Forward premium puzzle in Brasil: risk premium and order flow

Imagem de Miniatura
Muinhos, Marcelo Kfoury
Título da Revista
ISSN da Revista
Título de Volume

The uncovered interest parity postulates the relationship between two currencies over time should be given by the nominal interest rate differential. However, for over four decades, the forward exchange rate fails as a good predictor of the future rate, in a sense to reject the uncovered interest parity and the risk neutrality efficiency market hypothesis. The incapacity to explain such failure combined with unexpected odd results became known as the Forward Premium Puzzle. This paper aims to understand the Brazilian case of the puzzle and evaluate the benefits of alternative approaches vastly used in past literature. The lines of explanation to understand this enigma are many and long dated, usually related to factors that promote the unpredictability of the future exchange rates, such as changes in risk premiums, the influence of order flows and other inefficiencies. Recent studies opened roads to challenge the standard model, broadly known as the Fama Regression, in order to mitigate disturbances on the variance of the premium component and even reduce the bias observed in the forward interest differential coefficient estimates. In order to contribute to the existing literature, this study will assess the effects of (1) risk premium and (2) impacts of order flows on the USDBRL currency pair, when introduced to the traditional model. The analysis covers most of the floating regime with and without segmentation in sub-periods and primarily uses Ordinary Least Squares as estimator. Another contribution is the alternative approach to the models via Markov-Switching, that specially addresses quite of frequent distress periods in emerging currencies. The results from estimated models support the hypothesis of relationship for shorter horizons between the order flow with exchange rate depreciation and the forward premium. The risk premium, although significant only for the longest horizon in the premium regression, its presence improves the overall behavior of interest differential as estimator. The forward bias is reduced with the new terms, but not entirely mitigated. The key takeaway from this work is the MS approach, that allows the UIP to hold during high-volatility regime and provides insights about when the UIP fails.

Área do Conhecimento