Capital re-allocation among business units and firm performance

(1)

Capital Re-Allocation

Among Business Units and

Firm Performance

Master Thesis

MSc. Business Economics

Finance Track

Amsterdam Business School

Faculty of Economics and Business

Universiteit van Amsterdam

Academic year 2013-2014

Supervisor: Dr. Ilko Naaborg

Student: Federico Iudica (10425977)

(2)

Capital Re-Allocation Among Business

Units and Firm Performance

FEDERICO IUDICA

University of Amsterdam

MSc. Thesis in Business Economics

Abstract

For years conglomerate discount has been a central topic in financial literature. The lower valuations of diversified companies relative to an adjusted portfolio of stand-alones have been attributed to two internal capital market inefficiencies: cross-subsidization and power struggles. Some argued that those problems could be solved with a rigid allocation of budget to each segment. Constructing two measures of capital allocation activity this thesis study whether flexibility or rigidity have had any impact on subsequent performance. I found that rigid firms outperformed their active counterparts. This finding is robust with different measures of allocation activity and performance. I also investigate if this changes during financial crisis when the relative importance of internal capital is greater. Although a rigid budget is important during normal times, flexibility yields better results in downturns. Finally, I test for CEO characteristics’ impact on performance through better management of the internal resource re-allocation process. Results suggest that younger CEOs are on average better at managing the internal capital market.

Acknowledgements

I would like to thank my supervisor Ilko Naaborg for his guidance and helpful comments and also my parents for the support during this second master program at the UvA.

(3)

1. Introduction

The search for the correct allocation of capital has been a key theme of economic research since its inception. Specifically, capital allocation across industries has received an immense amount of attention. Its study has implication for society from both a macro and a micro point of view. It is relevant for companies, governments, regulators and financial institutions. Too often though, it is thought that allocation among sectors is only done through the interaction of these players in the external capital markets. In the external markets, funds are allocated simply through portfolio exposure. Namely, the purchase or sale of equities and fixed income securities of firms in different industries. In the last twenty years, equity issuance has approximately stood at 85 billions of USD per year, and corporate debt issuance at 535 billions1_.

There is another way to allocate capital among sectors: the internal capital market. In this market, companies that are active in more than one business shift funds across their own business units. Multi-business companies have shifted a total of 640 billions per year since 1992. The annual capital budget is the instrument they use to do so. However, their ability and effectiveness at channeling these resources has long been debated. The existence of at least two-layers of agency conflicts and the private benefits of empire building undermine the efficiency of the budget process. It is well documented that CEOs want to run larger companies (Shleifer and Vishny (1989)). Nonetheless, larger companies are made up of different businesses and necessitate a monitoring effort that a headquarters is often not able to make, as in Scharfstein and Stein (2002) and Rajan, Servaes and Zingales (2000).

Monitoring is made necessary by the existence of a second layer of agency problems. While the first layer is the conflict between shareholders and top management, the second is the conflict between top management and units’ executives competing for more resources (Scharfstein (2000) and Ozbas (2005)). Both types of inefficiencies are at the roots of what was named the diversification discount (Lang and Stulz (1994) and Berger and Ofek (1995)).

Some researchers, such as Ayal and Rothberg (1986), Ozbas (2005) and Marino and Matsusaka (2005), have put forward some proposals to address these issues

1_{McKinsey Global Institute (2012)}

2_{Effort toward activities that do not contribute to organizational objectives}

(5)

from a practical point of view. They suggested that a CEO could minimize the rent-seeking behaviors of units’ executives by designing a rigid budget structure to avoid the fight for resources. Furthermore, Scharfstein and Stein (2002) argue that socialism and cross subsidization from better to worse performing divisions could be weakened with a less flexible capital budget process. This thesis investigates on several features of the internal capital market through an examination of the yearly changes in capital budget. Specifically, I examine multi-business companies, operating in a range of different industries. I explore whether they perform better when they allocate funds across segments consistently year after year following the same pattern or when they deviate to adjust resources based on opportunities. In sum, I wonder which strategy, rigidity or continuous adaptation, will yield the better results in the medium and long term?

To answer this question I employ two simple measures of resource re-allocation intensity. One comes from the observation of practitioners in the consulting business, while the other is an adjusted version of that in Billet and Mauer (2003). I start by showing that rigidity is prevalent in most companies and that slow capital flows’ variations across segments are better than large ones. The difference in performance (Total Return to Shareholders) between very active resource reallocation and those at the receiving end of a constant slice of the budget is around 2% per year in favor of the latter, for at least five years. This difference seems to be robust to the use of different measures of allocation and of performance. Hence, it seems that it is better to make a good plan and then stick to it, rather than changing strategy more often and be subject to the lobbying of segments’ management.

The importance of internal capital is relative to the availability of external funding (Huang et al (2013)). As companies have to cut costs and concentrate their efforts on their sources of value, crises can represent a significant boost to a reshuffling of resources. Hence, my second research question is to verify the course of resource allocation patterns during the two most recent downturns (dot-com bust and housing bubble). Did those who changed more actively experienced better returns and are those different than those earned over growth years? The answer is yes. Although rigidity has a key role in curbing agency costs in normal economic times, flexibility has a superior role during a crisis. The performance of the conglomerates that showed more flexibility outperformed their rigid

(6)

counterparts by as much as 5.63% during the years following the crises. An investor who observed the reshuffling of resources of a corporation, purchased its equity and held it for the next 5 years, would now enjoy a 29% premium over an investor who did the same for the average passive company.

In the labor market for executives there is a link between current performance and future pay (Holmstrom (1999)). However it is hard to distinguish what share of the current performance is due to skills and what is due to external factors (Bertrand and Mullainathan (2001)). Since a CEO’s abilities are revealed slowly over time a younger CEO could have more interest to obtain a better performance. Previous literature has been inconclusive on the connection between age and performance, because of too many firm-specific and sector-specific factors. Does the relation between age of the Chief Executive Officer and conglomerates’ performance happen through better skills in managing the internal allocation? I find that the younger managers are on average better at managing the internal capital market. The difference with a manager at the top quartile of the age distribution is a budget rigidity of about 10% for both measures.

Finally, I check whether a similar reasoning can be followed for CEO tenure. Based on the findings of (Hermalin and Weisbach (1991) that a long mandate reduces operating performance. The idea here is that resource reallocation is a cornerstone of corporate strategy and a CEO may become more dynamic at the beginning of his mandate and less willing to reconsider its strategy as the years go by. The results do not seem to go in this direction. CEO tenure does not appear to affect performance in the short, medium and long term.

This thesis finds its origins in the groundbreaking papers of Lang and Stultz (1994) and Berger and Ofek (1995). It is based on the strands of theoretical literature on internal capital market inefficiencies (Scharfstein (2000), Rajan Servaes and Zingales (2000) and Ozbas (2005)); and contributes to the works examining the functioning of internal capital markets (Stein (1997), Whited (2001), Maksimovic and Phillips (2002), Billett and Mauer (2003), Villalonga (2004) Guedj and Scharfstein (2007) and Ozbas and Scharfstein (2010)). Specifically, I examine in greater detail the solutions to inefficiency caused by agency problems proposed by Harris and Raviv (1996), Ozbas (2005) and Marino and Matsusaka (2005). The finding that stock prices do not completely incorporate the benefits of internal capital allocation, this paper relates to research on the inefficiencies of external capital markets Daniel and Titman

(7)

(2006) and Hoberg and Phillips (2008). Finally, by examining the impact of CEO career concerns on firm performance this thesis relates to the literature examining characteristics of successful CEOs (Malmendier and Tate (2005), Simsek (2007) and (Serfling, 2012).).

The rest of the paper is organized as follows: section 1 is a review of the relevant literature, section 2 and section 3 cover methodology and data, in section 4 I present the results of my analysis and section 5 concludes and elaborates on the limitations of this study.

2. Literature Review

2.1 Bright and Dark Sides of Internal Capital Markets

Capital can be allocated across industries when investors increase or decrease their exposure among sectors by buying and selling equity and debt. Equity, fixed income and related products make up the external capital markets. However, capital can also be allocated through the internal capital market. That is, through the transfer of resources across segments of the same conglomerate as long as they operate in diverse industries. Since the seminal contributions of Lang and Stulz (1994) and of Berger and Ofek (1995) a large body of empirical research has confirmed the existence of a diversification discount. They found that conglomerates trade at a 15% discount over an adjusted portfolio of stand-alone firms. It was later found that this hypothesis held over time and geographical areas.

In the last twenty years, researchers and practitioners in strategy and corporate finance have been treating the finding that diversification reduces value as an obvious fact. Many of the channels that make this possible are subjects of debate. Nonetheless, a broad agreement is reached that the main cause is the existence of inefficiencies in the internal capital market of a company. Those are usually referred to as the dark side of internal capital markets (Scharfstein, 2000). But internal markets do not have to function poorly and in certain conditions they allow conglomerates to deliver superior performances relative to stand-alone counterparts. This positive aspect has been named the bright side of internal capital markets. Those two aspects have been thoroughly examined by theoretical and empirical papers.

(8)

On the bright side Stein (1997) and Choe et al (2009) are particularly important theoretical contributions. In the model proposed by Stein (1997) the self-interest of business units’ executives is able to bring substantial benefits to a conglomerate firm. A CEO can harness managers’ efforts to compete for private benefits towards maximization of shareholder value by engaging in a winner-picking strategy. Since the level of external funds for a firm is constrained, executives compete for scarce resources, a good CEO, is argued, could create value simply by transferring resources from the worse to the best performing projects (winner-picking). In this framework, the less projects the CEO monitors, the better results his supervising activity will yield. Choe et al (2009) develop a model where diversified firms tend to have a better resource allocation than stand-alone firms. The idea is that an internal capital market can break the budget constraint for a single division. This ultimately enlarges the set of feasible investments and allows for pooled resources to be redeployed across units more efficiently.

With regard to empirical evidence, data availability has always been a concern. Nonetheless, there are many significant studies. Maksimovic and Phillips (2002) use plant-level data to study whether the productivity patterns of conglomerates are consistent with optimal internal resource allocation. They find that multi-business companies invest in the sectors in which they have a comparative advantage with respect to stand-alone firms. The stock market valuations of a diversified company seem to be inferior to that of a pool of analogous single segment firms, this is a result of lower productivity in smaller divisions. In their neoclassical setting, lower productivity firms can exist in equilibrium thanks to decreasing returns to scale. This result is coherent not only with optimal resource allocation, but also with the corporate discount hypothesis.

Guedj and Scharfstein (2007) study their investment behavior of pharmaceutical firms and their relative performance. They divide their sample between single-project and multi- single-project firms. In the first group, managers are excessively reluctant to abandon marginally uneconomic drugs during the first stage of clinical trials. A possible reason is that they are hesitant to return money to stockholders and lose the benefits generated by running projects. In the second group, managers do not benefit from the success of one particular drug. Hence, they will take trials forward only for those drugs with the highest value for the business. This is exactly the winner-picking behavior described above.

(9)

Villalonga (2004) shows that diversified firms do not trade at a discount because of diversification itself. In fact, those firms do not diversify at random but their choice is dependent on their characteristics, on those of the market they operate and in which they chose to expand. Thus, conglomerates and single segment companies do not share the same market opportunities. Matching them only in terms of size and industry introduced a bias. Hence, she matches them using measures of their propensity to diversification as well. Results show how companies that diversify already trade at a discount before their decision to expand in different sectors. Afterwards, they seem to enjoy a premium. In the author’s view, the existence of a diversification premium indicates how the benefits of internal capital markets outweigh the costs.

On the dark side Scharfstein (2000), Rajan Servaes and Zingales (2000) and Ozbas (2005) are the most renowned theoretical contributions. Scharfstein (2000) models the organizational process that leads to the allocation decision by the CEO in two levels. Firstly, the CEO needs to obtain funding from outside investors. Secondly, the units’ executives have to obtain funding from the CEO. This leads to two layers of agency costs because both are agents. Hence, when the CEO wants something from a business unit manager, he will pay for it by directing an extra share of firm’s resources. One interesting implication of their research is that the larger the relative strength of one segment over the other divisions, the more the resources diverted to the weaker divisions. This form of socialism takes the name of cross-subsidization.

Rajan, Servaes and Zingales (2000) analyze budget’s allocation under the assumption that the CEO cannot perfectly monitor the projects. Indeed, the allocation of resources is concluded before the project is undertaken and, most importantly, this allocation process is conducted through negotiations with the divisions’ executives. Their conclusion is that internal power struggles lead to an excessive flow of resources towards the most inefficient units. Eventually, this leads to a destruction of the potential value of a diversified firm. The more unrelated the segments and the more diverse their investment opportunities, the larger the value loss will be.

Ozbas (2005) puts forward a model that relates the imperfectly informed CEO where units’ specialist managers have a tendency to ex-ante exaggerate the return on their projects. The intensity of competition between executives can lead to inflated statements on future returns despite possibly unfavorable career

(10)

consequences ex-post. In his framework the presence of many units (thus, many managers) creates the incentives to overstate the payoffs in order to obtain resources.

Turning to the empirical evidence regarding the inefficiencies of internal capital markets, Lamont (1997) studies the oil industry around the 1986 oil price decrease. In his findings he reports how with high oil prices, non-oil divisions of diversified oil conglomerates appear to invest more than their industry peers. This result is consistent with cross-subsidization among business units. Van der Stede (2001) analyzes 37 Belgian multi-business companies finding that diversification is positively associated with slack2_{in business units’ budgets. He}

argues that headquarters’ managers are unlikely to be familiar with the various projects of every division, so detecting slack is harder. In order to prevent slack, management needs more information. A higher amount of data puts pressure on the information processing capacity at the top. In order to reduce the information overload at the top, corporate managers are willing to tolerate some slack.

Ozbas and Scharfstein (2010) are the first to empirically test the link between managerial incentives and efficiency of internal capital markets. Their study focuses on unrelated segments in conglomerates. Unrelated segments are those that are not vertically integrated such that the output of one is not the input of another. These authors find that unrelated segments invest less in high-Q industries compared to their stand-alone counterparts. However, their most important result is that effects are stronger when the top managers own a smaller share of the conglomerate. The straightforward interpretation is that inefficiencies in resource allocation in conglomerates may be due to agency problems.

Empirical evidence has not only given contrasting results, but it has also been widely criticized. Whited (2001) argues that most empirical findings are characterized by endogeneity. Investment opportunities are usually measured through Tobin’s Q, but this may be a poor proxy. It is the unobserved marginal Q that matters and not the absolute observed level of the Q. Since the two may diverge substantially most of the research that used this proxy may report results that are due to measurement error. Through the use of a GMM regression to increase the precision of estimators, the author concludes that divisions do not

(11)

under or over invest relative to single-segment firms, given their investment opportunities.

Campa and Kedia (2002) reflect on the choice of most empirical work where authors compared conglomerates with stand-alone firms. This sample selection may be biased. Firms self-select where to operate (single or multiple industries) and this is a choice that often responds to changes in the economic environment that also cause variations in firm value. Companies move away from low growth sectors with high exit rates and enter into new ones. So they should be compared also with firms that decided to remain focused on the original sector and with firms who exited the sector. Therefore, the correlation between firm performance and diversification may not be causal.

2.2 Implications for Capital Budgeting

The most important implication of Stein (1997) is that of budget flexibility. In order to be able to effectively engage in winner-picking and in order to adjust to market opportunities the CEO must be actively managing its portfolio of divisions. The studies by Maksimovic and Phillips (2002) and Guedj and Scharfstein (2006) both show how a flexible budget structure creates value by reducing the agency problem between managers and shareholders that leads to over-investment. This is especially true when firms have a limited amount of projects that the headquarters’ managers are able to control.

The case for a more rigid budget structure has found supporters as early as almost thirty years ago. Ayal and Rothberg (1986) sustained that centralized ex-ante decisions may be better for R&D expenses because the headquarters is better able to evaluate the merit of the project. Only at the conglomerate top management level it is possible to assess strategic significance, interrelatedness, external implications and the newnesss (in terms of divergence from current technologies) of the different R&D projects. Later in the years, Harris and Raviv (1996) explore the theory behind the capital budgeting process in the presence of incentive problems and decentralized information. Units’ managers try to attract more resources than necessary to their divisions, because of private benefits or because of over-investment. The authors argue that a solution is to fix a limit to capital spending. This introduces rigidity in the budget allocation process. The amount of rigidity depends on the cost of auditing the claims of the managers who push to obtain more resources.

(12)

According to Ozbas (2005) and Marino and Matsusaka (2005) when there is competition among segments a rigid budget can be the optimal solution. The latters put forward a model with asymmetric information between a principal and a division manager with superior information. The central feature is that the agent can try to make the project look better and overspend. Their final result is that this process can be improved by commitment of the principal to limit rent-seeking behavior from division managers. Doing so consents to limit inefficient investments.

2.3 Internal Capital Markets in Crisis Years

The importance of internal capital is relative to the availability of external funding. Huang et al (2013) find that the amount of intra-company inter-divisional subsidies is larger in crisis years than during normal years, suggesting an increase in activity. Assuming that the external capital market is more efficient than the internal capital market (Stein, 1997), when the first offers less resources the second may improve its allocation process. Hence, the efficiency of internal funding can be greatly enhanced in years of economic downturn (ie, the internal winner-picking process has to improve). Additionally, the dark side of internal capital markets is smaller under financial crisis. Agency costs should decrease because managers are more likely to be disciplined by recessionary market conditions (Lee et al, 2009)

2.4 CEO Career Concerns and Firm Performance

Holmström (1999) put forward the basic framework that describes the relationship between current performance and future wages. The straightforward idea is that managers have abilities that are revealed slowly over time, through multiple observations of their performance. A problem may arise whenever there is a possibility that an individual’s concern for human capital returns and a firm concern for financial return clash. Since the two kinds of returns are not strongly correlated for top management positions, it seems reasonable to assume that younger CEOs have a stronger incentive to exhibit a greater attention to reveal their abilities by increasing their firm performance compared with their older colleagues.

Increasing age is positively associated with various measures of CEO Power Concentration and Entrenchment (Harjoto and Jo, 2009). Empirical evidence by

(13)

Gompers, Ishii, and Metrick (2003) and Bebchuk, Cohen, and Ferrell (2005) documents a negative relationship between CEO power measures, Bebchuk famous Entrenchment Index and market value or firm operating performance. Huson, Malatesta, and Parrino (2004) study firm performance after CEO turnover. Their evidence suggests that the more entrenched the departing CEO was, the larger the increase in operating returns in the following three years. Hence, age is positively associated with entrenchment and the latter is negatively associated with profitability measures. It seems straightforward to test whether a direct relationship exists between age and performance with the methodology outlined below. Existing evidence indicates that age has substantial effects on performance and on firms’ financial policies. The result is an under-investment problem: older CEOs invest less than younger CEOs and this is especially true in firms with larger growth opportunities (Serfling, 2012).

Previous analyses of the relation between tenure and performance largely agree on the fact that CEO Tenure does not affect operational profitability at lower level of tenure. However, when a CEO occupies the same position for at least twelve years, each additional year on the job seems to reduce profitability (Hermalin and Weisbach, 1991).

3. Methodology

3.1 Model

In this section, I present the equations that will be used to compare the economic performance of firms with different level of internal capital market activity. Economic performance is measured with a stock-market based measure and with an accounting based measure. Internal capital market activity is defined using two measures: one is used in some consultancies for clients’ segmentation and the other is a slightly modified version of that introduced by Billet and Mauer (2003). Both performance and activity will be described in deeper detail in the following sections.

Many authors – Scharfstein and Stein (2002) and Rajan, Servaes and Zingales (2000) – have highlighted the dark sides that undermine the efficiency of internal capital allocation. Inefficiencies, they argue, are generated by informational asymmetries between units’ executives and corporate headquarters. Based on this premise, other authors have suggested the benefits of reducing agency costs by

(14)

(2)

(3) specifying an ex-ante rigid medium term budget. Other authors – Stein (1997), Maksimovic, Phillips (2002) and Guedj (2007) – have stressed the benefits that superior management skills in allocating capital can bring. To demonstrate its ability in allocating funds a CEO needs to be able to engage in winner-picking, that is possible only with a flexible budget and active decisions.

The equation to investigate which strategy – inertia or continuous adaptation – will yield the better results in the medium and long term, takes the following form.

𝐹𝑖𝑟𝑚 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒!!!= 𝛼 + 𝛽𝑅𝑒𝑎𝑙𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛!,!!!+ 𝛾𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠 + 𝜀!

As exposed earlier, during financial crisis the external funding constraint is increasingly binding. In these circumstances, the relative importance of an efficient internal capital market is heightened. Hence, making the right moves and preventing immobility is crucial. To understand the importance of re-shuffling internal resources during crisis years, I estimate:

𝐹𝑖𝑟𝑚 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒_!!! = 𝛼 + 𝛽𝑅𝑒𝑎𝑙𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛_!,!!! + 𝜗𝑅𝑒𝑎𝑙𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛_!,!!!× 𝐶𝑟𝑖𝑠𝑖𝑠 + 𝛾𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠 + 𝜀!

Finally, is the degree of corporate resources allocation influenced by certain corporate governance characteristics such as CEO age or tenure? The idea is that career concerns are decreasing with respect to CEO age, thus a younger chief executives has more incentives to perform better. This reasoning is strictly linked to models, such as Holmstrom (1999), where future wages are linked to current performance by an implicit contract. Using CEO tenure consents to verify whether he is more likely to actively manage his resources at the beginning of his mandate, becoming more resilient to reconsider his long-term strategy as time goes by.

𝑅𝑒𝑎𝑙𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛!,!!! = 𝛼 + 𝛽𝐶𝐸𝑂𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 + 𝛾𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠 + 𝜀!

Where CEO characteristic is a dummy that assumes a positive value only when (i) CEO is in the bottom quartile of age; (ii) CEO is in top quartile of age

(15)

(4) distribution; (iii) CEO is in the first three years of its mandate; (iv) CEO is in the top quartile of tenure. Controls are: profitability, size, governance, amount of capital expenditures, crisis years dummy, diversity and Tobin’s Q.

3.2 Measures of Capital Re-Allocation

Measuring the activity of internal capital markets is not the easiest task. First of all, some variables are unobserved. Think of the amount of lobbying put in by units’ executives. Secondly, it is almost impossible to obtain the true amount of business segments for a large number of firms. Only educated estimates can be made. In this research, activeness measures are centered on the deviations of capital allocated to each unit from a mechanical decision rule. Drastically departing from this passive benchmark involves active decision making by the corporate headquarters.

3.2.1 Corporate Allocation Flexibility

In order to evaluate the level of resource reallocation within a conglomerate it is necessary to have a benchmark. The first measure of internal capital market activity uses the capital allocation of each firm in the past year as a benchmark. Earlier in this paper, I have written that previous authors, namely Harris and Raviv (1996) and Marino, Ozbas (2005) and Matsusaka (2005 have argued in favor of a rigid process to determine the portion of capital budget each unit received from the headquarters. Those authors have argued that less flexibility may be the solution to conglomerates where units’ executives possess more information on their businesses than the headquarters does. Therefore, a stricter budget process may limit agency problems.

To capture the level of flexibility in the resource re-allocation process, I introduce my measure of Corporate Allocation Flexibility (CAF). This takes the following form for each firm in each year:

𝐶𝐴𝐹! = 1 2 𝐶𝑎𝑝𝑒𝑥 𝑈𝑛𝑖𝑡_! 𝐶𝑎𝑝𝑒𝑥 𝐹𝑖𝑟𝑚!− 𝐶𝑎𝑝𝑒𝑥 𝑈𝑛𝑖𝑡_!!! 𝐶𝑎𝑝𝑒𝑥 𝐹𝑖𝑟𝑚!!! !"#$ ∈!"#$

Where Capex Unit represents the level of capital expenditures per segment in each firm in each year. Its is necessary that firms appear both in year t and in

(16)

(5) year t –1. Capital expenditures are always above zero, hence CAF always takes values between 0 and 1.

This indicator is then averaged out over three years to capture the general characteristics of the company strategy.

𝐶𝐴𝐹!,!!!=

1

3 𝐶𝐴𝐹!+ 𝐶𝐴𝐹!!!+ 𝐶𝐴𝐹!!!

But how to exactly interpret this indicator? The answer is simply that when the value is low, the firm is following a rigid strategy and there is no significant resource reallocation over time. Hence, once the headquarters designs the strategy set-up, each unit has to stick with its slice of the budget. Instead, if the value is high, the conglomerate is flexibly shifting resources with respect to the previous year.

For instance, think of a corporation with two divisions that subdivides capital expenditures evenly in year t-1 and allocates resources according to a 60-40 split in year t. In this case its CAF measure between t and t-1 would be 0.10. Whereas, a firm splitting evenly its capital expenditures in t-1 but reshuffling resources to obtain a 80-20 division in year t will have a CAF equal to 0.30. The first one can be thought of as a rigid firm, the latter as a flexible one.

However, a firm classified as rigid according to this indicator may still be characterized by a large degree of redistribution among units. The source of redistribution may be the divergence in the level or in the variation of cash flows in the various divisions. This measure only considers the inputs received by each unit, to broaden the horizon and solve this potential source of estimation bias I introduce the next measure.

3.2.2 Corporate Allocation Activeness

The second measure of internal capital market activity takes as a benchmark the free cash flow that each unit generates. Corporate Allocation Activeness (CAA) is defined as the deviation of capital expenditures of every segment from the cash flow it generated in the past year. The dark side of capital markets described by Scharfstein and Stein (2002) and Rajan, Servaes and Zingales (2000) is cross-subsidization: the subdivision of available free cash flow across business units.

(17)

(6)

(7) The goal of the CAA measure is to capture to what extent this process of cross-subsidization occurs.

Every year and for each firm CAA takes the following form:

𝐶𝐴𝐴! = 1 2 𝐶𝑎𝑝𝑒𝑥 𝑈𝑛𝑖𝑡_! 𝐶𝑎𝑝𝑒𝑥 𝐹𝑖𝑟𝑚!− 𝐶𝑎𝑠ℎ 𝐹𝑙𝑜𝑤 𝑈𝑛𝑖𝑡_!!! 𝐶𝑎𝑠ℎ 𝐹𝑙𝑜𝑤 𝐹𝑖𝑟𝑚!!! !"#$ ∈!"#$

Where Capex Unit has the same meaning as before. Firms need to appear in the dataset both in year t and in year t –1. For each segment, Cash Flow Unit is calculated by subtracting interest and tax expenses from the sum of operating income and depreciation. Cash Flow Firm is computed with the same approach, but using firm level data. Billet and Mauer (2003) compensate for the scarcity of data on units’ interest and tax expenses by applying firm level tax rate and passive interest rate to each segment. Although this may be a source of data distortion, it is also an approach widely used in literature –for instance, see Berger and Hann (2007) or Xuan (2009).

Averaging over a three years period to depict the general features of the strategy.

𝐶𝐴𝐴_!,!!! =1

3 𝐶𝐴𝐴!+ 𝐶𝐴𝐴!!!+ 𝐶𝐴𝐴!!!

This measure can be interpreted in the following way. When the value is low each business unit is inherently behaving as a stand-alone firm, it invests based on the resources it receives from the market. Instead, when the value is high, the headquarters of a conglomerate is being active in shifting capital from unit to unit I provide a numerical example to clarify what CAA measures. A firm with two units evenly subdivides capital expenditures between them in year t. However, in year t-1, the first unit contributes to 10% of the total cash flow, while the second brings the remaining 90%. Such a firm will have a CAA of 0.4 with one division subsidizing the other.

3.3 Measures of Corporate Performance

Measuring corporate performance often exposes to criticism as every measure has downfalls. In order to minimize failures and maximize results’ consistency, I use two measures of performance. One is stock-market based: Total Return to

(18)

(9) (8) Shareholders. One is accounting based: Return On Equity. Both are described below.

3.3.1 Total Return to Shareholders

With regards to the choice of a stock market based measure of performance the best choice was Total Return to Shareholders (TRS). TRS includes both stock price appreciation and monthly reinvestment of dividends over a certain timeframe. It is probably the best measure for a comparison of multi-business firms. The reason is that it allows shares to be compared even though some have low growth and high dividends and others have high growth but low dividends. The straightforward calculation of this measure is:

𝑇𝑅𝑆! =

𝑃𝑟𝑖𝑐𝑒_!− 𝑃𝑟𝑖𝑐𝑒_!!!+ 𝑅𝑒𝑖𝑛𝑣𝑒𝑠𝑡𝑒𝑑 𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑𝑠 𝑃𝑟𝑖𝑐𝑒!!!

In the equations presented above the relevant measures of future shareholder return will be the one-year, three-years and five-years annualized total return. According to the terminology of many investment world practitioners these thresholds will be named respectively: short-term, medium-term and long-term return.

3.3.2 Return on Equity

With regards to the choice of an accounting based measure of profitability the choice was to use Return on Equity (ROE). ROE is easily one of the most commonly used measures of a company performance in terms of earnings. At its core, it tells us how well is shareholders money being employed. The reason for choosing this ratio over other viable candidates (namely Return on Assets and Return on Sales) is that it is generally considered more appropriate for similar companies, industry and market comparisons.

ROE is calculated as:

𝑅𝑂𝐸 ! =

𝑁𝑒𝑡 𝐼𝑛𝑐𝑜𝑚𝑒_! 𝐵𝑜𝑜𝑘 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐸𝑞𝑢𝑖𝑡𝑦!

The main recognized weakness of this indicator is that a capital structure that disproportionately favors the amount of debt leads to a smaller equity base. Hence, a higher ROE for each level of Net Income. Since every profitability ratio has its pros and cons, my belief is that a combination of TRS and ROE represents a sound approach.

(19)

3. Data

Following previous literature, I obtain data for business units for the period of time between 1992 and 2013 from Compustat Business Segments Annual Database. Per Statement of Financial Accounting Standards 131, conglomerates are required to publicly disclose data on their business lines. Business segment reporting depends on the firm’s internal organization. An operating segment is defined as a division of the firm that generates revenues and incurs in costs. After aggregating the figures of all the different segments, a conglomerate must report on business units that account for one of: ten percent of revenues, ten percent of operating profits or ten percent of assets. According to SFAS 131, since 1992, a reportable division “may aggregate two or more operating segments if their products and services, production processes, type of customer, distribution and regulatory environments are similar.” According to the same rule, reported divisions should total at least 75% of total external revenues; if this is not the case additional divisions must be reported. In this research, it is necessary that for each firm the sum of total sales of all units is in a one percent interval from the total sales of the conglomerate as a whole.

The approach for data management was the same as in most studies on conglomerates and it is based on Berger and Ofek (1995). Not only firms and firms with segments in the financial industry (SIC codes between 6000 and 6999) are dropped, but also firms that operate in the utilities sector or can be considered as quasi-public (SIC codes 4900-4999 and larger than 9000). Firm-year outliers for capital structure and for market to book ratio are dropped (book leverage above 100% and market-to-book above 10. Small companies, with sales less than $20 millions or total assets less than $10 millions, are left out. Conglomerates should have no anomalous accounting data: capital expenditures larger than total sales, gross capital expenditures smaller than zero and need to be composed of at least three segments. Finally, I drop firms with incomplete accounting data. It is necessary to properly define financial crises years. Since one of the research questions of this paper compares firms’ characteristics during financial turmoil years with non-crisis years. Consistently with the definition brought forward by Berger and Bouwman (2010), financial crises are when companies find difficulties in obtaining funding on external financial markets. From 1992 to 2013 they determine the existence of two crises: burst of tech bubble and of housing bubble.

(20)

Checking the annual return on the S&P500 Index, negative returns are observed in 2000, 2001, 2002, 2008 and 2009. Therefore, those years will make up the two crisis periods covered in my dataset.

Table 1 shows summary statistics for the overall sample. The entire sample ranges from 1992 to 2010. It comprises 50169 firm-year observations and 1287 conglomerates. The variables that describe internal capital allocation, CAF and CAA, have respectively a mean of 0.211 and 0.491. Throughout the empirical analyses of this paper, I will consider only the three-year mean of those measures for each firm in order to seize the stable features of conglomerates only, rather than one-time fluctuations.

Table 1

Summary Statistics for All Variables

CAF is Corporate Allocation Flexibility as defined by equation (5). CAA is Corporate Allocation Activity as defined by equation (7). Number of segments is the number of operating and business segments included in Compustat Annual Segment Database, geographical segments are combined and reported as one unit. Size is the logarithm of Net Sales. G-Index is a measure of shareholder rights and is obtained from Prof. Metrick website. M/B is market-to-book ratio is the proportion of market value of assets to their book value. D / E is the ratio of debt to equity and controls for capital structure. ROA is Return on Assets defined as Net Income over total assets. Total Return to Shareholders is a common measure of investors’ return and is defined as in equation (8). ROE is Net Income on book value of equity for each year.

Variable Obs. Mean St. Dev. Variable Obs. Mean St. Dev.

CAF 31,897 0.211 0.125 Capex / Assets 50,169 0.055 0.043 CAA 21,485 0.491 0.287 Capex %∆ 40,037 0.128 0.574 N. Seg. 50,169 5.401 2.598 TRSt,t+1 50,169 12.273 37.333 Size 50,169 7.866 1.530 TRSt,t+3 50,169 8.840 19.203 G-Index 14,708 9.890 2.550 TRSt,t+5 50,169 8.523 13.940 M / B 50,169 1.500 0.685 ROEt,t+1 50,169 3.468 3.799 D / E 50,169 1.012 1.013 ROEt,t+3 29,560 3.980 7.366 ROA 50,169 4.227 6.643 ROEt,t+5 17,606 4.206 5.825

CEO Age 42,077 57.266 7.207 CEO Tenure 45,639 7.234 7.823

Young 12,274 49.221 3.425 Short Tenure 17,226 1.598 1.066

(21)

The number of segments averages at 5.401 over the complete set, but has an upward trend over the year. In 1992 the average firm had between 3 and 4 segments, whereas the average firm now has almost 6 segments. The pattern of an increasing number of segments described in Servaes (1996) still holds in my data. Capital expenditures have been rather constant relative to assets, growing at 12.8% in nominal terms. The typical firm has a capital structure divided equally between debt and equity, but the high standard deviation suggests a large dissimilarity among firms.

As a proxy for the quality of corporate governance, I employ the Index. The G-Index is a measure constructed with data compiled by the Investor Responsibility Research Center (IRRC). A company receives a score based on the number of shareholder rights-decreasing provisions a firm has. The higher the score, the poorer corporate governance is. Although it is widely accepted and is the de-facto standard in academic finance research literature, its use lured extensive criticism. Specifically, Klausner (2013) states that many elements of this index do not have potential to reduce firm value3_{or entrench management.}

To give the reader a better comprehension of the data I divide the sample in three groups based on their average three-years Capital Allocation Flexibility for all years in my sample. The average CAF for every year for the three groups is illustrated in Graph 1. Firstly, one can observe that most firms are extremely slow at re-allocating their capital expenditures. This is evident as the correlation among segment’s capital expenditures is close to one for both the rigid and the medium group. Secondly, firms defines as flexible show a significantly different resource allocation pattern compared with rigid conglomerates. The characteristics of each group are going to be discussed in the quintile portfolio comparison in the next chapter.

3_{For example, the presence of a poison pill is equal to the latent possibility that a firm}

(22)

4. Results

In this chapter, I compare the stock market and the operating performances of flexible and rigid firms. Observing performance enhances our knowledge about the efficiency of internal resource reallocation. When managers have the ability to reallocate capital to its best use, resource allocation is efficient. Hence, firms with a higher level of activity are expected to outperform more passive conglomerates. Conversely, when management is involved in lobbying, power struggles, demanding cross-subsidization or it simply lacks the managerial skills to identify business opportunities, rigidity might yield a superior payoff. Therefore, it will be passive firms that perform better.

In the following paragraphs I report the results of the tests described in the previous section to understand the link between the amount of resource reallocation in conglomerates and firm profitability. Firstly, I divide the firms into quintile portfolios based on their activeness and report on their respective profitability. Secondly, I perform fixed-effects regressions where I relate CAF and CAA measures with subsequent performance of conglomerates. Thirdly, I introduce in the regressions a dummy that only takes a positive value in crises years and report on the results. Finally, I analyze the impact of two CEO characteristics, age and tenure, on activeness.

(23)

4.1 Quintile Portfolio Comparison

In this section, a portfolio approach is employed to minimize the effect of outliers and to understand how well conglomerates allocate their resources. I sort firms in quintile portfolios based on their three-year CAF level. Firms with the higher CAF are the most rigid and firms with the lower CAF are flexible. In other words, flexible firms are defined as those corporations who actively deviate from the benchmark of last year expenses.

Table 2

Comparison of Quintile Portfolios

Data are subdivided in five subsamples based on their Capital Allocation Flexibility measure. Summary statistics of each portfolio are reported below. TRS is Total Return to Shareholder as defined in equation (5) in the following intervals: t and t+1, t and t+3, t and t+5. The numbers are in percentage points and are annualized. Size is the logarithm of Net Sales. M / B is market-to-book ratio. Lev is market leverage, defined as Total Assets over Total Liabilities. G-Index is a measure of shareholder rights and a proxy for corporate governance quality. ROA is Return on Assets defined as Net Income over Total Assets. N. Seg. is the number of segments reported in Compustat Annual Segment Database. Below the table I report the difference in financial performance between Rigid and Flexible firms and the respective t-statistics. The three asterisks represent significance at the 1% level.

Portfolio TRSt+1 TRSt+3 TRSt+5 Size M / B Lev. Capex / _Assets _IndexG- ROAt-1 N. Seg.

Rigid 14.06 10.19 8.32 8.49 1.54 0.40 0.05 10.04 5.12 5.41 2 13.41 8.30 6.22 8.30 1.49 0.40 0.05 10.27 5.35 5.35 3 13.58 8.57 6.17 8.07 1.52 0.39 0.05 10.13 4.93 5.32 4 13.13 8.08 6.24 7.97 1.49 0.40 0.05 10.23 4.63 5.38 Flexible 11.64 7.44 5.33 7.88 1.46 0.41 0.05 9.89 4.37 5.64 Difference (R – F) 2.42 2.75 2.99 t-stat (3.77) _*** (6.64) _*** (8.81) _***

In Table 2, the average values for each portfolio are reported. Rigid conglomerates tend to be slightly larger. However, the sample is made up of multi-business firms substantially larger than the average company and with total assets of around 13.06 billion dollars. Market-to book ratio in year t is only 5.4% higher than their active counterparts. So the stock market was not fully incorporating the subsequent increase in performance of the next five years. Additionally, it suggests that the level of investment opportunities across corporations is not what drives different resource reallocation decisions. Conglomerates in each

(24)

portfolio share the same level of market leverage and of capital expenditures relative to total assets.

Comparing the total return of active and passive firms can lead to a deeper understanding of internal capital market performance. It is evident how rigid companies exhibit superior future performance. The difference in TRS is persistent and increases with the timeframe considered. An investor deciding to acquire equity in firms that have re-allocated substantial amount of capital would have enjoyed a cumulative 24.02% return after three years and about 30% after five years. Investing in firms with a rigid budget over the previous periods would have yielded about 34% in three and 49.12% in five years. Therefore, if the two had bought shares for the same, now the first would have 14% less than the latter.

4.2 Internal Capital Allocation and Performance Regressions

The following regressions include fixed-effects to control for observed and unobserved firm characteristics. Those may have caused both the decision to be active and financial performance. The specification of the equation is as in equation (1) in the Methodology section. The dependent variable is Total Return to Shareholders in columns (a), (b) and (c) and Return on Equity in columns (d), (e) and (f) of Table 3. Capital allocation is measured by CAF as in equation (5). Control variables include: size, measured as the logarithm of sales; market leverage, defined as the ratio of book debt and total assets plus market equity minus book equity; and capital expenditures growth rate. The regression also features number of segments, a measure of internal business structure. Management has a certain level of discretion in deciding over the segmentation of a business. As reported in summary statistics the number of business units has increased over the years from 3.44 to 5.51. Each control variable has a one-year lag relative to the start year of the performance variable. Governance is the G-Index obtained from Prof. Metrick website. The idea of cross subsidization is that the private interests of the executives can distort decisions of the headquarters away from efficiency. Finance literature has generally held the position that weaker shareholder rights are value reducing because they incentivize a larger agency conflict between management and owners. This fact is even more valid in conglomerates, which are afflicted by a two-tiered agency problem. One is

(25)

between owners and management and one between management in the headquarters and management in business units.

Table 3

The effect of Capital Allocation Flexibility on Firm Performance

The dependent variable in columns (a), (b) and (c) is Total Return to Shareholders, a stock market based performance measure. In columns (d), (e) and (f) the dependent variable is Return on Equity, an accounting based measure. CAF is Corporate Allocation Flexibility, as defined in equation (5). M / B is market to book ratio, generally referred to as Tobin’s Q, it proxies for market firm-wide market opportunities. Size is the logarithm of Net Sales. G-Index is a measure of shareholders’ rights, it is a proxy of good corporate governance. A higher score reflects weaker rights. Leverage is market leverage defined as Total Assets over Total Liabilities. Capex Growth is the rate of change of Capital Expenditures. N. Seg. is the number of business segments reported by each company in the Compustat Annual Segment Database. Heteroskedasticity-consistent standard errors are reported in parenthesis. All regressions include a constant and firm fixed-effects. * indicates rejection of the null at the 10% significance level; ** indicates 5% significance level, and *** indicates 1% significance level.

Regressors (a) (b) (c) (d) (e) (f)

CAF _(6.477)9.480 -3.752** _(1.992) -6.482*** _(1.657) -7.099*** _(1.829) -3.411*** _(1.009) -1.906*** _(0.896) M / B -6.673*** _(0.870) -0.292** _(0.016) 0.374*** _(-0.136) -1.820*** _(0.206) _(0.083)-0.108 -0.116** _(0.060) Size -23.903*** _(1.786) -1.024*** _(0.175) -1.247*** _(0.147) 3.288*** _(0.546) 0.327 (0.274) 0.419** _(0.239) G-Index _(0.829)-1.047 0.266*** _(0.086) 0.269*** _(0.074) _(0.238)-0.005 _(0.143)-0.559 _(0.096)-0.374 Leverage -14.829*** _(7.417) -16.041*** _(1.473) -14.655*** _(1.187) 25.318*** _(5.129) 4.574*** _(1.323) 3.234*** _(0.831) Capex Growth (1.116) -0.183 -3.706*** (0.531) -2.750*** (0.463) 1.209*** (0.366) 0.998*** (0.217) 0.530*** (0.081) N. Seg. 0.087 (0.411) _(0.110)-0.031 _(0.089)0.70 -0.418*** _(0.144) -0.269*** _(0.066) -0.167*** _(0.047) R2 _0.23 _0.06 _0.08 _0.04 _0.04 _0.04 Obs. 7,910 6,558 4,398 9,715 7,081 5,307

Table 3 reports the multivariate fixed-effects regression estimation. In all columns except for (a) I found a negative and significant coefficient for CAF. This supports the idea that more flexible firms have lower future financial and operating performance. The coefficient of size ratio confirms the common saying among investors that ‘size is the enemy of performance’ as they are both negative. The negative coefficient on market to book ratio can be interpreted to support the idea that stock prices do not fully reflect the degree of flexibility of a conglomerate. Better governance, a lower value on the G-Index, is value increasing. Hence, shareholders find it more convenient to be invested in a

(26)

company where they have more rights. Some other control variables have statistical significance, but their contradictory results do not leave room for an economical explanation. As they were not at the center of my analysis they are irrelevant.

This model allows for measuring some numerical economic effects of the decisions taken when executives choose their medium to long-term strategy. Given the numbers in Table 3 – column (c) – for instance, think of a company that sets its Corporate Allocation Flexibility to change to the 10th_{percentile – thus very rigid –}

from the 90th_{percentile. Ceteris paribus, its shareholders could have received an}

additional 1.98% return for 5 years. Considering that, as reported in Table 1, the average 5-year TRS has been 8.52%, the gain of being rigid corresponds to an extra 23.24%.

Table 4

The effect of Capital Allocation Activeness on Firm Performance

The dependent variable in columns (a), (b) and (c) is Total Return to Shareholders, a stock market based performance measure. In columns (d), (e) and (f) the dependent variable is Return on Equity, an accounting based measure. CAA is Corporate Allocation Activeness, as defined in equation (7). M / B is market to book ratio, generally referred to as Tobin’s Q, it proxies for market firm-wide market opportunities. Size is the logarithm of Net Sales. G-Index is a measure of shareholders’ rights, it is a proxy of good corporate governance. A higher score reflects weaker rights. Leverage is market leverage defined as Total Assets over Total Liabilities. Capex Growth is the rate of change of Capital Expenditures. N. Seg. is the number of business segments reported by each company in the Compustat Annual Segment Database. Heteroskedasticity-consistent standard errors are reported in parenthesis. All regressions include a constant and firm fixed-effects. * indicates rejection of the null at the 10% significance level; ** indicates 5% significance level, and *** indicates 1% significance level.

CAA -9.687 *** _(3.453) -3.341** _(1.685) _(1.476)-0.247 -6.150*** _(1.086) -1.322*** _(1.532) -1.431*** _(0.496) M / B -10.897*** _(1.112) -6.853 ** _(0.511) -2.652*** _(-0.539) -2.318*** _(0.306) _(0.092)-0.009 -0.069** _(0.067) Size -20.926*** _(2.215) -12.414*** _(1.273) -9.073*** _(0.953) 1.316*** _(0.918) 0.591 (0.301) 0.270 (0.239) G-Index _(1.004)-1.045 -1.793*** _(0.452) -1.507*** _(0.295) 0.018 (0.277) -0.135** _(0.123) _(0.107)-0.276 Leverage -17.405*** _(7.564) -61.628*** _(4.025) -49.683*** _(3.755) 24.834*** _(7.843) 4.813*** _(1.781) 3.839*** _(1.050) Capex Growth (1.259) -0.420 -1.343*** (0.447) -0.985*** (0.342) 1.525*** (0.366) 0.469*** (0.255) 0.142 (0.081) N. Seg. _(0.518)-0.154 _(0.304)-0.386 _(0.226)0.079 -0.078*** _(0.138) -0.173*** _(0.071) -0.137*** _(0.058) R2 _0.24 _0.44 _0.45 _0.03 _0.03 _0.03 Obs. 5,300 4,222 2,752 6,433 4,752 3,608

(27)

The results of the specification using CAA as the allocation activity variable are described in Table 4. Consistent with the finding that the lower future profitability of active firms is not fully incorporated into stock prices, we find that investors in active firms have lower returns than investors in equity of passive firms. Particularly, TRS decreases monotonically with both versions of resource the re-allocation measure.

As earlier, to give an economic interpretation of the findings in Table 4 think again of a company passing from activeness to rigidity. Holding other things equal, a conglomerate setting this kind of strategy would enjoy a decrease in its Return on Equity of 0.97%. A striking difference of about 18% relative to the median of my measure of operating performance. Evidence suggests similar results using CAF and CAA versions, as both capture critical characteristics of multi-business enterprises that are related to financial and operating performance.

One common criticism of previous papers was that firms that choose to become conglomerates are different from stand-alone firms – Campa and Kedia (2002) and Chevalier (2004) – or that multi-business companies and single-business companies have different investment opportunities – Whited (2001) and Villalonga (2004). The results of this paper are based on a comparison between different conglomerates, instead of relating them to a portfolio of stand-alones. Comparing multi-business corporations only to their peer group should eliminate this issue. Although it is still possible that activeness is caused by some external factor that neither control variables nor fixed-effects are able to catch.

4.3 Internal Capital Allocation and Performance During

Financial Crises

The second main question this study tries to answer is whether the crisis has affected the impact that internal capital market activity has on firm performance. The results of the regressions reported in Table 5 and Table 6 include an interactive variable between capital allocation measures and crises years. Consistently with the findings exposed in non-crisis years, both CAF and CAA are associated with negative coefficients. Thus, again there is a link between higher yearly re-allocation of resources and subsequent financial and operating performance.

(28)

Although re-allocating resources is generally negative, this does not appear to be the case in periods of financial downturn. After adding the interaction variable with a dummy that takes a value of one in 2000, 2001, 2002, 2008 and 2009 and zero for the other years in the sample, an active internal market has a positive and significant effect on the successive performance of corporations after a crisis.

Table 5

The Effect of Flexibility During Crises

The dependent variable in columns (a), (b) and (c) is Total Return to Shareholders, a stock market based performance measure. In columns (d), (e) and (f) the dependent variable is Return on Equity, an accounting based measure. CAF is Corporate Allocation Flexibility, as defined in equation (5). CAF × Crisis is simply CAF multiplied by a dummy variable that takes positive value in years 2000, 2001, 2002, 2008 and 2009. M / B is market to book ratio, generally referred to as Tobin’s Q, it proxies for market firm-wide market opportunities. Size is the logarithm of Net Sales. G-Index is a measure of shareholders’ rights, it is a proxy of good corporate governance. A higher score reflects weaker rights. Leverage is market leverage defined as Total Assets over Total Liabilities. Capex Growth is the rate of change of Capital Expenditures. N. Seg. is the number of business segments reported by each company in the Compustat Annual Segment Database. Heteroskedasticity-consistent standard errors are reported in parenthesis. All regressions include a constant and firm fixed-effects. * indicates rejection of the null at the 10% significance level; ** indicates 5% significance level, and *** indicates 1% significance level.

CAF -4.723 -4.506 -8.038 -7.095 -2.629 -0.604 (6.650) (2.771)* (2.175)*** (1.182)*** (0.857)*** (0.548) CAF × Crisis 17.578 5.080 15.961 0.455 -1.182 -0.401 (4.598)*** (1.828)*** (1.515)*** (0.853) (0.492)** (0.331) Market to Book -6.451 -4.709 -1.796 -1.319 -0.118 -0.162 (0.805)*** (0.466)*** (0.330)*** (0.133)*** (0.058)** (0.041)*** Size -23.165 -13.628 -7.853 3.061 0.819 0.488 (1.666)*** (0.921)*** (0.652)*** (0.324)*** (0.228)*** (0.159)*** G-Index -0.895 -1.007 -1.058 -0.520 -0.314 -0.217 (0.781) (0.442)** (0.267)*** (0.128)*** (0.097)*** (0.080)*** Leverage -138.881 -73.673 -49.902 18.112 6.417 3.252 (8.190)*** (4.227)*** (3.076)*** (1.655)*** (0.933)*** (0.681)*** Capex Growth -2.455 -1.029 -0.917 0.649 0.596 0.272 (1.406)* (0.391)*** (0.282)*** (0.200)*** (0.095)*** (0.070)*** N. Seg. -0.494 -0.414 -0.096 -0.055 -0.150 -0.135 (0.419) (0.210)** (0.159) (0.066) (0.052)*** (0.038)*** R2 _0.24 _0.44 _0.48 _0.08 _0.06 _0.05 Obs. 8,136 6,593 4,118 9,526 6,927 5,132

The rationale behind it is probably that when external capital is abundant, managers exploit firms’ resources to seek private benefits. Instead, during bad years, executives struggle for maintaining their positions within the firm and

(29)

funds are re-directed towards more productive uses as a survival instinct prevails in management. Resource re-allocation make these firms perform better not only during the aftermath of the crises, but also in the long-term.

Table 6

The Effect of Activeness During Crises

The dependent variable in columns (a), (b) and (c) is Total Return to Shareholders, a stock market based performance measure. In columns (d), (e) and (f) the dependent variable is Return on Equity, an accounting based measure. CAA is Corporate Allocation Allocation, as defined in equation (5). CAA × Crisis is simply CAF multiplied by a dummy variable that takes positive value in years 2000, 2001, 2002, 2008 and 2009. M / B is market to book ratio, generally referred to as Tobin’s Q, it proxies for market firm-wide market opportunities. Size is the logarithm of Net Sales. G-Index is a measure of shareholders’ rights, it is a proxy of good corporate governance. A higher score reflects weaker rights. Leverage is market leverage defined as Total Assets over Total Liabilities. Capex Growth is the rate of change of Capital Expenditures. N. Seg. is the number of business segments reported by each company in the Compustat Annual Segment Database. Heteroskedasticity-consistent standard errors are reported in parenthesis. All regressions include a constant and firm fixed-effects. * indicates rejection of the null at the 10% significance level; ** indicates 5% significance level, and *** indicates 1% significance level.

CAA -11.828 -3.557 -2.902 -1.225 0.608 -1.808 (3.729)*** (1.604)** (1.560)* (0.616)** (0.338)* (0.477)*** CAA × Crisis 11.197 2.798 8.850 0.910 -0.360 0.730 (2.272)*** (0.819)*** (1.015)*** (0.328)*** (0.165)** (0.346)** Market to Book -9.641 -6.731 -2.504 -1.760 -0.019 -0.077 (1.068)*** (0.471)*** (0.544)*** (0.167)*** (0.052) (0.066) Size -21.363 -11.953 -7.181 2.779 1.283 0.276 (2.181)*** (1.175)*** (1.096)*** (0.323)*** (0.167)*** (0.219) G-Index -1.282 -1.881 -1.438 -0.535 -0.108 -0.264 (0.945) (0.465)*** (0.308)*** (0.130)*** (0.082) (0.109)** Leverage -105.038 -56.176 -35.545 17.511 0.752 5.250 (8.149)*** (4.246)*** (4.209)*** (1.991)*** (0.711) (1.292)*** Capex Growth -4.058 -1.335 -1.331 0.149 0.607 0.164 (1.872)** (0.424)*** (0.427)*** (0.241) (0.088)*** (0.091)* N. Seg. -1.155 -0.502 -0.416 -0.006 -0.081 -0.154 (0.533)** (0.293)* (0.242)* (0.077) (0.039)** (0.058)*** R2 _0.25 _0.46 _0.46 _0.08 _0.10 _0.04 Obs. 5,46 4,441 2,345 6,331 4,094 3,602

As earlier, in harmony with previous corporate governance studies, firms with a higher G-Index, weaker rights for shareholders, perform worse. Also, firms with more debt and less segments have had better results. Multi-business companies with ex-ante higher valuations did poorly in the next periods. Finally, the number