Abstract
The objective of this paper is to find evidence whether Lithuanian municipalities use new technologies to disclose all relevant budget information in a timely, systematic, and comprehensive manner. The article creates a budget transparency index for each municipality using criteria from theory and previous research and then builds an empirical model to identify what determining factors make the budget process more transparent in some municipalities than in others. The results show that the percentage of population living in rural areas and turnout at local elections are negatively related to budget transparency while the level of debt is positively related to the level of transparency. The findings also provide evidence that revenue per capita and level of intergovernmental grants are negatively related to budget transparency. The research contributes to the existing literature by adding Lithuania to the relatively small set of countries that have developed budget transparency indices for subnational governments.
Notes
1 To be clear, Albalate del Sol (Citation2013) provides separate least square estimates for different areas of transparency, including fiscal transparency.
2 The detailed explanation of scoring strategy is available upon request.
3 The explanation how individual scores are calculated for each of the criteria is available upon request.
4 Descriptive statistics of scores for each criterion are available upon request.
5 Given the risk of heteroscedasticity in regression analysis with cross-section data, the robust standard error estimator has been used. Multicollinearity is not a serious concern, according to tolerance and variance inflation factors (VIF). The rule of thumb is not clear on this test. Some authors say that VIF greater than 10 indicates multicollinearity. Other authors say that VIF greater than five indicates the presence of the problem (Studenmund, Citation1997, p. 276). Besides, despite near collinearity, OLS estimator still retained the BLUE property (Gujarati, Citation2006). In this analysis, the average VIF score is 3.22, and passes a stricter of the two tests. Although one variable in this estimation demonstrates that VIF higher than five (edu) and the presence of multicollinearity may cause this variable to become statistically insignificant, it was decided to leave this variable in the equation to avoid a model specification error (Gujarati, Citation2006, p. 380).
Additional information
Notes on contributors
Liucija Birskyte
Dr. Liucija Birskyte is an Associate Professor in the Department of Economics and Business at Mykolas Romeris University, Vilnius, Lithuania. Liucija Birskyte earned her Ph.D. in Public Affairs at the University of Indiana, Bloomington, USA in 2008. She specializes in local government finance and budgeting, tax administration and tax compliance, transparency in public finance. She can be reached at [email protected].