Economic reforms and human development: evidence from transition economies

ABSTRACT

Do market-oriented economic reforms result in higher levels of human well-being? This article studies the impact of macro-level institutional and infrastructure reforms on the economic, educational and health dimensions of human well-being among 25 transition economies. We use panel data econometrics based on the LSDVC technique to analyse the effects of market-oriented reforms on the human development index (HDI), as a measure of human well-being, from 1992 to 2007. The results show the complexity of reform impacts in transition countries. They show that institutional and economic reforms led to positive economic effect and significant impacts on other dimensions of human development. We find some positive economic impacts from infrastructure sectors reforms. However, not every reform measure appears to generate positive impacts. Large-scale privatizations show negative effects in health and economic outcomes. The overall results show the importance of the interaction among different reform measures and the combined effect of these on human development.

KEYWORDS:

• Human development
• transition economies
• institutions
• market reforms

JEL CLASSIFICATION:

• C33
• O15
• P20

Acknowledgements

We gratefully acknowledge the inputs from Matthew Cable during the early stages of this research. We would also like to thank two anonymous reviewers for their helpful comments. The usual disclaimer applies.

Notes

¹ The countries included can be divided into three groups: Central Eastern Europe and Baltic States (CEB) comprising Croatia, Estonia, Hungary, Latvia, Lithuania, Poland, Slovak Republic and Slovenia; South-Eastern Europe (SEE) comprising Albania, Bosnia and Herzegovina, Bulgaria, FYR Macedonia, Serbia, Romania and Montenegro; Commonwealth of Independent States (CIS) comprising Armenia, Azerbaijan, Belarus, Georgia (left the CIS in 2009), Kazakhstan, Kyrgyzstan, Moldova, Russia, Tajikistan, Turkmenistan, Ukraine and Uzbekistan. Besides these countries, Turkey and Mongolia are included in the transition economies as per European Bank of Reconstruction and Development (EBRD) areas of operation.

² An explanation on how the HDI is calculated is presented in Table A2.

³ For a complete review of critiques of the HDI, see Kovacevic (2010).

⁴ See Milanovic (1998) for a discussion of income, poverty and social transfers in communist regimes.

⁵ Data on public expenditure are scarce for transition economies, especially for education. The gap between spending and efficiency is also a reason why spending amounts should not be taken into account.

⁶ The data for Figures 1–5 include a set of 25 transition economies explored further in the article. We exclude Bosnia, Serbia, Montenegro and Turkey. The explanations are provided in Section V.

⁷ It is common to take logarithms of GDP to proceed with analysis. However, the GDP index bounded between 0 and 1 already includes a form of logarithmic income discounting.

⁸ This specific issue affects Croatia and Macedonia in 1992 and 1993 (affecting a residual 2% share of the data set). Limited UNESCO data for 1993 does not lead us to reject the assumption of constancy.

⁹ Annual gross enrolment ratios for all levels combined (except pre-primary) are combined with limited data for literacy ratios (that are mostly constant through time) from the UNESCO database according to the education index formula to replicate it. This is done for Uzbekistan from 1999 onwards and for the whole sample period for Mongolia.

¹⁰ There is a railway infrastructure reform index, which is not included in our analysis due to lack of data.

¹¹ The EBRD now releases sectoral transition scores in their transition reports, which give further insight about progress of reforms, but that data are not available before 2010.

¹² Results of multicollinearity tests are present in Table A2. Allison (1998) points that a VIF value above 2.5 is problematic, although different authors give different rules of thumb. Belsley, Kuh, and Welsch (1980) points that a condition index above 30 is a serious problem.

¹³ Arellano–Bond and Anderson–Hsiao estimators can also perform this task. However, results are very similar between estimators, as Bruno (2005) points in Monte Carlo studies, therefore only the results using Blundell–Bond is presented.

¹⁴ A test for strict exogeneity is discussed in Wooldridge (2002, pp.284–285). Leads of regressors are added to the original regressions and a joint test of their significant is conducted. Failure to reject the null implies strict exogeneity holds.

¹⁵ Besides Treelet, other unsupervised methods like principal component analysis (PCA) lead to interpretation problems as each basis vector is a linear combination of all variables. In theory, supervised dimension reduction methods are superior. The sliced inverse regression (SIR) method (Li 1991) was attempted but the issues with interpretation did not greatly improve so we follow an unsupervised method that provides more intuition for this specific dataset. Sparse PCA (SPCA) could be an equivalent method of similar performance to achieve similar goals, but not as readily available on software packages.

¹⁶ Staehr (2005) points that the EBRD indexes are scored in the middle of the year, which already implies some distance in time to the determination of the dependent variables.

¹⁷ The CIS group also includes Ukraine (a founding state, but only associate since 1993), Turkmenistan (founding state, but only associate since 2005) and Georgia (founding state, left the CIS in 2009).

¹⁸ One of the effects of multicollinearity is that the SEs might be too high, leading to the lack of significance of variables that are significant.