The Dynamics of Firm-Level Risk

Table of Contents

List of Tables

List of Figures

List of Abbreviations

List of Symbols

1 Introduction
1.1 Overview and General Research Objective
1.2 Essay 1: Research Questions and Main Findings
1.3 Essay 2: Research Questions and Main Findings
1.4 Essay 3: Research Questions and Main Findings

2 Cross-Border Bank Acquisitions and Changes in Bank Risk
2.1 Introduction
2.2 Related literature
2.3 Data and methodology
2.3.1 Sample and data
2.3.2 Measurements of (bank) risk
2.4 Results
2.4.1 Univariate analysis of acquiring bank’s risk
2.4.2 The acquiring bank’s risk, functional diversification, and relative riskiness
2.4.3 Robustness Tests
2.4.3.1 Cross-sectional regression on risk changes after cross-border bank M&A
2.4.3.2 Quantile regression of revenue diversity
2.5 Conclusions
Appendix Essay 1 | I
Appendix Essay 1 | II

3 Risk Dynamics Surrounding the Issuance of Convertible Bonds
3.1 Introduction
3.2 Related Literature
3.3 Sample and Research Methods
3.3.1 Sample Construction
3.3.2 Matched Firm Selection
3.3.3 Measurements of Risk
3.4 Results
3.4.1 Descriptive Analysis
3.4.2 Analysis of Systematic Risk Dynamics
3.4.2.1 Univariate Analysis of Systematic Risk Dynamics
3.4.2.2 Cross-sectional Analysis of Post-issuance Beta Changes
3.4.3 Systematic Risk and its Effects on Short- and Long-term Studies
3.4.3.1 Effects on Short-term Studies
3.4.3.2 Effects on Long-term Studies
3.5 Conclusion

4 Terror Attacks and the Efficiency of Risk Adjustment
4.1 Introduction
4.2 Related Literature
4.3 Sample and Research Methods
4.3.1 Event Definition and Sample Construction
4.3.2 Measurements of Systematic and Total Risk
4.4 Results
4.4.1 Descriptive Analysis
4.4.2 Analysis of Systematic Risk
4.4.2.1 Univariate Analysis
4.4.2.2 Regression Analysis
4.4.2.3 Implications for Event Study Methodology
4.4.3 Analysis of Total Firm and Systemic Risk
4.5 Conclusion

5 Conclusion

Appendix

References

List of Tables

Table 2.1: Cross-border Acquisitions by Acquirer and Target Country

Table 2.2: Cross-border Acquisitions by Year of Acquisition

Table 2.3: Decomposing Risk of Acquiring Banks

Table 2.4: Acquirer Risk Per Event Year

Table 2.5: Acquirer’s Risk Changes per Event Year

Table 2.6: Acquirer’s Risk Changes Conditional on Changes in Revenue Diversity

Table 2.7: Acquirer’s Risk Changes Conditional on Relative Market Risk

Table 2.8: Cross-sectional Regression on Risk Changes

Table 2.9: Quantile Regression of Total Volatility Changes

Table 3.1: Number of Issues per Year

Table 3.2: Number of Issues per Industry

Table 3.3: Pre-issue Firm and Market Characteristics

Table 3.4: Comparison of Operating Performance for CBO and SEO firms

Table 3.5: Cross-Sectional Regressions on Post-Issuance Beta Change

Table 3.6: CAARs for CBOs and SEOs Using Different Beta Estimates

Table 3.7: Five-year BHARs Grouped by Equity Size and Book-to-market Quintiles

Table 3.8: Cross-Sectional Regressions on BHR and Wealth Relatives

Table 3.9: Factor-mimicking Portfolios and Macroeconomic Variables

Table 3.10: Calendar-time Regression Analysis

Table 4.1: Number of Firms per Country and Firm Type

Table 4.2: Long-term Risk Reaction after Terrorist Attacks

Table 4.3: Systematic Risk Regressions

Table 4.4: CAARs using Different Methods in Estimating Beta Coefficients

List of Figures

Figure 2.1: Risk Dynamics of Acquiring Banks

Figure 2.2: Risk Dynamics Conditional on Changes in Revenue Diversity

Figure 2.3: Risk Dynamics Conditional on Relative Market Risk

Figure 3.1: Beta Dynamics for CBO and SEO Firms

Figure 3.2: Beta Dynamics and Investor Sentiment

Figure 4.1: SBETA Dynamics for US and European Financial Institutions

Figure 4.2: SBETA Dynamics for US Financial Institutions

Figure 4.3: SBETA Dynamics for European Financial Institutions

Figure 4.4: Volatility Dynamics for US and European Financial Institutions

Figure 4.5: Volatility Dynamics for US Financial Institutions

Figure 4.6: Volatility Dynamics for European Financial Institutions

Figure 4.7: Average Return Correlations over Time.

List of Abbreviations

Abbildung in dieser Leseprobe nicht enthalten

List of Symbols

Abbildung in dieser Leseprobe nicht enthalten

1 Introduction

1.1 Overview and General Research Objective

The seminal research of Markovitz (1959), Sharpe (1964), Lintner (1965), and Black (1972) marks the birth of modern finance theory which addresses the fundamental question of how the risk of an asset relates to its expected return. The notion that risk-averse investors choose portfolios that are “mean-variance-efficient” led to the field’s first coherent model (Perold (2004)) addressing this question: the Capital Asset Pricing Model (CAPM). It provides a formal framework for the idea that asset prices are affected only by certain risks, i.e. risks that cannot be diminished by diversifying, and states that the risk of an asset should be estimated relative to a “market portfolio.”^[1] Unfortunately, though, empirical tests have failed to find evidence to support the CAPM. Research has shown that the variation in expected returns is hardly related to the market beta. Many stock market anomalies have been documented that starkly contradict the theoretical predictions of the CAPM. For example, Basu (1977) shows that future returns on high E/P stocks are too high to be explained by the CAPM. Banz (1981) documents a size effect yielding returns for small stocks that exceed those explained by the CAPM. Furthermore, Stattman (1980) and Rosenberg, Reid and Lanstein (1985) find that high book-to-market stocks yield returns that are higher than those predicted by the CAPM. These findings lead Fama and French (1996) to the conclusion that “the average-return anomalies of the CAPM are serious enough to infer that the model is not a useful approximation . ” (p. 1957).

The conclusion that the static beta is dead has led to the development of several expansions of the model, for example, the intertemporal capital asset pricing model by Merton (1973) and the three-factor model by Fama and French (1993). Moreover, these developments have sparked a wide discussion regarding the assumption that the relation between risk and return remains constant over time. It is now widely accepted that risk factors, from the perspective of either a single-factor (e.g., CAPM) or a multi-factor model (e.g., the Fama-French model), vary substantially over time.^[2] For example, Jostova and Philipov (2005) and Ang and Chen (2007) find that systematic risk is highly volatile and that fluctuations in systematic risk differs across industries. Also, Campbell et al. (2001) find that firm-level idiosyncratic and total return risk fluctuate substantially. Consequently, considerable research focuses on the implications of time-varying risk. Some authors argue that time-varying betas (conditional CAPM) help explain many of the documented stock market anomalies. For example, Zhang (2005) finds that stocks with high book-to-market ratios are riskiest during recessions, helping to explain the value premium. Also, Avramov and Chordia (2006) show that the empirically documented book-to-market and size effect may be a consequence of fluctuations in systematic risk. Furthermore, Jagannathan and Wang (1996) and Lettau and Ludvigson (2001) argue that allowing betas to vary over time improves the overall performance of the CAPM.^[3] Learning about the dynamics of risk is one of the central issue in finance research since it is impossible to reason whether market prices reflect rational investment behavior without information about risk levels. It is important to understand the central relationship between the riskiness of an asset and its expected return. After all, as noted by Fama and French (2004a): “The CAPM, like Markowitz’s portfolio model on which it is built, is nevertheless a theoretical tour de force” (p. 44).

My dissertation contributes to this central issue of gaining a better understanding of the risk dynamics in capital markets. Although different in focus and approach, my first two essays analyze how corporate investments affect the riskiness of the firm. My first essay (Chapter 2) examines whether bank acquisitions lead to changes in bank risk and analyzes the determinants of these risk changes. My second essay (Chapter 3) is devoted to another important corporate finance event and analyzes the relationship between the issuance of convertible debt and changes in risk. Both essays build on recent research by Carlson, Fisher and Giammarino (2004) and Hackbarth and Morellec (2008) on how corporate investments relate to the riskiness of the firm. In my third essay (Chapter 4), I take a different perspective and analyze the time and size of the risk adjustment process to market-wide information events. In particular, I am interested in whether the adjustment process is consistent with the notion of efficient capital markets.

Overall, my research objective is to improve the understanding of the effects of corporate events on the riskiness of the firm. My research has important implications for several interest groups. Knowledge regarding the relationship between corporate events and the dynamics of risk is particularly important for: (1) Researchers who attempt to understand the valuation consequences of corporate finance effects. For example, puzzling return behaviors that occur after various corporate finance events can be related to fluctuations in risk. (2) Investors who need to ensure rational investment decisions. Without knowledge about the adequate risk levels, such rational investment decisions are impossible. (3) Bank supervisors and regulatory authorities who have to monitor the stability of financial institutions. Learning about risk changes subsequent to bank acquisitions may lead to regulatory policy changes.

1.2 Essay 1: Research Questions and Main Findings

The first essay, entitled “Cross-Border Bank Acquisitions and Changes in Bank Risk”, examines whether bank acquisitions lead to changes in the riskiness of the acquiring bank. From a theoretical perspective it is unclear how the acquisition of foreign banks changes the riskiness of the acquiring bank. On the one hand, it is argued that geographic diversification reduces risk, since it lowers the co-variation of returns on loans. On the other hand, this effect might be offset by managers who have incentives to increase risk as long as the regulatory safety net and its implicit and explicit guarantees are underpriced. Furthermore, geographic diversification may increase monitoring costs. Geographic distance as well as cultural and organizational differences might adversely affect risk and hence ultimately reduce economic growth. Also, several determinants may have a positive or negative effect on bank risk. In particular, the change in functional diversification and relative riskiness between target and acquirer may be important determinants of risk changes. Given these theoretically ambiguous findings it is remarkable that only limited research has examined the relationship between geographic bank diversification and risk. My first essay fills this research gap and ask the following two questions:

(1) Do cross-border acquisitions change the riskiness of the acquiring bank?

(2) Are the degrees of functional diversification and relative riskiness determinants of any of these changes in bank risk?

Analyzing 264 cross-border bank acquisitions from 33 countries I find no changes in bank risk subsequent to cross-border acquisitions in general. However, conditioning the sample on measures of functional diversification, I find that acquirers who diversify functionally experience large, significant, and persistent increases in firm-specific and total risk. I show that the positive relationship between the increase in functional diversity and the change in risk remains robust and independent of the absolute level of functional diversity prior to the merger. I also find that the relative riskiness between acquirer and target has no measurable impact on the riskiness of the acquiring bank.

1.3 Essay 2: Research Questions and Main Findings

The second essay, entitled “Risk Dynamics Surrounding the Issuance of Convertible Bonds”, investigates the risk dynamics accompanying the issuance of convertible bonds. Despite extensive research on the issuance decision of equity-linked financing instruments, we still possess a limited understanding of the issuing motives behind convertible bond offerings and the choice between issuing either convertible bonds or seasoned equity offerings. In particular, the empirical observed pre- and post-share price performance contradicts theoretical predictions of traditional convertible bond theories. Based on a real option framework that predicts substantial changes in systematic risk surrounding seasoned equity offerings, I analyze whether the issuer of convertible bonds experiences similar fluctuations in systematic risk. I also investigate whether the fluctuations in risk provide a rational explanation for the observed share price patterns of convertible bond issuers. My second essay contributes to the broad research conducted on the issuance decision of equity-linked financing instruments and asks the following two questions:

(3) Are the dynamics of systematic risk of firms that issue convertible bonds similar to those found for firms that conduct seasoned equity offerings?

(4) Do these risk dynamics have an impact on short- and long-term returns, and do they provide a rational explanation for the empirically observed share price patterns?

Analyzing a dataset of 1,148 convertible bond issuers over a period from between 1980 and 2002, I investigate the systematic risk dynamics surrounding convertible bond offerings and compare them to those observable in a dataset of 2,905 seasoned equity offerings. My findings confirm that firms that issue convertible bonds have risk dynamics very similar to those of firms that offer seasoned equity. Furthermore, I document that the apparent contradiction between traditional theories regarding the issuance decision of convertible bonds and extant empirical findings regarding the share price patterns surrounding convertible bond offerings are, to a large degree, caused by neglecting the particular dynamics of systematic risk surrounding the issuance.

1.4 Essay 3: Research Questions and Main Findings

The third essay, entitled “Terror Attacks and the Efficiency of Risk Adjustment”, takes yet another perspective and analyzes whether the process of risk adjustment is consistent with the assumption of efficient capital markets. The (semi-strong) efficient market hypothesis states that prices should always adjust immediately to information announcements to reflect all (publicly) available information. Consequently, much research has been undertaken examining the adjustment of share prices to new information. Yet remarkably little is known regarding the adjustment of (systematic) risk to new information. In an efficient market, any announcement of information may also theoretically induce a re-pricing of risk. For example, new information about changes in the micro- as well as the macroeconomic environment should naturally lead to adjustments of betas. My third essay consequently asks the following two questions:

(5) Does risk adjust to new information?

(6) Is the adjustment process consistent with the assumption of efficient capital markets?

To this end, I examine the dynamics of firm-level (systematic) risk surrounding three terrorist attacks in New York, Madrid and London. Theoretically, a risk shift is justified by changes in political and economical conditions. Analyzing 334 of the largest US and European financial institutions, I find that the attacks on September 11th had a strong short- and medium-term effect on the riskiness of insurance companies, possibly due to their expected loss exposure. On the other hand, I do not find any significant positive risk shifts subsequent to the terror attacks in Madrid and London. This may be explained by the fact that neither attack increased uncertainty regarding the political and economic situation, since the inherent possibility for political and economical instability had been immediately priced after 9/11. Overall, I find that the adjustment of risk is remarkably consistent with the assumption of efficient capital markets.

2 Cross-Border Bank Acquisitions and Changes in Bank Risk

2.1 Introduction

Geographic bank diversification through mergers and acquisitions (M&A) has been scrutinized extensively. Most theoretical and empirical work addresses performance-related questions of cross-border bank M&A (Eun, Kolodny and Scheraga (1996), Berger et al. (2000), Cybo-Ottone and Murgia (2000), Beitel, Schiereck and Wahrenburg (2004), Campa and Hernando (2006)). However, knowledge regarding the relationship between bank risk and geographic diversification is scarce. It remains unclear whether bank supervisors or regulatory authorities should limit or encourage international expansion of financial institutions via M&A. This ambiguity is particularly critical because cross-border transactions typically involve large financial institutions (Berger, DeYoung, Genay and Udell (2000)), and a possible risk increase may greatly affect the stability of the financial system.

Theoretical discussion of the relationship between bank risk and cross-border bank M&A is ambiguous. While it is generally assumed that cross-border bank mergers have the potential to reduce risk due to lower co-variation of returns on loans and a consequent reduction of the likelihood of insolvency (Vander Vennet (1996), Berger (2000)), other factors may potentially offset this reduction in bank risk. A moral hazard perspective suggests that managers of financial institutions have incentives to increase risk as long as the regulatory safety net and its implicit and explicit guarantees are underpriced (Keeley (1990), John and John (1991), John, Saunders and Senbet (2000)). As noted by Amihud, DeLong and Saunders (2002), one possibility for exploiting this underpricing is to acquire other potentially risky foreign banks. Geographical diversification may also be associated with an increase in monitoring costs. Factors such as geographic distance as well as cultural and organizational differences might adversely affect risk and ultimately reduce economic growth (Winton (1999)).

Notwithstanding the apparent importance of bank risk in the context of cross-border diversification, only limited empirical research has examined this relationship. To date, only Amihud, DeLong and Saunders (2002) have analyzed the effects of cross-border bank M&A on bank risk.^[4] They find that geographic bank diversification is not associated with changes in systematic or total risk of acquiring banks and conclude that bank regulators should not adopt systematic policies to limit cross-border acquisitions.

I analyze the relationship between cross-border bank M&A and bank risk in greater depth. Specifically, my interest lies in two particular aspects not examined by Amihud, DeLong and Saunders (2002): functional diversification and relative riskiness. First, the degree of functional diversification of banks has been shown to affect bank risk and value (Stiroh (2006b), Baele, De Jonghe and Vander Vennet (2007), Laeven and Levine (2007)). An analysis of the relationship between cross-border bank M&A and bank risk cannot be conducted without controlling for this determinant. I raise the question of whether concurrent diversification on a geographic and a functional scale raises integration and monitoring costs to a level where they exceed potential benefits. Second, the relative riskiness of the acquiring bank (or home market) and the target bank (or home market) may be a determinant of risk changes related to cross-border bank M&A. As Repullo (2001) argues, the decision to enter into a cross-border takeover depends on the riskiness of the target bank compared to the domestic bank. Furthermore, Hackbarth and Morellec (2008) document that, depending on the relative riskiness between acquirer and target, M&As are associated with significant dynamics in firm-level betas. In particular, acquirers who have higher (lower) betas compared to their targets experience an increase (decrease) in systematic risk prior to announcement of the transaction and a sharp decrease (increase) thereafter. These findings regarding firm-level dynamics strongly suggest that the relative riskiness of acquirers and allowances for time-varying risk measures should be accounted for.

I explicitly analyze the effects of functional diversification and relative riskiness on the changes in risk of acquiring banks in cross-border transactions. To this end, I examine 264 cross-border bank mergers from 33 countries. For the entire sample, I confirm findings from Amihud, DeLong and Saunders (2002) and document that the marginal changes in risk subsequent to the announcement of a cross-border M&A cannot be distinguished from unconditional fluctuations in risk. However, conditioning the sample on measures of functional diversification I find that acquirers who diversify functionally experience large, significant and persistent increases in firm-specific and total volatility. The risk of the acquiring bank’s home market relative to the target market does not have any effect on bank risk. I further investigate the robustness of my results and control for firm- and market-specific factors using cross-sectional regression analysis. I show that the positive relationship between changes in functional diversity and risk remains robust and independent of the absolute level of functional diversity prior to the merger. Also, I observe that bank size and prior share price appreciation have a significant positive effect on changes in bank risk. Finally, quantile regression analysis shows that the observed relationships hold for other points of the distribution and hence are not driven by outliers.

The results have important implications for bank supervisors and regulatory authorities. Banks that enter into cross-border transactions to functionally diversify experience significant and large increases in firm-specific as well as total return volatility. A supervision policy regarding international bank expansion via M&A should therefore involve close monitoring regarding the degree of functional diversification associated with the particular transaction. Furthermore, my results provide evidence that future research should incorporate diversifying bank acquisitions as an important determinant for bank risk.

The remainder of the paper is organized as follows. Section 2.2 presents related literature and derives the research hypotheses. Section 2.3 introduces the data and describes the methodology. Section 2.4 discusses the results and Section 2.5 presents my conclusion.

2.2 Related literature

Amihud, DeLong and Saunders (2002) demonstrate that systematic market risk and total return volatility are not significantly affected by geographic bank diversification. They conclude that regulators and bank supervisors should not adopt a general policy for cross-border bank mergers, but should decide on a case-by-case basis whether international bank expansion is encouraged. However, neglecting the many determinants of bank risk, their study does not control for any cross-sectional differences in bank- or market-specific factors.

Theoretical and empirical research suggests that changes in bank risk through geographical expansion cannot be analyzed independent of other factors. Much research has been conducted on the determinants of bank risk in general. Saunders, Strock and Travlos (1990), analyzing the relationship between bank ownership and risk-taking, find that stockholder-controlled banks are riskier than manager-controlled banks and that the documented differences become even greater during periods of deregulation. Their applied indicators of risk are market measures, i.e., total stock return, firm-specific volatility, and systematic market and interest rate risk. Numerous other studies focus on the relationship between the level of functional diversification and bank risk by using accounting-based measures of bank risk (DeYoung and Roland (2001), Stiroh (2004), Stiroh and Rumble (2006)). In contrast, the focus of this discussion is on those studies applying capital market-based indicators of risk, because these measures represent a timely and undistorted approximation of (bank) risk and, therefore, are used in this analysis.

Studies examining the relationship between functional diversification and bank (capital market) risk yield ambiguous results.^[5] Templeton and Severiens (1992) analyze bank holding companies from 1979 and 1986 and report that diversification is associated with lower stock return variance. Similarly, Kwan (1998) presents diversification benefits for commercial banks due to a low return correlation between securities and bank subsidiaries. On the other hand, Stiroh (2006a) argues that an increase in functional diversification (measured as the share of non-interest revenues) is associated with a significant increase in average return volatility but no increase in average returns. He suggests that banks have likely overstretched their diversifying efforts. In a comprehensive analysis, Baele, De Jonghe and Vander Vennet (2007) examine the relationship between functional bank diversification and systematic market risk, total risk and firm-specific risk. Their results are ambiguous. While the relationship between systematic risk and diversification is apparently positive, idiosyncratic and total risk is negatively affected by diversification. They conclude that diversification (measured as non-interest revenues and loans to total assets) can make banks safer. However, they argue that the relationship is non-linear and that diversification can easily exceed optimal thresholds, leading to an increase in risk. Overall, findings from prior studies suggest that cross-sectional differences in bank-specific factors and especially revenue diversification should be controlled for when analyzing the relationship between geographic diversification and bank risk.

While not specifically addressing bank M&A, Hackbarth and Morellec (2008) have recently suggested that the M&A itself changes the nature and riskiness of the acquiring firm’s assets. They propose a real option model assuming that the bidding firm holds a takeover (call) option. At the time of the transaction, the bidding firm exercises this option, which in turn affects the riskiness of the firm’s assets. Contradicting common intuition that this option exercise should trigger a decrease in (systematic) risk and expected returns, they argue that the sign of the risk change is dependent on the relative riskiness of the acquiring and the target firm. Firm-level betas of acquirers are shown to increase (decrease) prior to announcement and decrease (increase) thereafter if the acquirer is more (less) risky than the target prior to the takeover. These results have important implications for any study of (systematic) firm-level risk. In the context of this analysis, a possible negative relationship between geographical diversification and bank risk, for example, may be offset by an increase in risk due to the acquisition itself. Consequently, the riskiness of the acquiring bank prior to the transaction must be taken into account.

To summarize, evidence from previous work suggests that the relationship between cross-border bank M&A and bank risk cannot be interpreted independent of other determinants of bank risk in general or the effects of M&A on firm-level risk in particular. I therefore extend the examination of effects from geographic bank diversification and bank risk by explicitly accounting for firm- and market- specific factors. This enables us to isolate the cross-border effect from other risk factors, such as functional diversification.

2.3 Data and methodology

2.3.1 Sample and data

The sample of bank mergers was taken from the Thomson One SDC Platinum Database.^[6] I consider world-wide transactions (a) with deal values of $50m, (b) that were filed between January 1, 1980 and December 31, 2005 and (c) in which over 50% of the bidding firm was acquired. Furthermore, I examine only mergers where both partners have an SIC-code between 6000 and 6999 (financial institutions) and at least one partner is a commercial bank (SIC-code 6000).^[7]

Abbildung in dieser Leseprobe nicht enthalten

I define a transaction to be cross-border if the two partners are headquartered in different countries. In addition, I isolate the cross-border effect on risk from a “general” control transaction effect (Hackbarth and Morellec (2008)) by further restricting the sample to acquirers that have not participated in a takeover in the one year period surrounding the announcement date. Market data constraints further restrict the sample. I only examine transactions where return data for the acquiring bank is available for the one-year period prior and subsequent to announcement.^[8] Also, I only consider countries where local market and financial indices are available. The return data for the acquiring bank as well as the home, world, and host indexes come from the Thomson Financial Datastream. The accounting data are from Thomson Financial Worldscope.

The final sample consists of 264 cross-border transactions from 33 countries and differs slightly in size and composition from the sample used by Amihud, DeLong and Saunders (2002). This may be attributed to the longer time span of the analysis and the more restrictive sampling that I conduct. Table 2.1 shows acquirers and targets by country. Similar to Amihud, DeLong and Saunders (2002), we observe that in some countries both acquirers and targets are affiliated (e.g., Germany, United Kingdom, United States), whereas in other countries banks are only targets in M&A transactions (e.g., Argentina or Mexico).

Overall, acquiring banks are based in 21 countries, and targets are based in all 33 countries. Table 2.2 shows the number of cross-border mergers by year of announcement. I find a steady increase in merger activity, with a peak in 2000 (38 mergers).

Abbildung in dieser Leseprobe nicht enthalten

where there are N daily returns rit in month t . I assume that each month has 20 trading days. The applied measure of monthly volatility is a measure of “realized volatility” as in Andersen and Bollerslev (1998) and Andersen et al. (2003).

The applied methodology of “realized” risk measures theoretically imposes concerns regarding negative autocorrelation and hence an overstatement of risk for illiquid firms, overstatement which might be caused by bid-ask bounces (Campell, Lo and MacKinlay (1997)). However, as noted by Carlson, Fisher and Giammarino (2006b), the results do not seem to be affected by microstructure effects and hence are considered reliable. To provide a further robustness check, I split the sample into three groups including firms with high, medium and low liquidity as measured by Pástor and Stambaugh (2003). I find no systematic differences in patterns of risk dynamics (refer to Appendix I of this Chapter). In particular, differences in absolute terms are quite comparable to those observed by Hackbarth and Morellec (2008).

2.4 Results

This section is organized as follows. Section 2.4.1 provides a univariate analysis on the effects of cross-border bank M&A on the acquiring bank’s risk. Section 2.4.2 discusses the effects of changes in functional diversity and the relative riskiness between acquirer and target on changes in bank risk after cross-border M&A. In Section 2.4.3, I present several robustness checks. I apply cross-sectional analysis to control for firm- and market-specific factors of bank risk. Furthermore, I use a quantile regression approach to show that the observed relationships hold for other points of the distribution and are not driven by outliers.

2.4.1 Univariate analysis of acquiring bank’s risk

Table 2.3 presents some summary statistics for systematic risk, firm-specific (idiosyncratic) volatility, and total volatility as derived from a standard market model, where the market is represented by the home financial institution index. I report my initial results for the standard market model (using the financial index of the acquirer’s country as the relevant market) to be able to compare the risk measures with Baele, De Jonghe and Vander Vennet (2007).

I document risk measures over time where each time interval of five years includes all banks of the sample with data available for that particular period. Logically, the average firm-level beta is close to 1. Systematic risk is highest for the beginning of the period (1.119) and decreases gradually until the end of the millennium (0.993) only to subsequently increase (1.057). Turning to the (annualized) measures of volatility, I find that idiosyncratic and total risk fluctuates substantially over time. For example, idiosyncratic volatility is 20.7% during 1986-1990 and increases to 24.5% in 1996-2000. The largest portion of total risk is unsystematic firm-specific risk (on average 75%). Unsystematic risk is highest in 1991-1995 (79.7%) and lowest in 2001-2005 (69.5%). These patterns are generally consistent with those found by Baele, De Jonghe and Vander Vennet (2007). However, I find that idiosyncratic risk in their sample of (European) banks seems to account for a higher portion of total risk (91%). This observation suggests that banks participating in cross-border transactions have a higher overall exposure to systematic (home) market risk.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.1 shows risk dynamics over event months relative to the transaction. I can generally confirm the findings from Table 2.4. No clear relationship between cross-border bank M&A and risk dynamics can be observed. Two remarks should be made, however. First, systematic (home) market risk increases sharply in the month prior to announcement and decreases rapidly thereafter. While this fluctuation is small, it is very similar to results from Hackbarth and Morellec (2008), who also find a small effect around the transaction.

Abbildung in dieser Leseprobe nicht enthalten

This may indicate that acquiring banks have on average a higher firm-level beta than their targets. This is intuitive, since the effect can only be observed for the home market beta. Second, both volatility measures gradually increase starting about one year after the transaction. This observation might be interpreted as an indication of a delayed positive relationship between cross-border bank M&A and return volatility. Overall, it is unclear if the increase in volatilities can be attributed to geographic diversification.

The relationship between geographic diversification and bank risk is further examined in Table 2.5. I report mean and median year-over-year changes in period comprising three years before and three years after the announcement of the transaction and I then test these changes for significance.

Abbildung in dieser Leseprobe nicht enthalten

Consistent with prior results, I find indications that (median) volatility risk significantly decreases in the post-transaction year. Both (mean and median) volatility measures significantly increase between year +2 and +3. World market and home market betas also change significantly in the year subsequent to the merger. However, the direction of these changes is the opposite of what one would expect. Median world market beta decreases, whereas median home market beta increases. Intuitively, one would expect the opposite, since international diversification should be associated with a higher covariation with the world market and a lower covariation with the home market. Again, this may be caused changing macroeconomic conditions driving these results. This further emphasizes why total risk is the preferred measure. It is not only methodologically the more robust measure, but moreover, it allows for greater ease of control for changes in market conditions.

Overall, the results regarding the relationship between geographic bank diversification and bank risk are consistent with prior findings by Amihud, DeLong and Saunders (2002). We observe no clear pattern in the effects of cross-border bank mergers on risk. Furthermore, I find significant variation in risk measures in most years, which is consistent with prior studies and shows that bank risk is affected by many determinants. I must control for those other determinants in order to isolate the cross-border effect. Therefore, in the next section I condition the sample on measures of return diversification and relative riskiness.

2.4.2 The acquiring bank’s risk, functional diversification, and relative riskiness

Amihud, DeLong and Saunders (2002) find that the risk of acquiring banks is not affected by cross-border M&A in general. Prior studies argue that bank risk has various determinants. But the effects of cross-border bank M&A on bank risk cannot be analyzed in isolation from these other determinants. I am mainly interested in two particular aspects: the effects of combined geographic and functional diversification and the (relative) riskiness of the acquiring bank on bank risk.

Baele, De Jonghe and Vander Vennet (2007) argue that banks may exceed the optimal level of (functional) diversification, leading to an increase in risk. Also, Laeven and Levine (2007) argue that diversification in financial conglomerates is associated with large agency conflicts. One might expect that cross-border bank M&A and its associated consequences for risk differ depending on the degree of functional diversification. The concurrent diversification on a functional dimension may increase the costs of diversification and hence offset the potential benefits of geographic diversification. This evidence raises the question of whether acquirers who diversify geographically and functionally at the same time overstretch diversification efforts and increase bank risk. Similar to Stiroh (2006a) and Baele, De Jonghe and Vander Vennet (2007), I approximate functional diversification by the share of non-interest revenue to total revenue. The higher the share of non-interest revenue, the more a bank diversifies away from traditional bank activities. The Thomson Financial Datastream and Worldscope only provide accounting items for 159 of the 264 acquirers, though this number is still larger than the sample of Baele, De Jonghe and Vander Vennet (2007), who base their analysis on a sample of 143 European banks.

In Figure 2.2, I show risk dynamics of acquiring banks conditioned on functional diversification, i.e., the change in the share of non-interest revenue one year prior to the transaction compared to one year following it. I split the sample of 159 acquiring banks into tertiles based on the percentage change in revenue diversity, measured as 1 - |2x – 1|, where x is the share of non-interest revenue. This measure is also applied by Baele, De Jonghe and Vander Vennet (2007) and is considered to be a good proxy for the degree of functional diversification associated with geographic diversification. It can only take values between 0 and 1 and assumes that the optimal level of diversification is an equal division between interest and non-interest revenue.^[11] The solid red line represents acquirers in the highest tertile, whereas the dashed dark line represents acquirers in the lowest tertile. I find that both measures of systematic risk seem not to be affected by the degree of functional diversification associated with the merger. Differences can be found for neither absolute levels of systematic risk nor risk changes induced by the transaction. This result differs from the findings of Baele, De Jonghe and Vander Vennet (2007), who argue that an increase in the share of non-interest revenue is accompanied by an increase in systematic risk.

The other two measures of return volatility are greatly affected by the degree of change in non-interest revenue associated with the merger. When considering functional diversification, we observe that firm-specific and total volatility differ in absolute levels as well as in the degree of change accompanied by the transaction. Acquirers whose merger is associated with an increase in the share of non-interest revenue have lower absolute levels of idiosyncratic and total volatility in the year prior to the merger. However, in the year following the transaction, both volatility measures gradually increase for these acquirers. Conversely, for acquirers who are in the lowest tertile (i.e., who decrease their share of non-interest revenue), volatility decreases in the year following the transaction. About 24 months after the transaction acquirers who increase non-interest revenue shares have higher risk levels than acquirers who decrease non-interest revenue shares.

Abbildung in dieser Leseprobe nicht enthalten

Table 2.6 reports mean and median changes in all risk measures as well as the associated significance of these changes. In Figure 2.2, I document that systematic risk is not affected by the degree of functional diversity associated with the transactions. While (world) market beta changes are negative and significant, no differences between the highest and the lowest tertile can be observed. We observe, however, that for acquirers who concurrently diversify on a geographic and functional dimension, both volatility measures, i.e., idiosyncratic and total volatility, increase significantly. Conversely, I detect a decrease (albeit not significant) in volatilities for acquirers in the lowest tertile, i.e., for acquirers who decrease their degree of revenue diversification.

Abbildung in dieser Leseprobe nicht enthalten

Remarkably, for acquirers in the highest tertile, the absolute level of revenue diversity is lower in the year prior to the announcement and in the year after the announcement. This finding suggests that it is not only the absolute level of diversification that determines bank risk (Stiroh (2006a) or Baele, De Jonghe and Vander Vennet (2007)). Moreover, it is the means by which diversification is accomplished. The associated costs of diversification of both dimensions at the same time seem to be larger than the potential benefits. A possible explanation for this observation may be that the monitoring and post-merger integration costs become too high when not only cultural differences and geographic distance must be dealt with, but also increases in agency costs due to the increasing complexity of the functional diversified organization. Consequently, banks that enter into cross-border transactions to diversify functionally experience significant and large increases in firm-specific as well as total volatility.

[...]

---

^[1] As noted by Fama and French (2004a), the term capital asset pricing model refers to every asset pricing model. However, the acronym CAPM is commonly reserved for the model developed by Sharpe (1964), Lintner (1965), and Black (1972).

^[2] This has also led to a variety of time-varying market models. See Chen and Lee (1982), Bollerslev, Engle and Wooldridge (1988), Harvey (1989), Ng (1991), Bodurtha and Nelson (1991), and Chan, Karolyi and Stulz (1992) for a variety of early single-factor models, and see Jostova and Philipov (2005) or Ang and Chen (2007) for more current developments.

^[3] Other researchers such as Lewellen and Nagel (2006) counter-argue that fluctuations in betas must be implausibly high to explain some of the market anomalies.

^[4] Note that the unpublished work of Choi, Francis and Hasan (2006) also examines the relationship between cross-border bank M&A and risk. However, as opposed to Amihud, DeLong and Saunders (2002) and this analysis, Choi, Francis and Hasan (2006) focus on bond returns instead of equity returns.

^[5] See DeYoung and Roland (2001) and Stiroh (2004) for detailed surveys of the relationship between functional diversification and bank risk.

^[6] The mergers and acquisitions section (SDC Platinum) of the Thomson One Banker Database was formerly known as the Securities Data Company’s (SDC) U.S. Mergers and Acquisitions Database.

^[7] Note that I restrict the sample to financial institutions only. This differs from Amihud, DeLong and Saunders (2002) who do not require the partner of the commercial bank to be a financial institution.

^[8] The procedure is also applied by Amihud, DeLong and Saunders (2002) since the survivorship problem that is induced is considered to be of minor importance.

^[9] Note that asset return volatility is likely to be an even more accurate measure. However, this measure is only available to bank supervisors and regulators with a time delay and might therefore not be appropriate for monitoring purposes.

^[10] In line with Baele, De Jonghe and Vander Vennet (2007) who argue that interest rate betas are small compared to market betas, I disregard the interest rate beta in the subsequent sections of the analysis and focus only on systematic market risk and firm-specific and total return volatility.

^[11] Note that the results remain similar when using the share of non-interest revenue.