移行経済における中央銀行の独立性 ― インフレーション抑制効果のメタ分析 ―
Central Bank Independence in Transition Economies: A Meta-Analysis of the CBI Disinflation Effect

Summary/Abstract: In this paper, we aim to evaluate the central bank reforms in Central and Eastern Europe (CEE) and the former Soviet Union (FSU) through a meta-analysis of previous research works that empirically examined the impact of central bank independence (CBI) on inflation. We will also conduct a comparison with other developed and developing economies in terms of effect size and statistical significance of the CBI disinflation effect in response to concerns arising from Cukierman et al. (Journal of Monetary Economics 49, no. 2 (2002), pp. 237–264) about a possible overrating of the legal independence of the central banks in transition economies. From the results of the meta-analysis using a total of 282 estimates collected from 10 studies of transition economies and 12 studies of non-transition economies, we reached the following findings: First, the synthesized effect size using the partial correlation coefficient and the combined t value of the collected estimates are negative and statistically significant in both study areas, suggesting that, irrespective of the target region, the negative impact of CBI on inflation is verified in the literature as a whole. However, it was also revealed that the effect size and the statistical significance of the transition studies are inferior to those of non-transition studies. Second, the meta-synthesis of the estimates obtained from the transition studies indicated the possibility that their empirical results are strongly affected by a series of study conditions, including the target country, the estimation period, the data type, and the type of index used to measure CBI. Third, the meta-regression analysis (MRA) also reproduced the close correlation between various study conditions and the estimates reported in the transition studies. More specifically, the heterogeneity among studies of transition economies is caused by the choice of estimator, inflation variable type, degree of freedom, and research quality, in particular. Fourth, in contrast with the above-mentioned results of the meta-synthesis, the MRA using the collected estimates of both transition and non-transition studies revealed that there is no statistically significant difference between the two types of studies as long as we control for the degree of freedom and the quality level of the study. It is worth mentioning that we obtained similar results even when we utilized only the estimates of the legal CBI index. And, finally, according to the results of the assessment of publication selection bias, we found that, while the transition studies contain genuine empirical evidence of the disinflation effect of CBI in their estimates beyond such bias, the non-transition studies failed to provide evidence of a non-zero effect of CBI, due to the strong influence of publication selection bias on their empirical results. Based on these results, we conjecture that the existence of a group of countries such as Hungary and Georgia, which deviate considerably from the normal reform process, might lead to the meta-synthesis results that the effect size and the statistical significance reported by the transition studies as a whole are inferior to those of the non-transition studies. This is due to the situation of the transition economies where, while inflation was effectively controlled even in those countries that had outwardly low levels of CBI, such as Hungary, it was not controlled as effectively as it could have been in countries that appeared to have a high level of CBI, such as Georgia. However, the fact that the MRA does not support the results of meta-synthesis, in which study conditions are not controlled for simultaneously, indicates that the influence of these unorthodox countries on the empirical results is not as significant as we had predicted. Therefore, we conclude that the issue raised by Cukierman et al. (2002) is unfounded, considering the development of central bank reform in the 2000s as well. To sum up, the results of our meta-analysis strongly support the argument that socioeconomic progress is substantial, in the sense that there exists a close relationship between CBI and inflation in the post-communist world.

• Details
• Contents