan countries.
  - EU average unemployment: around 6 percent.
  - 2023 Western Balkan unemployment rates: around 10 percent in Albania, Kosovo, and Serbia; over 13 percent in the other countries.
  - Kosovo’s unemployment rate fell by as much as two thirds over the past ten years.
- Population revisions and demographics:
  - Census revisions: Albania (2023), North Macedonia (2021), Kosovo (2024) revised populations by over 10 percent; Montenegro (2023) and Serbia (2022) by +1 and -3 percent respectively; Bosnia and Herzegovina last census 2013.
  - From 2015 to 2023, the working-age population declined by more than 10 percent on average in Western Balkan countries versus 2 percent in the EU.
  - Western Balkan countries aged more than EU countries over the past decade: 3.8 years versus 2.5 years.
- Remittances and informality:
  - Remittances average 10 percent of GDP in Western Balkan countries; Kosovo has the highest remittances-to-GDP ratio.
  - Informality estimates range from 20 to over 50 percent of employment.
- Education and skills:
  - 21.1 percent of the working-age population have tertiary education—ranging from 25.9 percent in Montenegro to 14.7 percent in Bosnia and Herzegovina—compared with 30.9 percent in the EU.
  - Over-qualification rates of Western Balkan migrants: between 35 and 60 percent.
- Gender, youth, and NEET:
  - Labor participation gap (Western Balkans versus EU): 18.1 percentage points; employment gap: 17.1 percentage points; unemployment gap: 2.4 percentage points. EU comparable gaps: around 9.6, 9.4, and 0.5 percentage points respectively.
  - Youth unemployment rate (15-24) average: 25.0 percent in 2023 versus 14.5 percent in the EU.
  - NEET rate average: 23.7 percent (range: 15.9 percent in Serbia to 37.4 percent in Kosovo) versus 11.1 percent in the EU.
- Migration motives and composition:
  - Around half of Western Balkan migrants to the EU report moving for employment in some sources; bilateral flows higher when labor market gaps (wage gap, unemployment gap) are large.
  - Migrants tend to be similar or less likely to have tertiary education than the sending-country population.

### Model structure, calibration, and estimation
- Model features:
  - Multi-country quantitative labor market model with search-and-matching (Mortensen and Pissarides, 1994) within local labor markets.
  - Households choose between inactivity and searching; unemployed enter matches via job-finding rate p_c(θ_c); firms post vacancies at cost c and fill at q_c(θ_c); matches destroyed at rate s_c.
  - Households decide each period whether to relocate internationally, paying bilateral migration cost τ_cc' and idiosyncratic cost ε_cc'.
  - Wages determined by Nash bargaining; home production b_c captures income while inactive/unemployed.
- Calibration and sample:
  - Calibrated to labor market outcomes and migration patterns across 35 European countries (27 EU members, six Western Balkan countries, Türkiye, and Switzerland); Ukraine, Moldova, and Georgia dropped due to data gaps.
  - Estimation yields country-specific parameter sets and country-pair migration costs.
- Key parameter interpretations used in calibration:
  - y_c: production of matched firm-worker pairs (related to net wage and output per capita).
  - M_c: matching efficiency (job-finding rates).
  - c: vacancy posting cost (relates to tightness).
  - b_c: home production (flows between inactivity and activity).
  - τ̅_c: transition cost between inactivity and activity.

### Empirical regression results on migration determinants
- Estimated migration equation (log migrant share on wage gap, labor market gaps, controls, fixed effects): key coefficients (selected):
  - log wage gap coefficients (across specifications): 0.561***, 0.246*, 1.259***, 1.396***, 1.131***, 0.375**.
  - unemp rate gap coefficient (selected): -0.0253***, -0.0140*, -0.0250***.
  - interaction log wage gap x tightness gap: 0.476***.
  - EU linked pair coefficient: 0.723***.
  - log distance coefficient (across columns): -1.074***, -1.084***, -1.068***, -1.068***, -1.099***, -1.072***.
  - R2 reported: 0.783, 0.784, 0.813, 0.812, 0.825, 0.788.
  - Observations (Obs) across columns: 6743, 6647, 3551, 3551, 3551, 6647.
- Key empirical magnitudes:
  - A one percent increase in the wage gap between a country pair correlates with an increase in migration by around 0.56 percent (in a representative specification).
  - EU-linked country pairs have almost double bilateral migration flows, all else equal.

### EU integration experiments — design
- Two scenarios for the six Western Balkan countries:
  - Scenario 1 (lower migration costs only): decline in migration costs calibrated to match a doubling of the emigration rate in each Western Balkan economy, corresponding to an average decline in migration costs by 6.1 percent.
  - Scenario 2 (lower migration costs + higher productivity): same decline in migration costs plus an overall increase in output-per-capita of 30 percent, implemented by increasing the parameter for production of matched firm-worker pairs by an average of 14.5 percent.

### EU integration experiments — simulated outcomes and tradeoffs (average impacts)
- Scenario 1 (lower migration costs only):
  - Emigration: sharp jump (doubling targeted).
  - Wage: increase of around 0.7 percent on average.
  - Unemployment rate: increase of around 1.8 percentage points on average.
  - Employment rate: decline of around 1.4 percentage points on average.
  - Labor participation: relatively flat.
  - Output-per-capita: decline of around 2.8 percentage points on average.
  - Immigration: increases but not enough to offset emigration → total population declines.
- Scenario 2 (lower migration costs + higher productivity):
  - Emigration: increases but less than in Scenario 1.
  - Wage: increase by more than 10 percent.
  - Unemployment rate: declines by 5.8 percentage points on average.
  - Employment rate: increases by almost 8 percentage points on average.
  - Participation rate: increases by about 4 percentage points on average.
  - Immigration: increases by 0.4 percentage points on average.
  - Output-per-capita: increase of 30 percent (targeted).
  - Population loss: partly offset by lower emigration and higher immigration.
- Key interpretive findings:
  - Increasing productivity is necessary to maximize benefits of EU integration and can more than offset negative impacts from higher emigration due to lower migration barriers.
  - Without productivity gains, EU integration can reduce output-per-capita and employment via higher emigration and reduced vacancies.
  - Even with strong productivity gains (Scenario 2), emigration is still expected to increase (emigration rate continues to increase by over 0.2 percentage points despite wages rising by more than 10 percent).

### Policy simulations — design and parameter changes
- Four policy categories simulated uniformly across the six Western Balkan economies:
  - Structural reforms (productivity boost): increase production of matched firms and workers by 1 percent.
    - Interpretation: would close around 2 percent of the model-implied wage gap with the EU.
  - Active labor market policies (ALMP): increase matching rate of unemployed workers and vacancies by 10 percent.
    - Interpretation: would close around one-third of the model-implied matching gap with the EU.
  - Business-promoting policies: decrease firms’ costs of posting a vacancy by 10 percent.
  - Policies to boost labor participation: decrease home production output (raises the value of employment relative to inactivity).
- Practical measures (examples from source):
  - Structural: judicial reforms, improving business/investment environment, reducing bureaucratic hurdles, improving property rights, investing in infrastructure, improving access to credit, tackling informality.
  - ALMP: dual vocational training, personalized job search, youth apprenticeships, labor market information systems, on-the-job and lifelong learning.
  - Business: one-stop shops for business registration, digital public support, simplifying tax systems, improving public infrastructure.
  - Participation: affordable childcare and eldercare, on-the-job training/reskilling, adjusting pension ages with life expectancy and gender equalization.

### Policy simulations — quantitative impacts and comparative results
- General outcomes (average impacts after ten years; cross-country ranges exist):
  - All simulated policies increase output per capita.
  - Most policies increase labor participation and employment, reduce unemployment, and increase wages — except for labor participation policies (modeled by decreasing home production output).
- Cautionary finding on participation-targeted policies:
  - Policies decreasing home production (to raise participation) have a negative impact on wages and increase net emigration because lower home production reduces the nonmarket outside option, encouraging migration.
- Policy implications:
  - No single policy dimension dominates; a combination across structural reforms, ALMP, business support, and participation measures can boost output and support labor markets.
  - Emphasis on creating domestic pull factors (higher productivity, better institutions, public services) rather than attempting to restrict emigration or directly subsidize staying migrants.

### Conclusions and policy recommendations
- Progress and risk:
  - Western Balkan countries have closed many labor market gaps with the EU while experiencing shrinking populations due to emigration; high emigration could be a headwind for future growth and may be exacerbated by EU integration.
- Central policy lessons:
  - EU integration should be coupled with productivity-enhancing structural reforms (trade integration, supply chains, financial deepening) to maximize benefits and counter potential negative migration effects.
  - Policymakers should prioritize low-cost, high-return policies that improve the attractiveness of working domestically (structural reforms, ALMPs, business environment improvements, public services).
  - Policies aimed at disincentivizing emigration or incentivizing immigration are not modeled directly and are unlikely to be effective given the large model-implied wage changes needed to offset migration.
- Areas outside the model and future work:
  - Role of remittances and diaspora FDI: remittances remain key but may mainly increase consumption and real estate investment rather than productive capacity.
  - Informality: reducing informality (one-stop shops, digital services, removing tax incentives for informal activity) is important to encourage formal investment and human capital accumulation.
- Key quantitative summary point reiterated:
  - Increasing labor productivity by 30 percent (average of previous EU accession cases) would increase wages by around 13 percent, close most of the unemployment gap, and lead to almost no increase in net migration.

### Model mechanics and calibration details (Annex IV highlights)
- Timing and stages per period: I. Search phase; II. Production and match destruction; III. Relocation phase.
- Matching and tightness:
  - Matching function M(U_t^c, V_t^c) with elasticity α; tightness θ_t^c = V_t^c / U_t^c; job-finding p(θ_t^c) = M(·)/U_t^c; vacancy fill q(θ_t^c) = M(·)/V_t^c.
- Worker and firm value functions, Nash bargaining wage determination, and free-entry condition for firms (V_t^c = 0 in equilibrium).
- Migration choice with idiosyncratic tastes drawn from a Type-I Extreme Value distribution; migration flows have closed-form logit expressions.
- Calibration targets and identification:
  - External parameters: {α η ν β}.
  - Internally calibrated parameters: {y_t^c, s_t^c, M_t^c, b_t^c, τ̅_t^c, τ̄_t^c_outside} (productivity, separation, matching efficiency, home production, transition and migration costs).
  - Moment mapping examples:
    - Wage equation linking w_t^c to η, y_t^c, b_t^c, Φ_t+1^c.
    - Tightness equation solving θ_t^c from M_t^c, κ_t^c, y_t^c, w_t^c, β, d_t^c.
    - Job-finding probability p̄_t^c = M_t^c (θ_t^c)1−α.
    - Inactivity flows map to b_t^c and τ̅_t^c via logit-like expressions.
- Goodness-of-fit:
  - Model replicates calibration moments for net wage rate, labor market tightness, job finding probability, flows between inactivity and activity, and emigration rates.
  - Out-of-sample checks: unemployment rate and output per worker closely replicated and track cross-country trends.

*Source: IMF Working Paper — "Labor Markets, Migration, and EU Integration in the Western Balkans" (Working Paper No. WP/2025/226).*

### 1. Introduction ........................................................................................................

### 1. Introduction

### Overview and motivation
- Recent censuses in Western Balkan countries (North Macedonia 2021, Albania 2023, Kosovo 2024) led to substantial downward revisions to population figures, reflecting high emigration rates, particularly to EU countries.
- Net emigration drains local labor resources and dampens growth, posing challenges as all six Western Balkan countries aspire to join the EU.
- Core questions addressed:
  - How do economic factors drive emigration in the Western Balkans?
  - How will further EU integration impact migration, population dynamics, and output?
  - How can policymakers support domestic labor markets and boost output in highly interconnected economies?
- Approach: combine theory and data using a quantitative multi-country labor market model of Europe that endogenously links inactivity, unemployment, employment, and bilateral migration.

### Labor market and migration context (key facts from the Introduction and Section 2 preview)
- Labor participation:
  - As of 2023, labor participation of the working-age (15-64) population amounted to around 40 percent in Kosovo and between 60 to 75 percent in the other Western Balkan countries.
  - Albania has closed the labor participation gap; Montenegro and Serbia have significantly reduced it; Bosnia and Herzegovina still has a substantial gap; progress in Kosovo and North Macedonia has stalled.
- Unemployment:
  - EU average: around 6 percent.
  - 2023 Western Balkan unemployment rates: around 10 percent in Albania, Kosovo, and Serbia; over 13 percent in the other countries.
  - Kosovo’s unemployment rate fell by as much as two thirds over the past ten years.
- Population and census revisions:
  - Albania (2023), North Macedonia (2021), and Kosovo (2024) censuses led to population revisions of over 10 percent.
  - Montenegro (2023) and Serbia (2022) led to smaller revisions of +1 and -3 percent, respectively.
  - Bosnia and Herzegovina’s last census was conducted in 2013.
- Structural factors:
  - Remittances average 10 percent of GDP in Western Balkan countries.
  - Kosovo has the highest remittances-to-GDP ratio, the lowest overall and women labor participation rates, and the highest NEET rate.
- Gender gaps (Western Balkans versus EU):
  - Western Balkans: labor participation gap 18.1 percentage points; employment gap 17.1 percentage points; unemployment gap 2.4 percentage points.
  - EU: around 9.6, 9.4, and 0.5 percentage points respectively.
- Youth and NEET:
  - Youth unemployment rate (15-24) average: 25.0 percent in 2023 versus 14.5 percent in the EU.
  - NEET rate averages 23.7 percent (range: 15.9 percent in Serbia to 37.4 percent in Kosovo) versus 11.1 percent in the EU.
- Empirical evidence on migration motives:
  - Around half of Western Balkan migrants to the EU report moving for employment.
  - Bilateral migration flows tend to be higher when labor market gaps (e.g., wage gap, unemployment gap) between countries are large.

### Model, calibration, and experiments (framework summary)
- Model features:
  - Multi-country quantitative labor market model with search-and-matching (Mortensen and Pissarides, 1994) within local labor markets.
  - Households decide between inactivity and searching for employment; households are homogeneous (reflecting similar skill distributions of emigrants and domestic workers in the Western Balkans).
  - Households decide each period whether to relocate internationally, weighing preferences, economic conditions, and migration costs.
  - Bilateral migration flows depend on endogenous local labor market conditions (labor market tightness, wages) and exogenous fundamentals (job separation rates, ease of finding a job, productivity, migration costs).
- Calibration and coverage:
  - Model is calibrated to match labor market outcomes and migration patterns across 35 European countries, comprising the EU, Western Balkan countries, Türkiye, and Switzerland.
- Experiments:
  - Experiment 1 (EU integration): two scenarios
    - Scenario A: lower migration costs only (simulate reduced migration barriers).
    - Scenario B: lower migration costs and increase productivity in Western Balkan countries.
  - Experiment 2 (policy simulations): simulate policies to boost output and support labor markets via four sets of policies:
    - (i) structural reforms to improve productivity;
    - (ii) active labor market policies to improve matching efficiency;
    - (iii) policies to support new job creation by existing firms or entrants;
    - (iv) policies to increase the relative value of employment compared to inactivity.

### Main quantitative findings and implications (from simulations and interpretation)
- Importance of productivity gains:
  - Increasing productivity is necessary to maximize the benefits of EU integration and can more than offset negative impacts from higher emigration due to lower migration barriers.
  - Without productivity gains, EU integration leads to further population declines as emigration increases, higher unemployment, and lower output per capita.
- Quantitative illustrative result:
  - A 30 percent increase in labor productivity—consistent with past EU accession cases—would result in a 13 percent increase in wages, close most of the unemployment gap with the EU, and offset emigration with increased immigration and labor participation.
  - Even with these large productivity and wage gains, emigration remains higher than in the baseline economy.
- Policy implications:
  - Attempting to directly counter emigration through financial incentives would require matching the large wage increases implied by productivity gains and would be exceedingly expensive and ineffective.
  - EU integration should be coupled with structural reforms that support productivity gains (lower trade barriers, supply chain integration, financial deepening), and with a robust structural reform agenda required by the EU accession process.
  - Policymakers have multiple options to support economic efficiency and create positive pull factors domestically via the four policy categories simulated; these broadly increase output per capita and improve labor market indicators, albeit with quantitative differences across policy types.

### Roadmap of the paper
- Section 2: stylized overview of labor markets, migration in the Western Balkans, and previous EU accession cases.
- Section 3: model description and calibration.
- Section 4: quantitative results on EU accession and policy simulations.
- Section 5: conclusions.

*Source: IMF Working Paper — "Labor Markets, Migration and EU Integration in the Western Balkans", Chapter 1: Introduction.*

### 21.1 percent of the working-age population having tertiary education—ranging from 25.9 percent in Montenegro

### wpiea2025226-source-pdf - 21.1 percent of the working-age population having tertiary education—ranging from 25.9 percent in Montenegro

### Education, skills, and informality
- 21.1 percent of the working-age population having tertiary education—ranging from 25.9 percent in Montenegro to 14.7 percent in Bosnia and Herzegovina—compared with 30.9 percent in the EU.
- Regional business surveys (World Bank's STEP Skills Measurement Program and BEEP) and weak PISA results highlight efficacy issues in educational systems in closing skill gaps and mismatches.
- Over-qualification rates of Western Balkan migrants are relatively high, ranging between 35 and 60 percent.
- Informality is prevalent and likely weighs on labor market outcomes, with estimates ranging from 20 to over 50 percent of employment.
  - Informality deters investment in the formal sector and may impede transitions of informal workers to formal employment.

### Labor force participation, employment, and wages
- Labor participation and falling unemployment have coincided with employment growth:
  - Employment grew by around 20 percent over the past 10 years, compared with 10 percent in the EU (Labor Force Survey data).
- Registered employment remained relatively resilient during the pandemic except for Montenegro, which experienced a large lockdown-driven drop due to reduced tourism.
- Post-pandemic registered employment growth has been strong, particularly in:
  - Kosovo (economic growth and policy support for formalization).
  - Montenegro (strong growth supported by the ‘Europe Now’ program and an influx of affluent immigrants from Russia and Ukraine).
- Real wage growth and convergence:
  - Real wage growth has been higher than in the EU in most Western Balkan economies since before the pandemic.
  - Albania saw a large increase in 2023 amid tight labor market conditions and public sector and minimum wage increases while inflation was comparatively low.
  - Montenegro experienced a steep increase in net wages in 2022 following abolition of mandatory health insurance contributions under the ‘Europe Now’ program.
  - During the period of high inflation following the pandemic, real gross and net wage growth turned negative in several Western Balkan economies and the EU.
- Earnings gaps:
  - Real PPP-adjusted net wages are only around 30 to 40 percent of the level in Germany.
- Productivity and wages:
  - Between 2015 and 2023, average real wage growth outstripped productivity growth in four of the Western Balkan economies, except for Kosovo and Montenegro.

### Migration, demographics, and population change
- From 2015 to 2023, the working-age population declined by more than 10 percent on average in Western Balkan countries compared to 2 percent in the EU.
- Emigration trends:
  - While emigration slowed following the pandemic, the trend resumed and mostly accelerated, reflected by first-time residence permit issuances in the EU and Switzerland.
  - Based on Eurostat data, 70-80 percent of emigrants from Western Balkan countries are of working age.
  - Western Balkan countries aged more than EU countries over the past decade: 3.8 years versus 2.5 years.
- Ukraine refugees:
  - The World Bank (2024) notes around 92,000 Ukrainian refugees in the Western Balkans as of 2023, with around two-thirds in Montenegro.
- Net migration rates vary substantially across the region:
  - Net migration broadly neutral in Serbia to peaking at an average of around 1.5 percent per annum in Kosovo.
- Fertility and median age indicators (figures in text):
  - Fertility (births per woman) and median population age developments are presented for ALB, BIH, KOS, MNE, MKD, SRB, and the EU.

### Motives for migration and demographic composition of migrants
- Survey and permit data on reasons for migrating:
  - Employment listed as primary reason by around 60 percent of migrants in Bosnia and Herzegovina but only around 20 percent in Albania.
  - In all six countries, employment has become a more important factor over the past decade.
  - Gender differences:
    - Men are three or more times likely than women to report moving for employment (e.g., around 30 percent for men vs. around 10 percent for women in Albania; almost 80 percent for men vs. around 30 percent for women in Bosnia and Herzegovina).
    - Women are more likely to list family reasons as the main factor in all six countries.
  - OECD (2022) survey: 79.5 percent of men and 78.3 percent of women respondents from the Western Balkans with current or past migration experience cited higher salaries as either very important or important in their migration decision.
- Skill composition:
  - Migrants tend to be similar or less likely to have tertiary education than the sending-country population as a whole, suggesting emigration is not strongly skewed toward the highly educated in the Western Balkans.
  - Potential factors include low migration costs, high local demand for higher education workers, foreign demand for lower-education occupations (construction, hospitality, retail trade, support services), and relatively low quality of Western Balkan education compared with the EU.

### Economic determinants of migration — empirical regression results
- Estimated migration equation (log migrant share regressed on wage gap, labor market gaps, controls, and fixed effects).
- Key empirical findings:
  - A one percent increase in the wage gap between a country pair correlates with an increase in migration by around 0.56 percent.
  - Unemployment rate and labor market tightness gaps indicate households move when likelihood of finding employment in destination countries is higher, with tightness correlated more when wage gaps are high.
  - EU-linked country pairs tend to have almost double the bilateral migration flows, all else equal.
- Selected coefficients and regression outputs from Table 1 (columns (1)–(6)):
  - log wage gap coefficients: 0.561***, 0.246*, 1.259***, 1.396***, 1.131***, 0.375**
    - Standard errors shown in table (not repeated here verbatim beyond parentheses in original).
  - unemp rate gap coefficient: -0.0253***, -0.0140*, -0.0250*** (selected columns).
  - vacancy rate gap coefficient: -0.0311 (in one specification).
  - tightness gap coefficients: 0.01870, 0.0324 (in selected specifications).
  - interaction log wage gap x tightness gap: 0.476*** (in a specification).
  - EU linked pair coefficient: 0.723***.
  - log distance coefficient: -1.074***, -1.084***, -1.068***, -1.068***, -1.099***, -1.072*** across columns.
  - Fixed effects: Year FE Yes; Origin FE Yes; Dest FE Yes.
  - R2 reported: 0.783, 0.784, 0.813, 0.812, 0.825, 0.788.
  - Observations (Obs): 6743, 6647, 3551, 3551, 3551, 6647 (as listed across columns).

### Lessons from previous EU accession cases and policy themes
- High-level findings from past accession episodes (not a causal analysis):
  - Average new member state experienced gains in output per capita and labor market outcomes with a slight decline in population.
  - Emigration tended to accelerate around accession, with some reversal later; immigration remained comparatively flat.
  - Output per capita growth in the average new member state was around 5 percent in the years leading to and following EU accession.
  - Labor participation increased and unemployment declined on average around accession.
  - Large variation across countries: population growth around accession ranged from around -1 percent to +1 percent among new member states.
- Policy themes implemented by new member states (summarized categories):
  - Structural reforms.
  - Attracting new capital and labor.
  - Increasing return migration.
  - Facilitating investment of diaspora.
  - Maintaining ties to diaspora.
- Note: Annex III provides a detailed table of policies; the main text cautions the list is indicative and not exhaustive.

### Model framework and next analytical steps
- A multi-country structural model of the European labor market with migration is introduced to analyze interactions between EU integration, migration, and labor market outcomes.
  - Model builds on Ayerst and Zhang (2025).
  - Households transition between employment, unemployment, and inactivity and decide whether to migrate.
  - Firms’ and households’ flows are modeled; country-specific parameters indexed by superscript c.
- The model is calibrated to match moments related to labor market dynamics and migration and is used to simulate:
  - Potential impact of further EU integration.
  - Policies to boost output and support labor markets in the Western Balkans.
- The model aims to highlight policy levers that maximize benefits from EU integration while mitigating risks from increased migration and population loss.

*Source: IMF Working Paper — Labor Markets, Migration and EU Integration in the Western Balkans (excerpted content).*

### 1. Labor participation block. Households choose to be active or inactive in the domestic labor market.

### 1. Labor participation block. Households choose to be active or inactive in the domestic labor market.

### Labor participation: setup and choice
- Households face a fixed cost 휏̅_c and an idiosyncratic cost 휖_c to enter the labor market if currently inactive, or to become inactive if currently active.
- Households entering the labor market do so as unemployed workers.
- Unemployed and inactive households receive income b_c, referred to as home production, which includes both direct (e.g., unemployment or retirement benefits) and indirect benefits (e.g., time for childcare).
- Inactive households may enter the labor market in future periods.
- Transition costs between inactivity and activity are represented by the common component 휏̅_c and the idiosyncratic component 휖_c.

### Employment and unemployment: matching, wages, and separations
- Unemployed workers and firm vacancies are matched through a search and matching process that follows Mortensen and Pissarides (1994).
- Unemployed households are matched with a firm at rate p_c(휃_c), becoming employed upon matching.
- Firms post vacancies at cost c to be matched with a worker at rate q_c(휃_c). Job-finding and vacancy-filling rates depend on endogenous local labor market tightness 휃_c (ratio of vacancies to unemployed workers).
- Matched firm-worker pairs produce output y_c and allocate revenues as wages and profits according to a Nash bargaining process until the match is exogenously destroyed at rate s_c.
- Firms incur vacancy posting costs c and receive profits while employed; firms may exit, and firms post new vacancies at cost c.

### Migration: costs and household decisions
- Unemployed and inactive households may choose to relocate across countries, paying a common bilateral migration cost 휏_c c' and an idiosyncratic cost 휖_c c' to move from origin country c to destination country c'.
- Households weigh economic factors (value of being a worker in a country—job finding rate, market wage rate, separation rate, option value of future moves) and non-economic factors (transition costs 휏̅_c and 휖_c, migration costs 휏_c c' and 휖_c c').
- Common migration costs 휏_c c' capture average difficulties (transportation costs, language/cultural barriers, difficulties finding work); idiosyncratic migration costs 휖_c c' capture household heterogeneity (willingness to move, familial connections, attachment).

### Theoretical predictions and selection
- Locations with:
  - lower employment frictions (higher matching efficiency, lower vacancy costs, lower separation rates), or
  - higher productivity,
  will tend to have higher labor participation, lower unemployment, and attract more migrants.
- Country pairs with lower bilateral migration costs tend to have higher migration flows, all else equal.
- Idiosyncratic household preferences generate supply curves of households willing to select into each country’s labor market that depend on the country’s relative economic fundamentals.

### Model estimation and cross-country parameter differences
- Sample: 35 European countries, including 27 EU members, EU candidates and potential candidates (the six Western Balkan countries and Türkiye), and Switzerland. Three EU candidate countries—Ukraine, Moldova, and Georgia—are dropped due to lack of data. Missing data are interpolated.
- The estimation yields country-specific estimates for five parameter sets and country-pair-specific migration costs.
- Parameters discussed and their empirical relationships:
  - Production of matched firms and workers y_c:
    - Related to net wage rate; reflects differences in total factor productivity and/or capital intensity.
    - Production y_c tends to vary positively with output per capita.
  - Matching efficiency M_c:
    - Determines rate at which unemployed workers are matched to vacancies.
    - Closely related to job finding rates in the data.
  - Vacancy cost c:
    - Determines firms’ rate of posting vacancies; related to labor market tightness in data.
    - Interpretable as firm entry cost before production can begin.
  - Scale of home production b_c:
    - Determines benefits of inactivity for unemployed or inactive households.
    - Captures direct benefits (unemployment or social welfare benefits) and indirect benefits (leisure, family care, remittances).
    - Closely related to flows between inactivity and activity and between unemployment and employment.
  - Transition cost 휏̅_c:
    - Cost paid by households to transition to and from inactivity.
    - Closely related to flows of workers between inactivity and activity.
- Empirical patterns:
  - Higher income countries tend to have:
    - higher production by matched firms and workers,
    - higher matching efficiency,
    - lower frictions for switching between inactivity and activity,
    - but higher vacancy costs and higher home production.
  - Higher vacancy costs in higher output-per-capita countries can reflect higher hiring expenses or higher rent of unused production space.
  - Higher home production could reflect higher opportunity costs of market provision of goods and services (e.g., family care costs).

### Bilateral migration costs and flows (estimation highlights)
- Estimated bilateral migration costs and calibrated bilateral migration rates are reported as heatmaps (Figure 9).
- Observations:
  - Destination choices are skewed toward a few popular destinations; popular destinations tend to be common across origin countries.
  - Example: Germany is a common destination for almost all countries in the sample; Annex I notes Germany, Switzerland, Italy, and Austria as among the most popular destinations for Western Balkan countries.
  - Popular destinations may have more accommodating institutions for migrants, stronger migrant networks, or simply be larger economies better able to absorb new workers.
  - Estimated migration costs are not strongly correlated with migration rates; migration flows are driven by economic factors, reflected in large wage gaps between Western Balkan countries and advanced economies.
- Calibration detail: bilateral migrant share of the origin country’s population is top coded at 0.1 percent in the reported figure; the smallest migration cost is normalized to zero in the reported figure.

*Source: wpiea2025226-source-pdf - 1. Labor participation block. Households choose to be active or inactive in the domestic labor market.*

### 4. Quantitative Analysis: EU Integration and

### 4. Quantitative Analysis: EU Integration and Policy Simulations

### Increased EU Integration — experiment design
- Two scenarios simulate further EU integration for the six Western Balkan countries:
  - Scenario 1: Only lower migration costs to the EU, productivity held constant.
    - Decline in migration costs calibrated to match a doubling of the emigration rate in each Western Balkan economy, corresponding to an average decline in migration costs by 6.1 percent.
  - Scenario 2: Same decline in migration costs plus an overall increase in output-per-capita of 30 percent.
    - Implemented by increasing the parameter that determines production of matched firm-worker pairs by an average of 14.5 percent.
- Rationale: Empirical evidence links EU accession with both lower migration costs and higher productivity (technological convergence, access to the EU Single Market, participation in value chains, higher FDI, structural reforms).

### Increased EU Integration — simulated outcomes and tradeoffs
- Scenario 1 (lower migration costs only):
  - Emigration: Sharp jump (doubling targeted by calibration).
  - Wage: Increase of around 0.7 percent on average.
  - Unemployment rate: Increase of around 1.8 percentage points on average.
  - Employment rate: Decline of around 1.4 percentage points on average.
  - Labor participation: Relatively flat.
  - Output-per-capita: Decline of around 2.8 percentage points on average.
  - Immigration: Increases but not enough to offset emigration → total population declines.
- Scenario 2 (lower migration costs + higher productivity):
  - Emigration: Increases but to a lesser extent than in Scenario 1.
  - Wage: Increase by more than 10 percent (reported as part of discussion).
  - Unemployment rate: Declines by 5.8 percentage points on average.
  - Employment rate: Increases by almost 8 percentage points on average.
  - Participation rate: Increases by about 4 percentage points on average.
  - Immigration: Increases by 0.4 percentage points on average.
  - Output-per-capita: Increase of 30 percent (targeted in experiment parameterization).
  - Population loss: Partly offset by lower emigration and higher immigration.
- Key interpretive findings:
  - Increasing productivity is necessary to maximize benefits of EU integration for output and employment.
  - Without productivity gains, lower migration costs can reduce output-per-capita and employment via higher emigration and reduced vacancies.
  - Even with strong productivity gains, emigration is still expected to increase (emigration rate continues to increase by over 0.2 percentage points despite wages rising by more than 10 percent).
  - Policies that attempt to restrict emigration or directly subsidize staying/returning migrants would be expensive and unlikely to be effective; targeting pull factors via structural reforms is more promising.
  - Strengthening institutions, improving rule of law, and tackling corruption can reduce non-economic push factors for migration.

### Simulation of policies to boost output and support the labor market — experiment design
- Policy categories simulated; parameter changes applied uniformly across the six Western Balkan economies:
  - Productivity-boosting structural reforms:
    - Implemented by increasing production of matched firms and workers by 1 percent.
  - Active-labor market policies (ALMP):
    - Implemented by increasing the matching rate of unemployed workers and vacancies by 10 percent.
  - Business promoting policies:
    - Implemented by decreasing firms’ costs of posting a vacancy by 10 percent.
  - Policies to boost labor participation:
    - Implemented by decreasing home production output (raises the benefits of being employed relative to inactivity).
- Magnitudes and interpretation:
  - Productivity adjustment (1 percent) would close around 2 percent of the model-implied wage gap between the Western Balkan countries and EU.
  - Matching efficiency adjustment (10 percent) would close around one-third of the model-implied gap between the Western Balkan countries and EU.
- Examples of practical measures (as described in the source):
  - Structural reforms: judicial reforms, improving business/investment environment, reducing bureaucratic hurdles, improving property rights, investing in infrastructure, improving access to credit and corporate bond markets, tackling informality.
  - ALMP: dual vocational and education training systems, personalized job search, youth training/apprenticeships, labor market information systems, on-the-job and lifelong learning.
  - Business promoting: one-stop shops for business registration, tackling informality, digitalizing public support, simplifying tax systems, improving public infrastructure.
  - Labor participation: increased access to affordable childcare and eldercare, on-the-job training/reskilling, adjusting pension ages with improved life expectancy and gender equalization.

### Simulation of policies — quantitative impacts and comparative results
- General outcomes across policy types (average impacts reported after a ten-year period; ranges vary across the six countries):
  - All simulated policies increase output per capita.
  - Most policies increase labor participation and employment, reduce unemployment, and increase wages — except for labor participation policies (see below).
  - The policies produce broadly similar magnitudes of change in output per capita, labor participation, employment, unemployment, emigration, and immigration.
- Specific cautionary result:
  - Policies targeting labor participation (modeled by decreasing home production output):
    - Negative impact on wages.
    - Increase net emigration due to lower home production reducing households’ outside option.
    - Rationale: lower home production reduces the nonmarket benefit of remaining in the domestic economy, encouraging migration despite higher participation.
- Policy implications from simulations:
  - There is no single policy dimension that dominates; a combination of measures across structural reforms, ALMP, business support, and participation policies can be used to support labor markets and boost output.
  - Emphasis on creating pull factors (higher productivity, better institutions, better public services) rather than attempting costly wage convergence or restrictive migration measures.
  - Policies to attract diaspora capital and return migration were noted as important in practice but are outside the scope of the current model and would be useful for future analysis.

*From: 4. Quantitative Analysis: EU Integration and Policy Simulations (wpiea2025226-source-pdf)*

### 5. Conclusions

### 5. Conclusions

### Progress and main risk
- Western Balkan countries have made substantial progress over the past decade in closing labor market gaps with the EU, notably labor participation and unemployment rates, and increasing output per capita.
- This progress has occurred alongside shrinking populations due to high emigration as highlighted in recent population censuses.
- High emigration could be a major headwind for future growth and could be potentially exacerbated by further EU integration, including through EU accession.

### Method and experiments
- The paper employs a quantitative multi-country labor market model to analyze interconnected and endogenous drivers of labor market outcomes.
- Two sets of experiments are simulated:
  - Experiment 1: Simulates further EU integration by mirroring productivity increases and migration cost decreases observed in previous EU accession countries. This experiment highlights the importance of increasing productivity during EU integration to maximize benefits.
  - Experiment 2: Assesses the effects of improving efficiency across labor market stages (production, employing workers, encouraging participation), showing such improvements can boost output and support labor market outcomes in integrated regional economies.

### Key quantitative finding
- Increasing labor productivity in the Western Balkan countries by 30 percent—the average of previous EU accession cases—would:
  - increase wages by around 13 percent,
  - close most of the unemployment gap,
  - and lead to almost no increase in net migration.
- Lower increases in labor productivity risk higher unemployment and a greater loss in population from higher net migration.

### Policy lessons and recommendations
- First: EU integration should be supported by productivity-enhancing structural reforms domestically.
  - EU integration alone may raise productivity through greater access to intermediate trade, supply chains, and financial integration, but productivity growth is critical to maximize benefits and counter potential negatives from higher emigration.
  - Reforms should be well-planned and ambitious.
- Second: Policymakers should prioritize low-cost, high-return policies that improve the attractiveness of working in the local economy.
  - The second experiment shows policies aimed at various labor market stages have broadly similar positive impacts on employment, labor participation, and output per capita.
  - Any policies that make working in the local economy more attractive are likely to achieve similar goals.
- Policies aimed at disincentivizing emigration or incentivizing immigration are not directly considered in the model and are unlikely to prove effective because the model-implied wages necessary to offset the increase in net migration are quantitatively large.

### Factors outside the model and areas for future work
- Remittances:
  - Remittances and FDI from the diaspora remain key growth engines and may lessen potential costs of emigration.
  - Diaspora ties often weaken over time, suggesting integration and structural reforms should aim to reduce reliance on remittances (Roldan, 2021).
  - Higher remittances primarily increase consumption and real estate investment rather than investment in productive capacity (OECD, 2022), and may act as an obstacle for long-term growth reforms (Fullenkamp, 2008).
- Informality:
  - Large informal sectors create barriers for formal firm investment and can constrain human capital accumulation and economic mobility (informal firms have less incentive to train workers and may stigmatize future employment).
  - Suggested actions to reduce informality include:
    - reducing bureaucratic barriers to firm entry in the formal market (such as through one-stop shops to business registration and digital public services),
    - removing tax incentives for informality,
    - improving access to public services.

*IMF Working Paper — Labor Markets, Migration and EU Integration in the Western Balkans*

### Annex IV. Model Description

### Annex IV. Model Description

### Economic Environment
- Economic setting:
  - Time is discrete and indexed by   ∈0  1  2...
  - Countries: 퐶 = {1 2 ... 퐶}
  - Households: 푁
푐 푡
 households per country-period.
  - Households receive i.i.d. taste shock 휖
푡
푐′
 over locations and search for employment in current market or relocate and search in foreign markets.
- Microstructure of the labor market:
  - Country-specific submarkets with random search between vacant firms and unemployed workers.
  - Matching function: 푀(푈
푡
푐

푡
푐
) = 푀
푡
푐
(푈
푡
푐
)훼( 
푡
푐
)1−훼.
  - Labor market tightness: 휃
푡
푐 = 푡
푐
/푈
푡
푐.
  - Job finding probability for unemployed: 푝(휃
푡
푐) = 푀(푈
푡
푐

푡
푐
)/푈
푡
푐.
  - Vacancy fill probability:  
(휃
푡
푐) = 푀(푈
푡
푐

푡
푐
)/ 
푡
푐.
  - Matches destroyed exogenously at rate .
  - Workers do not leave current jobs for unemployment (domestic or foreign) following Chodorow-Reich and Wieland (2020).
- Timing (three stages within each period; value functions written at end of second stage):
  I. Search phase:
    - Begins with unemployed from preceding period (separations or migrations), exogenous newly unemployed entrants, and deaths.
    - Firms observe numbers of unemployed and post 푡
푐
 vacancies.
    - Household search within their current submarket; matched with probability 푝(휃
푡
푐); wages bargained.
  II. Production and match destruction phase:
    - Each matched firm-worker pair produces output 푡
푐
.
    - Matches destroyed with probability  
푡
푐
.
  III. Relocation phase:
    - Unemployed draw idiosyncratic preferences 휖
푡
푐′
 over each country-sector 푐′ and choose whether to relocate.
    - Migration cost when relocating from c to 푐′: permanent τ
c c

.
    - Migrant workers enter new labor markets as unemployed.

### Model Equilibrium
- Worker’s problem:
  - Workers are risk neutral, infinitely lived; states: employed or unemployed.
  - State values: 푡
휔 푐
 for 휔 ∈ {푈 푊}.
  - Value of unemployment at period :  
    푡
푈 푐
 = 푏
푡
푐
 + 훽퐸[ max
푐 { 푡+1
푈 푐
   τ
c c

 + 휈휖
푡
푐 } ]
    - 푏
푡
푐
 denotes home production.
  - Continuation value entering search in period +1:  
    푡+1
푈 푐
 = 푝(휃
푡+1
푐
) 푡+1
푊 푐
 + (1 푝(휃
푡+1
푐
)) 푡+1
푈 푐
  - Value of employment:  
    푡
푊 푐
 = 푤
푡
푐
 + 훽(1  
푡
푐
) 푡+1
푊 푐
 + 훽 퐸[ max
푐 { 푡+1
푈 푐
   τ
c c

 + 휈휖
푡
푐 } ]
  - With 휖
푡
푐
 drawn from a Type-I Extreme Value distribution, the continuation term has closed form:  
    퐸[ max
푐 { 푡+1
푈 푐
   τ
c c

 + 휈휖
푡
푐 } ] ≡ Φ
푡
푐
 = 휈 [ log ∑
( exp{ 
푡+1
푈 푐
  τ
c c

}
)1/휈
]
- Firm’s problem (DMP structure):
  - Values: matched firm 푡
퐽 푐
 and vacant firm 푡
푉 푐
.
  - Value of a vacant firm:  
    푡
푉
 = max 
푐 {  − 푡
푐
 + (휃
푡+1
푐
)( 푡+1
퐽 푐
 − 푡+1
푉 ) }
    - 푡
푐
 represents vacancy posting cost.
    - Free-entry condition: 푡
푉
 = 0 in equilibrium.
  - Value of matched firm:  
    푡
퐽 푐
 =  푡
푐
 − 푤
푡
푐
 + 훽[ (1  −  
푡
푐
) 푡+1
퐽 푐
 +  
푡
푐
 푡+1
푉 ]
- Wage bargaining:
  - Nash bargaining with solution to maximize:
    푚 푥 
    푤
푡
푐
( 푡
퐽 푐
(푤
푡
푐
))1−휂 ( 푡
푊 푐
(푤
푡
푐
)  − 푡
푈 푐
)휂.
  - Outcome implies ( 푡
푊 푐
(푤
푡
푐
)  − 푡
푈 푐
)휂 = 푡
퐽 푐
(푤
푡
푐
) (1 휂).
- Labor market dynamics:
  - Migration flow from submarket c to c′ at time : 푡
푐 푐

.
  - Evolution of employment:  
    퐸
푡+1
푐
 = (1  −  
푡
푐
) 퐸
푡
푐
 + 푝(휃
푡
푐
) ∑ 
푐′
 푡
푐′ 푐

( 푈
푡
푐′
 +  
푡
푐′
 퐸
푡
푐′
 + 퐵
푡
푐′
 − 퐷
푡
푐′ )
    - 퐵
푡
푐 and 퐷
푡
푐 denote birth and death respectively.
  - Evolution of unemployment:  
    푈
푡+1
푐
 = (1  − 푝(휃
푡
푐
)) ∑ 
푐′
 푡
푐′ 푐

( 푈
푡
푐′
 +  
푡
푐′
 퐸
푡
푐′
 + 퐵
푡
푐′
 − 퐷
푡
푐′ )
  - With 휖
푡
푐
 Type-I Extreme Value, migration flows:  
    푡
푐 푐

 = exp ( 
푡
푐
 τ
c c

)1/휈 / ∑
푐̃
 exp ( 
푐̃
휏
푐 푐̃
)1/휈
.
- Equilibrium definition:
  - A stationary equilibrium consists of {푤
푡
푐
 휃
푡
푐
 푈
푡
푐
 퐸
푡
푐
 푡
퐽 푐
 푡
푉 푐
 푡
푐
 푡
푊 푐
 푡
푐 푐

 푡
푐
} such that:
    (i) worker values 푡
푈 푐
 푡
푐
 푡
푊 푐
 solve worker problems;
    (ii) firms enter if optimal, satisfying free-entry and setting 휃
푡
푐;
    (iii) wages 푤
푡
푐
 determined by Nash bargaining;
    (iv) unemployed choose submarket to search, determining 푡
푐 푐

;
    (v) masses 푈
푡
푐 and 퐸
푡
푐 evolve according to laws of motion;
    (vi) goods market clears.

### Model Calibration
- Calibration targets and procedure:
  - Calibrated to data from European Union and Western Balkan countries.
  - Some parameters set from literature; others calibrated internally to match data moments.
  - Table I in Annex II lists all model parameters (referenced in source).
- Externally calibrated parameters: {α 휂 휈 β}
  - Interpretations: matching function elasticity (α), bargaining power (η), scale of the Extreme Value distribution (ν), and time discounting factor (β).
  - Set to fixed values corresponding to existing literature.
- Internally calibrated parameters: { 
푡
푐

푡
푐
 푀
푡
푐
 푏
푡
푐
 휏ˉ
푡
푐

ˉ
푡
푐
}
  - Represent productivity, vacancy posting cost, matching function efficiency, home production value, migration cost and the outside search option for workers.
- Relationships between data moments and parameters (expressed as given):
  - Wage:  
    푤
푡
푐
 = 휂 푡
푐
 + (1 휂)(1 훽(1
푡
푐
)) 푏
푡
푐
 + 훽(1 훽)(1
푡
푐
)(1 휂) Φ
푡+1
푐 풜.
  - Labor market tightness:  
    휃
푡
푐
 = ( 푀
푡
푐
 / (휅
푡
푐
 푦
푡
푐
 − 푤
푡
푐
 1−훽 (1−푑
푡
푐
)) ) )1/훼.
  - Job finding probability:  
    푝ˉ
푡
푐
 = 푀
푡
푐
(휃
푡
푐
)1−훼.
  - Flow into inactivity:  
    푡
푐 풜 푐ℐ
 = exp ( 푏
푡
푐
 + 훽 Φ
푡+1
푐
 − 휏ˉ
푡
푐
)1/휈 / exp ( Φ
푡
푐 풜
)1/휈.
  - Flow out of inactivity:  
    푡
푐 ℐ 푐 풜
 = exp ( ̃푉
푡
푐 풜
 − 휏ˉ
푡
푐
)1/휈 / exp ( Φ
푡
푐 ℐ
)1/휈.
  - Outflow rate:  
    푡
푐 풮 − 푐 풮
 = exp ( 푉
ˉ
푡
)1/휈 / exp ( Φ
푡
푐 풜
)1/휈.
- Implications for calibration:
  - Given Φ
푡+1
푐
, the wage rate pins down productivity.
  - Labor market tightness pins down vacancy posting.
  - Job finding probability pins down matching efficiency.
  - Inactivity rate pins down home production.
- Migration costs backed out as:  
  휏
푡
푐 푐

 = 휈 log[ 푡
푐 푐

 / 푡
푐 푐
 × exp( 
푡
푐
)1/휈 / exp( 
푡
푐
)1/휈 ].

### Comparison of Model and Data Moments (Calibration and Goodness-of-Fit)
- The model replicates calibration moments used (Figure III.1 reported in source) for:
  - a) Net wage rate
  - b) Labor market tightness
  - c) Job finding probability
  - d) Activity to inactivity rate
  - e) Inactivity to activity rate
  - f) Emigration rates
- Goodness-of-fit (Figure IV.2 / Figure III.2 in source) checks:
  - a) Unemployment rate (model vs. data)
  - b) Output per worker (model vs. data)
  - Neither unemployment rate nor output per worker is directly targeted in calibration; both serve as out-of-sample goodness-of-fit checks.
  - Source reports: in both cases, the model is able to closely replicate the data and the cross-country trend.

*Source: Annex IV. Model Description — IMF Working Paper "Labor Markets, Migration, and EU Integration in the Western Balkans" (Working Paper No. WP/2025/226).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025226-source-pdf.pdf_