The Impact of Abandoning a Fixed Exchange
Rate Regime on Stock Returns: Evidence from
Switzerland
Akkelyn Tabak1 Thesis MSc. Finance University of Groningen
January 2016
Supervisor: Prof. Dr. W. (Wolfgang) Bessler
Abstract
This study reviews the impact of the change in exchange rate regime on January 15th 2015 in Switzerland. The impact is measured for listed Swiss firms. Short-term and long-term analyses are performed using daily data on stock prices. The results of the short-term analysis indicate that the impact was different across industries and that the percentage of foreign sales and the percentage of foreign assets negatively influenced abnormal returns on a firm level. The long-term analysis explores whether the impact of exchange rates on the Swiss Performance Index (SPI) and the Swiss Market Index (SMI) changed after January 15th 2015. Results show there was a structural break for both of the exchange rates (CHF/EUR and CHF/USD), which indicates that the impact of exchange rates on market prices changed after unpegging the Swiss franc from the euro on January 15th 2015.
Key words: Exchange rate risk, exchange rate regimes, Swiss franc, abnormal returns, structural breaks
JEL-Classifications: F31, F33, G12
2
Introduction
On January 15th 2015 the Swiss National Bank (SNB) made the decision to discontinue its relatively fixed exchange rate regime2. The purpose of this study is to create an understanding of the impact of this decision. By unpegging the Swiss franc the SNB adjusted their relatively fixed exchange rate regime to a floating exchange rate regime, which led to an appreciation of the Swiss franc of more than 20%. This appreciation caused the whole stock market to collapse, with some firms experiencing a larger impact than other firms. This study provides a general understanding of the industries and individual firms that were hurt the most by this decision of the Swiss National Bank on the short-term and investigates whether the change in exchange rate regime had an impact on the Swiss stock market in the long-term.
Exchange rate risk and the resulting exchange rate exposure are well-researched topics in the literature of finance and economics. While exposure can be defined as the impact of a change in exchange rate on the balance sheet and cash flows (Dufey, 1972, Heckerman, 1972, Shapiro, 1975, 1977), it can also be defined as the impact it has on the market value of a firm (Jorion, 1990). Many studies focus on the impact of changes in the exchange rate on stock returns but there have not been many convincing results. Some researchers found a weak link between changes in the exchange rate and stock returns (Jorion, 1990), while others found mixed results (Bartov and Bodnar, 1994), or even an interdependent relationship (Bahmani and Sohrabian, 1992). In the case of Switzerland, with such a large unexpected exchange rate shock resulting in a substantially large drop of the stock market, we assume that the relation is present, at least in the short-term. This research will therefore not focus on finding evidence for the relationship between exchange rates and stock prices in the short-term, but will explore whether industries were impacted differently as a result from the exchange rate regime change. The second part of the short-term research will investigate which variables impacted abnormal returns resulting from the exchange rate regime change. For the long-term, a more general study will be conducted to find out whether the impact of the exchange rate on the market changed because of this event.
For the short-term research we find that industries indeed experience different impacts. On a firm-level, firms with a large proportion of foreign sales and/or foreign assets experience more negative abnormal returns than firms with low international activity. If we combine the industry-level analysis with the firm-level analysis, we see that industries with relatively large exports are impacted the most. For the long-term research we find that the impact of the exchange rate on the Swiss market changed when comparing a period before January 15th 2015 to a period after the event.
3 This paper contributes to the existing literature because the situation in Switzerland provides us with a unique dataset. The decision of the SNB resulted in an appreciation of the Swiss franc3 of around 20% in just hours of time. Such a large currency shock in such a short period of time is very uncommon, in particular for a major currency. Nobody (e.g. investors, firms and financial analysts) expected this situation, which means the exposure resulting from the shock was not anticipated by market participants. These aspects make the Swiss situation fairly unique. Hence, the results of this study will help understand the impact of major currencies changing their exchange rate regime unexpectedly on the various industries and individual firms in a country.
The remaining of the paper is structured as follows. It will start with background information on the situation in Switzerland. Next, a literature review is provided, including hypotheses. After that the methodology is presented. The next chapter will contain data and descriptive statistics followed by a section that will contain the results and discussion. The last chapter provides a conclusion.
2. Background information
After three years of pegging the Swiss franc to the euro, the SNB unexpectedly announced to change their policy from a fixed exchange rate regime into a floating exchange rate regime4. At the same time the SNB announced that it would lower the interest rates on sight deposits from -0.25% to -0.75% to ensure that abandoning the fixed exchange rate would not lead to an “inappropriate tightening of monetary conditions”. On January 15th of 2015 these decisions of the SNB caused the Swiss franc to appreciate from being worth CHF 1.20 per euro to CHF 0.85 per euro at one point and closing at CHF 1.04 per euro. The graph below displays the development of the exchange rate of the Swiss franc over time.
3 Compared to the euro
4 Graph 1
History of the Swiss Exchange rate
The graph shows the value of the Swiss franc per euro during the period of January 2014 until mid-November 2015.
According to the SNB the fixed exchange rate was introduced in a time where financial markets were very uncertain and the Swiss franc was largely overvalued4. To protect the Swiss economy, the SNB decided to introduce a floor in 2012. This led investors to believe the Swiss franc was a low-risk asset (www.economist.com). As a result, demand for Swiss francs increased drastically, putting an upward pressure on the exchange rate. To keep the exchange rate above the floor of CHF 1.20 per euro, the SNB had to print more Swiss francs to buy foreign reserves and thus reduce the pressure. On January 15th this policy of maintaining the Swiss exchange rate was discontinued by the SNB. The SNB stated this was the result of increasing divergences of the monetary policies of the major currency areas. The depreciation of the euro against other major currencies, for example the US dollar, had led the Swiss franc to weaken to these currencies as well, and the SNB concluded that the fixed Swiss franc exchange rate against the euro was no longer justified.
Abandoning the fixed exchange rate will have and has already had consequences for the Swiss economy. In an article of Reuters on January 15th was predicted that the two largest Swiss banks, Credit Suisse and UBS, will suffer majorly from the Swiss franc appreciation (www.reuters.com). These banks derive much of their earnings overseas, which leads to lower reported earnings and higher costs denominated in francs. Not only banks will suffer from the appreciation, according to The Guardian, exporters and the tourism industry will also face large losses (www.theguardian.com). The chief executive officer of the export oriented Swatch group, which owns various watch and jewellery brands, even described the action of the SNB as a financial tsunami, hurting the entire country
0,85 0,9 0,95 1 1,05 1,1 1,15 1,2 1,25 1,3
5 (www.reuters.com). These examples bring us to the aim of this paper, which is to explore the impact of the SNB changing their exchange rate regime on Swiss firms.
3. Literature review
The purpose of this section is to provide an understanding of the variables related to the research topic. First, exchange rate risk and the resulting exposure are discussed. Then, the relation between exchange rates and stock prices is investigated. Thereafter, the impact of a change in exchange rate regime is discussed. Then, industry-level impact is researched and lastly the hypotheses are stated.
3.1 Exchange rate risk and exposure
Exchange rate movements can result in exposure, which is why firms experience exchange rate risk. Exposure can be defined in multiple ways of which the following two are broadly recognized (Dufey, 1972, Heckerman, 1972, Shapiro, 1974, 1977). The first, which is the accounting definition of exposure, concerns impacts on the balance sheet because of a change in exchange rate. The various foreign assets and foreign liabilities on the balance sheet are affected differently by a change in the exchange rate, which might lead to a loss5. The second type of exposure is the influence of a change in the exchange rate on future cash flows. If a large proportion of future cash flows is obtained in a foreign currency, exposure increases. Dufey (1972) argues in his study that these two types of exposure can have opposing effects. A loss on the balance sheet may be offset by increased profitability due to, for example, higher revenues. These opposing effects complicate measuring and comparing the impact of a change in exchange rate across firms, because every firm will be affected differently.
In addition to exposure as a result of foreign operations, domestic firms without foreign operations can be affected by a change in exchange rate as well (Hodder, 1982, Adler and Dumas, 1984, Bardov and Bodnar, 1984). For instance, through a change in exchange rate affecting other major economic factors (e.g. interest rates), resulting in at least some exposure. Furthermore, firms without foreign operations might also be indirectly impacted because its customers or suppliers could be affected by the change in exchange rate. Hertzel et al. (2008) found that negative news concerning one firm can have negative consequences for other firms in the supply chain, which is called supply chain contagion. Contagion can also occur between competitors. Akhigbe et al. (2014) found that an unexpected negative change in stock price resulted in negative abnormal returns for competitors. Another form of contagion is financial contagion, which is mainly present in the banking sector. This involves banks impacting other banks’ balance sheets through claims that they have on each other. (Allen and Gale, 2000). Hence,
6 if one bank is impacted by the exchange rate change, it is likely that other banks are impacted as well, indirectly.
While exposure is one of the risks associated with exchange rate movements, the exchange rate itself can also be of risk (Adler and Dumas, 1989). When changes in the exchange rate are expected, a firm can decide to hedge against it, hence reducing losses when the change occurs. In contrast, when a change in exchange rate is not expected, a firm can occur large unanticipated losses. Therefore, a strong currency is not necessarily less risky than a weak currency. The case of Switzerland demonstrates that. Because the Swiss franc is a major currency, nobody expected this change to happen, and considered the Swiss franc to be a relatively safe currency. When the unexpected did happen, not many firms were prepared for the resulting shock,
To measure the total exposure resulting from exchange rate fluctuations, various techniques can be applied. A general method for measuring exposure is through decomposing the future value of a firm into two parts: a part that is independent from future random fluctuations in the exchange rate and a part that is directly related to the fluctuation. This second part is the value of the exposure (Dumas, 1978). This exposure can be calculated either from values obtained from the balance sheet and income (or cash flow) statement or from market values (stock prices). The next part of this chapter will focus on exposure measured by a change in the market value (stock price) of the firm.
3.2 The impact of exchange rate risk on stock returns
It is a common belief that a change in exchange rate impacts stock prices. While many studies focus on this topic, most findings on this relationship are not significant or provide mixed or weak results. Jorion (1990) found evidence for a weak relationship between the exchange rate and stock returns in the United States. In addition, he found that this relationship differed systematically across firms. One of his findings was that multinationals with a large proportion of foreign sales experienced a stronger effect from a change in the exchange rate. Bartov and Bodnar (1994) studied the relation between exchange rates and firm value in the United States as well, using a different sample selection method. For their sample they selected only firms with high reported impacts on their income from currency changes, which is how they identified the firms which would most likely experience the largest effects in firm value. The results of their study were not significant. However, when they included a lagged value of the exchange rate into the regression, they did find significant results, suggesting some form of mispricing. A possible explanation provided for this mispricing is that investors need time to comprehend the influence a change in exchange rate will have on the firm.
7 (1972) investigated whether stock prices of high intensity multinationals6 in the United States reacted differently to a change in exchange rate than low intensity multinationals. While they found instances where the performance of the stocks in the high and low companies were significantly different from the market, in general, they found no consistent pattern of performances for these groups. Therefore, Franck and Young (1972) concluded that the relationship between major changes in the exchange rate and stock returns depend on the particular event.
More recent research on the relation between exchange rates and stock returns provides similar mixed results. Lee et al (2012) found no significant results for exchange rates impacting the profitability of foreign operations in a study of 261 US multinationals in 2012. More recent research is also much more focussed on countries other than the US. Doukas et al. (1999) found a significant relation between exchange rates and stock returns in Japan and Doukas et al. (2003) found that the relation between foreign exchange rates and stock prices was stronger for multinationals and firms in Japan with high foreign linkages. Fribergh and Nadyhl (1999) found that stock markets in a relatively open economy experience stronger effects from exchange rate changes than stock markets in a relatively closed economy7. They attribute this effect to the fact that firms in a relatively open economy are typically more related to foreign trade, and hence experience a larger exposure. Similar results are found by Donnelly and Sheely (1996), when they analysed the relation between changes in the exchange rate and stock returns in the United Kingdom, which they classify as a far more open economy than the United States. Dominguez and Tesar (2001) studied exchange rate exposure, measured by weekly stock returns, in eight countries8. They found significant evidence for a relationship between exchange rates and stock returns and they also found that the smaller countries in their sample appeared to have a smaller exchange rate exposure than the larger countries in their sample.
Instead of investigating exchange rates impacting stock return, some studies investigated whether the causality is not the other way around: a change in stock returns impacting the exchange rate. Bahmani and Sohrabian (1992) argue that previously conducted studies came to the conclusion of exchange rates impacting stock returns by regressing stock prices on exchange rate risk without investigating causality first. Their study provides evidence for a dual causal relationship in the United States in the short run, implicating that exchange rates and stock returns impact each other. In addition, they did not find a long run relationship between exchange rates and stock returns. Ajayi et al. (1996)
6 The intensity of multinationals was measured by combining variables concerning foreign sales, foreign assets and other foreign earnings.
7 Their sample consisted of monthly data of 11 industrialized countries for the period of 1973-1996, where openness of a country was measured by the total export and import divided by total GDP.
8 found similar significant dynamic relations, for the short as well as the long run, combining data from eight countries9. Alagidede et al. (2011) found no evidence for a long-term relationship between the variables. They did find causality for exchange rates impacting stock returns in the short run for Canada, Switzerland and the UK, and a weak relationship of stock returns impacting the foreign exchange rate in the short run for Switzerland. Hence, while most studies assume a causal relationship of exchange rates impacting stock returns, this does not necessarily have to be the case.
3.3 Impact of a change in exchange rate regime
While many studies focus on the impact of exchange rate risk and exposure in general, other studies investigate the impact of changing from a fixed exchange rate regime into a floating exchange rate regime. Chortareas et al. (2010) conducted an event study to investigate the impact of changing a fixed exchange rate into a flexible exchange rate on stock returns in MENA countries. They found evidence for abnormal returns as a result of the announcement in Turkey and Egypt. However, no evidence was found for Morocco. These findings might be explained by the fact that the announcements in Turkey and Egypt were less anticipated than the announcement in Morocco. According to Chortareas these findings have direct implications for policy makers, which should keep in mind a sudden announcement can have a large impact on the stock market. Bartov and Bodnar (1996) found that after the breakdown of the Bretton-Woods system10, the change from a fixed exchange rate regime into a variable exchange rate regime resulted in increased volatility in stock prices for firms in the United States. In addition, they found that multinational firms experienced higher variability in stock returns than domestically oriented firms. Hence, consistent with general findings on impact of a change in exchange rate, changing the exchange rate regime from fixed to variable, should especially impact multinational firms.
3.4 Impact on an industry-level
Bodnar and Gentry (1993) conclude in their research that industries with high export levels, are negatively impacted in case of an appreciation of the home currency. Industries reliant on import, experience opposite effects. Hence, so called global industries should experience larger impacts in case of an exchange rate change. The concept of global industry can be defined in multiple ways and has changed over time. According to Porter (1986) a global industry is a collection of firms that are linked to each other and compete with each other on a worldwide basis. Yip (1992) defines a global industry based on the extent to which an industry is linked across countries. Another similar definition comes
9 These are the following countries: Canada, France, Germany, Italy, Japan, Netherlands, UK and US.
9 from Morison and Roth (1992), which define a global industry as “distinct competitive environments that are differently interdependent”. To conclude, global industries should experience a larger impact, either negative or positive, from exchange rate changes than industries that are not very reliant on imports or exports
3.5 Hypotheses
Based on the literature provided, the following hypotheses can be formulated:
Hypothesis 1: The mean abnormal returns of industries, resulting from the change in exchange rate, are significantly different from each other.
Hypothesis 2a: The percentage of foreign sales of a firm impacts abnormal returns resulting from the change in exchange rate.
Hypothesis 2b: The percentage of foreign assets of a firm impacts abnormal returns resulting from the change in exchange rate.
Hypothesis 2c: An interaction term of percentage of foreign sales and foreign assets impacts abnormal returns resulting from the change in exchange rate.
Hypothesis 3: The impact of exchange rates on the Swiss stock market changed after January 15th 2015.
These hypotheses will be elaborated on in the next chapter.
4. Methodology
This section contains an elaboration on the hypotheses and the corresponding methodology. As mentioned before, this research exists of a part focused on the short-term impact and a part focused on the long-term impact of the exchange rate change. The research on the short-term impact contains two broad hypotheses. The first hypothesis investigates whether industries were impacted differently by the change in exchange rate. The other hypothesis is focused on variables that impact abnormal returns on a firm-level. The second part of the research will focus on the long-term effects of the change in exchange rate regime. Here, the impact of the exchange rate in general will be investigated. The resulting hypothesis is whether the impact of exchange rates on the market of Switzerland changed during the period after January 15th.
4.1 Short-term impact
As mentioned in the introduction, the main focus of the short-term study is not finding evidence for a relationship between the exchange rate and stock prices, but rather to explore whether industries were impacted differently and which underlying variables impacted abnormal returns.
10 The first research question is whether there is a difference between industries in the impact of the change in exchange rate. The exchange rate exposure will be measured by abnormal returns, which will be calculated using the market model. Mean abnormal returns will be compared across several industries followed by a conclusion on whether these groups are significantly different from each other. This results in the following hypothesis:
Hypothesis 1: The mean abnormal returns of industries, resulting from the exchange rate change, are significantly different from each other.
To compare the impact of the exchange rate on the various industries, mean abnormal returns of the industries are calculated, where abnormal return is defined as actual return minus expected return. To calculate the expected return, beta’s from the firms in the sample were estimated. The beta of a particular stock is defined as:
𝛽𝑖 = 𝐶𝑂𝑉𝑖,𝑚
𝑉𝐴𝑅𝑚 (1)
where:
βi = Beta of the firm (i).
COVi,m = Covariance of the firm (i) with the market (m),
VARm = Variance of the market (m).
The estimated beta’s were used to calculate the abnormal returns per firm:
𝐴𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑅𝑖 = 𝐴𝑐𝑡𝑢𝑎𝑙 𝑅𝑖− 𝛽𝑖∗ 𝑅𝑚 (2)
where:
Abnormal Ri = abnormal return of the firm (i).
Actual Ri = actual return of the firm (i).
βi = beta of the firm (i)
Rm = return of the market (m).
To compare the calculated mean abnormal returns of the various industries, a one-way ANOVA-analysis will be performed11.
With an ANOVA analysis the means of the different industries will be compared, followed by a conclusion whether or not those are significant from each other. This leads to the following hypothesis:
11 𝜇1= 𝜇2= 𝜇3 = 𝜇𝑛
where:
𝜇1= mean abnormal return of industry 1
𝜇2 = mean abnormal return of industry 2
𝜇3 = mean abnormal return of industry 3
. . . 𝜇𝑛 = mean abnormal return of industry n
If the mean abnormal returns are significantly different from each other, a conclusion is drawn that there is evidence for industries reacting differently to a change in exchange rate regime. However, significant results only imply that at least one of the industries experiences a different impact from the change in exchange rate. No conclusions can be drawn on which industries respond differently. To find out which industries are different from each other, we will compare each industry separately with all other industries, using a student’s t-test. This leads to the following hypotheses:
𝜇1= 𝜇2
𝜇1 = 𝜇3
𝜇1 = 𝜇4
. . . 𝜇𝑛 = 𝜇𝑚
The difference with the previous test is that the mean abnormal return of one industry is compared directly to one other industry. Conclusions can be drawn on which industries experienced similar abnormal returns.
4.1.2 Factors impacting abnormal returns
The literature review discusses a number of factors that increase or decrease exchange rate exposure. Many studies emphasize that multinational firms experience larger impacts from changes in the exchange rate, which can be measured by the following variables:
Percentage of foreign sales
12
Percentage of foreign assets
As discussed in the literature review, the value of foreign assets might change when a change in exchange rate occurs. In the case of Switzerland, a high percentage of foreign assets at the moment of the appreciation of the Swiss franc, should have led to a loss on the balance sheet. Hence, we expect that the percentage of foreign assets has an impact on abnormal returns in case of an unexpected exchange rate change.
Because there might be a chance that the impact of the percentage of foreign sales on the abnormal return depends partly on the percentage of foreign assets and the other way around, an interaction term is included in the model. If this coefficient appears to be significant, these two variables interact with each other in impacting abnormal returns.
The following hypotheses can be formulated:
Hypothesis 2a: The percentage of foreign sales of a firm impacts abnormal returns. Hypothesis 2b: The percentage of foreign assets of a firm impacts abnormal returns.
Hypothesis 2c: An interaction term of percentage of foreign sales and foreign assets impacts abnormal returns.
Since the number of firms per industry is low and the number of missing values for the variables percentage of foreign sales and percentage of foreign assets is relatively large, industry effects will not be included in the model. To increase the prediction capabilities of the model, we also added control variables to the regression. Because size is a frequently used control variable in predicting stock returns, the control variables total sales and total assets are included in the model. These control variables were scaled by dividing them all by 1,000,000.
To test the hypotheses an Ordinary Least Squares (OLS) regression will be conducted. The following model can be specified:
𝐴𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑅𝑖 = 𝛼 + 𝛽𝐹𝑆𝐹𝑆 + 𝛽𝐹𝐴𝐹𝐴 + 𝛽𝑇𝑆𝑇𝑆 + 𝛽𝑇𝐴𝑇𝐴 + 𝛽𝐼𝑇𝐼𝑇 + 𝜀𝑖 (3)
where:
𝐴𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑅𝑖 = the abnormal return of firm (i).
𝛼 = the intercept
13 𝐼𝑇 = interaction term of foreign sales and total sales.
𝜀𝑖 = the error term.
Before estimating the regression a few tests will be conducted to test whether the assumptions of OLS are not violated. We will test whether heteroscedasticity is an issue, test for multicollinearity and test whether the data is normally distributed. The Breusch-Pagan test (Breusch and Pagan, 1979) is conducted to test whether there is heteroscedasticity present in the sample. If this is the case, robust standard errors will be used to correct for the impact of heteroscedasticity. Multicollinearity is tested for using the Variance Inflation Factor (VIF). With a value for VIF>10 (Kutner et al., 2004), multicollinearity might be an issue.
Abnormal returns will be measured for three days, which are the day before the event (January 14th, 2015), the day of the event (January 15th, 2015) and the day after the event (January 16th, 2015). The day before the event is measured to control for possible information leakage. If abnormal returns are low compared to the abnormal returns on January 15th and January 16th, we assume the event was unexpected. We will also perform a separate regression for the sum of the abnormal returns of January 15th and January 16th. For some industries impacts might be more difficult to estimate and therefore experience a lagged impact. By summing up the abnormal returns of January 15th and January 16th, this difference disappears. To summarize, four separate regressions will be performed for abnormal returns on January 14th, January 15th, January 16th and the sum of January 15th and January 16th.
4.2 Long-term impact of the change in exchange rate
The hypothesis concerning the long-term impact of the exchange rate is:
Hypothesis 3: The impact of exchange rates on the Swiss market changed after January 15th 2015.
To test this hypothesis, a Chow-test will be performed (Brooks, 2008). The Chow-test is used to investigate whether there is a structural break at some point in time.
The Chow-test shows whether the coefficient of the exchange rate (EUR/CHF and USD/CHF) impacting the Swiss stock market in the period before the event is equal to the coefficient of the exchange rate (EUR/CHF and USD/CHF) in the period after the event. The model can be specified as follows:
14 where:
𝑀𝐼𝑡 = the value of the market index
𝛼𝑡 = the intercept
𝐹𝑋1𝑡 = exchange rate of Swiss franc with euro
𝐹𝑋2𝑡 = exchange rate of Swiss franc with US dollar
𝜀𝑡 = error term
Using the Chow-test we explore whether there is evidence that there is an alternative specification for two groups, one being the period before January 15th and the other group, the period after. We will hence include a dummy variable indicating those two groups. The model to be tested is then specified as follows:
𝑦𝑖 = 𝑥𝑖′𝛽 + 𝑔𝑖 𝑥𝑖′𝛾 + 𝜀𝑖
Where:
𝑥𝑖′ = the vector of explanatory variables
𝑔𝑖 = the dummy variable indicating the suspected breakpoint.
We will perform a joint hypothesis test whether gamma is significantly different from zero. If we are able to reject the null-hypothesis, we found proof that there was indeed a structural break on January 15th.
4. Data and descriptive statistics
Since the study is focused on Switzerland, the sample contains only Swiss firms that are listed on the Swiss Stock Exchange. The data was collected from Thomson Reuters’ Datastream and Orbis. First a list of firms, identified by ISIN numbers was obtained from Orbis, which was then imported into Datastream to obtain the necessary data on the variables. The total sample consisted of observations on 218 Swiss listed companies, which are the firms that are included in the Swiss Performance Index minus the foreign firms that had a primary listing in Switzerland but were not Swiss from origin.
For the short-term analysis daily stock data was collected from the first of January 2014 to January 16th 2015. To conduct an ANOVA- analysis the data was classified into industries using the Industry Classification Benchmark (ICB), which is a frequently used tool in research12 (www.icbenchmark.nl). There are four levels of classification, which are the industry, supersector, sector and subsector. Because the total sample of Swiss firms is relatively small the broadest classification was used, which is the industry-level. This results in ten categories of which some groups are relatively large, while others are small or even zero (see table 1).
15 To compare the industries, mean abnormal returns were calculated with an estimated beta. The market index used in the estimation of the beta’s was the MSCI Europe excluding Switzerland. The Swiss Market Index was not appropriate for this research because of endogeneity issues, which is why a market index that did not include Swiss stocks was used. We assume that this index was not or minimally influenced by the event. The MSCI excluding Switzerland captures stocks from 14 developed countries in Europe and contains almost 85% of the free float-adjusted market capitalization of the European Developed Markets (excluding Switzerland). The UK, Germany and France are the largest contributors to the MSCI ex Switzerland (www.MSCI.com). With the MSCI ex Switzerland beta’s were estimated using daily stock data from January 2014 to December 2014. Since not all firms in the sample were listed since January 2014, those firms (7) were excluded from the sample. Since one of the industries (Telecommunication) contains only one observation, we excluded the remaining firm in this category from the sample as well. This results in a final sample of 210 firms that are divided over 8 industries. The composition of the final sample can be found in table 1.
Table 1
Observations for each of the Industries
The table displays the various industry categories with the number of Swiss firms for that industry, the total number of observations available for each industry and lastly the final sample. Totals are displayed at the bottom of the table.
Industry Number of Swiss Firms Observations Final Sample
Oil and Gas 0 0 0
Basic Materials 12 12 12 Industrials 56 55 55 Consumer Goods 19 19 19 Health Care 22 21 21 Consumer Services 15 15 15 Telecommunication 2 1 0 Utilities 6 6 6 Financials 73 69 69 Technology 13 13 13 Totals 218 211 210
The mean number of observations per industry is 21.1, while the median is 14. The maximum number of observations is 69, which is the industry Financials. The minimum number of observations is 0, which is Oil and Gas.
16 for all of the variables, which led to a smaller sample than the sample used to compare abnormal returns across industries. Because the number of observations for foreign sales as a percentages of total sales are considerably higher than the number of observations for foreign assets as a percentage of total assets, first a regression with solely foreign sales and the control variables was conducted13. After that a regression with foreign sales, foreign assets and an interaction term of foreign sales and foreign assets, was conducted. The dependent variables remain equal to the abnormal returns that were estimated in the first part of the short-term analysis.
The long-term impact of the change in exchange rate regime was measured for the Swiss market as a whole, which means that other variables were needed than for the short-term study. The dependent variable is the Swiss market, which was estimated by values of the Swiss Market Index (SMI) and for robustness purposes also the Swiss Performance Index (SPI). While the Swiss Market Index only includes 20 of the largest firms in Switzerland, the SPI includes approximately 230 listed firms. One drawback of using the SPI, is that it also includes a few companies that are not Swiss from origin, but do have a primary Swiss listing. From both the SMI and the SPI, daily values were collected from the first of April in 2014 (event -10.5 months) until the first of November in 2015 (event +10.5 months)14. These values were collected from either Datastream (SMI) or the website of the Swiss Exchange (SPI)15 (www.six-swiss-exchange.com). Besides daily values of the market indices, daily values on the exchange rate of Switzerland (CHF/Euro and CHF/USD) were also collected for the same period. These values were obtained from Datastream as well. Combining the daily data on the market indices and the daily data on the foreign exchange rates results in a sample of 397 observations for each of the variables.
5. Results and Discussion
The results are divided into two sections. The first section provides the results of the short-term impact, by looking at the impact across industries and the OLS regression conducted on a firm-level. The results of the industry-level analysis will be compared with the results of the firm-level analysis as well and we will take a closer look at two industries. The second section contains the results of the long-term impact of the exchange rate change.
5.1.1 Short-term impact across industries
13 A table with the number of observations available can be found in appendix 2 (Table 1)
17 Abnormal returns were calculated for three days, January 14th, January 15th and January 16th 2015, using the MSCI market index excluding Switzerland. The sum of the abnormal returns of January 15th and January 16th was calculated as well.
Before conducting the ANOVA-analysis to compare the means of each industry per abnormal return category, the data was tested on violations of the assumptions of ANOVA. The first assumption of independent observations was not violated, since none of the firms overlap with each other. The second assumption of all categories having equal variances was not violated as well16. However, the third assumption of all categories being normally distributed was violated. Evidence was found that the data of one or multiple categories was not normally distributed for each of the categories17. To take this non-normality in account the Kruskal-Wallis H test was conducted, which is a non-parametric test. The results of a Kruskal-Wallis H test can be interpreted similar to a one-way ANOVA. The table below displays the average abnormal returns per industry and the outcomes of the Kruskal-Wallis H test.
Table 2
Mean Abnormal Returns per Industry
The table displays the mean abnormal returns per date for each industry calculated with the MSCI Europe index excluding Switzerland. The total abnormal returns (sum of January 15th and January 16th) are displayed per industry as well. The total average abnormal returns of all industries can be found below the industries. At the bottom of the table the Kruskal-Wallis chi-squared values are displayed and whether or not these values were significant and at what level.
Mean Abnormal Returns
Industry January 14th January 15th January 16th Total
Basic Materials 0.10% -5.15% -6.13% -11.28% Industrials -0.27% -6.89% -5.50% -12.39% Consumer Goods -0.12% -5.52% -5.25% -10.77% Health Care -0.78% -8.15% -5.07% -13.22% Consumer Services -0.11% -3.92% -3.36% -7.28% Utilities 1.54% -3.52% -2.91% -6.43% Financials -0.14% -4.13% -2.94% -7.07% Technology 0.90% -6.93% -4.67% -11.60% Average -0.18% -5.53% -4.48% -10.01%
Chi-Squared Value of the
Kruskal-Wallis H Test 10.322 18.192** 23.229*** 24.461***
***,** and * denote the significance level at the 1, 5 and 10 percent levels, respectively, for the reported chi-squared values.
The table displays the mean abnormal returns per industry and the total abnormal returns for the four categories. If we compare the abnormal returns of January 14th with January 15th and January 16th
16 This assumption was tested for using Bartlett’s test in Stata, which provided no evidence for the categories not having equal variances. Hence, the hypothesis of equal variances was not rejected.
18 we see a very large difference, indicating that the event was indeed unexpected. The results show that all of the industries were impacted negatively by the appreciation of the home-currency, with negative abnormal returns on January 15th and January 16th. Furthermore, we see that the distribution of abnormal returns per industry were significantly different from each other. Hence, the null-hypothesis of all industries experiencing the same impact is rejected. However, no conclusions can be drawn on which of these industries reacted differently because the test only provides evidence for at least one category being statistically different from the other categories. Also, this test provides no information on reasons why some industries would be impacted differently than other industries.
To investigate which specific industries respond differently compared to other industries, we compared the total abnormal returns (the sum of January 15th and January 16th) industry by industry. For the total abnormal returns, only one category (Financials) is not normally distributed. This is why we decided to use the students t-test because we expect that the consequences of one category with no normal distribution are minimal. The table below summarizes the results of the test.
Table 3
19 This table provides the results of the comparison of means test. A students t-test was used to compare each of the industry abnormal returns (sum of January 15th and January 16th) with each other. The cells in the tables report the absolute difference between the mean abnormal returns of the industries. The hypothesis tested is whether the absolute difference is larger than zero
Bas ic Ma ter ial s Indus tr ia ls C onsum er G oods H ea lt h C ar e C onsum er Ser vi ce s U ti li ti es Fina nc ia ls Techno logy Basic Materials 0.0112 0.0050 0.0194 0.0399* 0.0484* 0.0420** 0.0033 Industrials 0.0112 0.0162 0.0082 0.0511*** 0.0596** 0.0532*** 0.0079 Consumer Goods 0.0050 0.0162 0.0244 0.0349* 0.0434* 0.0370** 0.0083 Health Care 0.0194 0.0082 0.0244 0.0593** 0.0678** 0.0614*** 0.0161 Consumer Services 0.0399* 0.0511*** 0.0349* 0.0593** 0.0085 0.0021 0.0432* Utilities 0.0484* 0.0596** 0.0434* 0.0678** 0.0085 0.0064 0.0517* Financials 0.0420** 0.0532*** 0.0370** 0.0614*** 0.0021 0.0064 0.0453** Technology 0.0033 0.0079 0.0083 0.0161 0.0432* 0.0517* 0.0453**
***,** and * denote the significance level at the 1, 5 and 10 percent levels, respectively, for the reported absolute differences.
Table 3 provides an overview of the absolute differences between the total mean abnormal returns of the various industries. Of these industries, consumer services, utilities, and financials are the categories that differ most from the other categories, with four significant differences. These industries are also the three industries that had negative abnormal returns closest to zero. The graph below provides an overview of the abnormal returns per industry on January 15th and January 16th and the sum of those. Here you can see that we can broadly identify two groups. Utilities, financials and consumer services can be grouped together and consumer goods, basic materials, technology, industrials and health care. This is consistent with the pairs of means test because the industries in one group are significantly different from all industries in the other group. It appears that industries that mainly provide services, were impacted less than production industries. A possible explanation could be that service industries generally do not export as much as production industries, and as such have less sensitivity to exchange rate changes.
Figure 1
20 This bar chart provides an overview of the abnormal returns per industry on January
15th, January 16th and it also provides the sum of January 15th and 16th.
5.1.2 Variables impacting abnormal returns
To investigate what variables impacted abnormal returns, a regression analysis was conducted. First a regression with the control variables and only foreign sales was conducted because there was more data available on the variable foreign sales than on the variable foreign assets. Second, a regression with the control variables, foreign sales, foreign assets and an interaction term (foreign sales * foreign assets) was conducted. The interaction term was included to explore whether the impact of the explanatory variables depend on each other’s value.
5.1.3 Regression with foreign sales
Before conducting the Ordinary Least Squares (OLS) regression with the control variables and foreign sales, it was tested whether the assumptions for OLS hold. The assumption of homoscedasticity for the residuals holds for three of four categories of abnormal returns, except for January 14th, which is why we use robust standard errors in the OLS for January 14th 18. Multicollinearity does not appear to be a problem because the Variance Inflation Factor is low (VIF<2) for each of the abnormal return
18 A Breusch-Pagan test was conducted to test for heteroscedasticity in the residuals and the null-hypothesis of homoscedasticity was not rejected for January 15th, January 16th and the sum of January 15th and 16th. Results of the Breusch-Pagan test can be found in appendix 3.
21 categories.19. Hence, an OLS regression was used to estimate the coefficient for foreign sales with abnormal returns being the dependent variable. The table below summarizes the results of the OLS-regression
Table 4
Results from the OLS-regression with Foreign Sales.
An OLS-regression was conducted with abnormal returns, calculated with the MSCI Europe index excluding Switzerland, being the dependent variable. The variables total sales and total assets were measured in millions of US dollars, while the variables foreign sales and foreign assets were measured as a percentage of total sales. For each of the four categories of abnormal returns, coefficients of the control variables and explanatory variables are provided in the table below. The standard errors are reported in parentheses. For January 14th heteroscedasticity was detected. Hence, the reported standard errors for January 14th are robust standard errors. Significant coefficients are marked.
Abnormal Returns
Variable January 14th January 15th January 16th Total
α 0.0081*** -0.0495*** -0.0423*** -0.0918*** (0.0030) (0.0075) (0.0062) (0.0112) Total sales 0.0000 -0.0004 -0.0001 -0.0006 (0.0001) (0.0004) (0.0003) (0.0000) Total assets 0.0000 0.0000 -0.0000 0.0000 (0.0000) (0.0000) (0.0000) (0.0000) % Foreign sales -0.0107** -0.0231** -0.0102 -0.0334** (0.0045) (0.0098) (0.0080) 0.0145 Total observations 152 152 152 152 F-value 2.13* 3.79** 1.29 3.72** R-Squared 0.0427 0.0713 0.0254 0.0701
***,** and * denote the significance levels at the 10, 5 and 1 percent levels, respectively, for the reported coefficients
The results in the table indicate that the impact of foreign sales on abnormal returns is negative and significant for all four abnormal return categories. For January 14th, the day before the event, the model is significant. The coefficient of foreign sales is significant as well, which could be an indication for foreign sales also having an impact on normal trading days. However, the coefficient is lower compared to January 15th, which could suggest a weaker relation. Also, the R-Squared is lower for January 14th indicating that the specification has a better fit for January 15th and the total abnormal returns.
For January 15th the model is significant as a whole. The coefficient of foreign sales is significant at the 5%-level and more negative than for January 14th and January 16th, which may indicate a stronger relationship. The R-Squared is also higher, which means a larger part of total variation was explained.
22 Hence, the percentage of foreign sales of firms seem to have an impact on abnormal returns on the day of an unexpected change in exchange rate.
The model does is not very appropriate to estimate the abnormal returns for January 16th because none of the coefficients is significant, the model is not significant as a whole, and the R-Squared is low. Hence, there is no evidence for foreign sales impacting abnormal returns the day after a large, unexpected exchange rate change. Since the abnormal returns of January 16th are very negative, it might be that the impact of an exchange rate change is easier to evaluate for firms that have a large proportion of foreign sales. Perhaps investors realized the day after the event that domestically oriented firms would also (indirectly) be impacted by the appreciation, which might be a possible explanation for the large abnormal returns on January 16th.
For the total abnormal returns, the sum of the abnormal returns on January 15th and January 16th, the coefficient of foreign sales was also significant and more negative compared to the other abnormal return categories. The R-Squared is comparable to the regression of January 15th, and higher than the other two categories, indicating a better goodness of fit. To conclude, the percentage of foreign sales of a firm is negatively related to the total abnormal returns, summing up the abnormal returns of the day of the appreciation and the day after.
Furthermore, from the OLS regressions it appears that the control variables, total sales and total assets, do not impact abnormal returns because the coefficients of these variables are extremely close to zero and the variables are not significant.
To conclude, it appears that the percentage of foreign sales is negatively related to abnormal returns that firms experience in case of a large, unexpected appreciation of the home currency.
5.1.4 Regression with foreign sales, foreign assets and interaction term.
Before conducting the second OLS regression, with the control variables, foreign sales, foreign assets and an interaction term, it was tested whether the assumptions of OLS hold. The assumption of no heteroscedasticity holds for January 16th and the total abnormal returns, but does not hold for January 14th and 15th, which is why we decided to include robust standard errors for the OLS of January 14th and 15th. Multicollinearity does not appear to be a problem because the Variance Inflation Factor is lower than 5 for all variables, including the interaction term20. Hence, the table below displays the results of the OLS-regression
23 Table 5
Results from the OLS-regression with Foreign Sales, Foreign Assets and an Interaction Effect. An OLS-regression was conducted with abnormal returns, calculated using the MSCI Europe index excluding Switzerland, being the dependent variable. For each of the three categories of abnormal returns, coefficients of the control variables and explanatory variables are provided in the table below. An interaction term is added for the explanatory variables: foreign sales and foreign assets. The standard errors are reported in parentheses. For January 14th and 15th heteroscedasticity was detected. Hence, the reported standard errors for January 14th and 15th are robust standard errors. Significant coefficients are marked.
Abnormal Returns
Variable January 14th January 15th January 16th Total
α 0.0145*** -0.0279*** -0.0298*** -0.0578*** (0.0045) (0.0060) (0.0088) (0.0143) Total sales 0.0000 -0.0002 -0.0001 -0.0003 (0.0001) (0.0002) (0.0003) (0.0005) Total assets 0.0000 0.0000*** 0.0000 0.0000 (0.0000) (0.0000) (0.0000) (0.0000) % Foreign sales -0.0188** -0.0704*** -0.0319** -0.1023*** (0.0089) (0.0131) (0.0134) (0.0218) % Foreign Assets -0.0237** -0.0973*** -0.0581** -0.1554*** (0.0106) (0.0192) (0.02356) (0.0383) Interaction Effect: % Foreign Sales
* % Foreign Assets 0.0233** 0.1241*** 0.0744*** 0.1985***
(0.0118) (0.0175) (0.0251) (0.0408)
Total observations 111 111 111 111
F-value 2.28* 24.51*** 2.30** 6.95***
R-Squared 0.0807 0.2613 0.0988 0.2487
***,** and * denote the significance levels at the 1, 5 and 10 percent levels, respectively, for the reported coefficients
The results in the table indicate that adding the variables foreign sales and the interaction term, improved the specification of the model for all four abnormal return categories because the R-Squared is higher for all categories. For January 14th the model is significant as a whole, and the variables foreign sales, foreign assets and the interaction term are all significant and negative, indicating that the explanatory variables might also have an impact on regular trading days. More research would be necessary to explore this possibility. However, the coefficients of the explanatory variables are much lower than the coefficients of the explanatory variables on January 15th. This would suggest that the relation might be stronger for January 15th. Besides that, the R-Squared for January 14th is much lower than for January 15th and the total abnormal returns, indicating that the specification of the model is a better fit for these abnormal return categories.
24 day of the event. This implies that firms with a relatively large percentage of foreign sales and foreign assets are impacted more negatively than firms that are less involved in international activities. The coefficient of the percentage of foreign assets is more negative than the percentage of sales, which may indicate that the relation is stronger for foreign assets and abnormal returns. By adding the variables foreign assets and the interaction term, the R-Squared increased from around 7% to 26%, indicating that this model is a better fit than the previous was. Furthermore, the coefficient of the interaction term is also positive, which means that if one of the variables increases, the impact of the other variable decreases. In this case, if foreign sales increases, the impact of foreign assets will decrease and the other way around.
In contradiction to the OLS with only foreign sales, this model is also significant in estimating abnormal returns on the day after the event. The variables are less significant compared to January 15th and the total abnormal returns, but still manage to explain almost 10% of the total variation. Both the impact of foreign sales and foreign assets appears to be lower than for January 15th, and again the impact of foreign assets is more negative than foreign sales. If we compare the value of the coefficient of the interaction term to January 15th, the interaction effect is stronger for abnormal returns measured on January 16th and significance is the same.
The regression with total abnormal returns shows a negative relation for foreign sales and foreign assets as well. Again, the impact of foreign assets appears to be larger than the impact of foreign sales. The R-Squared for the regression with total abnormal returns increased to 25%, when adding the extra variables, implying an improvement of the specification of the model. Again the interaction term is positive, indicating a diminishing impact of one variable if the other increases.
To conclude, adding the extra variables, foreign assets and the interaction term, improved the specification of the model and makes this a more suitable model to estimate abnormal returns compared to the regression with only foreign sales. Both the percentage of foreign sales and foreign assets appear to have a negative impact on abnormal returns as a result of a large, unexpected appreciation of the home-currency, which corresponds with the expectations resulting from the literature review.
5.1.5 Combing the industry-level analysis with the firm-level analysis
25 values of the percentage of foreign assets across industries is lower, meaning the lowest and largest value are much closer together than for foreign sales.
Table 6
Observations for each of the industries
The table displays the mean values of the explanatory variables per industry. Mean total abnormal returns are also reported, which is the sum of the abnormal return of January 15th and January 16th.
Mean Values
Industry
Foreign Sales Foreign Assets
Total Abnormal Return Basic Materials 72% 41% -11% Industrials 72% 28% -12% Consumer Goods 69% 34% -11% Health Care 78% 26% -13% Consumer Services 40% 29% -7% Utilities 50% 23% -6% Financials 51% 27% -7% Technology 76% 26% -12% Average 66% 28% -10%
5.1.6 A closer look at the health care and financial industry
Because health care was the industry with the largest total abnormal return we decided to take a closer look at this industry. We also wanted to zoom in on the financial industry because this is by far the largest category, containing 73 firms, and the total abnormal return was relatively low compared to other industries.
26 pharmaceuticals in Switzerland but the whole health care sector experienced a large impact. Another interesting industry in Switzerland is the financial industry. This industry is very large and contains a broad range of firms, which can be divided into banks, equity investment instruments, financial services, life insurance, non-life insurance and real estate investments & services. In the table below the abnormal returns of these separate sectors are displayed. We see that there is a very large dispersion of mean abnormal returns within the financial industry. While life insurance has a very large total abnormal return, real estate investments and services experience much less impact. These observations correspond to the explanatory variables, with life insurance having much higher percentages than real estate & investments.
Table 7
Mean Abnormal Returns of Health Care and Financial Sectors
The table displays the mean abnormal returns per date for each Health Care and Financials sectors calculated with the MSCI Europe index excluding Switzerland. The total abnormal returns (sum of January 15th and January 16th) are displayed per sector as well. The table also shows the values of the variables foreign sales and foreign assets for each of the Health Care and Financials sectors.
Mean Abnormal Returns Explanatory Variables
Industry January 14th January 15th January 16th Total Foreign Sales Foreign Assets Health Care: -0.78% -8.15% -5.07% -13.22% 77.52% 26.41% - Pharmaceuticals and Biotechnology -0.34% -10.82% -4.77% -15.59% 72.21% 28.33% - Health Services and
Equipment -0.56% -9.48% -4.92% -14.40% 84.36% 23.34% Financials: -0.14% -4.13% -2.94% -7.07% 51.47% 26.87% - Banks 0.12% -3.57% -3.76% -7.33% 32.95% 22.49% - Equity Investment Instruments -2.28% -3.98% -0.18% -4.16% 170.94% 79.20% - Financial Services 0.47% -6.72% -4.31% -11.03% 41.48% 28.78% - Life Insurance 0.40% -8.78% -6.77% -15.55% 40.31% 2.48% - Non-Life Insurance -0.21% -6.00% -5.25% -11.25% 45.60% 19.40% - Real Estate Investments and Services 0.26% -1.37% -1.10% -2.47% 34.75% 13.77%
27 and activities. The bar chart below provides a graphical overview of the total abnormal returns (sum of January 15th and January 16th) of the banking sector. Clearly, there are very large differences in impact within the banking sector. We see that the large banks like UBS Group, Credit Suisse, EFG International and Julius Bar Gruppe, were much more impacted than the other banks. Possibly because these banks are much more involved in international activities than the more local banks. Hence, the large banks indeed experienced very negative impacts, but because the banking sector of Switzerland exists mostly of smaller banks that were not impacted as much, this does not show in the average abnormal return of the banking sector as a whole.
Figure 2
Bar Chart of Abnormal Returns in Swiss Banking Sector
This bar chart provides an overview of the banking sector in Switzerland and the impact of the change in exchange rate regime, measured by the sum of abnormal returns of January 15th and January 15th.
5.3 Long-term impacts of exchange rate change
This section provides the results on the long-term impact of the exchange rate change. To illustrate the relation between the exchange rate and the market, a graph was composed using daily data on the Swiss Performance Index (SPI) and the exchange rate of the Swiss franc with the euro. From the graph below it seems clear that there is a relation between the exchange rate and the market. However,
-30,00% -25,00% -20,00% -15,00% -10,00% -5,00% 0,00% B an k C o o p B an k L in th B an q u e C an to n d e Gen èv e B an q u e C an to n VE B an q u e C an to n ale d u J u ra B asellan d sch af tlich e KB B asler KB B er n er KB C em b ra Mo n ey B an k C red it S u is se B an k E FG I n ter n ati o n al Gr au b KB Hy p o th ek ar b an k L en zb u rg Ju liu s B ar Gr u p p e L u ze rn er KB Sch weiz er is ch e Nat . B an k ST Gall er KB UB S Gr o u p Vali an t Vo n to b el Ho ld in g W allis er KB Z u g er KB
28 it is more interesting to study whether the relation between the exchange rate and the market changed after January 15th.
Graph 2
The Swiss Market and the Exchange Rate
The graph shows the values of the Swiss Performance Index (SPI) and the exchange rate of the Swiss franc and the euro (CHF/EUR) over time. The x-axis consists of the dates, which is from the time period of December first 2014 to February 23th 2015. The y-axis on the right displays the value of one euro in Swiss francs, while the y-axis on the right displays the value of the SPI.
29 Table 7
Test for a Structural Break
The table below shows the results of the Chow-test. The Chow-test was conducted both for the SMI and the SPI, with explanatory variables being the exchange rate of the Swiss franc with the euro and the exchange rate of the Swiss franc with the US Dollar. The F-value is the test statistic that indicates whether or not the coefficients of the two exchange rates are equal comparing the period before the event (January 15th) with the period after the event.
F-Value
SMI 110.06***
SPI 113.7***
***,** and * denote the significance level at the 1, 5 and 10 percent levels, respectively, for the reported coefficients.
The results in the table indicate that there is evidence that the coefficients of the two exchange rates in the period before the event of the exchange rate change are not equal to the coefficients after the event. To conclude, there is evidence that the null-hypothesis, of no change in impact of the exchange rate on the market after the event should be rejected. This result is in accordance with most of the literature written on this subject, for example on the Bretton-Woods system.
6. Concluding Remarks
After the decision of the SNB to discontinue the floor on the Swiss franc, the market collapsed. The purpose of this study was to create a general understanding of the impact of the change in exchange rate regime and variables that influence this relation. The study was divided into a short-term research part and a long-term research part. The short-term research part was mainly focussed on investigating whether industries and firms were impacted differently and if we could find any underlying variables that influenced the abnormal returns. The long-term research part was focussed on whether the impact of exchange rates on the market in general changed after January 15th. These concluding remarks will provide a summary of the results, the implications of the research and limitations combined with ideas for future research.
6.1 Summary of results
30 explanatory variables (percentage of foreign sales, percentage of foreign assets and an interaction term) and two control variables. Results indicate that both foreign sales and foreign assets have a negative relationship with the abnormal returns of Swiss listed firms as a result of the appreciation of the home-currency. Hence, a more internationalized company would experience a larger impact from a sudden change in exchange rate than less internationalized companies. Combining the industry-level results with the firm-level results we saw that the industries with the largest negative abnormal returns, were indeed the most exporting industries. The long-term research was focused on whether the impact of the exchange rate changed, comparing a period of 10.5 months before the event with a period of 10.5 months after the event. After conducting a Chow-test, we concluded that the coefficients of the exchange rates (EUR/CHF and USD/CHF) were significantly different comparing the period before and after the event. It appears that due to the exchange rate regime, the impact of the exchange rate on the market has changed. This result is in accordance with most of the literature described.
6.2 Implications
This research has implications for both policy makers and firms. One of the implications for policy makers is that they should keep in mind that adjusting the exchange rate regime can have a large impact on the economy. This depends partly on whether the economy of the specific country is very much dependent on international ties, for example by having a high export level. Also, the impact for Switzerland was huge because the change was not expected at all. Policy makers should consider whether they want to include such a surprise-effect. Without the surprise effect, Swiss firms could have been able to hedge their positions to decrease the impact or perhaps the impact would have occurred more gradually instead of one large shock.
This research also has implications for firms. For example, firms that operate in a country with a strong currency maintaining a fixed exchange rate regime should be aware that they are not completely without exchange rate risk. Even though it is not the most likely case, policy makers can always decide to switch to a floating exchange rate regime. Firms should be aware of the consequences of such a decision. In the case of Switzerland this is exactly what happened. Almost nobody predicted that the SNB would change their policy, which is why the impact was very large.
6.1 Limitations and Future Research
31 day, which is January 14th. Further research is necessary to compare the impact of foreign assets and foreign sales on regular trading with the impact on days where there is a large exchange rate change.
Further research on this topic could also focus on adding more variables to the model. For example, the variable foreign liabilities could be included in the regression to see if this would have an opposing effect (compared to foreign sales and foreign assets). The variable foreign liabilities was not included in this research because it was very difficult if not impossible to obtain data on this variable for the firms in the sample.
Another possible avenue for future research could be doing the same sort of research for a different country because similar outcomes would strengthen the results of this study. Switzerland is quite a unique country which is why it would be interesting to see if the impact is the same for countries with other characteristics. Also, it would be interesting to see what happens in case of a large unexpected depreciation (instead of appreciation). Specifically if this would lead to an impact opposed to the impact in Switzerland because it might also be the case that it does not matter in what direction the currency moves, but that the shock solely would be enough to cause a large negative impact. Lastly, this study is based on quantitative data. To fully comprehend the impact of a change in exchange rate regime, a qualitative research could be helpful. For example, in the case of Switzerland, some firms decided that the employees had to work half an hour longer per day for free21. For those firms this was a measure to cope with the negative impact of the currency shock. This example is just to illustrate that the annual report and the stock market do not provide a complete picture of the situation. Numbers might not fully capture extraordinary measures like the example, which is why a qualitative research would be interesting.
References
Adler, M., & Dumas, B. (1980). The exposure of long-term foreign currency bonds.Journal of Financial and Quantitative Analysis,15(04), 973-994.
Adler, M., & Dumas, B. (1984). Exposure to currency risk: definition and measurement.Financial management, 41-50.
