The role of legal system and global financial crisis on managerial risk taking

(1)

The role of legal system and global financial crisis on managerial

risk taking

Abstract: This paper investigates the impact of the company’s legal system origin on the managerial risk-taking behaviour. The empirical results of this study provide evidence that support the claim that legal system origin influences managerial risk-taking behaviour. Furthermore, evidence shows that companies originating in common law countries exhibit higher managerial risk-taking compared to companies from civil law countries. Next, the results show that the financial crisis had a recognizable effect on managerial risk-taking. Finally, this paper provides evidence that legal system origin is a significant determinant of managerial risk-taking during the global financial crisis and remained significant after the end of the global financial crisis.

Key words: managerial risk-taking, legal systems, global financial crisis, regulations Name: Matej Svoboda

Student number: s3499855

E-mail: m.svoboda.1@gmail.student.rug.nl Study programme: MSc Finance

(2)

1. Introduction

The general concept of risk has a long history. Understating the risk and risk assessment have developed over the centuries. Jammalamadaka, Sreenivasa, and Bernstein (1999) presents a mastery of a risk; an approach that future is not determined by the will of gods. Understanding, measuring and managing risk is what distinguishes the modern society from the thousands of years of human history (Jammalamadaka, Sreenivasa, and Bernstein, 1999). In the modern business environment, decisions are taken on a basis of financial fundamentals, therefore decision makers must assess whether the risk involved in the project is worth taking (Merna, Tony, Faisal, and Al-Thani, 2008).

The simple common definition of a risk is the chance of something bad happening in a certain period. Merna, Tony, Faisal, and Al-Thani, (2008) further distinguish three main components of the risk; the likelihood of an undesirable event, consequences of such an event and the period when the event may occur.

Corporate risk-taking behaviour has important implications for the company’s performance, growth and survival (Bromiley, 1991). Moreover, corporate risk-taking behaviour is a crucial factor of the long-term economic growth, therefore it is of a critical importance to understand the determinants of corporate risk-taking in order to identify the approaches to enhance the long-term economic welfare through policy changes and regulations (Faccio, Mura, and Marchica, 2011).

Previous studies have examined the effect of shareholder’s rights, regulation changes, globalization, accounting disclosure rules and creditor’s rights on the corporate risk-taking behaviour (John, Litov, and Yeung, 2005; Bruno and Shin, 2014; Yung, Kenneth, and Chen, 2017). However, the literature on the impact of the legal system on the managerial risk-taking behaviour is still limited.

(3)

There are essential differences between the legal systems, that can affect managerial risk-taking decisions. In this study, I focus mainly on the differences between the common and civil law countries. Magill, Quinzii, and Rochet (2015) argue that common law is a discretion-oriented legal system that is mainly focused on shareholders’ rights and places fewer restrictions on managerial behaviour while at the same time discouraging management from unacceptable and unethical behaviour by using sanctions and litigations. The authors further argue that shareholders’ wealth maximization leads the company to act in the interest of stakeholders such as consumers and employees. In the contrast, civil law systems are more regulation-based, where the government plays an important role in coordinating and regulating the markets (Liang and Renneboog, 2017).

The fundamental differences between the legal systems regarding regulations and market participants’ rights may lead corporations to take different levels of risk. Therefore, the aim of this paper is to identify a link between a company’s legal system origin and managerial risk-taking behaviour. Hence, the main research question in this study is: Does the company’s legal origin impact managerial risk-taking behaviour?

Additionally, the recent, global financial crisis had serious consequences on the macroeconomic and company level. Khademian and Anne (2011) summarize that millions of workers lost their jobs and were unable to find a new job, while at the same time governments injected millions of dollars into the economies. Furthermore, they show that for example by the year 2011, the U.S. government invested 600 million US dollars into the economy to recover from the crisis. Global financial crisis drastically affected the growth of the gross domestic product (GDP). For example, GDP growth of EU countries decreased on average by 50% by the beginning of 2009 (Gennard, 2009). Furthermore, the author reveals a decline in industrial production at the beginning of the financial crisis by 1.6% for the EU countries, while the decline was most recognizable in the Czech Republic, where the industrial production decreased by more than 17%. Gennard (2009) further points out that the most affected European sectors were the banking and motor car industries.

(4)

Furthermore, Apergis and Corray (2018) show, that interest rate spreads are more sensitive to its determinants after the global financial crisis than it was before and during the crisis. Further analysis by Apergis and Corray (2018) reveals that a safer regulatory environment leads to lower interest rate spreads. Therefore, companies operating in countries, where strict regulations and creditors’ protection are enforced, face lower credit spreads. Cociuba, Shukayev, and Ueberfeldt (2018) further highlight that risk-taking behaviour is affected by interest rates and explain that such behaviour can be managed through interest rate policy and further regulations.

The consequences of the global financial crisis and its aftermath affect managerial risk-taking behaviour and may affect the relationship between the legal system and managerial risk-taking. Therefore, the second research question is: Does the global financial crisis influence the managerial risk-taking behaviour under various legal systems?

In order to answer these two research questions, I examine 204,314 events for companies from 74 countries and various legal systems for the period between January 2000 and December 2017. I employ company-specific and macroeconomic variables as an explanatory device of the risk-taking behaviour under different legal systems. Furthermore, I adjust the risk-taking variable to the industry in order to avoid bias, due to the different risk-taking pattern among industries. I perform OLS unbalanced panel regressions to analyse the data. Additionally, I utilize multivariate analysis to show different managerial risk-taking under common and civil law and I use the Wald test to explain the different impact of common and civil law on the risk-taking.

This study reveals two sets of empirical findings regarding two research questions, which illustrate the link between the managerial risk-taking behaviour and corporate legal origin and the impact of the global financial crisis on the corporate risk-taking. Alongside the main specification of managerial-risk taking proxy, I use the alternative specification of risk-taking proxy as a robustness test.

(5)

Secondly, the results reveal that the effect of the global financial crisis on managerial risk-taking behaviour was significant. Additionally, I examine periods before, during and after the global financial crisis separately. The empirical findings show that the impact of legal origin was not significant before the financial crisis and becomes significant during the financial crisis and remained significant after the financial crisis. Lastly, the Wald test supports the finding and reveals that the impact of common and civil law systems become significantly different during the global financial crisis.

The rest of the research paper is structured in the following way: Section 2 reviews previous literature, discusses the theoretical links between legal system origin and managerial risk taking and develops hypotheses; Section 3 describes data and methodology; Section 4 presents and discusses empirical findings and lastly Section 5 concludes the research paper.

2. Background and hypothesis development

2.1 Background

2.1.1 Impact of legal system on financial performance and cost of equity and debt (CSR perspective)

The main differences between the legal systems in relation to the managerial risk-taking is the difference in state interventions into the shareholders’ and stakeholders’ economic life. Common law can be defined as a legal system, where shareholders’ protection plays an important role in contrast to the civil law system, where stakeholders’ protection is favoured (Liang and Renneboog, 2017). Moreover, the authors investigate the level of CSR of companies operating under various legal systems. They find evidence that companies exhibiting a high level of CSR are most likely based in the civil law countries, while companies operating under common law countries tend to achieve lower CSR levels.

(6)

However, high level of CSR is not always in the best interest of shareholders. Goss and Roberts (2011) call this an overinvestment view of CSR and argue that managers sometimes invest more resources into CSR than necessary and hence, harm the company. Furthermore, when the loan is extended to the low-quality borrowers, then there are high agency risks involved and lenders punish borrowers for investing into the CSR. However, high quality borrowers do not face such penalties (Goss and Roberts, 2011). Based on the above arguments, CSR enhances corporate value, when investments into the CSR are reasonable.

El Ghoul, Guedhami, Kim, and Park (2018) investigate the impact of corporate environmental responsibility, as a part of CSR, on financial performance and cost of equity financing. Similarly, to Liang and Renneboog (2017), they define corporate social responsibility as a risk hedging device. They estimate the cost of capital financing by using residual valuation models and derive a conclusion, that environmentally responsible companies face lower cost of equity. Based on their research, this relation holds across different legal systems, geographic and economic settings. The authors further argue that the benefits associated with the CSR investments exceed the costs.

Furthermore, Heinkel, Kraus, and Zechner (2001) show that if 25% of investors would invest their funds into the green investments only, then the toxic companies would be forced to become green due to the high cost of equity financing. They add, however, that at that time, there were only 10% of green investors. Kempf and Osthoff (2007) test a performance of the responsible portfolio. They construct a responsible portfolio by long positions in the stocks of the companies, that perform best in KLD ratings and short positions in the stocks of the companies, that perform worst in KLD ratings. The empirical findings show, that responsible portfolio achieve significant, positive abnormal returns.

Conclusion of Liang and Renneboog (2017) and empirical findings of El Ghoul, Guedhami, Kim, and Park (2018) and Goss and Roberts (2011) create a theoretical link between the legal system of origin and the cost of capital and debt. Companies operating under a civil law country are likely to have higher CSR and be greener than companies operating under a common law country, thus their cost of equity and debt financing is likely to be lower.

(7)

On the other hand, Acharya, Amihud, and Litov (2009) show that relaxing the creditors’ rights leads companies to take more risk and acquisitions are in average more profitable. Furthermore, González (2014) claims that the corporate cost of capital depends on the efficiency of the legal system. His research reveals that companies operating under the legal system, that enforces strict regulations for creditors’ protection face a lower cost of debt. He argues that strong protection of creditor’s rights leads to a creditors’ willingness to lend on the favourable terms. These conclusions are in line with the findings of Apergis and Corray (2018), that safe regulatory environment leads to lower interest rate spreads. González (2014) further explains that more favourable terms of lending are mainly due to lower agency costs. Such costs arise from a situation, in which a borrower has an information, that is not available to a lender which is known as adverse selection and change of borrower’s behaviour during the term of loan known as moral hazard. Therefore, reducing agency problems can lead to lower cost of capital. Jensen and Meckling (1986) state that the agency problem exists in all organizations and corporations. Such problems are associated with agency costs arising mainly from asymmetric information that lead to a moral hazard problem (Windram, 2005). He further argues that, in general, the agency problems are linked to the managerial risk-taking incentives, therefore understanding and management of agency problems is the key to explaining incentive structures. Strong shareholders’ rights can influence agency problem to some degree; however, Windram (2005) concludes that incentive structures can lead to either increase or decrease of managerial risk taking.

2.1.3 Global Financial crisis

(8)

2.2 Hypotheses development

The literature shows that the company’s legal system origin affects the cost of equity and debt financing. The differences in cost of capital between the companies from the various legal system origins are mainly due to differences in legislation and regulations. Based on the presented literature, companies operating under the common law system face higher cost of capital compared to those companies operating under civil law system. Furthermore, the evidence shows that creditors’ protection is associated with managerial risk taking and cost of debt. Moreover, corporate agency problems and shareholders’ protection are associated with managerial-risk taking incentives. Therefore, I hypothesize that there exists a relationship between the company’s legal system origin and the corporate risk-taking behaviour, so the legal origin is a significant determinant of the managerial risk-taking.

H1a: Company’s legal system origin affects managerial risk-taking.

Moreover, the evidence shows that civil law countries enforce stricter regulations and creditors’ protection than the common law countries. The literature further shows that companies operating under civil law countries have higher CSR, therefore exhibit lower cost of capital and debt. This leads to divergence in debt and equity financing and can lead to the different managerial risk-taking. Strong stakeholders’ protection in civil law countries and the difference in regulations leads to the following hypothesis:

H1b: Companies operating under civil law countries exhibit lower risk-taking than companies

operating under common law countries.

Lastly, Gennard (2009) shows that the consequences of the global financial crisis were overwhelming. Many countries experienced a high level of unemployment and low levels of GDP growth. During the financial crisis, government policies and regulations play a crucial role in smoothing the impact of financial crisis and preventing future financial crisis. Therefore, I hypothesize that the financial crisis is associated with different managerial risk-taking behaviour compared to the pre-crisis and post-crisis time period and that companies operating under civil law countries exhibit lower managerial risk-taking than companies operating under common law countries in each of the three time periods.

H2a: Period of global financial crisis affected managerial risk-taking.

H2b: The effect of legal system on managerial risk-taking is larger for companies operating

(9)

3 Data and methodology

3.1 Data

3.1.1 Data collection

I use DataStream’s LAST4ESG companies in my sample. The data for the company’s age and size are collected from the DataStream database and all the remaining company related data are obtained from the WRDS database. The macroeconomic data for GDP per capita and inflation are collected from The World Bank database. In order to cover years before the global financial crisis, the period of the financial crisis and the subsequent years after the end of the financial crisis, the sample covers the period January 2000 to December 2017. Furthermore, I winsorize the variables to limit the effect of outliers.

My source of concerns is the possibility of endogeneity issues, which are common in finance studies as pointed out by Reeb, Kwok, and Baek (1998). I reduce this problem by lagging the independent variables by one period, as proposed by Bruno and Shin (2014).

3.1.2 Industry, Country and Legal systems event distribution

My sample covers a total of 204,314 events distributed among 9 industries, 3 legal systems and 74 countries. Table 1 below shows the distribution of the events among industries. Most events belong to the manufacturing industry, which includes 35.86% and finance, insurance and real estate industry covering 23.24% of the events, while the least events belong to the agriculture, forestry and fishing industry, which includes only 0.38% and public administration industry which covers 1.52% of the events.

Table 1: Industry distribution

This table presents a distribution of the events among nine industries.

Industry Observations Percent

Agriculture, forestry and fishing 771 0.38%

Mining 16,410 8.03%

Manufacturing 73,266 35.86%

Transportation, communications, electric, gas and sanitary service 28,843 14.12%

Wholesale trade 4,672 2.29%

Retail trade 11,075 5.42%

Finance, insurance and real estate 47,485 23.24%

Services 18,696 9.15%

Public administration 3,096 1.52%

(10)

The sample includes companies operating under countries with common, civil and hybrid legal systems. Majority of the events belong to the common law systems, which accounts for 56.28% of the total sample size and the remaining 43.72% of the events are fairly distributed between civil and hybrid law systems.

My data comprise of events occurring before, during and after the global financial crisis. The events are reasonably distributed between these three time periods, where 38% of the events belong to the period after the global financial crisis and the rest of the sample is distributed among the periods before and during the global financial crisis with 33% and 29% of the events respectively.

3.1.3 Risk-taking variables

My main proxy for managerial risk-taking is the standard deviation of industry adjusted return on assets σ(𝑅𝑂𝐴𝐴𝑑𝑗). I adjust the ratio to the industry in order to reduce the effect of industry

in managerial risk-taking, therefore my risk-taking variable captures only company-specific managerial risk-taking. The main dependent variable is calculated over 16 overlapping quarterly periods. Furthermore, I employ two alternative specifications of the main dependent variable, which are calculated over 24 and 32 overlapping quarterly periods. The choice of main risk-taking variable is based on John, Litov, and Yeung (2006), Faccio, Mura, and Marchica (2011) and Bruno and Shin, (2014)

(11)

3.1.4 Independent variables

I use common, civil and hybrid legal systems dummy variables to distinguish between various legal systems in my sample. Bhagat and Bolton (2012) explain that large corporations are less risk-averse than small corporations and leverage is positively correlated with the risk-taking initiatives. Therefore, I use company specific independent variables as described by Faccio, Mura, and Marchica (2011) to control for such effects. Furthermore, Gupta and Krishnamurti (2018) show that risk-taking behaviour is affected by macroeconomic conditions.

The variables used in this study are as follows: (1) ln(Size) is measured as a natural logarithm of market capitalization, which is calculated as a market value of all outstanding shares of a company, (2) Leverage, that is represented by the ratio of total debt to total assets, (3) SalesGrowth, which is computed as a growth rate of sales between the periods t-1 and t, and (4) ln(1+age), that is defined as natural logarithm of 1 + number of years since incorporation of the company. Faccio, Mura, and Marchica (2011) suggest to use ln(1+age) variable to control for differences in risk-taking throughout company’s life cycle. They argue that young companies take more risk, while older companies tend to be more risk averse. Lastly, I control for macroeconomic conditions. I employ (5) GDP which is defined as a GDP per capita reported in US dollars to control for country economic development and (6) inflation to control for price stability.

3.1.5 Descriptive Statistics

(12)

Table 2: Descriptive statistics of independent variables

This table summarizes my independent macroeconomic and company-specific variables used in the analysis. This table shows only the values of independent variables used for analysis of main and robustness risk-taking variables calculated over 16 overlapping periods. The table includes mean, standard deviation, minimum, 0.25 quartile, median, 0.75 quartile and maximum. Company-specific variables capture size, total debt to total assets ratio, growth of sales and the age of the company and the macroeconomic variables control for economic development and price stability of the country, where the company operates.

Quantiles

Variable Mean S.D. Min 0.25 Median 0.75 Max Main risk-taking variable

ln(size) 14.71 1.62 9.26 13.81 14.83 15.76 18.22 leverage 0.23 0.18 0.00 0.08 0.22 0.35 0.77 salesgrowth 0.05 0.30 -0.76 -0.05 0.02 0.11 2.04 ln(1+age) 3.28 0.88 0.69 2.77 3.26 3.95 4.85 GDP 34,744 20,376 1,000 12,027 39,720 48,168 88,416 Inflation 0.02 0.02 -0.01 0.00 0.01 0.02 0.12

Robustness risk-taking variable

Table 3: Descriptive statistics of risk-taking variables.

This table summarizes my dependent variables used in the analysis. The table includes mean, standard deviation, minimum, 0.25 quartile, median, 0.75 quartile and maximum. ROA stands for main dependent variable and CAPEX stands for robustness dependent variable. The number in the name of the variable represents number of quarterly periods required to calculate each specification of the main / robustness variables.

Quantiles

Variable Observations Mean S.D. Min 0.25 Median 0.75 Max

ROA16 140,967 0.24 0.66 0.01 0.02 0.04 0.11 3.48 ROA24 109,150 0.27 0.69 0.01 0.03 0.05 0.21 4.19 ROA32 79,919 0.3 0.69 0.01 0.03 0.07 0.19 4.01 CAPEX16 100,528 4.66 9.84 0.03 0.15 0.44 2.74 37.66 CAPEX24 80,147 7.31 18.18 0.04 0.31 0.59 2.56 78.68 CAPEX32 60,648 9.94 26.17 0.04 0.3 0.83 6.23 109.69

(13)

3.1.6 Pearson’s correlation

I analyse the correlation between the variables to address multicollinearity problem. Table 4 presents Pearson’s correlation matrix. Measures of correlation between risk-taking variables, firm’s specific variables and macroeconomic variables do not suggest any multicollinearity problem. As anticipated, correlation between 3 measures of risk-taking is high, this however, does not suggest any problem, because only one variable is used in each regression. Among the independent variables, ln(Size) has the highest correlation with the dependent and independent variables, however, the correlation is not sufficiently high to suggest multicollinearity issues.

Table 4:Pearson's correlation matrix

This table presents correlation matrix of company specific variables capturing size, total debt to total assets ratio, growth of sales and the age of the company and the macroeconomic variables, which control for economic development and price stability of the country, where the company operates and main dependent variable denoted ROA16 and alternative specifications of main dependent variable denoted ROA24 and ROA32.

Variables ln(Size) Leverage Salesgrowth ln(1+age) GDP Inflation ROA16 ROA 24 ROA 32 ln(Size) 1 Leverage 0.0813 1 Salesgrowth -0.0875 -0.0305 1 Ln(1+age) 0.1748 0.0603 -0.0625 1 GDP 0.1711 -0.0307 -0.0242 0.0713 1 Inflation -0.0072 0.0211 -0.0052 0.0016 -0.2978 1 ROA16 -0.2169 -0.131 0.0959 -0.0404 0.0075 0.0259 1 ROA24 -0.2362 -0.1362 0.0944 -0.0481 0.0057 0.018 0.9542 1 ROA32 -0.2344 -0.1594 0.0984 -0.0531 0.0284 0.0486 0.8457 0.9071 1 3.2 Methodology

My methodology is inspired by John, Litov, and Yeung (2006), Faccio, Mura, and Marchica (2011) and Bruno and Shin, (2014). Firstly, I determine the legal system, under which the company operates. Secondly, I calculate ROA as a ratio of EBIT to total assets and calculate industry adjusted ROA by subtracting average industry ROA for the same period from the company ROA. Subsequently, I proceed with the last step of my main dependent variable calculation, which is defined as a standard deviation of the adjusted return on asset σ(𝑅𝑂𝐴𝐴𝑑𝑗),

(14)

Subsequently, I compute the three specifications of robustness risk-taking variable. Firstly, I calculate CAPEX to sales ratio and adjust it to the industry by subtracting industry average of CAPEX to sales ratio from the company’s ratio. Then, I follow the same procedure as I use for the calculation of the main dependent variable to derive robustness risk-taking variable calculated over 16, 24 and 32 overlapping periods. I use the robustness risk-taking variable to add robustness to my research, and thus confirm results derived from the main risk-taking variable.

Lastly, I proceed with a calculation of 𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒_𝑖,𝑡, which is defined as a ratio of total debt to total assets and 𝑆𝑎𝑙𝑒𝑠𝐺𝑟𝑜𝑤𝑡ℎ_𝑖,𝑡, that is defined as a growth rate of sales from period t-1 to period t.

In order to test the first hypothesis, I begin with regressing legal systems dummy variables, company-specific variables and macroeconomic variables on the risk-taking variable for all three specifications of main and robustness risk-taking variable.

𝑅𝑖𝑠𝑘 − 𝑡𝑎𝑘𝑖𝑛𝑔 = 𝛼 + 𝛽₁𝐷_𝐶𝑜𝑚𝑚𝑜𝑛_𝑖,𝑡+ 𝛽₂𝐷_𝐶𝑖𝑣𝑖𝑙_𝑖,𝑡 + 𝛽₃ln (𝑆𝑖𝑧𝑒)_𝑖,𝑡 + 𝛽₄𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_𝑖,𝑡 + 𝛽₆𝑆𝑎𝑙𝑒𝑠𝐺𝑟𝑜𝑤𝑡ℎ_𝑖,𝑡+ 𝛽₇ln(1 + 𝑎𝑔𝑒)_𝑖,𝑡 +

𝛽₈𝐺𝐷𝑃_𝑖,𝑡+ 𝛽₉𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛_𝑖,𝑡+ 𝜖_𝑖,𝑡 (1)

I test the first hypothesis by performing the test of joint significance of the dummy variables. If the legal system dummy variables are jointly significant, then the legal system is a determinant of risk-taking and the first hypothesis is confirmed.

I carry out a multivariate analysis and the Wald test to examine, whether there is different managerial risk-taking under common and civil law countries. The multivariate analysis is performed on medians of Common and Civil by nonparametric equality-of-medians test in STATA. The Wald test, on the other hand, compares the regression coefficients of the two legal systems, in order to examine if the effect of the two legal systems on risk-taking is different.

(15)

𝜎(𝑅𝑂𝐴_𝑎𝑑𝑗)

𝑖,𝑡 = 𝛼 + 𝛽1𝐷_𝐶𝑜𝑚𝑚𝑜𝑛𝑖,𝑡+ 𝛽2𝐷_𝐶𝑖𝑣𝑖𝑙𝑖,𝑡 + 𝛽3ln (𝑆𝑖𝑧𝑒)𝑖,𝑡 +

𝛽₄𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_𝑖,𝑡 + 𝛽₅𝑆𝑎𝑙𝑒𝑠𝐺𝑟𝑜𝑤𝑡ℎ_𝑖,𝑡+ 𝛽₆ln(1 + 𝑎𝑔𝑒)_𝑖,𝑡+

𝛽₇𝐺𝐷𝑃_𝑖,𝑡+ 𝛽₈𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛_𝑖,𝑡 + 𝛽₉𝐷_𝑏𝑒𝑓𝑜𝑟𝑒_𝑖,𝑡 + 𝛽₁₀𝐷_𝑎𝑓𝑡𝑒𝑟_𝑖,𝑡 + 𝜖_𝑖,𝑡 (2) I test hypothesis 2a by performing the test of joint significance of the time dummy variables. If the time dummy variables are jointly significant, then the periods before, during and after are determinants of the managerial risk-taking. Next, I test legal system dummy variables for joint significance to examine, if the legal system is still a determinant of risk-taking after accounting for different time periods in the model.

In order to test hypothesis 2b, I examine each of the three periods separately following the regression (2) if before = 1, during = 1 and after = 1. Subsequently I test each of the three periods for joint significance of legal systems separately. Lastly, I perform the Wald test for each of the three periods to verify if the impact of common and civil law system on managerial risk-taking is different in each of the three periods.

3. Results

4.1 Role of legal system in managerial risk-taking

In order to verify hypothesis 1a, I begin with two unbalanced panel regressions, where I use main risk-taking variable in the first regression and robustness risk-taking variable in the second regression. The results of the subsequent test for joint significance of legal systems reveals if the legal system is determinant of the managerial risk-taking.

(16)

Table 5: Main risk-taking proxy and legal systems

This table presents results of the OLS unbalanced panel data regression for main dependent variable and alternative specifications of main dependent variable. Model 1 shows results for dependent variable calculated over 16 periods model 2 shows results for dependent variable calculated over 24 periods and model 3 shows results for dependent variable calculated over 32 periods. I use company-specific variables capturing size, total debt to total asset ratio, growth of sales and age alongside with macroeconomic variables capturing economic development and price stability as the independent variables. Common and Civil are dummy variables representing legal system under which a company operates. Hybrid is omitted. *** indicates that the coefficient is significant at 1% level, ** indicates that the coefficient is significant at 5% level and * indicates that the coefficient is significant at 10% level. Model 1 2 3 Variable RT16 RT24 RT32 ln(Size) -0.1104*** -0.0814*** -0.0689*** (0.0039) (0.0039) (0.0029) Leverage -0.0085 -0.0732** -0.0521** (0.0314) (0.0304) (0.0216) Salesgrowth 0.0151* 0.0330*** 0.0205*** (0.0084) (0.0077) (0.0050) ln(1+age) -0.0677*** -0.0919*** -0.0862*** (0.0109) (0.0116) (0.0105) GDP -0.0000*** -0.0000*** -0.0000*** (0.0000) (0.0000) (0.0000) Inflation -0.7758*** -0.2845* 0.8529*** (0.1933) (0.1704) (0.1100) Civil 0.0371 0.1132* 0.1535** (0.0557) (0.0610) (0.0721) Common 0.1047** 0.2033*** 0.2576*** (0.0509) (0.0560) (0.0669) Constant 2.1624*** 1.8411*** 1.6295*** (0.0754) (0.0779) (0.0768)

(17)

Both macroeconomic variables are negatively related to managerial-risk taking, however model (3) shows that inflation increases managerial risk-taking in the long run. Negative coefficients of GDP show that companies operating under low-income economies are likely to take more risk compared to those countries, where GDP per capita is high. The GDP coefficient seems economically insignificant. Inflation have largest effect on the managerial risk-taking; however, I do not measure inflation as a percentage, thus one must divide the coefficient by 100 to derive the effect of 1% change in inflation on managerial risk-taking.

Table 6: Robustness risk-taking proxy and legal systems

This table presents results of the OLS unbalanced panel data regression for robustness dependent variable and alternative specifications of robustness dependent variable. Model 1 shows results for dependent variable calculated over 16 periods model 2 shows results for dependent variable calculated over 24 periods and model 3 shows results for dependent variable calculated over 32 periods. I use company-specific variables capturing size, total debt to total asset ratio, growth of sales and age alongside with macroeconomic variables capturing economic development and price stability as the independent variables. Common and Civil are dummy variables representing legal system under which a company operates. Hybrid is omitted. *** indicates that the coefficient is significant at 1% level, ** indicates that the coefficient is significant at 5% level and * indicates that the coefficient is significant at 10% level. Model 1 2 3 Variable RT16 RT24 RT32 ln(Size) 0.5595*** 1.1828*** 2.8893*** (0.0337) (0.0881) (0.1390) Leverage -1.6218*** -3.4678*** -5.3293*** (0.2544) (0.6689) (1.0196) Salesgrowth -0.2170*** -0.1961 -0.6601*** (0.0655) (0.1668) (0.2416) ln(1+age) -0.5278*** 1.1688*** 0.5844 (0.1080) (0.2842) (0.4774) GDP 0.0001*** 0.0002*** 0.0003*** (0.0000) (0.0000) (0.0000) Inflation 24.0556*** 31.6515*** 52.0422*** (1.5268) (3.7365) (5.2753) Civil -3.1206*** -7.8906*** -17.5327*** (0.9421) (1.7580) (2.9554) Common 1.3674 1.2306 -3.6083 (0.8482) (1.6018) (2.7436) Constant -3.8245*** -16.9430*** -34.1928*** (0.9313) (1.9818) (3.3135)

(18)

For the main specification of the risk-taking variable, the effect of Civil and Common law system is significant in all models, except Civil in the first model. The effect of legal system dummy variables becomes larger for specifications where dependent variables are calculated over longer periods. Lastly, the test for joint significance of legal system dummy variables reports that the first model is significant at 5% and the second and third models are significant at 1%. This test confirms hypothesis 1a, that the legal system is determinant of the managerial-risk taking.

The effect of Civil law system on managerial risk-taking is solely significant in all three specifications of the robustness risk-taking, however Common is not significant. The test for joint significance of the legal systems reveals that the legal systems are jointly significant at 1% level in all specifications. This provides robustness to the claim that legal system is determinant of the managerial risk-taking.

Table 7: Multivariate analysis

This table displays results of the multivariate analysis. Variables ROA stand for the main dependent variables and CAPEX stand for robustness variables. The number in the name of the variable represents number of quarterly period required to calculate each specification of the main / robustness variables. The numbers under Common and Civil stand for the medians of each specification of risk-taking variable under the common and civil law system. I derive the p-value from STATA nonparametric equality-of-medians test.

Common Civil Diff p-value

ROA16 0.044039 0.031574 0.012465 0.00 ROA24 0.054477 0.036414 0.018063 0.00 ROA32 0.077931 0.060553 0.017378 0.00 CAPEX16 0.520842 0.391664 0.129178 0.00 CAPEX24 0.831649 0.398724 0.432925 0.00 CAPEX32 1.079478 0.604409 0.475069 0.00

(19)

The multivariate analysis illustrates different managerial risk-taking under common and civil law countries. Furthermore, I use the Wald test to compare impact of civil and common law variables on managerial risk-taking in my models. The Wald test is conducted on the Common and Civil coefficients of the prior two regressions.

The results in Table 8 show that for the main specification of risk-taking variable, the different impact of Common and Civil coefficients on managerial risk-taking is confirmed at a 5% level for variables calculated over 24 and 32 periods and at a 10% level for variables calculated over 16 periods. The robustness risk-taking specifications show, that the difference is significant at 1% level for all specifications. This test reveals that the effect of Common on managerial risk-taking is significantly higher than the effect of Civil. Furthermore, this test supports the claim of the multivariate analysis that companies operating under common law countries exhibit higher managerial risk taking.

Table 8: Wald test

This table displays results of the Wald test. Variables ROA stand for the main dependent variables and CAPEX stand for robustness variables. The number in the name of the variable represents number of quarterly periods required to calculate each specification of the main / robustness risk-taking variables. Difference is calculated from the coefficients of common and civil derived from unbalanced panel data regressions displayed in Table 5 and Table 6.

Common Civil Difference p-value

ROA16 0.1047 0.0371 0.0676 0.0599 ROA24 0.2033 0.1132 0.0901 0.0199 ROA32 0.2576 0.1535 0.1041 0.0166 CAPEX16 1.3674 -3.1206 4.488 0 CAPEX24 1.2306 -7.8906 9.1212 0 CAPEX32 -3.6083 -17.5327 13.9244 0

(20)

The presented findings are in line with the previous literature. Firstly, Liang and Renneboog, (2017) argue that there are fundamental differences between the legal systems in terms of shareholder’s and stakeholder’s protection. Common law countries are mainly focused on the shareholder’s rights, while civil law countries aim to protect stakeholders such as employees, customers and creditors. These differences in legal systems affects managerial risk-taking. Lastly, Acharya, Amihud, and Litov (2009) conclude that strong creditor’s protection decreases managerial risk-taking incentives. Furthermore, Windram (2005) shows that shareholder’s rights have an important implication on agency problems and managerial risk-taking behaviour. My findings show that companies operating under civil law countries experience lower managerial risk-taking, than companies operating under common law countries. This is in line with arguments of Acharya, Amihud, and Litov (2009) and Windram (2005), because civil law countries enforce more and stricter regulations to protect stakeholders.

4.2 Role of global financial crisis in managerial risk-taking

In order to verify hypothesis 2a, I begin with the unbalanced panel data regressions that I use to confirm first hypothesis; however, I extend these regressions by a time dummy variable to capture the effect of global financial crisis and the periods before and after the crisis. For these regressions, I only use main and robustness risk-taking variable calculated over 16 overlapping quarterly periods because the other two specifications do not provide sufficient number of observations for the period after the global financial crisis due to time constrains. Next, I perform the test for joint significance of the time dummy variable and legal system dummy variable to examine if the time dummy variable is determinant of the managerial risk-taking and if the legal system is determinant of the risk-taking after introducing time dummy variable. Additionally, I examine each of the three periods separately.

(21)

Table 9: Main risk-taking and legal systems during various time periods

This table presents results for OLS unbalanced panel data regression for main specification of dependent variable. Common and Civil are dummy variables to capture legal system under which a company operates. I use company-specific variables capturing size, total debt to total asset ratio, growth of sales and age alongside with macroeconomic variables capturing economic development and price stability as the independent variables. Hybrid is omitted. Before and after are dummy variables capturing time period of before and after the global financial crisis respectively. Model 1 shows results of the whole model, while models 2,3,4 examine each period separately. *** indicates that the coefficient is significant at 1% level, ** indicates that the coefficient is significant at 5% level and * indicates that the coefficient is significant at 10% level.

Model 1 2 3 4

Variable Full model Before During After ln(Size) -0.10737*** -0.14294*** -0.00879*** -0.00183** (0.0038) (0.0070) (0.0031) (0.0008) Leverage 0.01892 -0.10305** -0.12813*** 0.02089*** (0.0304) (0.0525) (0.0274) (0.0047) Salesgrowth 0.00778 0.03764*** -0.02950*** 0.00047 (0.0081) (0.0114) (0.0078) (0.0006) ln(1+age) 0.03101*** 0.02253 0.00047 -0.01618*** (0.0116) (0.0195) (0.0062) (0.0046) GDP 0.00000 -0.00000*** 0.00000*** 0.00000*** (0.0000) (0.0000) (0.0000) (0.0000) Inflation 0.56389*** 2.34322*** 0.93393*** 0.01896 (0.1893) (0.3965) (0.1409) (0.0405) Civil -0.00139 0.0999 -0.11673*** -0.18965** (0.0541) (0.0869) (0.0224) (0.0752) Common 0.10135** 0.17816** -0.05870*** 0.07525 (0.0493) (0.0820) (0.0206) (0.0683) before 0.27615*** (0.0071) after 0.20479*** (0.0078) Constant 1.54281*** 2.34606*** 0.17534*** 0.34679*** (0.0770) (0.1243) (0.0535) (0.0648) Observations 26346 9639 12002 4705

(22)

The effect of explanatory variables differs among the periods. Leverage is negatively related to the managerial risk-taking before and during the global financial crisis, however the coefficient becomes positive after the crisis. The opposite holds for the company’s age variable, where age of the company is positively related to the risk-taking, however insignificant and becomes negative and significant after the financial crisis.

The effect of GDP is negative in the period prior to the financial crisis; however, it becomes negative during the crisis and remains negative in the post-crisis period. It is obvious, that certain company specific and macroeconomic determinants of managerial risk-taking have different effect in various time periods.

Table 10: Robustness risk-taking and legal systems during various time periods

This table presents results for OLS unbalanced panel data regression for robustness specification of dependent variable. I use company-specific variables capturing size, total debt to total asset ratio, growth of sales and age alongside with macroeconomic variables capturing economic development and price stability as the independent variables. Common and Civil are dummy variables to capture legal system under which a company operates. Hybrid is omitted. Before and after are dummy variables capturing time period of before and after the global financial crisis respectively. Model 1 shows results of the whole model, while models 2,3,4 examine each period separately. *** indicates that the coefficient is significant at 1% level, ** indicates that the coefficient is significant at 5% level and * indicates that the coefficient is significant at 10% level.

Model 1 2 3 4

(23)

Table 10 displays results of the unbalanced panel data regression, where the robustness risk-taking proxy is used as a dependent variable. Again, the magnitude and trend of some macroeconomic and company-specific variables coefficients are not consistent with the model using the main risk-taking variable. However, the effect of the legal system dummy variables is consistent with the main specification. The difference between the coefficients for Common and Civil is positive. This suggests that companies operating under common law countries exhibit larger managerial risk-taking.

Test for joint significance of legal systems reveals that the legal systems are jointly significant at 10% level in the period before the global financial crisis in the main model and at 1% level for the remaining two periods of the main model and in all three periods of the robustness model. Test for joint significance of time dummy variable is significant at 1% level for main and robustness model. This provides evidence to confirm the hypothesis 2a that global financial crisis and the periods prior and after the crisis are determinants of the managerial risk-taking. Lastly, Table 11 shows that the differences between the common and civil law systems in the main model prior the global financial crisis were not significant, however during and after the financial crisis, these differences become recognizable, which is supported by the p-value of 0.00. The alternative specification supports the findings from the main specification for the periods during and after the global financial crisis, however alternative specification reports significant differences between the legal systems for the period before the financial crisis as well. All differences in Table 11 are positive, therefore effect of common and civil law systems is different, but also common law countries have larger effect on managerial risk-taking than civil law countries during all three time periods. This provides evidence to support the hypothesis 2b.

Table 11: Wald test

This table displays results of the Wald test. Variable ROA stand for the main dependent variable and CAPEX stand for robustness dependent variable. Numbers under Common and Civil shows

coefficients of unbalanced panel data regressions for common and civil law systems for periods before, during and after the global financial crisis. The coefficients are taken from Table 9 for main risk-taking variable and Table 10 for robustness risk-taking variable.

ROA CAPEX

(24)

In summary, my results provide evidence to support the claim that there are differences between the managerial risk-taking behaviour in various legal systems. The evidence shows that legal system has a recognizable implication on risk-taking, furthermore, the evidence reveals that companies operating under a country with the civil law system exhibit lower managerial risk-taking than the companies operating under a country with common law system. These findings are in line with the previous literature arguing that civil law countries employ more regulation and are focused on stakeholder’s protection, while common law countries are mainly focused on shareholder’s rights. The strict regulations lead companies to decrease risk-taking.

Lastly, my findings show that the global financial crisis has a significant effect on managerial risk-taking behaviour and that the implication of legal system remains significant. Furthermore, I find that the effect of legal systems on managerial risk-taking becomes more significant during the financial crisis. Moreover, the Wald test reveals that the effect of common and civil law countries became significantly different during the financial crisis and remained significant in the period after the crisis. This is in line with the previous literature. Countries have to implement new regulations and policies as a response to the global financial crisis (Claessens, Stijn, Kose, Laeven, and Valencia, 2013). Therefore, it is anticipated that change in regulatory environment affected managerial risk-taking behaviour.

4. Conclusion

I study the relationship between the managerial risk-taking behaviour and legal system under which the company operates. I employ macroeconomic and company specific variables to examine this relationship. I use two different measures of risk-taking, where each is calculated over three different time periods to add robustness to the research. Furthermore, I study the impact of the financial crisis on managerial risk-taking and the effect of legal systems under the conditions of the global financial crisis and the time periods before and after the crisis. I see my contribution in explaining the role of the legal system in managerial risk-taking decisions.

I conclude that legal system is a determinant of managerial risk-taking and that companies operating under common law countries exhibit higher risk-taking compared to those companies that operate under civil law countries. These findings hold across all three specifications of main and robustness risk-taking variable.

(25)

5. References

Acharya, V., Amihud, Y., & Litov, L. (2011). Creditor rights and corporate risk-taking. Journal of Financial Economic 102, 150-166

Apergis, N., & Cooray, A. (2018). The behaviour of interest rate spreads prior to and after the financial crisis: evidence across OECD countries. The Manchester School 86(5), 559-585 Bhagat, S., Bolton, B. J., & Lu, J. (2012). Size, leverage, and risk-taking of financial institutions. SSRN Electronic Journal.

Bromiley, P. (1991). Testing a causal model of corporate risk taking and performance. Academy of Management Journal 34(1), 37-59.

Bruno, V., & Shin, H. S. (2014). Globalization of corporate risk taking. Journal of International Business Studies 45(7), 800-820

Chen, Y., Truong, C., & Veeraraghavan, M. (2015). CEO risk-taking incentives and the cost of equity capital. Journal of Business Finance & Accounting 42(7-8), 915-946.

Claessens, S., Kose, M. A., Laeven, L., & Valencia, F. (2013). Understanding financial crises: causes, consequences, and policy responses. SSRN Electronic Journal.

Cociuba, S. E., Shukayev, M., & Ueberfeldt, A. (2018). Managing risk taking with interest rate policy and macroprudential regulations. Economic Inquiry 57(2), 1056-1081.

Faccio, M., Mura, R., & Marchica, M. (2011). Large shareholder diversification and corporate risk-taking. SSRN Electronic Journal.

Gennard, J. (2009). The financial crisis and employee relations. Employee Relations 31(5), 451-454.

Ghoul, S. E., Guedhami, O., Kim, H., & Park, K. (2014). Corporate environmental responsibility and the cost of capital: international evidence. SSRN Electronic Journal. Gonzzlez, V. M. (2014). Firm and country determinants of debt maturity. international evidence. SSRN Electronic Journal.

Goss, A., & Roberts, G. S. (2011). The impact of corporate social responsibility on the cost of bank loans. Journal of Banking & Finance 35(7), 1794-1810.

Grauwe, P. D., Ji, Y., & Steinbach, A. (2016). The EU debt crisis: testing and revisiting conventional legal doctrine. SSRN Electronic Journal.

(26)

Heinkel, R., Kraus, A., & Zechner, J. (2001). The effect of green investment on corporate behavior. The Journal of Financial and Quantitative Analysis 36(4), 431.

Jammalamadaka, S. R., & Bernstein, P. L. (1999). Against the gods: the remarkable story of risk. The American Statistician 53(2), 171.

Jensen, M. C., & Posner, R. A. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The American Economic Review 76(2), 323-329.

Jensen, M. C. (1968). The performance of mutual funds in the period 1945-1964. The Journal of Finance 23(2), 389.

John, K., Litov, L. P., & Yeung, B. Y. (2005). Corporate governance and managerial risk taking: Theory and Evidence. SSRN Electronic Journal.

Khademian, A. M. (2011). The financial crisis: a retrospective. Public Administration Review 71(6), 841-849

Lian, C., Ma, Y., & Wang, C. (2016). Low interest rates and risk taking: evidence from individual investment decisions. SSRN Electronic Journal.

Liang, H., & Renneboog, L. (2017). On the foundations of corporate social responsibility. The Journal of Finance 72(2), 853-910.

Litov, L. P., John, K., & Yeung, B. Y. (2006). Corporate governance and corporate risk taking: theory and evidence. SSRN Electronic Journal.

Magill, M., Quinzii, M., & Rochet, J. (2015). A theory of the stakeholder corporation. Econometrica 83(5), 1685-1725.

Merna, T., & Al-Thani, F. F. (2008). Corporate risk management. West Sussex, England: John Wiley & Sons.

Porta, R. L., Lopez-De-Silanes, F., & Shleifer, A. (2007). The economic consequences of legal origins. Journal of Economic Literature 46(2), 285-332

Reeb, D. M., Kwok, C. C., & Baek, H. Y. (1998). Systematic risk of the multinational corporation. Journal of International Business Studies 29(2), 263-279.

Windram, R. (2005). Risk-taking incentives: A Review of the Literature. Journal of Economic Surveys 19(1), 65-90.

(27)

6. Appendix

Appendix 1: Descriptive statistics of independent variables

This table summarizes my independent macroeconomic and company-specific variables used in the analysis. This table shows values of independent variables used in the analysis of main and robustness risk-taking variables calculated over 24 and 32 overlapping periods. The table includes mean, standard deviation, minimum, 0.25 quartile, median, 0.75 quartile and maximum. Company-specific variables capture size, total debt to total assets ratio, growth of sales and the age of the company and the macroeconomic variables control for economic development and price stability of the country, where the company operates.

Quantiles

Variable Mean S.D. Min 0.25 Median 0.75 Max Main risk-taking variable (24 periods)

Main risk-taking variable (32 periods)

Robustness risk-taking variable (24 periods)

Robustness risk-taking variable (32 periods)