Regional gaps

Details

Economic inclusion

The final stage of the analysis attempts to measure regional inequality in terms of institutions, education and services, which probably reflects inequality of opportunity linked to people’s place of birth and place of residence within a country. This involves addressing the following two complications.

• First, internationally comparable data on institutions, education and services are rarely available at the regional level.
• Second, where such data exist, indices of intra-country inequality will depend on the definition of administrative regions, which may differ widely across countries. Consider two countries with identical intra-country inequality at the level of local institutions. These will appear to have very different levels of internal inequality if one country is divided into 10 regions, while the other is divided into just three. The level of inequality measured in the latter will be lower, because inequality within a region is not recorded.

To circumvent these problems, the next analysis is based primarily on LiTS (2010) data at the level of primary sampling units (PSUs). Imagine PSUs as micro-regions, each numbering about 20 respondent households, which are spread across a country to give a representative impression of the country as a whole. The fact that the PSUs are collectively representative and of equal size solves the problem that comparing administrative regions of different sizes may create spurious differences in inequality.

In addition, the LiTS contains plenty of information on households’ perceptions of local institutions and services, which is internationally comparable. The main disadvantage, though, is that it does not contain data for the SEMED countries.

The analysis focuses on four dimensions: differences in the quality of local institutions; access to, and quality of, services (such as utilities or health care); labour markets (local unemployment and the extent of informal employment); and education (quantity and perceived quality). With the exception of the quantity of education, which is drawn from an extensive regional-level dataset – see Gennaioli et al. (2013) – all data are drawn from the 2010 LiTS (see Table 5.5).

Regional inequality is measured in two ways: a Gini coefficient based on means for PSU (regional-level) data; and the difference between the mean of the top quintile of regions (that is to say, the 20 per cent at the top of the regional distribution for an indicator) and that of the bottom quintile. For the LiTS data, which comprise 50 PSUs in most countries, this means comparing the top ten PSUs (ranked according to a specific indicator) with the bottom ten. For the Gennaioli et al. (2013) data, the top and bottom regions were combined in artificial regions representing about 20 per cent of the population at both ends; means were then calculated and compared for these combined regions.

Although conceptually the benchmark against which inequality is measured is perfect equality, regions may be different as a result, for example, of geography and resource endowments. Therefore, the benchmarks against which gaps are measured are set empirically, based on the lower end of the observed distributions for the top-to-bottom difference and the Gini coefficient of each indicator (see Annex 5.2). The two resulting gap measures per indicator are subsequently averaged.

Table 5.6 shows the results. Across institutional dimensions regional gaps are largest in relation to labour markets, particularly in SEE countries, the Caucasus, Tajikistan and Uzbekistan. Gaps for access to local services are “medium” to “large” across most EBRD countries of operations – except for Belarus and Slovenia, where they are “negligible”. Regional gaps with regard to the quality of local institutions are mostly “medium” – with the exception of Bosnia and Herzegovina, Serbia and Uzbekistan, where they are “large”.

There are “small” education gaps in most CEB countries and about half of the SEE region, but Egypt, FYR Macedonia, Georgia, Moldova, Morocco, Serbia, Turkey and Uzbekistan all have “large” gaps.

• Prev
• Next