Default risk and interconnectedness in the US financial sector: Is financial institutions’ default risk systemic or systematic?

Jannes Rauch1 Mary A. Weiss2 Sabine Wende3

Draft prepared for submission to the 2015 FMA European Conference

November 24, 2014

1

Cologne Graduate School in Management, Economics and Social Sciences,University of Cologne, AlbertusMagnus-Platz, 50923 Cologne, Germany, Tel.: +49-221-470-3833, Fax: +49-221-428-349, [email protected] 2 Temple University, 624 Alter Hall, 1801 Liacouris Walk, Philadelphia, PA 19122, Tel: +1-215-204-1916, Fax: +1-610-520-9790, [email protected] 3 Faculty of Management, Economics and Social Sciences, University of Cologne, Albertus-Magnus-Platz, 50923 Cologne, Germany, Tel.: +49-221-470-2330, Fax: +49-221-428-349, [email protected]

Default risk and interconnectedness in the US financial sector: Is financial institutions’ default risk systemic or systematic?

Abstract

We analyze if financial institutions’ default risk is systemic and systematic risk in the US financial sector. In contrast to previous papers, we particularly disentangle between systemic and systematic risk. Furthermore, we analyze the interconnectedness between banks, investment banks, life insurers and property-liability insurers by analyzing if default risks can spillover from one sector to another sector. Using a dataset of 7,941 firm year observations for the years 2004 until 2012 and regression analyses, we find that default risk in the banking sector is systematic, systemic and idiosyncratic while it is systemic and idiosyncratic in the life insurance sector during the financial crisis and idiosyncratic only in the property-liability insurance industry. Our results have implications for regulators regarding the regulation of financial firms and for investors regarding diversifiable and non-diversifiable risks in their portfolios. Hence, we provide additional evidence on the necessity of macro-prudential regulation in the financial industry, as our results indicate the need for regulatory measures in the banking sector, while the insurance sector does not pose systematic risk and should thus not be subject to tight regulation.

1. Introduction The financial crisis of 2008 showed the vulnerability of the overall financial sector towards the financial distress of individual, large financial firms: The failure of Lehman Brothers and the financial distress of AIG (American Insurance Group) caused failures and financial distress of other companies in the financial sector, hence providing evidence of systemic linkages through the financial sector. Thus, the crisis showed that the default risk of major financial institutions can spill over to other financial institutions and create systemic risk in the financial industry and the overall economy. Given the high degree of interconnectedness between the firms in the financial sector, a financial firms’ default risk might not only affect the company itself, but also companies from the same as well as from other sectors, e.g. insurance companies or banks. In this research, we examine the effect of a financial institutions’ (banks and insurance companies) default risk on systemic and systematic risk.1 A vast amount of studies (e.g. Huang, Zhou and Zhu 2009; Acharya et al., 2010) examine systemic risk in the financial sector and analyze how the bank’s riskiness can affect other banks, in particular during the financial crisis of 2008 and the surrounding years. However, in contrast to previous studies, we follow Fiordelisi and Marquéz-Ibañez (2013) and Campbell et al. (2001) and particularly disentangle between systemic and systematic risk. We measure the link between these different types of risk and bank’s (insurer’s) default risk. Furthermore, several studies examine interconnectedness in the financial sector, i.e. systemic linkages between banks and insurance companies. Billio et al. (2012) and Chen et al. (2013) develop measures of systemic risk and interconnectedness in the overall financial sector based on financial market data, finding a high degree of interconnectedness and that the impact of banks on insurers is much stronger than vice versa. However, these studies do not 1

Following Fiordelisi and Marquéz-Ibañez (2013), we define systemic risk as sector-specific risk (banking / insurance sector risk, respectively) and systematic risk as non-diversifiable risk (the market risk, i.e. the overall stock market).

1

disentangle between systemic and systematic risk. Hence, we analyze if the default risk of banks (insurers) can affect insurance (banking) companies, and if this effect is based on systemic or rather systematic risk. We focus our analysis on the financial crisis of 2008 and the years surrounding the crisis, as default risk in the financial sector during this period showed to have severe consequences for the overall financial industry. Analyzing the interconnectedness in the financial sector and disentangling between systematic and systemic risk with respect to default risk is important for several reasons. First, a crisis that emerges in one sector might affect only a single or a small number of firms from the same sector, the whole sector or even firms from other sectors, and the overall economy. For example, banking distress might affect firms from the insurance sector particularly strong, as some insurers hold a large amount of bank debt (Cummins and Weiss, 2014), hence bank default risk might affect insurers while the rest of the economy remains relatively unaffected. Moreover, default risk in the life insurance sector might affect other life insurers due to the fact that most life insurers hold similar asset portfolios (Schwarcz and Schwarcz, 2014), but might not affect firms from other sectors due to a lack of connectedness between these firms. In the property-liability insurance sector, firm default risk could be triggered by a large catastrophe and thus a large amount of firms from the sector might be affected, depending on the catastrophe’s severity and the spread of coverage (Cummins et al., 2002), while firms from other sectors or the rest of the economy should not be affected. In these cases, default risk would be systemic, but not systematic. Previous papers just analyzed the relations between firms from the banking and insurance sectors, without separating the effects on the overall market and other sectors. Second, our approach allows us to analyze the contagion channels of risks in the financial sector in detail. A financial institution’s default risk might affect other financial firms directly (e.g. the failure risk of a bank might affect insurers negatively due to the fact that insurers hold a large share of bank assets) or indirectly (e.g. the failure risk of a bank might have an economy-wide effect, i.e. systematic, and thus affects the 2

insurer via its impact on the overall economy). Depending on the contagion channel, regulators need to undertake different activities, i.e. focusing their regulatory affords on reducing the links between financial institutions (in case default risk is systemic) or imposing regulatory actions to decrease the connectedness between banks and the rest of the economy (if default risk is systematic) e.g. by imposing higher capital standards and thus decreasing bank lending. The study closest to ours is Fiordelisi and Marquéz-Ibañez (2013). They show that measures of individual bank risk are systemic and systematic using data from the European banking sector. However, our study extends their analysis in several aspects. First, while their study is based on banks only, we also include insurance companies in our analysis. Thus, we do not only examine the effect of banking risk on systemic and systematic risk in the banking sector, but also examine the effect of insurer’s default risk on systemic risk in the insurance sector and systematic risk. Also, we analyze spillover effects, i.e. if bank (insurance) risk affects the systemic risk in the insurance (banking) sector. While several papers (e.g. Bijsma and Muns, 2011; Buhler and Prokopczuk, 2010) analyze systemic risk in different sectors, including the banking and insurance sector, these studies do not disentangle systemic and systematic risk and also do not examine spillover effects between the two sectors. Moreover, Fiordelisi and Marquéz-Ibañez(2013) focus on commercial banks in the European Union only. We focus on the US financial sector, as the recent financial crisis emerged in this sector and thus our analysis provides valuable results regarding the crisis impact. Also, we do not restrict our analysis to commercial banks, but also include firms from the investment banking sector, as these types of banks might be engaged in more risky businesses and thus, their financial distress should rather be able to affect other companies. In addition, Fiordelisi and Marquéz-Ibañez (2013) restrict their analysis to the years 1997 until 2007, thus excluding the financial crisis of 2008 and its consequences. However, analyzing the relationship between default risk and systemic (systematic) risk is particularly interesting in times of financial 3

crisis, as the failure of Lehman Brothers and the bailout of AIG showed to affect other firms in particular in that time. Hence, our analysis includes the years 2004-2012 to examine the effect of the financial crisis and its consequences for the financial sector. We use stock price and accounting data provided by SNL financial for all US-listed banks, investment firms, property-liability insurers and life insurers for the years 2004 till 2012, using different measures of default risk for robustness. In total, our sample includes 7,941 firm-year observations. We use regression analyses and variance decomposition approaches in our analysis. In addition, we analyze if our results are different during the years 2007-2010, as these years mark the peak of the financial crisis. Furthermore, we analyze Systemically Important Financial Institutions (SIFIs)2 in more detail, as these firms should pose a significant amount of systemic and systematic risk, in particular during the financial crisis. By way of preview, our results indicate that banks’ default risk is systematic, systemic and idiosyncratic, while investment banks’ risk is only systematic (and idiosyncratic) in noncrisis times, but also systemic in crisis-times. In addition, we find negative spillover effects for both types of banks to the insurance sector. For insurance firms, we find no spillover effects to the overall economy or to the banking sector, as expected given the findings in previous papers (Billio et al., 2012 and Chen et al., 2013). However, while property-liability insurers’ default risk is only idiosyncratic, default risk in the life insurance industry is also systemic during the financial crisis. This could be due to the fact that life insurers are strongly exposed to common exposures, i.e. they hold similar asset portfolios and thus show comparable risk exposures during economic downturn. The findings contribute to the literature in several ways: First, we extent the literature on systematic risk and holistic regulation in the financial sector. In the case a single banks’

2

Systemically Important Financial Institutions have been defined by the Financial Stability Board as financial firms whose failure might trigger a financial crisis.

4

(insurers’) distress can affect other firms in the sector; one should rethink the regulation of individual firms and implement capital surcharges related to their threat for other firms. Hence, we provide additional evidence on the necessity of macro-prudential regulation in the financial industry. Furthermore

we

provide

additional

evidence

on

spillover

effects

and

interconnectedness in the financial sector. We extend the literature by providing additional evidence on the degree of interconnectedness in the US financial sector. We show that banks’ financial distress in particular can endanger other banks and insurers, indicating strong systemic linkages in the financial system. As insurers do not pose systematic risk or spillover effects to the banking sector, our results do not indicate the necessity of further regulation in the insurance sector. In addition, we disentangle between systemic and systematic effects, providing information for regulators whether regulation should be focused on the individual sectors itself or rather on the overall financial industry. We show that the spillover effects are due to systemic and systematic linkages. In addition, we use a different methodology than previous papers to analyze interconnectedness and spillovers in the financial industry. Moreover, we provide evidence for asset managers regarding the benefits of portfolio diversification. Following Fiordelisi and Marquéz-Ibañez (2013), we show that the equity premium in the US financial industry is influenced by idiosyncratic risk because default risk is systematic in the banking sector and is therefore non-diversifiable. In the insurance sector, this risk is not systematic; hence we provide evidence that default risk in the US insurance industry is still diversifiable. This has important implications for portfolio managers that manage portfolios with a large share of financial companies, in particular in times of crisis. The paper proceeds as follows. The next section provides a summary of the background and previous literature on systemic risks, idiosyncratic risk and default risk in the financial sector. The following section describes the data and methodology used in our analysis, and the ensuing section presents the results. The final section concludes. 5

2. Literature Background and Hypothesis Development 2.1. Background on Systemic Risk and Interconnectedness in the Financial Sector Literature on systemic risk and interconnectedness in the financial sector is vast and growing, in particular since the failure of Lehman Brothers and the beginning of the financial crisis in 2008. For the banking sector, several papers (e.g. Huang, Zhou and Zhu, 2009; Acharya et al.; 2010) develop measures of systemic risk. Given the consequences of the financial distress of AIG for the overall financial sector, another strand of literature (e.g. Baluch, Mutenga and Parsons, 2011; Cummins and Weiss, 2014; The Geneva Association, 2010; Harrington, 2009) analyzes systemic risk in the insurance sector. They mainly find that insurers do not create systemic risk, in particular due to the fact that they are usually not large enough and not as interconnected with each other as banks.3 Moreover, another strand of literature (e.g. Billio et al., 2012; Chen et al.; 2013; Grace, Rauch and Wende, 2014) focuses on interconnectedness and spillover effects between firms from different sectors of the financial industry. These papers analyze systemic risk in the overall financial sector (i.e. including both banks and insurers) and how firms from these sectors affect each other. The main finding of these papers is that the financial sector is highly interconnected, and that banks can affect insurers stronger than vice versa.

2.2. Background on Systematic Risk, Idiosyncratic Risk and Default Risk Disentangling firms’ stock risk into several components has been extensively studied in previous papers. Campbell et al. (2001) disaggregate the volatility of stocks into market, industry and firm level components, finding a noticeable increase in firm-level volatility over the last decades. Bartram, Brown and Stulz (2012) analyze why US stocks are more volatile than stocks from other countries, thereby disentangling between idiosyncratic and systematic

3

However, several insurance firms have been designated as systematically important financial institutions (SIFIs) by the regulatory authorities (Cummins and Weiss, 2014).

6

risk. Bekaert, Hodrick and Zhang (2012) do not find an upward trend in idiosyncratic volatility of US stocks. Analyzing systematic, systemic and idiosyncratic components of stock risk is important for several reasons. In particular, these types of risk are important for investors that fail to diversify their stock investments in a way which is recommended by portfolio theory. As only systematic risk is rewarded, investors that hold large amounts of stocks with high levels of systemic or idiosyncratic risk bear more risk than rewarded by the returns of their stocks (Campbell et al., 2001; Bartram, Brown and Stulz, 2012). This affects for example fund managers of funds specialized in a certain sector or employees and managers who hold a large portion of their wealth in their own company due to stock options or pension plans. Moreover, analyzing systematic, systemic and idiosyncratic risk is important in the context of aggregate outputs in macroeconomic models, option pricing and applying event studies (Campbell et al., 2001). Regarding default risk, several papers analyze if default risk is systematic or idiosyncratic (e.g. Asquith, Gertner and Scharfstein, 1994; Opler and Titman, 1994; Dichev, 1998). However, these papers all exclude financial firms from their analyses, while our analysis focuses on the financial service sector and all its specifics that might influence the effect of default risk on the different components of risk (e.g. interconnectedness in the banking sector). Moreover, these studies do not examine if default risk is also systemic, while we in particular disentangle between systematic, systemic and idiosyncratic risk. Furthermore, these studies do not include the period of the recent financial crisis, while our analysis focuses on that period given the prominent role of banking risk in these years.

2.3. Hypothesis Development Previous papers showed the importance of research on systemic risk and interconnectedness in the financial sector and disentangling overall firm risk into its 7

components. However, previous literature has not analyzed the question whether firm default risk in the US financial sector is systematic, systemic or idiosyncratic and if spillover effects between the banking and insurance sectors exist with respect to these types of risk. Fiordelisi and Marquéz-Ibañez (2013) analyze the impact of default risk on systematic, systemic and idiosyncratic risk in the European banking sector in pre-crisis times, but they do not consider the different components of the financial sectors regarding their different business models and contagion effects between these firm types in times of financial crisis. Similarly, Bijlsma and Muns (2011) and Buhler and Prokopczuk (2010) analyze systemic risk in the several sectors of the economy, but do not analyze if spillover effects between financial firms are systemic or systematic and do not examine the effect of financial institutions’ default risk on the overall economy. In our research, we first examine the relation between default risk and systematic risk and systemic risk in the same sector for banks and insurers separately.4 Subsequently, we analyze spillover effects, i.e. how default risk in one sector of the financial industry affects systemic risk in another sector. A separate analysis of the different firms of the financial sector regarding the effect of default risk of banks, investment banks, life insurers and property-liability insurers on systemic and systematic risk is desirable with respect to their different business models and their role for the economy.5 Banks (and to a certain degree also investment banks) are usually larger than insurance firms and, due to their role of credit supplier for the economy, also stronger interconnected with firms from the overall economy. Additionally, the recent financial crisis provided evidence for significant spillover effects from banks to other sectors of the overall economy, hence indicating that banks’ default risk is systematic. Moreover, 4

We do not discuss hypothesis regarding idiosyncratic risk because we assume that this risk should always be affected by the firm’s default risk. 5 In particular, Cummins and Weiss (2014) discuss several indicators of systemic risk and factors contributing to vulnerability to systemic risk identified in Financial Stability Board (2009). The primary indicators contain size, interconnectedness and lack of substitutability, while the contributing factors contain leverage, liquidity risks and maturity mismatches, complexity, and government policy and regulation

8

banks are usually closely interconnected with each other, e.g. due to inter-bank activities and security lending (Cummins and Weiss, 2014). Thus, given this strong connection between banks, default risk of banks is likely to affect other banks, indicating that this risk is systemic in the banking sector. This can also hold for banks from the investment banking sector, as financial problems in the regular interbank market can lead to runs on these shadow banks (Cummins and Weiss, 2014). Moreover, investment firms strongly rely on short-term borrowing, making them strongly vulnerable for crises in the “normal” banking sector. Hence, our first hypothesis states Hypothesis 1: Banks’ default risk is systematic. Moreover, banks’ default risk is systemic in the banking sector.

Regarding insurance firms, we do not expect that default risk of insurers is systematic. Compared to banks, insurers are smaller, less interconnected with other parts of the economy and do not suffer from a comparable lack of substitutability as the banking sector (Cummins and Weiss, 2014). This is consistent with the view that insurers have only moderate effects on firms from other sectors, compared to banks.6 However the question if default risk in the insurance sector is systemic is ambiguous and might be different for life and property-liability insurers. First, insurers are less interconnected with each other than banks given the lack of an equivalent to the interbank market for insurers. Insurers might have business relations with reinsurers, but in general to a much lesser extent with other primary insurers. Thus, we might assume that default risk in the insurance sector is not systemic. However, the literature provides several arguments for the existence of systemic risk in the insurance sector, even though it might not spill over to other sectors of the overall economy and is thus not systematic. For life insurers, the fact that most life insurers hold similar asset portfolios and 6

Schwarcz and Schwarcz (2014) however state that, even though individual insurers do not endanger the overall economy, the insurance sector as a whole might be a source of systematic risk due to the correlation among individual insurers. In addition, while the insurers’ core business activities are not considered as systemically relevant, non-core activities like underwriting CDS can contribute to systemic risk.

9

follow similar strategies (Schwarcz and Schwarcz, 2014) might not only endanger a single firm in case of default risk, but the overall sector would be affected at the same time due to their comparable exposure, would make default risk in the life insurance sector systemic. In the property-liability sector, the firms might have strong, common exposures towards catastrophic events. Depending on their severity and the spread of coverage (Cummins et al., 2002), many insurers might be burdened with high claim payments that endangers their financial health at the same time.7 Thus, default risk would affect the property-liability insurance sector, while firms from other sectors remain unaffected, hence posing systemic, risk in the property-liability sector, but not systematic risk in the overall economy. In addition, events like the liability crisis in the US insurance sector in the 1980s might affect the property-liability sector, but not firms from other sectors. Overall, our second hypothesis states: Hypothesis 2: Insurers’ default risk is not systematic. Whether insurers’ default risk is systemic is ambiguous and depends on the insurance sector (life or property-liability).

Furthermore, following previous papers, we develop hypothesis regarding spillover effects, i.e. how default risk in one sector of the financial industry affects systemic risk in another sector. Regarding spillover effects from banks and investment banks to the insurance sector, previous papers (Billio et al., 2012; Chen et al., 2013) found that banks affect insurance firms strongly. Again, there might be differences between life and property-liability insurers regarding the extent to which they are affected by default risk in the banking sector. Life insurers hold much larger shares of bonds of banks in their portfolios than propertyliability insurers, and the life insurance industry shows a much larger correlation to the banking sector (Cummins and Weiss, 2014).8 In addition, the property-liability insurance

7 8

However, Cummins et al. (2002) state that the insurance industry could handle even very large catastrophes. In addition, property-liability insurers are on average better capitalized than life insurers.

10

sector was much less affected by recent financial crises than by sector-specific crises (e.g. liability crises or catastrophes). Overall, our third hypothesis states: Hypothesis 3: Default risk in the banking sector is systemic in the insurance sector.

For insurer default risk, we do not expect systemic consequences in the banking sector. Previous papers (Billio et al., 2012; Chen et al.; 2013; Grace, Rauch and Wende, 2014) find that insurance firms can affect banks to a much lesser degree than vice versa. Also, neither previous literature nor historical experience provides arguments that the insurance sector can significantly endanger the banking industry. Hence, we state our fourth hypothesis as: Hypothesis 4: Default risk in the insurance sector is not systemic in the banking sector.

3. Data and Methodology 3.1. Data We use daily stock price returns and financial statement data for all publicly traded banks (excluding thrifts), investment banks,9 life insurers and property-liability insurance companies in the US provided by SNL Financial. Our observation period covers the years from 2004 to 2012. We drop firms with negative or missing total assets, equity and revenue. Furthermore, the data is winsorized at the 99% percentile to remove the effect of extreme outliers. Moreover, we drop firms if the stock prices are not available during our observation period, as the estimation of our systemic and systematic risk components requires daily stock returns for each firm individually.

9

Our sample of investment banks covers all listed broker-dealers and banks with a substantial amount of investment banking activities.

11

3.2. Methodology We first follow Fiordelisi and Marquéz-Ibañez (2013) and estimate the systemic, systematic and idiosyncratic components in our analysis.10 However, while previous papers estimate these components in a single regression, we estimate the systemic and systematic components separately, as most indices are highly correlated, to avoid the problem of multicollinearity. Hence, we estimate the risk components using the following regression analyses: Ri,t= MKTi,t* RM,t+ ei,t

(1)

Ri,t= INDbank,i,t* Rbank,i,t+ ei,t

(2)

Ri,t=INDins(life),i,t* Rins(life),i,t+ ei,t

(3)

Ri,t= INDins(pc),i,t* Rins(pc),i,t+ ei,t

(4)

With: where Ri,t are the daily stock market logarithmic abnormal returns11 from each bank (insurer) i. RM,t are the daily stock market abnormal returns from the broad stock market index (the S&P 500).12 Rbank,i,t, Rins(life),i,t andRins(pc),i,t are the daily stock market abnormal returns from the banking (insurance) industry stocks listed, using indices from the respective sector.13 The term ei,t is the bank(insurance) specific residual.14 For each bank (insurance) i, our systematic and systemic risk components (MKTi,t, INDbank,i,t, INDins(life),i,t and INDins(pc),i,t respectively) are calculated by running separate regressions on daily data for each year separately. In this way we can match our risk proxies with the other individual bank (insurer)

10

Similar approaches to disentangle between systemic, systematic and idiosyncratic risk components are for example used in Campbell et al. (2001) or Bartram, Brown and Stulz (2012). 11 We use US 10 year T-note yields to calculate abnormal returns throughout our analysis. 12 Given that Fiordelisi and Marquéz-Ibañez (2013) use a sample of banks from different countries of the European Union, their broad stock market indices and banking industry indices vary between countries, denoted by subscript “c”. As we focus on firms from the US only, our indices are the same for all firms in our analysis. 13 For the banking sector index, we use the S&P 500 Bank index; for life insurers, the S&P 500 Life Insurance index; and for property-liability insurers the S&P 500 Property-Liability Insurance index. 14 The firm-specific component, in contrast to the other components, is estimated from a regression including all indices to capture the residual, non-systemic and non-systematic components of stock risk.

12

variables. Moreover, following e.g. Brandt et al. (2009) or Campbell et al. (2001), we measure idiosyncratic risk (FIRMi,t) as firm-specific volatility of the respective stock. To measure if measures of a financial institutions’ distress risk can affect systemic (IND), systematic (MKT) and idiosyncratic (FIRM) risk, we follow Fiordelisi and MarquézIbañez (2013) and use the following model for each bank, investment bank, life insurer and property-liability insurer:15

MKTi,t= α + β * RISKi,t+ Σγi * CONTj,i,t+ εi,t

(5)

INDbank,i,t= α + β * RISKi,t+ Σγi * CONTj,i,t+ εi,t

(6)

INDins(life),i,t= α + β * RISKi,t+ Σγi * CONTj,i,t+ εi,t

(7)

INDins(pc),i,t= α + β * RISKi,t+ Σγi * CONTj,i,t+ εi,t

(8)

FIRMi,t= α + β * RISKi,t+ Σγi * CONTj,i,t+ εi,t

(9)

Using the firms’ Z-score16 as an indicator for default risk (RISK)17 and a vector of control variables (CONT) that may influence systemic and systematic risk. Hence, while previous papers focus on the banking sector only (Fiordelisi and Marquéz-Ibañez; 2013) or do not disentangle between systemic and systematic risk (e.g. Billio et al. 2012; Chen et al., 2013), our approach allows us to not only analyze if bank (insurer) default risk affects the banking (insurance) sector, but at the same time we can also analyze spillover effects from the banking to the insurance sector vice versa, thereby disentangling between systemic and systematic risk.

15

Given that our dataset is a panel, we use clustered standard errors on firm level. The Z-Score has been used in various papers to measure a financial institution’s default risk (see e.g. Boyd et al., 2006; Berger, Klapper and Turk-Ariss, 2009; Uhde and Heimeshoff, 2009). The Z-Score is defined as the ratio of the sum of the firm’s return on asset plus its capital ratio divided by the standard deviation of its return on assets. 17 Fiordelisi and Marquéz-Ibañez (2013) use Moody’s probability of default ratings (PDR) and 1-Year ahead Expected Default Frequency as alternative measures of firm risk. However, as these measures are not available to most firms in our analysis, we mainly rely on the Z-Score in our analysis in order to not restrict our sample to too few observations. 16

13

We conduct several robustness checks. First, we analyze if our results change during the financial crisis of 2008, as default risk in the financial sector is particularly important during financial crises. Hence, we repeat our regression analyses for the years 2007-2010 only, which we define as the “crisis period”. In addition, we follow Turk-Ariss (2010) and use an alternative measure of financial institutions’ default risk to see if our findings are robust to the inclusion of another measure of risk. This alternative measure of bank stability (“Risk adjusted ROE”) is defined as ROE divided by its standard deviation. Given that financial institutions’ equity capital strongly suffered during the recent financial crisis, a measure based on equity (instead of assets as the ordinary Z-score) will provide additional evidence regarding our research question and the firm’s low equity level in crisis times. Finally, we restrict our sample to all firms designated as Systemically Important Financial Institutions (SIFIs) by the Financial Stability Board.18 We will examine if these firms’ riskiness can be systemic and systematic in crisis and non-crisis times and thus provide evidence on their threat to the overall economy and the financial sector. For our control variables, we follow Fiordelisi and Marquéz-Ibañez (2013) and use the same set of variables that might affect systematic risk. The vector of control variables contains factors that capture the firms’ efficiency, market power, business model and size.19 Regarding efficiency, financial institutions might adjust their risk-taking profile by increasing profits and reducing costs (Firodelisi, Marquéz-Ibañez and Molyneux, 2011), Hence, we include the cost to income ratio (defined as operating costs divided by operating income) for banks and investment banks. For insurers, we use commonly used measures of efficiency in the insurance industry: for life insurers, we use the insurers expense ratio

18

We include firms classified as Global Systemically Important Banks (G-SIBs), Domestic Systemically Important Banks (D-SIBs) and Global Systemically Important Insurers (G-SIIs). 19 Fiordelisi and Marquéz-Ibañez (2013) include industry-level and macroeconomic factors in addition to firmspecific variables. However, as our dataset contains only firms from one country (the US) while their analysis is based on a cross-country panel, we do not include industry-level or macroeconomic variables as they would be identical for each firm in a given year in our dataset.

14

(defined as expenses divided by premiums) and for property-liability insurers the combined ratio (defined as losses and expenses divided by premiums). To measure market power (Salas and Saurina, 2003; Turk-Ariss, 2010), we include a Lerner index, i.e. a measure of the firm’s ability to set prices above its marginal costs.20 Moreover, we include each firm’s individual market share, i.e. its revenue divided by the sector’s revenue, instead of the Herfindahl-Hirschmann Index as we include firms from a single country only. To account for the relation the financial institutions’ business models and systematic risk (Bertrand and Schroar, 2003), we include the ratio of non-interest to operating income ratio as a measure of income diversification for banks. For investment banks, we use a Herfindahl index based on different sources of revenue (e.g. underwriting fees, underwriting fees or interest income). For life and property-liability insurers, we include a Herfindahl index based on premium income from the different major lines of insurer business (Berry-Stölzle et al., 2012). In addition to income diversification, we also account for liability diversification. For banks and investment banks, we follow Fiordelisi and Marquéz-Ibañez (2013) and use non-deposit to total deposits ratio. For life and property-liability insurers, we include a Herfindahl index based on reserves from the different major lines of life insurer business. To measure the financial institutions’ size, we include the natural logarithm of the firm’s total assets (Fiordelisi and Marquéz-Ibañez, 2013). Moreover, we control for the financial institutions’ capitalization by including the natural logarithm of net equity in our analysis (Fiordelisi and Marquéz-Ibañez, 2013). For life insurers, we use surplus plus asset valuation reserve as a measure of equity, and for property-liability insurers we add unearned premium reserves to surplus (Cummins and Weiss, 2012).

20

See Appendix A for details on the calculation of the Lerner indices for the different industries.

15

4. Results 4.1.

Descriptive Statistics Table 1 presents the summary statistics for all variables used in our regression

analyses for the years 2004-2012 for all banks, investment banks, life insurers and propertyliability insurance companies. The main variable of interest, Z-Score, shows average values between 4.642 (investment banks) and 9.4 (banks), depending on the industry. These values are comparable to the Z-Scores used in comparable papers (e.g. Fiordelisi and MarquézIbañez, 2013; Hryckiewicz, 2014). Moreover, it can be seen that banks are on average larger than insurance firms. [insert table 1 here] [insert table 2 here]

4.2.

Variance Decomposition We begin our analyses by presenting the results of an approximation of the variance

decomposition as in Campbell et al. (2001) for each year separately. In this way, we provide further evidence on the development of the systemic, systematic and idiosyncratic components of stock market risk with respect to our analysis. Table 3 shows the variance decompositions for the years 2004-2012. The decomposition distributes the stock price volatility into three components: MKT denotes market-wide risk, IND the variance of the respective sector (banking and insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector (bank, investment bank, life insurer or property-liability insurer).21 [insert table 3 here] [insert table 4 here] 21

Given that idiosyncratic risk is much higher than the other two components in several years, table 8 provides three figures: Panel A shows all three components; Panel B shows only systematic and systemic components; and Panel C shows idiosyncratic components only. This might facilitate the interpretation of the tables for the reader. Table 8 provides details on the risk components regarding their size.

16

Consistent with Fiordelisi and Marquéz-Ibañez (2013), who find higher risk in times of the Asian and South American crises in the late 1990s, and Bekaert, Hodrick and Zhang (2012) who find idiosyncratic risk skyrocketing in crises times (e.g. the dot-com bubble in the early 2000s), we find strong increases for all three risk components between 2008 and 2009, the years of the financial crisis. All three components decrease again after 2009, but remain on a relatively higher level than before the crisis. In particular, as expected by previous papers, we can see higher levels of systemic risk in the banking sector compared to the insurance sector (Panel B). Also, we see higher idiosyncratic risk for banks compared to insurers (Panel C). This shows the importance of an analysis of systemic, systematic and idiosyncratic risk in times of the financial crisis of 2008. It also is an indicator the riskiness of (financial sector) stocks during that period, represented by low stock market valuations.

4.3.

Empirical Results Tables 5-8 present the preliminary results of our empirical analysis for banks,

investment banks, life insurers and property-liability insurers separately. We first discuss the results on the relation between default risk and systematic risk and systemic risk in the same sector for banks and insurers separately. Subsequently, we discuss the results on hypotheses regarding spillover effects, i.e. how default risk in one sector of the financial industry affects systemic risk in another sector. Sector-Specific Analysis Table 5 shows that default risk in the banking sector is systematic, systemic and idiosyncratic, as indicated by the significantly negative estimates for Z-Score in the columns for MKT (systematic), IND_Bank (systemic) and FIRM (idiosyncratic). Hence, we find support for our first hypothesis. This is consistent with the findings in previous papers and the experience during the recent financial crisis. Regarding investment banks (table 6), we find that default risk is systematic and idiosyncratic, but surprisingly not systemic, thus only 17

limited support for hypothesis 1. Hence, the effect of default risk on other banks is smaller than in the commercial banking sector. One potential explanation might be that investment banks are usually not linked with each other as closely as commercial banks, but still affect the overall economy due to their role as shadow banks. However, a closer examination of this finding is necessary. [insert table 5 here] [insert table 6 here] Regarding life insurers, we find that default risk is not systematic, hence providing evidence for our second hypothesis (table 7). In addition, we find that default risk is neither systemic. Similarly, for property-liability insurers, table 8 indicates that default risk is only idiosyncratic, but neither systematic nor systemic. Hence, our results indicate that default risk in the insurance sector threats only the firms itself, but does not spill over to other firms from the same sector or the overall economy. [insert table 7 here] [insert table 8 here] Analysis of Spillover Effects Next, we examine spillover effects in the US financial sector. For banks (table 5), our results indicate spillover effects from banks to both life and property-liability insurers (the coefficient of Z-Score in column 4 and 5, IND_LH and IND_PC, is significant). Hence, we find evidence for our third hypothesis, consistent with prior papers. The same holds for default risk in the investment banking sector, which is systemic in the insurance sector (table 6). Regarding life and property-liability insurers (table 7 and 8, respectively), our results do not indicate spillover effects from insurers to banks, hence providing evidence for hypothesis 4. Robustness: Analysis of the Crisis Period

18

We extent our analysis by focusing on the relation between default risk and systemic and systematic risk during the years 2007-2010, as the recent crisis showed how financial institutions’ default risk can affect other firms and sectors. Tables 9-12 show the results. For banks (table 9) and property-liability insurers (table 12), the results are consistent with the non-crisis period, hence providing further evidence for hypotheses 1-4. However, investment banks’ default risk is also systemic during the crisis (table 10), providing support for hypothesis 1. This might be due to the fact that financial problems during the crisis in the regular interbank market can lead to runs on these shadow banks (Cummins and Weiss, 2014). Moreover, investment firms strongly rely on short-term borrowing, making them strongly vulnerable for crises in the “normal” banking sector. For life insurers, we find that default risk is also systemic during the crisis (table 11). Hence, we find that the risk of one life insurer can spill over to other life insurers – potentially due to their similar asset portfolios and similar business strategies (Schwarcz and Schwarcz, 2014). [insert table 9 here] [insert table 10 here] [insert table 11 here] [insert table 12 here]

Robustness: Alternative Measure of Default Risk Next, we follow Turk-Ariss (2010) and use an additional measure of financial institutions’ default risk to see if our findings are robust to the inclusion of another measure of default risk during the financial crisis. Table 13-16 show the results of the regression analyses using risk-adjusted ROE as a measure of bank stability as the main variable of interest. The results are consistent with the previous results, except for investment banks (table 14). Using this measure, the results indicate that default risk in this sector is neither systematic nor systemic during the crisis. Thus, a closer examination of this finding is necessary. 19

[insert table 13 here] [insert table 14 here] [insert table 15 here] [insert table 16 here]

Robustness: Analysis of Systemically Important Financial Institutions (SIFIs) Finally, we analyze the results for all Systemically Important Financial Institutions (SIFIs) in our dataset separately Table 13-14. We find that SIFIs pose systematic and systemic risk in the insurance sector during non-crisis and crisis times, hence providing additional evidence regarding their threat to the overall economy in all market environments. However, our results indicate that SIFIs’ default risk is only systemic in the banking sector in crisis times. Thus, a closer examination of this finding is necessary. [insert table 17 here] [insert table 18 here]

5. Conclusion We examine the effect of the financial institutions’ (banks and insurance companies) default risk on systemic and systematic risk in the US financial sector. In contrast to previous studies, we follow Fiordelisi and Marquéz-Ibañez (2013) and Campbell et al. (2001) and particularly disentangle between systemic and systematic risk, and analyze if bank’s (insurer’s) default risk affects these types of risk. Furthermore, we analyze the interconnectedness between banks, investment banks, life insurers and property-liability insurers by analyzing if default risks can spillover from one sector to another sector. We use a dataset of 7,941 firm year observations for the years 2004 until 2012 and regression analyses. Consistent with our hypotheses, we find that default risk in the banking 20

sector is systematic, systemic and idiosyncratic, and only idiosyncratic in the life insurance sector and the property-liability insurance industry. However, when focusing on the years surrounding the financial crisis, we find that life insurers’ default risk is also systemic. Moreover, consistent with previous research (e.g. Billio et al., 2012; and Chen et al., 2013), we find spillover effects from the banking sector to both insurance sectors, while we do not find the reverse. Our results mostly hold for an alternative measure of default risk and the subsample of firms in our analysis that have been classified as Systemically Important Financial Institutions (SIFIs). We contribute to the literature on systemic risk in the financial sector in several ways. First, we extend the literature on systematic risk and holistic regulation in the financial sector. In the case a single banks’ (insurers’) distress can affect other firms in the sector; one should rethink the regulation of individual firms and implement capital surcharges related to their danger for the other firms. This provides further support for the implementation of a holistic regulatory approach that does not only consider the “stand-alone” regulation of individual companies, but rather taking into consideration the connections within the financial system. We also provide further evidence on the existence of these interconnections within the financial service industry. Finally, we also provide implications for investors by providing information on the riskiness of investments in the US financial sector. Given that default risk in the banking sector is systematic, it cannot be diversified away, while investors can diversify default risk in the insurance sector as it is not systematic.

21

References Acharya, V.V., L.H. Pedersen, T. Philippon, M. Richardson, 2010. Measuring Systemic Risk. Working Paper. Asquith, P., R. Gertner, D. Sharfstein, 1994. Anatomy of financial distress: an examination of junk-bond issuers. Quarterly Journal of Economics 109, 625-658. Baluch, F., S. Mutenga, C. Parsons, 2011. Insurance, Systemic Risk and the Financial Crisis. Geneva Papers on Risk and Insurance - Issues and Practice 36, 126-163. Bartram, S.M., G. Brown, R.M. Stulz, 2012. Why are U.S. Stocks more volatile? Journal of Finance 67(4), 1329-1370. Bekaert, G., R.J. Hodrick, X. Zhang, 2012. Aggregate Idiosyncratic Volatility, Journal of Financial and Quantitative Analysis 47(6), 1155-1185. Berger, A., L.F. Klapper, R. Turk-Ariss, 2009. Bank competition and financial stability. Journal of Financial Services Research 35, 99-118. Berry-Stölzle, T.R., A.P. Liebenberg, J.S. Ruhland, D.W. Sommer, 2012. Determinants of Corporate Diversification: Evidence from the Property–Liability Insurance Industry. Journal of Risk and Insurance 79(2), 381-413. Bertrand, M., A. Schoar, 2003. Managing with style: the effect of managers on firm policies. The Quarterly Journal of Economics 118(4), 1169-1208. Bijlsma, M., S. Muns, 2011. Systemic Risk across Sectors; Are Banks Different? Netherlands Bureau for Economic Policy Analysis CPB Discussion Paper 175. Billio, M., M. Getmansky, A.W. Lo, L. Pelizzon, 2012. Econometric Measures of Connectedness and Systemic Risk in the Finance and Insurance Sectors. Journal of Financial Economics 194(3), 535-559. Boyd, J., G. De Nicolo, A.M. Jalal, 2006. Bank Risk-Taking and Competition Revisited: New Theory and New Evidence. International Monetary Fund Working Paper 06/297. Brandt, M.W., A. Brav, J.R. Graham, A. Kumar, 2009. The Idiosyncratic Volatility Puzzle: Time Trend or Speculative Episodes? Review of Financial Studies 23(2), 863-899. Buhler, W., M. Prokopczuk, 2010. Systemic risk: is the banking sector special? Mimeo. Campbell, J.Y., M. Lettau, B.G. Malkiel, Y. Xu, 2001. Have individual stocks become more volatile? An empricial exploration of idiosyncratic risk. Journal of Finance 56(1), 1-43. Chen, H., J.D. Cummins, K.S. Viswanathan, M.A. Weiss, 2014. Systemic Risk and the InterConnectedness between Banks and Insurers: An Econometric Analysis. Journal of Risk and Insurance 81(3), 623-652. Cummins, J.D., N. Doherty, A. Lo, 2002. Can Insurers Pay for the ‘‘Big One’’? Measuring the Capacity of the Insurance Market to Respond to Catastrophic Losses. Journal of Banking & Finance 26(2-3), 557-583. Cummins, J.D., M.A. Weiss, 2012. Analyzing firm performance in the insurance industry using frontier efficiency and productivity methods. Handbook of insurance. Springer New York, 795-861. Cummins, J.D., M.A. Weiss, 2014. Systemic Risk and the U.S. Insurance Industry. Journal of Risk and Insurance 81(3), 489-527.

22

Dichev, I., 1998. Is the risk of bankruptcy a systematic risk? Journal of Finance 53,11411148. Financial Stability Board, 2009. Guidance to Assess the Systemic Importance of Financial Institutions, Markets and Instruments: Initial Considerations (Basel, Switzerland: Financial Stability Board). Fiordelisi, F., D. Marqués-Ibañez, 2013. Is bank default risk systematic? Journal of Banking & Finance 37, 2000-2010. Fiordelisi, F., D. Marquéz-Ibañez, P. Molyneux, 2011. Efficiency and risk in European banking. Journal of Banking and Finance 35 (5), 1315-1326. Grace, M.F., J. Rauch, S. Wende, 2014. Systemic Risk and Interconnectedness in the Financial Industry: Implications on the Regulation of Financial Conglomerates, Working Paper. Harrington, S. E., 2009. The Financial Crisis, Systemic Risk, and the Future of Insurance Regulation, Journal of Risk and Insurance 76(4): 785-819. Hryckiewicz, A., 2014. What do We Know about the Impact of Government Interventions in the Banking Sector? An Assessment of Various Bailout Programs on Bank Behavior. Journal of Banking & Finance 46(9), 246-265. Huang, X., H. Zhou, H. Zhu, 2009. A Framework for Assessing the Systemic Risk of Major Financial Institutions. Journal of Banking and Finance 33(11), 2036-2049. Opler, T., S. Titman, 1994. Financial distress and corporate performance. Journal of Finance 49, 1015–1040. Salas, V., J. Saurina, 2003. Deregulation, market power and risk behavior in Spanish banks. European Economic Review 47, 1061-1075. Schwarcz, D., S.L. Schwarcz, 2014. Regulating Systemic Risk in Insurance, Working Paper. The Geneva Association, 2010. Systemic Risk in Insurance – An Analysis of Insurance and Financial Stability. Turk-Ariss, R., 2010. On the implications of market power in banking: evidence from developing countries. Journal of Banking and Finance 34(4), 765-775. Uhde, A., U. Heimeshoff, 2009. Consolidation in banking and financial stability in Europe: empirical evidence. Journal of Banking and Finance 33(7), 1299-1311.

23

Appendix A

This section describes the calculation of the Lerner Index of Monopoly Power for each financial institution’s market power.22 This index measures the firm’s ability to set prices above marginal costs. A Lerner index equal to 0 indicates perfect competition, while a Lerner index of 1 indicates a price-monopoly. We follow Fiordelisi and Marquéz-Ibañez (2013) and Turk-Ariss (2010) and calculate the Lerner index (LER) as −

=

With p as the price of output Q and MC is the marginal cost. To calculate MC, the following translog function is estimated: =

+

1 + + 2

+

×

1 × " 2

+ "

#

+

+ $

+

+ ψ

×

+ &

Where TC is total cost; α, β, δ, γ, ρ, t, θ, ψ are coefficients to be estimated; εit is an error term. Pj are the firm’s input prices. MC are finally defined as =

= '

+

+

+ $

+ & (

Following Fiordelisi and Marquéz-Ibañez (2013) and Turk-Ariss (2010), we calculate the efficiency-adjusted Lerner index to avoid endogeneity bias that result from a single structural model as =

)

*+

)

−

*+

*+

Where AR denotes the firm’s average revenues. For banks and investment banks, we use the approach in Fiordelisi and MarquézIbañez (2013) and use the same set of variables in our analysis. We use total revenue (interest plus non-interest income) divided by the firm’s total assets for p and total assets as Q (assuming a single output). For Pj, three input prices are defined: P1 is the price of labor; P2 is the price of physical capital; and P3 is the price of of funds. Table 19 provides an overview on the variables included and table 20 provides definitions of these variables for banks (Panel A) and investment banks (Panel B). 22

A detailed discussion on the Lerner index is out of the scope of this paper. Please refer to Turk-Ariss (2010) for further information.

24

[insert table 19 here] [insert table 20 here]

For insurance firms, we use a different set of variables, given the large amount of literature on efficiency studies in the insurance sector (see Cummins and Weiss, 2012, for a comprehensive literature review.). Hence, detailed information on the insurers’ prices and quantities for inputs and outputs are available. This allows us to capture the specific characteristics of insurance business and will increase the quality of our estimation results. Following Cummins and Weiss (2012), we use 6 different types of output quantities and prices for life insurers in our analysis: one for each major lines of business offered by life insurers – individual life insurance, individual annuities, group life insurance, group annuities, and accident and health insurance – and one for the element of intermediation. We use 4 different input quantities and prices for life insurers: We account for administrative labor, agent labor, materials and financial capital. For property-liability insurers, we use different output

quantities and prices for personal lines short-tail losses, personal lines long-tail losses, commercial lines short-tail losses, and commercial lines long-tail losses, and one for the element of intermediation.23 For Inputs, we use the same types of quantities and prices as for life insurers. Table 21 provides an overview on the variables included and table 22 provides

definitions of these variables for life insurers (Panel A) and property-liability insurers (Panel B). [insert table 21 here] [insert table 22 here]

23

We do not provide detailed description or discussion on the variables that we include in our calculation as this would be out of the scope of this paper. Please refer to Cummins and Weiss (2012) for a detailed discussion and literature review on efficiency studies and inputs and outputs in the insurance industry.

25

Appendix B Table 1: Variables for Regression Analyses–Summary Statistics The table shows the summary statistics for banks (Panel A), investment banks (Panel B), life insurers (Panel C) and property-liability insurers (Panel D) used in our regression analyses. The variables are defined in table 2.

Variable - Banks Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Risk adjusted ROE Variable - Investment Banks Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Risk adjusted ROE Variable - Life Insurers Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Risk adjusted ROE Variable - PC Insurers Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Risk adjusted ROE

Panel A Obs Mean 6674 9.42204 6684 0.483738 6652 0.731384 6652 0.196162 6696 0.271353 6709 20.2417 6689 17.9131 6685 0.000492 6529 10.11639 Panel B Obs Mean 474 4.64531 483 0.699741 465 0.737823 475 0.395376 474 0.222845 476 21.2396 450 19.562 483 0.019192 474 2.88699 Panel C Obs Mean 325 5.13834 329 0.7333 327 13.2107 329 0.592883 329 0.839721 326 17.1613 326 15.1912 329 0.000103 325 2.104 Panel D Obs Mean 439 4.02471 445 0.775964 445 106.275 445 0.356173 445 0.482349 443 15.9869 444 13.9358 445 0.000069 442 1.11208

26

Std. Dev. Min Max 10.0512 0.040335 61.3875 0.132823 0 0.76181 0.197634 0.388907 1.66095 0.113125 0.011104 0.630807 0.161372 0.090791 1.16081 1.54788 17.7631 25.7472 1.62578 12.879 26.1911 0.002223 3.70E-06 0.016928 169.6895 -130.9017 11427.75 Std. Dev. Min 6.13248 0.246666 0.185099 0.205498 0.290193 2.89414 2.52879 0.066262 6.21261

-1.21223 0 0.001329 0 0 14.7927 13.7851 3.70E-06 -38.5197

Max

Std. Dev. Min 4.98733 0.285992 7.90557 0.251945 0.193109 1.80049 1.65799 0.000164 3.873

0.205777 0 1.24053 0.240012 0.113973 12.9436 10.8951 8.00E-07 -10.034

48.7365 1 1.66095 0.98628 1.16081 27.7442 25.0569 0.43661 37.0763 Max 30.8336 1 34.0058 1.57239 1.00748 20.5643 18.7266 0.000893 30.757

Std. Dev. Min Max 3.17849 -0.234163 15.5928 0.205821 0 1 44.0516 41.7427 394.057 0.284205 0.092408 1 0.282368 0.164636 1 1.85883 11.786 20.5579 1.58002 1.03E+01 17.9234 0.000125 5.10E-07 0.000539 1.7736 -6.56244 11.6651

Table 2: Variables for Regression Analyses -Definitions The table shows the definitions of the variables for banks (Panel A), investment banks (Panel B), life insurers (Panel C) and property-liability insurers (Panel D) used in our regression analyses.

Variable - Banks Z-Score Risk adjusted ROE Lerner Index

Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Variable - Investment Banks Z-Score Risk adjusted ROE Lerner Index

Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Variable - Life Insurers Z-Score Risk adjusted ROE Lerner Index

Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Variable - PC Insurers Z-Score Risk adjusted ROE Lerner Index

Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share

Panel A Description Ratio of the sum of each bank return on assets (ROA) and its capital ratio (E/TA) divided by the standard deviation of the return on assets (σROA) Ratio of the sum of each bank return on equity (ROE) divided by the standard deviation of the return on assets (σROE) Indicator of bank market power, measuring the extent to which a bank is able to fix a price above its marginal cost. To address endogeneity concerns, we calculate the efficiency-adjusted Lerner index estimating bank efficiency and market power simultaneously Operating costs to operating income ratio Non-interest to operating income ratio Non-deposit to total deposits ratio Natural logarithm of total assets Natural logarithm of net equity Firm revenue divided by industry revenue Panel B Description Ratio of the sum of each bank return on assets (ROA) and its capital ratio (E/TA) divided by the standard deviation of the return on assets (σROA) Ratio of the sum of each bank return on equity (ROE) divided by the standard deviation of the return on assets (σROE) Indicator of bank market power, measuring the extent to which a bank is able to fix a price above its marginal cost. To address endogeneity concerns, we calculate the efficiency-adjusted Lerner index estimating bank efficiency and market power simultaneously Operating costs to operating income ratio A Herfindahl index based on different sources of revenue (e.g. underwriting fees, underwriting fees, interest income…) Non-deposit to total deposits ratio Natural logarithm of total assets Natural logarithm of net equity Firm revenue divided by industry revenue Panel C Description Ratio of the sum of each insurer return on assets (ROA) and its capital ratio (E/TA) divided by the standard deviation of the return on assets (σROA) Ratio of the sum of each insurer return on equity (ROE) divided by the standard deviation of the return on assets (σROE) Indicator of insurer market power, measuring the extent to which an insurer is able to fix a price above its marginal cost. To address endogeneity concerns, we calculate the efficiency-adjusted Lerner index estimating insurer efficiency and market power simultaneously The insurer's expense ratio. A Herfindahl index based on premium income from the different major lines of life insurer business. A Herfindahl index based on reserves from the different major lines of life insurer business. Natural logarithm of total assets Natural logarithm of net equity (surplus plus asset valuation reserve) Firm revenue divided by industry revenue Panel D Description Ratio of the sum of each insurer return on assets (ROA) and its capital ratio (E/TA) divided by the standard deviation of the return on assets (σROA) Ratio of the sum of each insurer return on equity (ROE) divided by the standard deviation of the return on assets (σROE) Indicator of insurer market power, measuring the extent to which an insurer is able to fix a price above its marginal cost. To address endogeneity concerns, we calculate the efficiency-adjusted Lerner index estimating insurer efficiency and market power simultaneously The insurer's combined ratio. A Herfindahl index based on premium income from the different major lines of p/c insurer business. A Herfindahl index based on reserves from the different major lines of p/c insurer business. Natural logarithm of total assets Natural logarithm of net equity (surplus plus asset unearned premium reserves) Firm revenue divided by industry revenue

27

Table 3: Systematic, Systemic (Sector) and Idiosyncratic Risk - Graphs The graphs show variance decompositions as in Campbell et al. (2001) for the years 2004-2012. The decomposition distributes the stock price volatility into three components: MKT denotes market-wide risk, IND the variance of the respective sector (banking and insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector (bank, investment bank, life insurer or property-liability insurer). Panel A MKT

IND(Bank)

IND(Insurance)

FIRM(Bank)

FIRM(Investment Banks)

FIRM(PC)

FIRM(Life)

0.200%

0.150%

0.100%

0.050%

0.000% 2004

2005

2006

2007

2008

2009

2010

2011

2012

2010

2011

2012

2011

2012

Panel B MKT

IND(Bank)

IND(Insurance)

0.140% 0.120% 0.100% 0.080% 0.060% 0.040% 0.020% 0.000% 2004

2005

2006

2007

2008

2009

Panel C FIRM(Bank)

FIRM(Investment Banks)

FIRM(PC)

FIRM(Life)

0.300% 0.250% 0.200% 0.150% 0.100% 0.050% 0.000% 2004

2005

2006

2007

2008

28

2009

2010

Table 4: Systematic, Systemic (Sector) and Idiosyncratic Risk - Table The table shows variance decomposition as in Campbell et al. (2001) for the years 2004-2012. The decomposition distributes the stock price volatility into three components: MKT denotes market-wide risk, IND the variance of the respective sector (banking and insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector (bank, investment bank, life insurer or property-liability insurer).

2004

2005

2006

2007

2008

2009

2010

2011

2012

MKT

0.005%

0.004%

0.004%

0.010%

0.068%

0.028%

0.013%

0.022%

0.006%

IND(Bank)

0.001%

0.001%

0.002%

0.006%

0.091%

0.120%

0.010%

0.015%

0.007%

IND(Insurance)

0.003%

0.002%

0.001%

0.002%

0.018%

0.011%

0.002%

0.003%

0.001%

FIRM(Life)

0.016%

0.014%

0.015%

0.019%

0.170%

0.126%

0.022%

0.021%

0.016%

FIRM(PC)

0.019%

0.018%

0.018%

0.036%

0.190%

0.125%

0.033%

0.038%

0.032%

FIRM(Investment Banks)

0.044%

0.037%

0.036%

0.049%

0.226%

0.200%

0.074%

0.074%

0.051%

FIRM(Bank)

0.027%

0.025%

0.021%

0.040%

0.281%

0.334%

0.110%

0.110%

0.067%

29

Table 5: Regression Results - Banks The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2004-2012 for banks. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.004*** (0.001) 0.224 (0.148) 0.289*** (0.094) -0.151 (0.108) 0.093

(2) (IND_Bank) -0.002*** (0.001) 0.110 (0.089) 0.149*** (0.055) -0.111* (0.065) 0.082*

(3) (FIRM_Bank) -0.000*** (0.000) 0.023*** (0.009) 0.063*** (0.007) -0.022*** (0.006) 0.002

(4) (IND_LH) -0.004*** (0.001) 0.196 (0.128) 0.272*** (0.078) -0.146 (0.098) 0.083

(5) (IND_PC) -0.002*** (0.000) 0.067 (0.064) 0.116*** (0.040) -0.088* (0.047) 0.049*

(0.072) 0.144*** (0.031) 0.174*** (0.031) -49.090*** (6.385) -5.942*** (0.251) 0.501 0.500 6471

(0.043) 0.083*** (0.018) 0.110*** (0.018) -20.075*** (3.966) -3.580*** (0.148) 0.501 0.501 6471

(0.003) 0.016*** (0.003) -0.016*** (0.003) -0.002 (0.319) -0.059*** (0.014) 0.229 0.228 6471

(0.061) 0.105*** (0.026) 0.169*** (0.026) -42.315*** (5.417) -5.101*** (0.213) 0.498 0.497 6471

(0.030) 0.059*** (0.014) 0.079*** (0.014) -17.166*** (2.674) -2.541*** (0.108) 0.489 0.488 6471

Table 6: Regression Results – Investment Banks The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2004-2012 for investment banks. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking and insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.013** (0.006) 0.133 (0.200) 0.018 (0.288) -0.077 (0.112) -0.468***

(2) (IND_Bank) -0.002 (0.004) 0.021 (0.128) -0.045 (0.156) -0.020 (0.076) -0.208*

(3) (FIRM_Bank) -0.001*** (0.000) 0.018** (0.007) 0.026*** (0.009) 0.003 (0.002) 0.008

(4) (IND_LH) -0.011** (0.005) 0.032 (0.179) 0.010 (0.238) -0.029 (0.113) -0.438***

(5) (IND_PC) -0.005* (0.003) 0.042 (0.097) 0.071 (0.109) 0.003 (0.058) -0.245***

(0.170) 0.083** (0.032) 0.114*** (0.028) -1.454** (0.650) -2.801*** (0.498) 0.386 0.375 462

(0.108) 0.066*** (0.022) 0.054*** (0.020) -0.826** (0.314) -1.765*** (0.303) 0.361 0.350 462

(0.006) -0.006*** (0.001) -0.000 (0.001) 0.060*** (0.020) 0.130*** (0.024) 0.392 0.381 462

(0.152) 0.081*** (0.030) 0.091*** (0.027) -1.442** (0.573) -2.443*** (0.428) 0.378 0.367 462

(0.082) 0.046*** (0.015) 0.052*** (0.014) -0.146 (0.240) -1.502*** (0.201) 0.442 0.432 462

30

Table 7: Regression Results – Life Insurers The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2004-2012 for life insurers. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.004 (0.006) 0.240 (0.213) -0.003 (0.003) -0.025 (0.239) 0.364

(2) (IND_LH) -0.005 (0.006) 0.162 (0.184) -0.002 (0.002) -0.004 (0.195) 0.276

(3) (FIRM_Ins) -0.001*** (0.000) 0.009** (0.004) 0.000 (0.000) -0.003 (0.004) 0.002

(4) (IND_Bank) 0.003 (0.003) 0.161 (0.116) -0.003* (0.001) 0.031 (0.145) 0.105

(5) (IND_PC) -0.003 (0.003) 0.131 (0.092) -0.002 (0.001) -0.025 (0.112) 0.107

(0.327) 0.076 (0.084) 0.010 (0.085) -203.342 (364.589) -0.355 (0.969) 0.091 0.067 323

(0.268) 0.084 (0.071) 0.003 (0.071) -8.615 (285.204) -0.452 (0.800) 0.128 0.106 323

(0.007) 0.000 (0.001) -0.003* (0.002) 10.306 (8.331) 0.059*** (0.021) 0.133 0.111 323

(0.195) 0.046 (0.044) -0.012 (0.046) 3.058 (196.667) -0.011 (0.533) 0.139 0.117 323

(0.136) 0.029 (0.036) 0.009 (0.036) -112.695 (153.308) -0.119 (0.399) 0.123 0.101 323

Table 8: Regression Results – PC Insurers The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2004-2012 for property-liability insurers. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_PC) -0.000 (0.012) 0.109 (0.181) 0.004** (0.001) 0.220 (0.254) 0.357

(2) (IND_PC) 0.002 (0.005) 0.151 (0.090) 0.002** (0.001) 0.124 (0.114) 0.144

(3) (FIRM_PC) -0.001*** (0.000) -0.002 (0.007) 0.000 (0.000) 0.027*** (0.009) -0.009

(4) (IND_Bank) 0.001 (0.007) 0.247*** (0.091) 0.002** (0.001) -0.002 (0.138) 0.262**

(5) (IND_LH) -0.008 (0.010) 0.217 (0.139) 0.003** (0.001) 0.201 (0.210) 0.243

(0.228) 0.121*** (0.031) 0.058 (0.047) -1160.625*** (420.189) -2.361*** (0.502) 0.329 0.316 438

(0.094) 0.059*** (0.014) 0.029 (0.021) -621.281*** (206.106) -1.290*** (0.265) 0.351 0.339 438

(0.006) -0.001 (0.001) -0.004** (0.001) 37.715** (16.251) 0.072*** (0.020) 0.344 0.332 438

(0.113) 0.063*** (0.018) 0.052* (0.029) -787.839*** (274.758) -1.619*** (0.329) 0.292 0.279 438

(0.184) 0.116*** (0.027) 0.064 (0.042) -1085.692*** (383.408) -2.349*** (0.441) 0.369 0.357 438

31

Table 9: Regression Results – Banks – Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for banks. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.006*** (0.002) 0.253 (0.163) 0.312*** (0.108) -0.120 (0.121) 0.052

(2) (IND_Bank) -0.002*** (0.001) 0.089 (0.081) 0.158*** (0.050) -0.097* (0.057) 0.049

(3) (FIRM_Bank) -0.001*** (0.000) 0.026*** (0.009) 0.063*** (0.010) -0.027*** (0.010) 0.003

(4) (IND_LH) -0.005*** (0.001) 0.160 (0.138) 0.249*** (0.086) -0.125 (0.104) 0.078

(5) (IND_PC) -0.002*** (0.001) 0.100 (0.071) 0.139*** (0.047) -0.085* (0.049) 0.039

(0.082) 0.162*** (0.041) 0.158*** (0.043) -26.593*** (7.827) -6.058*** (0.306) 0.521 0.520 3041

(0.039) 0.081*** (0.019) 0.082*** (0.020) -7.167 (4.377) -3.082*** (0.144) 0.553 0.552 3041

(0.005) 0.018*** (0.003) -0.018*** (0.003) 0.147 (0.375) -0.065** (0.026) 0.217 0.215 3041

(0.068) 0.138*** (0.033) 0.141*** (0.033) -23.510*** (6.516) -5.234*** (0.250) 0.536 0.535 3041

(0.033) 0.064*** (0.018) 0.067*** (0.019) -4.370 (3.216) -2.490*** (0.127) 0.517 0.516 3041

Table 10: Regression Results – Investment Banks– Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for investment banks. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.051** (0.024) -0.026 (0.204) -0.102 (0.218) 0.148** (0.069) -0.327

(2) (IND_Bank) -0.025** (0.012) -0.095 (0.104) -0.164* (0.097) 0.155*** (0.030) -0.140

(3) (FIRM_Bank) -0.003*** (0.001) 0.013* (0.007) 0.019** (0.009) 0.006*** (0.002) 0.017**

(4) (IND_LH) -0.045** (0.022) -0.087 (0.178) -0.146 (0.174) 0.211*** (0.058) -0.324*

(5) (IND_PC) -0.017* (0.009) -0.104 (0.090) -0.080 (0.084) 0.117*** (0.029) -0.219**

(0.207) 0.048 (0.032) 0.149*** (0.027) -1.323 (0.853) -2.511*** (0.467) 0.482 0.462 217

(0.103) 0.034** (0.014) 0.070*** (0.012) -1.158*** (0.283) -1.299*** (0.227) 0.546 0.528 217

(0.007) -0.007*** (0.001) 0.000 (0.001) 0.093*** (0.029) 0.163*** (0.027) 0.449 0.427 217

(0.179) 0.048* (0.026) 0.118*** (0.020) -1.310** (0.585) -2.066*** (0.394) 0.506 0.488 217

(0.093) 0.036** (0.014) 0.068*** (0.013) -0.421 (0.326) -1.433*** (0.190) 0.604 0.589 217

32

Table 11: Regression Results – Life Insurers– Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for life insurers. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.028 (0.021) 0.497* (0.289) -0.004 (0.003) -0.465* (0.261) 0.485

(2) (IND_LH) -0.029* (0.017) 0.407* (0.233) -0.003 (0.002) -0.396* (0.211) 0.390

(3) (FIRM_Ins) -0.002*** (0.001) 0.015* (0.007) 0.000 (0.000) -0.016** (0.007) 0.002

(4) (IND_Bank) -0.012 (0.009) 0.205 (0.124) -0.003** (0.001) -0.222* (0.127) 0.157

(5) (IND_PC) -0.011 (0.009) 0.210 (0.125) -0.002 (0.001) -0.197* (0.114) 0.149

(0.388) 0.075 (0.095) 0.092 (0.092) -702.717* (413.393) -1.126 (1.056) 0.248 0.204 146

(0.311) 0.073 (0.079) 0.071 (0.077) -498.266 (321.723) -0.902 (0.860) 0.356 0.318 146

(0.010) 0.000 (0.002) -0.004 (0.002) 12.639 (15.628) 0.082** (0.033) 0.197 0.150 146

(0.160) 0.034 (0.042) 0.038 (0.041) -278.969 (165.982) -0.399 (0.414) 0.395 0.360 146

(0.153) 0.035 (0.041) 0.038 (0.038) -300.057* (160.087) -0.509 (0.418) 0.293 0.252 146

Table 12: Regression Results – PC Insurers– Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for property-liability insurers. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification Liability Diversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_PC) -0.017 (0.018) 0.209 (0.301) 0.002* (0.001) 0.644* (0.326) 0.074

(2) (IND_PC) -0.009 (0.008) 0.125 (0.131) 0.001* (0.001) 0.375** (0.149) 0.000

(3) (FIRM_PC) -0.002*** (0.001) 0.014* (0.008) 0.000 (0.000) 0.035*** (0.010) -0.017**

(4) (IND_Bank) -0.009 (0.009) 0.238 (0.146) 0.001 (0.001) 0.348** (0.160) 0.010

(5) (IND_LH) -0.017 (0.016) 0.363 (0.260) 0.002 (0.001) 0.621** (0.278) -0.039

(0.295) 0.153*** (0.031) 0.063 (0.054) -1357.299** (554.379) -2.675*** (0.680) 0.426 0.405 230

(0.120) 0.070*** (0.015) 0.035 (0.026) -669.519*** (236.728) -1.391*** (0.291) 0.500 0.482 230

(0.008) 0.000 (0.001) -0.003* (0.002) 30.908 (21.086) 0.051*** (0.018) 0.391 0.369 230

(0.128) 0.061*** (0.016) 0.053* (0.029) -744.606** (288.981) -1.547*** (0.340) 0.453 0.433 230

(0.245) 0.130*** (0.028) 0.073 (0.049) -1270.890** (498.334) -2.573*** (0.610) 0.462 0.443 230

33

Table 13: Robustness: Regression Results – Banks – Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for banks using an alternative measure of default risk. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Risk adjusted ROE Lerner Index

Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.006*** (0.002) 0.246

(2) (IND_Bank) -0.002*** (0.001) 0.087

(3) (FIRM_Bank) -0.001*** (0.000) 0.025***

(4) (IND_LH) -0.005*** (0.001) 0.157

(5) (IND_PC) -0.002*** (0.001) 0.098

(0.162) 0.265** (0.110) -0.081 (0.121) 0.043 (0.082) 0.174*** (0.038) 0.146*** (0.040) -26.609*** (7.670) -6.079*** (0.302) 0.524 0.522 3041

(0.081) 0.145*** (0.050) -0.085 (0.057) 0.047 (0.039) 0.087*** (0.018) 0.076*** (0.019) -7.070 (4.316) -3.100*** (0.141) 0.554 0.553 3041

(0.009) 0.057*** (0.010) -0.022** (0.010) 0.002 (0.004) 0.019*** (0.003) -0.018*** (0.003) 0.115 (0.384) -0.065** (0.026) 0.231 0.229 3041

(0.138) 0.223** (0.088) -0.100 (0.104) 0.073 (0.068) 0.150*** (0.031) 0.130*** (0.031) -23.340*** (6.407) -5.269*** (0.245) 0.537 0.536 3041

(0.071) 0.125*** (0.047) -0.073 (0.050) 0.037 (0.033) 0.070*** (0.017) 0.061*** (0.018) -4.295 (3.173) -2.506*** (0.124) 0.518 0.517 3041

Table 14: Robustness: Regression Results – Investment Banks– Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for investment banks using an alternative measure of default risk. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and propertyliability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Risk adjusted ROE Lerner Index Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.010 (0.009) 0.128 (0.199) -0.015 (0.281) -0.060 (0.114) -0.454** (0.171) 0.082** (0.034) 0.111*** (0.030) -1.145 (0.708) -2.750*** (0.487) 0.381 0.370 462

(2) (IND_Bank) 0.003 (0.006) 0.031 (0.128) -0.037 (0.155) -0.017 (0.078) -0.222** (0.110) 0.068*** (0.022) 0.052** (0.020) -0.911** (0.391) -1.773*** (0.304) 0.362 0.351 462

34

(3) (FIRM_Bank) -0.001*** (0.000) 0.017** (0.007) 0.022** (0.009) 0.004 (0.002) 0.010* (0.005) -0.006*** (0.002) -0.000 (0.001) 0.091*** (0.027) 0.135*** (0.024) 0.391 0.381 462

(4) (IND_LH) -0.009 (0.008) 0.027 (0.179) -0.019 (0.234) -0.015 (0.114) -0.424*** (0.154) 0.080** (0.032) 0.089*** (0.029) -1.173* (0.609) -2.400*** (0.420) 0.374 0.363 462

(5) (IND_PC) -0.002 (0.004) 0.045 (0.096) 0.066 (0.107) 0.009 (0.059) -0.248*** (0.084) 0.047*** (0.016) 0.050*** (0.015) -0.101 (0.276) -1.491*** (0.200) 0.436 0.426 462

Table 15: Robustness: Regression Results – Life Insurers– Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for life insurers using an alternative measure of default risk. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and propertyliability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Risk adjusted ROE Lerner Index Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.021 (0.018) 0.452** (0.169) 0.025*** (0.008) 0.022 (0.204) 0.048 (0.313) 0.527*** (0.104) -0.382*** (0.098) -400.408 (359.061) -2.485** (1.001) 0.377 0.340 146

(2) (IND_LH) -0.022* (0.011) 0.374*** (0.126) 0.018*** (0.006) -0.022 (0.166) -0.015 (0.266) 0.457*** (0.080) -0.333*** (0.074) -282.594 (301.336) -1.955** (0.794) 0.500 0.471 146

(3) (FIRM_Ins) -0.003*** (0.001) 0.009* (0.005) 0.000 (0.000) -0.003 (0.006) -0.008 (0.009) 0.008*** (0.002) -0.010*** (0.003) 6.897 (12.478) 0.048 (0.032) 0.351 0.313 146

(4) (IND_Bank) -0.008 (0.005) 0.181** (0.074) 0.009*** (0.003) -0.024 (0.093) 0.093 (0.151) 0.230*** (0.041) -0.169*** (0.038) -123.922 (154.689) -1.077*** (0.388) 0.535 0.507 146

(5) (IND_PC) -0.009 (0.006) 0.174** (0.072) 0.010*** (0.003) -0.004 (0.081) 0.022 (0.129) 0.236*** (0.036) -0.171*** (0.034) -173.868 (142.830) -1.138*** (0.399) 0.443 0.410 146

Table 16: Robustness: Regression Results – PC Insurers– Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for property-liability insurers using an alternative measure of default risk. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Risk adjusted ROE Lerner Index Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_PC) -0.018 (0.025) 0.284 (0.260) 0.002* (0.001) 0.641* (0.325) 0.086 (0.294) 0.153*** (0.032) 0.062 (0.054) -1345.388** (587.701) -2.754*** (0.696) 0.422 0.401 230

(2) (IND_PC) -0.010 (0.012) 0.165 (0.113) 0.001 (0.001) 0.373** (0.148) 0.006 (0.119) 0.069*** (0.015) 0.035 (0.025) -663.286** (254.681) -1.433*** (0.303) 0.494 0.476 230

35

(3) (FIRM_PC) -0.003*** (0.001) 0.022*** (0.006) 0.000 (0.000) 0.035*** (0.010) -0.015* (0.008) 0.000 (0.001) -0.003* (0.002) 31.656 (22.340) 0.043** (0.019) 0.380 0.358 230

(4) (IND_Bank) -0.007 (0.015) 0.279** (0.129) 0.001 (0.001) 0.345** (0.161) 0.016 (0.127) 0.061*** (0.016) 0.052* (0.028) -736.492** (306.967) -1.594*** (0.348) 0.447 0.427 230

(5) (IND_LH) -0.006 (0.024) 0.437* (0.230) 0.002 (0.001) 0.613** (0.282) -0.028 (0.241) 0.132*** (0.029) 0.068 (0.048) -1253.247** (531.119) -2.662*** (0.627) 0.454 0.435 230

Table 17: Regression Results – SIFIs The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2004-2012 for Systemically Important Financial Institutions. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.030*** (0.010) -0.419 (0.360) -0.022 (0.064) 0.076 (0.331) -0.556** (0.214) -0.329** (0.124) 0.250*** (0.069) 1.060* (0.530) 4.587* (2.654) 0.126 0.084 177

(2) (IND_LH) -0.021** (0.009) -0.410 (0.317) 0.033 (0.052) -0.124 (0.256) -0.392** (0.162) -0.202** (0.091) 0.175*** (0.050) 0.950** (0.446) 2.854 (1.935) 0.140 0.099 177

(3) (FIRM_Ins) -0.001*** (0.000) -0.014 (0.013) -0.001 (0.002) -0.005 (0.010) -0.001 (0.006) -0.005 (0.003) 0.001 (0.002) 0.020 (0.017) 0.137* (0.074) 0.136 0.095 177

(4) (IND_Bank) 0.001 (0.005) -0.079 (0.121) -0.066** (0.029) -0.154 (0.132) 0.001 (0.097) -0.146** (0.055) 0.055* (0.027) 0.196 (0.252) 3.494*** (0.996) 0.118 0.077 177

(5) (IND_PC) -0.012** (0.005) -0.242 (0.151) -0.043 (0.033) -0.109 (0.166) -0.100 (0.087) -0.138** (0.061) 0.084** (0.031) 1.217*** (0.329) 2.526* (1.306) 0.263 0.228 177

Table 18: Regression Results – SIFIs – Crisis Period The table shows the results of pooled OLS regressions using clustered standard errors at firm-level for the years 2007-2010 for Systemically Important Financial Institutions. The variables are defined in table 2. MKT denotes market-wide risk, IND the risk of the respective sector (banking, life insurance and property-liability insurance sector) and FIRM denotes the idiosyncratic risk of a firm from the respective sector. ***, ** and * denotes significance at the 1%, 5% and 10% level.

Z-Score Lerner Index Cost-Income Ratio Income Diversification LiabilityDiversification Asset Size Capitalization Market Share Constant R2 Adj. R2 Observations

(1) (MKT_Bank) -0.071*** (0.019) 0.761 (0.604) -0.137* (0.070) 0.323 (0.467) -0.563* (0.277) -0.490*** (0.125) 0.278*** (0.082) -0.227 (0.785) 7.623*** (2.617) 0.200 0.107 78

(2) (IND_LH) -0.053*** (0.015) 0.559* (0.322) -0.134** (0.055) 0.063 (0.309) -0.045 (0.145) -0.296*** (0.097) 0.104** (0.042) 0.049 (0.494) 6.432*** (1.971) 0.303 0.222 78

(3) (FIRM_Ins) -0.002*** (0.001) 0.005 (0.014) -0.006** (0.003) -0.001 (0.016) 0.006 (0.011) -0.011** (0.005) -0.001 (0.003) 0.030 (0.024) 0.314*** (0.103) 0.290 0.208 78

36

(4) (IND_Bank) -0.031*** (0.009) 0.419** (0.178) -0.068* (0.038) -0.268 (0.207) -0.151 (0.112) -0.173** (0.074) 0.092** (0.036) -0.875** (0.312) 3.231** (1.359) 0.407 0.338 78

(5) (IND_PC) -0.026*** (0.008) 0.410 (0.239) -0.079 (0.046) -0.212 (0.165) -0.157 (0.125) -0.170** (0.071) 0.099** (0.037) -0.062 (0.224) 2.826* (1.512) 0.302 0.221 78

Table 19: Banks’ and Investment Banks’ Variables to create Lerner Indices – Summary Statistics The table shows the summary statistics for banks (Panel A) and investment banks (Panel B) to create Lerner indices. Numbers larger than 1,000,000 are rounded to the nearest million. The variables are defined in table 20.

Variable - Banks Output Price of Output Q Total Cost Price oflabor Pricefphysicalcapital Price offunds Variable - Investment Banks Output Price of Output Q Total Cost Price oflabor Price ofphysicalcapital Price offunds

Panel A Obs Mean Std. Dev. Min Max 6,729 1,700,000,000 3,000,000,000 75,000,000 14,000,000,000 6,729 0.043 0.009 0.022 0.061 6,729 32,000,000 55,000,000 1,500,000 250,000,000 6,729 0.016 0.004 0.007 0.026 6,729 0.006 0.002 0.002 0.013 6,729 1.557 1.161 0.000 4.341 Panel B Obs Mean Std. Dev. Min Max 343 73,000,000,000 220,000,000,000 4,600,000 1,100,000,000,000 343 0.932 1.605 0.020 15.364 339 5,800,000,000 10,000,000,000 8,000,000 47,000,000,000 343 0.425 0.750 0.005 3.750 343 910,000,000 4,400,000,000 1,100,000 47,000,000,000 339 0.377 0.711 0.011 4.444

Table 20: Banks’ and Investment Banks’ Variables to create Lerner Indices - Definitions The table shows the definitions of the variables for banks (Panel A) and investment banks (Panel B) to create Lerner indices.

Variable - Banks Output Price of Output Total Cost Price of labor Price of physical capital Price of funds

Panel A Description Total assets. Total revenue (interest plus non-interest income) divided by total assets. The sum of personnel expenses, other administrative expenses, other operating expenses and price of funds Personnel expenses over total assets. Other administrative expenses plus other operating expenses over total fixed assets. Interest expenses over total funds. Panel B

Variable - Investment Banks Output Price of Output Total Cost Price of labor Price of physical capital Price of funds

Description Total assets. Total revenue (interest plus non-interest income) divided by total assets. The sum of personnel expenses, other administrative expenses, other operating expenses and price of funds. Personnel expenses over total assets. Other administrative expenses plus other operating expenses over total fixed assets. Interest expenses over total funds.

37

Table 21: Life and PC Insurers Variables to create Lerner Indices – Summary Statistics The table shows the summary statistics for life insurers (Panel A) and property-liability insurers (Panel B) to create Lerner indices. Numbers larger than 1,000,000 are rounded to the nearest million. The variables are defined in table 22.

Variable - Life Insurers Output Price Life Individual Life P1 Output Price Life Group Life P2 Output Price Life Individual Annuities P3 Output Price Life Group Annuities P4 Output Price Life Accident Health P5 Output Price Life Intermediary P6 Output Quantity Life Individual Life Y1 Output Quantity Life Group Life Y2 Output Quantity Life Individual Annuities Y3 Output Quantity Life Group Annuities Y4 Output Quantity Life Accident Health Y5 Output Quantity Life Intermediary Y6 Input Price Life Labor W1 Input Price Life Agent W2 Input Price Life Material W3 Input Price Life Financial Capital W4 Input Quantity Life Labor X1 Input Quantity Life Agent X2 Input Quantity Life Material X3 Input Quantity Life Financial Capital X4 Variable - PC Insurers Output Price PL Personal Short Tail P1 Output Price PL Personal Long Tail P2 Output Price PL Commercial Long Tail P3 Output Price PL Commercial Short Tail P4 Output Price PL Intermediary P5 Output Quantity PL Personal Short Tail Y1 Output Quantity PL Personal Long Tail Y2 Output Quantity PL Commercial Long Tail Y3 Output Quantity PL Commercial Short Tail Y4 Output Quantity PL Intermediary Y5 Input Price PL Labor W1 Input Price PL Agent W2 Input Price PL Material W3 Input Price PL Financial Capital W4 Input Quantity PL Labor X1 Input Quantity PL Agent X2 Input Quantity PL Material X3 Input Quantity PL Financial Capital X4

Panel A Obs Mean Std. Dev. Min Max 529 0.786 1.284 0.017 8.702 529 1.030 2.710 0.040 24.130 529 0.311 0.781 0.003 6.499 529 0.109 0.267 0.004 2.530 529 1.475 2.689 0.117 16.022 529 0.058 0.010 0.015 0.114 529 745,013 1,100,000 214 4,200,000 529 163,824 305,309 6 1,400,000 529 1,600,000 2,800,000 32 12,000,000 529 1,300,000 2,900,000 10 13,000,000 529 933,695 1,800,000 186 8,800,000 529 28,000,000 49,000,000 40,503 310,000,000 529 1521.990 382.246 447.000 2686.000 529 1149.420 68.522 1026.000 1253.000 529 109.490 8.252 83.277 220.076 529 0.122 0.023 0.082 0.170 529 217 273 4 1,254 529 510 579 6 2,054 529 2,666 3,570 37 17,584 529 2,200,000 2,500,000 17,922 11,000,000 Panel B Obs Mean Std. Dev. Min Max 810 1.556 2.500 0.259 19.465 810 0.573 0.512 0.104 3.519 810 0.322 0.505 0.023 4.540 810 1.399 1.796 0.097 11.515 794 0.047 0.014 0.006 0.103 810 108,759 208,333 48 1,000,000 810 245,334 573,446 81 3,600,000 810 460,307 822,990 83 4,600,000 810 181,301 303,241 79 1,500,000 794 7,200,000 16,000,000 11,660 140,000,000 810 1287.800 208.045 847.225 1930.210 810 1001.680 10.394 982.682 1024.180 797 95.179 4.972 67.414 116.401 810 0.124 0.024 0.082 0.170 810 202 329 2 1,468 810 250 395 1 1,733 810 2,463 4,214 28 17,313 810 2,200,000 3,300,000 26,912 14,000,000

38

Table 22: Life and PC Insurers Variables to create Lerner Indices – Definitions The table shows the definitions of the variables for life insurers (Panel A) and property-liability insurers (Panel B) to create Lerner indices.

Variable - Life Insurers Output Price Life Individual Life P1 Output Price Life Group Life P2 Output Price Life Individual Annuities P3 Output Price Life Group Annuities P4 Output Price Life Accident Health P5 Output Price Life Intermediary P6

Output Quantity Life Individual Life Y1 Output Quantity Life Group Life Y2 Output Quantity Life Individual Annuities Y3 Output Quantity Life Group Annuities Y4 Output Quantity Life Accident Health Y5 Output Quantity Life Intermediary Y6 Input Price Life Labor W1

Input Price Life Agent W2 Input Price Life Material W3

Input Price Life Financial Capital W4

Input Quantity Life Labor X1 Input Quantity Life Agent X2 Input Quantity Life Material X3 Input Quantity Life Financial Capital X4

Panel A Description The sum of premiums and investment income minus output for each line divided by output. The sum of premiums and investment income minus output for each line divided by output. The sum of premiums and investment income minus output for each line divided by output. The sum of premiums and investment income minus output for each line divided by output. The sum of premiums and investment income minus output for each line divided by output. The expected portfolio rate of return is the weighted average of the debt and equity returns, weighted by the proportion of the portfolio invested in debt securities and stocks. The sum of incurred benefits and additions to reserves for the major lines of business offered by life insurers. The sum of incurred benefits and additions to reserves for the major lines of business offered by life insurers. The sum of incurred benefits and additions to reserves for the major lines of business offered by life insurers. The sum of incurred benefits and additions to reserves for the major lines of business offered by life insurers. The sum of incurred benefits and additions to reserves for the major lines of business offered by life insurers. Average real invested assets are used to measure the quantity of the intermediation output for life insurers. U.S. Department of Labor (DOL) data on average weekly wages for Standard Industrial Classification (SIC) class 6311 before 2001 and North American Industry Classification System (NAICS) class 524113 since 2001. DOL average weekly wage rate for insurance agencies and brokerages (SIC class 6411 and NAICS class 524210). The weighted average of price indices for business services from the component indices representing the various categories of expenditures from the expense page of Best’s Aggregates and Averages. The component price indices are from the DOL and the U.S. Department of Commerce, Bureau of Economic Analysis. The cost of capital for year t is calculated as the 30-day Treasury bill rate at the end of year t-1, plus the long-term (1926 to the end of year t-1) average market risk premium on large company stocks, plus the long-term (the 1926 through end of year t-1) average size premium from Ibbotson Associates.

Total expenditures on labor from the regulatory annual statement. Total expenditures on agent labor from the regulatory annual statement. Total expenditures on materials from the regulatory annual statement. The average of the beginning and end-of-year equity capital, plus the asset valuation reserve.

39

Table 22: Life and PC Insurers Variables to create Lerner Indices – Definitions (continued) The table shows the definitions of the variables for life insurers (Panel A) and property-liability insurers (Panel B) to create Lerner indices.

Variable - PC Insurers Output Price PL Personal Short Tail P1

Output Price PL Personal Long Tail P2

Output Price PL Commercial Long Tail P3

Output Price PL Commercial Short Tail P4

Output Price PL Intermediary P5

Output Quantity PL Personal Short Tail Y1 Output Quantity PL Personal Long Tail Y2 Output Quantity PL Commercial Long Tail Y3 Output Quantity PL Commercial Short Tail Y4 Output Quantity PL Intermediary Y5 Input Price PL Labor W1 Input Price PL Agent W2 Input Price PL Material W3

Input Price PL Financial Capital W4

Input Quantity PL Labor X1 Input Quantity PL Agent X2 Input Quantity PL Material X3 Input Quantity PL Financial Capital X4

Panel B Description Output price is pi = [Pi - PV(Li)]/PV(Li), where pi is the price of output i, Pi = premiums in line i, Li = incurred losses in line i, and PV is the present value operator. Output price is pi = [Pi - PV(Li)]/PV(Li), where pi is the price of output i, Pi = premiums in line i, Li = incurred losses in line i, and PV is the present value operator. Output price is pi = [Pi - PV(Li)]/PV(Li), where pi is the price of output i, Pi = premiums in line i, Li = incurred losses in line i, and PV is the present value operator. Output price is pi = [Pi - PV(Li)]/PV(Li), where pi is the price of output i, Pi = premiums in line i, Li = incurred losses in line i, and PV is the present value operator. The expected portfolio rate of return is the weighted average of the debt and equity returns, weighted by the proportion of the portfolio invested in debt securities and stocks. Present value of real losses incurred by line of P-L insurance. Present value of real losses incurred by line of P-L insurance. Present value of real losses incurred by line of P-L insurance. Present value of real losses incurred by line of P-L insurance. Average real invested assets are used to measure the quantity of the intermediation output for P-L insurers. DOL data on SIC class 6331 before 2001 and NAICS class 524126 since 2001. DOL average weekly wage rate for insurance agencies and brokerages (SIC class 6411 and NAICS class 524210). The weighted average of price indices for business services from the component indices representing the various categories of expenditures from the expense page of Best’s Aggregates and Averages. The component price indices are from the DOL and the U.S. Department of Commerce, Bureau of Economic Analysis. The cost of capital for year t is calculated as the 30-day Treasury bill rate at the end of year t-1, plus the long-term (1926 to the end of year t-1) average market risk premium on large company stocks, plus the long-term (the 1926 through end of year t-1) average size premium from Ibbotson Associates

Total expenditures on labor from the regulatory annual statement. Total expenditures on agent labor from the regulatory annual statement. Total expenditures on materials from the regulatory annual statement. The average of the beginning and end-of-year equity capital, plus unearned premium reserves.

40