Assesment of factors influencing emigration in the Baltic States

This paper analyses the causes of emigration from the Baltic States during 1998–2013 period. The theme of emigration, the causes and consequences of this phenomenon are not recent. However, it was noticed that the number of econometrical research dealing with the causes of emigration is insufficient. Most of the existing research is not complex, mainly focused on surveys excluding secondary data statistical analysis. Basic statistical methods, such as correlation analysis or binary linear regression analysis are entirely used in most of the research. It was also noticed that only non-lagged regressors are employed in most of the research. The research paper consists of three parts. In the first part, theoretical analysis of economical and non-economical causes of emigration is represented. By comparing scientific literature it was emphasised that the approach to the causes of emigration among researchers is very different. Five theories that are employed to explain the causes of emigration were analysed: neoclassical economic theory, dual labour market theory, migration network theory, transnational migration theory and the theory that explains emigration by income inequality. It was pointed out that none of the analysed theories can thoroughly explain emigration phenomenon, however, these theories are compatible. In the second part, the methods of empirical research are introduced. The research was conducted for 3 states: Lithuania, Latvia and Estonia. Secondary data from Eurostat, United Nations Development Programme and the World Bank databases were employed. Regression analysis was the main method used in the research. Multiple linear regression equations were constructed. The change of emigration number per 1000 people was chosen as a dependant variable. Unemployment rates, average monthly salary, Human Development Index (HDI), GINI coefficient were used to construct regressors. Accumulated emigration from Lithuania, Latvia, Estonia and dummy variables that reflect joining the European Union and Schengen zone effects were included as well. Separate multiple linear equations were made for all 3 states. Separate equations with lagged regressors were also constructed. Equations with the regressors of the same time lag were made, t–1 and t–2 time lags were chosen. The decision to use lagged variables was made because of the assumptions that the indicators of previous periods such as unemployment rate, average monthly salary, etc. can cause present emigration from Lithuania, Latvia and Estonia. F–statistics, t–statistics, adjusted R squared were used to assess fitness of regression equations. In the third part the results of empirical research are represented. By using t–statistics and F–statistics p values it was found that regressors should be considered as statistically insignificant. Adjusted R squared values are low. They vary from –1,075 to 0,546. It was found that signs of β coefficients of some actual equations do not match the signs of β coefficients of corresponding theoretical equations. Therefore, theoretical assumptions regarding the impact of regressors on dependant variable were not confirmed. Statistically insignificant regressors, low adjusted R squared values can be explained by the following reasons: 1) officially undeclared emigration from the Baltic States (emigration shock from Lithuania in 2010, when declared emigration per 1000 people increased from 6,58 to 25,3 confirms this assumption); 2) too short time series that was analysed (only annual values of some indicators such as Human Development Index (HDI), GINI, etc. during the period 1998–2013 were available); 3) factors that also affect emigration from Lithuania, Latvia and Estonia were not included into the model