The Rise of Private Property in Russia. Part 4
«The rise of income and wealth inequality in Russia»
We now present our results regarding the evolution of income and wealth inequality in Russia. We begin with income inequality and the very long-run trends, before moving to a closer analysis of the recent decades, comparison with other countries, and finally wealth inequality.
Our general results on the long-run evolution of inequality in Russia over the 1905-2015 period are summarized on Figures 8a-8b. The basic picture is pretty obvious: income inequality was high under Tsarist Russia, then dropped to very low levels during the Soviet period, and finally rose back to very high levels after the fall of the Soviet Union.
According to our benchmark estimates, the top 10% income share was about 45-50% in 1905, dropped to around 20-25% during the Soviet period, and rose again to 45-50% in the 1990s before stabilizing at this very high level since then. The top 1% income share was somewhat below 20% in 1905, dropped to as little as 4-5% during the Soviet period, and rose spectacularly to 20-25% in the recent decades.
Several remarks are in order. First, these broad orders of magnitude can be considered as reliable, but small variations should not be taken too literally, given the strong limitations of our data sources. In particular, our benchmark estimates suggest that inequality levels in Tsarist and post-Soviet Russia are roughly comparable. Very top income shares seem if anything somewhat larger in post-Soviet Russia. One can interpret this finding as showing that modern economic and financial technologies (including international oil markets and offshore wealth) are able to generate more extreme monetary inequality than traditional societies like Imperial Russia. One could also argue that extreme inequality is maybe less dramatic (and more acceptable) when average living standards are much higher.
However we should also make clear that the differences between the two periods may not be fully significant, first because the lack of detailed income tax data - and the general lack of financial transparency - make our estimates for the recent period relatively imprecise (we will later return on this); and next and most importantly, because the estimate for 1905 is at least as imprecise. It relies not on actual income tax data, which was never implemented in Tsarist Russia, but on income tax projections that were made by Imperial tax administration at the time the regime was considering the possibility to implement such a tax. Similar estimates were made in a similar context in other countries in the late 19th and early 20th centuries (e.g. in France), and the comparison between these projections and the actual income tax data generated by the application of the new fiscal system revealed that the tax administration was significantly underestimating top income levels (see Piketty, 2001). Of course we will never know what would have happened if an income tax had been implemented in Tsarist Russia, but these is a possibility that the same result would have prevailed. It seems safer to conclude that inequality levels in Tsarist and post-Soviet Russia are both very high - and roughly comparable, possibly with a somewhat higher level in the later period. (Lindert and Nafziger (2012) argue that the 1905 official inequality estimate might be somewhat ovestimated. However on the basis of similar estimates done by tax administrations in other countries (such as France, see above), we tend to reach the opposite conclusion. In any case, the data seems too fragile to draw a definitive conclusion about the comparison between levels of monetary inequalities prevailing in 1905 and 2005-2015).
Finally, it is worth stressing that the measures of monetary inequality depicted on Figures 8a-8b obviously neglect non-monetary dimensions of inequality, which may biases comparisons of inequality over time and across societies. For instance, inequalities in personal status and basic rights (including mobility rights) were pervasive in Tsarist Russia, and persisted long after the official abolition of serfdom in 1861. (For instance, according to the 1861 reform, the serfs were made responsible for compensation to landlords for loss of labor, with “redemption payments” to be made annually for 49 years (this resembles the compensation that Haiti had to pay to its former French slave-owners in order to obtain independence). These payments were later renegotiated, with extensive regional and local variations, but the general point is that the abolition of serfdom was a very gradual process, which in some cases reinforced the rights of landlords (rather the rights of ex-serves). In particular, there is ample evidence that landlords retained for several decades extensive coercion power to restrict the mobility rights of peasants (who were subject to a specific legal status and court system based upon unwritten “customary law” and largely controlled by local elites)). Summarizing such inequalities with a single monetary indicator is clearly an over-simplification of a complex set of power relations and social domination, and should be kept in mind when making historical and international comparisons.
The same general remark applies to the Soviet period. Monetary inequality was reduced to very low levels under Soviet communism (and also in other communist experiences, as we shall later see). For instance, a top 1% income share around 4- 5% means that top 1% income holders earn only 4-5 times the average income of the time, as compared to 20 times when the top 1% share is equal to 20%. This reluctance to rely on extended monetary hierarchies is a feature that is confirmed by all Soviet household surveys and administrative documents on salary scales. In addition, the Soviet regime abolished private ownership (except in some cases for small capital holdings) and therefore suppressed top capital incomes (which in other societies always represent a large fraction of top incomes). It also compressed very significantly the hierarchy of salaries and labor incomes.
However this obviously does not mean that the Soviet elite did not have access to superior goods, services and opportunities. This could take different forms - access to special shops, vacation facilities, etc. - which in effect could allow the Soviet top 1% to enjoy living standards that in some cases might have been substantially higher than 4-5 times average incomes (though probably quite a bit lower than under Tsarist or in post-Soviet Russia). Unfortunately we have no way to quantify this.
Finally, it is worth pointing out that although monetary inequality has been very low throughout the Soviet period, there are interesting medium term variations. Namely, we observe a very strong compression of the distribution of income during the first stage of the Revolution (resulting into a large inequality decline between 1905 and 1925), followed by a relative enlargement of income hierarchies between 1925 and 1956 during the Stalinist period, a gradual decline between 1956 and 1980, and a rise during the 1980s and at the beginning of the economic reforms. This periodization has already been noted by other scholars exploiting Soviet sources on the distribution of income and wages.
Who benefited from Post-Soviet transition?
We now look into more details at the recent period. First, it is striking to see that the rise in income inequality occurred very fast after the fall of the Soviet Union. According to our benchmark estimates, the top 10% income share rose from less than 25% in 1990-1991 to more than 45% in 1996 (see Figure 8a).
It is also worth pointing out that this enormous rise came together with a massive collapse of the bottom 50% share, which dropped from about 30% of total income in 1990-1991 to less than 10% in 1996, before gradually returning to 15% by 1998 and about 18% by 2015.
There is no doubt that hyper-inflation played a key instrumental role in the collapse of bottom incomes. Between 1990 and 1996, prices were multiplied by a factor of nearly 5000. Inflation was particularly high in 1992-1993 after official price liberalization occurred on January 1st 1992. A large part of bottom 50% income classes were made up of pensioners and low-wage workers whose nominal incomes were not fully indexed to price inflation, resulting into massive redistribution and impoverishment for dozens of millions of Russians households (particularly among the retired population). See Appendix A and section 3.
Low-end pensions and wages then benefited from a gradual recovery process between 1996 and 2015, but they never fully returned to their 1990-1991 relative income share.
Together with this process of rapid collapse and partial recovery for bottom income groups, we observe a more gradual and continuous process of rising top 1% income shares, from less than 6% in 1989 to about 16% in 1996 and over 26% in 2008. The top 1% share then dropped in the aftermath of the 2008-2009 financial crisis and stabilized around 20-22% since 2010 (see Figure 8a).
If we consider the period 1989-2016 as a whole, average per adult national income has increased by 41% according to our benchmark estimates, i.e. at about 1.3% per year. However the different income groups have enjoyed widely different growth experiences. The bottom 50% earners benefited from very small or negative growth, the middle 40% from positive but relatively modest growth, and the top 10% from very large growth rates.
From that viewpoint, the 1989-2016 looks very different from the 1905-1956 period, when most of the growth went to the bottom 90%, and also from the 1956-1989 period, when the distribution was approximately constant and growth was relatively balanced over all groups.
The fact that the growth incidence curve over the 1989-1996 period displays a strong upward-sloping profile is fully consistent with recent findings presented in the 2016 EBRD report on inequality dynamics in transition economies. There are two differences, however.
First, the growth incidence curve reported on Figure 9a is even more strongly tilted toward top incomes than the one presented in the EBRD report. This is because we use corrected inequality series combining survey data with income tax data and wealth data, while the EBRD growth incidence curve relies solely on self-reported survey data.
Next, the EBRD report uses a different income concept that we do and comes with a higher cumulated growth of average income over the 1989-2016 (i.e. about +70% instead of +41%). We think it is preferable to use per adult national income (as we do), and we recognize that it is very difficult to compare real incomes for the Soviet and post-Soviet periods in a satisfactory manner. E.g. if we were to evaluate the welfare costs of shortages and queuing in 1989-1990, then it is possible that our aggregate growth figures might jump from +41% to +70% or more (one reason why the EBRD report comes with higher cumulated real growth estimates over the 1989-2016 is because they look at household income, whose share in GDP and national income was unusually small in 1989- 1990. However to the extent that other components of national income also ultimately benefit to households it seems more justified to look at national income rather than household income). More generally, we should make clear that there is little doubt in our view that the welfare of the vast majority of the population has improved since the end of Communism. The interesting question is whether they could have improved even more and in a more balanced and egalitarian manner with different policies and a different inequality trajectory.
We should also point out that the income-tax-data correction plays a much bigger role than the wealth-data correction in our corrected inequality estimates.
This reflects the fact that the income tax tabulations include a significant number of declarations very high business and capital income flows. This is also reinsuring, in the sense that the data available for the wealth correction (namely Forbes billionaire data) is relatively limited and uncertain. In the Appendix, we provide detailed robustness checks and a number of alternative variant series for the incometax-data corrections. In all variants, corrected inequality levels are substantially higher than raw survey levels, and stand relatively close in magnitude to our benchmark series (by international and historical standards). See Appendix B, Figures B40-B42.
Finally, it is interesting to note that our corrected Gini coefficient reaches its peak value in 1996, due to the very low bottom 50% share measured for this year.
This contrast with top 10% and top 1% income share series, which reach their peak levels in 2007-2008 (see Figures 10a-10b). This illustrates the need to go beyond synthetic inequality estimates and to look separately at the different segments of the distribution.
Co-authored with Gabriel Zucman.
To be continued...
