The goal of the analyses presented in this section is to assess whether and to what extent there have been any opportunities for intermediaries in the EMU financial sector to reduce market risk through cross-border or cross- sector diversification. To this end, we first look for differences in market risk
Table 4.5 Banks market risk by period and domicile: index model estimates Acquirer
country
Period
1997–2007 1995–8 1999–2002 2003–7
Beta R2 Beta R2 Beta R2 Beta R2
Germany 1.30 62.9% 1.24 60.0% 1.21 60.1% 1.54 63.7%
Netherlands 1.20 52.6% 1.23 51.7% 1.24 52.0% 1.06 45.7%
Spain 1.12 66.2% 1.29 61.4% 0.98 69.6% 1.09 62.1%
Greece 1.08 26.2% 1.41 23.6% 0.55 14.2% 1.64 57.6%
France 1.07 60.0% 1.43 70.3% 0.89 50.6% 1.08 61.1%
Italy 1.04 62.3% 1.4 63.0% 0.9 61.5% 0.81 46.6%
Ireland 1.01 4.8% … … 1.26 2.60% 0.8 36.8%
Belgium 0.8 44.2% 0.92 56.6% 0.57 27.9% 1.1 55.8%
Portugal 0.77 34.6% 1.11 50.1% 0.47 25.0% 0.86 23.1%
Austria 0.54 12.7% 1.03 31.1% … … 0.85 33.0%
Finland – – 1.13 27.8% – – – –
Average 0.99 42.6% 1.22 49.5% 0.9 40.4% 1.08 48.5%
Notes: OLS estimates with significance tests based on Newey&West robust standard errors.
A dash means that estimation has not been performed for lack of data.
Dots indicate that beta estimates are statistically indistinguishable from zero.
Source: Own processing of Dow Jones indices provided by Datastream.
differences would provide evidence of opportunities for risk reduction from cross-country or cross-sector M&As.
For our conclusions to be meaningful, we need to adopt a comparable definition of market risk for any country and any sector. Therefore, we define market risk as the beta coefficient of a simple (single) index model, where the EMU stock market is represented by the Dow Jones Euro Stoxx price index; consistently, the country-specific Dow Jones sector price indices for banking and for insurance are used as proxies for the respective market values.5
Classical regression estimates of banks’ and insurers’ beta over various time periods are displayed in Table 4.5 and Table 4.6, along with R-squared coefficients; all regressions have been run on monthly price relatives.
Considering several periods allows checking for beta variability over time, especially before and after the introduction of the euro in 1999.
Prima facie, it is apparent that there are substantial cross-country differ- ences in beta for both banks and insurers, which are often large enough to be economically relevant; in several cases, pair-wise t-tests lead us to reject
the hypothesis of identical sector-specific beta for two countries.We have set the significance threshold for every test to the conventional value of 5 per cent.
Over the whole period (1997–2007) there is a remarkable variability in beta by country. For banks, the ratio of the highest to the lowest beta is 2.4, for insurance companies it is 2.2; pair-wise t-tests for the difference are statis- tically significant in 26 per cent and 39 per cent of cases respectively – much more than the 5 per cent random rejection rate implied by the significance level of the tests. German banks and insurance companies are first in the ranking by beta, with values well above unity, while Dutch intermediaries come second and have the same level of market risk. The banking sectors of Spain, Greece, France, Italy, and Ireland have mid-ranking beta, in the range from 1.00 to 1.10, and are not statistically distinguishable from each other;
it is worth remarking that, with the excepion of Italy, t-tests for all these countries against Germany are not significant either. At the lowest level, Portugal, Belgium and Austria have beta below unity that are statistically dif- ferent from those of the top-three countries (Germany, the Netherlands and Spain). In the insurance sector, the beta of Germany, the Netherlands and France are in the range from 1.29 to 1.37, and much higher than the mid- and low-ranking beta of Italy, Finland, Spain and Ireland. These cases are all Acquirer
country
Period
1997–2007 1995–8 1999–2002 2003–7
Beta R2 Beta R2 Beta R2 Beta R2
Germany 1.37 49.7% 0.99 45.8% 1.11 54.10% 2.54 57.6%
Netherlands 1.35 53.6% 0.97 51.3% 1.18 45.0% 2.11 70.9%
France 1.29 59.4% 1.03 46.6% 1.34 56.20% 1.48 71.2%
Italy 0.92 52.4% 1.17 57.5% 0.85 51.10% 0.79 32.4%
Finland 0.91 28.1% 1.45 40.0% 0.57 13.70% 0.97 36.2%
Spain 0.77 23.8% 1.37 57.4% … … 1.22 48.2%
Ireland 0.62 13.5% 0.64 22.9% 0.47 5.0% 0.88 22.4%
Portugal – – 1.09 39.0% – – – –
Belgium – – 1.07 54.7% – – – –
Average 1.03 40.0% 1.09 46.1% 0.92 37.5% 1.43 48.4%
Notes: OLS estimates with significance tests based on Newey&West robust standard errors.
A dash means that estimation has not been performed for lack of data.
Dots indicate that beta estimates are statistically indistinguishable from zero.
Source: Own processing of Dow Jones indices provided by Datastream.
From the comparison of beta estimates by periods (1995–8, 1999–2002 and 2003–7) it emerges that beta change over time; the ranking of intermediaries from different countries by market risk is also different from period to period. The comparison of the pre-(1995–8) and the post-changeo- ver (1999–2002) beta reveals that, on average, the transition to the euro has generally reduced market risk at the country level for both banks (by 0.30 points) and insurers (by 0.15 points); insurance companies in Germany, the Netherlands and France are the only cases of increasing beta values.
From 1999 to 2007 beta are quite unstable. With the exception of Italian intermediaries and Dutch banks, beta increase in all countries and sectors during the last five-year period (2003–7); this trend is stronger in the case of insurance companies. In addition, market risk diversity widens across countries. For banks, rejections in t-tests rise from 4 per cent in the pre-EMU period (1995–8) to 20 per cent after the changeover (1999–2002) before jumping to 35 per cent over the last five years (2003–7); for the insurance sector, rejection frequencies in the same periods are 11 per cent, 12 per cent and 47 per cent respectively.
Table 4.7 illustrates differences between beta of the banking and the insurance sectors in each country; values of standard significance t-tests are also shown. Over the whole period (1997–2007) banks appear riskier than insurance companies in Ireland, Spain and Italy, while the opposite holds in Germany, the Netherlands and France. During the pre-changeover period (1995–8) the banking sector was riskier in four out of five countries,
Table 4.7 Cross-sector differences in market risk by country and period: banks vs insurance companies
Acquirer country
Period
1997–2007 1995–8 1999–2002 2003–7
Beta t-test Beta t-test Beta t-test Beta t-test
Ireland 0.38 1.408 – – 0.79 1.208 ⫺0.08 ⫺0.362
Spain 0.35 1.732 ⫺0.08 ⫺0.28 – – ⫺0.13 ⫺0.645
Italy 0.12 1.047 0.23 0.972 0.05 0.34 0.03 0.157
Germany ⫺0.08 ⫺0.296 0.25 0.901 0.1 0.347 ⫺1 ⫺1.437
Netherlands ⫺0.14 ⫺0.559 0.26 0.986 0.06 0.139 ⫺1.06 ⫺3.005 France ⫺0.22 ⫺0.908 0.40 1.347 ⫺0.45 ⫺1.109 ⫺0.4 ⫺2.554 Notes: OLS estimates with significance tests based on Newey&West robust standard errors.
A dash means that estimation has not been performed for lack of data.
Source: Own processing of Dow Jones indices provided by Datastream.
beta of insurers are the highest everywhere – Italy being the only exception.
Even if differences in beta are not small in absolute terms, we cannot confidently reject homogeneity among sectors within most countries and periods.
To sum up, from 1997 to 2007 there may have been opportunities for EMU banks and insurers to reduce risk through cross-border M&As and, perhaps, through cross-sector operations – but our evidence is statistically weak in this respect. Moreover, the potential benefits for acquirers belong- ing to a particular country or to a particular sector were not constant over time. Specifically, such benefits were likely to be higher from 2003 to 2007 than before, as the magnitude and statistical significance of beta differen- tials suggest.
Comparing these results to the account of M&A operations presented in the previous section, the coincidence of market risk reduction in both bank- ing and insurance after 1999 with the peak period of M&A activity (2001–4) is apparent. In addition, two important analogies emerge. First, the number of cross-border M&As, either in- or cross-sector, was highest from 2005 to 2007, when significant differences in beta (and consequently the potential for risk-reduction) were most evident. Second, domestic cross-sector aggrega- tions involving banks and insurance companies were, in general, a minority of M&A activity; our estimates of the risk-reduction potential within coun- tries also show it to be poor or absent.
These findings do not conflict with the claim that cross-border M&As (could) have had an effect on market risk reduction in the EMU financial sector. A closer inspection of cross-border deals by sectors reveals that countries with high beta banks or insurers tend to be net acquirers, while the opposite holds for mid- and low-beta countries. In the banking sector, Belgium, Germany, the Netherlands and Spain are net acquirers and have an average beta of 1.10, while France, Greece Ireland, Italy and Portugal are net targets, with an average beta of 0.99. As for insurance companies, Germany, the Netherlands and France have beta averaging 1.33 and are net acquirers, while Finland, Ireland, Italy and Spain are net targets with an average beta of 0.80. Therefore, high levels of equity market risk may have represented an incentive for some EMU financial intermediaries to diversify through cross- border M&As, and not a hindrance to external growth.