On the effectiveness of capital controls during the Great Recession: The Brazilian experience (2007–2013)

ABSTRACT

In this study, we analyze the impact in Brazil of a tax on external capital flows called “Imposto Sobre Operações Finaceiras” (IOF), by following data on a monthly basis from 2007 to 2013. Our goal is to determine whether a change in tax legislation can produce significant effects on the pattern of financial inflows from abroad. Using univariate structural models, our results show that changes in the IOF produced structural breaks on foreign portfolio investments.

KEYWORDS:

• Capital account
• capital controls
• capital flows
• macroprudential policies

JEL CLASSIFICATIONS:

• E44
• F41
• O16

Notes

¹Boughton analyzes the perspectives of Dexter White and John Maynard Keynes on international finance before and during the Bretton Woods Agreement negotiations. Accordingly, both economists agreed that exchange rates should be stabilized and private capital flows should be tamed in the new global financial order. Nevertheless, “what was second best to White was second nature to Keynes. In contrast to White’s acceptance of controls as occasionally to be tolerated, Keynes regarded them as essential for stable international finance” (Boiughton, 2002, p. 10).

²Gallagher (2015, p. 49) states: “For much of the twentieth century, the dominant view in macroeconomics was that cross-border finance needed to be regulated. This was seen as the way to balance ‘impossible trinity’ first sketched by John Maynard Keynes in his two books on monetary theory—especially in the post-Depression industrialized countries seeking to achieve full employment. The dominant view in Development Economics during the same period was that cross-border capital flows needed to be regulated for similar reasons and also to mobilize domestic resources for development … . In contrast, the view that capital mobility was something to be constrained had fallen out of favour in mainstream economics by the 1980s and 1990s.”

³This tends to happen when domestic banks come to regard borrowing abroad in foreign currency as a low-cost source of funds. However, if capital outflows force a devaluation, foreign-currency–denominated debts increase when measured in the domestic currency, which may lead to bank failures. Because financial institutions and private agents might take excessive risks on convertible currency (as seen in the 2007/2009 crisis), governments may need to bail them out (IMF, 2011, 2012).

⁴A proxy based directly on the nominal aliquot of the IOF has been discarded. A 2 percent aliquot in a given year, which was later increased to 6 percent, could seem, at first, of greater effectiveness, because it would make economic transactions more difficult. However, economic circumstances could just have made the 2 percent aliquot ineffective in affecting relative prices between the domestic economy and foreign financial markets and thus altered the dynamics of capital flows. In that case, the increase to 6 percent may only constitute the new adjusted efficiency point. Consequently, both aliquots can be equivalent regarding their efficiency in controlling a given amount of capital inflow through time. In this case, we choose not to base our proxy on the aliquots themselves insofar as, facing different circumstances, IOF aliquots of 2 percent or 6 percent may be equivalent to tame capital flows, considering the interest gap and other economic conditions that change over time. Therefore, the more appropriate alternative would be to grant the same importance to regulations that individually introduce different aliquots, independently of the specific aliquot each regulation has set forth.

⁵“Contribuição Provisória sobre a Movimentação ou Transmissão de Valores e de Créditos e Direitos de Natureza Financeira” (CPMF) was another tax on financial transaction, imposed on any removal of funds from bank accounts maintained in Brazil, in order to improve the central government’s tax collection.

⁶Particularly after decree 6,306/2007 replaced decree 4,494/2002.

⁷We also tested for loan inflows, but the results were not statistically significant. Due to lack of space, here we present only estimations for portfolio inflows.

⁸All series were seasonally adjusted by the method of X-12 Arima. It should be noted that the limited number of observations did not allow us to introduce a seasonal component (or to characterize business cycles).

⁹Source: Banco Central do Brasil, BC-8216 series, monthly, in million U.S. dollars.

¹⁰Source: Instituto Brasileiro de Geografia e Estatística, Sistema Nacional de Índices de Preços ao Consumidor (IBGE/SNIPC) (www.ibge.gov.br).

¹¹Source: Banco Central do Brasil, 17632 series—Central Bank’s Economic Activity Index (IBC-Br) index—with seasonal adjustment.

¹²Source: Banco Central do Brasil e Ipeadata, BC 4189 series—Selic interest rate accumulated in the month and annualized, by percentage in the year, divulged by Demab; United States–basic interest rate-Federal Funds—fixed by the FOMC (percentageper year).

¹³Source: Ipeadata, Country Risk (Risco Brasil), calculated by JPMorgan. Ipeadata EMBI+Risco-Brasil Series, monthly average.

¹⁴Source: Banco Central do Brasil, monthly average of the daily closing data for Ptax800 (Brazilian official foreign exchange rate) published by the Brazilian Central Bank, based on the BC-1series–Free Exchange.

¹⁵Series: VIXCLS. Source: Chicago Board Options Exchange. Not Seasonally Adjusted.

¹⁶This procedure corrects potential endogeneity. Moreover, it is consistent with exchange rate determination models that explicitly consider expectations. According to the seminal study by Meese and Rogoff (1983), economic fundamentals—such as the money supply, trade balance, and national income—are of little use in forecasting exchange rates, at least in the short and medium terms. They compared existing models to one of their own, which excluded fundamentals, and any exchange rate changes are purely random. Nevertheless, as soon as exchange rate expectations are considered, it is important to note that “the empirical evidence concerning which expectation mechanisms are best for modelling exchange rate expectations is very mixed. The most plausible story is that the appropriate mechanism is itself time-variant, with market participants sometimes having static expectations, sometimes extrapolative expectations, sometimes regressive expectations and so on” (Pilbeam, 2013, p. 227). Therefore, EX first difference is a plausible proxy for exchange rate expectations.

¹⁷We verified the presence or absence of autocorrelation in the time series, through the Box–Ljung test, which follows a chi-square distribution. The homoskedasticity test follows the F-distribution (for details, see Commandeur and Koopman [2007] and Durbin [2012]).

Additional information

Notes on contributors

Cesar Rodrigues van der Laan

Cesar Rodrigues van der Laan is an advisor in economics, Federal Senate of Brazil.

André Moreira Cunha

André Moreira Cunha is an associate professor in the Department of Economics and International Relations/UFRGS, and a CNPq research fellow.

Marcos Tadeu Caputi Lélis

Marcos Tadeu Caputi Lélis is an associate professor in the Department of Economics/Unisinos.