Banking crises, regulation, and growth: the case of Russia

Abstract

Recent empirical analyses of the relationship between financial system development and economic growth find that financial system development causes economic growth, is a good predictor of growth and that its impact is relatively large. Moreover, the empirical literature predicts that the adverse effects of banking crises on economic growth will rise in the absence of an adequate response by the government. Hence, given the Russian government's failure to respond adequately to the 1998 banking crisis, Russia's strong economic growth since the crisis is a puzzle. This study uses simulations to conclude that the growth costs of the 1998 crisis were larger than previously suggested. The adverse effects were compensated by expansionary effects. The findings corroborate the importance of financial development in promoting growth in transition countries.

Acknowledgements

For their constructive comments the author would like to thank – without implication – two anonymous referees, participants of the conference of the European Economics and Finance Society in Gdansk in May 2004 and his colleagues at DIW Berlin.

Notes

¹ Russia's indicators 1 and 4 in 2002 were still at a level comparable to that of the group of very slow growing countries. And regarding indicators 2 and 3 a substantial improvement occurred only recently during 2001 and 2002.

² Indicators 1–4 in Table 2 are averages for the whole year, so that movements within the year are not visible: Indicator 1 was 0.18 in mid-1998, and fell to 0.15 in the third quarter of 1998. The impact of the crisis on indicator 2 is clearly visible. Indicator 3 shows a substantial decrease already two years prior to the crisis in 1996! This is due to banks’ buying of short-term Russian government treasury bills, which crowded out lending to the private sector and which was one cause for the banking crisis due to the government's default on its debt. The decrease of indicator 3 during two years prior to the crisis is thus highly interconnected with the crisis. This is a relevant aspect in the following simulations of developments that assume absence of the banking crisis. The reader may ask why indicator 4, which has the same numerator as indicator 3, did not decrease in 1996. The reason is that the denominator of indicator 4, nominal GDP, was strongly dampened due to the drastic reduction of inflation during 1996 and 1997 so that this ratio remained constant. Indicator 4 decreased only transitory at the end of 1998 and in the beginning of 1999 with the surge of inflation that raised the denominator. Indicator 6 is highly dependent on the oil price, which largely explains its volatility.

³ See Buchs (1999), Chapman and Mulino (2001), and Thießen (2000). Causes of the crisis and its immediate effects are described in Thießen (2004, appendix A), which also provides additional evidence of inadequate banking supervision in the years following the crisis so that the relatively poor grades given by the EBRD for Russia's banking sector reforms shown at the bottom of Table 2 may even have been quite generous.

⁴ There are two main explanations for the recovery. The first is that the real currency depreciation ‘liberated’ the economy from the brakes put on it by the previous long lasting overvaluation. The depreciation caused import substitution and growth of real non-energy exports. The second explanation is the strong rise of oil prices during 1999 and 2000 associated with a continuous moderate increase in oil production since October 1998 boosting energy export revenues. Econometric analysis below attempts to clarify the role of these factors during the recovery.

⁵ See Hoggarth et al. (2002) for estimates of the relatively high output losses caused by banking crises.

⁶ Many close observers of Russia's financial system development argue that the forbearance during and after the crisis in 1998 was clearly further damaging the financial system. See, for instance, Russian European Centre for Economic Policy (1999) and MFK Renaissance Capital (2000). Their analyses reject very strongly any idea of arguing that forbearance may have had any positive macroeconomic effects.

⁷ Owing to the relatively few degrees of freedom, objections concerning estimating such regressions for one country and a relatively short period cannot be convincingly rejected, although the regressions passed a battery of statistical tests concerning their validity. But since this study has a second pillar on which it is built, namely the simulations of the growth costs of the banking crisis using growth coefficients from the literature and not from any regression for Russia, and since these two approaches yield the same result in a scenario that assumes a continued moderate improvement of financial system development during 1998–2002, the objections could mean to forgo using a potentially important piece of evidence regarding the interpretation of economic developments and economic policies.

⁸ These include initial per capita GDP and the initial level of educational attainment (such as school enrolment), variables that control for economic policies such as measures of government size, inflation, the black market exchange rate premium, and openness to international trade, and also variables that measure political stability and ethnic diversity.

⁹ Also, the estimated coefficient values of these variables appeared to be more robust than the coefficient values of the other potential explanatory variables.

¹⁰ Especially investment but also the oil price could have been expected to affect growth with longer lags. Regarding investment an explanation why this was not found here could be that at least for the period since 1999 the strong investment growth since then was concentrated in the oil, gas and metals industries (Hanson, 2003, p. 367), and these investments may have been largely capacity increasing allowing increases in production in the same quarter.

¹¹ Each of the four indicators was transformed into an index set to 1 for December 1994 and these four indices were given equal weight (multiplied by 0.25 and then added).

¹² The individual financial indicators were also used. However, in the regression for the longer period 1995–2003, the coefficients and significance of the first differences of all financial indicators were fragile in robustness tests and especially financial indicators 2 and 3 were consistently statistically insignificant with an often changing estimated sign.

¹³ All tests discussed in this section and their results are reported in Thießen (2004), pp. 12–13.

¹⁴ The F-statistic (4, 23) was 2.74, which is below the critical value at 5% significance of 2.80.

¹⁵ The — statistic (2, 13) was 25.79, above any critical value.

¹⁶ For the test results see Thießen, 2004, p. 13.

¹⁷ Since the authors did not include indicator 3 in their study this scenario is built on indicators 1, 2 and 4 only.

¹⁸ In other words, only the impact on economic growth, which is solely due to changing the financial indicator values, is considered. When using actual growth rates the differences are about 3 percentage points higher in the three simulations.

¹⁹ This may not be an inconsistency: Since the growth coefficients from the literature are averages for a large country sample and a long time period, it is possible that behind this average exists a large and unknown deviation between the largest and lowest coefficient for individual countries. Other factors that may be important are that Russia's circumstances during transition may have been different than those average circumstances during the 35 year period used by Levine et al. (2000) and that the latter derived their result by using a large set of statistically significant explanatory variables, many of which were either not significant or not available in the Russian case.