International Fixed Income Topic IIC Empirical Analysis How

International Fixed Income Topic IIC: Empirical Analysis: How Well Do the Parity Relations Hold?

OUTLINE • Review of Evidence on PPP • Alternative Exchange Rate Approaches • Forecasting – Forward Premium Puzzle

I. Recall the Evidence on PPP • Deviations from purchasing power parity can be substantial in the short run. • Real exchange rates can take several years to return to equilibrium. • Real exchange rates are autoregressive (that is, they depend on previous levels).

The empirical Record of RPPP in the long run

Evidence on RPPP Deviations

RXRs Over the (Very) Long Run: 1900 -1990

Implementation Issues A few problems/questions arise • The choice of a price index – choices include CPI, PPI, GDP • The choice of a base year/reference level – chose an “equilibrium” period. . . but (I) how do we know? , and (II) what does disequ’m mean? – some fundamental structural changes might have occurred • Relative prices might have changed – hence PPP is not expected to hold

PPP: an Exchange Rate Determination Theory • Causality is Goods market ==> Currency Market The adjustment mechanism: real$ is weak and domestic prices are low => export will rise & import will fall => a trade surplus will generate a real and nominal $ appreciation • We completely ignore – the role of financial transactions – the forward-looking nature of exchange rates – the stickiness of prices

Recall the Evidence on the Fisher Relation Difference in realized quarterly inflation (p£-p$) Difference in 3 -month Eurocurrency interest rates (i£-i$)

II. Alternative Approaches to Modeling XR Determination • Main challenge is to relate – Macro variables: • consumption, savings, taxes, investments, export, import, gvt’ deficit ==> supply and demand of FX THE FUNDAMENTALS APPROACH TO XRs DETERMINATION – Financial variables: • • interest rates, asset flows, volatility, risk premiums, stockmarket behavior central bank policy and intervention THE ASSET APPROACH TO XRs DETERMINATION • . . . and to provide theory(ies) which explain and reconcile – short, medium and long term behavior of XRs and RXRs – puzzles

The Balance of Payments Approach • Current Account: (sales and purchases of goods and services) – CA= X - M X - export of goods and services M - import of goods and services • As the RXR rises (real appreciation) X__, M__ => CA__ • In Equ’m: Y = C + I + G + CA GNP= consumption+investment+gvt’spend Open economy: add CA surplus(deficit) • RXR <==> CA(RXR) = Y - C - I - G. . . in order to (improve CA) <=> (reduce RXR) we need to increase Y and/or reduce C and/or reduce I and/or reduce G • Causality is the big leap of faith

The Asset Approach • The traditional approach: XRs adjust to eliminate trade imbalances ==> Countries with CA surplus will experience an appreciating currency • Empirical evidence is counterfactual to this theory consider especially the US during 1980 -85 (deficit + appreciation) • The asset approach treats XRs as a financial asset – attribute a FORWARD LOOKING role to XRs, consistent w/ other financial theories – explain large moves relative to fundamental

III. FORECASTING • What do we (think we) know ? – Higher interest rates ==> strengthening currency – Inflation rises ==> currency weakens …but for given real rates, high inflation <=> high interest rates – Trade deficit ==> weaker currency … but recall the experience in the US in the early 80’s and recent opposite experience in FE countries

Merrill’s March 98 3 mon Forecasts • 3 out of 5 forecasts in the “right direction” • Mean change is 3. 364% • Mean absolute error of forecast 4. 102% • HOW DO THESE FORECASTS COMPARE TO MARKET-WIDE AVAILABLE FORECASTS ? (e. g. , CURRENT XRs and FORWARD rates) Source: March 5 th, 1998 “Currency and Bond Market Trends” biweekly review

Exchange Rate Forecasting Market-Based Exchange Rate Forecasts E[Std/f] = S 0 d/f - The random walk hypothesis E[Std/f] = Ftd/f - The expectations hypothesis E[Std/f] = S 0 d/f [(1+pd)/(1+pf)]t - RPPP Model-Based Exchange Rate Forecasts • Technical analysis - uses the past history of exchange rates to predict future exchange rates • Fundamental analysis - uses macroeconomic data to predict future exchange rate changes

March 98 3 mon Forward Forecasts • 3 out of 5 forecasts in the “right direction” • Mean change is 3. 364% • Mean absolute error of forecast 3. 337%

Forward vs. Merrill • 3 out of 5 forecasts in the “right direction” in both • Forward forecast outperforms

Merrill vs. Random Walk • Random Walk MAE is 3. 170% • Outperforms both Forward and Merrill’s forecasts. . .

1 -month forward rates as spot rate predictors: Yen per dollar exchange rates 1970 -1995 Actual change in the spot rate Forward premium or discount

The Forward Premium Puzzle • The Expectations hypothesis Ft+1/ St= Et[St+1] / St • Consider the regression St+1/ St = a + b (Ft+1 / St) + errt • Random walk ==> a=___, b=___ • Exp’ Hypothesis ==> a=___, b=___

The Forward Premium Puzzle

The Forward Premium Puzzle: Long Horizon Evidence

Implications and Interpretations • With a beta=-1. 8, on average when i*>i by 1%, then the XR appreciates by 0. 88% instead of depreciating by 1% ==> Sell a currency trading at a premium(UIP predicts it should appreciate, but empirically the reverse happens), and buy a currency at a discount