1
Foreign exchange rate exposure in Hong
Kong, Japan and Singapore: firm and industry
level analysis
Thesis 30 Cr | | Master Programme | Spring 2011
Master Thesis in Economics Title: Exchange rate exposure in Hong Kong, Japan and Singapore: firm and industry level analysis Author: Tao Xie 870928‐1565 Supervisor: Xiang Lin Date: 2011‐06 Keywords: Foreign exchange rate exposure, stock returns, stock market, exchange rate regime, total exposure Abstract
This paper analyzes the extent of foreign exchange rate exposure in Hong Kong, Japan and Singapore in both firm level and industry level in the period of January 1996 to January 2011 by regressing the stock return of a particular industry or firm on exchange rate changes while controlling for overall stock market movements. It is found that exchange rate movements do affect firm and industry value in a manner consistent with expectation and the extract of unexpected exchange rate changes from actual exchange rate changes have little influence on the testing results of exposure. It is also proved that exchange rate regime plays an irreplaceable role in drawing the structure of exchange rate exposure of a country.
Table of Contents
1. Introduction ... 5
1.1 Purpose and research questions ... 6
1.2 Outline ... 7
2. Background of exchange rate regime in the three regions ... 8
2.1 Hong Kong ... 8
2.2 Japan ... 8
2.3 Singapore ... 8
3. Previous researches ... 10
4. Theory ... 12
4.1 Concept of foreign exchange rate exposure ... 12
4.2 Measure of foreign exchange rate exposure ... 13
4.2.1 Cash flow approach ... 13
4.2.2 Capital market approach ... 14
4.3 Model and related data used in this article ... 15
4.3.1 Bilateral exchange rate and trade-weighted exchange rate ... 16
4.3.2 Actual and unanticipated exchange rate changes ... 19
4.3.3 Firm level and industry level analysis ... 21
5. Data and Method ... 23
5.1 Data ... 23
5.2 Method ... 26
6. Empirical testing analysis ... 27
6.1 Test of stationarity ... 27
6.2.1 Industry level analysis using bilateral exchange rate changes ... 30
6.2.2 Industry level analysis using NEER ... 35
6.3 Firm level analysis ... 40
6.3.1 Hong Kong ... 40
6.3.2 Japan ... 41
6.3.3 Singapore ... 42
6.3.4 Summary of firm level analysis ... 43
7. Deficiency and further suggestion ... 45
8. Conclusion ... 46
1. Introduction
The rapid globalization in recent years boosts the economies in many countries, and also brings some new topics regarding risk management. Measuring and managing foreign exchange rate exposure is a new emerging issue caused by internationalization. Changes in exchange rate through its effect on the costs of inputs, outputs, or substitute goods have impact on the competitive position of domestic firms, individual investors, and also exporters and importers.
Up to now most previous studies of exchange rate exposure have focused on western well‐industrialized countries. This article extends the analysis of exposure to three Asian regions, which are Hong Kong, Japan and Singapore. The growing significance of Asia in the stage of international trade over the past decades has made Asia an attractive candidate for study. The analysis of Hong Kong, Japan and Singapore in this article therefore offers different prospect to observe foreign exchange rate exposure. What is more, these three regions provide variations in exchange rate regimes across economies. As it is a conventional wisdom that exchange rate regimes have great influence on exchange rate exposure, the analysis of these three regions provides an opportunity to explore how different exchange rate arrangements affect the extent of foreign exchange rate exposure. For all of the reasons mentioned above, it is expected that the evidence of exposure among Asian firms will be interesting and may differ from existing literature. To gauge a region’s exchange rate exposure, this paper follows in the tradition of Adler and Dumas (1983) to construct a capital market model which allows us to estimate exposure in a two factor regression framework by using trade weighted exchange rate changes and bilateral exchange rate changes respectively.
specifically, when using trade weighted exchange rate changes, approximately 23 percent and 29 percent of firms in the sample are significantly exposed to actual exchange rate changes and unanticipated exchange rate changes in Hong Kong; 34 percent and 33 percent of firms are reported in Japan; 16 percent and 14 percent of firms are reported in Singapore. When using bilateral exchange rate, 9 percent and 6 percent of firms are significantly exposed to actual exchange rate changes and unanticipated exchange rate changes in Hong Kong; 41 percent and 39 percent are reported in the case of Japan; 33 percent is reported in Singapore for both actual exchange rate and unanticipated exchange rate changes.
The main contribution of this paper is to employ whether the decomposition of unanticipated exchange rate from actual exchange rate influence the testing results and to find whether exchange rate regimes have impact on the level of exposure. The findings are as follows,
1. In most cases exposure to unexpected exchange rate changes and actual exchange rate changes are quite similar, but further studies are still suggested since in the case of Hong Kong, the results of two sectors show some differences between using actual change and unanticipated change.
2. Exchange rate regime plays an irreplaceable role in drawing the structure of exchange rate exposure of a country.
Furthermore, to the best of the author’s knowledge, no study has yet conducted an industry level analysis of the exposure by using unanticipated exchange rate changes. This paper fills the gap in this field.
1.1
Purpose and research questions
rate changes in a significant scale.
Hong Kong, Japan and Singapore are implementing different exchange rate regimes. Thus, another task of this paper is to measure whether exchange rate arrangements have impact on the extent of foreign exchange rate exposure by doing a horizontal analysis among the three regions. The research questions addressed in this paper are as follows, 1. Do the change in exchange rates, namely bilateral exchange rate and trade weighted exchange rate, influence stock returns in Hong Kong, Japan and Singapore? 2. Are the results of empirical testing different between using bilateral exchange rate and trade weighted exchange rate in these three economies? Why?
3. Are the results of empirical testing different between using actual exchange rate changes and unanticipated exchange rate changes in these three economies? Why? 4. Are the results of empirical testing different between firm level analysis and industry
level analysis in these three economies? Why?
5. Do different exchange rate arrangements have impact on the extent of foreign exchange rate exposure of these three economies?
1.2
Outline
2. Background of exchange rate regime in the three regions
This section gives a brief introduction of exchange rate regimes applied in the three regions. In section of empirical testing analysis the information given in this part will be used to explain the differences in the level of exchange rate exposure in Hong Kong, Japan and Singapore.2.1
Hong Kong
Hong Kong's monetary policy objective is to maintain currency stability. Given the highly externally oriented nature of Hong Kong’s economy, regime aiming at stabilize external value of Hong Kong dollar in terms of its exchange rate in the foreign exchange market against the US dollar at around 7.80 Hong Kong dollars to one US dollar was adopted to pursue this objective. (Monetary Policy, 2004) This exchange rate policy is so called linked exchange rate regime.2.2 Japan
Japan maintained a fixed exchange rate of 360 Japanese yens per US dollar until August 1971. After that, Japanese yen was allowed to float above its fluctuation ceiling and the effective rate of yen was also set to be floating freely afterwards.2.3 Singapore
Since 1985, Singapore started to apply a more market‐oriented exchange regime, which allowed Singapore dollar to float within an undisclosed bandwidth of a central parity. (Historial Exchange Rate Regime of Asian Countries, 2000) Since then the Singapore dollar is managed against a basket of currencies of its major trading partners. The weights of one foreign currency in the basket depend on the trading volume between that particular country and Singapore. The composition of the basket is revised periodically to catch the most updated changes in Singapore’s trade patterns by Monetary Authority of Singapore (MAS). (Singapore's exchange rate policy, 2001) According to the classification of IMF, the summary of exchange rate regimes is described in the table below.Table 1 Exchange rate regimes and monetary policy frameworks1
Country Exchange rate arrangement Monetary policy
1 IMF “De Facto Classification of Exchange Rate Regimes and Monetary Policy Frameworks”;
framework Hong
Kong Currency board arrangement
Exchange rate anchor: US dollar
Japan Independently floating Other
Singapore Managed floating with no pre‐determined path for the exchange rate
Exchange rate anchor: Composite Notes: (defined by IMF)
Exchange rate anchor: The monetary authority stands ready to buy or sell foreign exchange at given quoted rates to maintain the exchange rate at its predetermined level or within a range (the exchange rate serves as the nominal anchor or intermediate target of monetary policy). These regimes cover exchange rate regimes with no separate legal tender, currency board arrangements, fixed pegs with or without bands, and crawling pegs with or without bands.
3. Previous researches
This section introduces some researches that contribute considerably to the analysis of exchange rate exposure. The first three articles are viewed as the “benchmark” researches which have great impact on other studies of exchange rate exposure; the authors of these three articles have either constructed or developed the measuring model of exchange rate exposure. After the presentation of basic ideas in these three articles, a brief description of other related researches is given, followed by the introduction of researches which have given a glimpse of the effect of exchange rate arrangements.
Adler and Dumas (1984) firstly analyze exchange rate risks in a way that conforms to the interests of stockholders and analysts. Their definition of exchange rate exposure has received much attention from other scholars and they are also the first one who introduced capital market model into the field of exchange rate exposure analysis. They redefine exposure properly in terms of market rather than book values and measure it as a regression coefficient which provide a single comprehensive measure that summarizes the sensitivity of the whole firm. (Adler & Dumas, 1984)
Jorion (1992) makes a further revision of the model constructed by Adler and Dumas (1984). He sets up a two factor model to gauge the extent of foreign exchange rate exposure in U.S. multinationals and identifies significant cross‐sectional differences in the relationship between the value of multinationals and the exchange rate. It is observed that the level of exposure is positively related with degree of foreign involvement.
Bodnar and Wong (2000) finds that the structure of empirical model has a crucial influence on the results of estimating of exchange rate exposures from stock returns by using the two factor model in a sample of 910 U.S firms over the period of 1977 to 1996. The majority of previous studies have only documented a weak link between exchange rate changes and stock returns. In some articles the reasons of empirical weak link have been analyzed. In the article of Bartov and Bodnar (1994), they attribute the observed insignificant relationship to different sample selection procedure and the mispricing caused by errors in estimating this linkage. Bodnar and Wong (2003) show that both return measurement horizon and model specification have noticeable impacts on estimates of exchange rate exposure.
4. Theory
There is a common belief that foreign exchange rate fluctuations impact firms’ value through various mechanisms, such as international trade with foreign companies. The coming of this belief is quite straightforward. For example, in a simplest case, a depreciation of domestic currency has a probability to increase the profits of a local exporter. In this section, the following questions will be answered,
First, what is the definition of foreign exchange rate exposure?
Second, what is the central model used to analyze the extent of foreign exchange rate exposure in this article?
Third, how is each component in the model defined and measured in this article?
4.1
Concept of foreign exchange rate exposure
Foreign exchange rate exposure is a relatively new‐emerging concept compared with many other economic and financial concepts. It became increasingly attractive in last two decades as international trade played a more important role than ever.
analysis of multinationals only. The reason is straightforward; the profitability of companies which have international imports or exports is inevitably affected by exchange rate fluctuation. However, in this article, the analysis also involves other listed firms.
Why include firms that have no international trade? Exchange rates affect profitability of a firm through many routes. Exchange rate fluctuation is definitely a source of risk for firms with foreign assets and liabilities, and also firms with overseas operations. But at the same time, firms without foreign revenues might also be indirectly affected by exchange rate changes through its impact on foreign competition or broader macroeconomic conditions. (Salsifu, Osei, & Adjasi, 2007) Hence, under an international operating environment companies face direct or indirect exchange rate risk regardless of whether they involve in international trade.
4.2 Measure of foreign exchange rate exposure
There are basically two main approaches to estimate foreign exchange exposure from the prospective of measurement; these two approaches are generally dependent on two different theoretical frameworks, which are cash flow approach and capital market approach. (Dominguez & Tesar, 2006)
4.2.1 Cash flow approach
The cash flow approach focuses on the impact of exchange rate changes on current cash flows. Dumas (1978), Hodder (1982), Hekman (1985) etc. are advocators of cash flow approach. This approach emphasizes to measure the influence of exchange rate changes on a firm from the point of view of the firm’s internal operations. This model uses the present value of cash flow as proxy of firm’s value, in order to gauge exposure. Thus, the simplest form of the model based on cash flow approach can be expressed as below, = + + , t = 1, ... , T (1)
Where CF represents the changes in operating cash flow or operating revenues of firm i in period t; it reflects changes in profitability of a firm, EX is the changes in exchange rate in period t. β and ε are intercept and error term respectively. β measures the level of exchange rate exposure.
By and large, cash flow approach gives researches an opportunity to do in‐depth analysis on how changes in exchange rate influence firms’ various operations.
complicated econometric equipments. However, this approach has some drawbacks. The major one is that it usually suffers from the unavailability of data. To obtain cash flow data, significant amounts of firm‐specific and competitor‐specific information are required, but such information is often available only to insiders. Consequently, this cash flow based method is not easily applicable to multifirm studies or large‐scale cross‐firm comparisons of exchange rate exposures. (Bodnar & Wong, 2003) Thus, for this article, a methodology that uses accessible information is applied.
4.2.2 Capital market approach
Based on the hypothesis that the present value of a firm’s future cash flow is equal to the value of the firm, Adler and Dumas (1984) overcome the limitation of analyzing exchange rate exposure from the view of internal operation. They defined exchange rate exposure as the effect of exchange rate fluctuations on the value of an asset. The exposure is calculated by regressing stock returns with respect to exchange rate changes. Stock return on the left side of the model is the proxy of value of a particular firm. In this way, they measure exchange rate exposure from the point of view of investors and stock analysts. The model can be expressed as,
, = + + , , t = 1, ... , T (2)
Where R, is the stock market return of firm or industry i in period t, EX is the rate
of change of domestic currency’s exchange rate against a foreign currency. ε, is an
independent and identically distributed error. Under the capital market approach, foreign exchange rate exposure of firm i is simply measured by the part of firm i’s stock return variance that is correlated to exchange rate fluctuations.
In Equation (1), δ is called total exposure elasticity of firm i. Generally, this total exposure of a firm comprises two effects. One effect is the average change in the present value of a firm caused by exchange rate change. The other effect is the non exchange rate related phenomena that affect valuations and are spuriously correlated with the exchange rate variable over the sample period, such as some macroeconomic effects which influence the valuation of all firms. (Bodnar & Wong, 2003)
2003) (Jorrison, 1992)
Bodnar and Wong (2003) reformulated the model by including a stock market portfolio in the regression model. The model is as below,
, = , + , + , + , , t = 1, ... , T (3)
Where R, is the stock return of firm i at time t, EX is the percentage change in an
exchange rate variable at time t, and R is the return on domestic market portfolio at time t. The coefficient associated to changes of the exchange rate β,, measures the foreign exchange exposure of firm or industry i, or equivalently, the elasticity of its stock returns to percentage changes in exchange rates. Compared with δ in equation (1) β, is so called residual exposure. Residual exchange rate exposure can be measured by the regression coefficient. (Chan‐Lau, 2005)
4.3 Model and related data used in this article
As mentioned above, cash flow approach requires significant amounts of inside information. So in this article capital market approach serves as the central model and to emphasize the distinctions between residual and total exposure and make the analysis more precisely, this article follows the model of Bodnar and Wong (2000). Thus, the central model of this article is,
, = , + , + , + , , t = 1, …, T2 (3) Most of previous researches on exchange rate risk suggest only modest exposure. The failure of most studies to establish significant exposure of firm value to changes in foreign exchange rate is largely due to the difference in a number of areas that scholars used in their analysis, such as the time horizon, the sample selection procedures or the portfolio construction. (Salsifu, Osei, & Adjasi, 2007) For example, even if the analysis is focusing on the same firm, subperiod analysis is necessary to construct a comprehensive understanding of a firm’s exposure; that is because the exchange rate regimes of a country may change over times. Furthermore, the choices between trade‐weighted exchange rate and bilateral exchange rate, multiple exchange rates and single currency proxy also have influence on the empirical testing results. Therefore, besides the introduction of the main methodology, further description of the model should be given
2 In some previous researches, this model is revised by deducting the risk free rate from actual return. Actually the
before analysis.
In the following subsections, details of the model’s components are introduced. 4.3.1 Bilateral exchange rate and trade-weighted exchange rate
First, let’s focus on how the exchange rate component (EX ) in the model is defined and selected in this article.
Generally speaking, three different types of exchange rate have been defined as EX in previous literature. They are trade‐weighted exchange rate, bilateral exchange rate and multiple exchange rates.
The trade‐weighted exchange rate, also known as the effective exchange rate, is a multilateral exchange rate. It is a weighted average of exchange rates of home and foreign currencies, with the weight for each foreign country equal to its share in trade. (Currency indices, 2011) It measures the average price of a home good relative to the average price of goods of trading partners, using the share of trade with each country as the weight for that country. The trade weighted exchange rate is used to make a complete comparison between one economy's currency and other currencies it interacts with.
Bilateral exchange rate is the exchange rate of home currency against another foreign currency. It compares only two currencies, for example, the Singapore dollar and US dollar.
Each coin has two sides. Although trade‐weighted exchange rate is thought to be a more comprehensive analysis than only comparing two currencies, the model using trade‐weighted exchange rate also suffers from some drawbacks. The domestic currency can appreciate against one foreign currency and at the same time depreciate against another foreign currency, making the weighted index unchanged even when there is huge fluctuation in exchange rate market. Consequently, the use of trade‐weighted exchange rate may underestimate the extent of foreign exchange exposure. (Salsifu, Osei, & Adjasi, 2007)
as the trade‐weighted exchange rate index. It measures the relation of one currency to a group of other currencies that are given weights reflecting the importance of these currencies in international trade.
A small amount of literatures, such as Parsley & Popper (2006), suggests using multiple exchange rates instead bilateral exchange rate in the model, for example, including both Singapore dollar and US dollar exchange rate and Singapore dollar and euro exchange rate in the model. Including more currency theoretically makes the model more persuasive. However, if the model contains several bilateral exchange rates, multicollinearity is expected to be a problem since the change of one bilateral exchange rate is usually highly related with that of another bilateral exchange rate. Standar errors are thought to be large when there is high multicollinearity, making it hard to sort out the individual effects of the explanatory variables and may consequently leaving the individual coefficients statistically insignificant. (Parsley & Popper, 2006)
In Table 2, the correlations of some major currencies (specifically euro, US dollar and sterling) against the home currencies of each region under analysis are reported. Table 2 Montly Exchange Rate Correlations, 1996/12-2011/01
US dollar Euro Pound
Hong Kong US dollar 1 Euro ‐0.188** 1 Pound ‐0.090 0.632*** 1 Japan US dollar 1 Euro 0.615*** 1 Pound 0.699*** 0.768*** 1 Singapore US dollar 1 Euro ‐0.031 1 Pound 0.266*** 0.515*** 1
As shown in Table 2, multicollinearity turns out to be a serious problem in this case. Most of the correlations are relatively high. Although the estimating results in the existence of multicollinearity are not biased, bilateral exchange rate is preferred in the analysis to avoid the potential influence of multicollinearity. As America plays an important role in the international trade market, the exchange rate of domestic currency against US dollar is chosen as the bilateral exchange rate. It should be noted that the exchange rate is expressed as the domestic currency’s price of US dollar, namely Hong Kong dollar/US dollar, Japanese Yen/US dollar and Singapore dollar/US dollar.
As a result, this paper constructs two models. In the first one, a trade‐weighted exchange rate serves as the exchange rate component EX . In the second one, the bilateral exchange rate against US dollar is used as EX . The first model is expressed as, , = , + , + , + , , t = 1, …, T (4)
The second model is expressed as,
, , + , + , + , , t = 1, …, T (5)
The stock returns for firm i or industry i for each period t is obtained in the following formula, , , , , where , = stock returns of company i or industry i for period t, , = stock closing prices of company i or industry i of period t, and , = stock closing prices of company i or industry i of period t‐1.
The rates of exchange rate changes in each period t are obtained in the following formular,
, ,
where
, = percentage change of Nominal effective exchange rate in period t; =
= Nominal effective exchange rate in period t; = exchange rate of domestic currency against US dollar in period t;
= Nominal effective exchange rate in period t‐1; = exchange rate of domestic currency against US dollar in period t‐1.
Model 4 and 5 are run separately for the selected firms and industries in the three countries.
4.3.2 Actual and unanticipated exchange rate changes
As mentioned in the section of purpose, one objectives of this article is to examine the effect of both unanticipated and actual exchange rate changes on common stock returns in the framework introduced in the last section and compare the results. Why is that meaningful to extract unexpected parts from actual exchange rate changes? That is because if the financial markets are assumed to be efficient, the use of unexpected changes in exchange rates are preferred than actual changes since the expected values of the exchange rate should have been reflected in stock prices and only the unexpected changes have effect on stock returns. (El‐Masry, 2006)
The analysis in this paper follows a two‐step procedure to apply an autoregressive moving average model to identify the unanticipated exchange rate from actual exchange rate. The first step consists of finding an ARMA model. Table 3 shows the estimators of ARMA model of both trade‐weighted exchange rate and exchange rate of home currency against US dollar in the three countries. Autocorrelation and partial correlation structures seem to suggest ARMA (1, 1) for exchange rate factors in some cases and ARMA (1, 0) in other cases.5
The ARMA (1, 1) and ARMA (1, 0) models can be expressed as,
(6) (7) where u is zero mean white noice process with variance σ .
Table 3 Estimators of ARMA model6
Coefficient Std. Error t‐Statistic Prob.
Japan NEER c 110.245252 9.711097748 11.35250153 1.11022E‐16 AR(1) 0.967767009 0.025426983 38.06063001 1.11022E‐16 MA(1) 0.291698906 0.079717408 3.659161954 0.000348405 Japan USD c 101.8581154 11.695806 8.708944006 4.66294E‐15 AR(1) 0.968444372 0.0229734 42.15508333 0 Hong Kong NEER c 120.0159609 45.946928 2.612056239 0.010059074 AR(1) 1.005462744 0.0100878 99.67101695 0 Hong Kong USD c 7.78363153 0.006412 1213.968427 0 AR(1) 0.85671585 0.045512 18.823984 0 5 ARMA models are found by using Excel ARMA add‐in functionl.
6 The building of ARMA model follows the four steps Box Jenkins Approach. (model identification, parameter
Singapore NEER c ‐49941.327 0 65535 0 AR(1) 1.000001807 1.26341E‐06 791510.0262 0 MA(1) 0.192633006 0.07397836 2.603910207 0.00999945 Singapore USD c 1.49826767 0.311806 4.805124778 3.273E‐06 AR(1) 0.99217378 0.01364 72.74036984 1.11E‐16 The fitted values of this model correspond to the expected changes. The residuals u in model 6 and 7 are then defined as the unanticipated changes. The second step involves the substitution of these residuals for the exchange rate variables in regression models. (El‐Masry, 2006)
So the percentage change of EX is the changes in actual exchange rate used in the model while percentage change of u is the changes in unanticipated exchange rate used in the model.
In the empirical testing section, the exposure will be analyzed by using actual exchange rate and unanticipated exchange rate respectively. And the residual obtained from ARMA model is the proxy of unanticipated exchange rate.
4.3.3 Firm level and industry level analysis
The majority of previous researches have focused on firm level analysis by selecting a specific amount of firms in domestic stock market. Compared with firm level analysis, industry level analysis could give an overview of the exposure condition. It provides the information of widespread exchange rate risk each industry is exposed to and helps to identify exports denominated industry and import denominated industry. Given the extent of industry exposure, investors may predict further development of a particular industry and expect how well the derivative market works.
5. Data and Method
This section begins with the introduction of data selection, following by the illustration of industries’ classification and the standards of firm selection. After that, the main method will be described briefly.
In this section, the following questions will be answered,
First, what is the market index for each country used in the model? Second, how are the firms selected?
Third, what is the suitable method applied to measure the extent of exposure for firm‐level and industry‐level analysis respectively?
5.1
Data
Table 4 shows some basic information of data used in this paper. All the data is selected from Ecowin Database.7 Monthly data of Hang Seng index, Nikkei 225 and Straits times are chosen as the market index for Hong Kong, Japan and Singapore respectively. All of them are composite index which consists of many stocks averaged to form a stock price representative of an overall market. (Watada & Wen, 2010) Hence, composite index is a useful statistical measure of the overall market performance over time. It measures and trackes changes in price levels for an entire stock market or sector. (Watada & Wen, 2010)Table 4 Definition of market index and number of firms
Country Number of companies in firm level analysis Market Index Description of the market index Hong Kong 35 Hang Seng
Japan 59 Nikkei 225
The Nikkei 225 is a stock market index for the Tokyo Stock Exchange (TSE). Currently, the Nikkei is the most widely quoted average
of Japanese equities.
Singapore 43 Singapore
Straits
The FTSE Straits Times Index is a stock market index that is regarded as the benchmark index for Singapore stock market. It tracks the performance of the top 30 companies listed on the Singapore Exchange. In the firm level analysis, all the firms selected are the listed companies in the domestic stock exchanges of the three countries. All of them are domestic large companies that constitute major composite indices. Compared with some western countries, the classification of sectors in Asian economies does not follow a unified standard. This is largely due to the size of the country and the limitation of geographic factors. For instance, some of Asia‐Pacific countries are relatively small‐sized; their economies are denominated by several major sectors (industries). The amount of firms in some sectors may be quite small, making the construction of industry index valueless. So when analyzing these economies, lack of some sectors’ data diminishes the number of sectors one can compare among countries. Table 5 Available sectors or industries for each country
Energy; Financials; Health Care; Industrials; Information Technologies; Materials; Telecommunication; Utilities Singapore Financials; Manufacturing; Electronics Table 5 shows the sector data available in these three countries. In the case of Japan, the classification of sectors follows the Global Industry Classification Standard. GICS is an industry taxonomy developed by Morgan Stanley Capital International (MSCI) and Standard & Poor's (S&P) for use by the global financial community. The GICS structure consists of 10 sectors, 24 industry groups, 68 industries and 154 sub‐industries. Appendix B gives some further description of this classification. In the case of Hong Kong, the Hang Seng Industry Classification System is used to reflect the stock performance among different market sectors. The classification is based upon the sales revenues from each business area and the sources of the business’s revenue. The following aspects are the difference between Hang Seng Industry Classification and Japan’s GICS classification: 1. In the case of Hong Kong, a consumer goods sector is set up by combining consumer discretionary and consumer staples sectors together. 2. The data of health care sector is not available in Hong Kong.
3. Conglomerates, property and construction and services are three distinctive sector constructed by Hong Kong stock exchange.
In the case of Singapore, data is available in only three sectors, which are financials, manufacturing and electronics.
5.2 Method
6. Empirical testing analysis
This section shows the empirical testing results, gives the analysis of these results and makes comparison among the three regions.6.1
Test of stationarity
It is a common belief that the raw data of exchange rate and stock price is thought to be nonstationary. But as in this paper percentage change is used in the estimating model, intuitively speaking, the model should not suffer from nonstationarity.8 Figure 1 to 3 below depicts the percentage change of NEER and bilateral exchange rate against US dollar and market return during the sample period in Hong Kong, Japan and Singapore. It is observable that all curves are fluctuating around zero and there is no obvious pattern. By and large, actual exchange rate change and unanticipated exchange rate change are moving together in all the three countries.Figure 1 Exchange rate fluctuations in Hong Kong
8 Besides, to test the stationarity of stock returns, all the industry level stock return and several representative firm
Figure 2 Exchange rate fluctuations in Japan
Figure 3 Exchange rate fluctuations in Singapore
Figure 4 fluctuations of stock return in Hong Kong, Japan and Singapore
augmented Dickey‐Fuller test (ADF) is a test for a unit root in a time series sample. It is an augmented version of the Dickey–Fuller test for a larger and more complicated set of time series models. The augmented Dickey–Fuller (ADF) statistic, used in the test, is a negative number. The more negative it is, the stronger the rejection of the hypothesis that there is a unit root at a certain level of confidence. (Greene, 1997)
The ADF test in this article consists of estimating the following regression:
∆ ∆
where ε is a pure white noise error term, t is the time trend term, is the drift term, ∆ is the difference between exchange rate in period t and period t‐1. By including ∑ ∆ , the ADF formulation allows for higher‐order autoregressive processes. The number of lags is determined by examining information criteria such as Akaike Information Criterion (AIC) and Schwarz Information Criterion (SIC) which are measures of the relative goodness of fit of a statistical model.
The null hypothesis of this test is
H0: δ 0, which indicates there is a unit root
Thus, the alternative hypothesis is H1 0, which indicates the series is stationary.
The results of ADF test is shown in the table below. The coefficients for market return and both actual and unanticipated exchange rate of US dollar and NEER are all statistically significant. Then the null hypothesis is rejected; exchange rate and market return series of all the three countries are stationary.
Table 6 ADF Test Results
Country
ADF Test Results
USD NEER
Rm Actual Unanticipated Actual Unanticipated
Hong Kong ‐10.974404*** ‐10.290929*** ‐9.961376*** ‐10.328675*** ‐9.3019*** Japan ‐13.422964*** ‐12.001840*** ‐8.977054*** ‐11.960945*** ‐11.1780*** Singapore ‐12.449802*** ‐12.378482*** ‐11.293741*** ‐13.636412*** ‐11.8076***
1. Using critical values by Mackinnon, 1996
2. Optimal lag length is selected based on minimum Schwarz Information Criterion (SIC)9
3. After observing figure 1 to 3, it is believed the model does not need to include trend and intercept.10
6.2 Industry level analysis
Using industry portfolio returns for Hong Kong, Japan and Singapore, this subsection examines the relation between exchange rate changes and changes in industry values and explores the potential reasons that determine the relations.
Before analyzing the results, some further description of data is necessary. As mentioned above, the trade weighted index is an economic instrument used by economists to compare their exchange rate against those of their major trading partners. Trading partners that constitute a larger portion of an economy's exports and imports receive a higher index. The trade weighted exchange rate index in this article is nominal effective exchange rate supplied by IMF.
One thing needed to be noted is the interpretation of trade‐weighted exchange rate and bilateral exchange rate against US dollar. The interpretation of the effective exchange rate is that if the index increases, the purchasing power of that currency is higher, or the currency strengthened against those of the region's trading partners. A lower index means that the currency depreciated so that one needs more of that currency to pay for imports. But in the case biletaral exchange rate, this paper selects domestic currency’s price of dollar, namely the amount of domestic currency required to buy one unit of dollar. Thus, an increase in the bilateral exchange rate means domestic currency depreciates against dollar while a decrease represents an appreciation of domestic currency.
6.2.1 Industry level analysis using bilateral exchange rate changes
Table 7 Industry level analysis results using exchange rate against US dollar
The equation estimated is ,= , + , + , + ,, t = 1, …, T
Country Industry β R
Actual Unanticipated Actual Unanticipated
*. Statistically significant at 0.1 level (2‐tailed) **. Statistically significant at 0.05 level (2‐tailed) ***. Statistically significant at 0.01 level (2‐tailed)
6.2.1.1 The case of Hong Kong using bilateral exchange rate
Theoretically speaking, the exposure of Hong Kong dollar to US dollar should be insignificant as Hong Kong dollar is linked to the value of US dollar during the whole period of our analysis. However, some empirical results are inconsistent with the intuition.
Three of the eleven sectors are statistically significant exposed to changes in exchange rate against US dollar, which are conglomerates, services and utilities sectors. Among these three sectors, the definition of conglomerates may be not that clear as the other two sectors. Thus, in the next paragraph, it is defined first. After that, some representative companies are listed to make a further description of each sector.
Conglomerates are sometimes referred to as multi‐industry companies which are involved in a variety of business functions simultaneously, usually involving a parent company and several subsidiaries. (Conglomerates sector, 2011) Conglomerates are often large and multinational. Hutchison Whampoa Ltd and CITIC Pacific Ltd, for example, are companies which play roles in constituting the conglomerate index in Hong Kong. In the case of utilities sector, the involvement of China Power International Development Ltd and Datong Intl Power can give us a sense of the classification’ standards of this sector. For services sector, Beijing Capital International Airport can be taken as the representative company.
continuously stronger than it was expected to be. Figure 6 shows the percentage change of Hong Kong dollar against US dollar. It is observed that the rate of change is also not that stable as it is expected under a linked exchange rate regime.
Figure 5 Fluctuation of exchange rate of Hong Kong dollar against US dollar
Figure 6 Fluctuation of percentage change of exchange rate against USD
Such deviations may have been tolerable or even welcome in some cases, but actually
Hong Kong dollar exchange rate does not deviate from the official 7.8 rate to the U.S. dollar. For example, any slight deviation from 7.8 will give private banks and investors incentive to arbitrage; they may be hurry to buy Hong Kong dollars from the market at a price higher than 7.8, and then sell it at 7.8 and thus earn a profit. This arbitrage process decreases the supply of the Hong Kong dollar and brings back the exchange rate in line with the peg. (Should the Hong Kong dollar be delinked) But in the case of Hong Kong, exchange rate deviated from its standard level for a long time. This deviation may be a possible explanation of the significance results of foreign exchange rate exposure in some sectors, such as conglomerate sector.
Companies constituting conglomerate sector are usually big sized companies which is often a combination of two or more corporations engaged in entirely different businesses together into one corporate structure. Thus, these companies’ ability to grab the arbitrage opportunities is thought to be higher than other small sized companies because arbitrage is a cash‐based operation. (Should the Hong Kong dollar be delinked) If the currency board system is not working perfectly to control the linked exchange rate within a particular band, conglomerates’ speculators could and would like to attack the Hong Kong dollar. Furthermore, conglomerates are usually multinationals; their use of US dollar as the trading currency may be more frequent than other companies. Their reliance on US dollar is consequently much more than others. Hence, when currency board is working inefficient, conglomerates’ exposure to fluctuations of exchange rate against US dollar may be significant.
Service sector is significantly exposed to actual exchange rate changes but insignificant to unanticipated exchange rate changes while utilities sector is significant exposed to unanticipated exchange rate change but significant to actual exchange rate changes. These results suggest that the decomposition of unanticipated change from actual change may have some impact on the empirical testing results in the case of Hong Kong when using bilateral exchange rate against US dollar.
6.2.1.2 The case of Japan using bilateral exchange rate
Only consumer discretionary sector and financials sector are reported significantly exposed to changes in bilateral exchange rate against US dollar. Consumer discretionary sector in Japan is an export oriented sector which plays a crucial role in total stock value.11 (Vanguard Japan stock index fund an index‐related fund, 2010) This is consistent with empirical testing result, which shows the value of stock in consumer discretionary sector is positively related to the depreciation of exchange rate against US dollar, indicating that Japan is the exporter of consumer discretionary goods.
The sign of financials sector’s coefficient is negative. This shows that value of stocks in the financials sector is negatively related to the depreciation of yen against US dollar. Among all the results, the insignificant result of energy sector is beyond one’s expectation. Japan has few domestic energy resources; only 16 percent energy is self‐sufficient. It is the third largest oil consumer in the world behind the United States and China and the third largest net importer of crude oil. It is also the world's largest importer of both liquefied natural gas (LNG) and coal. (Country analysis briefs Japan, 2011) Due to Japan’s geographic factors and its high reliance on the imports of energy resources, it is beyong the expectation that energy sector is insignificantly exposed to foreign exchange rate changes.
Besides, based on the empirical testing results, there is no obvious difference between using actual exchange rate change and unanticipated exchange rate in the case of Japan. 6.2.1.3 The case of Singapore using bilateral exchange rate
According to the empirical testing results, the coefficients of all the three sectors available in Singapore are statistically significant exposed to changes in exchange rate against US dollar. And all the three coefficients are positive, indicating that the value of firms in these three sectors benefits from the depreciation of Singapore dollar against US dollar.
6.2.2 Industry level analysis using NEER
Table below shows the results of industry level analysis when using trade‐weighted exchange rate changes as the proxy of exchange rate component.
Table 8 Industry level empirical testing results, using NEER as the exchange rate
The equation estimated is ,= , + , + , + ,, t = 1, …, T
Country Industry β R
Actual Unanticipated Actual Unanticipated
***. Statistically significant at 0.01 level (2‐tailed)
6.2.2.1 The case of Hong Kong using traded weighted exchange rate
The significant sectors change drastically when using trade weighted exchange rate instead of bilateral exchange rate against US dollar in the case of Hong Kong. When using NEER, conglomerates and utilities sectors’ coefficients become statistically insignificant; property and construction sector turns to be significantly exposed to exchange rate shocks; services sector is reported much more sensitive to exchange rate changes as its significance level increases from 10 percent to 1 percent.
The big difference between using bilateral exchange rate and trade weighted exchange rate is largely due to the linked exchange rate regime. Although the objective of linked exchange rate regime is to stabilize exchange rate, it only guarantees the stability of exchange rate against US dollar, but not other currencies. If the US dollar appreciates or depreciates against other currencies, the exchange rate of Hong Kong dollar against other currencies will also fluctuate significantly. Thus, when using trade‐weighted exchange rate, testing results are expected to be different from those using the bilateral exchange rate against US dollar, which is indeed proved by empirical testing results. Only four sectors are significantly sensitive to foreign exchange rate changes in the case of Hong Kong. As Hong Kong is an international trading centre and the second regression is not using the exchange rate against its linked currency, the number of significant coefficient is expected to be bigger than the empirical results. For example, the coefficient of financials sector is usually statistically significant in previous researches on other countries.
There is a potential reason of this unexpected result. Although the companies’ stocks under analysis are all traded on Hong Kong stock exchange, most of their businesses are focusing on Mainland China. Appendix C shows the servicing areas (sometimes also the location of headquarter) of 45 companies which constitute the Hang Seng composite index. All of these companies also attribute to make up the sector index. As shown in Appendix C, almost 69 percent of these 45 companies have services in Mainland China. Although the selected 45 companies are only a small proportion of total firms published on Hong Kong stock exchange (Appendix D shows the number of listed companies by industry classification in Hong Kong), one can still anticipate that the monetary policy, trading characteristics and type of enterprise in Mainland China have great influence on the testing results.
important role in Hong Kong stock market. Due to the strong economic growth in the Mainland and robust fund‐raising activities on the Exchange, Hong Kong expect more Mainland enterprises including SMEs (small and medium sized enterprises) to continue to view Hong Kong as the ideal platform for securing investment funds or extending their international reach.
One major characteristic of Mainland companies on Hong Kong stock exchange is that the majority of them are state‐owned enterprises. For example, in the financials sector, all the Mainland banks published on Hong Kong stock exchange are state‐owned banks. Their trades with other countries are at least partially backed by the government. And the services they offer are highly focusing on China. Thus, the fluctuations in foreign exchange rate have less impact on them than banks in other countries. This could also at least partially explain the insignificant results of some sectors.
When using NEER, three of the four significant coefficients are negative. As mentioned before, NEER represents the relative value of a home country's currency compared to the other major currencies being traded. A lower index means that the currency devaluates so that domestic investors need more money to pay for imports. Thus, the negative coefficient means that an appreciation of domestic currency against its trading partners’ currencies has a negative impact on the sectors’ stock returns while a depreciation of domestic currency is positive related to sectors’ stock returns. This is consistent with the results in using bilateral exchange rate against US dollar.
Furthermore, the results of significance are the same between using actual exchange rate change and unanticipated exchange rate change, indicating that unanticipated trade weighted exchange rate change is actually moving together with actual trade weighted exchange rate change.
Four sectors’ coefficients are statistically significant, which are materials, property and construction, services and telecommunication. It is not surprised that the coefficient of telecommunication is statistically significant. In the last two decades, the objective of Information Technology and Broadcasting Bureau in Hong Kong is to enable Hong Kong to be recognised as a world class telecommunications centre for doing business. (Telecommunication, 2011) This objective has been achieved successfully over these years. But it should be noted as the coefficient is statistically significant positive; telecommunication is thought to be an import oriented industry.
the authority. And the import of construction services does not appear to be very active and has consistently stayed at less than 4% of the total construction volume. This is consistent with the testing result. The coefficient of property and construction sector is significantly negative, which indicates that this industry is export‐oriented and is significantly exposed to foreign exchange rate risk. (Business review 2010 Annual report, 2010)
Besides, services sector in Hong Kong is a well known export oriented industry and its exports has been growing rapidly over these years. For example, the exports of services grew by 10.8% in real terms in 2008, which is really a high growth rate. (Anson, et al., 2008)
6.2.2.2 The case of Japan using traded weighted exchange rate
Compared with the testing results by using bilateral exchange rate, coefficient of consumer discretionary is still significantly when using NEER. The negative coefficient indicates that Japan’s firms in consumer discretionary sector benefits from depreciation of yen against its trading partners’ currencies. When using NEER, coefficient of financials sector is no longer significant while coefficient of telecommunication tends to be significant.
6.2.2.3 The case of Singapore using trade weighted exchange rate
When using traded weighted exchange rate, all the coefficients become insignificant. This is not surprised since Singapore manages its currency against a basket of its trading partners’ currencies, meaning it virually shield domestic firms from shocks in trade weighted exchange rate.
6.2.2.4 Summary of industry level analysis
After analyzing the industry level empirical testing results, several conclusions can be drawn.
First, the decomposition of unanticipated exchange rate changes from actual exchange rate changes has little influence on the empirical results in Japan and Singapore. But in the case of Hong Kong, two sectors are affected by the decomposition. Therefore, the extract of unanticipated exchange rate changes is still suggested in the further studies of foreign exchange rate exposure.
to domestic market, and consequently determines the extent of exposure measured by that currency.
Third, exchange rate regimes do have influence on the extent of exposure. Empirical testing provides clear evidence which shows the level of exposure is considerably affected by linked exchange rate and managed exchange rate regimes.
6.3 Firm level analysis
This subsection analyzes the firm level testing results of the three countries. 6.3.1 Hong Kong
Table 9 Exchange risk exposures of individual firms in Hong Kong, NEER Table 9 Exchange Risk Exposure Hong Kong, NEER
Total Firms Significant Exposure Positive Exposure
Negative Exposure
(% Total Firms) (% Significant) (% Significant)
1.Actual Exchange Rate Changes
35 8 (22.86%) 2 (25%) 6 (75%)
2. Unanticipated Exchange Rate Changes
35 10 (28.57%) 2 (20%) 8 (80%)
Table 10 Exchange risk exposures of individual firms in Hong Kong, USD Table 10 Exchange Risk Exposure Hong Kong, USD
Total Firms Significant Exposure Positive Exposure
Negative Exposure
(% Total Firms) (% Significant) (% Significant)
1.Actual Exchange Rate Changes
35 3 (8.57%) 2 (66.67%) 1 (33.33%)
2. Unanticipated Exchange Rate Changes
35 2 (5.71%) 1 (50%) 1 (50%)
Table 9 and 10 show firm level exchange rate exposure condition by using bilateral exchange rate and trade‐weighted exchange rate in Hong Kong.
to US dollar its exposure to changes in exchange rate against US dollar is under controlled. Firms are considered to be less exposed to US dollar fluctuations. But the linked exchange rate regime also makes Hong Kong dollar expose to other currencies, so the number of firms sensitive to changes in NEER is larger.
The percentages of positive and negative results are also consistent with industry level analysis. When using NEER, the number of positive coefficient is much more than the number of negative coefficient. This shows the fact that the value of most firms in Hong Kong benefits from the depreciation of Hong Kong dollar. Thus, Hong Kong may be an export denominated region.
6.3.2 Japan
Table 11 Exchange risk exposures of individual firms in Japan, NEER
Table 11 Exchange Risk Exposure Japan, NEER
Total Firms Significant Exposure Positive Exposure
Negative Exposure
(% Total Firms) (% Significant) (% Significant) 1.Actual Exchange Rate Changes
59 20 (33.9%) 4 (20%) 16 (80%)
2. Unanticipated Exchange Rate Changes
59 19 (32.2%) 4 (21.05%) 15 (78.95%)
Table 12 Exchange risk exposure of individual firms in Japan, USD
Table 12 Exchange Risk Exposure Japan, USD
Total Firms Significant Exposure Positive Exposure
Negative Exposure
(% Total Firms) (% Significant) (% Significant)
1.Actual Exchange Rate Changes
59 24 (40.68%) 16 (66.67%) 8 (33.33%)
2. Unanticipated Exchange Rate Changes
59 23 (38.98%) 16 (69.57%) 7 (30.43%)
changes when using bilateral exchange rate against US dollar.
After comparing the percent of positive exposure and negative exposure, it is observable that a larger proportion of firms in the sample are positively affected by the depreciation of yen against other currencies. These companies experience an adverse valuation effect when the domestic currency appreciates and benefit when the domestic currency depreciates. They perhaps import inputs of production, or use internationally priced raw materials, more than they export. Since the information that helps us to distinguish net exporters from net importers is not available, it is unable to explore the reasons for the observed phenomenon. But one can still conclude that Japanese firms’ exposure to exchange rate changes is predominately positive when using bilateral exchange rate and negative when using NEER, indicating that firms benefit when the yen depreciates.
6.3.3 Singapore
Table 13 Exchange risk exposures of individual firms in Singapore, NEER Table 13 Exchange Risk Exposure Singapore, NEER
Total Firms Significant Exposure Positive Exposure
Negative Exposure
(% Total Firms) (% Significant) (% Significant) 1.Actual Exchange Rate Changes
43 7 (16.28%) 3 (42.86%) 4 (57.14%)
2. Unanticipated Exchange Rate Changes
43 6 (13.95%) 3 (50%) 3 (50%)
Table 14 Exchange risk exposures of individual firms in Singapore, USD Table 14 Exchange Risk Exposure Singapore, USD
Total Firms Significant Exposure Positive Exposure
Negative Exposure
(% Total Firms) (% Significant) (% Significant)
1.Actual Exchange Rate Changes
43 14 (32.56%) 7 (50%) 7 (50%)
2. Unanticipated Exchange Rate Changes
43 14 (32.56%) 7 (50%) 7 (50%)