MSc. International Business and Management
Specialization International Financial Management
The impact of the introduction of the
Euro on foreign exchange rate risk
exposures
- A decade later -
By
Cristina Toma S1711725
January 2010
2
The impact of the introduction of the Euro on foreign
exchange rate risk exposures
- A decade later -
Abstract
The uncertainty of dealing with several European currencies caused various risks that translated into costs for businesses worldwide. A decade after the introduction of the Euro, quantifying its alleged impact on foreign exchange rate exposure is appropriate. By using an extensive dataset of 780 nonfinancial companies from 20 European and Non-European countries, this study analyzes whether the adoption of the Euro triggered significant changes in stock return volatility, market risk and foreign exchange rate risk exposures. This study documents reductions in market risk for companies within and outside Europe, in spite of increased total stock return volatility. Nonfinancial companies with foreign sales in Europe exhibit higher decreases in market risk exposure than companies with no foreign sales in the same geographic area. It also indicates that the advent of the single European currency led to a decrease in foreign exchange rate risk exposure for companies with foreign sales in Europe. In general, this study contributes significantly to the exchange rate risk management literature by adding valuable insights.
JEL classification: F3, F4, G3
3
Table of Contents
1. Introduction ... 5
2. Literature review ... 9
2.1. Exchange rate exposure ... 9
2.2. Model specification ... 11
2.3. Value or equally weighted market return ... 11
2.4. Single-currency versus trade-weighted index ... 12
2.5. Cash Flows ... 14
2.6. Nominal versus real exchange rates ... 14
2.7. Firm-level versus portfolio analysis ... 14
2.8. Choice of time-horizon ... 15
2.9. Temporal instability – the use of sub-periods ... 16
2.10. Economic openness ... 16
2.11. Firm strategy... 17
2.12. Industry ... 17
2.13. Reasons for lack of empirical support ... 19
2.14. Hedging ... 19
2.15. Studies on the Euro... 22
2.16. Concluding remarks ... 23
3. Research Design ... 27
3.1. Sample Construction ... 27
3.2. Hypothesis Construction ... 29
4. Data and descriptive analysis ... 32
4.1. Dataset construction and methodological issues ... 32
4.2. Descriptive statistics ... 34
5. Results ... 38
5.1. Changes in stock return volatility ... 38
5.2. Changes in market risk ... 41
5.3. Changes in foreign exchange rate exposure ... 44
6. Discussion ... 49
7. Conclusion ... 53
References ... 55
4
List of Tables
Table 1: Overview of exchange rate exposure as documented in the empirical literature ... 25
Table 2: Distribution of companies in the multinational and control groups within the raw sample and final sample. ... 33
Table 3: Descriptive statistics of firms in test and control samples ... 35
Table 4: Descriptive statistics of stock market indices ... 37
Table 5: Analysis of stock return variances ... 39
Table 6: Regressions of stock market returns on market indices ... 42
Table 7: Regressions of stock returns on market indices and foreign exchange rates ... 46
5
1. Introduction
The volatility of exchange rates and its associated risks has been a popular subject in international financial management since the breakdown of the Bretton Woods fixed-parity system at the beginning of the 1970s. The increased currency volatility of the last few decades triggered the empirical interest in the potential vulnerability of multinational companies to unexpected changes in exchange rates (Allayannis & Ofek, 2001; Bartov & Bodnar, 1994; Doidge et al., 2006; He & Ng, 1998; Nguyen et al., 2006; Bartram & Bodnar, 2007). From a theoretical perspective, the value of the firm should be affected by exchange rate fluctuations (Shapiro, 1975). In contrast to expectations, empirical research often results in mixed and conflicting findings.
This study focuses on the exposure puzzle by analyzing the potential impact of the introduction of the Euro. More specific, it performs a firm-specific analysis of changes in overall stock return volatility in connection to market risk and foreign exchange rate risk and how it differs depending on the extent of economic activity the sample firms incur within Europe. The substantial body of research on currency volatility combined with the rise of the Euro as one of the most influential currencies, indicate the necessity of knowledge in multinational companies all over the world.
6
increased volatility in global currency markets, would be absorbed without job losses or lower growth.
As an outcome of the recent economic crisis, foreign exchange volatility has reached historical levels. With it, the Euro has received new Europe-wide popularity since almost all countries which are not members of the Eurozone have recently increased their pace for Euro adoption. UK, Sweden and Denmark, which have opted not to sign up for the common currency, are now rethinking their position with respect to the Euro. The same holds true for the new EU Member States which are reported to have suffered the most from the recent financial turmoil compared to the Eurozone Member States. What is common about these countries is that they consider the Euro as potential protection in a time of crisis because adopting a single currency ultimately leads to the elimination of foreign exchange risk within intra-zone trade and investment.
7
Keeping in mind the small amount of research on Euro exposure, the different shortcomings of the previous studies and the necessity of knowledge regarding the potential currency stabilization of nonfinancial companies worldwide, it is important to study the impact of the introduction of the Euro on the exposure of nonfinancial firms, 10 years after the adoption of the common European currency. The object of this thesis is thus to test whether significant changes in stock return volatility, market risk and foreign exchange rate risk exposures occurred prior and after the introduction of the Euro. The empirical results have important policy implications for nonfinancial firms as evidence of low exposures can allow them to take higher business risk or increased levels of debt. Similarly, a direct effect of reduced market risk is a decrease in the cost of capital which would benefit firm valuations and corporate investments. This extensive study contributes to the existing literature in a number of ways. It uses detailed data on a data set of 780 firms from 18 European countries, the United States and Japan. Furthermore, the analysis spans over a period of 19 years divided into two almost equal periods which have the benefit of incorporating any transition phase that might influence the results both before and after the event. Finally, due to a detailed comparison and analysis of nonfinancial firms with foreign sales in Europe and without foreign sales in Europe, matched by firm size and geographic location, this study contributes significantly to understanding the link between changes in currency regimes and foreign exchange rate exposures. The main Research Question of this study is:
“What is the impact of the introduction of the Euro on stock return volatility, market risk and foreign exchange rate risk exposures of nonfinancial multinational companies?”
8
9
2. Literature review
In this section, the findings of earlier studies regarding exchange rate exposure are discussed. In order to analyze the impact of the introduction of the Euro on foreign exchange rate risk exposures of companies, it is important to identify the issues which have been analyzed within the field in question and the constraints which have to be overcome in the construction of an accurate research design. This section formulates the basis for the second part of the study by providing an extensive overview of the empirical research conducted so far in the field of currency risk in general but also by giving an overview of the studies which have been performed on the introduction of the Euro. Consequently, this will highlight once more the necessity of studying the phenomena in great detail 10 years after the actual event date. This section will close with a table summarizing the exchange rate exposure results as documented by the empirical literature.
2.1. Exchange rate exposure
10
According to Prasad and Rajan (1995) and Bodnar and Gentry (1993), the movements in exchange rates may impact on firm value in several ways. Firstly, they can affect the future cash flows of companies by changing the firm’s economic and competitive conditions in the market for its products or factors of production. Secondly, they may change the domestic currency value of cash flows from foreign operations. Lastly, they may result in translation gains or losses when converting the assets or liabilities of a foreign subsidiary into the domestic currency. Their research shows that export-oriented industries gain from a depreciation in home currency while import-oriented industries gain from an appreciation of the home currency.
In contrast to theoretical beliefs, the empirical findings show conflicting results. Amihud (1994) finds no significant exchange rate exposure for a sample of 32 large US exporters for the period 1982-1988. According to other empirical articles, there is either little or no exposure of individual firms or industries towards currency volatility (Jorion, 1990; Amihud, 1994; Miller & Reuer, 1998; Muller & Verschoor, 2008; Doidge et al., 2006; Dominguez & Tesar, 2006). Choi and Prasad (1995) examine a dataset of 409 US multinational firms for the 1978-1989 period. They document that only 15% of firms have significant exchange rate exposure and that a higher percentage of firms with significant sensitivity gain from a depreciation of the US dollar.
11
Williamson (2001) argues that in order to evaluate the effect of an exchange rate shock on firm value, the shocks that are permanent and unanticipated must be identified. Therefore, what is really measured is the net exposure to exchange rates, namely the exposure left after the firm has engaged in some form of hedging activity, by using either derivatives or its own operations.
2.2. Model specification
Since the breakdown of the Bretton Woods fixed exchange rate regime in 1973, the international economic environment has been characterized by high exchange rate volatility. Surprisingly, it took eleven years for scholars to develop a model to measure exchange rate exposure. The pioneers of this model are Adler and Dumas (1984) whose article is most often the foundation of almost all the other articles written on the subject. The linear regression they have developed relates the stock return of a company to the change of the exchange rate.
In 1990, Jorion developed this model further, by incorporating the market return. He argues that macroeconomic variables may impact stock returns and exchange rates at the same time. Consequently, the exposure coefficient in Adler and Dumas’s (1984) model incorporates not only the exposure of a company towards currency fluctuations but also the impact of macroeconomic variables such as investor sentiment, the market risk premium or the risk-free rate (Bodnar & Wong, 2003). According to Bodnar and Wong (2000), there are two main reasons behind the inclusion of the market portfolio return variable. Firstly, it controls for macroeconomic effects that are correlated with the exchange rate over the estimated time horizon. Secondly, the market return explains most of the volatility of a company’s stock return and its inclusion reduces the residual variance of the regression and thus, improves the precision of the residual exposure estimates. Therefore, the percentage of firms with statistically significant residual exposure estimates slightly increase after the inclusion of the market return. This phenomena is explained by the fact that other macroeconomic variables might co-vary with exchange rate fluctuations and stock returns, and controlling for these co-movements refines the number of estimates attributable to foreign currency movements.
2.3. Value or equally weighted market return
12
There are however, several variations of this model, especially in the choice of the exchange rate variable, be it a single currency or an exchange rate index, or a value- or equally-weighted market return (Bodnar & Wong, 2003). The value-weighted market return used by Jorion (1990) induces a positive bias in exposure coefficients because large firms are over-represented and thus are given more importance. Similarly, an equally-weighted market return gives more importance to smaller firms. The core of the discussion lies in the fact that larger firms are usually export oriented while smaller firms are import oriented (Bodnar & Wong, 2003) and the choice of the market risk factor might augment the cash flow exposure of the two types of firms, thus biasing the results.
2.4. Single-currency versus trade-weighted index
When specifying the exchange rate variable, a choice has to be made between a trade-weighted exchange rate index and one single currency. The proponents of a single-currency method underline the importance of one country acting as a trading partner, and the dominance of mainly one currency over the value of the firms included in the sample (Glaum et al., 2000; Williamson, 2001). Dominguez and Tesar (2006) compare these choices and their results show that the trade weighted index has a slightly higher exposure than the US dollar or the currency of the main trading partner. However, Miller and Reuer (1998) consider that a trade-weighted index disregards the problem of low and negative correlations among exchange rates over time. Thus, such a model may underestimate exposure. Similarly, Williamson (2001) states that the use of such an index lacks power if a firm is exposed to only a few of the currencies within the basket.
There are also proponents of creating firm-specific exchange rates. For instance, Makar and Huffman (2008) estimate the individual exposure of 44 large UK multinationals to the main currency to which they are exposed as opposed to other studies that analyze exposure either to a broad exchange rate index or on the geographic location of subsidiaries. They show that incorporating firm-specific currencies leads to an increase in the number of companies with significant exposures in contrast with the exposure estimates determined by using a broad exchange rate index.
13
yen, Canadian dollar and Mexican peso. Including these currencies in the model yields a higher proportion of exposed US firms than reported in previous research. Consequently, the empirical evidence from US manufacturing firms indicates that 13 to 17 percent of firms are exposed to currency volatility. However, the inclusion of several currencies in the model may lead to a correlation problem, as experienced by Miller and Reuer (1998).
Chow, Lee and Solt (1997) analyzed the exposure of 213 US multinationals towards an exchange rate index computed as a weighted average of six bilateral real exchange rates, defined as US dollars per unit of foreign currency for the British pound, Canadian dollar, French franc, German deutsche mark, Italian lire and Japanese yen. Muller and Verschoor (2006a) look at the exposure of 817 European firms towards the Euro’s bilateral continuously compounded exchange rate with the US dollar, the Japanese yen and the UK pound (i.e. Euro/ US dollar, Euro/Japanese yen and Euro/ UK pound).
14
2.5. Cash Flows
There are also other variations of the Adler and Dumas (1984) and Jorion (1990) model specification. For instance, Chow and Chen (1998) include business condition variables such as the prevailing dividend yield and term premium in the Adler and Dumas (1984) model. Bartram (2008) examines the impact of exchange rate changes on corporate cash flows with no significantly different results as compared to the stock return approach. However, linking currency volatility to cash flows is considered to be past-oriented as compared to stock returns which are forward-looking (Muller & Verschoor, 2006c). Chow et al. (1997) argues that unless measured as cash flows net of hedging, the cash flow exposure will show insignificant exposure results due to effective hedging.
2.6. Nominal versus real exchange rates
In the specification of the currency risk factor several choices have to be made. One of them is the choice between nominal and real exchange rates. Most studies use nominal returns for several reasons. Firstly, if the exchange rate fluctuations are measured in real terms, the equation has to be consistent and thus all the variables have to be also measured in real terms (Muller & Verschoor, 2006c). Secondly, since the volatility of inflation rates is low, financial markets do not observe inflation instantaneously and thus, most of the changes in nominal exchange rates also translate into changes in real exchange rates (Glaum et al., 2000; Bodnar & Gentry, 1993). As a result, it is widely agreed that the choice between real versus nominal exchange rates has a negligible effect on estimating exchange rate exposure (Amihud, 1994; Bodnar & Gentry, 1993; Choi & Prasad, 1995; Griffin & Stulz, 2001).
2.7. Firm-level versus portfolio analysis
15
that comprised of companies with no international sales during periods of currency depreciation. However, the constraint of this approach is that within a portfolio, some companies with international sales have offsetting exposures if they also have operations abroad, while firms with no international sales may face significant exposures if they have more foreign competitors. Choi and Prasad (1995) also argue that by aggregating firm data into portfolios, the resultant loss of information represents the main reason why studies fail to support the theoretical inference that exchange rate volatility has an impact on firm value.
2.8. Choice of time-horizon
One other important issue in analyzing exposure is the time horizon used. A vast portion of the existing literature focuses on monthly returns (He & Ng, 1998; Jorion, 1990; Amihud, 1994; Miller & Reuer, 1998; Makar & Huffman, 2001; Friberg & Nydahl, 1999). However, recent empirical studies have also considered the use of weekly returns (Dominguez & Tesar, 2006; Muller & Verschoor, 2006a; Muller & Verschoor, 2007a; Griffin & Stulz, 2001; Faff & Marshall, 2005) and even annual returns (Makar et al., 1999). Chamberlain et al. (1997) promotes the use of daily data in order to estimate the exposure of US and Japanese banking institutions to currency volatility.
16
Di Iorio and Faff (2001) and Glaum et. al (2000) show that less exchange rate exposure is detected when using monthly data as compared to daily data. In contrast, Bodnar and Wong (2003) argue that exposure is evident in longer time horizons as it takes time for information to be incorporated into stock returns, while the noise in high-frequency data can be avoided by using data gathered over longer time-horizons. Long-horizon regressions capture the long swings that currencies experience and are able to better incorporate the long-term relationship between exchange rates and firm value (Muller & Verschoor, 2006c).
2.9. Temporal instability – the use of sub-periods
Most of the literature on currency risk also takes into account the question arising from the potential temporal instability of a company’s exchange risk exposure. Jorion (1990) assumes in his model that exposure is constant through time. However, Glaum et al. (2000) stresses the fact that it is unrealistic to assume that a firm’s exchange rate exposure remains constant over a long period of time. The same line of reasoning is employed by Bartov and Bodnar (1994) who underline the fact that exposure is bound to change over the sample period as long as there are changes in the overall economic environment, the firm’s competitive position, its operational structure and its hedging policy. Therefore, Glaum et al. (2000) divide their 24-years time series into four equal sub-periods. The same is employed by other researchers as they test for constant exchange rate exposure over different sub-periods (Amihud, 1994; Choi & Prasad, 1995; He & Ng, 1998). In his analysis of the automotive industry from the US and Japan, Williamson (2001) finds that for each separate sub-period the exposure is linked to the competitive environment of the sector.
2.10. Economic openness
17
which European firms are exposed to fluctuations in foreign exchange rates differ across countries. As such, multinationals based in France, Germany, the Netherlands and Spain appear to be significantly exposed to currency risk as compared to the other European Union Member States. Multinationals based in Ireland and Luxembourg experience no significant exchange rate exposure.
Donnelly and Sheehy (1996) study 68 UK large exporting firms during the period 1980-1992 and their results indicate that there is a relationship between the foreign exchange rate and the market value of large exporters. They attribute this difference to the fact that the UK is a more open economy than the US and that the companies in the sample are more export-intensive than the US firms previously researched. Moreover, they consider that US companies have an advantage in avoiding currency risk since the dollar makes it easier for them to denominate their sales in terms of their domestic currency. Consequently, only US exporters with price-elastic exports are affected by currency shocks.
2.11. Firm strategy
According to Muller and Verschoor (2006b), a multinational’s reaction to exchange rate fluctuations depends on the strategy of the firm. If an exporting firm pursues a market-share objective, it will prefer to maintain its export price in its currency and allow the export price in the foreign currency to fall. Therefore, the firm can gain market share when its currency depreciates. Similarly, an exporting firm may decide to allow its markup to absorb the effect of small changes in currency rates, while large exchange rate fluctuations may force the exporter to pass-through part of the currency change in export prices.
2.12. Industry
18
and Japan, out of which 208 represent non-traded goods industries and 93 represent traded goods. Their findings show that exchange rate shocks have an insignificant impact of the value of the industries across the world. One probable explanation could be that exchange rate changes have a significant impact on profit margins but at the same time they have little or no effect on shareholder wealth. This can be possible if other sources of variation in stock prices are much more important than exchange rate shocks or if the stock market is efficiently incorporating the impact of exchange rate shocks on stock prices but these shocks are not significant for shareholders. Griffin and Stulz (2001) also show that common industry effects dominate competitive effects by finding a positive relationship between industries from different countries. Their conclusions hold across different regression specifications, exchange rate benchmarks, measurement intervals and sub-periods. For instance, with weekly returns only 2.4% of the variation in stock returns is explained, while annual returns explain only 1.5% of the variation.
Since empirical results of the effect of currency volatility on firm value have not been strong or consistent to theoretical expectations, Williamson (2001) raises the question regarding the failure of the existing test to capture the effects of such volatility on firm value. Hence, he suggests that testing one specific industry should allow for better analyses regarding exchange rate changes. Consequently, he studies the exposure of firms from the automotive industry from the US and Japan and finds that exposure varies across firms from different countries.
19
2.13. Reasons for lack of empirical support
According to Bartov and Bodnar (1994) investors do not have sufficient information in order to react timely to a firm’s currency exposure and thus, there is a delayed market response to exchange rate volatility. Consequently, if investors react incorrectly to exchange rate movements, systematic errors are introduced. Hence, they suggest that lagged changes in exchange rates are introduced in the model specification. Similarly, Allayannis and Ofek (2001) conclude that firms actively manage their exchange rate risk by using foreign currency derivatives and other hedging techniques. The lagged relationship between currency fluctuations and firm value has also led to conflicting results. Nydahl (1999) show that 26% of their Swedish sample are significantly exposed to exchange rate changes. In contrast, He and Ng (1998) find that less than 4% of the Japanese companies in their sample experience a significant lagged response to currency changes. Actually they conclude that including lagged currency changes has no significant impact on the explanation power of the model.
On the other hand, Ihrig (2001) and Dewenter et al (2005) associates the puzzling lack of support to foreign exchange exposure to misspecifications in the estimation model or the correct identification of the currencies to which a company is vulnerable. The identification of possible asymmetries in the impact of currency changes, the difference between temporary and permanent currency movements and the impact of the exchange rate shocks on a company’s competitive and economic environment represent a difficult task for shareholders (Bartov & Bodnar, 1994).
2.14. Hedging
20
Empirical findings show that both operational and financial hedges are associated with lower exchange rate risk exposure (Kim et al., 2005). Exchange rate exposure is reduced by using currency derivatives (Nydahl, 1999). Makar et al. (1999) argue that there is a significant and positive relationship between the use of foreign exchange derivatives and the level of foreign exchange exposure and geographic diversification natural hedges.
Makar and Huffman (2008) show that their sample of UK multinationals make effective use of financial currency-hedge techniques in order to diminish the risk associated with the volatility of the bilateral exchange rates to which they are most exposed. They base their study on publicly available UK accounting disclosures. In contrast, Hagelin and Pramborg (2004) use a financial hedge indicator variable which takes the value 1 if derivatives or foreign denominated debt is used. They base their indicator on a questionnaire documenting the effective hedging of Swedish multinationals to a broad exchange rate index. Similarly, De Jong et al. (2006) retrieve data regarding the use of foreign-denominated debt and other hedges from a questionnaire sent to Dutch companies. Pantzalis et al. (2001) conclude that operational hedges determine the exposure estimates of US multinationals.
Data on hedging activities are usually incomplete or difficult to obtain. Thus, several studies use variables that proxy for a company’s incentives to hedge (He & Ng, 1998; Chow & Chen, 1998; Makar & Huffman, 2008). Firms are prone to hedge especially if the benefits of such activities are higher than the costs. The variables usually used are the size of the company, the dividend payout ratio, the quick ratio, the book value per share and the debt ratio. European firms with low dividend payout ratios or short-term liquidity positions are less inclined to hedge and thus, more exposed to exchange rate fluctuations (Muller & Verschoor, 2006a).
21
Allayannis and Ofek (2001) examine the link estimated exposure coefficients with data on foreign derivate usage and find that firms use foreign derivative instruments for hedging and not for speculative purposes.
Foreign sales are considered to be the main determinant of exchange rate exposure. However, the literature shows conflicting evidence. For instance, Chow, Lee, Solt (1997) and Makar and Huffman (2008) prove that the magnitude of exposure is unrelated to foreign sales. Nydahl (1999) finds a positive, significant relation between the estimated exposure and the ratio of foreign sales to total assets. Williamson (2001) argues that foreign sales increase exposure while foreign operations reduce exposure. Consequently, it is generally assumed that companies employ a type of natural hedge by matching their foreign sales and foreign assets and thus, minimize their currency exposure.
He and Ng (1998), Choi and Kim (2003) and Muller and Verschoor (2006a) argue that high leveraged firms and companies with weak short-term liquidity positions have more incentives to hedge and thus are less exposed to foreign currency fluctuations. This results are backed up by the study of Muller and Verschoor (2007a) who analyze the Asian foreign exchange risk exposure and find that consistent with optimal hedging theories, Asian firms with strong liquidity positions in terms of low dividend payout ratios or more profitable firms, have less incentives to hedge and thus are more exposed to foreign exchange volatility. However, in contrast to findings on US multinationals, Asian companies with high leverage ratios or companies with a lower quick ratio have higher exposure to exchange rate risk.
22
diversification while smaller firms have more incentives to hedge and thus should be less exposed to currency risk (Pantzalis et al., 2001).
One recent finding with regard to hedging is that of Bartram et al. (2010) who look at the annual reports of 1150 firms from 16 countries. They argue that firms combine three instruments in order to mitigate exchange rate risk. Therefore, firms can either, pass through part of their currency risk to customers, match their foreign costs and revenues, use financial products such as foreign currency denominated debt and foreign derivative, or a combination of them. Accordingly, pass-through reduces exposures by about 10-15%, operational hedging reduces exposure also by 10-15%, while financial risk management leads to a 40% reduction in exposure. Consequently, firms have the capacity to reduce their exchange rate exposure by up to 70% if all three hedging mechanism are employed. Thus, Bartram et al. (2010) manage to give a relevant solution to the exposure puzzle documented in the literature.
2.15. Studies on the Euro
The Euro was launched on 1 January 1999 and became the official currency of 11 Member States after a long converge process beginning in 1979 with the creation of the European Monetary System. Hence, the economic convergence has influenced the macroeconomic structure of some European countries. Morana and Beltratti (2002) find clear signs of an increase in volatility around the introduction of the euro along with a stabilization of the European stock markets in the months following the introduction of the common currency. Nguyen et al. (2006) analyze the exposure of 99 French multinationals before and after the introduction of the euro during 1990-2001. Their findings show that the number of firms that are vulnerable to exchange rate fluctuations drops from 32% during the pre-Euro period to 11% after the introduction of the Euro. Additionally, French firms tend to decrease their use of foreign currency derivatives from 39.5% to 10% during the two sub-periods.
23
risk is part of systematic risk. In the last stage, the researchers focus on the measurement of incremental foreign exchange rate exposures after accounting for market risk. Their findings show that the incremental exchange rate exposure is slightly reduced. Moreover, consistent with economic theory, firm characteristics such as total sales, foreign sales in general and in Europe, industry characteristics such as competition and trade, and regional factors, such as geography and the strength of the currency, determine the level of exposure. They find no support for the hypothesis that market risk is affected by changes in financial leverage.
The introduction of the Euro can be compared with the opposite phenomenon from 1973 when the breakdown of the Bretton Woods system allowed previously fixed exchange rates to float. Bartov et al. (1996) study this phenomenon using 109 US multinationals over two-equal periods spanning around the year 1973. In line with economic intuition and similar to Bartram and Karolyi (2006), the stock return volatility increased from 0.77 to 1.07 per month, along with a 10% increase in market risk.
Using a sample of 1154 European firms from 11 countries, Hutson and O’Driscoll (2009) estimate the firm-level exchange exposure during the pre-euro period 1990-1998 as compared to the post-euro period 1999-2008. Contrary to expectations, they find that in the post-euro period, multinationals based in the Euro-area have higher currency exposure than multinationals outside the Euro-area. They explain this by the fact that the non-Eurozone countries in their sample have corporate governance systems that give better rights to creditors than those located in the Eurozone countries. Therefore, these companies face high bankruptcy costs and managers are more likely to employ activities that reduce risk and thus diminish their company’s exposure. Moreover, their findings show that exposure has increased as an outcome of the introduction of the Euro. They argue that the increase in exposure is due to increasing competition brought about by the intensification of trade and international investment. As they look at firm-level characteristics that might explain the results, Hutson and O’Driscoll (2009) find no difference between the firm-specific exposure of Eurozone and non-Eurozone companies.
2.16. Concluding remarks
24
literature it is possible to see that the majority of empirical studies dealing with exchange rate exposure have contradictory findings (e.g. Amihud, 1994; Dominguez and Tesar, 2006; Muller and Verschoor, 2008; Makar & Huffman, 2008). Even though a large number of studies document low or insignificant exchange rate exposures, the results vary widely with the model specification employed and dataset selected (Muller & Verschoor, 2006c, Dominguez & Tesar, 2006). Thus, there is an ongoing debate regarding the suitability of the sample collection and analysis structure that has to be employed in order to lead to significant findings.
27
3. Research Design
The previous section indicates that there are several differences in the research design which have to be considered before carrying out the investigation. The research design and data collection are of outmost importance in conducting an insightful study in the field of currency risk. The first subsection presents the methodology used in constructing the sample. Subsequently, a number of hypotheses are formulated and will mark the three stages of the research. Testing the hypotheses will shed light in the expected differences in exposure of multinational companies with foreign sales in the Euro area against those with no foreign sales in the Euro zone.
3.1. Sample Construction
As indicated in the introduction, the analysis will be done on a regional basis. The firms are grouped according to their geographic location, firm size and their degree of multinationality, more specifically, the extent to which they carry foreign business activities in Europe no matter their country of domicile. It is expected to find that following the launch of the Euro, nonfinancial firms experience reductions in foreign exchange rate exposures, and that these reductions hold especially for firms which are located within the Euro-zone countries and companies domiciled outside the Euro area which have foreign sales in the Euro area. This assumption also holds for the companies located outside the Euro area because the common European currency was designed to eliminate the exchange rate risk within Eurozone Member States and thus, its introduction cancels the risk of investment in the foreign European market. Consequently, it reduces the uncertainty related to foreign costs and earnings. The data set and descriptive findings will be discussed in section 4 of this study.
28
Chamberlain et al., 1997). In their study on the impact of the introduction of the Euro, Bartram and Karolyi (2006) use weekly data, while Hutson and O’Driscoll (2009) consider monthly returns. In order to be consistent and able to compare this study with the extensive study of Bartram and Karolyi (2006), the data frequency used in this paper is weekly.
Many studies document a positive relation between currency exposure and foreign sales of nonfinancial firms (Jorion, 1990; Choi & Prasad, 1995; He & Ng, 1998; Williamson, 2001; Dominguez & Tesar, 2006; De Jong et al., 2006; Bodnar & Wong, 2003). Consequently, it is important to select firms that have foreign sales in Europe. Following Bartram and Karolyi (2006), companies in the Telecom, Media and Technology sector are excluded from the sample in order to control for the extreme effects of the rise or fall of the internet sector stocks on these companies. Also, firms which are part of the financial service industries such as banks or insurance companies are excluded from the sample due to their complex financial risks. Consequently, the remaining companies matching the criteria listed above are referred to as “Multinational Firms”. These companies are then grouped into three geographic regions according to where they are domiciled. The first geographic region is the “Euro area” which includes companies located in the European countries which adopted the Euro on 31 December 1998: Austria, Belgium, Netherlands, Finland, France, Germany, Italy, Ireland, Luxembourg, Portugal and Spain. Greece is not included in the sample countries since the Euro membership for this country was determined on 19 June 2000. The second geographic region is the “Non-Euro “Non-Europe” which consists of companies domiciled in the following seven “Non-European countries: Denmark, Norway, Poland, Sweden, Switzerland, Turkey and the United Kingdom. The last geographic region will be referred to as “Outside Europe” and comprises firms from Japan and the United States. The choice of the total 20 countries is in line with the study of Bartram and Karolyi (2006), which will prove to be useful in comparing the results.
29
measured by market capitalization. These benchmark firms are referred to as “Size Control firms”. All the companies in the multinational sample and control sample which are not matched for various reasons2 are eliminated from the analysis.
3.2. Hypothesis Construction
Firstly, the introduction of the common European currency was designed to reduce currency shocks for those companies which have a percentage of their sales in Europe. These firms are more sensitive to unexpected foreign exchange rate changes since they conduct foreign business activities outside the country they are domiciled in. The introduction of the Euro is expected to affect multinationals in the three geographic regions (Euro-area, Non-Euro Europe and Outside Europe) in the same way because the common currency marks not only the elimination of exchange rate risk within the Eurozone but also increasing international trade and a more stable investment climate for companies with foreign activities within Europe. Even though the magnitude of the impact is expected to be different, the adoption of the common currency is expected to instill greater confidence and less risk for companies seeking foreign investments within the Eurozone. In order to quantify the increase or reduction in stock return volatility as a consequence of the introduction of the Euro and in line with Bartov et al. (1996) and Bartram and Karolyi (2006), the variances of stock returns of individual firms are calculated for the sample periods before and after the introduction of the Euro. A decrease or significantly lower increase in stock return variances of firms in the multinational sample as compared to companies in the size control sample would prove that firms which have foreign sales in Europe benefit from the introduction of the common currency (Bartram & Karolyi, 2006). Consequently, foreign exchange rate volatility is assumed to be positively related to the stock return volatility of multinational firms. As a result, the first testable hypothesis is formulated as:
H1: Nonfinancial firms that are domiciled in Euro area countries and/or those domiciled outside the Euro area that have foreign sales/assets in Euro area countries should exhibit a larger reduction in stock return volatility than firms with no foreign sales/assets in the Euro area.
2
30
In line with Bartram and Karolyi (2006), the Chi squared test for change in variance is used and is based on the null hypothesis that the variance of an individual firm does not change significantly. Moreover, Wilcoxon rank-sum tests, mean tests and median tests are performed across subperiods and samples to test the differences in variances and also variance ratios, thus comparing the control sample with the sample of multinationals.
Secondly, it is important to test whether the market risk of the companies in the sample has changed after the introduction of the Euro and thus, if there is any indication that the exchange rate risk is partially non-diversifiable. More specifically, a decrease in currency risk would be positively related to a reduction in market betas. According to Bartov et al. (1996), because multinational firms have substantial revenues and assets denominated in foreign currency, the volatility of exchange rates contributes to their non-diversifiable, systematic risk by increasing their market beta as compared to non-multinational or domestic firms. In line with Bartram and Karolyi (2006), the sample of multinational companies with foreign sales in Europe should experience a larger decrease in market risk exposure as compared to the size control firms. As a result, the second testable hypothesis is formulated as:
H2: Nonfinancial firms that are domiciled in Euro area countries and/or those domiciled outside the Euro area that have foreign sales in Euro area countries should exhibit a larger reduction in market risk exposures than firms with no foreign sales in the Euro area.
In order to test this hypothesis the stock market betas are estimated by using the Ordinary Least Squares estimation method (OLS) on equation (1):
(1)
where Rijt is the return of stock i in country j, RMjt is the return of the market portfolio in country
j, and DEurot is a dummy variable that takes the value 1 after 1/1/1999 and 0 otherwise. Standard
31
significantly positive and negative coefficients are reported according to a significance level of 1%, 5% and 10%.
Lastly, the traditional approach of Adler and Dumas (1984) of estimating the incremental foreign exchange rate exposures after accounting for market risk is employed. If foreign exchange rate risk is partially non-diversifiable and lower currency risk does not translate into smaller market betas, it is expected that lower incremental foreign exchange rate exposures for multinational firms within and outside the Euro area occur for firms with foreign sales in Europe (Bartram & Karolyi, 2006). As a result, the third testable hypothesis is formulated as:
H3: Nonfinancial firms that are domiciled in Euro area countries and/or those domiciled outside the Euro area that have foreign sales/assets in Euro area countries should exhibit a larger reduction in incremental foreign exchange rate risk exposures than firms with no foreign sales/assets in the Euro area.
In order to test this hypothesis, the following equation (2) will be estimated by using OLS:
(2)
where Rijt is the return of firm i in country j, RMjt is the return of the market portfolio in country j,
RFXjt is the percentage change of a trade-weighted exchange rate index for country j, and DEurot is
a dummy variable that takes the value 1 after 1/1/1999 and 0 otherwise. As in regression (1) standard errors are corrected for autocorrelation and heteroscedasticity with the Newey-West procedure. Results are reported by area, sample, and variable. Foreign exchange rate coefficients are reported separately for positive and negative foreign exchange rate betas and corresponding changes. For each variable, the median coefficient is reported. Furthermore, Wilcoxon rank-sum tests and Chi-squared tests of equal coefficients of firms in different quartiles are performed. In addition to these tests, the percentages of significantly positive and negative coefficients before the Euro (δij) are reported separately according to a significance level of 1%, 5% and 10%. This
32
4. Data and descriptive analysis
This section presents and discusses the dataset and descriptive statistics of this study. Firstly, the methodological issues which, at times, constrained the construction of the dataset are discussed. Secondly, the descriptive statistics of the firms in the test and control samples are presented depending on the geographical area where they are domiciled and according to their market capitalization, sales, total assets, foreign sales and foreign Europe sales. Similarly the average weekly returns and return variances of stock market indices and their implications are also discussed.
4.1. Dataset construction and methodological issues
33
Table 2: Distribution of companies in the multinational and control groups within the raw sample and final sample.
Raw sample Final sample Eliminated companies MNCs Controls Total MNCs Controls Total MNCs Control Total Euro area 353 197 550 116 116 232 237 81 318
Non-Euro Europe 145 226 371 138 138 276 7 88 95
Outside Europe 191 166 357 136 136 272 55 30 85
Total 689 589 1278 390 390 780 299 199 498
34
companies change proportionally over the sample period in terms of size and foreign activity within Europe.
4.2. Descriptive statistics
In order to understand the characteristics of the dataset and check the accuracy of the research design, the descriptive findings of the dataset are presented below. This is done for all the companies in the multinational and control samples which are further divided into the three geographical locations that correspond to where they are domiciled.
Table 3 provides descriptive statistics of the final sample of multinational firms as well as for the size control sample. More specific, the two groups of companies are compared in terms of market capitalization, sales, total assets, the percentage of total foreign sales, and foreign sales in Europe as a fraction of total sales. The median multinational firm has 25.7% foreign sales in Europe. Across regions, companies domiciled in the Euro area and Non-Euro Europe have median foreign Europe sales of 38.3% and 33.5%, respectively, with only 16.6% for firms from outside Europe. This validates the expectation according to which firms located within Europe are more likely to incur increased foreign trade in this region as compared to companies from the United States and Japan. However, if one compares these results with those of Bartram and Karolyi (2006), one can conclude that there has been an overall increase in the amount of foreign trade within Europe in the time frame 2001 - 20083, especially for firms domiciled in countries from the European continent.
In terms of market capitalization, size control firms match multinationals. However, there are considerable differences in the level of sales and total assets. For instance, while the median multinational has sales of €1614.1 million and total assets of €1408.4 million, the median size control firm has almost half, namely sales of €773.8 million and total assets of €774.8 million. Multinationals tend to be large firms and consequently, the companies in the multinational sample have higher market capitalization, sales and total assets compared to the size control sample across all geographic regions.
3
36
As mentioned in section 3, weekly stock returns, market indices and trade-weighted exchange rate indices are used. All returns are measured on a weekly basis covering the period from January 1990 to December 2008. The returns are computed based on Friday-to-Friday closing prices and quotes and are continuously compounded. The actual event date is considered to be 1 January 1999. Consequently the sample period is divided into two almost equal periods, the pre-Euro period spanning from January 1990 to December 1998 and the post-pre-Euro period spanning from January 1999 to December 2008. The study makes use of trade-weighted exchange rate indices measured in local currency relative to the basket of foreign currencies from the Bank of England and in the case of Turkey and Poland from JP Morgan’s trade-weighted currency index because the information from the Bank of England was not available. In order to retrieve the latter two indices, Datastream was used. Stock returns and market indices are obtained from Thomson ONE Banker.
37
Table 4: Descriptive statistics of stock market indices
1990-2008 1990-1998 1999-2008
Mean Variance Mean Variance Mean Variance
Austria 0.0116 1.3581 0.0068 1.3896 0.0160 1.3323 Belgium 0.0103 1.0960 0.0850 0.6511 -0.0570 1.4889 Denmark 0.0488 1.2316 0.0938 0.8382 0.0083 1.5845 Europe 0.0378 1.1245 0.0855 0.6697 -0.0051 1.5321 Finland 0.0638 3.2417 0.1422 2.2161 -0.0067 4.1604 France 0.0327 1.2993 0.0720 0.9728 -0.0026 1.5929 Germany 0.0258 1.3531 0.0716 0.9229 -0.0155 1.7393 Ireland 0.0189 1.5850 0.1119 1.0892 -0.0648 2.0194 Italy 0.0121 1.6555 0.0646 1.7692 -0.0351 1.5516 Japan -0.0477 1.5034 -0.0802 1.4180 -0.0185 1.5814 Luxembourg 0.0573 1.0460 0.1217 0.4673 -0.0006 1.5619 Netherlands 0.0271 1.2676 0.1141 0.7142 -0.0513 1.7550 Norway 0.0457 1.8378 0.0547 1.6350 0.0377 2.0237 Poland 0.0037 3.1259 -0.0362 4.0524 0.0395 2.2953 Portugal 0.0139 1.0068 0.0684 0.9708 -0.0353 1.0360 Spain 0.0533 1.2977 0.1086 1.2536 0.0036 1.3346 Sweden 0.0496 2.0410 0.1007 1.9028 0.0036 2.1648 Switzerland 0.0539 1.1139 0.1272 0.8488 -0.0120 1.3454 Turkey 0.3150 8.2918 0.4585 10.4510 0.1858 6.3291 UK 0.0299 0.9498 0.0838 0.6688 -0.0185 1.1996 US 0.0474 1.0245 0.1261 0.5818 -0.0234 1.4142 World 0.0249 0.8596 0.0563 0.5770 -0.0034 1.1139
38
5. Results
In this section, the results of the analyses are presented. Firstly, the question whether nonfinancial firms with foreign sales in Euro area countries exhibit a larger reduction in stock return volatility than firms with no foreign sales in the Euro area (H1) will be answered. This will be done by analyzing the pre-Euro and post-Euro introduction stock return variances of the individual firms in the multinational and control samples. Secondly, the hypothesis according to which the market risk exposures decrease for firms which have foreign sales in the Euro area compared to companies with no foreign sales in the Euro area (H2) will be tested. Finally, the incremental change in foreign exchange rate risk exposure for companies with foreign sales in Europe will be evaluated against that of firms with no foreign sales in the Euro area (H3). All the three-subsections include one table with the aggregate results of the tests and regressions conducted on the individual companies in the multinational and control samples.
5.1. Changes in stock return volatility
In the first stage of the research, the volatility of the stock markets of the 20 countries in the sample and stock returns of nonfinancial firms is analyzed. Table 5 presents the analysis of stock return variances for the individual firms in the dataset for the two periods before and after the introduction of the Euro. In order to control for outliers, the variances above the 99th percentile and below the 1st percentile are excluded from the analysis. The findings are reported by geographic region.
40
Across regions multinational companies exhibit different degrees of volatility from one period to the other. Both before and after the introduction of the Euro, the stock return variances are quite similar. However, the multinationals domiciled in the non-Euro Europe area have higher pre- and post-Euro volatility than the firms from the Euro area. In the pre-Euro period, the mean (median) multinational firm domiciled in the Euro area has a variance of 20.9 (18.2) compared to 29.5 (18.7) for the non-Euro Europe area, while in the post-Euro period the mean (median) multinational firm located in the Euro area has a variance of 30.2 (23.1) compared to 36.1 (28.4) in the non-Euro Europe area. This could be attributed to the macroeconomic stability characterizing European countries that strived for the adoption the common currency. However, no such inferences can be made when comparing firms located in the Euro area with those from Outside Europe.
Next, the multinational sample is compared with the benchmark control firms for both periods. For companies domiciled in the Euro area, the mean individual stock return variances in the benchmark sample are similar to those of the multinationals. The only exception is the group of firms located in the non-Euro Europe area. Here the means for the pre-Euro (post-Euro) periods for multinationals compared to the size control firms is 29.5 (36.1) compared to 43.7 (44.5). It can be concluded that the companies from the non-Euro Europe area which have foreign sales in the Euro-zone experience increased volatility after the introduction of the Euro than those which have no foreign sales in the Euro-zone.
41
Similarly to the discussions from section 4, according to which stock market volatility has increased in the period after the introduction of the Euro, stock return volatility of nonfinancial firms has also grown after the adoption of the common currency. Moreover, there is little evidence that nonfinancial firms which have foreign sales in Europe have lower increases in volatility as evaluated against the firms in the control sample. These findings provide little support towards the first formulated hypothesis (H1).
The findings are consistent with those of Bartram and Karolyi (2006) who also find little differences between multinational and control firms. However, this comes in contrast to the results of Bartov et al. (1996) who, in line with economic thinking, find that firms located in the United States experience increased stock return volatility as a consequence of the breakdown of the Bretton Woods system. Unfortunately, the only other study on the firm-level exchange rate exposure of companies after the introduction of the Euro, does not conduct any analysis on the changes in the volatilities of the stock returns of individual firms (Hutson & O'Driscoll, 2009). Still, these findings can be biased by the different components of market risk and foreign exchange rate risk that might have been influenced by the introduction of the Euro. Consequently, the next two stages of this study aim at differentiating between the changes in systematic and unsystematic risk in the context of the introduction of the Euro.
5.2. Changes in market risk
The second stage of the analysis examines whether the market risk has changed after the introduction of the Euro and thus, if there is any indication that the foreign exchange rate risk is partially non-diversifiable. This is done by estimating equation (1) with OLS for every company in the dataset. Table 6 reports the aggregate results of the regression estimates after correcting the standard errors for autocorrelation and heteroskedasticity with the Newey-West procedure. The median market beta coefficients (β) along with the median changes in market beta coefficients after the introduction of the Euro (βEuro) are reported by area and sample. The
median values of β and βEuro are reported along with the associated percentages of significant
43
It can be observed that the firms in the multinational sample have median β estimates of 0.616 for those located in the Euro area, 0.579 for the Non-Euro Europe area and 0.955 for Outside Europe. In all cases, more than 87% are statistically significant and positive under the 10% significance level. If the results are aggregated at the 1% significance level, the number of significant positive β coefficients drops to 76%. This is especially the case for the companies domiciled in the Non-Euro Europe area. For the other geographic regions, the percentages change only slightly.
The findings are different for the size control sample. Compared to the multinationals, control companies have lower β coefficients of 0.372 for the Euro area, 0.456 for the Non-Euro Europe area and 0.733 for firms located Outside Europe. According to the Chi-squared and Wilcoxon rank sum tests, the β coefficients of firms in both samples are significantly different in all geographic regions. The percentage of significant positive coefficients for size control firms is less than that for multinationals but overall higher than 60%.
After the introduction of the common currency, the results show that the market risk is reduced for multinationals located in the Euro area and Outside Europe. In contrast, multinationals from the Non-Euro Europe area do not experience a reduction in market risk. However, the changes in market risk for multinationals are quite moderate, -0.094 for the Euro area, 0.052 for the Non-Euro Non-Europe and -0.069 for firms from Outside Non-Europe.
The pattern in market risk changes following the introduction of the Euro is similar for companies in the size control group with the exception of firms located Outside Europe. This group has a positive βEuro of 0.039 as compared to -0.069 for multinationals. It can be observed
44
The percentages of firms with significant βEuro coefficients is on average less than 40% for both
samples for all geographic regions, mostly equally distributed among positive and negative coefficients.
In general, the findings suggest that the reduction in foreign exchange rate risk triggered by the adoption of the European common currency is linked to decreases in market risk for companies located in the Euro Area and Outside Europe. It can also be noticed that the decrease in market risk is greater for companies with significant foreign sales in Europe as compared to companies of similar size with no foreign sales in Europe. Under the same line of reasoning, for the Non-Euro Non-Europe area the increase in market risk is less for firms with foreign business activities in Europe than for companies with no foreign activities in the same region. However, the companies located in the Non-Euro Europe area experience an increase in market risk. Consequently, the results support the second formulated hypothesis (H2) only partially. Thus, it can be argued that foreign exchange rate risk is partially systematic risk which cannot be diversified away.
The results correspond to those of Bartov et al. (1996) who analyzed the opposite case. They find that the breakdown of the previously fixed exchange rates under the Bretton Woods system led to significant positive changes in market risk for US multinationals. Thus, the adoption of a common currency has the reverse effect, by diminishing market risk. Similarly, the findings are consistent with Bartram and Karolyi (2006) who also find evidence of a reduction in market risk of non-financial firms as an outcome of the adoption of the Euro. Likewise, Hutson and O’Driscoll (2009) show that the increase in firm-level currency risk exposure after the introduction of the Euro was offset by a reduction in systematic risk. The importance of a reduction in market risk betas underline the increased opportunity for multinational firms to sustain higher business risk in the form of reduced cost of capital and increased financial leverage.
5.3. Changes in foreign exchange rate exposure
45
(Bartram & Karolyi, 2006; He & Ng, 1998; Prasad & Rajan, 1995). As such, equation (2) is estimated with OLS for every company in the dataset. The incremental exchange rate coefficient is then interpreted as the residual effect of the adoption of the common European currency on the value of multinationals with foreign sales in Europe and companies with no foreign business activities in Europe. Table 7 reports the results of the firm-level regression after correcting the standard errors for autocorrelation and heteroscedasticity with the Newey-West procedure. δ represents the overall exchange rate exposure while δEuro symbolizes the change in the exposure
after the introduction of the Euro. The median δ and δEuro are reported by sample, geographic
area and coefficient sign. Because the coefficient results for market betas and changes in market betas are close to those presented in Table 6, these findings are not reported in Table 7. Moreover, the exchange rate exposures δ and δEuro are accounted separately for positive and negative
foreign currency rate coefficients and the corresponding changes. Also, the percentage of positive and negative coefficients is given depending on three different significance levels, 1%, 5% and 10%. The table also reports the probability values of a Chi-squared test and Wilcoxon rank-sum test for determining whether the coefficients of the δ and δEuro groups are similar.
In general, compared to the results for market risk and changes in market risk, the exposures around the Euro launch are small and considerably lower. Also, the percentage of negative coefficients appears to be slightly larger than the percentage of positive foreign exchange rate coefficients. Furthermore, the median positive and negative coefficients are quite similar in magnitude. However, the median positive coefficients are larger for firms in the Euro area as compared to companies locate in Non-Euro Europe and Outside Europe (0.228, 0.212, 0.188, respectively). As far as the negative coefficients are concerned, these are larger for firms in the Euro area compared to those from Outside Europe but smaller than those from the Non-Euro area (-0.260, -0.210 and-0.287, respectively). The fraction of multinational firms with significant coefficients at the 1% level ranges between 0% and 2.9% for positive exposures (δ+) and between 0.9% and 13.2% for negative exposures (δ-). When the significance level is increased to 10%, the number of multinational firms with significant coefficients also increases. Consequently, δ+ ranges between 7.8% and 10.1%, while δ- is between 6.9% and 22.1%.
47
and outside Europe. All the exposure coefficients are smaller than those of multinational firms, with the exception of the positive exposure coefficients for companies located Outside Europe, which have higher median δ+ than that for multinationals (0.218 compared to 0.188, respectively). However, in this case the Wilcoxon and Chi-squared tests indicate that the differences are not large enough to be significant. This also holds for negative exposures in the Euro area. Still, the picture is mixed when it comes to the other regions for which either one or both of the two tests indicate that the differences are significant. For instance, the negative exposures Outside Europe are significantly smaller for size control firms, with p-values of 0.033 and 0.006, respectively.
The change in the foreign exchange exposure after the introduction of the common European currency (δEuro) has the opposite sign as the exposure (δ). The only exception is the δEuro+
coefficient, which corresponds to Euro-area size control firms, that is close to zero, namely 0.002. Consequently, it can be concluded that the change is mean-reverting to zero. For companies in the multinational sample, which are locate in the Euro area, the change for positive and negative exposures are -0.196 and 0.410. In Non-Euro Europe, the values for these changes are -0.324 and 0.173, while Outside Europe they are -0.338 and 0.107. Measured at the 1% significance level, the number of significant changes in exposure for multinationals is 0% for positive changes of positive exposures and ranges between 0% and 7.4% for negative changes of positive exposures. On the other hand, the percentages range between 1.7% and 5.1% for positive changes of negative exposures and between 0% and 2.9% for negative changes of negative exposures. By far, it can be observed in both cases that the companies domiciled Outside Europe have the highest percentage of significant changes. This observation holds true even if the significance level is increased to 5% and 10%.
48
located in the Euro area and Non-Euro Europe area with p-values of 0.002 and 0.012, and 0.008 and 0.021, respectively. In both cases, the δEuro+ coefficient for size control firms is larger than
that for multinational companies (0.002 as compared to -0.196 in the Euro area and -0.148 as compared to -0.324 in the Non-Euro area).
Generally, the findings suggest that the introduction of the common European currency brought about a reduction in exchange rate exposure. On the whole, companies with no foreign sales in Europe have higher coefficients for changes in exchange rate exposure, which translates to lower changes in currency exposure. Consequently, firms with foreign sales in Europe exhibit a larger reduction in incremental foreign exchange rate risk exposures than companies with no foreign sales in the Euro area. This is particularly true for companies located in the Euro area and Non-Euro Non-Europe area. However, the results also indicate that there are many companies with foreign sales in Europe which exhibit no significant levels of exchange rate exposure nor significant changes in exposure after the introduction of the Euro. Consequently, the findings support the third formulated hypothesis (H3) only partially.
