1. Doctoral School of Finance and Banking Bucharest Uncovered interest parity and deviations from uncovered interest parity MSc student: Alexandru-Chidesciuc Nicolaie

2. Presentation contents Introducing UIP Deviations from UIP Methodology, data and empirical results

3. Why UIP? UIP is the cornerstone of international finance (it appears as a key behavioral relationship in almost all of the prominent current-day models of exchange rate determination) Since UIP reflects the market’s expectations of exchange rate changes, it represents the benchmark from which any analysis which depends on future exchange rate values must begin. Because of this, if there are reasons to believe UIP will not hold precisely, an investor must be able to identify the source of deviation and respond accordingly.

4. Notation used S t – nominal spot exchange rate at time t expressed as the price, in “home-country” monetary units, of foreign exchange (ROL against USD); S t e – expected nominal spot exchange rate at time t; F t – forward rate at time t; i t, respective r t – nominal interest rate at time t, respective real interest rate at time t in home country; i t *, respective r t * – nominal interest rate at time t, respective real interest rate at time t in foreign country.

5. Covered interest parity (CIP) The difference in interest rates between two countries is equal to the expected appreciation as measured by the forward exchange rate. In principal, this condition always holds because of arbitrage (no risk involved). The difference (f t – s t ) is called forward premium/discount

6. Uncovered interest parity (UIP) The difference in interest rates between two countries is equal to the expected rate of appreciation/depreciation in the spot market (if market participants are risk neutral). Thus, UIP ex ante is:

7. Uncovered interest parity (UIP) The version that appears in leading econometric models: - the disturbance term, which might represent time-varying risk premia or other effects

8. Forward premium puzzle If both CIP and UIP hold, a common test of UIP considers the following regression: In practice, for a wide range of currencies, is found significantly less than zero This is called the forward premium puzzle (forward premium anomaly) – interest differential predicts the wrong direction in which exchange rate moves

9. Deviations from UIP foreign exchange risk premia systematic forecast errors transaction costs intervention in the foreign exchange market capital does not flow freely across borders There are many reasons why Uncovered Interest Parity will not hold exactly, and can be even expected to fail:

10. Methodology, data and empirical results Empirical analysis has been made using monthly data from 1995/01 to 2000/12 for: – the average nominal exchange rate, – average passive interest rate used by banks for LEI operations (dpm), – loan interest rate in USA (Bank prime loan rate)(mprime) Test of UIP hypothesis Why do deviations occur? Joint tests of three parity conditions

11. Estimation of UIP I specified the regression according to Flood and Rose (1994) and Meredith and Chinn (1998) The above equation incorporates rational expectations I changed the interest rate series from annual percent to monthly percent. In this purpose we used two methods

12. Estimation of UIP Is there any connection between exchange rate change and interest rate differential? – Change in exchange rate (in logarithm) with respect to interest differential – Change in exchange rate (in logarithm) and the interest differential Properties of the regression variables; for this purpose we will perform unit-root tests: augmented Dickey-Fuller and Phillips-Perron

13. Unit-root tests Unit-root test for exchange rate change (in log) Unit-root test for nominal interest rate differential

14. Regression specification and result I tested the following regression: Dummy variable were included because of the shocks in early 1997 (d97) and end of 1998 and early 1999 (d99) The result: UIP doesn’t hold in case of Romania (it’s a standard result in international finance)

15. UIP estimation View correlogram

16. UIP estimation There is another way to test UIP (very simple) If sample mean of (ex post deviations from UIP) is statistically different from zero Is a stationary process?

17. Why do deviations occur? UIP equation can be written in terms of the real interest rate differential and real exchange rate growth ex post deviation from UIP is Where: is real interest differential logarithm of the real exchange rate

18. The real exchange rate If real exchange rate is random walk, then all movements in real exchange rate are unexpected I estimated to see the effect of current information dataset on

19. Estimation of real exchange rate real exchange rate change is not a random walk test reveals that UIP deviations are predictable, but doesn’t show how important is the predictable component

20. Sources of variances in UIP deviations I decomposed ex post deviations from UIP further into anticipated and unanticipated components of real exchange rate growth (Tanner (1998)) anticipated unanticipated

21. Joint tests of three parity conditions The test consists of parameter restrictions risk premia only affect nominal and real interest rate differential, but not inflation differential systematic forecast errors of exchange rate only affect nominal interest differential and inflation differential, but not real interest differential

22. Joint tests of three parity conditions The system that connects deviations from parity conditions to the current information set (I included here interest rate differential and inflation differential) is as follows: includes interest rate differential and inflation differential

23. Joint tests of three parity conditions – results Here the coefficients are:

24. Joint tests of three parity conditions – results There are no common factors to generate deviations from two parity conditions. I found evidence of systematic departures from all three parity conditions and this is consistent with the coexistence of both foreign exchange risk premia and systematic forecast errors in the foreign exchange markets.

25. Conclusions In line with results of other studies UIP doesn’t hold for Romania either – capital markets aren’t fully integrated with the internationals ones – there are bounds imposed to natural and legal persons regarding their investments in other countries – capital account isn’t fully liberalized

26. Conclusions The main components of deviations from UIP: the variance of the real exchange rate (anticipated and unanticipated) the risk premium bears an important influence too Joint tests of three parity conditions had shown: both factors are present on the foreign exchange market (risk premium and forecast errors )

27. Key Points Uncovered Interest Parity is the benchmark from which to view future exchange rate behavior; it requires having a clear understanding when deviations from UIP can/do occur, so that we can adjust our analysis accordingly; Ex-post deviations from Uncovered Interest Parity can be identified as being generated by systematic forecast errors and by risk premia

28. Nominal exchange rate (ROL against USD) from 1995:01 to 2001:12

29. Nominal exchange rate change (in logarithms)

30. Average passive interest rate used by banks with their clients

31. Interest rate in USA (Bank prime loan rate) – averages of daily figures

32. Change in exchange rate (in logarithm) and the interest differential

33. Change in exchange rate (in logarithm) with respect to interest differential

34. System estimation (UIP and RIP)

35. System estimation (UIP and PPP)

36. System estimation (PPP and RIP)

37. Correlogram for UIP regression

38. Deviations from UIP, PPP and RIP