Determinants of Unemployment in European Countries
A Panel Data Analysis using Eurostat and World Bank Data
Christian Castillo
2026-05-26
1 Abstract
This paper analyses the determinants of unemployment in selected European countries during the period 2010-2024 using panel data econometric techniques. The empirical analysis combines macroeconomic data from the World Bank and Eurostat. The dependent variable is the unemployment rate, while the explanatory variables are GDP growth, inflation, tertiary education attainment, and government expenditure as a percentage of GDP.
The study first presents a descriptive analysis of the data using summary statistics, maps, distribution plots, time-series graphs, scatter plots, and a correlation matrix. Then, Fixed Effects and Random Effects panel models are estimated. The Hausman test is used to select the most appropriate specification. The results support the use of the Fixed Effects model, suggesting that unobserved country-specific characteristics are relevant when explaining unemployment differences across European countries.
2 Introduction
2.1 Presentation
Unemployment is one of the most important macroeconomic and social problems faced by European countries. A high unemployment rate is not only an economic indicator of weak labour market performance, but also a source of social instability, income inequality, poverty, and lower living standards. For this reason, understanding the determinants of unemployment is essential for governments, policymakers, and international institutions.
The European context is especially interesting because countries share certain institutional and economic frameworks, but still differ significantly in terms of labour market structure, fiscal policy, educational attainment, productivity, and macroeconomic performance. Some countries have historically experienced persistent unemployment problems, while others have maintained more stable labour markets. These differences make Europe a suitable region for panel data analysis.
This study analyses the relationship between unemployment and a set of macroeconomic variables. In particular, the paper focuses on GDP growth, inflation, tertiary education, and government expenditure. These variables were selected because they represent different channels through which unemployment may be affected: economic activity, price dynamics, human capital, and fiscal policy.
The main objective of this paper is to estimate the effect of these variables on unemployment in European countries between 2010 and 2024. By using panel data methods, the analysis controls for unobserved country-specific characteristics that may remain constant over time, such as institutional quality, labour market culture, or structural differences between economies.
2.2 Literature Review
The relationship between economic growth and unemployment has been widely discussed in macroeconomic literature. One of the most relevant theoretical foundations is Okun’s Law, which states that higher economic growth is generally associated with lower unemployment. When output increases, firms tend to demand more labour, which reduces unemployment. Therefore, GDP growth is expected to have a negative effect on unemployment.
Inflation is also commonly related to unemployment through the Phillips Curve. According to this framework, inflation and unemployment may display an inverse relationship in the short run. When aggregate demand increases, production and employment may rise, while inflationary pressures also increase. However, this relationship is not always stable and may depend on monetary policy, external shocks, and expectations.
Education is another key factor in labour market outcomes. Human capital theory suggests that individuals with higher education levels are more productive, more adaptable, and more likely to find employment. At the country level, a higher share of population with tertiary education may reduce structural unemployment by improving the matching process between workers and firms.
Government expenditure may also affect unemployment through fiscal policy. Public spending can stimulate aggregate demand, finance employment programmes, and support economic activity during recessions. However, the relationship may also be complex. Countries with high unemployment may increase public expenditure as a response to economic weakness, which may generate a positive association between government expenditure and unemployment in empirical data.
Overall, previous literature suggests that unemployment is influenced by both cyclical and structural factors. This paper contributes to this discussion by applying panel data methods to a sample of European countries and by combining data from Eurostat and the World Bank.
2.3 Hypotheses
Based on economic theory and previous empirical findings, the following hypotheses are proposed:
H1: GDP growth has a negative and statistically significant effect on unemployment.
Higher economic growth is expected to increase labour demand and reduce unemployment. This hypothesis is consistent with Okun’s Law.
H2: Inflation is negatively associated with unemployment in the short run.
According to the Phillips Curve, higher inflation may be associated with lower unemployment, although the relationship may vary across countries and over time.
H3: Higher tertiary education attainment reduces unemployment.
A more educated labour force is expected to have better employment opportunities, reducing structural unemployment.
H4: Government expenditure influences unemployment.
Government spending may reduce unemployment through fiscal stimulus, although the empirical sign may depend on whether expenditure is used as a response to high unemployment.
3 Data Analysis
3.1 Data Description
The empirical analysis uses a panel dataset composed of selected European countries observed annually between 2010 and 2024. The sample includes Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Netherlands, Poland, Portugal, Romania, Spain, and Sweden.
The variables used in the analysis are collected from two main databases: the World Bank and Eurostat. The unemployment rate, GDP growth, and inflation were obtained from the World Bank. Tertiary education attainment and government expenditure were obtained from Eurostat.
# World Bank data
wb_data <- WDI(
country = c("ES","FR","DE","IT","PL","PT","NL","BE","AT",
"SE","DK","FI","IE","GR","CZ","HU","RO"),
indicator = c(
unemployment = "SL.UEM.TOTL.ZS",
gdp_growth = "NY.GDP.MKTP.KD.ZG",
inflation = "FP.CPI.TOTL.ZG"
),
start = 2010,
end = 2024,
extra = FALSE
)
# Eurostat education data
education_data <- get_eurostat("edat_lfse_03", time_format = "num")## indexed 0B in 0s, 0B/sindexed 1.00TB in 0s, 63.01TB/s education_filtered <- education_data %>%
filter(
sex == "T",
age == "Y25-64",
unit == "PC",
isced11 == "ED5-8"
) %>%
select(geo, TIME_PERIOD, values) %>%
rename(
country = geo,
year = TIME_PERIOD,
education = values
)
# Eurostat government expenditure data
gov_data <- get_eurostat("gov_10a_main", time_format = "num")## indexed 0B in 0s, 0B/sindexed 1.00TB in 0s, 11.13TB/s gov_filtered <- gov_data %>%
filter(
sector == "S13",
na_item == "TE",
unit == "PC_GDP"
) %>%
select(geo, TIME_PERIOD, values) %>%
rename(
country = geo,
year = TIME_PERIOD,
gov_expenditure = values
)
# Final panel dataset
wb_clean <- wb_data %>%
rename(
country_name = country,
country = iso2c
) %>%
select(country, country_name, year, unemployment, gdp_growth, inflation)
panel_data <- wb_clean %>%
left_join(education_filtered, by = c("country", "year")) %>%
left_join(gov_filtered, by = c("country", "year")) %>%
filter(year >= 2010 & year <= 2024) %>%
drop_na()
head(panel_data)## country country_name year unemployment gdp_growth inflation education
## 1 AT Austria 2010 4.883 1.8089824 1.8135344 19.1
## 2 AT Austria 2011 4.637 2.9274680 3.2865791 19.2
## 3 AT Austria 2012 4.909 0.6282462 2.4856756 19.8
## 4 AT Austria 2013 5.367 -0.2507264 2.0001562 20.6
## 5 AT Austria 2014 5.674 0.7557989 1.6058118 29.9
## 6 AT Austria 2015 5.802 1.3035233 0.8965633 30.6
## gov_expenditure
## 1 53.4
## 2 51.3
## 3 51.8
## 4 52.4
## 5 52.4
## 6 51.2The final dataset contains 240 observations and 16 countries. Each observation corresponds to one country in one year.
The dependent variable is the unemployment rate, measured as a percentage of the total labour force. GDP growth is measured as annual percentage growth. Inflation is measured through the annual consumer price index growth rate. Tertiary education represents the percentage of people aged 25-64 with tertiary education. Government expenditure is expressed as a percentage of GDP.
3.2 Descriptive Statistics
Before estimating the econometric model, it is important to examine the main characteristics of the variables. The table below presents the descriptive statistics of all variables included in the model.
descriptive_table <- psych::describe(panel_data[, c("unemployment", "gdp_growth", "inflation", "education", "gov_expenditure")])
descriptive_table## vars n mean sd median trimmed mad min max range
## unemployment 1 240 7.56 4.00 6.84 6.98 2.72 2.02 26.09 24.08
## gdp_growth 2 240 1.95 3.42 1.83 1.94 1.97 -10.94 24.62 35.56
## inflation 3 240 2.56 2.93 1.83 2.03 1.75 -1.54 17.12 18.67
## education 4 240 32.39 9.63 33.25 32.43 11.05 13.60 56.60 43.00
## gov_expenditure 5 240 47.62 7.51 48.95 48.31 6.60 20.60 64.90 44.30
## skew kurtosis se
## unemployment 1.87 4.80 0.26
## gdp_growth 0.99 9.40 0.22
## inflation 2.24 6.07 0.19
## education -0.04 -0.86 0.62
## gov_expenditure -1.08 1.91 0.49The descriptive statistics show that unemployment varies considerably across countries and years. This confirms the existence of important differences in labour market performance within Europe. GDP growth also displays significant variation, especially due to crisis periods such as the sovereign debt crisis, the COVID-19 shock, and the subsequent recovery.
Inflation presents more moderate values on average, although some years show higher inflationary pressure. Education and government expenditure also differ across countries, reflecting structural differences in education systems and fiscal policy models.
3.3 Distribution of the Variables
To better understand the distribution of the variables, histograms and boxplots are used. These graphs help identify dispersion, outliers, and skewness in the data.
panel_data %>%
select(unemployment, gdp_growth, inflation, education, gov_expenditure) %>%
pivot_longer(cols = everything(), names_to = "variable", values_to = "value") %>%
ggplot(aes(x = value)) +
geom_histogram(bins = 25, fill = "skyblue", color = "white") +
facet_wrap(~ variable, scales = "free") +
theme_minimal() +
labs(title = "Distribution of the Main Variables", x = "Value", y = "Frequency")
The histogram of unemployment shows that most observations are concentrated at moderate unemployment levels, although some countries and years present much higher values. GDP growth has a wider dispersion due to economic downturns and recovery periods. Inflation also shows some extreme values, especially in the most recent years.
panel_data %>%
select(unemployment, gdp_growth, inflation, education, gov_expenditure) %>%
pivot_longer(cols = everything(), names_to = "variable", values_to = "value") %>%
ggplot(aes(x = variable, y = value)) +
geom_boxplot(fill = "lightgreen") +
facet_wrap(~ variable, scales = "free") +
theme_minimal() +
labs(title = "Boxplots of the Main Variables", x = "Variable", y = "Value") +
theme(axis.text.x = element_blank())
The boxplots confirm the presence of heterogeneity across observations. Unemployment and GDP growth present several extreme values, which is expected in macroeconomic data. These differences justify the use of panel data models because they allow the analysis to exploit both cross-country and time variation.
3.4 Evolution of Unemployment over Time
The following graph shows the evolution of unemployment rates by country. It allows us to visually compare labour market dynamics across European economies.
ggplot(panel_data, aes(x = year, y = unemployment, group = country, color = country)) +
geom_line(linewidth = 0.8) +
theme_minimal() +
labs(title = "Evolution of Unemployment Rate by Country", x = "Year", y = "Unemployment rate (%)", color = "Country")
The time-series plot shows that unemployment evolved differently across countries. Some economies experienced persistently high unemployment during certain periods, while others maintained lower and more stable unemployment rates. This pattern reinforces the importance of controlling for country-specific effects in the econometric model.
3.5 Maps of Unemployment in Europe
Maps are useful to visualise the geographical distribution of unemployment across European countries. The first map shows unemployment in the latest available year, while the second map shows average unemployment over the full period.
latest_year <- max(panel_data$year)
map_data_latest <- panel_data %>%
filter(year == latest_year) %>%
select(country, unemployment)
# Eurostat country map
country_map <- get_eurostat_geospatial(
nuts_level = 0,
year = 2021,
resolution = "60",
output_class = "sf"
)
map_join_latest <- country_map %>%
left_join(map_data_latest, by = c("CNTR_CODE" = "country"))
ggplot(map_join_latest) +
geom_sf(aes(fill = unemployment), color = "white", linewidth = 0.2) +
scale_fill_viridis(option = "plasma", na.value = "grey90") +
theme_minimal() +
labs(title = paste("Unemployment Rate in", latest_year), fill = "Unemployment (%)")
map_data_avg <- panel_data %>%
group_by(country) %>%
summarise(avg_unemployment = mean(unemployment, na.rm = TRUE))
map_join_avg <- country_map %>%
left_join(map_data_avg, by = c("CNTR_CODE" = "country"))
ggplot(map_join_avg) +
geom_sf(aes(fill = avg_unemployment), color = "white", linewidth = 0.2) +
scale_fill_viridis(option = "viridis", na.value = "grey90") +
theme_minimal() +
labs(title = "Average Unemployment Rate, 2010-2024", fill = "Average unemployment (%)")
The maps reveal clear geographical differences in unemployment across Europe. Countries with higher unemployment tend to appear concentrated in specific areas, while other countries display lower average unemployment. These spatial differences suggest that national labour market structures and country-specific conditions matter, which again supports the use of Fixed Effects.
3.6 Correlation Analysis
A correlation matrix is used to examine the linear relationship between the variables and to detect potential multicollinearity problems.
cor_matrix <- cor(panel_data[, c("unemployment", "gdp_growth", "inflation", "education", "gov_expenditure")])
print(cor_matrix)## unemployment gdp_growth inflation education
## unemployment 1.00000000 -0.12574234 -0.275697111 -0.029381411
## gdp_growth -0.12574234 1.00000000 0.048526267 0.124483507
## inflation -0.27569711 0.04852627 1.000000000 -0.004448084
## education -0.02938141 0.12448351 -0.004448084 1.000000000
## gov_expenditure 0.13547670 -0.41142163 -0.077679061 0.038726684
## gov_expenditure
## unemployment 0.13547670
## gdp_growth -0.41142163
## inflation -0.07767906
## education 0.03872668
## gov_expenditure 1.00000000corrplot(
cor_matrix,
method = "color",
type = "upper",
addCoef.col = "black",
tl.col = "black",
number.cex = 0.7
)
The correlation matrix shows that GDP growth is negatively correlated with unemployment. This preliminary result is consistent with Okun’s Law, since stronger economic growth tends to reduce unemployment. Inflation also shows a negative correlation with unemployment, which is consistent with the short-run logic of the Phillips Curve.
Education presents a weak negative correlation with unemployment. This suggests that education may reduce unemployment, although the relationship is not very strong in simple correlation terms. Government expenditure shows a slight positive correlation with unemployment, which may indicate that countries with higher unemployment also tend to spend more through public budgets and social protection mechanisms.
Importantly, none of the correlations among explanatory variables are extremely high. This suggests that multicollinearity is unlikely to be a serious problem.
3.7 Scatter Plots
Scatter plots are used to visually examine the relationship between unemployment and the explanatory variables.
ggplot(panel_data, aes(x = gdp_growth, y = unemployment)) +
geom_point(alpha = 0.6) +
geom_smooth(method = "lm", se = FALSE, color = "red") +
theme_minimal() +
labs(title = "Unemployment and GDP Growth", x = "GDP growth (%)", y = "Unemployment rate (%)")
The scatter plot between GDP growth and unemployment suggests a negative relationship. Countries and years with higher GDP growth tend to be associated with lower unemployment rates. Although the relationship is not perfectly linear, the trend is consistent with economic theory.
ggplot(panel_data, aes(x = education, y = unemployment)) +
geom_point(alpha = 0.6) +
geom_smooth(method = "lm", se = FALSE, color = "red") +
theme_minimal() +
labs(title = "Unemployment and Tertiary Education", x = "Tertiary education (%)", y = "Unemployment rate (%)")
The relationship between education and unemployment is less visually clear, but the slope is slightly negative. This suggests that countries with a higher share of tertiary educated population may experience lower unemployment, although other country-specific factors also play an important role.
ggplot(panel_data, aes(x = inflation, y = unemployment)) +
geom_point(alpha = 0.6) +
geom_smooth(method = "lm", se = FALSE, color = "red") +
theme_minimal() +
labs(title = "Unemployment and Inflation", x = "Inflation (%)", y = "Unemployment rate (%)")
The inflation scatter plot shows a negative relationship with unemployment. This result is consistent with the Phillips Curve framework, although the relationship should be interpreted carefully because inflation dynamics differ across countries and periods.
ggplot(panel_data, aes(x = gov_expenditure, y = unemployment)) +
geom_point(alpha = 0.6) +
geom_smooth(method = "lm", se = FALSE, color = "red") +
theme_minimal() +
labs(title = "Unemployment and Government Expenditure", x = "Government expenditure (% of GDP)", y = "Unemployment rate (%)")
The scatter plot between government expenditure and unemployment shows a slightly positive relationship. This may reflect the fact that countries with higher unemployment require higher public expenditure, particularly through unemployment benefits and social protection programmes.
4 Model Estimation
4.1 Panel Data Model
The empirical model estimated in this paper is the following:
\[ Unemployment_{it}=\beta_0+\beta_1GDPGrowth_{it}+\beta_2Inflation_{it}+\beta_3Education_{it}+\beta_4GovExpenditure_{it}+u_{it} \]
where \(i\) represents the country and \(t\) represents the year. The dependent variable is unemployment. The explanatory variables are GDP growth, inflation, education, and government expenditure.
Two panel data estimators are considered: Fixed Effects and Random Effects. The Fixed Effects model controls for unobserved country-specific characteristics that do not change over time. The Random Effects model assumes that these country-specific effects are not correlated with the explanatory variables.
fixed_model <- plm(
unemployment ~ gdp_growth + inflation + education + gov_expenditure,
data = panel_data,
index = c("country", "year"),
model = "within"
)
random_model <- plm(
unemployment ~ gdp_growth + inflation + education + gov_expenditure,
data = panel_data,
index = c("country", "year"),
model = "random"
)
summary(fixed_model)## Oneway (individual) effect Within Model
##
## Call:
## plm(formula = unemployment ~ gdp_growth + inflation + education +
## gov_expenditure, data = panel_data, model = "within", index = c("country",
## "year"))
##
## Balanced Panel: n = 16, T = 15, N = 240
##
## Residuals:
## Min. 1st Qu. Median 3rd Qu. Max.
## -5.050492 -0.959716 -0.078774 0.928671 6.789067
##
## Coefficients:
## Estimate Std. Error t-value Pr(>|t|)
## gdp_growth -0.038819 0.038082 -1.0194 0.30915
## inflation -0.088533 0.044168 -2.0045 0.04625 *
## education -0.293571 0.031185 -9.4137 < 2.2e-16 ***
## gov_expenditure 0.153377 0.033263 4.6110 6.794e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Total Sum of Squares: 1280.8
## Residual Sum of Squares: 672.09
## R-Squared: 0.47526
## Adj. R-Squared: 0.42994
## F-statistic: 49.8139 on 4 and 220 DF, p-value: < 2.22e-16summary(random_model)## Oneway (individual) effect Random Effect Model
## (Swamy-Arora's transformation)
##
## Call:
## plm(formula = unemployment ~ gdp_growth + inflation + education +
## gov_expenditure, data = panel_data, model = "random", index = c("country",
## "year"))
##
## Balanced Panel: n = 16, T = 15, N = 240
##
## Effects:
## var std.dev share
## idiosyncratic 3.055 1.748 0.205
## individual 11.813 3.437 0.795
## theta: 0.8698
##
## Residuals:
## Min. 1st Qu. Median 3rd Qu. Max.
## -3.98649 -0.99012 -0.22712 0.80407 8.43061
##
## Coefficients:
## Estimate Std. Error z-value Pr(>|z|)
## (Intercept) 8.963593 2.232916 4.0143 5.962e-05 ***
## gdp_growth -0.036444 0.038841 -0.9383 0.34810
## inflation -0.109623 0.044614 -2.4571 0.01401 *
## education -0.263696 0.030169 -8.7406 < 2.2e-16 ***
## gov_expenditure 0.157324 0.032864 4.7871 1.692e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Total Sum of Squares: 1323.8
## Residual Sum of Squares: 749.72
## R-Squared: 0.43367
## Adj. R-Squared: 0.42403
## Chisq: 179.953 on 4 DF, p-value: < 2.22e-164.2 Fixed Effects Model
The Fixed Effects model is particularly appropriate when unobserved country characteristics are expected to influence unemployment and may be correlated with the explanatory variables. In the European context, these characteristics may include labour market institutions, social protection systems, industrial structure, and long-term policy differences.
The results of the Fixed Effects model indicate that education is negatively associated with unemployment. This supports the hypothesis that higher levels of tertiary education contribute to reducing unemployment. Government expenditure appears positively associated with unemployment, which may reflect the fact that governments tend to spend more in countries or periods with weaker labour market conditions.
4.3 Random Effects Model
The Random Effects model is also estimated for comparison. This model assumes that country-specific effects are random and uncorrelated with the regressors. However, this assumption may be restrictive in macroeconomic applications because country characteristics are often related to variables such as education, public spending, and unemployment.
4.4 Hausman Test
The Hausman test is used to choose between Fixed Effects and Random Effects. The null hypothesis states that the Random Effects estimator is consistent. If the null hypothesis is rejected, the Fixed Effects model is preferred.
hausman_test <- phtest(fixed_model, random_model)
print(hausman_test)##
## Hausman Test
##
## data: unemployment ~ gdp_growth + inflation + education + gov_expenditure
## chisq = 82.197, df = 4, p-value < 2.2e-16
## alternative hypothesis: one model is inconsistentThe Hausman test strongly rejects the null hypothesis. Therefore, the Fixed Effects estimator is selected as the preferred specification. This result indicates that country-specific effects are correlated with the explanatory variables, making the Fixed Effects model more reliable for this analysis.
5 Diagnostic Tests
5.1 Heteroskedasticity
A Breusch-Pagan test is performed to check whether the variance of the residuals is constant.
bptest(fixed_model)##
## studentized Breusch-Pagan test
##
## data: fixed_model
## BP = 9.3917, df = 4, p-value = 0.05202The Breusch-Pagan test does not provide strong evidence of heteroskedasticity at the 5% significance level if the p-value is above 0.05. If the p-value is close to the threshold, the result should be interpreted cautiously. In this case, the test suggests that heteroskedasticity is not severe.
5.2 Multicollinearity
Multicollinearity is checked using the Variance Inflation Factor (VIF). Since VIF is calculated from an OLS model, the following auxiliary regression is estimated.
ols_model <- lm(
unemployment ~ gdp_growth + inflation + education + gov_expenditure,
data = panel_data
)
vif(ols_model)## gdp_growth inflation education gov_expenditure
## 1.233498 1.006424 1.025906 1.220575The VIF results are close to 1 for all variables, which indicates that there is no serious multicollinearity problem. This means that the explanatory variables are not excessively correlated with each other and that coefficient estimates can be interpreted with reasonable confidence.
6 Residual analysis
fixed_residuals <- as.numeric(residuals(fixed_model)) fixed_fitted <- as.numeric(fitted(fixed_model))
plot( fixed_fitted, fixed_residuals, main = “Residuals vs Fitted Values”, xlab = “Fitted values”, ylab = “Residuals”, pch = 19 )
abline(h = 0, lty = 2)
hist( fixed_residuals, main = “Histogram of Residuals”, xlab = “Residuals”, col = “lightblue”, border = “white” )
qqnorm(fixed_residuals) qqline(fixed_residuals, col = “red”) ```
The residual plots provide additional information about the adequacy of the model. If residuals are randomly distributed around zero, this suggests that the model captures the main relationship between the variables. The histogram and Q-Q plot allow us to check whether residuals are approximately normally distributed.
7 Interpretation and Conclusion
7.1 Interpretation of the Results
The objective of this study was to identify the main determinants of unemployment in European countries using panel data econometric techniques. The analysis focused on GDP growth, inflation, tertiary education, and government expenditure.
The results confirm that country-specific effects are important when explaining unemployment. The Hausman test supports the Fixed Effects model, indicating that unobserved national characteristics are correlated with the explanatory variables. This is economically reasonable because European countries differ in their labour market institutions, educational systems, fiscal policies, and structural economic conditions.
The negative relationship between GDP growth and unemployment is consistent with Okun’s Law. Higher economic growth generally increases production and labour demand, which helps reduce unemployment.
Education is also an important factor. The negative coefficient of education suggests that higher tertiary education attainment is associated with lower unemployment. This result supports the human capital argument that a more educated labour force has better employment opportunities and greater adaptability to economic change.
Government expenditure shows a positive relationship with unemployment in the empirical results. This does not necessarily mean that public expenditure causes unemployment. A more plausible interpretation is that governments tend to spend more when unemployment is higher, especially through social protection, unemployment benefits, and countercyclical fiscal measures.
Inflation displays a negative relationship with unemployment in the descriptive analysis, which is consistent with the Phillips Curve. However, the strength and significance of this relationship may depend on the econometric specification and the period analysed.
7.2 Conclusion
This paper analysed the determinants of unemployment in selected European countries between 2010 and 2024 using panel data methods. The study combined data from the World Bank and Eurostat and estimated both Fixed Effects and Random Effects models.
The empirical evidence suggests that unemployment is influenced by both macroeconomic and structural variables. GDP growth and education are especially relevant, while government expenditure appears to be positively associated with unemployment, probably reflecting the response of public spending to weaker labour market conditions.
The Hausman test indicated that the Fixed Effects model is the preferred specification. This confirms the importance of controlling for country-specific characteristics when studying unemployment across Europe.
Overall, the results show that reducing unemployment requires not only short-term economic growth but also long-term improvements in human capital and institutional conditions. Future research could expand the model by including additional variables such as labour market regulation, minimum wages, demographic structure, or foreign direct investment.
8 References
Eurostat Database. (2024). European statistical database. Retrieved from https://ec.europa.eu/eurostat
World Bank. (2024). World Development Indicators. Retrieved from https://data.worldbank.org
Okun, A. M. (1962). Potential GNP: Its Measurement and Significance. Proceedings of the Business and Economic Statistics Section of the American Statistical Association.
Phillips, A. W. (1958). The Relation between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom. Economica, 25(100), 283-299.
Wooldridge, J. M. (2010). Econometric Analysis of Cross Section and Panel Data. MIT Press.
Baltagi, B. H. (2021). Econometric Analysis of Panel Data. Springer.
9 Appendix: Full R Code
All the R code used to download, clean, analyse, and model the data is included in the code chunks of this RMarkdown document. When the document is knitted, the tables, maps, graphs, and econometric results are generated automatically.