Appendix 4.B Additional Outputs
5.4 The relation between black and official markets
In economies with price controls it is common that black markets emerge as paral-lel markets, in which people are able to trade at different rates than official rates.
The reason why black markets emerge is excess demand for the good whose price is regulated. This is also what happened in Belarus’ foreign exchange market before the website was launched. (Ioffe & Yarashevich 2011) note that demand for foreign currencies exceeded supply at the official rates right before the website was launched:
“Belarusian banks [...] reported a shortage of hard currency. In late March 2011, hard currency altogether disappeared from the country’s exchange outlets, whereupon the end of unobstructed access to hard currency provoked a consumer panic with Belaru-sians promptly stocking up on such necessities as sugar, salt, and vinegar.” Since it was not possible to trade in official market at official rates the emerged black market acts as a substitute to the black market. Economically, however, black markets are not only relevant as substitutes for official markets. (Reinhart & Rogoff 2004) show that in most of countries that enacted price controls (pegs) in the foreign exchanges market after WWII black market exchange rates were better indicators of monetary policy and economic conditions than the official foreign exchange were. This makes
black markets an indirect factor for exchange rate-setting behavior.
Table 5.1: Probit estimation results.
premia w.r.t. Bloomberg premia w.r.t. NBB
(1) (2) (3) (4)
constant -2.219*** -32.846*** -5.631*** -33.962***
(0.314) (6.842) (1.378) (6.175)
7d premium -0.052 19.388* 9.572*** 20.921*
(0.731) (11.497) (3.123) (11.665)
7d volume -3.48×10−7*** -3.58×10−6** -8.89×10−7*** -4.01×10−6**
(1.06×10−7) (1.46×10−6) (2.83×10−7) (1.71×10−6) 7d range 4.81×10−3*** 0.029*** 2.34×10−3* 0.039***
(1.03×10−3) (6.19×10−3) (1.36×10−3) (5.22×10−3)
refinancing rate 0.503*** -0.077
(0.138) (0.220)
balance of payments -0.011*** -9.67×10−3***
(2.68×10−3) (1.99×10−3)
foreign reserve assets -1.08×10−3* 4.92×10−3*
(6.23×10−4) (2.86×10−3)
log likelihood -52.643679 -16.204368 -38.852761 -13.313116
observations 201 201 201 201
Note: The dependent variable is an indicator that becomes one during the seven days before each of the three major devaluations that occurred on May 24, September 21 and October 12 (beginning of the free-floating regime). The independent variables are averages of black market premia, volumes and ranges measured using volume-weighted standard deviations over the preceding seven days. The time period spans from April 29 to November 15, 2011. Specifications (1) and (3) show robust standard errors in parentheses, specifications (2) and (4) show standard errors clustered at the monthly level.
Following (Reinhart & Rogoff 2004), we perform probit regressions in which we regress dummies (which are equal to one during the seven days before the three devaluations and zero otherwise) on two different sets of variables. The first set of variables consists of average black market premia, average volume and average range during the preceding seven days. We use the black market premia with re-spect to the Bloomberg and NBB exchange rates separately. The corresponding estimation results of this specification are given in column (1) and (3) in Table 5.1.
The two estimation results are qualitatively very similar but quantitatively slightly different. The estimated coefficients indicate that black market premia and range
correlated with upcoming devaluations. The coefficient on black market premia is, however, not always statistically significant. The negative sign on the effect of vol-ume is probably due to the decrease in black market activity after the announcement of the free-floating regime and the special trading sessions with two segments in the official exchange rate market on August 30.
Note that, since black market premia depend crucially on official exchange rates, the dependent variable in our probit regressions is to some extent defined based on one of the right-hand side variables. This does not imply that black market premia have explanatory power by construction (black market rates might jump simultaneously).
But it is remarkable that black market volume and range display even higher levels of statistical significance than black market premia. Robustness checks for the less active black markets for euros and Russian rubles can be found in Appendix 5.C. In these markets, black market premia are the most significant predictors of upcoming devaluations.
The question arises whether the black market just captures information that is avail-able in other fundamental data. To investigate this question, in specifications (2) and (4) of Table 5.1 we include three fundamental variables in our probit regres-sions: the Belarusian refinancing rate and balance of payments, which are available at the monthly level, as well as foreign reserve assets, for which we have quarterly data.8 Our three measures of black market activity remain statistically significant predictors of devaluations even when we control for fundamental variables.
Fundamental variables are known to be of little use in predicting changes in free-floating exchange rates (Meese & Rogoff, 1983b; Meese & Rogoff, 1983a). In our setting with a fixed exchange rate regime, however, fundamentals are statistically significant predictors of devaluations even in the short run. Asset pricing models of exchange rates (Engel & West, 2005) explain both, the Meese-Rogoff puzzle and why it does not apply in our context: current fundamentals were priced in the black market but not in the fixed official exchange rates.
The findings in Table 5.1, however, do not necessarily mean that policy makers based their decisions on black market activity. The coefficients would be significant even if the black market were small and exogenous (Dornbusch, Dantas, Pechman, de Rezende Rocha & Sim˜oes, (1983)). But they imply that the black market effi-ciently anticipates the devaluations and regime changes that we observe in our data.
8We obtained these data on fundamental variables from Bloomberg.
Following the probit regressions, we conduct several additional statistical analy-ses to establish the time series properties and examine whether the black market is helpful in forecasting the devaluations. Similar investigations have been con-ducted by (Akgiray, Aydogan, Booth & Hatem 1989), (Booth & Mustafa 1991), (Moore & Phylaktis 2000), (Dawson, Millsaps & Strazicich 2007) and (Caporale &
Cerrato 2008).
We first test for the presence of unit roots in our exchange rate time series. Aug-mented Dickey-Fuller tests indicate that all series are integrated of order one (see Appendix Table 5.D.1). However, it has been shown that in the presence of struc-tural breaks, which we are likely to have in our data due to the devaluations and the regime change (see Figure 5.1), unit-root tests too often fail to reject the null hypothesis (Glynn, Perera & Verma, 2007). We thus apply the Lagrange multiplier unit root test with two structural breaks by (Lee & Strazicich 2003) where the break dates are endogenously determined.9 The results of these two-break LM unit root tests with estimated break dates can be found in Appendix Table 5.D.2.
In Figure 5.1, we plot the time series of exchange rates in the black market for dollars with estimated break dates and devaluations. The solid vertical lines indicate the estimated dates of the two structural breaks, the dashed vertical lines indicate the devaluations.10 The first structural break happened slightly before the first devalu-ation, while the second structural break happened approximately one month before the second devaluation. The timing of these estimated structural break dates in the black market exchange rate time series indicates that the black market was able to efficiently process information about upcoming devaluations.
9The technical details of the test can be found in (Lee & Strazicich 2003). A brief summary of the implementation is given in the Appendix 5.D. For the application of the test, we used GAUSS codes shared with us by the authors.
10Corresponding graphs for euros and Russian rubles, as well as for official exchange rates can be
300040005000600070008000900010000BM Exchange Rates USD/BYR
May1 Jul1 Sep1 Nov1
Date (2011)
Black Market Exchange Rates USD/BYR
Figure 5.1: Exchange rates and estimated break points: Dashed vertical lines are estimated break points. Solid vertical lines are the important dates; May 24: first devaluation, September 21: second devaluation, October 12: final devaluation.
The results of tests for Granger causality, which can be found in Appendix 5.E further support this interpretation. To apply the Granger causality tests, we first removed all trends in our time series according to the results of the LM unit root tests. We find that black market exchange rates Granger cause Bloomberg official exchange rates (see Appendix Table 5.E.4), but only weak evidence of Granger causality in the opposite direction. This does not necessarily imply economic causality, i.e. it is not necessarily the case that the Belarusian authorities devalued in response to the black market premium. But the black market rates reflected market conditions, to which the official exchange rates were adjusted only with some lag.
The analysis in this section builds on the argument that, while the official foreign exchange rates were still fixed, the black market processed available information effi-ciently whereas the official foreign exchange market did not. Additional evidence for efficiency in the black market, besides the stability of black market premia during normal times, can be found in Appendices 5.E and 5.F. Using tests for Granger causality (Granger, 1969) we show that information flowed from markets, in which there was more activity to less active markets. In Appendix 5.E, Table 5.E.1, we demonstrate that information spreads form the black market for dollars to the black markets for euros and Russian rubles. In Appendix 5.F, we demonstrate that
infor-mation spreads form the country’s economic center, Minsk, to the five next largest cities in Belarus.
5.5 Conclusion
Using data from the website Prokopovi.ch, we investigate the relation between black market and official exchange rates during the Belarusian currency crisis of 2011.
Prokopovi.ch, which was set up in April 2011, allowed Belarusian citizens to find potential trading partners in the black market for foreign exchange. Technological progress has, thus, helped Belarusians get around government restrictions more eas-ily: the creation of a website like Prokopovi.ch, which reduced transaction costs in the black market, would not have been possible a few years earlier.
During the time interval under consideration, Belarus experienced three devalua-tions, a period with fixed exchange rates in two segments and an exchange rate regime change. Various statistical tests suggest that the devaluations of the Belaru-sian ruble in the official market were anticipated by the black market. It is hardly necessary to repeat that we are dealing with highly interdependent forces. On the one hand, black market activity reflected adverse economic conditions during the crisis. These economic developments would have occurred even in the absence of a black market, but they also made the black market more valuable as an alternative for Belarusians who wanted to exchange foreign currency. On the other hand, the emergence of the black market may have worked as a catalyzer which accelerated the course of events that ultimately resulted in a move to free-floating exchange rates.
The episode we describe resembles developments that took place in many other countries around the world earlier and over a longer period (see Reinhart & Rogoff, 2004). (Gwartney, Lawson & Easterly 2006) and (Shleifer 2009) document that black markets for foreign exchange have vanished in nearly all countries (Belarus being an exception) between 1980 and 2005. This was preceded by a broad move to free-floating exchange rate regimes, at first glance a sign of financial liberalization.
Regular interventions in the foreign exchange market, however, suggest that, like many other central banks in the rest of the world, the National Bank of Belarus has merely switched towards more subtle measures to maintain control over foreign ex-change rates. The NBB would not be able to afford getting the exex-change rates back to the levels of the first half of 2011, which had highly overvalued the Belarusian ruble. It has, however, been able to keep the exchange rates within relatively narrow ranges since November 2011.
Bibliography
Akgiray, V., Aydogan, K., Booth, G. & Hatem, J. (1989), ‘A causal analysis of black and official exchange rates: The turkish case’,Weltwirtschaftliches Archiv 125, 337–345.
Booth, G. & Mustafa, C. (1991), ‘Long-run dynamics of black and official exchange rates’,Journal of International Money and Finance 10(3), 392 – 405.
Caporale, G. M. & Cerrato, M. (2008), ‘Black market and official exchange rates:
Long-run equilibrium and short-run dynamics’, Review of International Eco-nomics 16(3), 401–412.
Dawson, J. W., Millsaps, S. W. & Strazicich, M. C. (2007), ‘Trend breaks and non-stationarity in the yugoslav black market for dollars, 1974–1987’, Applied Economics39(1), 43–51.
Ding, S. & Kovtun, D. (2010), Republic of belarus: Selected issues, Imf country report no. 10/16, International Monetary Fund.
Dornbusch, R., Dantas, D. V., Pechman, C., de Rezende Rocha, R. & Sim˜oes, D. (1983), ‘The black market for dollars in brazil’, The Quarterly Journal of Economics98(1), 25–40.
Engel, C. & West, K. D. (2005), ‘Exchange rates and fundamentals’, Journal of Political Economy113(3), 485–517.
Glynn, J., Perera, N. & Verma, R. (2007), ‘Unit root tests and structural breaks: a survey with applications’,Faculty of Commerce-Papers p. 455.
Granger, C. W. J. (1969), ‘Investigating causal relations by econometric models and cross-spectral methods’, Econometrica 37(3), pp. 424–438.
Gwartney, J., Lawson, R. & Easterly, W. (2006), Economic freedom of the world: 2006 annual report, Technical report, Vancouver: Fraser Institute, www.freetheworld.com.
Husain, A. & Arora, V. (2012), Staff report for the 2012 article iv consultation and second post-program monitoring discussions, Imf country report no. 12/113, International Monetary Fund.
Ickes, B. W. & Ofer, G. (2006), ‘The political economy of structural change in russia’, European Journal of Political Economy22(2), 409–434.
Ioffe, G. & Yarashevich, V. (2011), ‘Debating belarus: An economy in comparative perspective’, Eurasian Geography and Economics52(6), 750–779.
Korosteleva, J. (2007), ‘Maximizing seigniorage and inflation tax: The case of be-larus’,Eastern European Economics 45(3), 33–50.
Korosteleva, J. (2011), ‘The global recession and the belarusian economy: Revealing cracks in the model’, Journal of Communist Studies and Transition Politics 27(3-4), 632–653.
Lee, J. & Strazicich, M. C. (2003), ‘Minimum lagrange multiplier unit root test with two structural breaks’,The Review of Economics and Statistics85(4), pp.
1082–1089.
Meese, R. A. & Rogoff, K. (1983a), ‘Empirical exchange rate models of the seventies:
Do they fit out of sample?’, Journal of International Economics14(1-2), 3–24.
Meese, R. & Rogoff, K. (1983b), The out-of-sample failure of empirical exchange rate models: Sampling error or misspecification?, in ‘Exchange Rates and In-ternational Macroeconomics’, NBER Chapters, National Bureau of Economic Research, Inc, pp. 67–112.
Moore, M. & Phylaktis, K. (2000), ‘Black and official exchange rates in the pa-cific basin: some tests of dynamic behaviour’, Applied Financial Economics 10(4), 361–369.
Reinhart, C. M. & Rogoff, K. S. (2004), ‘The modern history of exchange rate arrangements: A reinterpretation’, The Quarterly Journal of Economics 119(1), 1–48.
Rogoff, K. (1998), ‘Blessing or curse? foreign and underground demand for euro notes’,Economic Policy 13(26), 261–303.
Schneider, F. (2005), ‘Shadow economies around the world: what do we really know?’, European Journal of Political Economy21(3), 598–642.
Shleifer, A. (2009), ‘The age of milton friedman’, Journal of Economic Literature 47(1), 123–35.
Shleifer, A. & Treisman, D. (2005), ‘A normal country: Russia after communism’, Journal of Economic Perspectives19(1), 151–174.
Strazicich, M. C., Lee, J. & Day, E. (2004), ‘Are incomes converging among oecd countries? time series evidence with two structural breaks’,Journal of Macroe-conomics 26(1), 131 – 145.
Zlotnikov, L. (2011), The foreign exchange crisis in belarus: Causes and effects, Technical report, Center for European Policy Analysis.