Risk-taking on listed companies during the global financial crisis

(1)

Risk-taking on listed companies during the global

financial crisis

--Evidence from the Netherlands and China

Master thesis

DD MSc IFM

20

th

_{June, 2014}

Author: Fei Wang

Student Number: 2299097

E-mail: f.wang.4@student. rug.nl

University: University of Groningen

Uppsala University

Faculty: Faculty of Economics and Business

Faculty of Social Sciences

Specialization: Double Degree of:

MSc International Business and Management, Program: International Financial Management;

MSc. International Business and Economics.

(2)

Abstract:

This paper examines whether there is significant difference in the periods before and after the global financial crisis from the evidence of China and the Netherlands. To this extent, this research employs ordinary least square (OLS) estimation approach through a sample data on listed companies over the period of 2003-2013 (year 2008 is excluded). The empirical results from analysis reveal that findings from four-year data sample seem to ha more significant impact on risk-taking than those from five-year data sample. Moreover, the results for China show more significant difference on corporate risk-taking, compared to those for the Netherlands. Finally, as a complement, the whole estimation for both countries is reported in the Appendix. The implications for corporate management are presented as well.

(3)

The whole thesis includes the following contents:

1. Introduction

Risk management is one of the most important issues in corporate finance nowadays. In recent decades, due to the ongoing globalization, risk management has undergone a dramatic expansion in its significance, being transformed from an aspect of management control to become a benchmark of good governance for firms, banks, and many other organizations. Still, it has been a great challenge for relevant researchers to create effective risk management systems to monitor and decrease the risks under control.

Meanwhile, there are no unique definitions of risk management. First of all, risk management deals with risk mainly in terms of the likelihoods of given undesirable outcomes, according to Gray and Wiedemann (1999), whereas Fatemi & Luft (2002) state that modern financial theory is based on three major paradigms: rational wealth maximization, the risk/return tradeoff, and the no-arbitrage principle. Additionally, both from a theoretical and an applied perspective, risk management can be best understood within the context of these paradigms and their extensions to each of the three major areas of finance: corporate finance, financial intermediation, and investments.

As a matter of fact, risk-taking is playing a crucial part in risk management, affecting strategic decisions, its operation and corporate investments; it is imperative to be further researched in worldwide settings. Numerous standards for risk management practice have been produced by a variety of transnational organizations. However, the global financial crisis that erupted from the U.S.A in September, 2008 attracts people more attention to the equity markets with functionally imperfect mechanism. Furthermore, the rise of risk management is often linked to the intensification of auditing and control processes.

(5)

investment opportunities. The different political climate and conditions of countries are related to the risk-taking of entrepreneurs and corporate managers, according to the work of Tan (2001). For instance, China and the Netherlands belong to a transitional economy and advanced economy separately. Thus, the different political environments in these two countries lead to distinct firms’ growth.

On the other hand, at the micro-level, like the firm-level, which I would focus on, is under the control of the manager. So it will help us better understand firms’ management around the world if firms can deal well with the particular risks at this level. For example, as mentioned by Mitton (2002), corporate governance can become more important and critical within a financial crisis. One feasible explanation is that a crisis can force investors to recognize and benefit from deficiencies in corporate governance that existed all along (Johnson et al., 2000), then give manager pressure to be less risky. Thus, corporate governance can be a crucial factor affecting firm value and risk-taking during a financial crisis. And corporate governance is China is significantly different from the situation in the Netherlands.

(6)

to improve the risk-taking for corporate managers from the comparison.

In fact, there are several reasons for selecting these two countries. In the first place, as a developing country, China is a leading emerging economy with fast national development and potential further growth in near future, making it deserve to be paid much more attention to. Else, among many advanced economies in Europe, the Netherlands is currently recovering from the financial and economic crisis. Second of all, as an export-oriented nation, the Netherlands has already become China’s second largest trade economy among the EU countries; and they would further strengthen this international business relationship. Next, with different institutions, two countries will continue to undergo a variety of challenges in financial management. Moreover, during the last five to six years, an increasing number of researchers from academics and institutions have been searching for an answer and the cause of one of the largest crises in modern history. Lastly, this financial crisis did impact on the level of risk-taking to listed companies across the globe. Therefore, the paper’s research question is formed:

Does the global financial crisis have different impact on the risk-taking of listed enterprises of China and the Netherlands?

The study employs dynamic and panel dataset of econometric techniques to analyze to what extent the financial crisis in 2008 has impacted on the risk-taking of listed companies. Ordinary least square (OLS) and sensitivity tests are run with two different periods: pre-crisis period, as well as post-crisis period to compare their differences. Finally, the results would have meaningful implications for corporate managers, through improving the risk-taking of the corporate risk management and encouraging investment at the firm level. Sound risk management system and overall growth are fundamental to reducing the risk.

(7)

results, following the work of Boubakri et al. (2013). Risk-taking in financial firms is affected by a variety of factors, some of which are hard to get access. Second, two reference periods (pre-crisis period, post-crisis period) are compared with each other, to capture the trend in risk-taking levels and other relevant variables. Third, the focus will be on these two selected countries, while other research has mainly focused either on East Asian companies or on U.S. companies during the financial crises.

The remainder of this paper is structured as follows. At first, Section 2 comprehensively discusses the theoretical framework of this research. Further on, the methodology and data of this paper is given in Section 3. Subsequently, Section 4 presents the results of the research, followed by discussion. Finally, in Section 5, a conclusion can be offered, which includes some implications of this study, its limitations and future research directions.

2. Theoretical framework

It is known the importance of effective risk management has never been greater. From the recent financial disasters of financial institutions, there is no doubt that the need for useful risk management must have been emphasized urgently. And as the nature of the risks has changed over time, methods of measuring these risks have to adapt to specific situations. Moreover, corporate risk can be measured in many ways. E.g., Christoffersen and Diebold (1998) assumed that if volatility fluctuates in a forecastable way, then good volatility forecasts can improve financial risk management, and vice versa.

(8)

2.1 Introduction of the financial crisis

To begin with, it is of necessarity to define what a financial crisis is. As reported by Mishkin (1992), a financial crisis is defined in the following:

“A disruption to financial markets in which adverse selection and moral hazard problems become much worse, so that financial markets are unable to efficiently channel funds to those who have the most productive investment opportunities.”

With this definition, it is obvious that the crisis of 2008 is a financial crisis. It is generally accepted by researchers and economists that the global financial crisis has started in 2008, when several renowned financial institutions collapsed. At first, in one year (August, 2007–August, 2008), the U.S.A mortgage crisis has eventually become a deep financial crisis. Next, in the third quarter of 2008, the United States entered into a recession. Quickly, the crisis spread from the U.S.A towards Europe. Financial institutions and non-financial companies came in troubled water, while globalization and trade interdependence have already contributed to a decline in the economic growth rate of developing countries. However, for the purpose of modeling, this study uses January 1st of 2008 as the starting point of the financial crisis, just as several other eminent economists did in their research. From then on, the direct consequence of this financial crisis is that stock markets declined significantly and caused a change in the market value of companies. That is, the collapse on the stock markets globally contributes to decreasing market value of a great number of companies.

(9)

instruments, such as providing various derivatives, as a result of increasing systemic risk.

Meanwhile, the financial crisis also largely impacted on people’s daily life. For instance, customer confidence and consumption decreased. Eventually, this causes lower sales, and thus a decrease in profits for a company. As a result, the amount of the cash flows or dividends to shareholders would decrease. In return, the stock price of a specific company would decline. But it is certain that each company takes different risk to reduce the loss in the crisis. I claim that one plausible explanation of cross-firm differences is corporate governance. Based on previous research, companies with high quality corporate governance are expected to handle risks better than companies with relatively poor quality corporate governance, even during a crisis (Mitton, 2002).

2.2 Various factors pertaining to corporate risk

As mentioned earlier, different ways could be adopted to measure corporate risk. First of all, it can be done on a country-level basis; secondly, on a firm-level basis (Aebi et al., 2012) or on both levels together (Boubakri et al., 2013). Many researches can be used to help understand how several factors impact on risk-taking & risk management in financial as well as non-financial corporates over time, particularly in commercial banks.

Aebi et al. (2012) investigate whether risk management-related corporate governance mechanisms are associated with a better bank performance during the financial crisis of 2007/2008, such as the presence of a chief risk officer (CRO) in a bank’s executive board and whether the CRO reports to the CEO or directly to the board of directors. Hereby, corporate governance factors like CEO ownership, board size, and board independence are employed from the firm level.

(10)

the time-series standard deviation of ROA/ ROE.

Most recent academic studies have been addressing corporate governance and risk-taking (Cornett et al., 2010; Aebi et al., 2012 etc.). However, in this research, I put emphasis on non-financial firms to investigate the differences of risk-taking during the period of 2003-2013, in which the financial crisis occurred. The upcoming sections will discuss the impact of political institutions and country effects to the corporate risk-taking.

2.3 Risk-taking, political institutions and country effects

Most of the papers focus on the risk management techniques in financial firms, especially in banks, to examine risk exposure and capital structure through several indictors (Cebenoyan and Strahan, 2004, Aebi et al., 2012). However, working on an external factor to firms, Boubakri et al. (2013) investigate the impact of political institutions on corporate risk-taking. Employing a large sample of non-financial firms from 77 countries covering the period from 1988 to 2008, the authors find that sound political institutions are positively associated with corporate risk-taking, and that this relation is stronger when government extraction is higher. Furthermore, in a subsample of 45 countries, they also find that politically connected firms are more likely to be risk-taking, which suggests that close ties to the government lead to less conservative investment choices.

(11)

risk-taking in decisions (e.g. Roe, 2003).

Hence, a country’s political institutions or the extent of political constraints on the government can affect corporate risk–taking through both direct and indirect mechanisms. The measure of political institutions is the proxy of political constraints constructed by Henisz (2010), which covers several countries over a long period of time, and incorporates the political interactions among independent governments.

Additionally, other country factors also play a role in affecting the risk-taking. Such as, the assessment of the law and order tradition in the country differs in each country, and then impacts on corporate risk-taking decisions. Second, companies in a country with great rules of law can be better protected and have relatively low corruption. In China, there are a big proportion of firms that are connected to governments; the extent of political restraints may be stronger than countries in Europe (e.g. the Netherlands).

2.4 Risk-taking and corporate governance

In modern academics, corporate governance is a concept that has been widely used for explaining several movements in the economy. Before the relationship of corporate governance and risk-taking is being discussed, it is good to know the concept of corporate governance. Simply speaking, corporate governance refers to the system of rules, practices and processes by which a company is directed and controlled. Corporate governance essentially involves balancing the interests of the many stakeholders in a company, which include its shareholders, management, customers, suppliers, financiers, government and the community1_. However, there are many other definitions of it, which will not be explained here (See Johnson et al., 2000; Mitton, 2002; and Gompers et al., 2003). Up to the present, corporate governance is becoming increasingly important within companies to control hostile takeovers, especially after the introduction of

1_{http://www.investopedia.com/terms/c/corporategovernance.asp}

(12)

Sarbanes-Oxley Act (SOX).

The financial crisis following the subprime meltdown in the U.S.A has led to a further growing awareness and need for appropriate risk management techniques and structures within financial organizations. Mongiardino and Plath (2010) show that the risk governance in large banks seems to have improved only to a limited extent despite increased regulatory pressure induced by the credit crisis. They outline best practices in banking risk governance and highlight the need to have at least (1) a dedicated board-level risk committee, of which (2) a majority should be independent, and (3) that the CRO (Chief Risk Officer) should be part of the bank’s executive board.

Roe (2003) concludes that, under governments characterized by few checks and balances, ownership concentration is higher, which in turn will lead controlling owners to pressure managers to increase risk-taking. Additionally, because labor unions and other interest groups can impose lower risk-taking on managers, and since these interest groups are less influential under less constrained governments, risk-taking by managers is expected to be higher under more authoritarian governments (e.g., Pagano & Volpin, 2005; Roe, 2003). Corporate governance can generate influence on risk-taking decisions. According to Bae et al. (2012), companies with higher quality corporate governance perform greatly better than companies with lower corporate governance quality during a financial crisis.

2.5 Corporate risk-taking and the crisis

Based on the prior research, corporate risk-taking is more crucial and cautious when a financial crisis occurs.

(13)

bigger than that in the Netherlands. For instance, state-owned and state-controlled companies are able to have more capacity to cope with those risky activities due to their government’s intervention, which is particularly significant when a financial crisis occurs.

Moreover, it is widely acknowledged that countries between China and the Netherlands have suffered obviously different levels of impacts on national economic growth during the financial crisis of 2008. China is still in its transitional period to boost economic and financial development, even with new policy reforms; whereas, the Netherlands is already a developed economy, which cannot achieve a domestic growth rate as fast as China does. Compared to China’s situation, the system of law rules in the Netherlands is maturer, which seems that it can better monitor the corporate risk-taking levels.

There is abundant empirical support for the relationship between corporate governance and company risk-taking. Companies with high quality corporate governance cope with risks better than companies with lower corporate governance quality during a financial crisis, according to Mitton (2002), as well as Mongiardino and Plath (2010). Thus, companies with poor corporate governance quality will perform worse during the financial crisis of 2008. However, according to Wei and Zhang (2008), quality of corporate governance mechanisms increased to some extents due to improved government regulations after the crisis. Hence, the corporate risk-taking in different countries is much worth to be further investigated, accompanying with the disastrous financial crisis of 2008.

3. Methodology

(14)

control variables. Proxies regarding the concept of risk-taking will be given. Moreover, every variable in the model will be explained and discussed in terms of measurement.

3.1 Data selection and description

In this study, the data for non-financial companies from two distinct countries are collected mainly through Orbis & “Datastream” database. These countries are rather different in ownership control and financial management systems. The time period is from 2003 to 2013 in order to examine the differences of risk-taking before, during and after the global financial crisis. Given that financial firms use different calculations on varieties of ratios as well as growth rates, compared with those in non-financial firms, and because financial firms are more complicated and quite sensitive to a country’s political institutions, I choose to exclude those firms (SIC primary codes between 6000 and 6999). Additionally, because these selected enterprises can only be analyzed when their financial information can be found in “Datastream” database, the sample companies are confined to those listed enterprises. Data availability should be taken significantly into account in the first place, and Table 1 gives an overview of three major stock exchanges.

Table 1: Data sample

Country Stock Exchange Number of

observations The Netherlands Euronext Amsterdam 81 China Shanghai & Shenzhen 2331 Total 2412 Not enough data 1139

Included in the study 1273

(15)

countries. As discussed before, the starting point of the global financial crisis is the 1st of January 2008. Although it is debatable whether this should be the exact day, for analytical purposes, the period will be divided into two groups: pre-financial crisis (2004-2007) versus post-financial crisis (2009-2012), note that the crisis year (2008) is not included for accuracy. These selected enterprises will be categorized by country type (namely China versus the Netherlands). The profiles of sample companies can be acquired from “Orbis” database (see the following).

Pre-crisis period: 1st of January 2004 – 31st of December 2007; Post-crisis period: 1st of January 2009 – 31st of December 2012.

Here, four years’ data started from 2004 before and after the global financial crisis are needed, and I aim to compare the differences of risk-taking in those companies over both periods. Keeping the sample balanced is considered, as well. In addition, five years’ data will also be used later as a comparison to increase the representativeness. As shown in Table 1 above, 1273 companies are reporting four years’ data for the total samples, which consist of 1201 Chinese companies and 72 Dutch companies; afterwards, five years’ data are employed, which include 762 firms (e.g. 717 companies in China and 45 firms in the Netherlands) for this research. Very importantly, the volatility of earnings is required to be available over these years.

3.2 Dependent variables

Corporate risk-taking can be calculated in different ways. This research exerts two different measures. First, following Faccio et al. (2011), I measure risk-taking (Risk1) as the standard deviation of the firm profitability (ROA) over periods from 2003 to 2013, where ROA is calculated as the ratio of net income before interest, taxes, depreciation, and amortization (EBITDA) to total assets. I also estimate another measures of corporate risk-taking (RISK2) for robustness.

(16)

the 4-year interval. Then, these risk-takings are segmented for each of the two periods (pre-crisis and post-crisis period). For each individual period, these two measures can be estimated for each company. A brief description of each of these variables used in the study is provided in the Appendix, as well as their data sources (see Table 2).

3.3 Independent variables

This study focuses on firm-level data from two entirely different countries (the Netherlands and China) to examine to what extent the risk-taking on listed enterprises has differed during the selected period. And this study is primarily based on Boubakri, Mansi, and Saffar (2013). Much importantly, their internal variables are relevant to corporate risk-taking.

Descriptive statistics for key variables that are associated with risk-taking will be reported to undertake the statistical analyses. That is, several financial ratios, such as ROA, DTA, and retained earnings etcetera, need to be calculated to observe the differences during these time periods. What’s more, leverage ratios are lack, since these ratios can perform better in financial companies, instead of non-financial firms.

Firm-level variables studied:

ROA: Ratio of net income before interest, tax, depreciation, and amortization to total assets (Industrial Companies).

DTA: Ratio of total debt to total assets.

GROWTH: Total assets growth rate over each year;

(Current Year's Total Assets / Last Year's Total Assets - 1) * 100. SIZE: Log of total assets in US$.

3.4 Variables regarding the political institutions and country factors

(17)

POLITICAL: Measures the degree of political constraint of a country. It is derived from a model of political interaction that incorporates information on the number of independent branches of government with veto power, and the distribution of preferences across and within those branches. Higher scores indicate stronger political constraints and sound political institutions.

LAWORDER: Assessment of the law and order tradition in the country. This variable ranges from 0 to 6, with higher scores indicating greater rule of law in the country.

GDPG: Growth rate of GDP. And GDP is measured in constant 2005 US dollars.

According to the data of Henisz (2012), in the first place, the political scores for the selected years of China are the same at 0.00; while those scores for the same years of the Netherlands are around 0.64 except for year of 2011 at 0.14. All these suggest that the Netherlands has stronger political constraints and better political institutions than China. On the other hand, the scores for variable “LAWORDER” in the Netherlands stay at 6.0, whereas these scores in China keep constant at 4.5 till year of 2009; afterwards, they slowly fall down. Thirdly, in effect, China has been one of the leading emerging economies with fast national economic development in last two decades, which is about 10% growth in GDP annually, and it has the potential further growth in the future. As a developed economy, the Netherlands cannot be such a case with an annual growth at 2%, and growth being negative in years of 2009 and 2012.

Nonetheless, Faccio et al. (2006) demonstrate that politically connected firms are much more likely to be bailed out by the government in cases of distress, thus having influences on their risk taking decisions. That is, politically connected firms in countries that rescue them in case of trouble are more likely to undertake risky activities. Furthermore, this finding is consistent with the results in the paper of Boubakri et al. (2013).

(18)

at the firm level. The differences of risk-taking between pre- and post-financial crisis will be elaborated, as well as the changes between two economies.

Below I state the empirical model of risk-taking variables, considered that different levels of risky corporate operations can affect the volatility of returns to capital:

RISK1=α+ β1Firmcontrols+ β2Politicalindex + β3CountryVariables + Dummy

Years + η (1)

where RISK1 is a proxy of the standard deviation of firms’ ROA for each firm over 4-year periods, where ROA is computed as the ratio of earnings before interest, taxes, depreciation, and amortization (EBITDA) to total assets, based on Faccio et al. (2011); POLITICAL is Henisz’ (2012) index of political constraints; Country variables include real GDPG and LAWORDER, the proxy for rule of law; Firm controls refer to the set of firm-level independent variables (SIZE, DTA, GROWTH, and ROA); Year, and IND are dummies that control for year, and industry, respectively; and η is an error term. The focus in the analysis is the coefficient β1, which measures the sensitivity of corporate risk-taking to the firm level prevalent in the country.

3.5 Multiple regressions

The risk-taking for managers after a financial crisis is prone to risk-aversion, partly due to the pressure of investor protection (John et al., 2008). It is acceptable that the number of company observations for the Netherlands is quite smaller than that for China, since China has a substantially bigger national area than the Netherlands. From this point of view, there are many more sample firms for China in this research than firms for the Netherlands.

(19)

summarize the main descriptive statistics.

Table 2.1 Descriptive statistics of major variables for China

China pre-crisis period post-crisis period

Four years Variables Mean Median Std. Dev. Mean Median St. Dev. RISK 1 8.15 1.97 102.85 8.35 2.04 96.24 Observations No.: RISK 2 17.44 4.38 207.81 17.95 4.49 199.07 1201 ROA 4.87 3.65 50.39 3.90 4.07 54.43 DTA 0.43 0.30 3.72 0.29 0.27 0.28 GROWTH 16.13 10.84 31.42 280.95 12.96 534.57 SIZE 5.33 5.30 0.45 5.67 5.65 0.57 POLITICAL 0.00 0.00 0.00 0.00 0.00 0.00 LAWORDER 4.50 4.50 0.00 4.07 4.08 0.44

Five years Risk 1 4.61 2.32 7.85 9.18 2.20 108.33

Observations No.: Risk 2 11.29 5.54 19.76 21.81 5.38 248.00 717 ROA 4.07 4.13 5.25 3.71 4.32 54.06 DTA 0.30 0.29 0.26 0.28 0.27 0.26 GROWTH 18.78 14.99 22.59 143.04 13.30 281.77 SIZE 5.33 5.30 0.44 5.76 5.74 0.57 POLITICAL 0.00 0.00 0.00 0.00 0.00 0.00 LAWORDER 4.50 4.50 0.00 4.07 4.08 0.44

By taking a close look at Table 2.1, the values of mean, median and standard deviations for the samples are reported separately via countries. Two time periods of descriptive statistics for each country are displayed and compared. Firstly, it is good to see all the values for the mean and median in both countries are positive for the whole periods. However, it is noteworthy to consider that the high values of standard deviations in both countries are meaning differently.

(20)

as well as high inflation rates. When it comes to the country variable, the values of “LAWORDER” in the mean and median become a little smaller after the crisis, leading to a change on standard deviation from 0.00 to 0.48. Lastly, the global economic recession caused by the crisis partly has an impact on foreign sales of China’s exports companies and influences corporate risk-taking as well.

Similarly, when taken a better look at Table 2.2, due to the decreasing observations from 4 years to 5 years, it can be seen that many variables have relatively smaller results in the five years.

For the Netherlands, the financial crisis has been making national economic recession, leading to decreasing total assets growth etcetera. Both risk variables have bigger values in the post-crisis period, which indicates this crisis has an impact on the risk-taking in the Netherlands. Another evident thing is, unlike China’s case, the “LAWORDER” variable always keeps stable in the whole periods. Nonetheless, the scores for “POLITICAL” variable have become smaller in the post-crisis period. Therefore, it can be said that the choice on different time periods (4 years or 5 years) does matter to each variable.

Table 2.2 Descriptive statistics of major variables for the Netherlands

Netherlands pre-crisis period post-crisis period

Four years Variables Mean Median St. Dev. Mean Median St. Dev. Risk 1 8.91 2.94 29.45 12.76 3.69 47.86 Observations No.: Risk 2 20.02 6.89 68.70 26.82 8.31 97.29 72 ROA 8.15 8.38 11.95 4.38 4.06 26.52 DTA 0.23 0.20 0.28 0.25 0.21 0.33 GROWTH 25.99 11.60 54.95 9.01 2.69 47.48 SIZE 5.64 5.65 0.99 5.78 5.78 1.02 POLITICAL 0.64 0.64 0.01 0.40 0.40 0.31 LAWORDER 6.00 6.00 0.00 6.00 6.00 0.00

Five years Risk 1 6.44 3.21 10.55 6.92 3.65 9.17

(21)

Observations No.: ROA 7.32 8.25 12.21 4.16 5.23 13.26 45 DTA 0.22 0.23 0.15 0.21 0.20 0.13 GROWTH 16.27 10.57 39.72 5.05 4.48 11.97 SIZE 5.99 6.11 0.94 6.19 6.28 0.93 POLITICAL 0.64 0.64 0.01 0.40 0.40 0.31 LAWORDER 6.00 6.00 0.00 6.00 6.00 0.00

In the regressions, whether there are significant differences in each individual variable during these two defined periods are estimated. Besides, political institutions and various country characteristics at these two periods (based on yearly data) are related to help explain the changes of corporate risk-taking around the financial crisis. On the other hand, the firm-level set of independent variables belongs to the financial perspectives of corporate governance, which could be briefly complemented to further understand the changes in risk-taking within this period.

I simplify the regressions model in the following (internal factors to firms), given that country-level variables almost do not vary across time for both countries.

RISK1 = α + γ1ROA + γ2DTA + γ3GROWTH + γ4SIZE + Dummy Years + η.

(2) Table 2.3 (in the Appendix) summarizes the correlation coefficients for all variables of model (2) in the pre-crisis period as well as in the post-period. Because part of this table shows strong correlation between RISK1 and ROA (higher than 0.80), it is necessary to exclude ROA. Hence, Table 2.4 is provided without ROA to avoid autocorrelation.

Table 2.4: Correlation coefficients variables reported 1273 obs 2004-2007

RISK1 DTA GROWTH SIZE RISK1 1.0000

DTA 0.5901 1.0000

GROWTH 0.1141 0.1155 1.0000

(22)

RISK1 DTA GROWTH SIZE RISK1 1.0000 DTA 0.0885 1.0000 GROWTH 0.0196 -0.0232 1.0000 SIZE -0.2140 0.1168 -0.0514 1.0000 762 obs 2003-2007

DTA 0.4102 1.0000

GROWTH -0.1173 -0.0763 1.0000

SIZE -0.2862 -0.0173 0.0226 1.0000 762 obs 2009-2013

DTA 0.0818 1.0000

GROWTH 0.0050 -0.0073 1.0000

SIZE -0.2419 0.1343 0.0156 1.0000

Generally speaking, the different independent variables show no strong correlation with each other. So these variables can be used in the regression. And in each period, RISK1 always has a negative correlation with corporate size.

4. Empirical results

(23)

than 90% data in the entire samples.

4.1 Results of the multiple regression analysis

The multiple regression analysis is performed mainly for two purposes: 1) to acquire more understanding of the differences of companies in two countries on risk-taking before and after crisis period; 2) to compare the effects of four years and five years on risk-takings. The results are displayed in the following tables.

Table 3-1 and Table 3-2 offer the independent variable (RISK1) in relationship with firm set of variables for each time-period. The results of an alternative independent variable (Risk 2) with firm-level variables are also reported in the appendix. When taking a deep look at both tables (for four years and for five years), it can be seen that the results of pre- and post-crisis periods are much different from each other. Moreover, the signs of values in the pre-crisis period are not necessarily the same. As described before, this crisis has brought bad an influence to firms’ financial performance; it would be impacting the coefficients of variables before and after 2008, as well.

Table 3-1: for four-year period

Dependent Variable: RISK1_Panel A1:

Variables Pre-crisis Post-crisis

(24)

Adjusted R2 _0.684 _0.656 Durbin-Watson stat 1.947 1.985

Observations 1201 1201

Panel B1:

Netherlands 2004-2007 2009-2012 DTA Coefficient 8.285 7.896 t Statistic 13.494 7.910 p value 0.000*** 0.000*** Growth Coefficient 0.078 0.356 t Statistic 2.550 3.354 p value 0.013** 0.001*** SIZE Coefficient -5.785 -9.772 t Statistic -3.406 -4.673 p value 0.001*** 0.000*** C Coefficient 10.137 15.534 t Statistic 1.815 3.656 p value 0.074* 0.001*** R-squared 0.502 0.575 Adjusted R2 _0.490 _0.567 Durbin-Watson stat 1.974 2.142 Observations 72 72

(25)

and from 0.097, 1.058, 1.243 to 0.742 during 2009-2012. Wide-range fluctuations brought influences on corporate risk-taking to some extent.

However, at the same time-periods, all significant differences for variables are found at different levels of significance in Panel B1. In consequence, the effect is relatively weaker on “GROWTH” (all coefficients are less than 0.5) than on the other variables. More importantly, this Table 3-1 illustrates that there are indeed differences on corporate risk-taking before and after the financial crisis.

On the other hand, the results for five-year period in two samples can be found in Table 3-2 as follows.

Table 3-2: for five-year period

Dependent Variable: RISK1_Panel C1

China 2003-2007 2009-2013 DTA Coefficient 11.592 17.710 t Statistic 12.382 20.497 p value 0.000*** 0.000*** GROWTH Coefficient 0.007 0.000 t Statistic 0.632 0.850 p value 0.528 0.396 SIZE Coefficient -4.447 -10.397 t Statistic -7.994 -9.737 p value 0.000*** 0.000*** C Coefficient 6.421 12.102 t Statistic 8.976 8.882 p value 0.000*** 0.000*** R-squared 0.357 0.411 Adjusted R2 _0.353 _0.410 Durbin-Watson stat 1.987 2.023 Observations 717 717 Panel D1:

Netherlands 2003-2007 2009-2013

DTA Coefficient 15.073 -12.971 t Statistic 1.412 -1.187

(26)

GROWTH Coefficient 0.047 0.058 t Statistic 1.205 0.512 p value 0.235 0.611 SIZE Coefficient -2.413 -2.534 t Statistic -1.412 -1.590 p value 0.166 0.120 C Coefficient 9.096 16.186 t Statistic 1.943 2.937 p value 0.059* 0.006*** R-squared 0.287 0.329 Adjusted R2 _0.216 _0.262 Durbin-Watson stat 2.111 2.134 Observations 45 45

Note: notation* implies the statistical significance at a 10% level;

Note 2: ** implies the statistical significance at a 5% level;

Note 2: *** implies the statistical significance at a 1% level.

First of all, for the country of China, all variables except for “GROWTH” (pre-crisis: t statistic = 0.632, p value = 0.528; post-crisis: t statistic = 0.850, p value = 0.396) show significant results in selected periods. Even the coefficient of “GROWTH” is almost zero (= 0.0004), which implies little economic effect at all.

However, for the Netherlands, not a single independent variable is significant during the pre- and post-crisis periods. At the same time, DTA has different signs in these two periods (coefficient =15. 073; coefficient = -12.971). One acceptable reason is that after the global financial crisis, foreign sales of many companies are harmed, affecting the decisions of risk-taking. Surprisingly, only constants in both periods have significant show-ups (pre-crisis: t statistic = 8.976, p value = 0.000; post crisis: t statistic = 8.882, p value = 0.000). Inferred from the panel D1, it could be the small sample size that caused all variables insignificant in any levels, with a low adjusted R2_{and observations (Panel D2 also has the similar} results in the appendix).

(27)

the risk-taking than those in the Netherlands. Furthermore, there is no autocorrelation in the residuals, since all statistics are around 2 (See the Durbin– Watson test results in Panels above).

To conclude, it seems from the above that the results from a four-year window are more statistically significant than those from a five-year window for both economies. The next section will combine the results presented in this part with the theory discussed earlier.

4.2 Discussion

As is shown in the research, entire sample of 1273 listed companies are used for 4-year’s estimation, and a subsample of 762 companies are employed for 5-year’s. The corporate risk-taking after the financial crisis has been different. In the Appendix, regression for the whole samples containing two countries (Table 3-3) and an alternative choice of risk-taking (RISK2: Table 4-1 and Table 4-2) are displayed. In this section, results will be linked with existing literature.

In the first place, the literature provides enough evidence on the impact of politically associated companies to the political environment prevailing in the country. For instance, Boubakri et al. (2013) investigate the effect of political connections on corporate risk-taking through a large sample of firms from 77 countries are examined in this research, leading that it is difficult to measure political connection with governments. As mentioned before, the variable “POLITICAL” in China is at 0.00 in both periods, whereas in the Netherlands, it is at 0.64 in the pre-crisis, and then becomes an average of 0.4 in the post-crisis. Greater scores suggest stronger political constraints and sound political institutions, thus affecting corporate risk-taking. Moreover, another country variable “LAWORDER” in the Netherlands (6.0 in both periods) gets higher scores than that in China (from 4.5 in the pre-crisis down to 4.07 in the post-crisis), having an influence on corporate risk-taking.

(28)

the crisis period. Nonetheless, the effect of variable “GROWTH” on risk-taking is not that important, with low coefficients in each period. Unlike the theory from Aebi et al. (2012), the financial crisis was triggered by financial institutions, e.g. banks. Apparently, this research focuses on non-financial companies at the firm-level. As a contrast, in the five-year data sample, two of three independent variables for China demonstrate significant effects on risk-taking in two different periods. Only “GROWTH” shows no difference during the periods before and after the crisis, which means it is not a good proxy. To the opposite, no significant effect for the Netherlands can be found in three given variables, which is surprising and not in consistence with the findings in four-year data. Several explanations are possible. First, when comparing with Boubakri et al. (2013), the sample size is small in this case, due to time constraints and data availability. Apart from this one, another explanation is that corporate risk-takings in developing country and developed nation could not have the same impacts to defined variables during a financial crisis. It can explain why some variables are not significant any more in other cases. Also the variable “SIZE” shows a negative relationship with corporate risk-taking in all estimations, being in accordance with correlation of regression models. Note that all these variables are briefly discussed in the articles of Faccio et al. (2011) and Boubakri et al. (2013).

(29)

managers were changing over time.

As a consequence, any factors that impact on risk-taking can be improved to enhance risk management. Mitton (2002) argues that there is a relationship between company performance and corporate governance. This can help to find a relation between corporate governance and risk-taking. That is, companies can perform better with a good corporate governance quality, thus affecting risk-taking decisions in a good way, especially when a financial crisis occurs. The global financial crisis has different impacts on the risk-taking in China and the Netherlands. Moreover, the impacts on China appear to be more significant than those in the Netherlands during the whole time-periods. One possible explanation is that the sample size for the Netherlands is much smaller and its rules of law in the country are greater. Therefore, the general level of corporate risk-taking in China is higher than in the Netherlands. Finally, with different country situations and management system, more relevant variables should be added to make the results more significant.

5. Conclusion

5.1 Summary of this study

(30)

results show that findings from four-year data sample have more significant impacts on corporate risk-taking before and after the financial crisis, compared with findings from five-year data sample. In addition, China seems to have more significant effect on risk-taking in both time-periods. On the contrary, independent variables show no significance for the Netherlands during the period of 2009-2013.

This research broadens the comparison with adding a pre-crisis period and post-crisis period via two choices on time-periods. Excluding the year 2008 can get an unbiased result, which cannot be ignored. From the perspective of financial management, the empirical results and analysis are applied to the non-financial firms to further understand the effect of this severe crisis on risk-taking, but these financial variables cannot completely capture the effect on risk-taking. Hence, it can be said that many more variables including dummy variables need to be added, which will be discussed later on.

5.2 Implications about corporate management

Considering the results and discussion above, this research has a few implications for corporate manager and relevant authorities as well. Corporate risk-taking is much of importance in risk management for managers, particularly when a financial crisis erupts (Tan, 2001). Apart from the country level factors to risk-taking, it is imperative to create a sound risk management system to monitor and inspect it from the internal aspects.

From the external perspective, political institutions are more likely to take risky corporate decision-making decisions, based on Boubakri et al. (2013). For these companies, managers should increase the transparency of contracts, increase investment opportunities and get gradual growth. On the other hand, governments need to adjust their preferential policy to facilitate business. Meantime, better rules of law can be pursued by governments to attract investment and attain profits.

(31)

governance quality is also necessary. For example, the experience of executives plays a great role in corporate governance (Aebi et al. 2012); when companies carry out a good corporate governance mechanism, more risky risk-taking decisions can be taken than a poor governance mechanism does.

To conclude, it is very hard to judge whether a corporate governance mechanism is good or not, all managers can do is to continuously improve the mechanism to fit in the real situations over time.

5.3 Recommendations to academics

There are some limitations in this study as in any other research. Firstly, as discussed previously, the number of independent and controlling variables could be more, like in the article of Boubakri et al. (2013). This may have affected the results. Secondly, industry should be added to the study as a dummy variable to check whether it is significant to the corporate risk-taking. Focus in this research has been on variables that where available and easily to collect. Moreover, this research only takes companies listed in these two countries into account. The results can therefore be significantly different for, for example firms cross-listed in US or in other European Union, with regards to the financial crisis of 2008. Finally, the samples could be more extensive so as to offer more detailed findings. Such as, domestic companies and international companies should be divided (firm types), or stock market variables could be involved.

(32)

6. References:

Aebi, V., Sabato, G. & Schmid, M. (2012). Risk management, corporate governance, and bank performance in the financial crisis, Journal of Banking & Finance (36): 3213–3226.

Allayannis, G. & Weston, J.P. (2001). The use of foreign currency derivatives and firm market value, Review of Financial Studies (14): 243-276.

Bae, K.H., Baek, J.S., Kang, J.K. & Liu, W.L. (2012). Do controlling shareholders’ expropriation incentives imply a link between corporate governance and firm value? Theory and evidence, Journal of Financial Economics (105): 412-435. Boubakri, N., Mansi, S.A. & Saffar, W. (2013). Political institutions, connectedness,

and corporate risk-taking, Journal of International Business Studies (44):195– 215.

Cebenoyan, A.S. & Strahan, P.E. (2004). Risk management, capital structure and lending at banks, Journal of Banking & Finance (28): 19-43.

Cornett, M.M., McNutt, J.J. & Tehranian, H. (2010). The financial crisis, internal corporate governance, and the performance of publicly-traded US bank holding companies. Working Paper, Boston College.

Christoffersen, P.F. & Diebold, F.X. (1998). How Relevant is Volatility Forecasting for Financial Risk Management? Review of Economics and Statistics (82-1): 12-22.

Durnev, A. & Fauver, L. (2011). Stealing from thieves: Firm governance and performance when states are predatory, Working Paper, McGill University. Faccio, M., Marchica, M. & Mura, R. (2011). Large shareholder diversification and

corporate risk-taking, Review of Financial Studies (24-11): 3601–3641.

Faccio, M., Masulis, R. & McConnell, J. (2006). Political connections and corporate bailouts, Journal of Finance (61-6):2597-2635.

Fatemi, A. & Luft, C. (2002). Corporate risk management: costs and benefits,

Global Finance Journal (13): 29–38.

(33)

business cycles in Asian emerging economies, Journal of Asian Economics (21-3): 293-303.

Gompers, P.A., Ishii, J.L. & Metrick, A. (2003). Corporate governance and equity prices, Quarterly Journal of Economics (118): 107-155.

Gray, P.C.R. & Wiedemann, P. M. (1999). Risk management and sustainable development: mutual lessons from approaches to the use of indicators, Journal

of Risk Research (2-3): 201-218.

Grigor’ev, L. & Salikhov, M. (2009). Financial Crisis 2008: Entering Global Recession, Problems of Economic Transition (51-10): 35-62.

Henisz, W. (2012). The political constraint index (POLCON) dataset. Philadelphia: University of Pennsylvania.

John, K., Litov, L. & Yeung, B. (2008). Corporate governance and risk-taking.

Journal of Finance (63-4): 1679-1728.

Johnson, S., Boone, P., Breach, A. & Friedman, E. (2000). Corporate governance in the Asian financial crisis, Journal of Financial Economics (58): 141-186.

Lardy, N.R. & Subramanian, A. (2011). Sustaining China's economic growth after the global financial crisis, Peterson Institute.

Mishkin, F.S. (1992). Anatomy of a financial crisis, Journal of Evolutionary

Economics (2): 115-130.

Mitton, T. (2002). A cross-firm analysis of the impact of corporate governance on the East Asian financial crisis, Journal of Financial Economics (64): 215-241. Mongiardino, A. & Plath, C. (2010). Risk governance at large banks: have any

lessons been learned? Journal of Risk Management in Financial Institutions (3): 116–123.

Pagano, M. & Volpin, P. (2005). The political economy of corporate governance,

American Economic Review (95-4): 1005–1030.

Roe, M. (2003). Political determinants of corporate governance, Oxford: Oxford University Press.

Stulz, R.M. (1996). Rethinking risk management, Journal of Applied Corporate

(34)

Tan, J. (2001). Innovation and risk-taking in a transitional economy: A comparative study of Chinese managers and entrepreneurs, Journal of Business

Venturing (16-4): 359-376.

Wei, K.C.J. & Zhang, Y. (2008). Ownership structure, cash flow, and capital investment: evidence from East Asian economies before the financial crisis,

Journal of Corporate Finance (14): 118-132.

7. ACKNOWLEDGEMENTS

(35)

8. Appendix

Table 2: all variable descriptions

Variable Definition Data source

Corporate risk-taking

RISK1

Calculated as the volatility of firms’ ROA for each firm over 4-year periods, based on Faccio, Marchica, and Mura (2011)

Datastream

RISK2 Calculated as the difference between the maximum and _{minimum ROA reported over the 4-year interval.} Datastream

Firm-level variables

ROA Ratio of net income before interest, tax, depreciation, _{and amortization to total assets.} Datastream DTA Ratio of total debt to total assets. Datastream GROWTH Growth rate of total assets in each year. Datastream SIZE Log of total assets in US$. Datastream

Country-level variables

POLITICAL

Measures the degree of political constraint of a country. Derived from a model of political interaction that

incorporates information on the number of independent branches of government with veto power, and the distribution of preferences across and within those branches. Higher scores indicate stronger political constraints and sound political institutions.

POLCON Dataset (2012)

LAWORDER

Assessment of the law and order tradition in the country. This variable ranges from 0 to 6, with higher scores indicating greater rule of law in the country.

POLCON Dataset (2012)

GDPG Growth rate of GDP. And GDP is measured in constant 2005 US dollars.

World

(36)

Table 2.3: Selected sample correlation of the multiple regression models

1273 obs 2004-2007 RISK1 ROA DTA GROWTH SIZE RISK1 1.0000 ROA 0.8793 1.0000 DTA 0.5901 0.6832 1.0000 GROWTH 0.1141 0.1658 0.1155 1.0000 SIZE -0.1467 -0.0697 -0.1149 0.0661 1.0000 1273 obs 2009-2012

RISK1 ROA DTA GROWTH SIZE RISK1 1.0000 ROA -0.7628 1.0000 DTA 0.0885 0.1247 1.0000 GROWTH 0.0196 0.0240 -0.0232 1.0000 SIZE -0.2140 0.1307 0.1168 -0.0514 1.0000 762 obs 2003-2007

RISK1 ROA DTA GROWTH SIZE RISK1 1.0000 ROA -0.3960 1.0000 DTA 0.4102 -0.1889 1.0000 GROWTH -0.1173 0.3769 -0.0763 1.0000 SIZE -0.2862 0.2986 -0.0173 0.0226 1.0000 762 obs 2009-2013

RISK1 ROA DTA GROWTH SIZE RISK1 1.0000

ROA -0.8211 1.0000

DTA 0.0818 0.1425 1.0000

GROWTH 0.0050 0.0006 -0.0073 1.0000

SIZE -0.2419 0.1834 0.1343 0.0156 1.0000

Four years’ data: See the whole estimation of RISK 1:

Table 3-3 Dependent

(37)

Variables Pre-crisis Post-crisis Whole sample 2004-2007 2009-2012 DTA Coefficient 2.631 7.787 t Statistic 15.028 11.712 p value 0.000*** 0.000*** Growth Coefficient -0.029 0.001 t Statistic -2.540 2.110 p value 0.011** 0.035** Size Coefficient -8.645 -10.952 t Statistic -11.495 -7.835 p value 0.000*** 0.000*** C Coefficient 13.759 14.534 t Statistic 10.811 7.481 p value 0.000*** 0.000*** R-squared 0.683 0.636 Adjusted R2 _0.683 _0.634 Durbin-Watson stat 1.930 1.968 Observations 1273 1273 Five years’ data

Panel C2:

(38)

Regressions of Risk 2 are reported below:

Table 4-1 For four-year period:

Dependent Variable: RISK2_Panel A2:

China 2004-2007 2009-2012 DTA Coefficient 4.266 12.176 t Statistic 13.037 9.731 p value 0.000*** 0.000*** GROWTH Coefficient -0.102 0.001 t Statistic -3.960 2.015 p value 0.000*** 0.044** SIZE Coefficient -9.652 -20.033 t Statistic -11.148 -8.200 p value 0.000*** 0.000*** C Coefficient 10.930 17.499 t Statistic 10.705 8.044 p value 0.000*** 0.000*** R-squared 0.684 0.646 Adjusted R2 _0.683 _0.645 Durbin-Watson stat 1.948 1.984 Observations 1201 1201 Panel B2:

(39)

Durbin-Watson

stat 1.981 2.171

Observations 72 72

Table 4-2 For five-year period:

Dependent Variable:RISK2_Panel C2

China 2003-2007 2009-2013 DTA Coefficient 10.123 17.523 t Statistic 12.792 20.297 p value 0.000*** 0.000*** GROWTH Coefficient 0.028 0.001 t Statistic 0.954 0.821 p value 0.340 0.412 SIZE Coefficient -11.241 -15.574 t Statistic -8.034 -9.787 p value 0.000*** 0.000*** C Coefficient 15.842 20.441 t Statistic 8.893 8.944 p value 0.000*** 0.000*** R-squared 0.358 0.606 Adjusted R2 _0.355 _0.605 Durbin-Watson stat 1.990 2.018 Observations 717 717 Panel D2:

(40)

Durbin-Watson

stat 2.091 2.152

Observations 45 45

Table A-1 Hetereoscedasticity Test --White Test

White Test

2004-2007 F-statistic 7.88300 Prob. F(3,1269) 0.0671 Obs*R-squared 23.28951 Prob. Chi-Square(3) 0.0523 Scaled explained SS 176.93200 Prob. Chi-Square(3) 0.0476 2009-2012 F-statistic 48.84740 Prob. F(4,1268) 0.2458 Obs*R-squared 169.96910 Prob. Chi-Square(4) 0.1346 Scaled explained SS 452.32000 Prob. Chi-Square(4) 0.0511 2003-2007 F-statistic 37.09679 Prob. F(4,757) 0.3682 Obs*R-squared 124.88690 Prob. Chi-Square(4) 0.3217 Scaled explained SS 855.32500 Prob. Chi-Square(4) 0.1096 2009-2013 F-statistic 76.94698 Prob. F(4,757) 0.4291 Obs*R-squared 220.26400 Prob. Chi-Square(4) 0.3286 Scaled explained SS 826.26000 Prob. Chi-Square(4) 0.1073

Table A-2 Autocorrelation Test-- Breusch-Godfrey Serial Correlation LM Test Breusch-Godfrey Serial Correlation LM Test:

2004-2007 F-statistic 1.54464 Prob. F(1,1268) 0.2142

Obs*R-squared 1.54885 Prob. Chi-Square(1) 0.2133

Risk-taking on listed companies during the global financial crisis