Description of Model
To satisfy the purpose of this thesis a quantitative analysis is carried out and is a deductive operation due to the amount of existing literature on the subject. The exposure coefficient is estimated by using a capital market approach as a framework (Dominguez & Tesar, 2005; Wong, Wong & Leung, 2008 and Chi, Tripe & Young, 2010) to isolate the relationship be-tween firm value (defined as stock returns) and exchange rate changes. The capital market approach not only allows for detection of the relationship between excess returns and the exchange rates but also shows the degree of the estimated exposure. The estimated expo-sure is represented by β2, and its value indicates the degree of exposure. The exposure es-timate is obtained by running regressions based on Equation (1) with stock returns as the dependent variable and return on market index and the individual exchange rates as inde-pendent variables:
As argued by Choi and Prasad on the use of a two-factor model ”…[Equation 1] is not a model of asset pricing but a factor model that allows measurement of factor sensitivities” (Choi & Prasad, 1995, pp. 78).
With the inclusion of a market variable the estimated betas will show the excess exposure faced by the banks. Thus a 2,i equal to zero does not mean the firm faces no exposure, but that the firm faces the same exposure as the market portfolio. The returns on the bank stock, the market and the exchange rates will be calculated as compounded by using the formula:
Seven regressions were performed on each bank, based on equation 1, and every regression using the individual currency pair in the form of domestic currency in terms of foreign cur-rency. Previous studies have been using a trade weighted currency index instead of using the individual exchange rates (e.g. see Doidge, Griffin and Williamson, 2002). To measure if the banks show different results in estimated exposure when using a trade weighted cur-rency index, I include a regression on Equation (1) with the TCW index return as the inde-pendent variable instead of the individual exchange rate return. TCW (Total Competitive-ness Weights) is a trade-weighted index measuring the value of the Swedish krona (SEK) to a basket of other currencies, where the IMF calculates the weights. TCW is an index where the SEK is compared to other currencies as a bilateral exchange rate, and the weight of each bilateral exchange rate is based on the amount of trade between Sweden and 21 other countries (a list of all currencies and respectively weight is provided in Appendix 1). Including the TCW-index contributes with one essential function, namely it checks the robustness of the estimated exposure. If the bank fails to show any significant exposure to any of the individual exchange rate, it does not imply that the bank has no exposure but that the cho-sen exchange rates are not significant to the bank, thus including the index checks if the bank shows exposure to a basket of other currencies and hence checks the robustness of the exposure.
The thesis further includes an exchange rate exposure estimation of a benchmark, which al-lows for a comparison on the exposure of banks’ and the benchmark. The benchmark con-sists of the four largest (based on market capitalization) non-financial corporations in Swe-den and is Atlas Copco, Ericsson, H&M and Telia Sonera. The benchmark corporations are also traded on the Nasdaq OMX Stockholm. This comparison is of interest since it provides a clearer picture on the amount of excess exchange rate exposure of the banks. The benchmark does not possess the same internal expertise on foreign exchange and the banks can be expected to have less exposure.
After estimations on the exposure of the major banks and the benchmark, the thesis con-tinues by applying determinants of exchange rate exposure to see if it can shed light on why the exposure diversify among the four major Swedish banks. The investigated determinants are proven to show a significant relation to the exposure coefficient in previous studies (de-terminants are identified in the Review of literature section).
The thesis hypothesizes that there is a correlation between the banks’ firm values and fluc-tuations in the investigated exchange rates. This is assumed since the banks are highly in-teracting on the foreign exchange market both for their own account and for client purpos-es. Because movements in exchange rates are unpredictable in the long run it can be hard to maintain a perfect hedge throughout. In addition, it has been proven that the size and the openness of the economy plays a role in the detection of exposure, and since these fac-tors can both be applied to Sweden, banks with Sweden as their home market are expected to show some degree of exposure. Moreover, the banks are frequently indirect exposed to exchange rate changes, with the exposure arising from providing services to global clients. The indirect exposure is problematic to control and hedge against, therefore it can also be expected that the four banks are more sensitive to exchange rate changes than the bench-mark.
The alternative to the statement above, is that the banks are unaffected by the fluctuations in the exchange rates or that the investors do not care about the exposure, in other words no significant relationship between the banks’ firm value and exchange rate changes is de-tected. An explanation to this could be that the banks are efficient in hedging hence no sig-nificant exchange rate exposure is to be detected among the banks. Furthermore, because of the banks’ internal expertise on the foreign exchange market, the banks can be expected to have properly hedged to those currencies in which the bank may experience losses when unfavourable market events arise. In theory it is possible to entirely hedge its foreign exchange exposure, which would show an insignificant correlation between firm value and exchange rates (Nydahl, 1998).
The stated null hypothesis follows:
H0: There is no relationship between firm value and exchange rate changes
H1: There is a relationship between firm value and exchange rate changes.
If we reject the null hypothesis above and conclude there is a significant relationship be-tween firm value and exchange rate movements, the thesis expects that the estimated expo-sure coefficient have a non-zero value. A correlation equal to zero would imply that the bank have no excess exposure than the overall exchange rate exposure faced by the market, but due to the nature of the financial markets the banks are expected to show higher expo-sure than the overall market.
The sample consists of the four major Swedish banks Nordea, SEB, Handelsbanken and Swedbank. Weekly data from 20th July 2004 to 13th March 2012 is used, where stock prices are collected on Tuesdays to avoid any occurrence of Monday-effects1. Weekly data is used because the issue of “the correct” time horizon has not yet been solved, thus weekly data is used over of daily data to have as many observations as possible to give power to the test (Rees & Unni, 1999). The estimated exposure is represented by the beta coefficient and ex-plains the change in firm value caused by changes in the exchange rate. The value of the beta coefficient represents the degree of exposure e.g. a positive coefficient indicates that a one per cent increase (decrease) in exchange rate is expected to cause a percentage increase (decrease) in stock return equal to the value of the beta coefficient.
The banks’ weekly stock prices and the market index prices are collected and compiled from Yahoo Finance and are adjusted for splits and dividends. The market index is repre-sented by the OMXS30 index, which is a market value-weighted index that represents the 30 most traded stocks on the Stockholm Stock Exchange. Returns are calculated as com-pounded returns and are used in the regression since this result in lower value (except for zero returns) thus implying that the effect of any outliers or data errors is reduced (Ryan & Worthington, 2004). The weekly cross exchange rates are collected from The Swedish Cen-tral Bank and are also converted into compounded returns. The TCW index (Total Com-petitiveness Weights) is weekly and derived from the Swedish Central Banks database where the same time horizon is used, 20th July 2004 to 13th March 2012 and is weekly aver-age data.
The weekly exchange rates are collected from The Swedish Central Bank and are also trans-lated into compounded returns for consistency. The included exchange rates are SEK/USD, SEK/EUR, SEK/GBP, SEK/DKK, SEK/NOK, SEK/JPY and SEK/CNY2.
Choice of Market Portfolio
This thesis analyse the effect of the exchange rate exposure on stock returns separately from the market exposure. This because the purpose of the thesis is to investigate in the individual banks excess exposure and further try to explain the reason to the excess expo-sure. The particular interest of the thesis is to investigate in the firm specific exposure ra-ther than the overall exposure since it is the firm specific exposure that is of interest for in-vestors and firm owners. The choice of using an equally weighted index or a value-weighted index shows no difference in a test by Dominguez & Tesar (2005). In this study the value-weighted index OMXS30 (representing the 30 most traded stocks on the Stock-holm Stock Exchange) is used to represent the market.
What defines the relevant exchange rates that should be included in the analysis? No em-pirical studies have yet provided a strategy on how to select the most relevant exchange rates. Widely used is a trade-weighted portfolio of currencies, however very poor results are found when using this method (Doidge, Griffin & Williamson, 2002). The results may be smoothed out and high exposure in one currency may be off set with no exposure in an-other currency. Further, using an exchange rate index will not provide an overlook on the firm’s responses to different exchange rates, meaning that the firm can be positively related to one currency and negatively related to another that is an effect that will be smoothed out by using an index. Another issue when choosing relevant currencies is that the firms are expected to be properly hedged and the risks limited to the currencies where the firm has much of its activities, while the currencies with no obvious linkage and more indirect ef-fects may be overlooked. With respect to these issues and to the aim of this thesis, which is to find the key currencies the Swedish banks are exposed to, currency pairs will be used in-dividually in the equation to avoid the currencies to offset each other. The chosen exchange rates for this investigation are those in which the Swedish banks have most of their foreign operations namely the US dollar (USD), the euro (EUR), the British pound (GBP), the Danish krone (DKK), the Norwegian krona (NOK). In addition the Chinese yuan (CNY) and Japanese yen (JPY) are included since these currencies are the two countries on the top 15 trading partners with the Swedish economy (Statistics Sweden, 2012). The Dan-ish krone has been pegged to the EUR since the introduction of the latter. This will result in very high correlation between the two currencies but since the regressions will be run on the exchange rate individually this will not be a problem. Both currencies are included in this investigation since they are both two important currencies to the Swedish banks and should therefore not be excluded. Though, the banks are expected to show similar expo-sure to the SEK/EUR and SEK/DKK exchange rates. Further the CNY is pegged to the USD but the same argument applies here as for including both the DKK and the EUR, thus banks expect to show close to similar exposures to the SEK/CNY and SEK/USD.
The exchange rate is the market price for which one currency can be traded for another. In this study the exchange rate are stated as indirect quotes (domestic currency being fixed at one unit and where the foreign currency is a variable amount). Therefore, an increase in the exchange rate indicates a strengthening of the domestic currency, alternatively a weakening of the foreign currency relative to the other currency.
As for testing the robustness of the estimated exposures, a currency index is included in the investigating, however it should be noted that the main focus lies on the individual ex-change rates.
Description of Benchmark
In this thesis a benchmark is included to allow for a comparison on the degree of exposure of the banks and other firms. The benchmark is represented by the four largest corpora-tions in Sweden (defined by market capitalization); Atlas Copco, Ericsson, H&M and Telia Sonera, which are all listed on the Nasdaq OMX Stockholm. The benchmark represents non-financial firms with both assets and liabilities denominated in foreign currencies and thus faces the market risk similar to the banks. The market divisions within the banks pro-vide financial solutions and expertise on hedging to global customers and are a part of the banks business, thus the banks have economies of scale in hedging. Due to the banks ex-pertise in hedging along with their economies of scale it is interesting to make a compari-son on the sensitivity in banks’ stock returns and the sensitivity of the benchmark stock re-turns. The benchmark exposures are measured in the same way as for the banks, using the same equation involving exchange rates and market portfolio3.
Exposure on Firm level
This model will not provide a measure of total exposure of the banks in the banking sector but instead measure the exposure on firm level. The advantage of a total approach would allow for comparison between different industries and reasons to the variations of expo-sure among sectors. However, looking at the exposure on industry level instead of firm lev-el may leave us with biased results due to the effect that the aggregation may even out the firm specific exposures (Dominguez & Tesar, 2001). Since the aim here is to investigate in the individual exchange rate exposure of the banks, the total approach is ignored.
Weak statistical significance
Previous studies have met some limitations in determining the statistically significance of the test on the relation between firm value and exchange rate movements. As discussed by Bartov and Bodnar (1994) there are some possible reasons to why some studies fail to detect the exchange rate exposure. The main reason tends to be poor research design e.g. weaknesses in sample selection such as inclusion of banks with no linkages to international markets and chosen time horizons. Longer time horizons seem to capture the exposure better than short horizons. Dominguez and Tesar (2005) find significant exposure for a fraction of firms, however, the firms affected and the currencies that carry the exposure changes over time. This indicates that firms adjust their hedging strategies over time, which should be a natural action for all firms.
Table of Contents
1.2 Problem Statement
1.4 Review of literature
2 Theoretical Framework
2.1 The Swedish Banks in the International Arena
2.2 Exchange Rate Exposure
3.1 Description of Model
3.3 Data Description
4 Estimated Exposure of the Swedish major Banks: Result and Analysis
4.1 The banks’ estimated exposure
4.2 Determinants of Exchange Rate Exposure
4.3 Bank exposure vs. Non-financial firms exposure
5.1 Suggestions for future studies
List of reference
GET THE COMPLETE PROJECT
Exchange Rate Exposure of Swedish Banks