Over the last fifteen years, economists have started to seriously study ethnic, linguistic, and religious fractionalisation. The seminal papers of Mauro (1995) and Easterly and Levine (1997) seemed to show that greater levels of ethno-linguistic fractionalisation hinder economic performance, but this direct effect has been seriously challenged. Easterly (2001) argues that ethnic fractionalisation only slows down economic activity in countries with “sufficiently bad” institutions, while Alesina et al. (2003) show that such negative impact is strong mostly in autocracies. Garcia-Montalvo and Reynal-Querol (2005), meanwhile, argue that the effect of fractionalisation on economic performance is weak and suggest focusing on polarisation instead.
Following the recommendations of Alesina and La Ferrara (2005), my research with Ahmad Saleh and Vitaliy Kuzeyev contributes to this debate in two ways.
- Improving the measurement of diversity – fractionalisation is typically measured using secondary data compiled by Soviet researchers in the early 1960s.
- Modelling diversity as an endogenous variable – fractionalisation is often treated as a time-invariant, exogenous variable.
Improving the measurement of fractionalisation
Our research (Campos et al 2009) puts together a unique data set based on primary data from censuses and demographic surveys in order to capture potential changes in ethnic fractionalisation over time. In each country, collection was carried out by the same agency employing consistent methodologies. The sample covers 26 former centrally-planned economies in Central Europe and in the former Soviet Union from 1989 to 2007. The data contains the percentage of the population belonging to each ethnic, linguistic, and religious group in each country, for each of four time periods, generating a panel with 104 observations.
We then use this data to calculate a whole family of fractionalisation and polarisation indexes. Fractionalisation measures capture the probability that two randomly selected individuals belongs to different ethnic, religious, or linguistic groups. They take values between zero for a perfectly homogeneous country and one for a highly heterogeneous country. In addition, because fractionalisation measures increase in the number of underlying groups, polarisation measures were also computed. In this case, the maximum is reached with two groups of equal size.
Does fractionalisation change over time?
Figure 1 calculates the averages of ethnic fractionalisation indexes between 1989 and 2007 using this new data. Overall, these countries become more ethnically homogeneous over a short period of time. Moreover, the Visegrad countries of the Czech Republic, Hungary, Poland and Slovakia have been consistently the most ethnically homogeneous, while the Baltic countries have been consistently the most ethnically heterogeneous group of countries in this sample. The Figure also shows that the countries in the Asia group have experienced more severe changes in ethnic fractionalisation than the other groups. The simple correlation between economic growth rates and the measures of fractionalisation and polarisation is negative and ranges from -.24 to -.37.
Figure 1. Ethnic fractionalisation in transition: 1989 to 2007
Note: Figure 1 plots average fractionalization for the following groups of countries: Asia (Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Mongolia, Tajikistan, Turkmenistan and Uzbekistan), Balkans (Albania, Bulgaria, Croatia, Macedonia, Moldova and Romania), Baltics (Estonia, Latvia and Lithuania), BUR (Belarus, Ukraine and Russia), and Visegrad (Czech Republic, Hungary, Poland, Slovakia and Slovenia).
These changes in fractionalisation suggest that there may be value in re-thinking the common assumption that fractionalisation levels are time-invariant. This also raises the question of why fractionalisation changes over time.
For this group of countries, one general cause is migration flows. These may be driven by better economic performance and opportunities in the destination country as well as by inferior economic performance and/or civil war and ethnic cleansing in the origin country. In developing countries, such a process could take decades to unfold. But there are circumstances in this sample of transition countries which allow for this process to take place in a much shorter period of time.
- First, with the collapse of communism, workers become free to move to other countries in search of better economic opportunities. While under communism, severe mobility restrictions were in place.
- Second, the ubiquitous Russian minorities seem to have been made to feel unwelcome and the new economic and political situation after 1991. This resulted in return migration, causing the share of Russians to fall in every country in the sample, with the exception of Moldova.
- A third important factor may have been violent conflict such as the wars in the Caucasus and former Yugoslavia. This particular conflict, for example, resulted in the share of Serbs in Croatia declining from 12.2% in 1991 to 4.54% in 2001.
New insight on the effect of fractionalisation
We treat fractionalisation as an endogenous variable by making use of this temporal variation and using latitude and lagged fractionalisation as instruments. Our results support the recent literature and show that static – exogenous – diversity is indeed not robustly correlated with economic growth. But, when accounting for changes over time and modelling ethnic diversity as an endogenous variable, we find that ethnic fractionalisation is negatively related to growth.
Once fractionalisation is instrumented with latitude and lagged fractionalisation, we find a significant negative effect on economic growth.
Does fractionalisation change fast in poorer countries too?
The dramatic changes in the ethnic composition over such a short period of time are caused by various characteristics of the transition countries. While we do not think it reasonable to expect a similar magnitude of changes to take place in other groups of developing countries over the next ten years or so, data is becoming available that will allow us to relax the assumption that the degree of ethnic homogeneity has not change meaningfully in poorer countries since 1960. This will be useful in re-assessing our results as well as the recent discussion about the channels through which fractionalisation affect growth.
It is also clear that the construction of census-based measures for larger samples of developing countries over longer periods of time is a task that still requires some concerted effort.
Our results support the idea that fractionalisation has a negative effect on growth. Crucially, our new methodology suggests that less weight should be attached to the widely held assumption that ethnic fractionalisation does not change over time – or that it is possible to safely ignore these changes.
Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat and Romain Wacziarg (2003), “Fractionalization”, Journal of Economic Growth, 8: 155-194.
Alesina, Alberto. and E. La Ferrara, “Ethnic Diversity and Economic Performance,” Journal of Economic Literature, 43 (3), 762-800, 2005.
Campos, Nauro, Ahmad Saleh, and Vitaliy Kuzeyev (2009), “Dynamic Ethnic and Economic Growth in the Transition Economies from 1989 to 2007”, CEPR DP 7586,.
Easterly, William (2001), “Can Institutions Resolve Ethnic Conflict?” Economic Development and Cultural Change, 49 (4): 687-706.
Easterly, William and Ross Levine (1997), “Africa's Growth Tragedy: Policies and Ethnic Divisions", Quarterly Journal of Economics. CXII (4): 1203-1250.
Garcia-Montalvo, Jose and Marta Reynal-Querol (2005), “Ethnic Polarization, Potential Conflict and Civil Wars,” American Economic Review, 95: 796-816.
Mauro, Paulo (1995), “Corruption and Growth,” Quarterly Journal of Economics, 110(3): 681-712.