Impactssof the Monetary Policy Committee decisions on the foreign exchange rate in Brazil.
| Autor | Vicente, Jose Valentim Machado |
-
Introduction
Central banks closely monitor the daily exchange rate (FX) movements, since they impact future price dynamics and, thus, help in setting appropriate interest rate policies (Groen and Matsumoto, 2004). Indeed, the relationship between the interest rate and the foreign exchange rate has been extensively studied in the literature.
The classical approach is focused on the interest rate parity, which is a noarbitrage condition representing an equilibrium state under which investors will be indifferent to invest on interest-bearing instruments in two countries. Such approach relies on certain assumptions, such as capital mobility, where investors can readily exchange domestic assets for foreign assets, and perfect substitutability of assets (that is, similar in riskiness and liquidity). (1)
In this paper, we focus on the link between interest rate surprises (i.e., after the interest rate decisions made by the MPC--Monetary Policy Committee) and the FX rate daily returns. The main idea is to capture the impact of non-anticipated interest rate monetary policy decisions on daily FX rate returns.
The empirical strategy adopted here is to estimate a conditional mean model of the FX-rate, using a standard OLS approach, at daily frequency, with focus on interest rate surprises after MPC decisions, under different model specifications. The robustness analysis is conducted in two main dimensions: (i) taking into account the FX-rate second moment using GARCH models; and (ii) enlarging the set of control variables with five new time series that represent government interventions in the FX market.
This way, we contribute to the literature in three manners. Firstly, we build new measures of interest rate surprises under two different concepts, using daily market and survey information. Secondly, our analysis is done using data from an emerging economy (Brazil) instead of developed markets. (2) As far as we know, this is the first paper to use such approach to investigate the impact of MPC decisions on the FX rate dynamics in Brazil. (3)
Thirdly, our setup allows investigating the existence of asymmetric effects on the referred statistical relationship, which are very important for the implementation of monetary policy, especially for an inflation targeting country (and emerging economy) like Brazil.
Our empirical findings point out that an unanticipated easing of 100 basis points leads to an FX rate depreciation of roughly 3.4 percentage points (p.p.). In other words, if a decrease of 0.50 p.p. is expected in the Selic target rate, but the actual decrease is of 1.50 p.p., this represents a negative surprise of -1.00 p.p., which leads to a 3.4 p.p. increase in the FX daily return.
The rest of the paper is organized as follows: Section 2 presents a brief literature review, Section 3 describes the methodology, Section 4 presents the results and Section 5 concludes.
-
Literature review
A wide range of factors can affect the FX-rate, for instance: economic fundamentals, speculative transactions, and currency interventions, among others. However, the failure of the standard economic theory to explain foreign exchange rate behavior using key economic fundamentals (e.g., money supply, trade balance, national income) has been shown in the international economics literature since the classical papers by Meese and Rogoff (1983a,b). (4)
While macroeconomic theory has proposed several potential predictors of exchange rates (e.g., based on the purchasing power parity (PPP) hypothesis, the uncovered interest rate (UIP) parity condition or the monetary model), the forecasting contributions of such approaches have been under question since the influential findings of Meese and Rogoff. (5) Consequently, a vast literature has studied the forecast performances of empirical exchange rate models, and several explanations have been put forward. Just to mention a few examples:
* Mark (1995) argues that fundamental relationships (parity conditions) hold better in the long run;
* Kilian and Taylor (2003) claim FX rates can be predicted after properly considering nonlinearities;
* Bacchetta and Van Wincoop (2004, 2006) propose the scapegoat theory; (6)
* Christoffersen and Mazzotta (2005) and Sarno and Valente (2005) provide statistical/financial models that beat the random walk for density forecasting FX rates using higher (intra-day) frequencies;
* Molodtsova and Papell (2009) show Taylor rule models perform better than the monetary, PPP and forward premium models (especially for longer horizons);
* Wang and Wu (2012) argue that economic variables contain information that is useful for forecasting the distributions of exchange rates; and
* Rossi (2013) reviews the empirical evidence on FX rate forecasting in the presence of instabilities, concluding that the predictive content of many predictors in macroeconomics, finance and international finance is unstable over time.
The literature on FX rate dynamics in Brazil has also been fruitful in recent years. For instance, see Minella et al. (2003), Bevilaqua and Azevedo (2005), Silva Jr. (2010), Wu (2012), Kohlscheen (2014b), and Gaglianone and Marins (2017).
Our paper is closely related to the study of Fatum and Scholnick (2008), which shows FX rates only respond to the surprise component of an actual U.S. monetary policy change. According to the authors, the failure to disentangle the surprise component from the actual monetary policy change can lead to an underestimation of the impact of monetary policy (or even to a false acceptance of the hypothesis that monetary policy has no impact on FX rates). (7)
Our work is also linked to the work of Kearns and Manners (2006), which investigates the impact of monetary policy on the exchange rate using an event study with intraday data of four countries (Australia, Canada, New Zealand, and the United Kingdom). According to the authors, carefully selecting the sample periods ensures the policy change is exogenous to the exchange rate. They conclude that an unanticipated tightening of 25 basis points leads to a rapid FX rate appreciation of around 0.35 percent.
In this paper, we also measure the impact of unanticipated interest rate monetary policy decisions on the FX rate. However, differently from Kearns and Manners (2006) or Fatum and Scholnick (2008), we go a step further here and seek to answer the following empirical questions: What is the short-term impact of monetary policy interest rate decisions on foreign exchange rate in Brazil? Is this impact symmetric in terms of positive or negative surprises? Has it changed in recent years, under a lower interest rate regime?
-
Methodology
We use daily data from January 2006 to August 2020. (8) We model the relationship between interest rate decisions and FX rate returns using a set of OLS models, in which the dependent variable is the (log) return of the Real-U.S. Dollar exchange rate (R$/US$) between days t and t + 1. (9) The set of regressors is the following: Surprise arising from Copom (10) interest rate decisions (defined later in this section); Selic rate target (Selic target); dummy which identifies Copom meeting decision dates (Dummy Copom); Selic rate target first-difference (Selic target variation); and dummy for low interest rates periods, which is equal to 1 if Selic rate is below 10% per year, and equal to 0 otherwise (Dummy Low Interest).
It is important to highlight at this point that the dependent variable and all regressors (in particular, the interest rate surprises) are considered in our models one-day lagged of each other, that is, not contemporaneously. In this sense, the dependent variable [DELTA]ln([FX.sub.t+1]), observed on day t + 1, will be explained by regressors [X.sub.t], available on day t. It is important to highlight that the disclosure of the Selic rate target occurs after the closing of the markets considered. This way, we avoid the potential issue of endogeneity, pointed out by Kohlscheen (2014a), (11) and thus refrain from doing exogeneity tests between the FX rate change and the surprise variables.
We also include in the model the following set of control variables: Emerging Market Dollar Index first-difference (12) (EME DXY variation) and return of five-year CDS Brazil (Return of CDS Br 5Y). (13) In the first robustness check, we estimate GARCH (1,1) and GARCH-in-Mean models to take into account the second moment of the FX rate returns.
As an additional robustness check, we...
Para continuar a ler
PEÇA SUA AVALIAÇÃOCOPYRIGHT GALE, Cengage Learning. All rights reserved.
