Homework answers / question archive / TASK A (30 points)  Uncovered Interest Rate Parity  We start by testing uncovered interest rate parity

TASK A (30 points)  Uncovered Interest Rate Parity  We start by testing uncovered interest rate parity

Statistics

Share With

TASK A (30 points) 
Uncovered Interest Rate Parity 
We start by testing uncovered interest rate parity. You are provided with quarterly data for the spot exchange rate USD/JPY in the period 01 1980 - Q4 2017. You can choose the convention that you prefer, either direct quote or indirect quote. Both are provided in the Excel attachment, together with quarterly data for the interest rates of US and Japan. Interest rates have been computed from bonds at the following maturities: 3 months, 1 year, 3 years, 5 years and 10 years. All interest rates that you find in the Excel file have been annualized. Assume that the reference investor is American, i.e. the United States are the home country. We define St as the spot rate at time t, and we denote by st the natural logarithm of the spot rate at time t, such that st = In St. 
Run the following regression 
ASt+k = a + 11 (it - it*) + Et+kl 
where ASt+k = - S t+k - St denotes the difference of the logarithm of the spot rate between t + k and t. We denote by it the interest rate in the domestic country, and by it the interest rate in the foreign country, both at time t. Run this regression separately for different values of k: 3 months, 1 year, 3 years, 5 years and 10 years. 
Please note that each regression needs consistency between the horizon of the interest rate differential and the horizon of the exchange rate variation. In order to ensure consistency, please proceed as follows, always keeping the interest rate differential at annual rates as provided in the Excel attachment. 
• For k equal to 3 months, compute an annualized spot exchange rate variation by multiplying Ast+k by 4, since we have quarterly data. Afterwards, regress this 4 • ASt+k on (it — it). 
• For k equal to 1 year, simply regress ASt+k on (it — it), given that both sides of the equation are already expressed as annual rates. 
• For k equal to 3, 5 and 10 years, divide ASt+k by, respectively, 3, 5 and 10. Afterwards, regress thisis 3' (or  or k ist+k ist+0 k .. s —1, respectively, for the cases with 5- and 10-year maturities) on (it - i• 
Question A1: Comment on the relevant regression outputs, comparing them with the predictions of UIP. Does the UIP generally hold? 
Question A2: Comment on the differences in results that you obtained across maturities, focusing only on the indicator(s) that is (are) needed to corroborate your statements. Are these findings across maturities surprising or expected? Comment and explain. 
Question A3: What is the practical implication of the result that you have found with short maturities, and how could one exploit such pattern?