This paper explores possibilities of election forecasting in Lithuania, post-communist country which has a party system characterized by high levels of electoral volatility and fragmentation. The main argument of the paper is that despite such unfavorable conditions, election forecasting has a perspective. Since the sample of national parliamentary elections is too small for the statistical modeling, possibilities of forecasting at the level of municipal council elections are discussed. Analysis of voting in Lithuania’s local elections reveals that two main types of economic voting (vital element for the election forecasting) could be observed: a) general tendency to punish the parties that are in the national government, but less when economy is improving and more when economy is deteriorating; b) change of unemployment level is important in the explanation of vote change of the dominant party in the municipal council. Both of these elements (altogether with the lagged vote share and party split) are integrated into a pilot model (method of OLS regression) that strives to predict the vote share of the party that holds mayor’s post (dominant political power in the municipal council). Presented model explains more than 50 percent of variance in the dependent variable. Standard Error of the Estimate (10.5 percent) is quite high, but case diagnostics reveal that model predicts part of the cases very accurately. Model’s potential to forecast election results in at least some municipalities is assessed with application of the function for the 2011 election to the municipal councils.
