Corporate Strategy and Performance of Western-European Hospitality Firms

(1)

Corporate Strategy and Performance of

Western-European Hospitality Firms

June 2009

Abstract

This thesis examines the effect of corporate strategy on firm performance. This is tested in the large and growing Western-European hospitality industry for the period of 2003-2008. Using a sample of 51 firms, the results indicate that corporate strategy does explain a small but significant amount of variance in performance. Corporate strategy is characterised by three key elements; growth, liquidity and leverage. The results also show a small significant relation between growth potential (market-to-book of total assets) and performance. In addition, a size effect is observed; on average, larger firms perform better than small firms. Finally, in 2007 and 2008, the hospitality firms performed significantly worse than in 2003, which is probably the result of the credit crunch.

JEL classification: G12, C33, L83

Keywords: Hospitality, Hotels, Restaurants, Unbalanced Panel Data, Corporate Strategy,

Fixed Effects, Random Effects, Jensen‟s Alpha

Author Research Supervisor

Annelien Kool (1358251) Lammertjan Dam

Faculty of Economics and Business Faculty of Economics and Business University of Groningen University of Groningen

The Netherlands The Netherlands

(2)

1. Introduction

In 2008, the hospitality industry was a $2.1 trillion service sector globally, in which Europe reached a respective value of $0.7 trillion. The name of this industry is an umbrella term for a broad variety of service industries, including traveller‟s accommodations and food services, such as hotel firms, restaurant firms and recreation parks.

During the last two decades, the hospitality sector has experienced a growing interest by empiricists, due to higher growth than the overall economic growth worldwide; a compound annual growth rate (CAGR) of 5%1 versus a CAGR of 3.1%2 for the period 2004-2008. Additionally, the related sectors like the travel and leisure sector were expanding as consumers enjoyed greater available income. Due to its size and growth, the hospitality industry is an interesting industry to investigate. Financial institutions confirm this as they have set up special business units which focus on this industry in order to offer their products and services. Therefore, this industry will be the main subject of this thesis. In particular, this thesis investigates the effect of differences in corporate strategy on firm performance.

Essentially, most research on the hospitality industry focused on the United States. One of the reasons for this is that the United States has a large market for which there is widespread information available. Although the US hospitality industry is an interesting market to investigate (Harvey, 2007), Western-Europe is also an attractive market due to its great share in the global tourism market. Europe is seen as a top destination for travellers and attracted more than half of the world‟s travellers in 2007, which means 480 million tourists, an increase of 19 million compared to the previous year.

For financial institutions and investors the hospitality industry is an interesting market. They are interested in the effect of investments (asset growth) and the way it is financed on the performance of the hospitality firms, both from a commercial point of view and from a risk point of view (related to granting loans). Obviously, management of these firms is also interested in the effect of investments and the way it is financed, as they want to know how they are performing compared to the market (competitors). As such a better understanding of the key performance drivers is needed.

This thesis is in line with Chathoth and Olsen (2002). They investigated the effect of environment risk, corporate strategy and capital structure on firm performance of US restaurants. However, this thesis will only address corporate strategy and capital structure

1_Datamonitor 2

(4)

based on the variables such as growth, liquidity and leverage. This complies with the main interest of the above mentioned parties. In this thesis, choices related to corporate strategy and capital structure based on growth, liquidity and leverage are referred to with the term „Corporate Strategy‟. Taken as a whole, corporate strategy focuses on the type of investments the company should make, how to finance these investments, the level of cash flows required to operate the company in the short run and how to manage the risks to maximise long-term value of the firm (Ross et al., 1999).

Theoretically, it is expected that the corporate strategy has no influence on the stock performance of a company. For example, according to the Capital Asset Pricing Model (CAPM) developed by William Sharpe (1964) and John Lintner (1965), the only factor that affects expected return of a security, is its sensitivity to market risk, which is reflected fully in the beta of the firm. Furthermore, in line with the CAPM, Modigliani and Miller (1958) established a theory - the MM theorem - that states that it is expected that financial decisions are irrelevant in determining firm value; which means that the value of the firm is unaffected by how the firm is financed.

The most frequently mentioned criticism against these theories is the oversimplification of the assumptions about how markets work (Barton and Gordon, 1987). Although these simplifications are needed to develop convenient mathematical models in finance, the assumptions, conclusions or implications are often not consistent with the real world observations. The MM theorem and the CAPM assume, among others, that markets are perfect and efficient and that investors are rational (risk-averse), which is not fully comprehended in the real world. For example, hospitality firms, which aim to grow, either by takeovers of other firms or by autonomous expansion, are in practice limited by the availability of target firms, market potential and financing capacity (the willingness of financial institutions to grant loans). This is especially the case in an industry that is sensitive to the economic cycle and movements in the market, such as the hospitality industry. Thus, management of individual firms cannot always make the choice that would be most successful. Management has to deal with competition and wants to do at least as well as its competitors. Therefore, the purpose of this thesis is to provide insights to managers, investors and financial institutions, in the hospitality industry, specifically in terms of the emphasis they need to put on issues pertaining corporate strategy.

(5)

together, this thesis focuses on the relation between the corporate strategy and (risk-adjusted) performance in the Western-European hospitality industry over the period 2003 to 2008. Corporate strategy will be characterised by three elements; growth, liquidity and leverage. The sample consists of 51 Western-European hospitality firms. The results of this thesis show that to some extent, firm performance can be explained by the corporate strategy. The results also show that the market-to-book of asset (growth potential) and firm size are positively correlated with the risk-adjusted performance measure. Furthermore, in the final two years of the sample period, 2007 and 2008, it can be seen that firms performed significantly worse than they did in 2003. This is most likely caused by the world wide credit crunch.

(6)

2. Theoretical literature review

The widely recognized and employed theories of Modigliani and Miller (1958) and the CAPM, imply that the only factor that affects the rate of return of a security is the market (systematic) risk (beta). Still, good management and effective corporate strategy are considered to be necessary for any successful business (Vaish, 2007). Especially in a highly competitive industry, such as the hospitality industry (Kayaman and Arasli, 2007), the effectiveness of the strategy is a major determination of the business success of a firm and its long-term survival. This section will delve into the theories and backgrounds of corporate strategy and its elements.

2.1 Corporate strategy

Corporate strategy is a widely interpreted term. Simply stated, it refers to the decisions the investment managers must make to properly allocate resources to the future of the company (Olsen, 2004). For example, the choices that the managers can make include mergers and acquisitions, growth plans, selling assets, investment in new products/ services, determining how to finance and/or restructure the organisation.

As already mentioned, the corporate strategy is concerned with the type of investment the company will make, how to finance this investment, the level of cash flows required to operate the company in the short run and how to manage the risks to maximise the long-term value of the firm (Ross et al., 1999). This complete statement is based on four key elements, namely growth, liquidity, capital structure and risk. The element „risk‟ will be included in the performance measure, which will be explained in section 4. The theories and backgrounds about the other three elements and their effects for the company are elaborated below.

2.2 Growth

In most industries, growth is considered to be the key ingredient of success (Chathoth and Olsen, 2002; Broussard et al., 2005). Chathoth and Olsen (2007) imply that the strategy of hospitality firms has focused on growth, growth and more growth; growth is an imperative for every manager of a hospitality firm. They state that growth has always been assumed to add value to the company. Additionally, Borde (1998) emphasizes that growth must be managed well, while developing an internal structure that is capable of dealing with the growth while maintaining control of operations of the firm.

(7)

expected returns as a function of the mix of firm growth options and assets in place. As firms invest, the importance of growth options relative to the existing assets in place declines, reduces overall risk, and induces a negative link between investment and expected return. A low market-to-book value of assets indicates the presence of more assets-in-place, and thus, lowers investment options for growth. Other theoretical papers emphasize on the risk-based explanations that state that the relationship between abnormal returns and asset growth rates should disappear once proper risk adjustments have been made.

2.3 Liquidity strategy

(8)

2.4 Leverage

One of the strategic choices a manager must make, is how to finance the investments (Olsen, 2004). This choice is dependent on the capital structure of the firm, which involves two key elements, namely debt and equity. Generally, it is expected that financial decisions are irrelevant in determining firm value; the value of the firm is unaffected by how the firm is financed (Modigliani and Miller, 1958). The Modigliani and Miller (MM) theorem is widely recognized in modern corporate finance. The theorem provides condition under which the financial decisions of a firm do not affect its value. Therefore, it has four propositions. The first proposition identifies, under certain conditions, that the debt-equity ratio does not affect its market value. The second proposition establishes that leverage has no effect on its weighted average cost of capital. The third proposition suggests that firm market value is independent of its dividend policy. Finally, the fourth proposition establishes that equity-holders are indifferent about the financial policy of the firm. This theory only holds in the absence of taxes, bankruptcy costs, and asymmetric information, and in an efficient market.

In reality, the choice between debt and equity is more complicated. The goal of management is to maximise the market value of equity and the market value of debt, so that firm value can be maximised (Ross et al., 1999). The management of a firm must create the optimal financing mix that results in the minimisation of the costs that are related to holding debt and equity. The costs of debt are associated with interest payments and with „financial distress‟ (bankruptcy risks). For example, when a company is highly leveraged, the risk of bankruptcy is high, which consequently has a higher expected rate of return. The benefit of debt is that it has a tax advantage of interest deductibility. This benefit means that companies that have high corporate tax rates probably use more debt for tax incentives (as long as they make any taxable profit).

(9)

3. Empirical literature review

Whether corporate strategy really has a positive effect on performance of a hospitality firm was investigated by some researchers (i.e. Chathoth and Olsen, 2002; Lee and Jang, 2007). However, more papers study the effect of one or two elements of corporate strategy on performance (i.e. Cooper et al., 2006; Barton and Gordon, 1988; Kim et al., 1998). The term „corporate strategy‟ as stated by Ross et al, (1999), is covered by four elements; growth, liquidity, leverage and risk. The element „risk‟ will be explained in section 4. Below empirical evidence about the effect of corporate strategy and its elements on performance will be elaborated. Relevant literature for various industries, also for the hospitality industry, will be discussed.

3.1 Corporate strategy

Prior studies investigated the effect of corporate strategy on performance. In such studies, the notion of differences in „corporate strategy‟ is captured in a quantitative (financial) way (Chathoth, 2002; Chathoth and Olsen, 2007; Kim et al., 1998) or in a more qualitative way (Rumelt, 1974; Lee and Jang, 2007; Barton and Gordon, 1988; Bowman and Helfat, 2001). Below, various corporate strategies that are dealt with in literature will be explained in more detail. In addition, their relation with commonly used performance measure will be discussed.

(10)

measure (free cash flow per share). Chathoth and Olsen (2002) did not test the relationship between the construct corporate strategy, thus growth and liquidity together in one regression, and firm performance. They separated the growth variables from the liquidity variable and tested the relation with firm performance individually. This will be explained further in this section.

Besides quantitative (financial) measures, more qualitative measures have also been used. One such measure that is used quite often is Rumelt‟s (1974) typologies of diversification strategies. The reason for using such a measure is that it reflects the management‟s attitude and their approach towards risks (Bettis, 1983). Rumelt (1974) categorises the diversification strategies into four main categories; Single (firms are committed to a single business in a single industry); Dominant (firms that are diversified to some extent, but are still focusing on their revenues from a single business in a single industry); Related (firms operating in several industries with linked activities); Unrelated (firms operating in several industries whose activities are not related). These categorical measures are generally accepted and used by many other researchers (i.e. Lee and Jang, 2007; Barton and Gordon, 1988; Bowman and Helfat, 2001; Singh and Gu, 1994). However, it seems that there is no consensus on the effect of diversification on firm performance. For example, Lee and Jang (2007) and Singh and Gu (1994) concluded that a diversification strategy does not improve profitability. They investigated the effect of a diversification strategy on corporate financial performance and stability for publicly traded US hotel firms and US restaurants firms respectively. They employed accounting measures, market measures and risk-adjusted performance measures. In contrast, Rumelt (1982) found that related diversified firms have an advantage over unrelated diversified firms on the profitability of the firm. He analysed the link between these diversification strategies and the profitability (return on invested capital) of 273 large industrial corporations in the United States over the period 1949-1974.

Although market diversification is considered to be an important business strategy for increasing market share and profitability (Ayal and Zif, 1979), this thesis will not further explore this type of strategy. This thesis will delve into the corporate strategy elements; growth, liquidity and leverage and their effect on firm performance.

3.2 Growth strategy

(11)

individually. When they tested the relations between the growth variables (sales growth, asset growth, growth potential), that partly represent the corporate strategy construct, and firm performance (free cash flow per share and return on equity), they did not find significant relations between them. This outcome was in contrast with their expectations, which means that their hypotheses were rejected.

Because growth is always assumed to add value to a firm, Chathoth and Olsen (2007) thoroughly investigated this notion. They did a study on whether growth strategies (sales growth and growth potential) are actually sustainable performance-enhancing strategies; whether growth strategies always add value to the firm in the long-run. This study hypothesised that the higher the growth potential, the higher capital expenditures, which will lead to lower free cash flow per share, implying a negative relation between growth potential and firm performance. In addition, Chathoth and Olsen (2007) hypothesised a positive relation between sales growth and return on equity. This research used the same variables that represent performance and corporate strategy (except for asset growth) and the same sample for the same period as their previous paper (Chathoth and Olsen, 2002). Again, the results showed that growth strategy variables are not significantly related to firm performance.

(12)

costly-arbitrage-based explanations3, they find a negative correlation between the asset growth rates and stock returns. They used a sample with US nonfinancial firms for the period of 1968 to 2006.

Furthermore, Barton and Gordon (1988) imply a negative relation between growth potential and firm performance; firms that have a high growth potential, the ability to grow in the future, will have higher capital expenditures to fund future growth. This will lead to a lower free cash flow per share. The sample in their study includes 279 large US firms. At last, Prahalad and Hamel (1990) suggest that firms need to invest in competitive methods and core competencies today to improve overall future growth, which sequentially will lead to a decrease in free cash flows.

3.3 Liquidity strategy

Kim et al. (1998) study the decision by firms whether or not to invest in liquid or illiquid assets. They performed a regression analysis with a sample of 915 industrial firms during the period from 1975 to 1994. As hypothesised, the results of Kim et al. (1998) showed a positive relation between liquidity and cash flow measures, i.e. the free cash flow; when firms maintain a high liquidity, it will have a positive effect on their free cash flow per share (Chathoth and Olsen, 2002). They also found a positive relation between liquidity and the cost for external financing (transaction cost motive). In this case, firm size and growth potential (market-to-book ratio of total assets) are used as a proxy for external cost of finance.

Chathoth and Olsen (2007) hypothesised that corporate liquidity is more performance-enhancing than corporate growth strategies, which consequently should be considered in the decision-making process. Although they did not find significant results for the relation between growth strategies and the performance measures, they did find a positively significant relationship between liquidity and performance (free cash flow per share). This means that the higher the level of liquidity, the higher the free cash flow of the firm. Moreover, the relationship between liquidity and growth strategies is tested positively, which indicates that as the liquidity of a firm increases, its growth potential and sales growth also increases. Kim et al. (1998) confirm this relation. Put all together, when a firm holds a good liquidity position, it will grow in a better way than when a firm ignores its liquidity situation.

3_{The costly-arbitrage-explanation utilizes the standard arbitrage logic that in a frictionless world, if a security is}

(13)

3.4 Leverage

Dalbor et al. (2007) explored the impact of long-term debt on firm value in the US lodging industry for the years 1980 through 2005. After controlling for risk and size, they found a positive relationship; debt increases the risk of the firm, and that increases the return of the security. In contrast, Abor (2005), who based his study on Ghanaian listed firms, found a negative relation between long-term debt and shareholder value; a higher level of debt reduces the return on equity (ROE). Ruland and Zhou (2005) support this negative relationship.

Damodaran (1997) argues that an increased debt level will increase interest payments, which will lower net income available. Additionally, it will increase the likelihood of the bankruptcy of the firm, which may further increase the costs related to the financing of the strategy of the firm; too much debt can have a negative impact on shareholders‟ wealth. Chathoth and Olsen (2002) also found a negative relation between the level of debt and the return on equity. Companies that have a higher debt level in their capital structure as compared to companies that use a lower debt level, will have a negative influence on their return on equity (Hall and Weiss, 1967).

Leverage is frequently studied in combination with the agency theory. The agency theory suggests that the level of debt should increase firm value through a reduction of agency costs. In Jensen‟s view, leverage reduces the agency problems of overinvestment because interest payments reduce the free cash flow available for discretionary spending; managers have fewer opportunities to overinvest or else misuse excess cash. However, debt cannot only reduce agency problems, it can also increase agency problems. For example, according to Jensen and Meckling (1976), agency costs can arise from conflicts of interest between (would-be) debt holders and (existing) equity holders. These costs are caused by the option value related to the equity position in a levered firm.

3.5 Hypotheses

(14)

Chathoth and Olsen (2002) tested whether corporate strategy (defined by growth and liquidity), environment risk and capital structure collectively, have a significant effect on firm performance. Their main conclusion is that the constructs together explain a significant amount of variance in hospitality firm performance. In this thesis, the three constructs used by Chathoth and Olsen (2002) are all inside the umbrella term corporate strategy and in the risk-adjusted performance measure (Jensen‟s Alpha). Thus, considering the results of Chathoth and Olsen (2002), the following hypothesis is developed:

Hypothesis 1: Differences in corporate strategy explain differences in the hospitality firm performance

The alternative hypothesis is that corporate strategy is irrelevant. Secondly, there is no consensus whether corporate growth has an effect on free cash flow. Although it is assumed that growth will add value to the firm, most empirical research found a negative relation between growth and firm performance. Firms that pursue a high growth strategy will have a negative effect on free cash flow. A negative effect of free cash flow will lead to a negative rate of return. Given these results, hypotheses 2 and 3 are developed:

Hypothesis 2: There is a significantly negative relation between asset growth and firm performance

Hypothesis 3: There is a significantly negative relation between growth potential and firm performance

The alternative hypotheses are that asset growth and growth potential are unrelated to performance. Furthermore, most studies about the relation between liquidity and performance suggest that firms that pursue a high level of liquidity will have a positive effect on free cash flow. A positive effect on free cash flow will reflect in a positive rate of return. Thus, hypothesis 4 is as follows:

Hypothesis 4: There is a significantly positive relation between liquidity and firm performance

(15)

that pursue a high level of debt will have a negative effect on free cash flow. This negative effect will lead to a negative rate of return, which further leads to hypothesis 5:

Hypothesis 5: There is a significantly negative relation between leverage and firm performance

(16)

4. Data

In this section the data, which will be used for the empirical part of the study on the corporate strategy and its effect on firm performance, will be described. The first paragraph deals with the actual data set used in this paper. The second paragraph will elaborate the measurements of the dependent and independent variables. Finally, in the last paragraph, the descriptive statistics will be shown and described.

4.1 Data set

The sample in this study consists of 51 hospitality firms listed in Western- Europe for the 5-year period of 2003 through 2008. Firms considered to be „Hospitality firms‟ in this thesis are: traveller accommodations and food services, such as hotels, bars and restaurants. Using Amadeus, 63 hospitality firms from 10 different countries in Western-Europe were selected. The main source that was used to collect firm specific data is FactSet Research Systems, a database that contains financial and economic information for companies worldwide.

This thesis uses listed firms to ensure data availability, however, there are still some missing data, reducing the data set from 63 to 51 firms (see Table A1, Appendix). This final data set contains 28 „restaurants, pubs & breweries‟ and 23 „hotels, resorts & cruiselines‟ (classification by FactSet Research Systems). In this thesis these two sectors are called „restaurants‟ and „hotels‟. Table I shows the number of hospitality firms organised per country in which the firms are listed.

Table I

Hospitality firms organised per country where the firms are listed

Country Number of firms

(17)

The data set in this thesis contains an unbalanced panel4 of 51 hospitality firms in a 5 year period, for a total of 2,550 monthly observations. Not all firms have been listed for the full 5 years, as the individual time series do not have equal length. For example, FishWorks PLC have been publicly traded since 2005 and Les Hotels de Paris since 2001. By still adding those firms that have not been publicly traded for the last 5 years to the sample, it benefits of a much larger data set and thus more observations can be used; a larger amount of observations permits greater estimation power, so the coefficients can be estimated more precisely. When ignoring those firms that are not publicly traded for the whole 5 years, much information will be lost.

As can be seen in Section 3, research about the hospitality industry all dealt with a limited sample size due to the lack of consistent financial data available (i.e. Chathoth and Olsen, 2002). This could have led to biased coefficients. This thesis will not use a much larger data set, but will make use of panel data analysis. Panel data provides a higher number of observations, which can be used for regression analysis. It is statistically proven that with a larger sample of observations coefficients are more reliable and accurate. Furthermore, panel data analysis, in combination with corporate strategy, was never done before in hospitality research.

4.2 Measures of the independent and dependent variables

Firm performance is the dependent variable and the corporate strategy variables are the independent variables. Corporate strategy is represented by three key variables: Growth (Asset Growth and Growth Potential), Liquidity and Leverage. Firm managers decide where to invest in, how liquid the assets of the firm should be and how much they need to rely on debt to obtain maximum value of the firm. These three key variables are explained in the following sub paragraphs. Table III (page 20) provides an overview of how the independent and independent variables are measured.

4.2.1 Firm performance

Hospitality research has used various performance measures, measures in terms of accounting performance and in terms of market performance. Accounting performance can be reflected by different accounting determined indices, i.e. return on investment, return on equity and return on assets. Market performance is measured by market indicators such as stock returns or free cash flow per share. Market performance measures evaluate, among others, the present

4

(18)

and expected future earning flows of a firm, the timing risks of these flows and the dividend policy of the firm (Hitt and Ireland, 2006). In other words, the stock price of the firm reflects the long-term performance of the company on behalf of the stockholders.

The goal of the company is to maximise the value of the firm, or the wealth of the investor. For an investor the value of the firm depends on two important factors: return and risk. Gu (1993) says: “the wealth of the investor is maximised when future operational cash flows are maximised and the variability of operational cash flows is minimised”. Rational investors aim at the highest possible return at the given level of risk. Consequently, when realising the value of the firm, it depends on return-risk characteristics of the company (Kim and Gu, 2003). Three widely used risk-adjusted performance measures are the Sharpe index (Sharpe, 1966), the Treynor index (Treynor, 1965) and the Jensen‟s Alpha (Jensen, 1968). The Treynor and the Jensen‟s index assume stocks are priced according to the Capital Asset Pricing Model (CAPM). Additionally, these two indices have been used for individual securities as well as for portfolios, whereas the Sharpe ratio is only applicable for portfolios. The Sharpe and Treynor indices will be employed to generate a risk-adjusted performance ranking table, while Jensen‟s Alpha is used to measure the excess return relative to the benchmark (CAPM). In the light of these statements, this paper will make use of the risk-adjusted performance measure Jensen‟s Alpha.

Thus, to measure the dependent variable Performance Jensen‟s Alpha is used, which is also known as Jensen‟s Performance Index (Kim and Gu, 2003). In 1968, Michael C. Jensen introduced this performance measure, which is still widely used. Jensen (1967) based his measure on a study of 115 US equity funds for the period of 1945 to 1964. Jensen‟s Alpha is a risk- adjusted performance measure that determines the excess return of a stock over the required return of the stock as determined by the CAPM. The formula for Jensen‟s Alpha looks the following:

(Jensen‟s Alpha) i ri rf i rm rf (1)

i

r = return of the security i i = risk measure of security i (systematic/ market risk) f

r = risk free rate rm = market return

The risk-free rate (rf ) is measured by the yield of the 10 year Euro Area Government Bond

and the market return ( rm) is measured by taking the local stock exchange where the

(19)

taken. The CAPM return (rf i (rm rf))

5

is adjusted for the relative riskiness of the security; a riskier security will have a higher expected return than a less risky security. Within the theoretically framework of the CAPM, Jensen‟s Alpha ( i) should be zero, which means

that the stock would perform exactly as the market would expect based on the market risk. If the return of a stock is higher than the risk adjusted return of the CAPM, that stock is said to have "positive alpha" or "excess returns". A positive Alpha indicates a high level of return given the level of risk (systematic risk or market risk) on the security of the firm. A negative Alpha implies a poor performance as compared to the risk.

The beta ( ) is provided by Factset Research Systems, that calculated the historical beta by using a regression analysis of 52-weeks historical returns of the security of the firm relative to the returns of the stock exchange where the firm is listed ( r_m). Below the descriptive statistics of the beta are shown. Table II also shows the descriptive statistics of the returns of the securities.

Table II

Summary (descriptive) statistics for the variables of the Jensen‟s Alpha

Minimum Maximum Mean Standard Deviation

Median

Beta -0.734 1.751 0.402 0.399 0.298

Stock Returns (%) -76.588 266.373 0.631 13.362 0.000

Note: There are 2,250 monthly observations. The stock returns are monthly percentages.

If the beta is 1, it indicates that the price of the security of that particular firm will move with the market. A beta greater than 1 means that the security of the firm is more volatile (riskier) than the market and vice versa when the beta is less than 1. Table II shows a wide range for the beta, from -0.734 to 1.751. Then, together with comparing the mean and the median, one could see that the betas differ among the firms. Also, there are negative values, which means that these firms follow the market inversely; when the market goes up, the value of the firm will decrease and vice versa.

The maximum value of the Stock Returns of 266% comes from Cains Beer Company PLC (sector restaurants). This company completed a large business deal; a reverse takeover of a large English pub operator6. This deal led to a formation of Cains Beer Company PLC, and probably to the high stock return. The second highest value of the Stock Returns is 75%.

5_CAPM: ₍ ₎ f m i f i r r r r , Expected 0 6

(20)

4.2.2 Corporate strategy variables

The element Growth is captured by Asset Growth and Growth Potential. The variable Asset Growth will be operationalised by measuring month-to-month percentage change of the book value of total assets7 (Cooper, et al., 2008). Higher growth rates indicate a growth strategy, and a negative growth rate reflects asset reduction or a downsizing strategy (Richard, 2000). Growth Potential indicates the ability of the firm to grow in the future, which is obtained by using the ratio of market value of assets to book value of assets, also known as market-to-book ratio (Chen and Zhao, 2006; Kim et al., 1998; Chathoth and Olsen, 2002). A high growth potential can be created by handling the existing asset base as efficiently as possible, which will increase the market value of the assets and therefore the market value of the company (market value of the equity). Here, the quality of the management makes the difference.

To measure Liquidity a liquidity ratio is calculated. Liquidity ratios attempt to measure the ability of a company to pay off its current liabilities. This ratio shows the proportion of the total assets of the firm that are highly liquid. Then again, it reflects how well the firm manages its liquidity position on a period-to-period basis. The liquidity ratio will be measured by dividing the cash and short term investments to the book value of total assets (Chathoth and Olsen, 2002). Generally, the higher the value of the ratio, the more precautious the firm is, in which it handles unexpected increases of costs and in which it is prepared for possible profitable future investment opportunities. On the other hand, a very high liquidity ratio can be costly because of the missed opportunity of making profits by using the cash for positive NPV-projects; the liquid assets earn a low rate of return.

The last independent variable is the Leverage of the firm that is measured by dividing total debt, long-term plus current debt, by its book value of total assets (Opler and Titman, 1994; Chathoth and Olsen, 2002). This ratio indicates how much the company relies on debt to finance total assets. Firms are said to be „highly leveraged‟ when they have a high ratio. The lower this debt ratio is, the less risky the firm is, since excessive debt can lead to very heavy interest and principal payments. In contrast, when a firm totally relies on equity, it gives up the tax reduction effect of the interest payments.

4.2.3 Control variables

In this thesis, a variable is also used to control for factors that could affect the dependent variable firm Performance other than the independent variables mentioned above. The control

7_{The book value of total asset is taken, instead of the market value of total assets, due to the high correlation}

(21)

variable Firm Size is included, which is obtained by taking the natural logarithm of the market value of the assets (Chathoth and Olsen, 2007). Small firms may have a different effect than large firms in terms of the relationship between firm performance and the explanatory variables. According to Fama and French (1992), firm size can have additional explanatory power on stock returns over beta. Banz (1981) and Reinganum (1981) found that stock returns are substantially higher for small firms even after controlling for the effect of beta risk.

Table III

Variables, measures and descriptions

Variable Measure Description

Performance PERFOR _{Measured by using Jensen‟s Alfa:}

f m i f i i r r r r

Asset growth ASSETGR Measured by the month-to-month percentage change of the book value of total assets.

Growth Potential

GRPOTEN Obtained by dividing the market value of assets to book value of assets (market value of assets/book value of assets). Market value of assets is measured by adding the difference between the market value of equity and the book value of equity to the book value of assets (book value of assets + (market value of equity – book value of equity))

Liquidity LIQRAT Obtained by dividing the cash and short term investments to book value of total assets (the cash and short term investments / book value of total assets)

Leverage DEBTRAT Calculated by dividing total debt by the book value of assets (total debt/book value of total assets)

Firm Size SIZE Calculated by taking the natural logarithm (LN) of the market value of assets

4.3 Descriptive statistics

Table IV provides the summary (descriptive) statistics of all hospitality firms for both the independent variables and the dependent variable. The data set contains of 51 companies with 2,550 monthly observations.

Table IV

Summary (descriptive) statistics of the dependent and independent variables for all hospitality firms

Minimum Maximum Mean Standard Deviation Median Performance -0.771 2.627 -0.020 0.130 -0.028 Asset Growth -0.307 8.543 0.012 0.197 0.000 Growth Potential 0.395 36.880 1.599 1.764 1.195 Liquidity 0.000 0.664 0.081 0.124 0.036 Leverage 0.000 0.896 0.310 0.191 0.325 Firm Size 1.736 9.695 5.230 1.899 4.777

Note: The variables are measured on a monthly basis. Performance is the dependent variable and is measured by Jensen‟s Alpha, a risk-adjusted performance measure. Asset Growth is the month-to-month percentage change of the book value of total assets. Growth potential is calculated by market value of assets/book value of assets. Liquidity is measured by the cash and short term investments/book value of total assets. Leverage is provided by dividing total debt by the book value of total assets. Size is the natural logarithm (LN) of the market value of assets.

(22)

values come from two firms in the sub sector restaurants, namely Nordic Service Partners Holding and Domino’s Pizza. This means that compared to the other firms, that in some months the market values of assets of these two firms are much higher than their book values of assets. These higher market values of assets were caused by a higher share price in those particular months. The variable Asset Growth also shows a high maximum value (8.543) as compared to the mean value (0.012). This high value comes from the restaurant Individual Restaurant Co. PLC. In June 2007 this firm significantly increased its asset base with 8.5%. The maximum value of Performance (Jensen‟s Alpha) is from Cains Beer Company PLC, and is connected to the high maximum value of the Stock Return (see Table II). The second highest value of the performance measure is 1.238.

To check whether there is a correlation among the variables, a correlation matrix is provided (Table V). When the correlation coefficient is high (e.g. ±0.8) the two variables are strongly related to each other, which means that a change in one variable causes a change in the other. In contrast, when the correlation coefficient is low (e.g. ±0.1), it indicates that the relationship between the two variables is weak or does not exist. Table V shows that the correlation coefficient between Liquidity and Leverage is relatively high (-0.453). This coefficient is negative, which indicates that the higher the level of liquidity, the lower the level of leverage will be. Furthermore, there exists no high correlation among the variables.

Table V

Correlation Matrix of the dependent and independent variables for all hospitality firms

Performance Asset Growth

Growth Potential

Liquidity Leverage Firm Size Performance 1.000 Asset Growth 0.008 1.000 Growth Potential 0.003 0.004 1.000 Liquidity 0.020 0.019 0.266 1.000 Leverage 0.000 -0.025 -0.082 -0.453 1.000 Firm size 0.081 -0.015 0.078 -0.075 0.302 1.000

(23)

Table VI

Summary (descriptive) statistics of the dependent and independent variables for hotels and for restaurants

Sub sector Minimum Maximum Mean Standard Deviation

Median

Performance Hotels -0.378 1.238 -0.015 0.115 -0.027

Restaurants -0.771 2.627 -0.024 0.139 -0.029

Asset Growth Hotels -0.239 1.495 0.004 0.059 0.000

Restaurants -0.307 8.543 0.018 0.255 0.000

Growth Potential Hotels 0.395 3.702 1.195 0.619 1.051

Restaurants 0.610 36.880 1.894 2.212 1.332

Liquidity Hotels 0.000 0.664 0.083 0.153 0.025

Restaurants 0.000 0.597 0.080 0.098 0.040

Leverage Hotels 0.000 0.802 0.321 0.177 0.328

Restaurants 0.000 0.896 0.302 0.200 0.320

Firm Size Hotels 2.486 9.695 5.313 1.665 4.837

Restaurants 1.736 9.623 5.169 2.057 4.747

The most notable observation is the difference between the maximum values of the Growth Potential between the two sub sectors. The Growth Potential values of the restaurants show higher values in comparison with the Growth Potential values of the hotels As mentioned above, these high Growth Potential values are caused by the firms Nordic Service Partners Holding and Domino’s Pizza. Another interesting observation from the sub sector restaurants, is the high maximum value of Asset Growth (8.543). This higher value comes from the restaurant Individual Restaurant Co. PLC. Moreover, the range of Performance for the restaurants is wider than for the hotels. Again, this is the result of the maximum value of the Stock Returns (Table II), the return of Cains Beer Company PLC.

Tables VII and VIII show the correlation coefficients between the dependent and independent variables for both sub sectors.

Table VII

Correlation matrix of the dependent and independent variables for the hotels

Growth Potential

Liquidity Leverage Firm Size Performance 1.000 Asset Growth 0.031 1.000 Growth Potential 0.034 -0.032 1.000 Liquidity -0.000 0.022 0.552 1.000 Leverage -0.018 -0.024 -0.189 -0.569 1.000 Firm Size 0.094 -0.080 0.360 -0.043 0.039 1.000

Note: Performance is the dependent variable and is measured by Jensen‟s Alpha, a risk-adjusted performance measure. Asset

Growth is the month-to-month percentage change of the book value of total assets. Growth potential is calculated by market

value of assets/book value of assets. Liquidity is measured by the cash and short term investments/book value of total assets.

Leverage is provided by dividing total debt by the book value of total assets. Size is the natural logarithm (LN) of the market

(24)

Table VIII

Correlation matrix of the dependent and independent variables for the restaurants

Growth Potential

Liquidity Leverage Firm Size

Performance 1.000 Asset Growth 0.007 1.000 Growth Potential 0.008 -0.002 1.000 Liquidity 0.040 0.028 0.293 1.000 Leverage 0.007 -0.027 -0.060 -0.376 1.000 Firm Size 0.094 -0.006 0.050 -0.114 0.437 1.000

Note: Performance is the dependent variable and is measured by Jensen‟s Alpha, a risk-adjusted performance measure. Asset

Growth is the month-to-month percentage change of the book value of total assets. Growth potential is calculated by market

value of assets/book value of assets. Liquidity is measured by the cash and short term investments/book value of total assets.

Leverage is provided by dividing total debt by the book value of total assets. Size is the natural logarithm (LN) of the market

value of assets.

It is noticeable that the correlation between Liquidity and Growth Potential has a higher coefficient for the hotels (0.552) than the correlation coefficient for the restaurants (0.293). This may mean that for the hotels Liquidity is more important for Growth Potential than it is for the restaurants. Moreover, there is a great difference in the correlation coefficients of Firm Size and Leverage for the hotels and the restaurants (0.039 vs. 0.437). This indicates that larger restaurants have a higher debt, and that this is not the case for hotels.

In the data set, there are 24 firms (with 1266 monthly observations) that are listed in the UK and 27 firms (with 1284 monthly observations) that are listed somewhere else in Western-Europe (see Table I). Table A2 (Appendix) shows descriptive statistics of these two groups, U.K. firms versus non-U.K. firms. It shows no significant differences between the two groups, which probably means that U.K. firms do not outperform non-U.K. firms or the other way around.

To check whether or not one year did much better or worse than another year, the summary statistics are given per year for all variables (see Table A3, appendix). An observation of Table A3 is that in the years 2006 and 2007 the standard deviation values of Asset Growth are higher compared to the standard deviation values of the other years. These higher values are caused by some maximum values that are extraordinarily high. There are a few extremely high Asset Growth values, which means that some firms increased their asset base greatly. Furthermore, there are no significant deviations between the 5 years.

(25)

(26)

5. Methodology

As already mentioned above, this study uses unbalanced panel data. Panel data have both a cross-sectional and a time series dimension; they have an individual dimension (i) and a time dimension (t) (Xit). The individual dimensions are 51 companies. There are several types of

panel data analytic models; the Pooled Model, the Fixed Effects Model, and the Random Effects Model (Baltagi, 2005). This section will explain these models. The regression formulas will be represented with the explanatory variables (Table III) used in this thesis.

First, all data are assumed to be normal time-series data; the data is pooled as if there were only a cross-sectional dimension (no differences in time) or only a time-series dimension (cross-sectional units are homogeneous). The fact that the data have two dimensions is ignored and is assumed to have only one dimension. In this case, the Pooled Model is used. The Pooled Model assumes that any company-date can be compared, and that all companies have the same relation with the independent variables, Asset Growth, Growth Potential etc. This model has constant coefficients; it estimates a single best-fitting regression line for the entire model. For the Pooled Model, the regression line looks as follows:

it it it it it it it SIZE DEBTRAT LIQRAT GRPOTEN ASSETGR PERFOR 5 4 3 2 1 (2)

This is just an ordinary least squares (OLS) regression model. This model, in combination with panel data, can give seriously biased estimates of the slope coefficients. Wrong conclusions can be drawn, because the explanatory variables can differ among the companies due to for example the sector the company belongs to (hotels or restaurants). The explanatory variables can also vary significantly between time periods, which affect all companies alike; the companies can do much better in one year than they do in another year. A reason for this variation can be the economic situation of a particular year. For example, after the „credit crunch‟ began at the end of 2007, the hotel sector as a whole started to see a decline in investment, development and demand (Kiessling, Balekjian and Oehmichen, 2009). These time differences, that may have an effect on the dependent variable (PERFORit), are called

time-specific (year) effects, and the company differences that may have an effect are called individual effects. The Pooled Model (such as formula 2) does not control for individual or time-specific effects; it just provides a single best-fitting regression line for the entire model.

(27)

variables must be created for each year (2003-2008). Consequently, the Pooled Model with time dummies would look like this:

it it it it it it it it it W W SIZE DEBTRAT LIQRAT GRPOTEN ASSETGR PERFOR ) 6 ( 6 ) 1 ( 1 5 4 3 2 1 ... (3)

Where the time dummy variables are called W(T) with T= 1,…, t, with t= 6 (years).

In this thesis, there may be unobserved heterogeneity across companies, which means that the Pooled Model will provide biased coefficient estimates. There are two main individual effects model that control for the heterogeneity across companies, called the Fixed Effects Model and the Random Effects Model. One or both of these individual effects models will be used, next to the Pooled Model. This will be decided based on the Hausman Specification test (Hausman, 1978), which will be explained further in this section. Table IX shows the characteristics of these two main individual effects models.

Table IX

The Fixed Effects Model and the Random Effects Model

Fixed Effects Model Random Effects Model

Simple regression it it i it X Y ( ) Y_it X_it ( _it _i)

Intercepts Varies across companies and/ or time-periods

Constant

Error variances Constant Varies across companies and/or time periods

Slopes Constant Constant

Estimation OLS (within effects, between effects)

GLS

Hypothesis test F-statistic Wald χ2 -statistic

The Fixed Effects Model, also called “within” estimator, is used when one wants to control for omitted variables that differ across individuals but are constant over time. It allows for each individual to have different regression lines. A Fixed Effects Model, for a simple regression, is written as follows: it it i it X Y with i i (4)

The difference with the Pooled Model (formula 2), is the _i instead of the constant . This

i varies across the individuals; it refers to the individual specific effect. The it is the

idiosyncratic risk, that captures the impact of unobserved variables which vary between companies and over time. The individual effects might lead to multiple best fitting regression lines with different starting points, which means different intercepts with the y-axis (different

(28)

To control the individual effects, the Fixed Effects Model uses dummy variables8. In this way the model examines the unobserved differences within the company. A dummy variable must be created for each company and must be included in a standard OLS regression. The (company) dummy variables are called D(j) with j = 1,…, N, with N= 51 (companies). As already mentioned, there may also be time-specific (year) fixed effects within the data set (maybe due to the „credit crunch‟); the independent variables vary over time periods, between the years. Just as within the Pooled Model, this can solved by creating dummy variables for each year (2003–2008) and subsequently include them into the regression formula, next to the company dummy variables. Then the fixed effects multiple regression formula with time dummy variables (W(T)) looks as follows:

it it it it it it it it t it it it W W SIZE DEBTRAT LIQRAT GRPOTEN ASSETGR D D D PERFOR ) 6 ( 6 ) 1 ( 1 5 4 3 2 1 51 ) 51 ( 51 ) 2 ( 2 ) 1 ( 1 ... ... (5)

This model will provide multiple regression lines with constant slopes. Furthermore, this is a common multiple regression model, only with a large amount of explanatory variables. This large amount of explanatory variables (N + T + k regular explanatory variables) eats up degrees of freedom; each extra dummy variable removes one degree of freedom from the model. Consequently, this causes an increase of standard errors, which reduces the efficiency of the model. However, this model controls for correlation between explanatory variables and unobserved company specific effects; the Fixed Effects Model is consistent, no matter if there is correlation.

In order to perform the test for the inclusion of time dummies in the Fixed Effects Model, a test must be applied whether or not the time dummy variables are jointly significant. The null hypothesis of this test assumes that the time dummies are not jointly significant, which means that the time dummies should not be included. Therefore, when the null hypothesis is rejected (p-value smaller than 0.10), the Fixed Effects Model should include time effects.

Instead of thinking of each company as having its own regression line (fixed effects), the Random Effects Model believes that each starting point is the result of a random deviation from some mean intercept; the individual effect is a random variable. The estimation of the coefficients can be done via a generalised least squares (GLS) regression. Moreover, it may

8_{A dummy variable is either a 1 or 0, it is a way of turning qualitative explanatory variables into quantitative}

(29)

also be possible that the inclusion of time dummies allows the use of a Random Effects Model in the individual effects. Again, the Random Effects Model should include the time dummy variables ( (T)

W ). The regression for the Random Effects Model with time dummy variables is presented: it it it it it it it it it W W SIZE DEBTRAT LIQRAT GRPOTEN ASSETGR PERFOR ) 6 ( 6 ) 1 ( 1 5 4 3 2 1 ... (6)

Where it i it. This model looks the same as a pooled regression (formula 2), only the

error term is divided into two parts; the intercept that the individuals share ( i), which is

constant over time, and a unique random component ( _it). It is assumed that _it and _i is uncorrelated with one another. Note that there is only one intercept with the y-axis ( ), and not many extra explanatory variables. In contrast to the Fixed Effects Model, the degrees of freedom are not eaten up by every extra dummy variable included; the Random Effects Model is more efficient. However, it is absolutely fundamental that the error term, the individual effects, is not correlated with the independent variables. If there is such correlation, then the Random Effects Model is not consistent.

To determine which of the individual effects models (Fixed or Random) is appropriate to use, the Hausman Specification test (Hausman, 1978) is performed. This test tests whether the individuals are correlated with the explanatory variables. The null hypothesis of the Hausman test is: “There is no correlation between the individual (and/or time-specific) effects and the explanatory variables”. The Fixed Effects Model controls for the correlation and the Random Effects Model assumes there is no correlation. When there is no correlation, both models are consistent and should give approximately the same results. However, the Random Effects Model is said to be more efficient than the Fixed Effects Model9, and is thus the most appropriate model to use. When there is such a correlation, the Random Effect Model is inconsistent and then the Fixed Effect Model is the right model to use.

When the Hausman test implies that the Random Effects Model is most appropriate to use, then the Breush and Pagan Langrange Multiplier (BP-LM) test for random effects must be done. This test indicates if there are really random effects and thus indicates whether or not the Random Effects Model should be used. If there are no random effects, then the Pooled

9_{The fixed effects model has a large number of explanatory variables, which eat up degrees of freedom and may}

(30)

(31)

6. Results

The main objective of this study is to test if the dependent variable Firm Performance is affected by the independent variables Asset Growth, Growth Potential, Liquidity and Leverage. First, this relationship is measured by the use of the Pooled Model. However, as already mentioned, this model can give seriously biased estimates. That is why an individual effects model, the Fixed Effects Model or the Random Effects Model, must be used. To test which of the two individual effects models should be used, the Hausman test is performed. Below, all three models will be shown and explained, for both the complete data set and the restaurants and hotels separately.

As mentioned before, there are a few outliers within the data set (of 2,550 observations). These 14 extreme observations are dropped from the data set. Although in the main text the results are shown from the regressions without the outliers, the results with the outliers are shown in the appendix (Table A4 and Table A5). There are two notable differences between the results of the two regressions, namely the significance of the explanatory variable Growth Potential and the different sign of the coefficient of Leverage. These two differences will further be explained in this section. Besides these two differences between the results, the regressions do not differ much from each other, and will thus not be further delved into.

(32)

Table X

The effect of corporate strategy variables on firm performance using the Pooled Model (OLS), including time dummies (without outliers)

Dependent Firm

Performance

(Jensen‟s Alpha)

Complete data set Restaurants Hotels

Pooled Model (OLS) Pooled Model (OLS) Pooled Model (OLS) Variable Coefficient (t-value) Coefficient (t-value) Coefficient (t-value) Asset Growth 0.034 (1.23) 0.021 (0.66) 0.066 (1.09) Growth Potential 0.007*** (3.06) 0.010*** (3.99) 0.004 (0.51) Liquidity 0.016 (0.75) 0.069* (1.94) -0.019 (-0.52) Leverage 0.007 (0.49) 0.016 (0.87) -0.016 (-0.66) Size 0.006*** (4.38) 0.006*** (3.82) 0.004* (1.85) Year 2004 -0.008 (-0.73) -0.008 (-0.58) -0.011 (-0.59) Year 2005 -0.002 (-0.16) -0.008 (-0.58) 0.002 (0.11) Year 2006 0.008 (0.76) 0.002 (0.18) 0.008 (0.48) Year 2007 -0.022** (-2.14) -0.043*** (-3.33) -0.002 (-0.13) Year 2008 -0.047*** (-4.20) -0.073*** (-5.17) -0.019 (-1.03) Constant -0.054*** (-4.63) -0.066*** (-4.77) -0.035 (-1.54) R-Squared 0.036 0.073 0.0125 Adjusted R-Squared 0.033 0.067 0.003 Statistical Test F(10, 2525)= 9.52*** F(10, 1450)= 11.44*** F(10, 1064)= 1.35 Number of Observations 2536 1461 1075

Note: Year 2004 to Year 2008 are dummy variables. Year 2003 is dropped due to collinearity. The dependent variable

Jensen’s Alpha is a risk-adjusted performance measure. Asset Growth is the month-to-month percentage change of the book

value of total assets. Growth Potential is calculated by market value of assets/book value of assets. Liquidity is measured by the cash and short term investments/book value of total assets. Leverage is provided by dividing total debt by the book value of total assets. Size is the natural logarithm (LN) of the market value of assets. The Hausman Test is a test for fixed over random effects. *** indicates significance at a 1% level, ** at a 5% level, * at a 10% level.

(33)

data set and the restaurants, the Pooled Model is a good model; the firm Performance is explained well by the explanatory variables. This is not the case for the hotels, where the F-statistic is not significant and adjusted R-squared is below 1%.

Nevertheless, it is suspected that there are individual effects in the data set, which may generate different intercepts across companies or time periods. There are two main individual effects models, the Fixed Effects Model and the Random Effects Model. To decide which of these two models should be used, the Hausman test is done. Table XI presents the results of both individual effects models, for as well the complete data set as for restaurants and hotels separately. Also, the results of the Hausman test and the Breush and Pagan Langrange Multiplier (BP-LM) test are shown in the last two rows of Table XI.

For the complete data set, and restaurants and hotels separately, Table XI shows no significant (χ2-)values for the Hausman test. This implies that there is no correlation between the individual effects and the explanatory variables and thus that the null hypothesis10 cannot be rejected. Therefore, for every sample, the Random Effects Model is most efficient and should be used.

Because the Hausman test implies that the Random Effects Model should be used, also the BP-LM test for random effects must be performed. For the complete data set, the BP-LM value is significant. This means that there exist random effects, and thus the Random Effects Model should still be used. For both sub sectors the BP-LM values are not significant. This indicates that there are no random effects, and that means that the Pooled Model should be used, instead of the Random Effects Model. Consequently, for the sub sectors, the conclusions will be drawn, based on the Pooled Model.

Furthermore, for all three samples, the results of the test, whether or not the time dummy variables are jointly significant, showed significant F-statistics (Table A6). This means that the null hypothesis11 must be rejected and that it is justified to include the dummies into the regression formulas. Below, first, the results for the complete data set are elaborated. After that, the results of the hotels and restaurants will be described separately.

10_H

0: There is no correlation between the individual (and/or time-specific) effects and the explanatory variables. 11

Corporate Strategy and Performance of Western-European Hospitality Firms