Guest Contribution: “Capital Controls in Brazil: Effective”

Today we are fortunate to present a guest contribution written by Marcos Chamon, Senior Economist in the Research Department of the International Monetary Fund, and Márcio Garcia, Associate Professor of Economics at PUC-Rio. The views expressed in this blog are solely those of the authors and do not necessarily represent the views of the IMF, its management, nor its Executive Board.

Emerging markets have experienced a strong recovery in capital inflows in the aftermath of the systemic sudden stop in late 2008-early 2009. Flows reached levels comparable to their pre-crisis peak, driven by a combination of relatively favorable fundamentals in emerging markets and a “search for yield” in the context of low interest rates in advanced economies. These flows should, in principle, bring numerous benefits, but they may also bring risks. One concern is that massive inflows can lead to a strong appreciation of the exchange rate and loss of competitiveness of the tradable sector. Given large adjustment costs, a strong but temporary appreciation may cause lasting damage to industries which may not recover even after the flows abate and the exchange rate returns to its equilibrium level. Large inflows can also complicate macroeconomic management by further stimulating an already overheating economy, particularly if efforts to control inflation through higher interest rates attract more inflows. On the prudential side, there are concerns that flows may be associated with risky external liability structures, and more generally that the flows may not be directed to productive uses, and end-up fueling consumption booms and asset price bubbles instead.

Brazil has been one of the leading countries in this effort to manage inflows, and one of the most vocal against the loose monetary policy in advanced economy policies that are pushing capital towards emerging markets (the Brazilian finance minister, Guido Mantega, coined the term “currency wars”). It sought to limit inflows in the aftermath of the crisis, adopting taxes on portfolio inflows in October 2009. Over the following two years, Brazil adopted a series of other measures to discourage inflows, starting gradually to dismantle them in 2012.

The recent Brazilian experience provides an ideal context to study the effect of capital controls and restrictions. No other country with a similar level of integration with global financial markets has ever experimented as actively with market-based capital controls, placing Brazil on a category of its own.

Brazilian capital controls

On October 2009, the Brazilian government began to introduce what would become an extensive set of controls on inflows of foreign capital. It started with a 2% tax on financial transactions on foreign investments in portfolio debt and equity, collected at the initial currency conversion, similar to a Tobin tax. Several other measures followed. But since 2012, many of the controls have been relaxed or eliminated, suggesting that one more cycle of capital controls may be coming to an end (previous cycles include 1993-1998 when high interest rates combined with an exchange rate peg also attracted large inflows to Brazil).

Chamon and Garcia (2015) (forthcoming, Journal of International Money and Finance) analyze the recent Brazilian experience. We compare prices for similar financial assets available in Brazil and in the US. The first comparison is between shares traded in Brazil with their respective American depositary receipts (ADRs), which are based on the same underlying shares but are traded in the US market. If the controls have been effective, a premium as large as the magnitude of the tax on financial transactions (2%) should have risen. We find such a premium, but only at times of excess foreign demand for Brazilian shares (which typically leads to more issuance of ADRs). In the fixed-income market, the spread between the interest rate in dollars in Brazil (Cupom Cambial) and in the US is lower than the tax rate on financial transactions (6%), and temporary spikes following some of the controls tend to be short lived. Overall, we document that capital controls did produce a wedge between the Brazilian and international financial market.

But the focus of our analysis is the effect of the controls on the exchange rate. The Brazilian authorities were, as a rule, candid about the main motivation for the imposition of controls: combating the appreciation of the real. Thus, it is natural to use the developments in the exchange rate as the main criteria to evaluate the effectiveness of the controls.

The measures adopted were transparent and market-based. The inflow tax increases were announced when the market was closed and became effective on the following day, with one exception. This makes these policies particularly suitable for daily-frequency analysis, where presumably the newly announced taxes were the main financial news on a particular day. Our estimates also control for a host of variables that can affect the exchange rate (e.g. commodity prices, VIX, other exchange rates), and includes daily sterilized intervention data.

We find little or no effect on the exchange rate in the aftermath of the first several measures. While the exchange rate seems to revert from an appreciation trend following some measures, we do not find significantly strong effects even on specifications that consider longer time windows for the first measures. But the exchange rate seems to respond strongly to the last restrictions adopted, beginning with a tax on the notional amount of derivatives. Our estimates point to a response of 10 percent or more, even after controlling for other variables that affect the exchange rate. While our approach allows us to estimate that effect, it does not allow us to disentangle the particular channels through which the controls operated. This strong response may be the result of a cumulative effect of the several restrictions. That is, the response may have been large because the last measures finally closed the main remaining channels to bypass the inflow taxes. That result may also have been supported by the beginning of a monetary policy easing cycle.

Our results are much stronger than those typically found in the capital controls literature. That literature typically finds some evidence that controls affect the composition of flows (e.g. controls on portfolio flows leading to a shift towards FDI or longer maturities for which the control is less burdensome, although part of the shift may just reflect a relabeling of flows). That literature typically struggles to an effect of controls on reducing the volume of flows, and exchange rate pressures. The large effects estimated in our paper may be driven by the broad and extensive nature of the measures adopted in Brazil. It is also possible that unique features in the recent Brazilian experience provided a setting with better identification prospects.


The recent Brazilian experience suggests that controls can succeed in depreciating the exchange rate in the face of capital inflows. But that success may require several rounds of fine-tuning to close loopholes (which can also increase the scope for unintended consequences and “collateral damage” to the supply of FX hedging for the real sector). It also suggests that controls are more likely to succeed when they are supported by an easing of monetary policy. However, there is also the possibility that capital controls may have had the effect of lowering capital inflows even when they are needed. Brazil savings rate is only 16%, and capital inflows are needed to finance investment.

This post written by Marcos Chamon and Márcio Garcia.

One thought on “Guest Contribution: “Capital Controls in Brazil: Effective”

  1. Michael

    I have only taken a brief look at the paper, but I think is is worth sharing the following doubt because I have the same problem with many papers even after close reading, and there ought to be a way to dispel it it fairly quickly where it does not apply.

    The main results behind these conclusions seem to be based on regressions of the daily change in the logarithm of the exchange rate on some conventional control variables and dummy varialbes for the capital control measures. There is one such dummy for each capital control measure, and it is equal to one only on the day the measure was announced , and zero for the other 800-odd trading days in the sample period. With so little variation in the regressor, it can only be statistically significant if its estimated effect is very large — and in this case I believe that means implausibly large.

    Suppose, for example, the IV regression in Column 3 of Table 4 is perfectly specified, so it is measure true causal effects. Then the very first capital control — the tax on inflows — caused the Real to depreciate by a full log-point, which is to say some 170 percent! But because the effect is not statistially significant, it is part of what gets summarized above as, “We find little or no effect on the exchange rate in the aftermath of the first several measures.”

    The impact that is reported as the start of controls being effective is, in that regression, estimated at 1.515 log points, or about a 350% depreciation. And the significant impact that follows it, from a tax on borrowing with five years maturity or less, is 2.186 log points, which implies nearly eightfold depreciation.

    I don’t see how any of these can be plausibly interpreted as causal impacts. No change nearly that ever actually happens. Indeed, the difference between the maximum and minimum exchange rate in your chart appears to be a mere 1/3 of a log point. Also, they seem greatly disproportionate to the taxes of a few percent of transaction value.

    Have I misread something fundamental?

Comments are closed.