Forex dealers' perspectives on exchange rate determination in Korea

This paper reports on a survey of dealers in the Korean foreign exchange market. The trade balance and foreign portfolio investment are reported to be the two most important variables in determining the exchange rate.

Although dealers believe that fundamental factors are important, their decisions, particularly in the short run, are also affected by 'noise' factors such as herding behaviour, the use of charts and trends, the advice of opinion leaders, and positions that have already been taken. Speculation is perceived positively as enhancing market efficiency and liquidity. Arbitrage is limited by risk aversion and the unpredictability of future exchange rate changes.

Keywords: Foreign exchange, Korea

JEL Code: F31

1 Introduction

The foreign exchange market in Korea is divided into a customer market and an interbank market. An important difference between the Korean foreign exchange market and most others is the thinness of the Korean market. As of April 2001, the average daily market turnover was about $US 10 billion, which was about one fifth that in the Australian market, one tenth that in the Singapore market and one fiftieth that in the UK market (Bank for International Settlements, 2002). The proportion of daily turnover in the Korean market to the sum of annual exports and imports was only 3.4 per cent, compared with 13.3 per cent (USA), 85.7 per cent (UK), 40.9 per cent (Australia), 42.5 per cent (Singapore), 19.5 per cent (Japan) and 8.3 per cent (Germany) (Bank for International Settlements, 2002). There are several reasons for the low level of market liquidity in Korea. First, the money markets that are connected with the foreign exchange market are not well-developed. The money market instruments and the reference swap rates corresponding to the various maturities do not always exist. Hence the opportunities for hedging and speculation are limited and market participants are often reluctant to pay the cost of hedging. Second, the development of the foreign exchange market was formerly restricted by various regulations that were only removed by the second stage of foreign exchange liberalization in January 2001 (Bank of Korea, 2002). Furthermore, non-bank financial institutions were not allowed to conduct foreign exchange business until 2002.

The majority of foreign exchange transactions are made in the interbank market, in which dealers generally transact with each other indirectly, through two foreign exchange brokerage firms. The fact that banks trade mainly through brokers reflects the thinness of the market: the use of brokers reduces search costs and allows banks to avoid the risks involved in actively quoting buy and sell rates. Out of more than 70 institutions in the Korean forex market, the largest 17 banks account for 75 per cent of total turnover. Since the market is thin in terms of trading volume, there is always the possibility that a large discrepancy between foreign exchange supply and demand could disrupt its regular operation. The Central Bank is normally ready to intervene in such circumstances.

According to the efficient market hypothesis, asset prices in financial markets are consistent with economic fundamentals in the sense that they fully reflect all available relevant information (Fama, 1991). Belief in the applicability of the efficient markets hypothesis and of conventional monetary and portfolio models to foreign exchange markets was severely dented by Meese and Rogoff (1983), who demonstrated that random walk models of the dollar prices of the pound, the mark and the yen generally provided better short-run predictions (that is, up to 12 months ahead) out of sample, than models based on uncovered interest parity, purchasing power parity or simple monetary and portfolio theories. This finding was all the more remarkable because the 'predictions' of the monetary and portfolio models used by Meese and Rogoff were based on the realized values of the explanatory variables and therefore used more up-to-date information than that available to the random-walk models. Many subsequent studies have produced similar findings (Frankel and Rose, 1995; Taylor, 1995). A second challenge to fundamentalists has been provided by evidence that foreign exchange markets are more unstable than they would be if only fundamentals were relevant (Eichengreen and Wyplosz, 1993). However, in contrast to their poor short-run performance, exchange-rate models that use the variables found in standard theoretical work can outperform random-walk models when the time horizon is extended to 3 to 4 years (Mark 1995).

An alternative to the view that the exchange rate is determined purely by fundamentals is provided by Shleifer and Summers (1990), who suggest that introducing some degree of irrationality gives more plausible explanations of financial markets. They argue that the excessive volatility of stock market prices documented by Shiller (1981) is consistent with the assumption that some traders try to forecast price movements using both fundamental and non-fundamental factors, or 'noise', and that arbitrage by those relying purely on fundamental factors is limited by risk aversion. By noise, we mean the use of technical analysis (for example, extrapolating price trends) and psychological factors (for example, indexes of investor sentiment, and the attribution of importance to prices crossing 'barriers', which may be historic highs or lows, or round numbers). With limited arbitrage, it becomes rational for all traders to monitor such variables if their use by some traders is widespread and correlated (De Long et al., 1990).

Empirical evidence for the self-fulfilling nature of technical trading in various foreign exchange markets is given by Frankel and Froot (1990) and by Taylor and Allen (1992). 'Herding', defined as doing something because others are doing it, can make the use of technical analysis self-fulfilling. Ahn et al. (2002) argue that excess demand or supply that is not cleared within the market can easily cause herding behaviour in a thin market like the one in Korea. Another explanation for herding behavior is following the advice of opinion leaders, that is persons or organizations that are believed to have better information about future price developments than other market participants. It is well known that in the US stock market many investors follow the advice of opinion leaders such as Merill Lynch, or of individuals like Joe Granville (Shleifer and Summers, 1990, p. 24).

There have been several surveys of forex markets to examine their microstructure, or dealers' views on how they behave. However, these surveys are mainly for highly developed markets such as UK (Taylor and Allen, 1992), USA (Cheung and Chinn, 1999), Germany (Menkhoff, 1998) and Australia (Hutcheson, 2001). Our survey was designed to provide a better understanding of the trading strategies of dealers, and the information that they rely on, in the relatively underdeveloped Korean forex market. (1) In particular, it examines:

* the relative importance of fundamentals, defined as a list of variables (see Section 3 and Table 3) found in mainstream macroeconomic models of exchange rate determination, and of apparently non-fundamental factors such as technical analysis, market sentiment and dealers' established positions;

* the limits to arbitrage and the unpredictability of the exchange rate when expectations are not fully rational; and

* the role of speculation and central bank intervention.

2 Overview of the Sample of Dealers Surveyed

A survey containing 24 questions was mailed to all active forex dealers in the Korea Forex Club to collect their views. The questions asked were similar to those asked by Shiller (1989), Taylor and Allen (1992), Cheung and Chinn (1999) and Menkhoff (1997, 1998). As a first step, we checked the relevance of the questionnaire to the Korean forex market after consultation with experienced foreign exchange dealers at the Bank of Korea and at the Korea Exchange Bank.