Regional differences between family firms in the 2008/2009 crisis

(1)

1

MASTER THESIS

Regional differences between family firms in the 2008/2009 crisis

Master in Business Economics: Organization Economics - 60 ECTS

Guilherme de Vasconcelos Parra

10827110

Supervisor: Thomas Buser

Abstract

Using a large sample of family and non-family companies from 30 countries in Asia, North America and Europe, I tried to identify differences in the behavior of family firms across regions during the 2008/2009 crisis. I show that there are regional differences in the way family firms made capital expenditures during the crisis and that this difference is robust and valid across different levels of ownership. These differences, however, were not found for the other two variables examined: research & development (R&D) investments and financial leverage.

(2)

2

I. Introduction

Studies have shown that this firms with concentrated ownership account for 44% of large Western Europe firms (Faccio & Lang, 2002), over two-thirds of firms in East Asian countries (Claessens et al., 2000), and 33% and 46% of the S&P500 and S&P1500 index companies, respectively (Anderson and Reeb, 2003; Chen et al., 2009). Understanding how these firms behave is then relevant from a global perspective. Moreover, the unexpected global financial crisis of 2008 and 2009, which originated in the advanced economies and hit the rest of the world powerfully, with some regions being more affected than others. This exogenous shock provided a nice setting for an analysis of how firms behave in an economic downturn.

There is a vast literature on how family firms performance compares to the other types of entities but so far the evidence on this matter evidence has been ambiguous. Some papers find that being a shareholder in a family firm can yield better returns while others find the opposite1. By having unique characteristics, families may behave differently as they can, for example, try to perpetuate their ownership in the firm instead of maximizing shareholder value.

While research on publicly traded family firms has largely focused on performance differences relative to nonfamily firms, little is known about the differences in decision making in family firms across regions. For example, when Stan Shih, founder of Taiwanese computer making company Acer, was questioned about the low price of the company’s shares and a potential acquisition from competitors his response was that "U.S. and European management teams usually are concerned about money, their CEOs only work for money. But Taiwanese are more concerned about a sense of mission and emotional factors"2_{. The} statement captures how cultural differences may affect the daily operation of firms in different regions.This research intends to provide a better understanding of the family firm institution by comparing regions, which to the best knowledge of the author, has not yet been done, and add to scarce literature in the area of cross regional and cultural family businesses.

1_{See Wagner (2015) for a meta-analysis}

2_{As reported by Reuters:}

(3)

3

Using 10% ownership as a minimum threshold for a firm to be labeled as family-owned and building on the findings of Lins, Volpin and Wagner (2013) that showed that families indeed cut more investments than their peers and this affected their firms’ value, this paper questions if family firms behaved differently than non-family firms across regions during the 2008/2009 crisis. I try to answer this by looking at two investment variables, capital expenditures and research and development, as well as a financing variable, financial leverage. These variables should reflect the company’s (and its shareholders’) views on the future to verify the existence differences from other types of ownership.

In order to measure the developments of the recent economic crisis in the behavior of firms, I make use of a dataset containing family and non-family listed companies between 2006 and 2009 from 30 countries in Asia, North America and Europe. I find that the effect of the crisis on the capex differed significantly between family owned and non-family firms in Asia and that this effect is significantly different from the way European and North American family firms reacted to the shock.

This paper is structured as follows. Section II presents related literature on the topic, section III introduces the data and methodology, section IV contains results and I discuss and conclude my findings in section V.

II. Related Literature

Investments in firms that have a majority shareholder

Jensen & Meckling`s (1976) agency theory, introduced the concept of agency costs and Shleifer & Vishny (1997) give a simple description of how agency costs arise: “An entrepreneur, or a manager, raises funds from investors either to put them to productive use or to cash out his holdings in the firm. The financiers need the manager's specialized human capital to generate returns on their funds. The manager needs the financiers' funds, since he either does not have enough capital of his own to invest or else wants to cash out his holdings. But how can financiers be sure that, once they sink their funds, they get anything but a worthless piece of paper back from the manager? The agency problem in this context refers to the difficulties financiers have in assuring that their funds are not expropriated or wasted

(4)

4

on unattractive projects” (Shleifer & Vishny, 1997). To overcome these difficulties, the financier chooses to monitor the manager, assuring financial performance.

Large shareholders have higher incentives to monitor managers given they are entitled to a larger portion of the results, just as minority investors have lower incentives to do so. Fama & Jensen (1983) laid the foundation of the idea of the separation of ownership and control in which large shareholders develop a close relationship with management to protect their interests. Fama & Jensen (1985) claimed that large and undiversified shareholders might decide for investments that favor their own preferences, contrary to market-based decisions, that would favor the entire firm and, consequently, all other shareholders. For example, Gompers & Lerner (1998) found that, in the venture capital industry, these large shareholders decide for the firm considering their investment horizon, perhaps damaging long-term value of the business. Using a dataset from mergers and acquisitions, Croci & Petzemas (2010) showed how conflicts of interest might arise between large and powerful owners versus other stakeholders. Aligned with this view, Berkman et al (2009), using a sample of publicly traded Chinese firms, showed how large (and usually controlling) shareholders use their influence to affect decision making of the firm, enjoying of private benefits at the expense of minority shareholders.

The separation of ownership and control is specifically risky for minority investors in companies that have different classes of shares. This means that by owning perhaps only 10% of total capital, an investor can control, for example, over 50% of the company. In firms with such ownership structure, control mechanisms between shareholders and managers must be implemented to reduce the potential agency costs that may arise. Barontini & Caprio (2006) investigated western European firms and pointed that family firms show larger separation of control and cash flow rights. In such way, controllers are expected to take riskier approach towards the business as their financial downside is lower but they may enjoy of private benefits such as status and recognition if the riskier strategy is successful.

The study of Faccio, Marchica and Mura (2011) showed how shareholders’ portfolio diversification have a direct impact on risk preferences these shareholders impose on the firms they control. In their words, “poorly diversified controlling shareholders may choose to forgo some positive net present value (NPV) projects simply because they are too risky.

(5)

5

In contrast, well-diversified controlling shareholders are likely to invest in all positive NPV projects, regardless of these projects’ riskiness”. The study was done using a large sample (over one million observations) of European companies between 1999 and 2007 and made use of both cross-sectional and panel regressions to investigate the causality of risk behavior. They also make use of various robustness methods by including control variables, using fixed effects, instrumental variables and also exploiting a natural experiment of succession in which they analyze the behavior of heirs after they have “received” control and concluded that portfolio diversification itself leads to risky behavior.

Family firms investment decision

What is a family firm? Shleifer & Vishny (1986) use a 5% threshold to determine whether a company is a considered family owned or not. That being, a family firm is one in which an individual, or group of individuals from the same family, holding at least 5% of the voting shares, regardless of the fact the that the family may or may not have senior management or executive positions. La Porta et al (1999), uses a 10% threshold based on (i) the idea that it is a significant amount of votes and also (ii) that most countries demand a formal disclosure when a shareholder reaches this level. Brazilian securities law, for example, guarantees a seat in the board of directors for every shareholder that holds at least 10% of the total capital. Faccio & Lang (2002) use a more conservative notion in which for a firm to be family owned, the family must own at least 20% of the voting shares. Lins et al (2013) go even further and uses 25% as a threshold. I will 10% ownership as the main reference throughout the paper but also adding different levels of ownership, 25% and 50%, to confirm if different cutoff points affect results.

Anderson, Duru and Reeb (2012) studied the impacts that risk aversion and long-term perspective may have in a family firm’s investment policy. On the one hand, the company should minimize their risks by investing in the most secure projects. On the other hand, the family should be committed to the long-term viability and health of the firm. These two factors are somewhat conflicting since every investment bears risks. To deal with this, shareholders may decide to invest in different categories of projects: physical assets or research and development (R&D). Kothari et al (2002) and also Lev & Sougiannins (1996)

(6)

6

claim that R&D expose firms and shareholders to greater levels of uncertainty so it is expected that family firms would invest more in physical assets than in R&D. However, when considering the long-term perspective, Poterba & Summers (1995) report that managers decide for projects that have a short-term impact as it is their belief that the market fails to recognize the value of long-term investments. Prior research has showed that R&D spending results in long-term benefits for the company and these investments will ensure the continuity of the business (Hall et al, 2005). That being, family firms would then invest more in R&D (long-term) than widely held firms. The results showed that family run companies tend to commit less capital to R&D than widely held businesses and in contrast, family business invest more in capital expenditures (capex, ie. physical assets) as the value of these assets are more tangible, reinforcing the idea of risk aversion (Anderson et al, 2012). This suggest that the risk aversion factor has a more relevant importance than the long-term perspective that shareholders may have. One of the potential explanations for this is that owners of family firms usually do not have a diversified wealth portfolio, which in turn, increases the families’ incentives to minimize risks as their wealth depends on firm’s survival (Anderson & Reeb, 2004).

One concern about family ownership and investment policy is endogeneity, in which families decide to take (and keep) participation in a firm for its low risk investment profile. Anderson et al (2012) test this by using an instrumental variable two-stage least square (IV-2SLS). They used an exogenous and significant instrument (median income of the population of the county where the firm maintains its headquarters) and conclude that, even after controlling for endogeneity, family firms still commit a smaller amount of resources to investments when compared to non-family firms.

Corporate decisions in period of crisis

Existing literature show that having liquid assets is crucial to finance corporate spending during times of financial constraints or when credit becomes scarce (Fazzari, Hubbard, and Petersen, 1988, Bernanke & Gertler, 1989). Using a survey with European CFOs in the first quarter of 2009 Campello et al (2012) showed that, while the capital markets were virtually frozen, firms were using credit lines they had available with banks. Smaller,

(7)

7

less profitable and non-investment grade firms were the biggest beneficiary of that liquidity source to survive and fund their investments, even if that means borrowing at higher at costs.

There are a number of papers describing how firms financed themselves and how they behaved towards investments during the 2008 crisis. In perhaps one of the most broad surveys regarding the decisions firms took after the crisis, Campello et al (2010), used a large sample of CFOs in the US, Europe and Asia to understand how credit constraint affected their decisions. The results showed that those companies with more limited funds planned to cut more investment, employment and spending than unconstrained firms. Nearly half of the companies surveyed also claimed they had to delay or cancel projects that would generate future value for the company. According to them, the results found are statistically similar in all three regions. In an analysis of US corporate investment Duchin, Ozbas, and Sensoy (2010) also find that firms reduced their investment in the period of 2007 and 2008 and that, like Campello et al (2010), the amount of cash held by these companies is significant to determine the level of reduction.

In a research similar to this, Lins, Volpin and Wagner (2013) showed that family firms underperformed widely-held firms during 2008-2009 crisis. Using the unexpected crisis as a natural experiment, they make use of a difference-in-differences model using the years of 2006 and 2007 as pre-crisis period and 2008 and 2009 as crisis years. In this period large shareholders remain the same (as confirmed by Faccio et al (2011)) and the exogenous shock also made it possible to determine whether performance was affected by family ownership or if families self-select into firms with lower performance. Their model also includes controls for profitability, size, and firm and industry-year fixed effects. Results touch on the reasons for this underperformance and the conclusion is that family firms reduced their investments more than non-family companies in the period. Additionally, there were no other significant differences that might help to explain the underperformance, including financing decisions, such as amount of cash, leverage, credit lines and debt maturity, suggesting that the reduction in investment was the mean found by families to reduce their risk exposure and increase the likelihood of retaining control of the firm.

(8)

8

A paper by Kose et al (2003) estimated the impact that world, region and country business cycles have on a country-specific output, consumption and investment. Their results show the existence of global growth business cycle (world factor) and that has a significant impact in the growth of the variables in question, accounting for 15% of total variation of output, 9% in consumption and 7% of the investments. That being the case, it expected that a global downturn would also affect the same variables negatively.

More importantly, global cycles affect different countries/regions with distinct intensities. The world factor plays a less important role in explaining economic activity in developing economies when compared to develop ones and there is a large variation in those results across different regions. Comparing G7’s (Canada, France, Germany, Italy, Japan, the United Kingdom and the United States) medians to global medians the authors find significant differences. While the world factor explains 37% of output growth the G7 economies, only 15% of output growth in other countries is attributable to the world factor. This pattern also extends to consumption and investment growth: global business cycles captures almost 19% of the variation in investment, and approximately 36% of consumption growth volatility in the G7 economies, while the global average is 7% and 9%, respectively.

When observing specifically for at the impact in investments, world, region and country effects respond for 7.3%, 1.9% and 35%. Another 46.7% are idiosyncratic factors, meaning that business cycles do not explain approximately half of the variation, but internal aspects, possibly legislation and culture play an important role. According to Kose et al (2003), “the idiosyncratic behavior of investment volatility (…) is consistent with observed cross-country investment correlations: these correlations are low and generally lower than the cross-country correlations of output (see Christodoulakis et al., 1995; Christian Zimmerman, 1995)”.

Family firms cultural differences between regions

Quantifying cultural differences is not an easy task. Hofstede (1984) managed to create a ranking of countries based on several dimensions, such as power distance (measuring authority), uncertainty avoidance, individualism and masculinity. In his research, Hofstede used data from a company with offices in several countries and found that within this company, there were substantial differences in the way employees behaved. The idea is that

(9)

9

these differences are passed to organizations according where the companies are located. Hofstede finds that all four dimensions show “significant and meaningful correlations with geographic, economic, demographic and political national indicators” (Hofstede, 1984: p 11).

Carr & Bateman (2009), provided a rough data about some potential variation between North America, Europe and Asia. They tested whether family and non-family firms differed across regions in six dimensions: Return on capital employed (ROCE), Sales Growth, “International orientation”, Capital expenditures (capex), Research and Development (R&D) and sales and general administrative costs as a fraction of total sales. With a small sample of 65 family owned firms (according to the criteria of at least 10% if the company is publicly trade or 51% if its private) and “matched” this company with a non-family owned competitor, based on criteria such as size and industry (with a total sample of 130 firms). They then compared the averages of these companies to verify if there was a difference between the companies across regions. Their results show differences in the capex and R&D. Despite the differences, they do not report the statistical significance of their results, which does allow further interpretation.

Although not trying specifically to compare the differences between family firms across regions, it is interesting to note that there are some intriguing differences across is the difference between family firms. North American firms spent on capex 4% and 8% more than European and Asian companies, respectively. When looking at R&D the differences become much more interesting as European families invest as much as 4.9% of their sales into R&D but and north American and Asian only invest 3.5% and 1% making this difference as large as five-fold. Again, Carr & Bateman (2009) do not report standard deviations on their measures so not much can be said about the statistical significance.

When running a meta-analysis of financial performance of family firms versus non-family firms, Wagner et al (2015) find that companies owned by families have an small economically advantage over their peer group. In their analysis, they also control for Hofstede’s cultural dimension, such as individualism, power distance, masculinity and uncertainty avoidance. Although they do not find any statistical significance on the first two variables their results suggest that firms based on a individualistic and with high power distance environment are more profitable. Moreover, he finds that “family firms show higher

(10)

10

performance in low uncertainty avoidance countries, which are characterized by low degrees of regulation and an entrepreneurship-friendly environment. Family firms, which are owner-managed or owner-governed, benefit from such settings.” suggesting that there may be differences in family versus non-family firm behavior after all.

Institutional reasons that affect family and non-family firms differently

The majority of research on family firms tries to answer whether this form of organization is better or worse than other types of structures, looking at profitability and/or decision making within these firms. Some authors, however, explore reasons that make the predominance of family firms more common in one region than another.

Bhattacharya & Ravikumari (2001) present a model in which the existence of less developed financial markets, family business are bigger and last longer. In line with this, Burkart et al (2003) studied managerial succession in family firms and showed that the entrepreneurs’ decision to keep or sell the shares owned at a firm depends on the level of legal protection of minority shareholders. Under strong protection, the family would be better off hiring a professional manager to run the company and selling off their share. If the minority investor protection is weak, the agency problems that arise are too strong and there would be a significant discount on the price, therefore the family should keep the shares and run the company. These results do not come as a surprise given that a less developed and less transparent market implies higher discounts on the valuation of the company and therefore the family has reduced chances of cashing out the investment, thus leading to a larger and older family firm.

Another explanation for the success of family firms in some regions comes from Gilson (2007), in which he claims that families’ businesses beneficiate from the reputation building more than non-family ones in regions with poor commercial law. The “bad law” argument is based on the idea that commercial transactions only take place when the contract is self-enforcing, meaning that it would be in the party’s best interest to engage in the exchange regardless of the legal implications. In this case, the reputation of those involved becomes the best predictor of how the transaction will happen. The reason why families beneficiate more from this “market” is that transferring the values of reputation from one family member

(11)

11

to another is less costly and/or more understandable by the market than in the case of a professional manager. The likelihood that a descendant has the same values as the founder is larger than a non-family CEO, for example. Additionally, the family will always have the goal of perpetuating the company, while a hired executive could act only considering a shorter time frame (such as the length of his/her contract).

Adding to the legal perspective of institutions, Fisman & Khanna (2004) analyzed the behavior of business groups in India and found evidence that firms associated with these groups, which are usually family owned, are more likely to invest in regions with poor physical infrastructure when compared to firms that do not belong to any business group. Their interpretation to this finding is that these groups are better able to cope with the challenges that arise from undeveloped regions, specifically “because the scale and scope of groups, and the de facto property rights enforcement within groups in environments where legal enforcement is lacking, permit them to overcome some of the difficulties that impair production in underdeveloped regions”.

To conclude this section, we know that families are risk averse when it comes to making investments, as they prefer secure investments (capex) to riskier ones, such as R&D. This is also supported by the fact that families usually have a high portion of their wealth into one firm and, as shown by Faccio et al (2011), lower portfolio diversification leads to lower risk taking behavior. By controlling or taking participation in firms, families can also enjoy of private benefits at the expense of other minority shareholders, affecting overall performance and, during crisis years, Lins et al (2013) showed that firms indeed reduced their investments more than non-family firms and this was the cause for which family owned firms showed lower returns. Profitability also seems to be affected by local values, as Wagner (2015) finds a significant difference in performance and country level uncertainty avoidance. We also know that the level of development of the institutions in which the companies operate can have diverse impacts on family and non-family owned business. Furthermore, that global economic growth (or crisis) affects different regions with different intensities, creating the perfect setting to answer the research question: did the behavior of family firms, relative non-family firms, differed across different regions during the 2008/2009 crisis? With the answer to this question I provide a better understanding of the family firm “institution”

(12)

12

by comparing regions, which to the knowledge of the author, has not yet been done, and add to scarce literature in the area of cross regional and cultural family businesses.

Based on the research question above, I will test the following hypotheses:

H1: There was a difference between regions:

1a: There was a difference between regions on how family firms made capital expenditures during the crisis relative to other firms. I expect families to

1b: There was a difference between regions on how family firms made R&D investments during the crisis relative to other firms

1c: There was a difference between regions on how family managed their debts during the crisis relative to other firms

H2: Stronger family control is associated with more conservative measures:

2a: Stronger family control is associated with higher cuts in capex during the crisis

2b: Stronger family control is associated with higher cuts in R&D during the crisis

2c: Stronger family control is associated with lower debt in the crisis

III. Data and methodology

To answer the research question if family firms behaved differently in the 2008/2009 crisis I run an empirical model using data from listed firms during the period of 2006 until 2009. For the purposes of this research, I focused on firms that were listed on any stock exchange in the years prior and after the 2008 crisis. When running robustness tests I alter the time period to check if the results hold even under different time frames.

The starting point of the research is the definition of a family firm. Following the La Porta et al (1999), I will use a 10% ownership threshold that, according to the authors, grants the owners a significant amount of votes in the company and possibly a seat on the board. In

(13)

13

addition, I also run the model in which a firm is considered to be family owned if has at least one shareholder holding a minimum of 25% or 50% of the voting capital.

The data was taken from the Osiris database. This database is global and provides information on listed, delisted and unlisted firms, in which is possible to sort firms according to their shareholder base. Osiris distinguishes a firm’s shareholders by its type. This is important because is how I can track if the shareholder belongs to a family or not. The different types of shareholders available in Osiris are: Banks and Financial companies, Insurance companies, Industrial companies, Private Equity firms, Hedge funds, Venture capital, Mutual & Pension Funds/Nominees/Trusts/Trustees, Foundations/Research Institutes, Public authorities, States, Governments, One or more named individuals or families, Employees/Managers/Directors, Public (publicly listed companies), Unnamed private shareholders and Other unnamed shareholders. According to Osiris, shareholders designated by more than one named individual or families are in the family category. “The idea behind this is that they would probably exert their voting power together”. By selecting only individuals or families, I am able to eliminate all firms that are not family owned.

The data extracted from Osiris tracks the control rights rather than cash flow rights. This means that if a company has different classes of shares, separating control rights from patrimonial rights, only the first is available. This is a relevant topic, as pointed out by Barontini & Caprio (2006), family firms are particularly prone to having this type of ownership structure. Osiris is able to track the direct and indirect ownership, following the structure of the shareholders until the very last level. This means that pyramidal ownership structures are captured and that the data reflects, in fact, the ultimate owner of the shares, and not only the direct ownership (which is the one usually required by financial regulators).

It is also possible to identify shareholders of the firms across time. Therefore, to be qualified as a family firm, the family or individuals must be listed in Osiris as shareholder with more than 10% (25% or 50%) of the voting shares in the year of 2008. This is also important as the shareholder base can change over time. Existing research (La Porta et. al., 1999) claims that this change is negligible but, having the option to do so, I have added this filter. The database is constantly being updated, as Bureau van Dijk, the company responsible for Osiris database, cannot instantly update changes in all firms as it demands some tracking

(14)

14

for the indirect control. By using the information on the year of 2008 I can assure that all the information reflects the real shareholder base at the time.

Following Franks et al (2012), Lins et al (2013) and Duchin et al (2010), I included minimum financial requirements for a company to be in the sample. I have excluded from the sample companies operating in the financial and insurance segments, companies that that have total revenues lower than USD10 million, negative assets and that did show an asset base variation larger than 100%, which could indicate restructuring. The revenue and negative asset filters apply to the year of 2008 while the asset variation filter refers to 2006 and 2010 period. Any observation with a negative value is also dismissed as this may indicate error in the database. Also dropped from the sample are companies based in countries for which there is no calculation for the anti-self-dealing index and firms based in countries with less than 100 listed companies (both family and non-family).

The final sample is composed by 13,560 firms, for four years, totaling 54,237 observations. Table 1 provides more details on the sample according to regions and ownership threshold criteria.

It is important to note a significant difference in the number of observations of the R&D variable. This information is manually collected by Osiris database and is on companies’ annual reports but the disclosure of values is, sometimes, not required by financial regulators. This means that some firms choose not to report the amount while others do. Additionally, some companies may report one expense in their income statement but the same company also details their R&D expenses on the notes section, which can have different values than the income statement. This indicates, that the criteria of what is considered R&D may not be properly defined within the company and the financial regulators, making this variable a weak indicator of a firm’s view of the future. Considering this variable alone, the sample is significantly reduced to 21,476 observations.

(15)

15 Model and definition of variables

I use difference-in-differences (DiD) estimation model to answer whether companies behaved differently after the crisis. To do so, I analyze three specific variables: capital expenditures, research & development (R&D) spending and financial leverage.

The DiD model allows me to compare two different groups (family and non-family) in two different periods (pre and post-crisis). I verify if the exogenous shock of the crisis have impacted both groups with different intensity. One important issue that arises from this method is that both groups (family and non-family) must share the same common trend prior to the crisis and, as shown in Figures 1 to 3, the common trend seems to be satisfied. DiD also allows me to remove any differences there may be between control and treatment groups and this is particularly good because it can compensate the fact that in some region, the distance between family and non-family behavior can be larger than in others. By using DiD I can also dismiss effects that are caused by characteristics that do not vary during the years being research, such as cultural behaviors. This allows me to run the model with less controls without losing precision.

The use of DiD also permits the correct interpretation of results when comparing family firms across different regions. By analyzing each region independently and then testing the coefficients for differences, I ensure that the coefficients reflect only the ownership criteria and can rule out other impacts to the variable. For example, by running only one simple model on the behavior of family firms and comparing these firms across regions, I could have biased coefficients by not including non-family firms. If so, the difference in the behavior could have simply been because one region was more affected by the crisis than the other,

Family Non-family Family Non-family Family Non-family

Asia 1,451 6,339 640 7,150 134 7,656

North America 199 3,142 72 3,269 25 3,316

Europe 399 2,030 263 2,166 131 2,298

Total 2,049 11,511 975 12,585 290 13,270

Region

Table 1 - Number of firms in the sample according to ownership criteria

(16)

16

making it impossible to distinguish if there was any difference in the behavior of family firms compared non-family ones.

I define the model according to the formula below, in which I run each model independently for each dependent variable (capex, R&D and leverage). Having data on three regions, I run a separate regression for each region and then test if the coefficients of these regressions differ among themselves. The model is defined as:

𝑌_𝑖 = 𝛽₀+ 𝛽₁(𝑓𝑎𝑚𝑖𝑙𝑦_𝑜𝑤𝑛𝑒𝑑) + 𝛽₂(𝑐𝑟𝑖𝑠𝑖𝑠) + 𝛽₃(𝑓𝑎𝑚𝑖𝑙𝑦_𝑐𝑟𝑖𝑠𝑖𝑠) + 𝛽₄(𝑐𝑜𝑢𝑛𝑡𝑟𝑦_𝑑𝑢𝑚𝑚𝑦) + 𝜇_𝑖

In which the dependent variable, Yi, is either Capex, R&D or Leverage.

To increase the precision of the model, I have included country level fixed effects, which, according to the existing literature, helps to explain decisions made by firms. These fixed effects are able to capture, for example, minority investor protection level and stock market development. Additionally I also added a dummy variable to control for pre and post crisis, a dummy for family owned and an interaction variable across the two dummies.

Capex: captures the amount spent in capital expenditures. Pre-crisis defined as 2006

and 2007 while crisis years are 2008 and 2009. Calculated in percentage of total annual assets.

RD: captures the amount spent in research & development. Pre-crisis defined as 2006

and 2007 while crisis years are 2008 and 2009. Measured in percentage of total annual revenue of the year of 2006. I keep the initial period as a reference to prevent that changes in the revenue affect the variable

Leverage: captures the amount of debt used to finance a firms’ assets. Pre-crisis defined

as 2006 and 2007 while crisis years are 2008 and 2009. Measured in percentage of total annual capital (debt+assets).

Crisis: dummy variable equal 1 if the years of 2008 and 2009. Zero in the year of 2006

and 2007.

Family: dummy variable equal 1 if the firm has at least one individual or family of

(17)

17 Family_crisis: interaction variable equal to one if firm is family owned in the years of

crisis. Zero otherwise.

Descriptive statistics

Table 2 provides a descriptive analysis the main variables. It is possible to note that the median firm has annual sales of USD193 million, assets worth USD223 million, annually invests the equivalent of 3.3% of its assets in capital expenditures and 1.8% of its revenues in research and development and has a financial leverage of nearly 35%.

Capex

Figure 1 and Table 3 provide a summary of the behavior of firms from 2006 until 2009. There is a clear trend, in all regions, of declining investments until 2009. From the average investment of 5.7% in 2006, one year prior to the crisis, average investment in the three regions was significantly cut to 3.8% in 2009, a remarkable 34% reduction.

Other differences arise when comparing regions. In Asia, family firms invest more than non-family firms do while in both North America and Europe, family companies are investing less. Additionally, there is clear difference in the average level of investments as North American firms invest more companies in other regions while Asia and Europe present similar level.

Figure 01 - Average capex spend per type of firm and region between 2006 and 2009. Capex measured as a percentage of total assets. Family firm at 10% ownership

Variable Unit N Mean 25th pctl. Median 75th pctl. SD

Revenues USD MM 54,240 1,932 27 193 762 10,146

Assets USD MM 54,240 2,346 71 223 865 12,873

Capex % assets 54,240 0.050 0.014 0.033 0.065 0.060 R&D % of revenues 21,476 0.047 0.005 0.179 0.051 0.084 Leverage % of total capital 54,240 0.332 0.268 0.348 0.405 0.105

Table 2 - Descriptive statistics of variables

Revenues and assets are in USD million. Capex is the ratio of capital expenditures to total assets. R&Dis the ratio of total research and development expenses to revenu. Leverage is the ratio of total debt to total capital.

(18)

18

Research and Development (R&D)

2006 2007 2008 2009 0.061 0.058 0.053 0.038 (0.068) (0.062) (0.055) (0.046) 0.058 0.057 0.053 0.041 (0.082) (0.082) (0.073) (0.062) 0.045 0.047 0.046 0.032 (0.054) (0.050) (0.047) (0.037) 0.058 0.559 0.052 0.037 (0.067) (0.062) (0.056) (0.046) 0.054 0.051 0.049 0.037 (0.068) (0.055) (0.051) (0.046) 0.064 0.062 0.060 0.041 (0.081) (0.071) (0.071) (0.049) 0.055 0.053 0.050 0.036 (0.075) (0.061) (0.052) (0.044) 0.057 0.055 0.052 0.038 (0.075) (0.061) (0.052) (0.044) Family threshold set at 10% ownership. Capex measured as a percentage of total assets. Standard errors reported in parenthesis.

North America Europe Family Total Non-family Total Non-family-owned firms Asia Europe Capex Family-owned firms Asia North America

(19)

19

In general, the investment in R&D shows less changes than capex, with few significant variations between 2006 and 2009 in both family and non-family firms. The exception is the behavior of family firms in North America, which showed a 22% reduction in 2008 when compared to 2007. This could possibly explained by the factor that the financial crisis originated in the US but only became a global economic crisis later.

R&D expenditures vary significantly between regions. The most notable difference is the level of investment chosen by North American firms. While European and Asian companies invest on average (median), 5.7% and 2.6% of their revenues into R&D, the average company in either US or Canada invest 9.9%. Comparing the global behavior of family and non-family firms it possible to note that the latter invests more but this is only true to Europe. In the US, family firms were investing more the non-family before the crisis but this changed after 2008. In Asia, both types show a similar increasing pattern after 2008. These numbers were reached by only using a sample of companies that reported any R&D expense higher than zero, which could bias the result upwards since this excludes two different reporting, the ones that did not report anything and the ones that that reported zero and actually invested nothing. Figure 2 and Table 4 provide more details on the numbers by regions.

Figure 2 - Average R&D spend per type of firm and region between 2006 and 2009. R&D measured as a percentage of total revenues.

(20)

20 2006 2007 2008 2009 0.025 0.024 0.027 0.030 (0.038) (0.035) (0.042) (0.048) 0.116 0.109 0.084 0.095 (0.183) (0.169) (0.109) (0.139) 0.046 0.039 0.040 0.044 (0.953) (0.054) (0.052) (0.059) 0.035 0.033 0.033 0.037 (0.075) (0.064) (0.053) (0.063) 0.023 0.023 0.026 0.027 (0.034) (0.036) (0.043) (0.042) 0.099 0.096 0.097 0.101 (0.123) (0.120) (0.120) (0.121) 0.055 0.054 0.056 0.057 (0.093) (0.098) (0.103) (0.103) 0.049 0.048 0.051 0.052 (0.088) (0.085) (0.088) (0.088)

Table 4 - Average R&D invested by firms

Asia

North America Europe Non-family Total

Family threshold set at 10% ownership. R&D measured as a percentage of total revenues. Standard errors reported in parenthesis

Non-family-owned firms R&D Family-owned firms Asia North America Europe Family Total

(21)

21

Leverage

When it comes to taking debt, family firms show a more conservative approach and this seems a stable treat in all regions as it possible to note in Figure 3 and Table 5. In 2008, with the capital markets virtually frozen, firms made use of their credit lines leading to a higher debt ratio (Campello et al, 2012). After that year, the behavior appear to follow the same trend it had before 2008.

Figure 3 - Average leverage of firms by type and region between 2006 and 2009. Leverage measured as a percentage of total capital (ratio of debt to debt+assets).

(22)

22

IV. Results

This chapter is divided in sections, one for each of the dependent variables being analyzed: capex, R&D and leverage.

Capex

In Table 6 I verify if family controlled firms have different policies regarding capital expenditures during the crisis relative to a sample of non-family controlled firms. It is possible to note how large was the impact of the crisis on the decision to invest. In all the regions being researched, firms reduced their spending from 0.01 up to 0.012 of their assets. This effect may seem small but the average pre-crisis investment level was between 5.3% and 6.2% of the assets, leading to an economical and statistical relevant reduction, across all

2006 2007 2008 2009 0.321 0.319 0.318 0.311 (0.096) (0.097) (0.101) (0.102) 0.309 0.315 0.326 0.319 (0.128) (0.135) (0.131) (0.134) 0.338 0.338 0.345 0.338 (0.096) (0.916) (0.093) (0.099) 0.323 0.322 0.324 0.317 (0.100) (0.101) (0.103) (0.106) 0.336 0.332 0.332 0.328 (0.097) (0.098) (0.103) (0.010) 0.320 0.323 0.336 0.329 (0.115) (0.116) (0.116) (0.123) 0.348 0.347 0.356 0.350 (0.100) (0.093) (0.094) (0.096) 0.334 0.332 0.338 0.332 (0.104) (0.102) (0.106) (0.109) Non-family Total Europe Non-family-owned firms Family Total Asia North America North America Europe

Table 5 - Average leverage of firms Leverage

Family-owned firms Asia

Family threshold set at 10% ownership. Leverage calculated as the ratio of debt to total capital (debt+assets). Standard errors reported in parenthesis

(23)

23

regions, of 19% for Asia, 19% for North America and 21% in Europe. These results are the same in all three different ownership levels, 10%, 25% and 50%.

Family ownership does seem to affect the level of investment of firms, although results vary according to the region. In Asia, I only find statistical significance at 25% ownership and with a moderate positive effect of 0.003, meaning managers of these firms invest 6% more than non-family firms, when compared to the region pre-crisis mean of 5.4%. In Europe, a family firm owned at the 10% and 25% level invest 0.007 less than other firms in the region, meaning a 13% reduction and at 50% ownership I find no significance. In North America, results are only relevant with the existence of a shareholder with 25% or more. In this region a family controlled firm would invest 0.021 less than a non-family firm, which means as 34% reduction. At the 50% ownership level, the difference is even higher as they invest 0.029 less, an impressive 47% difference.

Moving to the variable of most interest, family_crisis, the Asian sample is the only one that presents statistical significance at 10% ownership. In this region, family firms decreased their capex by 0.004 more than other firms and this would be equivalent of reducing 7% of the region average pre-crisis investment level. However, even though there is no significance in the North American and European coefficients individually, when running a joint test on the three coefficients to verify if they differ from each other I find a p-value of 0.015, meaning I can reject the hypothesis that they are the same at the 5% significance level. The results are robust and significant across all three specifications of ownership (p-values of 0.017 and 0.0995 for 25% and 50% ownership, respectively), thus confirming the hypothesis H1a in which family firms in different regions did behave differently in the 2008/2009 crisis when compared to non-family firms.

Looking at how level of ownership affects capital expenditures of family firms, I move to the 25% threshold. Just as in the 10% ownership, Asian family firms are the only ones with statistical significance, with a coefficient of -0.007, which is equivalent to 13% of the pre-crisis investment level. With 50% control none of the regions are statistically significant, although the coefficient of Asian is slightly more negative at -0.008. With these results, there is weak evidence to support hypothesis H2a.

(24)

24

A closer look on the tests across regions shows that, at 10% ownership, only Asian and European family firms differed between each other relative to non-family firms (p-value 0.005). When moving to 25% it is possible to note a significant difference between Asian versus North American (p-value 0.043) and Asian versus European (p-value 0.014). And finally at 50% ownership only Asian and North American firms differed (0.037). There were no significant differences between European and North American family firms (relative to non-family firms).

In order to confirm the results that support hypothesis H1a I also run a robustness test, using a subsample of the data. The dataset is composed by the same firms as in Table 6 but only using half of the observations (years of 2006 and 2009). As seen in Figure 1, the effects of crisis on capex are more severe in 2009, while in 2006 all regions had their highest investment level of the four-years sample. By doing so, I expected the results to be enhanced as there is a greater distance in the capex of pre-crisis and crisis. Table 9, in the appendix, confirms results as the tests on family_crisis coefficients differences maintain their significance (p-values of 0.014, 0.002 and 0.099 for 10%, 25% and 50% ownership respectively). In addition, the new specification also gave statistical meaning to North American coefficients for the 25% and 50% ownership threshold.

Capex Asia N. Amer. Europe Asia N. Amer. Europe Asia N. Amer. Europe

(1) (2) (3) (4) (5) (6) (7) (8) (9) family_owned 0.000 -0.006 -0.007*** 0.003* -0.021*** -0.007*** -0.002 -0.029*** -0.005 (0.001) (0.004) (0.002) (0.002) (0.005) (0.002) (0.004) (0.004) (0.003) crisis -0.010*** -0.012*** -0.011*** -0.010*** -0.012*** -0.011*** -0.010*** -0.012*** -0.011*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) family_crisis -0.004*** 0.002 0.004 -0.007*** 0.005 0.003 -0.008 0.008 0.003 (0.002) (0.005) (0.003) (0.002) (0.006) (0.003) (0.005) (0.005) (0.004)

Country level fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 31,160 13,361 9,716 31,160 13,361 9,716 31,160 13,361 9,716

Adjusted R2 _0.061 _0.008 _0.041 _0.061 _0.009 _0.040 _0.060 _0.008 _0.040

Mean pre-crisis capex 0.054 0.062 0.053 0.054 0.062 0.053 0.054 0.062 0.053

Tests on family_crisis across regions F-stat p-value F-stat p-value F-stat p-value

Joint (all three regions) 8.34 0.015 8.16 0.017 4.61 0.100

Asia vs N. America 1.23 0.267 4.11 0.043 4.34 0.037

Asia vs Europe 7.84 0.005 6.05 0.014 2.43 0.119

Europe vs N. America 0.17 0.684 0.19 0.664 0.55 0.459

Capex is the amount spent in capital expenditures calculated in percentage of total assets. Crisis period takes the value of one for years 2008 and 2009 and the value of zero for years 2006 and 2007. All regressions control for country level fixed effects. Robust standard errors reported in parenthesis. *, **, *** indicates that coeficients are significant at 10%, 5%, 1% respectively

Family threshold @ 10% Family threshold @ 25% Family threshold @ 50%

(25)

25

R&D

The results for R&D are not as strong as the ones for capex as, at the 10% ownership level, I only find significant results for the crisis variable. From Table 7 it is possible to note that the crisis showed that Asian firms increased the amount of share of their revenue committed to R&D by 0.3%, which is an 14% increase over the region’s average investment. Firms based in North American also increased their investments in 0.8% which accounts for an 8% increase over pre-crisis average. For European companies I find no statistical meaningful results. It is important to highlight that this increase in R&D was not caused by a reduction in revenue levels as I use the year 2006 as a reference year.

Being family owned does not seem to impact how firms invest in R&D as both

family_owned and family_crisis show no statistical meaning in any region. This is somewhat

surprising as it contrasts with the findings of Anderson et al (2012) that claims that families commit less money to R&D as they prefer the tangible value of physical assets to the risk of a research project. Further on the paper I explain some of the potential reasons for why I found no significance on these variables.

The tests to verify if the coefficients from family_crisis differ from each other are not statistically significant, providing the basis to reject hypothesis H1b (p-values of 0.474, 0.818 and 0.305 for 10%, 25% and 50% ownership levels respectively). An analysis on the individual comparisons between regions also show that none of the regions differ between each other.

Additionally, all other coefficients are statistically insignificant for all three different levels of ownership, 10%, 25% and 50%, which means I also reject hypothesis H2b. Robustness test in appendix (Table 10) confirms the rejection of both hypothesis using only the data from 2006 and 2009.

Leverage

In Table 8 I verify if family controlled firms have different policies regarding their leverage during the crisis relative to a sample of non-family controlled firms.

(26)

26

It is possible to note that crisis is statistically significant in all regions across all three different levels of ownership. The impact is negative for Asian firms (-0.004), meaning they reduced their debt, and positive for North American and European (0.011 and 0.006, respectively), which means firms in these regions slightly increase their debt. Although statistically relevant, the impact of such changes are only marginal to firms as the average pre-crisis leverage level were between 0.32 and 0.35.

Being family owned or not does have an impact on the decision of how much debt a firm take but the economic consequences of such results are minimal and the results do not hold under all different ownership levels I only find statistical significance in Asia and at the 10% threshold (-0.004) and in Europe (-0.011).

The tests to verify if the coefficients from family_crisis differ from each other are not statistically significant, providing the basis to reject hypothesis H1c (p-values of 0.990, 0.943 and 0.966). Additionally, all coefficients from the three regions are statistically insignificant for all three different levels of ownership, 10%, 25% and 50%, which means I also reject hypothesis H2c. Robustness test in appendix confirms the rejection of both hypothesis using only the data from 2006 and 2009.

R&D Asia N. Amer. Europe Asia N. Amer. Europe Asia N. Amer. Europe

(1) (2) (3) (4) (5) (6) (7) (8) (9) family_owned -0.001 0.007 -0.004 -0.004** -0.002 0.010 0.000 0.008 -0.009 (0.001) (0.015) (0.007) (0.002) (0.022) (0.010) (0.003) (0.044) (0.006) crisis 0.003*** 0.008** 0.004 0.003*** 0.007* 0.004 0.003*** 0.007** 0.003 (0.001) (0.004) (0.004) (0.001) (0.004) (0.004) (0.001) (0.004) (0.004) family_crisis 0.000 -0.024 -0.002 -0.002 -0.019 -0.003 -0.003 -0.063 0.005 (0.002) (0.020) (0.009) (0.002) (0.028) (0.013) (0.004) (0.045) (0.010)

Observations 12,960 5,245 3,271 12,960 5,245 3,271 12,960 5,245 3,271

Adjusted R2 _0.039 _0.007 _0.025 _0.039 _0.007 _0.025 _0.039 _0.007 _0.025

Asia vs N. America 1.47 0.225 0.39 0.532 1.76 0.185

Asia vs Europe 0.03 0.872 0.01 0.913 0.53 0.467

R&D captures the amount spent in research & development measured in percentage of total revenue. Crisis period takes the value of one for years 2008 and 2009 and the value of zero for years 2006 and 2007. All regressions control for country level fixed effects. Robust standard errors reported in parenthesis. *, **, *** indicates that coeficients are significant at 10%, 5%, 1% respectively

Table 7 - Effects of family ownership on firms' R&D investment decisions in periods of crisis

(27)

27

Leverage Asia N. Amer. Europe Asia N. Amer. Europe Asia N. Amer. Europe

(1) (2) (3) (4) (5) (6) (7) (8) (9) family_owned -0.004** -0.005 -0.011*** -0.008*** 0.012 -0.007 -0.005 0.028 -0.010* (0.002) (0.007) (0.004) (0.003) (0.011) (0.005) (0.006) (0.022) (0.006) crisis -0.004*** 0.011*** 0.006*** -0.004*** 0.011*** 0.006*** -0.004*** 0.011*** 0.005*** (0.001) (0.002) (0.002) (0.001) (0.002) (0.002) (0.001) (0.002) (0.002) family_crisis -0.002 -0.001 -0.002 -0.002 -0.003 0.000 0.003 0.003 0.006 (0.003) (0.009) (0.005) (0.004) (0.015) (0.006) (0.009) (0.028) (0.008)

Observations 31,160 13,361 9,716 31,160 13,361 9,716 31,160 13,361 9,716

Adjusted R2 _0.051 _0.012 _0.034 _0.051 _0.012 _0.033 _0.051 _0.012 _0.032

Asia vs N. America 0.01 0.941 0.00 0.960 0.00 0.997

Asia vs Europe 0.01 0.914 0.11 0.743 0.07 0.797

Table 8 - Effects of family ownership on firms' financial leverage in periods of crisis

Leverage captures the amount of debt used to finance a firms’ assets, measured in percentage of total capital (debt+assets). Crisis period takes the value of one for year 2009 and the value of zero for year 2006. All regressions control for country level fixed effects. Robust standard errors reported in parenthesis. *, **, *** indicates that coeficients are significant at 10%, 5%, 1% respectively

(28)

28

Discussion and Conclusion

Using a large sample of family and non-family companies from 30 countries in Asia, North America and Europe, I tried to identify differences in the behavior of family firms relative to other firms across regions during the 2008/2009 crisis. I show that there are regional differences in the way family firms made capital expenditures during the crisis and that this difference is robust and valid across different levels of ownership. These differences, however, were not found for the other two variables examined: research & development (R&D) investments and financial leverage.

These results have some potential limitations to their interpretation. Perhaps the most relevant in this case the omission of control variables. This means that by not controlling for other characteristics the coefficients of the regressions may lose their significance or even change direction. Similar research in the field of family business make use of different controls for their modelling such as whether the owner of the shares is the founder or his/her descendants, whether the family is involved in the firm through the board and/or through management, whether the decision of investing affected only particular industry (industry level fixed effects). Such information on the ownership is done mainly through unique databases and thus makes it harder to be included in this research.

Other factor to be considered is that families may have a preference to be in businesses and industries that require less capital in times of crisis. As pointed out by many papers (Anderson, Duru & Reeb, 2012) families are particularly risk averse and, as shown by Lins et al (2013), a reduction in investments can negatively impact returns, as they tend to prefer remaining relevant owners of smaller business than insignificant shareholders of a larger organization. However, even the evidence on the profitability of family firms compared to other types of organizations is still debatable. It is not yet clear whether a family run business over or underperform they peers. A meta-analysis by Wagner et al (2015) show family have an “economically weak, albeit statistically significant, superior performance compared to non-family firms”.

When looking at R&D and leverage, there could a number of reasons for not finding significance in both. First, the original data for R&D may not be 100% accurate as pointed out by the data provider. Some companies can disclose this information on the annual report

(29)

29

in two different sections (in the Income Statement section and also on the Notes section) and there are cases in which these figures differ. This could be caused for different definitions of R&D by the securities and tax authorities and the firms’ view on what is research and development. Second, I have not included in the regressions firms that reported zero investment in R&D. This means that I excluded firms that actually invested zero as well as firms that made investments but consciously decided not to disclose this number on their annual reports, adding noise to the analysis. Third, firms may have a passive behavior towards their target leverage, meaning that this measure would be a consequence of other decisions. Measuring debt levels in the short/medium term have this downside. As an example, in order to reduce debt, managers may decide to cut costs and/or investment for a few years instead of issuing more shares to raise cash and diluting current shareholders. Thus, costs or investment would be a better predictor of a firms view on the future. And, as shown, there is a regional difference in the way these firms execute their investments. Fourth, I have excluded from the dataset firms that may have had actually made a conscious choice of reducing/increasing leverage by excluding form the data companies that showed a large variation in assets, meaning I would only capture small changes and therefore, find little significance.

On the variable in which I did find significance, capex, I find that the results are somewhat in line with Lins, Volpin & Wagner (2013) in which family owned firms cut more investment then non-family owned ones and that there is no difference in firms’ leverage during the crisis. The reason for being so cautious about this comparison is that individually, the only significant coefficient (Asia) is showing the same signal and even similar impact but testing if those coefficients are equal across regions showed statistical difference between Asia and Europe. Although not explicitly saying the behavior of family is firms is the same across all regions, they imply it, by stating that the underperformance of family firms is “a global effect, consistently distributed around the world” (Lins et al, 2013) and that this is caused by reduction in investments. Even if this research does not test the effects on profitability, it raises the question whether the effect of reducing investments is truly global.

One interesting result found when jointly comparing regions is that, whenever there is a significant difference between family firms’ behavior relative to non-family firms, this

(30)

30

difference is driven by the conduct of Asian companies. Looking at the region-to-region results, I found no significant difference on Europe-North America comparisons on any of the variables. The statement made by the founder of Acer that European and North American management teams are different from the Taiwanese seems to hold for family firms, suggesting a similar behavior of “western culture” versus “oriental culture”. In the end, the discussion falls on how different aspects affect decision making in the business environment. During crisis periods, uncertainty avoidance, one of the cultural aspects pointed by Hofstede (1984) plays a significant role, as shown by Wagner (2015), there is a significant effect on companies’ performance with the interaction of family management and low uncertainty avoidance. In addition, each region share other unique cultural values that affects the way these companies do business.

The impacts of these results for the shareholding families are, to some extent, insignificant. Shareholder’s decisions should reflect their values and beliefs but knowing that they are more or less risk averse than a similar firm in a different region would, most likely, not change their decision making. It would simply reveal to them that they are, in fact, different.

These findings, however, have implications for the financial and management literature, especially for institutional investors (such as pension funds), that invest overseas. When investing in firms in a different region than their own, these investors should be aware of the existence of any significant shareholder and, in the case of a crisis, they should increase monitoring so firms do not pass on opportunities to that can impact firm value. Managing joint ventures with companies from different regions could also be a delicate situation in periods of crisis as some of the partners may choose to be more cautious while others prefer a more aggressive approach, increasing the possibility of internal conflicts. Lastly, external investors (with no influence on the board) could have more difficulty in predicting a company’s investment policy in periods of crisis if that same company has shareholders from different regions in board, which could lead to higher discounting by the market.

This research is one of the few that compares regional differences in the context of family firms and further work need to be done in family cross country/regional business decisions. It would also be relevant to investigate if these differences hold over time and if

(31)

31

the presence of family on the board and/or management have any relevant effect on the investment decision in periods of crisis.

(32)

32

Appendix

Capex Asia N. Amer. Europe Asia N. Amer. Europe Asia N. Amer. Europe

(1) (2) (3) (4) (5) (6) (7) (8) (9) family_owned 0.000 -0.007 -0.009*** 0.005 -0.024*** -0.010*** -0.003 -0.031*** -0.010** (0.002) (0.006) (0.003) (0.003) (0.006) (0.003) (0.006) (0.004) (0.004) crisis -0.017*** -0.022*** -0.019*** -0.017*** -0.022*** -0.019*** -0.018*** -0.022*** -0.019*** (0.001) (0.002) (0.002) (0.001) (0.002) (0.002) (0.001) (0.002) (0.002) family_crisis -0.006*** 0.006 0.006 -0.009*** 0.015** 0.006 -0.004 0.018*** 0.007 (0.002) (0.007) (0.004) (0.004) (0.008) (0.004) (0.008) (0.007) (0.005)

Observations 15,880 6,682 4,858 15,880 6,682 4,858 15,880 6,682 4,858

Adjusted R2 _0.085 _0.027 _0.046 _0.084 _0.028 _0.045 _0.084 _0.028 _0.045

Asia vs N. America 2.17 0.141 8.31 0.004 4.59 0.032

Asia vs Europe 7.37 0.007 7.52 0.006 1.43 0.232

Table 9 - Effects of family ownership on firms' capex investment decisions in periods of crisis (only years of 2006 and 2009)

Capex is the amount spent in capital expenditures calculated in percentage of total assets. Crisis period takes the value of one for year 2009 and the value of zero for year 2006. All regressions control for country level fixed effects. Robust standard errors reported in parenthesis. *, **, *** indicates that coeficients are significant at 10%, 5%, 1% respectively

R&D Asia N. Amer. Europe Asia N. Amer. Europe Asia N. Amer. Europe

(1) (2) (3) (4) (5) (6) (7) (8) (9) family_owned -0.001 0.013 -0.003 -0.003 0.002 0.011 -0.001 0.060 -0.010 (0.002) (0.022) (0.010) (0.002) (0.031) (0.014) (0.004) (0.080) (0.007) crisis 0.005*** 0.011** 0.009* 0.005*** 0.010* 0.009* 0.005*** 0.010* 0.008 (0.001) (0.005) (0.006) (0.001) (0.005) (0.005) (0.001) (0.005) (0.005) family_crisis 0.000 -0.038 -0.004 -0.003 -0.023 -0.005 -0.002 -0.117 0.007 (0.002) (0.026) (0.013) (0.003) (0.040) (0.018) (0.006) (0.080) (0.014)

Observations 6,499 2,626 1,619 6,499 2,626 1,619 6,499 2,626 1,619

Adjusted R2 _0.043 _0.006 _0.019 _0.043 _0.006 _0.020 _0.043 _0.006 _0.019

Asia vs N. America 2.04 0.154 0.26 0.612 2.04 0.153

Asia vs Europe 0.01 0.768 0.02 0.901 0.39 0.534

R&D captures the amount spent in research & development measured in percentage of total revenue. Crisis period takes the value of one for year 2009 and the value of zero for year 2006. All regressions control for country level fixed effects. Robust standard errors reported in parenthesis. *, **, *** indicates that coeficients are significant at 10%, 5%, 1% respectively

Table 10 - Effects of family ownership on firms' R&D investment decisions in periods of crisis (only years of 2006 and 2009)

(33)

33

Leverage Asia N. Amer. Europe Asia N. Amer. Europe Asia N. Amer. Europe

(1) (2) (3) (4) (5) (6) (7) (8) (9) family_owned -0.005 -0.007 -0.011** -0.009** 0.008 -0.009 -0.008 0.015 -0.014 (0.003) (0.009) (0.005) (0.004) (0.015) (0.007) (0.009) (0.026) (0.009) crisis -0.008*** 0.009*** 0.003 -0.008*** 0.009*** 0.002 -0.008*** 0.009*** 0.002 (0.002) (0.003) (0.003) (0.002) (0.003) (0.003) (0.002) (0.003) (0.003) family_crisis -0.002 0.001 -0.002 -0.003 0.010 0.002 0.008 0.027 0.012 (0.004) (0.013) (0.007) (0.006) (0.022) (0.009) (0.013) (0.035) (0.012)

Observations 15,580 6,682 4,858 15,580 6,682 4,858 15,580 6,682 4,858

Adjusted R2 _0.051 _0.012 _0.031 _0.051 _0.012 _0.029 _0.051 _0.012 _0.029

Asia vs N. America 0.05 0.826 0.30 0.587 0.27 0.605

Asia vs Europe 0.00 0.956 0.19 0.661 0.01 0.795

Leverage captures the amount of debt used to finance a firms’ assets, measured in percentage of total capital (debt+assets). Crisis period takes the value of one for years 2008 and 2009 and the value of zero for years 2006 and 2007. All regressions control for country level fixed effects. Robust standard errors reported in parenthesis. *, **, *** indicates that coeficients are significant at 10%, 5%, 1% respectively

Table 11 - Effects of family ownership on firms' financial leverage in periods of crisis (only years of 2006 and 2009)

(34)

34 Region Country Asia China India Indonesia Japan Korea, Republic of Malaysia Pakistan Philippines Singapore Sri lanka Taiwan Thailand Vietnam Europe Austria Belgium Cyprus Denmark Finland France Germany Greece Ireland Netherlands Norway Sweden Switzerland Turkey United Kingdom North America Canada