The Misalignment of the Korean Won: Is It Overvalued? Taizo MOTONISHI Kansai University September 2006
Motivation There is much discussions on exchange rate misalignment Is Korean Won Overvalued? Is Japanese Yen Undervalued? Is Chinese Renminbi Undervalued? Are there misalignments for other Asian currencies?
Importance of Misalignment (1) For countries with free-floating exchange rate regime, misalignment is important for forecasting future economic conditions For countries which intervene foreign exchange market, misalignment is the benchmark of intervention For countries with fixed exchange rate regime, misalignment is the measure of sustainability of the current regime
Importance of Misalignment (2) For future possible financial cooperation among Asian countries, exchange rate arrangement can be one key issue
The Importance of Misalignment for Korea Recent Won appreciation Misalignment increasing or decreasing? No misalignment? Large amount of Yen denominated debt of Korean Banks Possible Won depreciation against Yen increases the burden How large is the potential burden?
The Main Result: Estimated Equilibrium Won Rate
How to Estimate Misalignment The estimation framework of this paper is based on Purchasing Power Parity (PPP) Balassa-Samuelson Model Modified version of PPP
Purchasing Power Parity Law of one price The price of a good is equalized in two countries The price of a basket of goods is also equalized Exchange rate equalizes the price of a basket of goods in two countries
Are Prices Really Equalized? Korea Japan China Bicycle (standard type) 100000 won 10000 yen (82000 won) 300 yuan (36000 won) Haircut (male) 8000-10000 won 3500 yen (28700 won) 5-15 yuan (600-1800 won)
The Balassa-Samuelson Model (1) Exchange rate tends to equalize tradables (e.g. bicycle) prices Exchange rate does not tend to equalize non-tradables (e.g. haircut) prices Basket of goods (which include bicycle, haircut, and others) are not equally priced due to the tradables price differences
The Balassa-Samuelson Model (2) What determines the non-tradables price? It tends to be high in high-income countries The price of a basket of goods also tends to be higher in high-income countries PPP rate is the exchange rate that equalizes the price of a basket of goods The exchange rate of high-income countries tend to be priced higher than the PPP rate
The Balassa-Samuelson Model (3) Why non-tradables are highly priced in developed countries? Economic growth is driven mainly by the productivity growth of tradables (e.g. bicycle) Tradables get cheaper by this productivity growth Exchange rate appreciates by this price change
Economic Growth High Non-Tradables Price High Currency Valuation
Basic Assumptions Assumption 1: Economic growth leads to high non-tradables price Assumption 2: (Balassa-Samuelson Model) High non-tradables price leads to higher currency valuation than PPP rate Under these assumptions, Kawai and Motonishi (2005) regress exchange rates (relative to PPP rate) on per capita GNI
2004 Per Capita GNI (PPP dollar) Nominal Exchange Rate/ PPP rate
Kawai-Motonishi (2004) The regression line can be interpreted as the equilibrium exchange rate Vertical distance from the regression line is the misalignment Korea is almost on the regression line
PPP conversion factor (1) Uniqueness of their analysis comes from the utilization of the PPP conversion factor A Numerical Example: 2004 Korea average exchange rate: 1145 2004 Korea PPP conversion factor: 0.693 2004 Korea PPP rate = 1145/0.693= 1652
PPP conversion factor (2) Calculation of PPP rate is very troublesome The PPP conversion factor is calculated by the International Comparison Program (IPC)
Questions to Kawai-Motonishi (2004) Is per capita income the only determinant? What is the source of the misalignment? How did the misalignment change? Is the error term homoskedastic? Is per capita GNI exogenous?
Modifications More determinants Foreign assets and Interest rates Decomposition of the equilibrium exchange rate changes 1990-2004 data More than 1300 samples (from 131 countries) Weighted Least Squares estimate More efficient than OLS under heteroskedasticity Instrumental Variable estimate A resolution to the possible endogeneity in per capita GNI
Existing Literature Many of the existing studies utilize price level changes, not price levels By using price levels, we can detect long run misalignment
Regression Framework Regress (Nominal Exchange Rate)/(PPP Rate) on Per Capita GNI (2004 PPP) Net Foreign Assets Real Interest Rate Yearly Dummy
Per Capita GNI Economic growth leads to high currency valuation (Assumptions 1&2) Denominated in 2004 US dollar Converted to the US dollar by PPP rates Converted to the 2004 price by US inflation rate Logarithmic conversion is performed
Net Foreign Assets (1) A country s accumulation of net foreign assets put upward pressure on the country s currency People dislike having a large amount of foreign currency denominated assets This leads to selling foreign currencies and buying home currency Home currency appreciation
Net Foreign Assets (2) Net foreign asset to GDP ratio is used Accumulative current account of the past 10 years is the proxy of net foreign asset Net foreign asset is obtainable only for a small number of samples
Real Interest Rates High interest rate makes a currency more attractive Lending rate minus inflation rate of the year is used
Yearly Dummy There is co-movement of exchange rates due to the dollar appreciations and depreciations against other currencies The co-movement is captured by the yearly dummy variable Estimated coefficients is close to the changes in the real effective exchange rate of the dollar
Data Database World Development Indicators (World Bank) International Financial Statistics (IMF) From 131 countries in the 1990-2004 period 1336 observations for non-iv estimation 564 observations for IV estimation
Estimation Method Weighted Least Squares (WLS) Instrumental Variable Estimation (IV)
Weighted Least Squares (WLS) Samples are weighted according to per capita GNI More efficient than OLS (Ordinary Least Squares) under heteroskedasticity
Instrumental Variable Estimation (IV) We obtain unbiased estimator even under the existence measurement error in PPP conversion factor Infant mortality is used as the instrumental variable for per capita GNI Criteria of selecting IV High correlation with per capita GNI No correlation with the measurement error Data availability
Regression Results All the explanatory variables are significant at 1% level The effects of foreign assets and interest rates are small but significant R-squared=0.57 The results are reasonably robust to specification and estimator
Decomposition Changes in the fitted rate are decomposed into five parts by using the regression results PPP Per capita GNI Foreign Asset Real Interest Rate Real Dollar Close to the effects of dollar real effective exchange rate
How to Read the Following Graphs The difference between the fitted rate (the pink line) and the actual rate (the blue line) is the misalignment Changes in the fitted rate (the pink line) from 1990 are decomposed into the five lines below
Korea
Korea (non-iv estimate)
Japan
China
Real Effective Exchange Rate of Dollar
Cambod China Philippi Singapo Thailand Misalignment in 2004 Korea Malaysia Indonesi Mongoli Japan
Summary (1) Korean Won is at equilibrium exchange rate in 2004 There may be overvaluation of Korean Won in 2006 Japanese Yen is overvalued by about 20% in 2004 The misalignment may have disappeared in 2006
Summary (2) Chinese Yuen is undervalued by 38% in 2004 The misalignment may not have changed significantly in 2005 and 2006
Summary (3) East Asian Countries with undervalued currencies in 2004 are Cambodia, China Philippines, Singapore, Thailand East Asian Countreis with overvalued currencies in 2004 are Japan, Mongolia
Risk of Yen Denominated Debt for Korean Banks? By rule of thumb, potential yen denominated debt burden is 10-30%
Thank You Very Much! Comments Welcome