Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries

J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Journal of International Financial Markets, Institutions & Money j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / i n t fi n

Managerial discretion, net operating assets and the crosssection of stock returns: Evidence from European countries q Georgios Papanastasopoulos a,⇑, Dimitrios Thomakos b a b

University of Piraeus, Department of Business Administration, Greece University of Peloponnese, Department of Economics, Greece

a r t i c l e

i n f o

Article history: Received 30 March 2015 Accepted 24 November 2016 Available online xxxx Keywords: Net operating assets Stock returns Managerial discretion European equity markets

a b s t r a c t We show that firms with higher NOA (net operating assets) subsequently experience lower stock returns in at least nine out of sixteen European countries, consistent with the U.S. evidence. This negative relation between NOA and future returns is strongly linked with crosscountry variation in factors capturing managerial discretion. However, once we adjust for risk, the effect of NOA on stock returns is substantially attenuated and becomes significant only in three European countries. Overall, our findings suggest that optimal investment by executives in response to discount rate changes could be an underlying source of return predictability attributable to NOA in Europe. Ó 2016 Elsevier B.V. All rights reserved.

1. Introduction The effect of managerial discretion on firm performance is an important topic of research in strategic and international management. At the same time, in market-based accounting research the concept of managerial discretion is used to interpret predictable returns associated with various accounting figures. In this paper, we combine these two lines of research in the context of return predictability associated with net operating assets in an international setting. The level of net operating assets is an important accounting figure: it is equal to the difference between operating assets and operating liabilities, while it also represents the cumulation over time of the difference between net operating income and free cash flow (see Penman, 2007). Thus, the level of net operating assets serves as a measure of past and current operating performance (i.e., a measure of balance sheet bloat). At the same time, operating asset/liability accounts are more subject to managerial discretion than other accounts constituting a firms’ net financial asset position. Both the level of net operating assets and the level of various operating asset/liability accounts, have informational content for future profitability and stock price performance (see Hirshleifer et al., 2004; Papanastasopoulos et al., 2011). Hirshleifer et al. (2004) show a strong negative relation between the level of net operating assets scaled by lagged total assets (NOA, hereafter) and future stock returns. As suggested by the literature, managerial discretion drives this relation. Hirshleifer et al. (2004) and Papanastasopoulos et al. (2011) attribute the relation between NOA and subsequent returns to misunderstanding of earnings management and/or overinvestment, while Wu et al. (2010) to optimal investment by

q The authors appreciate helpful comments from the seminar participants at the XXI International Conference on Money, Banking and Finance (2012), at the 37th European Accounting Association Annual Congress (2014), at the Rimini Conference in Economics and Finance (2014), at the 2nd Paris Financial Management Conference (2014) and at the 1st International Conference on Business & Economics of the Hellenic Open University (2015). The authors thank Gikas Hardouvelis, Vasileios Papadakis and two anonymous reviewers for insightful comments and suggestions. The usual disclaimer applies. ⇑ Corresponding author. E-mail addresses: [email protected] (G. Papanastasopoulos), [email protected] (D. Thomakos).

http://dx.doi.org/10.1016/j.intfin.2016.11.013 1042-4431/Ó 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

2

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

executives in response to discount rate reduction. The sample in these studies is, however, limited on firms in the U.S. capital market. Hambrick and Finkelstein (1987) conceptualize managerial discretion, as latitude of action, in an effort to reconcile the long-running discussion in the literature about the influence of executives over the performance of their firms. Before the seminal work of Hambrick and Finkelstein (1987), economists relied on Williamson (1963) and used the term of managerial discretion to define how much leeway executives have in the pursuit of their personal objectives versus the objectives of the stakeholders of their firms. Shen and Cho (2005) integrated both points of view concerning discretion into a single framework, while recently Finkelstein and Peteraf (2007) point out that they address similar antecedents. Hambrick and Finkelstein (1987) argue that managerial discretion originates from three levels: individual, organizational and environmental. Research on environmental determinants of discretion is primarily conducted in terms of industry-level factors (e.g., Hambrick and Abrahamson, 1995). Possible national sources of variation in discretion are largely ignored, although the literature clearly suggests that the environment within the country where firms operate has a substantial impact on their strategies and performance (e.g., Makino et al., 2004). The recent studies by Crossland and Hambrick (2007) and Crossland and Hambrick (2011) constitute the first systematic attempt in the literature that explores the determinants of managerial discretion at the country-level. Both studies indicate that certain formal and informal national institutions – culture, ownership structure, legal origin, employer flexibility – greatly influence managerial discretion, and in turn the effects by firm executives on corporate performance. In the context of our work their findings are very important since they provide a mediating role for managerial discretion on firm performance in an international setting. Following up on studies about managerial discretion at the industry level, Zhang (2006) shows that within the U.S. stock market, both the cross industry and the within industry components of NOA are strong negative predictors for future stock returns. Nevertheless, to the best of our knowledge, we are not aware of any study investigating the possible influence of cross-country variation in managerial discretion on the relation between NOA and stock returns. This issue forms our essential motivation to examine two research questions: (1) whether the negative relation between NOA and subsequent stock returns generalizes to other countries; (2) whether the occurrence of the negative relation between NOA and subsequent stock returns is associated with differences in important country-level factors capturing managerial discretion. In addressing the first research question, we conduct our work on a sample of fifteen countries of the European Union (EU) prior to its enlargement in 2004, plus Switzerland, that are developed economies with a code-law tradition. In doing so, we can assess whether return predictability attributable to NOA constitutes a more pervasive asset pricing regularity or just a ‘‘freak” occurrence in the U.S.; a developed economy, but with common-law tradition and different accounting standards with those in EU (before and after IFRS adoption). At the same time, by conducting our analysis in European equity markets that enjoy some homogeneity regarding the status of the economy and legal origin we intend to eliminate any effects arising from disparities in this respect and to focus more on effects arising from variation in managerial discretion. In our work, we also control for other well-known determinants of stock returns in the cross section and take into account cross-country variation in transaction costs. In addressing the second research question, we blend blending the work of Crossland and Hambrick (2007, 2011) with the strand of literature dealing with implications of managerial discretion on accounting puzzles. Based on Crossland and Hambrick (2011), EU countries face systematically different degrees of constraint on firm executives (see Table 1, p. 806). Institutional theory suggests that both informal institutions (i.e., enforced by the society) and formal institutions (i.e., enforced by the state) can be related with the extent of leeway that firm executives possess. In our work, we investigate the impact of factors that are associated with cultural environment (individualism), equity-market setting (market development, ownership structure), productivity and growth potential (competitiveness), and accounting regimes (quality of accounting standards). We also directly investigate the impact of the national level of managerial discretion over earnings on the relation of NOA and future returns. In this regard, we seek to gain insights on the rationale of the relation between NOA and stock returns. The remainder of the paper is organized as follows. In the next we develop our hypotheses, while in Section 3 we provide details about the research design. Section 4 describes the data and variable measurement. Section 5 provides our empirical results. Section 6 summarizes and concludes the paper.

2. Theory and hypothesis development A substantial stream of the literature in strategy investigates how, whether and when managerial discretion has effect on corporate performance. Another body of literature in accounting interprets the operating and stock price performance attributable to accounting figures as a consequence of executive activities and decision making behavior. We interconnect these two lines of research, by investigating the role of managerial discretion on return predictability associated with NOA in an international setting.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

3

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx Table 1 Summary statistics on NOA. Country

Firm-Year Obs.

% of Total Obs.

Mean

St. Dev.

25th Percentile

Median

75th Percentile

Austria Belgium Denmark Finland France Germany Greece Ireland Italy Netherlands Norway Portugal Spain Sweden Switzerland United Kingdom Country-Average All Countries

1022 1403 2009 1562 9029 7596 2717 754 2951 2367 2047 891 1660 3619 2745 20,238 3193 62,610

1.632% 2.241% 3.209% 2.495% 14.421% 12.132% 4.340% 1.204% 4.713% 3.781% 3.269% 1.423% 2.651% 5.780% 4.384% 32.324% 6.25% 100%

0.585 0.646 0.659 0.637 0.545 0.550 0.776 0.741 0.608 0.591 0.691 0.739 0.687 0.622 0.616 0.658 0.647 0.627

0.311 0.382 0.300 0.269 0.297 0.403 0.319 0.468 0.274 0.289 0.495 0.277 0.318 0.444 0.267 0.501 0.351 0.408

0.401 0.443 0.496 0.478 0.371 0.313 0.595 0.506 0.441 0.426 0.424 0.570 0.514 0.384 0.46 0.418 0.453 0.414

0.583 0.626 0.642 0.637 0.541 0.504 0.761 0.669 0.594 0.58 0.629 0.729 0.688 0.565 0.615 0.602 0.623 0.595

0.752 0.799 0.793 0.782 0.692 0.708 0.926 0.857 0.737 0.722 0.844 0.868 0.834 0.756 0.764 0.787 0.789 0.771

Notes: Table 1 reports univariate statistics (mean, standard deviation, 25th percentile, median, 75th percentile) on net operating assets (NOA). The sample consists of 62,610 firm-year observations over the period 1988–2009. Net operating assets (NOA) are defined in Appendix A.

Several findings from research in capital markets-based accounting suggest that both the level and the growth of balance sheets are strong negative predictors of stock returns in the cross-section. Within the U.S. capital market, Hirshleifer et al. (2004) first report evidence for the level of NOA, while Cooper et al. (2008) for the growth rate in total assets.1 Whether the relation of NOA with future returns reflects mispricing or compensation for risk is still an issue under debate in the literature (see Hirshleifer et al., 2004; Papanastasopoulos et al., 2011; Wu et al., 2010). Our first hypothesis concerns whether return predictability attributable to NOA is limited to the U.S. capital market or constitutes a more pervasive empirical regularity. In this respect, Chudek et al. (2011) among others claim that generalizability of return predictability attributable to various accounting figures in a global setting continue to attract researchers’ interest. The existing literature on the predictive ability of balance sheet measures for future returns in an international setting focuses on the asset growth rate (see Gray and Johnson, 2011; Yao et al., 2011; Titman et al., 2013; Watanabe et al., 2013). We need to stress, that we are not aware of any other study investigating the relation of NOA with future returns outside U.S. Our analysis is conducted on fifteen countries of the European Union (EU) prior to its enlargement in 2004 plus Switzerland that constitute a set of developed economies with a code-law tradition. 2 European countries are characterized by both differences and similarities with U.S. in institutional environment, regulation, capital market setting, business practices, accounting structure, etc. Consider as examples the following cases. A code law tradition is more likely to exist in Europe, while U.S. has a common law tradition. Like (unlike) U.S., Nordic countries (South European countries) are expected to have high (low) economic development and competitiveness. U.K. is an Anglo-Saxon common law country with a large stock market, high trading liquidity and low transaction costs. Ireland is also an Anglo-Saxon common law country, but with a small stock market, low trading liquidity and high transaction costs. Accounting standards in European countries differ with those in U.S., although differences between IFRS and US GAAP are reduced after the convergence projects between IASB and FASB. Importantly, these country-level factors could be linked with the occurrence of return predictability attributable to NOA. Thus, is questionable whether the negative relation between and NOA and subsequent returns documented in the U.S. stock market, can be generalized in Europe, and if it is, that it can be generalized in all European equity markets. This leads to the first hypothesis (expressed as the alternative) of the study: H1. The negative relation of NOA with future returns also occurs in European capital markets. Regarding the source of the negative relation of NOA with future returns, Hirshleifer et al. (2004) argue that it may be driven from investors’ misunderstanding of earnings management or inability to make full use of available accounting information. The earnings management explanation suggests that high NOA could be exploited as managers opportunistically inflate earnings; when earnings management reverses, stock prices adjust downward. 1 Additional empirical evidence can be found in Chan et al. (2008), Fama and French (2008), Lam and Wei (2011), Li and Zhang (2010), Zhang (2006) and Wu et al. (2010). 2 Regarding the status of the economy, the great majority of EU countries are identified by IMF and OECD as developed economies. Nevertheless, during the recent financial crisis, three EU countries, Greece, Irelandand Portugal, requested a bailout loan package from EU and IMF. Regarding legal origin, the great majority of EU countries are classified by La Porta et al. (1998) among others as code-law countries, while only Irelandand United Kingdom are classified as common-law countries. At the same time, as pointed out by Arce and Mora (2002) and by Ball et al. (2000), Netherlands is formally classified as a code-law country, but for accounting purposes it should be classified as a common-law country. We need to stress here, that we do not assume pure homogeneity among EU countries regarding the status of the economy, and legal origin. In particular, we focus on each country separately and then, we pool them all-together.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

4

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Building on q-theory of investment, Wu et al. (2010) provide a rational interpretation associated with managerial discretion for the negative relation of NOA with future returns; executives optimally adjust investment upwards (i.e., leading to a higher level of NOA) in a rational response to discount rate reduction. Recently, Papanastasopoulos et al. (2011) claim that the negative relation of NOA with future returns may arise from investors’ misunderstanding of earnings management and/or agency-related overinvestment. Agency-related overinvestment suggests that high NOA could be derived as managers with empire building incentives engage in wasteful capital expenditures; when investors learn about the implications of overinvestment, they downwardly revise their valuation on high NOA firms. It is clear that managerial discretion must be a key in understanding return predictability associated NOA. Managerial discretion is conceptualized in the literature through two approaches: a behavioral - oriented and an economic - oriented. Based on the behavioral - oriented approach offered by Hambrick and Finkelstein (1987), managerial discretion refers to the latitude of actions available for firm executives in the pursuit of organizational objectives. According to the economicoriented approach that builds on Williamson (1963) managerial discretion refers to the amount of freedom available to firm executives to pursue their own objectives. Shen and Cho (2005) reconcile both approaches by arguing that they are orthogonal; the economic-oriented approach is aligned along an axis with latitude of objectives and the behavioral-oriented approach along another axis with latitude of actions. In particular, Shen and Cho (2005, p. 846) argue that ‘‘latitude of objectives addresses the performance pressure faced by managers, whereas latitude of actions addresses the range of strategic options available to managers as they strive to bring about the performance demanded by stakeholders”. The above points suggest that managerial discretion is not, per se, necessarily good or bad. On the one hand, greater discretion associated with open-mindedness could result to more competitive and innovative strategies such as optimal investment in value-enhancing projects. On the other hand, greater discretion associated with recklessness and hubris could result to strategies that lack stakeholder buy-in. In the latter case, executives may engage in opportunistic earnings manipulation and/or wasteful spending in value-destroying projects to serve their own interests. Such practices could have clear effects on stock price performance attributable to NOA. Hambrick and Finkelstein (1987) claim that the extent of managerial discretion emanates from three set of factors: the individual (e.g., tolerance of ambiguity, locus of control, cognitive complexity, professional aspiration, power base, political acumen, commitment); the organization (e.g., firm age, size, resources, capital intensity, corporate governance) and the environment (e.g., product differentiability, market growth, industry structure, demand instability, quasi-legal constraints, powerful outside forces). Regarding the third level, the great majority of research is conducted in terms of industry characteristics. Implications of managerial discretion on stock price performance attributable to various accounting figures (e.g., book to market ratio, accruals, NOA) at the industry level are clearly shown in Chan et al. (2006), Fama and French (1997), Zhang (2006) and Papanastasopoulos et al. (2011). Recently, Crossland and Hambrick (2007) investigate managerial discretion at the country level and show that there are systematic differences between German and Japanese CEO on corporate performance. As argued there, these differences arise from variation in culture, ownership structure and board governance. In follow up research, Crossland and Hambrick (2011) show a great impact of several informal institutions (individualism, uncertainty tolerance, power distance, cultural looseness) and several formal institutions (ownership structure, legal tradition and employer flexibility) in the level of managerial discretion across 15 countries (U.S., Japan, Canada, Australia, Korea, Singapore, Switzerland and 7 countries in Europe). More importantly, they clearly demonstrate that within countries with more (less) managerial discretion, executives have more (less) impact on corporate performance. Our second hypothesis concerns the role of managerial discretion on stock return predictability attributable to NOA in European capital markets. Hirshleifer et al. (2004), Papanastasopoulos et al. (2011) and Wu et al. (2010) attribute the negative relation between NOA and subsequent returns within the U.S. capital market to executive discretion. Crossland and Hambrick (2007, 2011) show that the level of discretion by firm executives is strongly influenced by the ‘‘national” environment. Lee (1997) provides evidence of cross-country differences in stock market response to managerial choices and actions. Thus, the extent of leeway that executives possess is expected to have severe implications on stock price performance attributable to NOA in an international setting. This leads to the following hypothesis: H2. The negative relation of NOA with future returns is associated with managerial discretion in a country. Our work makes at least two contributions in the strategic management and accounting literature. First, we provide evidence on the occurrence of the negative relation between NOA and subsequent returns in a set of European countries and thus, provide new insights behind the possible generalization of this prominent empirical regularity. Second, we demonstrate the prominent role of managerial discretion under return predictability attributable to NOA in an international setting. 3. Research design In the first part of our work, we investigate whether the negative relation between NOA and future returns occurs in European equity markets via cross-sectional regressions and firm-level portfolio tests. Following the existing literature on the generalization of asset pricing regularities in international equity markets (McLean et al., 2009; Titman et al., 2013; Watanabe et al., 2013), we consider in our empirical tests subsequent holding period raw returns. We initially analyze each country separately and then, we perform country-average analysis. We also conduct analyses by pooling countries all-together. Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

5

We consider cross-sectional regressions of subsequent holding period raw returns on NOA. In regressions, we control for other determinant of stock returns in the cross section such as size and book to market ratio. In this respect, we assess whether return predictability attributable to NOA is independent of these well-known factors. In our firm-level portfolio tests, we follow the common convention of assessing future raw returns of portfolios ranked on the level of NOA. We report our regression and portfolio results by taking into account cross-country variation in transaction costs. By considering transaction costs we seek to examine whether investors can benefit from possible return predictability attributable to NOA. Consideration of transaction costs can also shed light on whether a relation between NOA and stock returns in European countries reflects systematic mispricing, an issue that is still debatable for the U.S. stock market (see Hirshleifer et al., 2004; Papanastasopoulos et al., 2011; Wu et al., 2010). Such a possibility, suggests that the NOA-return relation should generalized in European countries with higher trading costs which will make more costly and risky for arbitrageurs to trade on the relation and thus, to correct mispricing. In the second part of our work, we examine whether the generalization of the negative relation between NOA and future returns could be associated with differences in country-level factors associated with managerial discretion. Documenting the existence of return predictability attributable to NOA in an international setting does not necessarily imply that the relation is driven by managerial discretion as in the U.S. In this respect, we first connect an asset pricing regularity attributable to an important accounting figure that attracts researchers’ interest in market-based accounting with factors reflecting discretion at the national level. We also employ a much larger dataset at the firm-level than previously considered by researchers in strategic and international management in assessing the effects of managerial discretion on corporate performance. Institutional theory suggests that both informal institutions (i.e., enforced by the society) and formal institutions (i.e., enforced by the state) can be related with the extent of leeway that firm executives possess. We focus on country-level factors covering culture dynamics (individualism), equity-market setting (equity-market development, ownership concentration), productivity and growth potential (competitiveness), and accounting regimes (quality of accounting standards). 3.1. Cultural environment Starting with culture, an informal institution that greatly influences decision making by firm executives (e.g., Shao et al., 2010), we choose to focus on one of the most well-studied indicators: individualism (e.g., Nguyen and Truong, 2013). In the spirit of Hofstede (1980, 2001), we argue that individualistic (collectivistic) cultures tolerate unilateral-based (consensusbased) decision making by firm executives and thus, will be characterized by more (less) managerial discretion. In this respect, executive discretion will vary upon the degree to which a society prefers autonomous vs. interdependent actions (e.g., Crossland and Hambrick, 2007, 2011). If individualism promotes suboptimal or optimal managerial decisions, is an issue under debate. Concerning investment, Matoussi and Jardak (2012) show that more individualistic countries exhibit stronger investor protection. At the same time, McLean et al. (2012) show that stronger investor protection promotes efficient investment. As result, in countries with high individualism, where investor protection is expected to be stronger, optimal investment by firm executives is more likely to be prevalent. The evidence from the existing literature on the presence of managerial discretion over earnings in individualistic countries is mixed. According to Han et al. (2010) individualism has a positive link with managers’ earnings discretion practices. In contrary, Callen et al. (2011) that it is negatively with earnings management. 3.2. Equity-market setting Regarding the equity-market setting, we use equity-market development and ownership concentration. The level of development in equity markets has clear implications for managerial discretion on investment. As the level of development increases, investors tend to be more sophisticated, and the information content of stock markets is expected to be higher. Put another way, in more developed equity markets the expected returns exhibit a stronger relation with underlying risk. As suggested by Watanabe et al. (2013), optimal investment by firm executives as a rational response to the reduction in the cost of capital is more likely to occur, when stock prices more accurately reflect all firms’ fundamental information. In a similar vein, Titman et al. (2013) argue that executives in more developed stock markets are more convenient to align their investment expenditures to changes in the discount rate. As result, in countries with higher equity-market development, optimal managerial discretion concerning investment is more likely to be prevalent. The impact of equity-market development on managerial decisions to engage in earnings manipulation is, a priori, ambiguous (see Degeorge et al., 2013). On the one hand, if financial development encourages analyst monitoring, then manipulation of earnings by firm executives is expected to be attenuated (Brown and Higgins, 2001, 2005). On the other hand, the importance of earnings for security pricing is higher in more developed equity markets and thus, executives can have additional motives to manipulate earnings upwards in order to meet analysts’ target price forecasts (Degeorge et al., 1999). Existing empirical evidence is more likely to support the former case: more developed capital markets exhibit lower levels of earnings management (Degeorge et al., 2013; Leuz et al., 2003). Ownership concentration refers to the amount of stock in publicly traded firms that are owned by few large-block shareholders vs. widely owned by many small investors. As suggested by La Porta et al. (1997) ownership concentration could be a Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

6

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

reflection of poor investor protection. Put another way, in countries with poor (strong) investor protection the ownership is more likely to be concentrated (dispersed). As a result, in countries with concentrated ownership, where investor protection is expected to be weaker, earnings manipulation and/or investment in value-destroying projects by executives is more likely to occur. Leuz et al. (2003), show that executives in countries with concentrated ownership structures engage more in earnings management. On the other hand, in countries with dispersed ownership, where shareholder protection is expected to be stronger, investment in value-enhancing projects by executives in response to reduction in discount rates is more likely to occur. As a result, optimal (suboptimal) managerial discretion is expected to be more prevalent in countries with low (high) concentration of share ownership. 3.3. Productivity and growth potential Productivity and growth potential is captured by competitiveness. A number of studies (e.g., Nickell, 1996; Griffith, 2001) show a strong relation between productivity and competition measures. In a competitive environment only firms that are efficiently managed can survive. Competition encourages managers to invest in value-enhancing projects and to avoid wasteful investments. Grullon and Michaely (2014) find that firm executives in a competitive environment are more likely to make decisions that serve their shareholder interests rather than their own interests. As a result, optimal managerial discretion concerning investment expenditures is expected to be more prevalent in highly competitive countries. The impact of competitiveness on managerial discretion over earnings is, a priori, ambiguous (see Markarian and Santalo, 2014). On the one hand, if investors and analysts have information on real firm output, competition might make harder the justification of discretionary accruals and thus, it can discourage earnings manipulation by executives (see Bagnoli and Watts, 2010; Kedia and Philippon, 2009). On the other hand, the importance of earnings for security pricing is higher in more competitive markets and thus, executives can have additional incentives to engage in earnings manipulation. Markarian and Santalo (2014) provides empirical evidence that supports both cases; competitiveness has a positive (negative) impact on managerial discretion over earnings when information is low (high). 3.4. Accounting regimes Regarding accounting regimes, we use the quality of accounting standards. La Porta et al. (2006) argue that higher (lower) quality of accounting standards indicate stronger (weaker) investor protection. The role of investor protection on managerial discretion has been already discussed in detail. As a result, optimal (suboptimal) managerial discretion is expected to be more prevalent in countries with better (poorer) accounting standards. To investigate the impact of country-level factors capturing managerial discretion on return predictability attributable to NOA in an international setting, we follow the novel approach of Pincus et al. (2007) in studying the global accrual anomaly. We construct an indicator associated with the significance of the occurrence of the negative relation between NOA and future returns in European countries.3 Then, we examine pair-wise correlations of this indicator with country-level factors capturing discretion. We also consider regressions of this indicator on each country-level factor. Further, we follow some recent studies that focus on conditional evidence within the U.S. capital market to examine the predictive power of several accounting figures (e.g., asset growth) for future stock returns. These studies investigate the effect of accounting figures during subperiods or in subsamples of stocks (e.g., Chan et al., 2008; Lam and Wei, 2011; Li and Zhang, 2010). In particular, we consider cross-sectional regressions of subsequent holding period raw returns on NOA, within groups of countries based on the level of each country-level factor. In a similar vein, we also consider regressions within groups based on the national level of managerial discretion over earnings. Again, in all regressions size and book to market ratio are included as additional control variables. In the final part of our work, we assess the performance of trading portfolios based on NOA by considering abnormal returns (along with transaction costs). We attempt to examine mispricing and risk aspects of the possible relation between NOA and future returns in Europe, as well as, the type of managerial discretion behind this relation. On the one hand, a mispricing-based consideration suggests suboptimal managerial discretion (i.e., earnings management and/or agency related overinvestment) as underling source of return predictability attributable to NOA. On the other hand, a risk-based consideration suggests optimal managerial discretion (i.e., optimal investment in response to discount rate changes) as underling source of return predictability attributable to NOA. 4. Data and variable measurement 4.1. Data Our sample consists of 16 major European countries (countries are listed in Table 1) and covers the period 1988–2009. We select non-financial common stocks that are listed on the major stock exchange in each country from both active and defunct research files of Datastream in order to avoid the survivorship bias. Closed-end funds, trusts, ADRs, REITs, units of 3

In Pincus et al. (2007) the respective indicator is associated with the significance of accrual overweighting in international equity markets.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

7

beneficial interest, other financial institutions and foreign firms are excluded from the sample.4 Note that, we also perform the data screenings procedures suggested by Ince and Porter (2006) for basic coding errors (see also Wu and Li, 2011). Specifically, we consider only domestic companies based on their home country information (GEOGC). We require for each firm, a type of instrument indicator (TYPE) equal to equity (EQ) and contain no words or phrases in its name (NAME) to ensure that it is not a common equity. In order to screen for coding errors in monthly stock returns, any return above 300% that is reversed within one month is treated as missing. In a similar vein, any return below 50% that is reversed within one month is treated as missing. Regarding delisted stocks, we exclude all the zero returns from the last observation to the first observation with nonzero return. Firm-year observations with missing market capitalization and no valid data to calculate net operating assets and annual one-year ahead raw and abnormal returns are excluded from the sample. All variables are expressed in US dollars.5 Further, each country is required to have at least 30 stocks in any year during the sample period, in order to ensure a reasonable number of firms for the subsequent portfolio and regression tests. All the above mentioned criteria yield a final sample with 62,610 firmyear observations (i.e., equivalent to 751,320 firm-month observations) with non-missing data on net operating assets, market capitalization and stock returns. The first two columns of Table 1 provide details about the final sample. As expected, France, Germany and the U.K. represent the largest part of the overall sample. U.K. is the largest stock market with about 32% of the total firm-year observations. France is the second largest and Germany is the third largest, accounting for about 14.5% and 12% of the total observations, respectively. Austria, Ireland and Portugal represent the smallest part of the overall sample, with each country accounting for about 1.5% of the total firm-year observations. Each of the remaining countries typically account for less than 6% of the total firm-year observations. 4.2. Firm-level variables Scaled net operating assets (NOA) are measured as the difference between operating assets (OA) and operating liabilities (OL), deflated by lagged total assets. OA are calculated as the residual amount from total assets after subtracting cash & cash equivalents (i.e., financial assets):

OAt ¼ TAt  CASHt

ð1Þ

where  TAt : Total assets at year t (Worldscope data item 02999).  CASHt : Cash and cash equivalents at year t (item 02001). OL are calculated as the residual amount from total assets after subtracting minority interest, preferred stock, total debt (i.e., financial liabilities) and total common equity:

OLt ¼ TAt  MINT t  TDt  OPSt

ð2Þ

where  MINT t : Minority interest at year t (item 03426).  TDt : Total debt at year t (item 03255).  OPSt : Ordinary and preferred shares at year t (item 03995). Thus, NOA are defined through the following expression6:

NOAt ¼ ðOAt  OLt Þ=TAt1

ð3Þ

Market capitalization (item 08001) is measured at the financial year-end. Size is the natural logarithm of market capitalization (SIZE). Book to market (BM) is the natural logarithm of the ratio of total book equity (W03501) to market capitalization at the financial year-end. In order to reduce the effect of extreme outliers, we follow the existing literature and winsorize (e.g., Doukakis and Papanastasopoulos, 2014) NOA, SIZE and BM at the top and bottom 1% of their distribution within each country. 4 According to these criteria, Luxembourgish excluded from the study, since the average percentage of foreign firms listed on Luxembourg stock exchange amounts up to 82% between 1995 and 2008. 5 All results remain qualitatively similar, if we instead use the local-currency converted firm-level variables for all countries. 6 Penman (2007, ch.9, pp. 302–306) show analytically how to distinguish between operating and financial asset/liability accounts in a firm’s balance sheet. However, some items are inherently difficult to classify as either operating or financing. Consider the cash account as an example: one has to decompose into working cash necessary for operations (i.e., operating item) and investments of excess cash over that required to meet liquidity demands (i.e., financial item). Firms often record cash and cash equivalents together, and hence, decomposition of cash into a working and an investing component could be a difficult exercise. Similar issues exist for other accounts (e.g., debt investments, short and long-term equity investments, leases, etc.), which pose a challenge to arrive at a more refined measurement of net operating assets. At the same time, databases do not provide sufficient data at the required level of detail to distinguish between operating and financial items. Recognizing these issues, we follow the common convention in the literature and define operating assets as the residual amount from total assets after subtracting cash & cash equivalents and operating liabilities as the residual amount from total assets after subtracting financial liabilities and equity (see also Hirshleifer et al., 2004).

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

8

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Monthly return data are obtained from Datastream (item RI), representing closing prices at the last trading day of the month. Based on Ince and Porter (2006) and McLean et al. (2009), we trim monthly returns at the top and bottom 1% of their distributions, within each country in order to eliminate extreme observations. Once we get firm-monthly returns, we calculate one-year ahead annual raw stock return (RET) using compounded 12-monthly buy-and-hold returns. Following, Titman et al. (2013) and Watanabe et al. (2013), the 12-month return accumulation period begins six months after the financial year-end.7 For the measurement of one-year ahead annual abnormal returns (ARET), we use a characteristic-based benchmark approach that controls for the effects associated with firm size and book to market (i.e., size & book to market adjusted returns). As argued by Fama and French (2008), these portfolio-adjusted average returns are similar to the alphas from the three-factor model of Fama and French (1993). In particular, six months after financial year-end, firms are first sorted into four portfolios (i.e., quartiles) by market capitalization (W08001) and in each of the resulted portfolios firms are further sorted into other four portfolios by the ratio of book equity (W03501) to market capitalization (W08001). Stocks are weighted both equally and according to their market capitalization within each of these 16 groups. The equal-weighted benchmarks are employed against equal-weighted portfolios, and the value-weighted benchmarks are employed against value-weighted portfolios. The one-year ahead abnormal annual return (ARET) for a firm is the difference between the one year ahead raw return (RET) and the matching return of the benchmark portfolio to which the firm belongs. If a firm delists during the period, then the last available return index before delisting is used to calculate the delisting return and the proceeds are reinvested into the benchmark portfolio. For the sake of brevity, we report results based on equally-weighted abnormal returns.8 Appendix A provides the definition of firm-level variables.

4.3. Country-level variables Regarding the possible generalization of return predictability attributable to NOA in an international setting, we consider a factor capturing the magnitude transaction costs at the country level. Regarding the role of managerial discretion behind this generalization, we consider factors that are associated with cultural environment, equity-market setting, productivity and growth potential and accounting structure. Further, we also consider an earnings management index at the national level. Data for these country-level factors are taken from various publicly available sources. Transaction Costs: We use an index of the level of transaction costs (TC) within a country, described in what follows. Transaction cost index (TC): is based on commissions, fees, and market impact costs. This index has been developed by Elkins-Sherry Co for the period 1996–1998. A higher value of the index indicates higher trading costs within a country. We consider this index under the assumption that transaction cost estimates typically do not change substantially and are still applicable to our analysis (see also Chan et al., 2005). In a similar vein, Chui et al. (2010) use this index to study the momentum effect on stock returns in an international setting. Data for the index are taken from Chan et al. (2005) and Chui et al. (2010). Cultural Environment: we use the individualism index (IDV) that is described in what follows. Individualism Index: comes from a cross-country psychological survey conducted by Geert Hofstede (see Hofstede, 1980, 2001), which contains results collected from 88,000 IBM employees in 72 countries between 1967 and 1973. It is calculated from the country mean scores on 14 questions about the employees’ attitudes towards their work and private lives and ranges from 0 to 100. A higher value of the index indicates a higher level of individualism. A criticism on the use of this index is that it may be outdated. However, Tang and Koveos (2008), claim that cultural indices, like the individualism index, provide information about a country’s relative position to other countries, which changes scarcely. Data for the index are available at the website of Hofstede (www.geert-hofstede.com). Equity Market Setting: we use two indices for equity-market development; the market capitalization to GDP ratio (MKT to GDP) and the easiness of access to equity market index (ACCESS). For ownership structure we use the ownership concentration ratio (OWCR). Those indices are calculated as follows. Market Capitalization to GDP Ratio: is calculated as the average of the ratio of stock market capitalization held by small shareholders to gross domestic product over the period 1989–2008. A higher value of the ratio indicates higher equity market importance. Data for the index are available at the website of World Bank (www.worldbank.org). Easiness of Access to Equity Market Index: is based on the extent to which business executives in a country agree with the statement ‘‘Stock markets are open to new firms and medium-sized firms.” Responses scale from 1 (strongly disagree) though 7 (strongly agree). A higher value of the index indicates greater easiness with which firms executives issue securities in local stock markets. The index continues to attract researchers’ interest in international studies (e.g., McLean et al., 2012). Data for the index are taken from La Porta et al. (2006). 7 In doing so, we assume that financial statement information related to NOA are not available to the market until this time. We claim that such an assumption is reasonable, since the filing deadline in the great majority of European countries in our sample is around 6 months after year end (see Pincus et al., 2007, Table 4, p. 184). Our results remain qualitatively similar, if we instead accumulate returns using windows that begin, e.g., one, two or three months after year end. 8 Results are qualitative similar for value-weighted abnormal returns.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

9

Ownership Concentration Ratio: is the median percentage of common shares owned by the three largest shareholders in the ten largest nonfinancial firms. A higher value of the ratio indicates higher ownership concentration. As in prior studies (e.g. La Porta et al., 2002), we assume that ownership structures are fairly stable over time. Data for the ratio are taken from La Porta et al. (2006). Productivity and Growth Potential: we use the competitiveness index (COMP) that is calculated as follows. Competitiveness Index: It is a weighted average of 12 components each of which contributes importantly to the complexity of a country’s degree of competitiveness: institutional environment, infrastructure, macroeconomic stability, health and primary education, higher education and training, goods market efficiency, labor market efficiency, financial market sophistication, technological readiness, market size, business sophistication and innovation. A higher value of the index indicates higher competitiveness. We use the index from the Global Competitiveness Report that covers the period 2006–2007. Data for the index are available at the website of World Economic Forum (www.weforum.org). Accounting Regimes: we use an index of the quality of accounting standards (ACCS) associated with disclosure that is calculated as follows. Accounting Standards Index: is the average number of 90 accounting and non-accounting items disclosed by a sample of large firms in their 1995 annual reports. A higher value of the index represents a higher quality of accounting standards. We need to stress here, that there is no cross-sectional variation in accounting standards across European countries for the post2004 period of our study, since all publicly-traded European firms adopt mandatorily IFRS. However, we claim that the selected accounting quality index could still capture some variation in executive discretion after 2004, since it also reflects voluntary disclosures that are strongly influenced by managerial incentives. Data for the index are given in Bushman et al. (2004). Managerial Discretion over Earnings: we use an earnings management index (EM) that is calculated as follows. Earnings Management Index: It is the percentage rank of two earnings smoothing metrics (reduced variability of reported earnings by altering accounting accruals and correlation between accounting accruals and operating cash flows) and two earnings discretion metrics (magnitude of accounting accruals relative to that of operating cash flows and small loss avoidance). A higher score of the index indicates higher earnings management. This index has been introduced by Leuz et al. (2003) and updated by Leuz (2010). Appendix B provides the definitions and the data for the country-level factors used in this study. 5. Empirical results In the first part, we present results on the possible generalization of the negative relation between NOA and future returns in European equity markets (Sections 5.1-5.3). In the second part, we present results on the role managerial discretion behind this possible generalization (Sections 5.4 and 5.5). We also report some additional results across both dimensions (Section 5.6). 5.1. Summary statistics on NOA Table 1 presents summary statistics on NOA (mean, standard deviation, 25th percentile, median, 75th percentile). The country-average mean and median value of NOA is equal to 0.647 and 0.623, respectively. The level of NOA is high in small economies like Greece and Portugal, where earnings management is high as documented by Leuz et al. (2003). The mean value of NOA in Greece and Portugal is 0.776 and 0.739, respectively. At the same time, the level of NOA is high in Norway and Ireland, where earnings management is likely to be low (see Leuz et al., 2003). The mean value of NOA in Norway and Ireland is 0.691 and 0.741, respectively. Surprisingly, the level of NOA appears to be high in countries with both low and high degree of managerial discretion over earnings. However, Norway is also generally perceived as a country that facilitates investment due to high institutional quality and economic development. Similarly, U.K the largest stock market in the sample has a level of NOA above the european countryaverage. In particular, it is equal to 0.658. According to Crossland and Hambrick (2011), U.K. has the highest score in general managerial discretion after U.S. Leuz et al. (2003) show a low degree of earnings manipulation by executives in U.K. Thus, greater managerial discretion in U.K. is not necessarily over earnings and it may be also associated with investment. The highest standard deviation of NOA is observed for the U.K. (0.501). The standard deviation of NOA for Germany (0.403) is higher than the country-average standard deviation (0.351), but almost equal to the standard deviation when all countries are considered as a union (0.408). The lowest standard deviation of NOA is observed for Switzerland. Overall, findings in Table 1 reveal a substantial variation of NOA across countries of the European Union plus Switzerland. 5.2. Cross-sectional regressions of future returns on NOA In Table 2, we report results from Fama and MacBeth (1973) regressions of one-year ahead raw returns (RET) on NOA after controlling for size (SIZE) and book to market (BM) as follows.

Model1 : RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

10

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Table 2 Regressions of future returns on NOA. Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 2

Country

Level of TC

Intercept

SIZE

BM

NOA

AdjR

Netherlands

24.5 27.1

Germany

30.6

Sweden

33

United Kingdom

34.1

France

35.7

Portugal

35.7

Switzerland

38.6

Spain

39.2

Denmark

40.8

Italy

41

Norway

41.5

Finland

45.2

Austria

53.2

Ireland

93.7

Greece

105.1

0.006 0.367 0.016** 0.011 0.014** 0.017 0.009 0.306 0.009* 0.098 0.012** 0.022 0.002 0.881 0.009* 0.075 0.010 0.391 0.006 0.528 0.016*** 0.004 0.006 0.505 0.003 0.690 0.013 0.218 0.016* 0.070 0.028 0.391 0.007** 0.026 0.006 0.103

0.021 0.321 0.026 0.111 0.036** 0.011 0.013 0.645 0.026* 0.067 0.040** 0.011 0.034 0.307 0.042** 0.034 0.046*** 0.005 0.053** 0.012 0.063*** 0.009 0.051 0.122 0.039 0.166 0.055** 0.021 0.032 0.171 0.003 0.855 0.036*** 0.000 0.043*** 0002

0.110*** 0.001 0.056*** 0.009 0.054** 0.046 0.103** 0.014 0.074*** 0.000 0.090*** 0.000 0.104 0.131 0.074** 0.019 0.093*** 0.000 0.051 0.196 0.039 0.116 0.103** 0.018 0.003 0.937 0.074 0.102 0.041 0.313 0.020 0.768 0.068*** 0.000 0.082*** 0.000

0.056

Belgium

0.111 0.256 0.062 0.456 0.073 0.296 0.077 0.559 0.025 0.749 0.015 0.845 0.152 0.280 0.056 0.463 0.054 0.780 0.080 0.588 0.136* 0.093 0.267** 0.047 0.104 0.343 0.017 0.907 0.050 0.721 0.417 0.388 0.064* 0.079 0.089 0.127

Country-Average All Countries

0.030 0.044 0.059 0.035 0.039 0.046 0.031 0.051 0.049 0.037 0.054 0.045 0.030 0.027 0.068 0.044 0.028

Notes: Table 2 presents results from Fama and MacBeth (1973) regressions of one-year ahead raw annual returns (RET) on net operating assets (NOA), after controlling for size (SIZE) and book to market ratio (BM). We estimate annual cross-sectional regressions and report the time-series averages of the parameter coefficients and adjusted R2 . We present separate coefficients for each country, averages of coefficients across countries and coefficients when countries are consider all-together. In a similar way, we present adjusted R2 . Countries are presented in an ascending order based on an index of trading costs (TC) within each country. Firm-level variables are defined in Appendix A. The index of trading costs (TC) is defined in Appendix B. *** Represents statistical significance at 1% level, respectively, two-tailed (p-values are reported in italics). ** Represents statistical significance at 5% level, respectively, two-tailed (p-values are reported in italics). * Represents statistical significance at 10% level, respectively, two-tailed (p-values are reported in italics).

We estimate annual cross-sectional regressions and report the time-series averages of the parameter coefficients and adjusted R2. We present separate coefficients for each country, averages of coefficients across countries and coefficients when countries are considered all-together. In a similar way, we present adjusted R2. Countries in Table 2, are presented in an ascending order based on the index of trading costs (TC) within each country. Coefficients on NOA are negative and statistically significant at the 1% level in Belgium, France, Netherlands, Spain and U. K. In Germany, Norway, Sweden and Switzerland are negative and statistically significant at the 5% level. Coefficients on NOA in Austria, Denmark, Finland, Greece, Ireland, Italy and Portugal are statistically indifferent from zero. Thus, the negative relation between NOA and stock returns significantly occurs in nine European equity markets, but does not occur in seven equity markets. Failure to detect return predictability attributable to NOA in some countries should not reflect a lack of power due to small sample sizes for those countries. Skipping the large capital markets (i.e., France, Germany and U.K.), one can see the presence of a significant relation between NOA and stock returns in Belgium and Spain, but not in Greece and Italy. Indeed, the sample in Belgium and Spain consists of fewer firm-year observations relative to Greece and Italy (see Table 1). Importantly, return predictability attributable to NOA is concentrated across countries with lower transaction costs. If one splits the sample into two equally-weighted groups based on the level of transaction costs, then it is observable that seven out of nine countries with a significant NOA-return relation are included in the low transaction cost group, while the remaining two are included in the high transaction cost group. This pattern is not consistent with a possibility of mispricing, which

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

11

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx Table 3 Regressions of future returns on NOA. Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 2

Country

Level of TC

Intercept

SIZE

BM

NOA

AdjR

Netherlands

24.5 27.1

Germany

30.6

Sweden

33

United Kingdom

34.1

France

35.7

Portugal

35.7

Switzerland

38.6

Spain

39.2

Denmark

40.8

Italy

41

Norway

41.5

Finland

45.2

Austria

53.2

Ireland

93.7

Greece

105.1

0.004 0.578 0.021** 0.013 0.008 0.274 0.011 0.209 0.009 0.131 0.007 0.256 0.014 0.328 0.010 0.118 0.001 0.812 0.009 0.354 0.012** 0.015 0.007 0.622 0.005 0.590 0.007 0.565 0.010 0.424 0.043 0.231 0.004 0.2541.315 0.005 0.199

0.075*** 0.009 0.097*** 0.000 0.099*** 0.000 0.093** 0.015 0.046** 0.021 0.072*** 0.001 0.092** 0.014 0.092*** 0.001 0.075*** 0.000 0.105*** 0.000 0.086** 0.013 0.108*** 0.003 0.114*** 0.003 0.107*** 0.000 0.046 0.254 0.065 0.457 0.086*** 0.000 0.072*** 0.000

0.126*** 0.000 0.109*** 0.000 0.165*** 0.000 0.149*** 0.000 0.107*** 0.000 0.123*** 0.000 0.088 0.226 0.105*** 0.009 0.171*** 0.000 0.048 0.448 0.089 0.109 0.167*** 0.001 0.064 0.325 0.080 0.259 0.122** 0.012 0.251*** 0.005 0.123*** 0.000 0.116*** 0.000

0.027

Belgium

0.171 0.176 0.057 0.566 0.115 0.194 0.129 0.361 0.059 0.490 0.112 0.169 0.017 0.907 0.094 0.315 0.239*** 0.003 0.066 0.675 0.048 0.505 0.356* 0.074 0.310** 0.042 0.071 0.713 0.096 0.558 0.821 0.105 0.159*** 0.008 0.135** 0.020

Country-Average All Countries

0.046 0.050 0.030 0.015 0.023 0.017 0.028 0.033 0.024 0.025 0.042 0.036 0.032 0.013 0.032 0.030 0.021

Notes: Table 3 presents results from Petersen (2009) regressions of one-year ahead raw annual returns (RET) on net operating assets (NOA), after controlling for size (SIZE) and book to market ratio (BM). In particular, we estimate regressions with the Ordinary Least Squares (OLS) approach clustered at firm and year level using the pooled sample over the whole time-period of our sample. We report separate coefficients for each country, averages of coefficients across countries and coefficients when countries are consider all-together. In a similar way, we report adjusted R2 . Countries are presented in an ascending order based on an index of trading costs (TC) within each country. *** Represents statistical significance at 1% level, respectively, two-tailed (p-values are reported in italics). ** Represents statistical significance at 5% level, respectively, two-tailed (p-values are reported in italics). * Represents statistical significance at 10% level, respectively, two-tailed (p-values are reported in italics).

put forward suboptimal managerial discretion to interpret return predictability attributable to NOA. At the same time, the findings suggest that the NOA-return relation occurs in the most advanced European economies, where optimal decisions concerning investment expenditures are more likely to be taken. To further examine the robustness of our results, we estimate cross sectional regressions using the clustered standard errors method of Petersen (2009). In particular, we estimate regressions with the Ordinary Least Squares (OLS) approach clustered at firm and year level using the pooled sample over the whole time-period of our sample. An advantage of the two-way clustering is that it uses all available information in one go (by data pooling), thus improving estimation precision and, in addition, computes standard errors by clustering observations based on their firm-year characteristics. As argued by Petersen (2009), this technique is preferred for estimating standard errors in finance applications using panel data. In Table 3, we report separate coefficients for each country, averages of coefficients across countries and coefficients when countries are consider all-together. In a similar way, we report adjusted R2. As one can see, regression results are qualitatively similar with those tabulated in Table 2. At the same time, the coefficients on NOA obtained using the Petersen (2009) approach, are in almost all cases more negative, relative to the respective coefficients obtained using the Fama and MacBeth (1973) approach. Overall, our findings in Tables 2 and 3 confirm the first hypothesis (H1) concerning the existence of the negative relation between NOA and stock returns in European equity markets. 9

9 In unreported tests, we find that the negative relation between NOA and stock returns in European equity markets continues for up to three years following the NOA measurement year, though the magnitude of the relation is attenuated in later years. The results are available from the authors upon request.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

12

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Table 4 Future returns of portfolios on NOA. Country

Level of TC

Lowest NOA Portfolio

Highest NOA Portfolio

Hedge (L-H) NOA Portfolio

Netherlands Belgium Germany Sweden United Kingdom France Portugal Switzerland Spain Denmark Italy Norway Finland Austria Ireland Greece Country-Average All Countries

24.5 27.1 30.6 33 34.1 35.7 35.7 38.6 39.2 40.8 41 41.5 45.2 53.2 93.7 105.1

0.125** 0.086* 0.070* 0.145* 0.075 0.138** 0.081 0.117** 0.125* 0.131** 0.029 0.156** 0.106* 0.096 0.135** 0.132 0.109*** 0.099**

0.059 0.031 0.015 0.054 0.017 0.060 0.070 0.056 0.062 0.075 0.008 0.049 0.131* 0.041 0.082 0.092 0.054*** 0.044

0.066** 0.055** 0.085** 0.091** 0.058*** 0.078*** 0.011 0.061** 0.063** 0.056* 0.021 0.107** 0.025 0.055 0.054 0.040 0.055*** 0.055***

Notes: Table 4 presents one-year ahead raw annual returns (RET) for country-specific portfolios, country-average portfolios and portfolios when countries are considered all-together. Countries are presented in an ascending order based on an index of trading costs (TC) within each country. Country-specific portfolios are formed as follows. Each year (six months after the financial year-end) firms are sorted on net operating assets (NOA) and allocated into five equal-sized portfolios (i.e., quintiles) based on these ranks. Then, we report time-series averages of one-year ahead raw annual returns for the lowest portfolio, the highest portfolio and the hedge (i.e., consisting of a long position in the lowest quintile and a short position in the highest quintile) portfolio. A ‘‘country-average” portfolio is formed as a portfolio that puts an equal weight on each country-specific portfolio. The ‘‘all-countries” portfolios are formed with the same procedure used for country-specific portfolios with firms from all countries (results are reported for lowest, highest and hedge NOA portfolio). *** Represents statistical significance at 1% level, respectively, two-tailed. ** Represents statistical significance at 5% level, respectively, two-tailed. * Represents statistical significance at 10% level, respectively, two-tailed.

5.3. Future returns of portfolios on NOA We next investigate the magnitude of the NOA effect on stock returns, in European equity markets. In Table 4 we present one-year ahead raw returns on NOA portfolios. Results are reported for country-specific portfolios, country-average portfolios and portfolios when countries are considered all-together. Country-specific portfolios are formed as follows. Each year (six months after the financial year-end) firms are sorted on NOA and allocated into five equal-sized portfolios (i.e., quintiles) based on these ranks. Then, we report time-series averages of one-year ahead raw returns for the lowest portfolio, the highest portfolio and the hedge (i.e., consisting of a long position in the lowest quintile and a short position in the highest quintile) portfolio. A ‘‘country-average” portfolio is formed as a portfolio that puts an equal weight on each country-specific portfolio. The ‘‘all-countries” portfolios are formed with the same procedure used for country-specific portfolios with firms from all countries. Countries in Table 4, are presented in an ascending order based on the index of trading costs (TC) within each country. The hedge raw return is significantly positive in ten out of sixteen countries with an average equal to 0.055. When all countries are considered together, the hedge return does not change. The largest hedge raw returns are observed in two Nordic countries. Norway has a hedge return of about 0.107, followed by Sweden with a return of about 0.091. For large stock markets, the hedge raw return for France and Germany is equal to 0.085 and 0.078, respectively. U.K has a lower return of about 0.058. For other countries, statistically significant hedge raw returns vary from 0.055 (Belgium) to 0.066 (Netherlands). Fig. 1 provides evidence on the stability of raw returns to the hedge portfolio formed on the magnitude of NOA, when countries of the European Union (prior to the 2004 enlargement, Luxembourg is excluded) plus Switzerland are considered all-together. Consequently, the average of the over the period 1988–2009 corresponds to the hedge return of 0.055 reported in the final cell of results in Table 6. The hedge portfolio is positive in 16 of the 21 years examined, illustrating that the relation is fairly stable over time. The evidence reported in Table 4 is entirely in accordance with regression results in Tables 2 and 3. The size of the sample within a country does not drive the magnitude and the significance of country-specific returns obtained from NOA hedge trading portfolios (i.e., compare for instance hedge returns and firm-year observations in Norway and Italy). NOA hedge returns are found positive and significant across countries with more developed economies and lower transaction costs. Overall, these findings confirm the first hypothesis (H1) and provide an economic summary of the negative relation between NOA and stock returns in Europe. Chan et al. (2008) point out, based on US data, that about half of the predictive power of NOA for future stock returns overlaps with that of asset growth rate (AGR hereafter, defined as annual percentage change in total assets). In follow up research, Papanastasopoulos et al. (2011) show in the US capital market, that across high (low) NOA firms only those that

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

13

Fig. 1. Time series of annual hedge raw returns of NOA portfolio. Notes: Fig. 1 plots the annual buy-and-hold raw return for the hedge trading strategy formed on the magnitude of NOA. As summarized in Table 4, a hedge trading strategy on NOA consists of a long position in the lowest NOA quintile portfolio and a short position in the highest NOA quintile portfolio.

have high (low) asset growth experience significantly negative (positive) stock returns in the future. Therefore, in order to investigate the possible relation of the predictive power of these balance sheet measures for future returns in European equity markets, we assess the stock price performance of quintile portfolios on NOA conditional on AGR and vise-versa (see Dechow et al., 2008). To implement a two-dimensional portfolio on NOA conditional on AGR, each year we first sort firms into five equal-sized portfolios based on AGR and subsequently sort them into five equal-sized sub-portfolios based on NOA. We then combine together all sub-portfolios on NOA of quintile rank 1, all sub-portfolios on NOA of quintile rank 2, e.t.c., and report oneyear ahead raw returns (t-statistics in italics) for each of these sub-quintiles. This technique allows substantial variation in NOA, while holding AGR relatively constant. Inversely, to implement a two-dimensional portfolio on AGR conditional on NOA, each year we first sort firms into five equal-sized portfolios based on NOA and subsequently sort them into five equal-sized sub-portfolios based on AGR. We then combine together all sub-portfolios on AGR of quintile rank 1, all sub-portfolios on AGR of quintile rank 2, e.t.c., and report one-year ahead raw returns (t-statistics in italics) for each of these sub-quintiles. This technique allows substantial variation in AGR, while holding NOA relatively constant. One year ahead raw returns from pure and two-dimensional portfolios on NOA and AGR for the European Union plus Switzerland are reported in Table 5.10 Panel A presents results for pure portfolios and Panel B for two-dimensional portfolios. The hedge return on NOA is equal to 0.055 and significant at the 1% level. The hedge return on AGR is 0.05 and significant at the 5% level. Conditional on AGR, the hedge return on NOA drops to 0.032 and remains significant at the 5% level. Conditional on NOA, the hedge return on AGR drops to 0.034 and is no longer statistically significant at conventional levels. Thus, after controlling for NOA, AGR has no significant ability to explain future returns In summary, our conditional firm-level portfolio analysis, in accordance with Chan et al. (2008) and Papanastasopoulos et al. (2011), suggests that the effect of the level of balance sheets on stock returns is related with the respective effect of growth in balance sheets.

5.4. NOA, future returns and country-level factors: summary statistics Table 6 presents summary statistics about an indicator associated with the significance of return predictability attributable to NOA in Europe and country-level factors associated with executive discretion. The indicator, labelled as NOAR, equals 1 for European countries that we find a significant negative relation between NOA and raw returns from crosssectional regression tests in the first part of our analysis (reported in Tables 2 and 3) and 0 otherwise. Country-level factors capturing discretion are: the individualism index (IDV), the market capitalization to GDP ratio (MKT to GDP), the easiness of access to equity market index (ACCESS), the ownership concentration ratio (OWCR), the competitiveness index (COMP) and the accounting standards index (ACCS). Panel A presents univariate statistics (mean, standard deviation, median) on NOAR and factors capturing discretion at the national level, while Appendix B provides the data on these factors. Starting with individualism index, as expected from prior studies (see Hofstede, 2001 for a review), European countries with the highest scores are U.K. (89) and Netherlands (80), while countries with the lowest scores are Portugal (27) and Greece (35). The index has a mean value around 65, a high standard deviation around 16 and a median value near 70. Not surprisingly, based on both values of MKT to GDP and ACCESS, Finland, Sweden, Switzerland, and U.K. constitute the most developed European equity markets. The mean value and median value of MKT to GDP is equal to 0.672 and 0.575, 10 Due to the independent nature of sorts, we present results for European stock markets only as a union and not in isolation as safeguard that portfolios are sufficiently diversified to yield reliable inferences.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

14

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Table 5 Balance sheet level vs. balance sheet growth and stock return predictability in the European equity markets. Panel A, RET of Portfolios on NOA and AGR Portfolio Rank

NOA

AGR

Lowest 2nd 3rd 4th Highest Hedge (L-H)

0.099** 0.094* 0.093* 0.087* 0.044 0.055***

0.079* 0.113** 0.105** 0.090* 0.029 0.050**

Panel B, RET of Portfolios on NOA, after controlling for AGR and vice-versa Portfolio Rank

NOA conditional on AGR

AGR conditional on NOA

**

Lowest 2nd 3rd 4th Highest Hedge (L-H)

0.082* 0.098** 0.101** 0.086* 0.048 0.034

0.100 0.080* 0.088* 0.080* 0.068 0.032**

Notes: Table 5 presents one-year ahead raw annual stock returns (RET) for portfolios based on net operating assets (NOA) and asset growth rate (AGR), when firms from European countries are pooled all-together. In Panel A, we report RET for portfolios on each measure in isolation. In Panel B, we report RET for portfolios on NOA, after controlling for AGR and vice-versa. To implement a two-dimensional portfolio on NOA conditional on AGR, each year we first sort firms into five equal-sized portfolios (i.e., quintiles) based on AGR and subsequently sort them into five equal-sized sub-portfolios based on NOA. We then combine together all sub-portfolios on NOA of quintile rank 1, all sub-portfolios on NOA of quintile rank 2, e.t.c., and report RET for each of these subquintiles. To implement a two-dimensional portfolio on AGR conditional on NOA, each year we first sort firms into five equal-sized portfolios based on NOA and subsequently sort them into five equal-sized sub-portfolios based on AGR. We then combine together all sub-portfolios on AGR of quintile rank 1, all sub-portfolios on AGR of quintile rank 2, e.t.c., and report RET for each of these sub-quintiles. *** Represents statistical significance at 1% level, respectively, two-tailed. ** Represents statistical significance at 5% level, respectively, two-tailed. * Represents statistical significance at 10% level, respectively, two-tailed.

Table 6 NOA, future returns and country-level factors: data & summary statistics. NOAR

IDV

Panel A: Univariate Statistics on Country-Level Factors Mean 0.563 65.063 St. Dev. 0.512 16.068 Median 1 69.500

MKTtoGDP

ACCESS

OWCR

COMP

ACCS

0.672 0.415 0.575

5.598 0.635 5.725

0.441 0.124 0.430

5.112 0.460 5.195

72.880 8.778 74.500

0.527** 0.571** 0.647*** –

0.495* 0.617** 0.562** 0.675*** –

0.431* 0.659*** 0.526** 0.807*** 0.729*** –

0.387 0.686*** 0.667*** 0.726*** 0.861*** 0.711*** –

Panel B: Pearson Correlations among Country-Level Factors NOAR – 0.449* 0.445* IDV – 0.399 MKTtoGDP – ACCESS OWCR COMP ACCS

Notes: Table 6 reports descriptive statistics (mean, median standard deviation, pair-wise correlations) on selected country-level factors capturing managerial discretion. Panel A presents data with univariate statistics, while Panel B Pearson pair-wise correlations. NOAR is 1 for countries that we find a significant negative relation between net operating assets and one-year ahead raw annual returns, from the regression tests reported in Tables 2 and 3, 0 otherwise. Country-level variables are defined in Appendix B. *** Represents statistical significance at 1% level, respectively (two-tailed) for correlation statistics. ** Represents statistical significance at 5% level, respectively (two-tailed) for correlation statistics. * Represents statistical significance at 10% level, respectively (two-tailed) for correlation statistics.

respectively. The mean value and median value of ACCESS is equal to 5.598 and 5.725, respectively. ACCESS is a more volatile measure than MKT to GDP. U.K. (Greece) has the lowest (highest) concentration of share ownership. The average of OWCR is 0.441 and the standard deviation is 0.124. U.K. and Switzerland (Greece and Portugal) are the most (least) competitive countries in Europe. COMP has a mean/median value around 5.1 and standard deviation equal to 0.46. Consistent with the existing literature (e.g., Watanabe et al., 2013), the quality of accounting information is the best in U. K. (ACCS = 85) and Sweden (ACCS = 83), while it is the worst in Greece (ACCS = 61) and Austria (ACCS = 62). ACCS has a mean equal to 72.88, a high standard deviation equal to 8.778 and a median equal to 74.5. Overall, results in Panel A reveal significant variation across the selected country-level factors capturing discretion.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

15

Panel B presents Pearson pair-wise correlations. Notably, NOAR is significantly correlated with all selected country-level factors capturing executive discretion, except the accounting standards index. In particular, NOAR is significantly and positively correlated with IDV (q = 0.449), MKT to GDP (q = 0.445), ACCESS (q = 0.527), and COMP (q = 0.431), while is significantly and negatively correlated with OWCR (q = 0.495). IDV has a significantly positive correlation with ACCESS and COMP. Although there is a positive relation between IDV and MKT to GDP, it is not significant. At the same time, IDV has significantly positive (negative) correlation with ACCS (OWCR). Thus, as expected, individualistic countries are characterized by higher equity-market development (see Titman et al., 2013), higher competitiveness and stronger investor protection (see Matoussi and Jardak, 2012). Measures of equity-market development (MKT to GDP, ACCESS) are positively correlated (q = 0.571). Both measures have a strong positive correlation with COMP and ACCS and a strong negative correlation with OWCR. Not surprisingly, investor protection is stronger and competitiveness is higher in more developed equity markets. OWCR is highly and negatively correlated with ACCS (q = 0.861). Recall that low OWCR and high ACCS are suggestive of stronger investor protection.

5.5. NOA, future returns and country-level factors: regression analysis In this subsection, we present results from regression analysis about the role of country-level factor of managerial discretion on the negative relation of NOA with future returns. Table 7 presents statistics of regressions of NOAR on each countrylevel characteristic in isolation, as follows.

Model2 : NOARc ¼ c0 þ c1 Factor c þ tc Regressions are estimated by the binary probit method with robust Huber-White standard errors & covariance.11 We find a positive coefficient on IDV that is statistically significant at the 5% level. Further, we find positive coefficients on MKT to GDP and ACCESS that are statistically significant at the 10% and 5% level, respectively. The coefficient on OWCR is negative and statistically significant at the 5% level. The coefficients on COMP and ACCS are positive and statistically insignificant at the 10% level. Thus, the negative relation of NOA with subsequent returns is more likely to occur in countries with higher individualism, higher equity-market development, lower concentration of share ownership, higher competitiveness and better quality of accounting standards. Overall, our evidence in Table 5 strongly supports the second hypothesis (H2) concerning the mediating role of managerial discretion on the occurrence of the negative relation between NOA and stock returns in European equity markets. 12 The dispersion on return predictability attributable to NOA (see Table 2 and 3) and on the selected country-level factors associated with managerial discretion (Table 4), enable us to investigate the negative relation of NOA with subsequent returns within groups of European countries formed on the magnitude of these factors. In order to allow sufficient variation across country-level factors of managerial discretion within the resulted groups, we classify countries into three groups; the lowest group with the bottom 25% of country-observations (i.e., four countries) based on the magnitude of each factor, the middle group with the next 50% of country-observations (i.e., eight countries) based on the magnitude of each factor, and the highest group with the top 25% of country - observations (i.e., four countries) based on the magnitude of each factor.13 Then, within each of the resulted group, we estimate Fama and MacBeth (1973) regressions of one-year ahead raw returns (RET) on NOA after controlling for size (SIZE) and book to market (BM) and report the time-series averages of the parameter coefficients along with their associated p-values in italics (see Model 1) and timeseries averages of adjusted R2. An advantage of this approach is that inferences are based on all available firm-year observations of our sample. In Table 8 we report regression results within group of countries based on culture. The coefficient on NOA in the lowest IDV group is statistically insignificant (p = 0.952). Thus, there is no significant relation between NOA and future returns across countries with the lowest individualism. The coefficient on NOA in the middle group and highest group is 0.084 (p = 0.000) and 0.075 (p = 0.000), respectively. The difference in the magnitude of the NOA coefficient between the highest and the lowest groups is equal to 0.071. Thus, in more individualistic countries, the negative relation of NOA with future returns becomes stronger in terms of both magnitude and significance. Higher individualism is more likely to be associated with optimal discretion concerning investment by managers. Table 9 presents regression results within groups of countries based on stock-market setting. Panel A reports results based on MKT to GDP. In the lowest group the coefficient on NOA is equal to 0.051 (p = 0.071). In the middle group the coefficient on NOA is equal to 0.1 (p = 0.002), while in the highest group is equal to 0.082 (p = 0.000). The spread in the magnitude of the NOA coefficient between the extreme groups equals 0.031. Similar results are reported in Panel C where countries are classified into groups based on ACCESS. Thus, the slope coefficient of NOA becomes lower in magnitude and significance in the subsample of countries with the lowest equity-market development. At the same time, the associa11 We avoid estimating the full model due to few observations (i.e., 16 countries) and possible multicollinearity as suggested by reported pair-wise correlations between country-level factors. 12 Our results remain qualitatively similar, if we instead estimate regressions by the binary logit method. 13 Our results remain qualitatively similar, if we instead use alternative cuts of data (e.g., quintiles or terciles by dropping the resulted middle groups, high group versus low group, etc.).

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

16

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Table 7 NOA, future returns and country-level factors: regression analysis. Model (2): NOARc ¼ c0 þ c1 Factor c þ tc Intercept IDV

10.753** 0.041 2.623** 0.037

1.303 0.106

6.981** 0.040

3.189** 0.015

6.914* 0.057

2.402* 0.078

MKTtoGDP

1.273** 0.040

ACCESS

6.762** 0.014

OWCR

1.380* 0.054

COMP ACCS McFadden  R2 n

19.864* 0.079

0.172 16

0.195 16

0.221 16

0.208 16

0.144 16

4.674* 0.079 0.126 16

Notes: Table 7 reports regression results about the impact of selected country-level factors capturing managerial discretion (IDV, MKTtoGDP, ACCESS, OWCR, COMP, ACCS) on the possible occurrence of a significant relation of NOA with future stock returns (NOAR). Regressions are estimated by the binary probit method with robust Huber-White standard errors & covariance. IDV and ACCS are entered in the regressions by taking the natural logarithm of their actual values. McFadden  R2 and n denote the McFadden R2 ratio and the number of country - observations, respectively. ⁄⁄⁄ Represents statistical significance at 1% level, respectively (p-values are reported in italics). ** Represents statistical significance at 5% level, respectively (p-values are reported in italics). * Represents statistical significance at 10% level, respectively (p-values are reported in italics).

Table 8 Regressions of future returns on NOA, conditional on culture. Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 Groups on IDV

Intercept

SIZE

BM

0.154 0.003 0.035 0.455 0.858 0.135 Middle Group (n = 8 countries, N = 29,361 firm-years) 0.057 0.009** 0.046*** 0.393 0.040 0.001 Highest Group (n = 4 countries, N = 26,959 firm-years) 0.039 0.008* 0.029** 0.562 0.063 0.033 Difference in the magnitude of the NOA coefficient between the highest and the lowest groups: 0.071

Lowest Group (n = 4 countries, N = 6290 firm-years)

2

NOA

AdjR

0.004 0.952 0.084*** 0.000 0.075*** 0.000

0.071 0.033 0.031

Notes: Table 8 presents results from Fama and MacBeth (1973) regressions of one-year ahead raw annual returns (RET) on net operating assets (NOA), within groups of countries based on individualism index (IDV). In all regressions, size (SIZE) and book to market ratio (BM) are included as additional control variables. The group formation procedure is as follows. Countries are classified based on the magnitude of IDV into three groups, the lowest group with the bottom 25% of observations, the middle group with the next 50% of observations and the highest group with the top 25% of observations. Within each group, we estimate annual cross-sectional regressions and report the time-series averages of the parameter coefficients and adjusted R2 . n denotes the number of country-observations in each group, while N the number of firm-year observation in each group. The last row provides a direct comparison of the magnitude of the NOA coefficient between the highest and the lowest groups. *** Represents statistical significance at 1% level, respectively, two-tailed (p-values are reported in italics). ** Represents statistical significance at 5% level, respectively, two-tailed (p-values are reported in italics). * Represents statistical significance at 10% level, respectively, two-tailed (p-values are reported in italics).

tion between NOA and subsequent returns is more severe in more developed equity markets. Higher equity-market development captures either managers’ willingness or ability to align optimally investment expenditures to the discount rate. Panel C reports results based on OWCR. The highest group has a statistically insignificant coefficient on NOA (p = 0.848). The coefficient on NOA in the middle group and lowest group is equal to 0.077 (p = 0.001) and 0.095 (p = 0.000), respectively. The difference in the magnitude of the NOA coefficient between the highest and the lowest groups is equal to 0.107. Thus, there is no significant association between NOA and subsequent returns across countries with highest ownership concentration ratio. At the same time, the negative relation between NOA and future returns is more pronounced as concentration of share ownership decreases. Ownership dispersion is more likely to be associated with optimal discretion by firm executives. Table 10 presents regression results within groups of countries based on competitiveness index. Within the lowest COMP group, the coefficient on NOA is statistically indifferent from zero (p = 0.979). Thus, there is no significant relation between NOA and future returns across countries with the lowest competitiveness. The coefficient on NOA in the middle group is 0.087 (p = 0.000) and in the highest group is 0.079 (p = 0.000). The spread in the magnitude of the NOA coefficient between the extreme groups equals 0.08. Thus, in more competitive markets, the negative slope of NOA coefficient

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

17

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx Table 9 Regressions of future returns on NOA, conditional on equity market setting. Panel A, Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 Groups on MKTtoGDP

Intercept

SIZE

BM

0.060 0.013** 0.045*** 0.320 0.010 0.007 Middle Group (n = 8 countries, N = 23,238 firm-years) 0.151** 0.003 0.047*** 0.047 0.458 0.005 Highest Group (n = 4 countries, N = 26,912 firm-years) 0.031 0.010** 0.034** 0.659 0.027 0.019 Difference in the magnitude of the NOA coefficient between the highest and the lowest groups: 0.031

Lowest Group (n = 4 countries, N = 12,460 firm-years)

2

NOA

AdjR

0.051* 0.071 0.100*** 0.002 0.082*** 0.000

0.042

NOA

AdjR

0.038 0.035

Panel B, Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 Groups on ACCESS

Intercept

SIZE

BM ***

***

0.041 0.012 0.065 0.547 0.001 0.000 Middle Group (n = 8 countries, N = 28,300 firm-years) 0.084 0.006 0.053*** 0.163 0.146 0.000 Highest Group (n = 4 countries, N = 27,786 firm-years) 0.049 0.009* 0.032** 0.492 0.053 0.035 Difference in the magnitude of the NOA coefficient between the highest and the lowest groups: 0.043

Lowest Group (n = 4 countries, N = 6524 firm-years)

2

**

0.045 0.031 0.073** 0.014 0.088*** 0.000

0.025

NOA

AdjR

0.035 0.034

Panel C, Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 Groups on OWCR

Intercept

SIZE

BM **

Lowest Group (n = 4 countries, N = 34,933 firm-years)

0.058 0.009 0.399 0.045 Middle Group (n = 8 countries, N = 19,584 firm-years) 0.052 0.009** 0.384 0.035 Highest Group (n = 4 countries, N = 8093 firm-years) 0.151 0.006 0.449 0.738 Difference in the magnitude of the NOA coefficient between the highest and the lowest groups: 0.107

**

0.034 0.020 0.051** * 0.000 0.041** 0.020

2

***

0.095 0.000 0.077*** 0.001 0.012 0.848

0.032 0.034 0.032

Notes: Table 9 presents results from Fama and MacBeth (1973) regressions of one-year ahead raw annual returns (RET) on net operating assets (NOA), within groups of countries based on equity market development index (MKTtoGDP), access to equity market index (ACCESS) and ownership concentration ratio (OWCR). In all regressions, size (SIZE) and book to market ratio (BM) are included as additional control variables. The group formation procedure is as follows. Countries are classified based on the magnitude of each index (MKTtoGDP, ACCESS and OWCR) into three groups, the lowest group with the bottom 25% of observations, the middle group with the next 50% of observations and the highest group with the top 25% of observations. Within each group, we estimate annual cross-sectional regressions and report the time-series averages of the parameter coefficients and adjusted R2 . n denotes the number of country-observations in each group, while N the number of firm-year observation in each group. Panel A reports regression results based on equity market development index (MKTtoGDP), Panel B based on access to equity market index (ACCESS), while Panel C based on ownership concentration ratio (OWCR). The last row in each Panel provides a direct comparison of the magnitude of the NOA coefficient between the highest and the lowest groups. *** Represents statistical significance at 1% level, respectively, two-tailed (p-values are reported in italics). ** Represents statistical significance at 5% level, respectively, two-tailed (p-values are reported in italics). * Represents statistical significance at 10% level, respectively, two-tailed (p-values are reported in italics).

Table 10 Regressions of future returns on NOA, conditional on productivity. Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 Groups on COMP

Intercept

SIZE

BM

0.141 0.004 0.039* 0.480 0.813 0.086 Middle Group (n = 8 countries, N = 27,837 firm-years) 0.044 0.009** 0.043*** 0.494 0.024 0.002 Highest Group (n = 4 countries, N = 26,554 firm-years) 0.027 0.011** 0.035** 0.710 0.039 0.019 Difference in the magnitude of the NOA coefficient between the highest and the lowest groups: 0.080

Lowest Group (n = 4 countries, N = 8219 firm-years)

2

NOA

AdjR

0.001 0.979 0.087*** 0.000 0.079*** 0.000

0.065 0.033 0.036

Notes: Table 10 presents results from Fama and MacBeth (1973) regressions of one-year ahead raw annual returns (RET) on net operating assets (NOA), within groups of countries based on competitiveness index (COMP). In all regressions, size (SIZE) and book to market ratio (BM) are included as additional control variables. The group formation procedure is as follows. Countries are classified based on the magnitude of COMP into three groups, the lowest group with the bottom 25% of observations, the middle group with the next 50% of observations and the highest group with the top 25% of observations. Within each group, we estimate annual cross-sectional regressions and report the time-series averages of the parameter coefficients and adjusted R2 . n denotes the number of country-observations in each group, while N the number of firm-year observation in each group. The last row provides a direct comparison of the magnitude of the NOA coefficient between the highest and the lowest groups. *** Represents statistical significance at 1% level, respectively, two-tailed (p-values are reported in italics). ** Represents statistical significance at 5% level, respectively, two-tailed (p-values are reported in italics). * Represents statistical significance at 10% level, respectively, two-tailed (p-values are reported in italics).

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

18

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Table 11 Regressions of Future Returns on NOA, conditional on Accounting Regimes. Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 Groups on ACCS

Intercept

SIZE

BM

0.145 0.007 0.036 0.423 0.674 0.135 Middle Group (n = 8 countries, 28,856 N = firm-years) 0.050 0.009** 0.047*** 0.403 0.024 0.000 Highest Group (n = 4 countries, N = 26,173 firm-years) 0.042 0.010* 0.031** 0.582 0.070 0.045 Difference in the magnitude of the NOA coefficient between the highest and the lowest groups: 0.115

Lowest Group (n = 4 countries, N = 7581 firm-years)

2

NOA

AdjR

0.028 0.697 0.087*** 0.000 0.087*** 0.000

0.069 0.034 0.035

Notes: Table 11 presents results from Fama and MacBeth (1973) regressions of one-year ahead raw annual returns (RET) on net operating assets (NOA), within groups of countries based on accounting standards index (ACCS). In all regressions, size (SIZE) and book to market ratio (BM) are included as additional control variables. The group formation procedure is as follows. Countries are classified based on the magnitude of ACCS into three groups, the lowest group with the bottom 25% of observations, the middle group with the next 50% of observations and the highest group with the top 25% of observations. Within each group, we estimate annual cross-sectional regressions and report the time-series averages of the parameter coefficients and adjusted R2 . n denotes the number of country-observations in each group, while N the number of firm-year observation in each group. The last row provides a direct comparison of the magnitude of the NOA coefficient between the highest and the lowest groups. *** Represents statistical significance at 1% level, respectively, two-tailed (p-values are reported in italics). ** Represents statistical significance at 5% level, respectively, two-tailed (p-values are reported in italics). * Represents statistical significance at 10% level, respectively, two-tailed (p-values are reported in italics).

becomes higher in magnitude and more significant. Higher competitiveness is linked with value-enhancing investment by firm executives. Table 11 presents regression results within groups of countries based on the accounting standards index. Within the lowest ACCS group, the coefficient on NOA is statistically indistinguishable from zero (p = 0.697). Thus, there is no significant relation between NOA and future returns across countries with the lowest quality of accounting standards. The coefficient on NOA in the middle group and highest group is equal to 0.087 (p = 0.000) and 0.087 (p = 0.000), respectively. The spread in the magnitude of the NOA coefficient between the extreme groups equals 0.115. Thus, in countries with better quality of accounting standards, the negative relation of NOA and subsequent returns is stronger and more significant.14 Higher quality of accounting standards is more likely to be associated with optimal managerial discretion. Further, we conduct regression analysis within group of countries to investigate the impact of the national level of managerial discretion over earnings on the relation of NOA and future returns. Leuz et al. (2003) introduced this index, while Leuz (2010) updated the index for the period 1990–1999 and for the period 1996–2005 (see p. 253). A closer look at the index reveals that there is significant variation in managerial discretion over earnings within Europe. We take an average value of the index between the two periods in order to estimate the regressions. Results are based on 15 countries, since Leuz (2010) does not report values of the earnings management index for Norway. We classify countries into three groups; the lowest (highest) group consists of four countries with the bottom (top) values on the earnings management index, while the remaining seven countries are allocated in the middle group. As one can see from Table 12, the negative relation between NOA and stock returns is stronger in countries with lower level of earnings management, while it becomes insignificant in countries with higher level of earnings management. Overall, findings in Tables 8–11 are in accordance with those in Table 7 and confirm our second hypothesis (H2) on the importance of managerial discretion behind return predictability attributable to NOA in European equity markets. Regarding the type of managerial discretion, earnings management and/or overinvestment appear not to be underlying driving forces. In contrary, optimal investment by firm executives seems to be the most consistent driving force as such strategic managerial behavior is more prevalent in countries with higher individualism, higher equity-market development, higher ownership dispersion, higher competitiveness and higher quality of accounting standards.

5.6. Future abnormal returns of portfolios on NOA In Table 13, we present one-year ahead abnormal returns for portfolios formed on the level of NOA. Results are reported for country-specific portfolios, country-average portfolios and portfolios when countries are considered all-together. The portfolio formation procedure is described earlier in Section 5.3. Countries in Table 13, are presented in an ascending order based on the index of trading costs (TC) within each country. The country-average hedge returns is equal to 0.026 and statistical significant at the 1% level. When all countries are considered together, its magnitude drops at 0.023 and significance at the 10% level. Comparing these results with those in Table 4, it is notable that once returns are adjusted for size and book to market, the magnitude of the NOA effect on stock returns in European equity markets reduces by more than 50%. 14

Similar results are found for a period up to 2004 when European countries mandatorily adopt IFRS (i.e., 1988–2004).

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

19

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx Table 12 Regressions of future returns on NOA, conditional on earnings management. Model (1): RET tþ1 ¼ c0 þ c1 SIZEt þ c2 BM þ c3 NOAt þ ttþ1 Groups on EM

Intercept

SIZE

0.042 0.010* 0.582 0.070 Middle Group (n = 7 countries, N = 26,809 firm-years) 0.033 0.010** 0.582 0.016 Highest Group (n = 4 countries, N = 7581 firm-years) 0.145 0.007 0.423 0.674 Difference in the magnitude of the NOA coefficient between the highest and the lowest groups:0.115

Lowest Group (n = 4 countries, N = 26,173 firm-years)

2

BM

NOA

AdjR

0.031** 0.045 0.047*** 0.000 0.036 0.135

0.087*** 0.000 0.083*** 0.000 0.028 0.697

0.035 0.034 0.069

Notes: Table 12 presents results from Fama and MacBeth (1973) regressions of one-year ahead raw annual returns (RET) on net operating assets (NOA), within groups of countries based on an earnings management index (EM). In all regressions, size (SIZE) and book to market ratio (BM) are included as additional control variables. The group formation procedure is as follows. Countries are classified based on the magnitude of EM into three groups; the lowest group consists of four countries with the bottom values on the earnings management index (EM), the highest group consists of four countries with the top values on the earnings management index (EM), while the remaining seven countries are allocated in the middle group (i.e., Norway is not included in these tests due to lack of data on the index). Within each group, we estimate annual cross-sectional regressions and report the time-series averages of the parameter coefficients and adjusted R2 . n denotes the number of country-observations in each group, while N the number of firm-year observation in each group. The last row provides a direct comparison of the magnitude of the NOA coefficient between the highest and the lowest groups. *** Represents statistical significance at 1% level, respectively, two-tailed (p-values are reported in italics). ** Represents statistical significance at 5% level, respectively, two-tailed (p-values are reported in italics). * Represents statistical significance at 10% level, respectively, two-tailed (p-values are reported in italics).

Table 13 Future abnormal returns of portfolios on NOA. Country

Level of TC

Lowest NOA Portfolio

Highest NOA Portfolio

Hedge (L-H) NOA Portfolio

Netherlands Belgium Germany Sweden United Kingdom France Portugal Switzerland Spain Denmark Italy Norway Finland Austria Ireland Greece Country-Average All Countries

24.5 27.1 30.6 33 34.1 35.7 35.7 38.6 39.2 40.8 41 41.5 45.2 53.2 93.7 105.1

0.007 0.006 0.029 0.018 0.052** 0.001 0.035 0.011 0.003 0.009 0.017 0.042 0.041 0.010 0.005 0.029 0.016*** 0.024*

0.043*** 0.053** 0.067*** 0.063** 0.058*** 0.048*** 0.010 0.057*** 0.039*** 0.040* 0.045** 0.055** 0.004 0.031 0.014 0.046 0.042*** 0.047***

0.050** 0.059** 0.038 0.045 0.006 0.049** 0.025 0.046 0.042 0.031 0.028 0.013 0.037 0.041 0.019 0.017 0.026*** 0.023*

Notes: Table 13 presents one-year ahead abnormal annual returns (ARET) for country-specific portfolios, country-average portfolios and portfolios when countries are considered all-together. Countries are presented in an ascending order based on an index of trading costs (TC) within each country. Countryspecific portfolios are formed as follows. Each year (six months after the financial year-end) firms are sorted on net operating assets (NOA) and allocated into five equal-sized portfolios (i.e., quintiles) based on these ranks. Then, we report time-series averages of one-year ahead abnormal annual returns for the lowest portfolio, the highest portfolio and the hedge (i.e., consisting of a long position in the lowest quintile and a short position in the highest quintile) portfolio. A ‘‘country-average” portfolio is formed as a portfolio that puts an equal weight on each country-specific portfolio. The ‘‘all-countries” portfolios are formed with the same procedure used for country-specific portfolios with firms from all countries (results are reported for lowest, highest and hedge NOA portfolio). *** Represents statistical significance at 1% level, respectively, two-tailed. ** Represents statistical significance at 5% level, respectively, two-tailed. * Represents statistical significance at 10% level, respectively, two-tailed.

Further, results reveal that abnormal returns are significant in only three European countries: Netherlands, Belgium and France. Thus, when we consider abnormal returns, the return performance of NOA portfolios is substantially reduced to insignificant levels in other six European equity markets, where the NOA-return relation is found to be present: Germany, Norway, Spain, Sweden, Switzerland and U.K. Hedge abnormal returns range from about 0.05 for Netherlands and France up to 0.059 for Belgium. For Netherlands and France the magnitude of hedge abnormal returns is lower relative to the respective hedge raw returns, but for Belgium is somewhat higher. Netherlands and Belgium are the European countries with the lowest transaction costs, while France is ranked in the sixth position out of 16 countries. Thus, again, significant abnormal returns from NOA hedge trading strategies are concentrated in countries with lower transaction costs.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

20

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Fig. 2. Time series of annual hedge abnormal returns of NOA portfolio. Notes: Fig. 2 plots the annual buy-and-hold abnormal return for the hedge trading strategy formed on the magnitude of NOA. As summarized in Table 4, a hedge trading strategy on NOA consists of a long position in the lowest NOA quintile portfolio and a short position in the highest NOA quintile portfolio.

Fig. 2 graphs the hedge abnormal returns from the NOA trading portfolio broken down by year, with firms from all European countries. The strategy is now profitable in 13 out of 21 years examined. Overall, our findings in Table 13 suggest the return predictability attributable to NOA in Europe, does not seem to constitute a systematic mispricing pattern. Interpreting them together with prior evidence discussed earlier, they put forward optimal investment decisions by executives as a possible driving force of the NOA-return relation. 6. Conclusions Hirshleifer et al. (2004) document a negative relation between NOA and stock returns in the U.S. capital stock market and claim that the interpretation of this relation accommodates, but does not require earnings management. Papanastasopoulos et al. (2011) show that return predictability associated with NOA is consistent with investors’ misunderstanding of earnings manipulation and/or overinvestment by firm executives. Wu et al. (2010) attribute the NOA-return relation to optimal investment adjustment by managers in response to discount rate changes. The purpose of our research is twofold. First, we investigate the possible generalization of the NOA effect on stock returns in Europe. Second, we examine the possible mediating role of executive discretion concerning the presence/absence of the NOA effect on stock returns in an international setting. We are motivated by a desire to extend the existing accounting literature on the NOA effect on stock returns, by focusing on data from Europe, and to bridge this literature with the literature from strategic management, particularly on the consequences of managerial discretion on firm performance. Regarding our first research objective, we show that the negative association of NOA with subsequent raw returns occurs in nine European stock markets: Belgium, France, Germany, Netherlands, Norway, Spain, Sweden, Switzerland and U.K. In the great majority of these stock markets the level of transaction costs is generally lower relative to those stock markets where the NOA-return relation does not occur. We also find that the occurrence of this negative relation is of great economic importance; the hedge raw return on NOA portfolios ranges from 0.055 for Belgium to 0.107 for Norway. When all European equity markets are considered together, the hedge NOA portfolio earns a raw return of about 0.055, suggesting that the NOA effect on stock returns in European equity markets is weaker than in U.S. Regarding our second research objective, we find that the negative relation of NOA with future returns is more likely to occur and be stronger in countries with higher individualism, higher equity-market development, lower ownership concentration, higher competitiveness, better quality of accounting standards and lower earnings manipulation. Further, this relation is not statistically significant in European countries with the lowest individualism, the highest concentration of share ownership, the lowest competitiveness, the lowest quality of accounting standards and the highest earnings manipulation. Furthermore, once we adjust returns for risk, we show that hedge abnormal returns from NOA trading portfolios are positive and significant only in three European capital markets: Belgium, France and Netherlands. However, caution is required by readers in interpreting abnormal returns from trading portfolios (see Fama, 1998). It is plausible that abnormal returns may exist due to miscalculation of risks. Conventional risk-related adjustments (e.g., size) that are typically made to arrive at abnormal returns may pose a challenge in countries, where the number of firms for a given year is quite low (e.g. Belgium). Thus, we suggest an interpretation of the results from abnormal returns tests together with the results on the generalization and the role of managerial discretion for return predictability attributable to NOA worldwide. Overall, our findings make at least three contributions to the existing literature. First, they suggest that return predictability attributable to NOA is not just a ‘freak’ occurrence for the U.S. market, but it generalizes in other markets as well. More interestingly the NOA-return relation generalizes in the European Union, where countries share some similarities regarding economic status, legal system and accounting standards, but face systematically different degrees of constraint on firm executives. Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

21

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Second, our results come to complement, with regard to the factors explaining the NOA effect on stock returns, the studies by Crossland and Hambrick (2007, 2011), as they clearly confirm their findings about the importance that both informal and formal institutions have on managerial discretion and, in the end, on corporate performance. Third, they provide a real challenge on interpretation of the negative relation between NOA and future returns. Earnings management and/or agency-related overinvestment do not seem to constitute driving factors of the possible occurrence of this relation in European equity markets. We conjecture, that optimal investment behavior by firm executives as a rational response to the reduction in the cost of capital could be an underlying source of return predictability attributable to NOA. Overall, our paper highly reinforce the importance for future research to bring even closer the results from capital markets based accounting with advances from strategic management on managerial discretion. A possible suggestion might be to attempt to find alternative proxies for managerial discretion that do not require international diversity (e.g., joint CEOChairmanship, captive boards of directors, CEO tenure, CEO pay as evidence of entrenchment, etc.) in order to investigate their impact on asset pricing regularities associated with various accounting figures in the U.S. capital market. Appendix A. Definition of firm-level variables Variable

Measurement

Total assets (TA) Cash & cash equivalents (CASH) Minority interest (MINT) Total debt (TD) Ordinary and preferred shares (OPS) Total equity (TE) Operating assets (OA) Operating liabilities (OL) Net operating assets (NOA) Asset Growth Rate (AGR) Market capitalization (MV) Book-to-market ratio (BV=MV) Natural logarithm of market capitalization (SIZE) Natural logarithm of book-tomarket ratio (BM) Monthly raw return

W02999 W02001 W03426 W03255 W03995

Annual one-year ahead raw return (RET) Annual one-year ahead abnormal return (ARET)

W03501 W02999  W02001 W02999  W03426  W03255  W03995 OA–OL (scaled by lagged total assets) Annual percentage change in total assets (W02999) W08001 (measured six months after financial year-end). W03501=W08001 lnðW08001Þ lnðW03501=W08001Þ Monthly return data are obtained from Datastream (item RI), representing closing prices at the last trading day of the month. RET is calculated using compounded 12-monthly buy-and-hold returns. The return cumulation period begins six months after financial year-end. Six months after each financial year-end, firms are first sorted into four quartile portfolios by MV and in each of the resulted quartile portfolios are further sorted into four additional quartile portfolios by BV/MV. This procedure results in 16 benchmark portfolios and the matching return is the annual one-year ahead weighted average return for each benchmark portfolio. ARET is the difference between the RET and the matching return of the benchmark portfolio to which the firm belongs.

Note: ‘‘W” denotes that the relevant data item comes from Worldscope.

Appendix B. Definition & data of country-level characteristics Definition of Country-Level Characteristics Variable

Measurement – Data Sources

Transaction cost (TC)

An index based on commissions, fees, and market impact costs within each country. Data Source: Chan et al. (2005) and Chui et al. (2010) Average score on a psychological survey about IBM employees’ attitudes towards their work and private lives.

Individualism (IDV)

(continued on next page)

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

22

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

Appendix B (continued)

Definition of Country-Level Characteristics Market capitalization to GDP ratio (MKTtoGDP) Access-to-equity market (ACCESS) Ownership concentration (OWCR) Competitiveness (COMP) Accounting Standards (ACCS)

Earnings Management (EM)

Data Source: Hofstede (1980, 2001), www.geert-hofstede.com Average ratio of stock market capitalization of listed domestic firms to gross domestic product over the period 1989–2008. Data Source: www.worldbank.org Average score on a survey published in the Global Competitiveness Report, about the ability of firms to raise equity in local stock markets. Data Source: La Porta et al. (2006) The median percentage of common shares owned by the three largest shareholders in the ten largest nonfinancial firms. Data Source: La Porta et al. (2006) Average score for competitiveness over the period 2006–2007. Data Source: www.weforum.org Average number of 90 accounting and non-accounting items disclosed by a sample of large firms in their 1995 annual reports Data Source: Bushman et al. (2004) Percentage rank of two earnings smoothing metrics (reduced variability of reported earnings by altering accounting accruals and correlation between accounting accruals and operating cash flows) and two earnings discretion metrics (magnitude of accounting accruals relative to that of operating cash flows and small loss avoidance). Leuz (2010) provides scores on this index for the period 1990–1999 and for the period 1996–2005. Data Source: Leuz (2010)

Data of Country-Level Factors Country

NOAR

TC

IDV

MKTtoGDP

ACCESS

OWCR

COMP

ACCS

EM 1990–1999

EM 1996–2005

Austria Belgium Denmark Finland France Germany Greece Ireland Italy Netherlands Norway Portugal Spain Sweden Switzerland United Kingdom

0 1 0 0 1 1 0 0 0 1 1 0 1 1 1 1

53.2 27.1 40.8 45.2 35.7 30.6 105.1 93.7 41 24.5 41.5 35.7 39.2 33 38.6 34.1

55 75 74 63 71 67 35 70 76 80 69 27 51 71 68 89

0.210 0.569 0.497 0.899 0.618 0.389 0.453 0.583 0.337 0.935 0.393 0.310 0.594 0.871 1.828 1.272

4.89 5.7 5.87 6.37 5.75 5.93 5.28 5.29 4.41 6.43 5.57 4.5 5.09 6.15 6.07 6.26

0.58 0.54 0.45 0.37 0.34 0.48 0.67 0.39 0.58 0.39 0.36 0.52 0.51 0.28 0.41 0.19

5.156 5.056 5.553 5.505 5.205 5.482 4.121 5.079 4.370 5.365 5.175 4.471 4.696 5.442 5.544 5.565

62 68 75 83 78 67 61 81 66 74 75 56 72 83 80 85

0.862 0.739 0.475 0.397 0.475 0.726 0.91 0.428 0.844 0.593 – 0.774 0.756 0.394 0.637 0.216

0.808 0.682 0.530 0.260 0.536 0.620 0.881 0.199 0.826 0.482 – 0.880 0.792 0.168 0.504 0.133

References Arce, M., Mora, A., 2002. Empirical evidence of the effect of European accounting differences on the stock market valuation of earnings and book value. Eur. Account. Rev. 11, 573–599. Bagnoli, M., Watts, S., 2010. Oligopoly, disclosure and earnings management. Account. Rev. 85, 1191–1214. Ball, R., Kothari, S., Robin, A., 2000. The effect of international institutional factors on properties of accounting earnings. J. Account. Econ. 29, 1–51. Brown, L., Higgins, H., 2001. Managing earnings surprises in the US versus 12 other countries. J. Account. Public Policy 20, 373–398. Brown, L., Higgins, H., 2005. Managers’ forecast guidance of analysts: international evidence. J. Account. Public Policy 24, 280–299. Bushman, R., Piotroski, J., Smith, A., 2004. What determines corporate transparency? J. Account. Res. 42, 207–252. Callen, J., Morel, M., Richardson, G., 2011. Do culture and religion mitigate earnings management? evidence from a cross-country analysis. Int. J. Disclosure Gov. 8, 103–121.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013

G. Papanastasopoulos, D. Thomakos / J. Int. Financ. Markets Inst. Money xxx (2016) xxx–xxx

23

Chan, K., Covrig, V., Ng, L., 2005. What determines the domestic bias and foreign bias? Evidence from mutual fund equity allocations worldwide. J. Finance 60, 1495–1534. Chan, K., Chan, L., Jegadeesh, N., Lakonishok, J., 2006. Earnings quality and stock returns. J. Bus. 79, 1041–1082. Chan, L., Karceski, J., Lakonishok, J., Sougiannis, T., 2008. Balance sheet growth and the predictability of stock returns. Working paper, University of Illinois. Chudek, M., Truong, C., Veeraraghavan, M., 2011. Is trading on earnings surprises a profitable strategy? Canadian evidence. J. Int. Financ. Markets Inst. Money 21, 832–850. Chui, A., Titman, S., Wei, J., 2010. Individualism and momentum around the world. J. Finance 65, 361–392. Cooper, M., Gulen, H., Schill, M., 2008. Asset growth and the cross section of stock returns. J. Finance 63, 1609–1651. Crossland, C., Hambrick, D., 2007. How national systems differ in their constraints on corporate executives: a study of CEO effects in three countries. Strateg. Manag. J. 28, 767–789. Crossland, C., Hambrick, D., 2011. Differences in managerial discretion across countries: how nation-level institutions affect the degree to which CEOS matter. Strateg. Manag. J. 32, 797–819. Dechow, P., Richardson, S., Sloan, R., 2008. The persistence and pricing of the cash component of earnings. J. Account. Res. 46, 537–566. Degeorge, F., Patel, J., Zeckhauser, R., 1999. Earnings management to exceed thresholds. J. Bus. 72, 1–33. Degeorge, F., Ding, Y., Jeanjean, T., Stolowy, H., 2013. Analyst coverage, earnings management and financial development: an international study. J. Account. Public Policy 32, 1–25. Doukakis, L., Papanastasopoulos, G., 2014. The accruals anomaly in the UK stock market: implications of growth and accounting distortions. J. Int. Financ. Markets Inst. Money 32, 256–277. Fama, E., MacBeth, J., 1973. Risk, return, and equilibrium: empirical tests. J. Polit. Econ. 81, 607–636. Fama, E., French, K., 1993. Common risk factors in the returns on stocks and bonds. J. Financ. Econ. 33, 3–56. Fama, E., French, K., 1997. Industry costs of equity. J. Financ. Econ. 43, 153–193. Fama, E., 1998. Market efficiency, long-term returns, and behavioral finance. J. Financ. Econ. 49, 283–306. Fama, E., French, K., 2008. Dissecting anomalies. J. Finance 63, 1653–1678. Finkelstein, S., Peteraf, M., 2007. Managerial activities: a missing link in managerial discretion theory. Strategic Org. 5, 237–248. Gray, P., Johnson, J., 2011. The relationship between asset growth and the cross-section of stock returns. J. Bank. Finance 35, 670–680. Griffith, R., 2001. Product Market Competition, Efficiency and Agency Costs: An Empirical Analysis. Institute for Fiscal Studies. Grullon, G., Michaely, R., 2014. He impact of product market competition on firm’s payout policy. Working paper, Rice University. Han, S., Kang, T., Salter, S., Yoo, Y., 2010. A cross-country study on the effects of national culture on earnings management. J. Int. Bus. Stud. 41, 123–141. Hambrick, D., Abrahamson, E., 1995. Assessing managerial discretion across industries: a multimethod approach. Acad. Manag. J. 38 (1427), 1441. Hambrick, D., Finkelstein, S., 1987. Managerial discretion: a bridge between polar views of organizational outcomes. In: Staw, B., Cummings, L.L. (Eds.), Research in Organizational Behavior, vol. 9. JAI Press, Greenwich, CT, pp. 369–406. Hirshleifer, D., Hou, K., Teoh, S., Zhang, Y., 2004. Do investors overvalue firms with bloated balance sheets? J. Account. Econ. 38, 297–331. Hofstede, G., 1980. Culture’s Consequences: International Differences in Work-Related Values. Sage Publication, Beverly Hills, CA. Hofstede, G., 2001. Culture’s Consequences: Comparing Values, Behaviors, Institutions, and Organizations Across Nations. Sage Publication, Beverly Hills, CA. Ince, O., Porter, B., 2006. Individual equity return data from Thomson Datastream: handle with care. J. Financ. Res. 29, 463–479. Kedia, S., Philippon, T., 2009. The economics of fraudulent accounting. Rev. Financ. Stud. 22, 2169–2199. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., Vishny, R., 1997. Legal determinants of external finance. J. Finance 52, 1131–1150. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. J. Polit. Econ. 106, 1113–1155. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., Vishny, R., 2002. Investor protection and corporate valuation. J. Finance 57, 1147–1170. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., 2006. What works in securities laws? J. Finance 61, 1–32. Lam, E., Wei, J., 2011. Limits to arbitrage, investment frictions, and the asset growth anomaly. J. Financ. Econ. 102, 127–149. Lee, P., 1997. A comparative analysis of layoff announcements and stock price reactions in the United States and Japan. Strateg. Manag. J. 18, 879–894. Leuz, C., 2010. Different approaches to corporate reporting regulation: how jurisdictions differ and why. Account. Bus. Res. 40, 229–256. Leuz, C., Nanda, D., Wysocki, P., 2003. Earnings management and investor protection: an international comparison. J. Financ. Econ. 69, 505–527. Li, D., Zhang, L., 2010. Does q-theory with investment frictions explain anomalies in the cross-section of returns? J. Financ. Econ. 98, 297–314. Makino, S., Isobe, T., Chan, C., 2004. Does country matter? Strateg. Manag. J. 25, 1027–1043. Markarian, G., Santalo, J., 2014. Product market competition, information and earnings management. J. Bus. Finance Account. 41, 572–599. Matoussi, H., Jardak, M., 2012. International corporate governance and finance: legal, cultural and political explanations. Int. J. Account. 47, 1–43. Mclean, D., Zhang, T., Zhao, M., 2012. Why does the law matter? Investor protection and its effects on investment, finance and growth. J. Finance 67, 313– 350. McLean, D., Pontiff, J., Watanabe, A., 2009. Share issuance and cross-sectional returns: international evidence. J. Financ. Econ. 94, 1–17. Nickell, S., 1996. Competition and corporate performance. J. Polit. Econ. 104, 724–746. Nguyen, N., Truong, C., 2013. The information content of stock markets around the world: a cultural explanation. J. Int. Financ. Markets Inst. Money 26, 1– 29. Papanastasopoulos, G., Thomakos, D., Wang, T., 2011. Information in balance sheets about future stock returns: evidence from net operating assets. Int. Rev. Financ. Anal. 43, 214–229. Petersen, M., 2009. Estimating standard errors in finance panel data sets: comparing approaches. Rev. Financ. Stud. 22, 435–480. Pincus, M., Rajgopal, S., Venkatachalam, M., 2007. The accrual anomaly: international evidence. Account. Rev. 82, 169–203. Penman, S., 2007. Financial Statement Analysis and Security Valuation. McGraw-Hill. Shao, L., Kwok, C., Guedhami, O., 2010. National culture and dividend policy. J. Int. Bus. Stud. 41, 1391–1414. Shen, W., Cho, T., 2005. Exploring involuntary executive turnover through a managerial discretion framework. Acad. Manag. Rev. 30, 843–854. Tang, L., Koveos, P., 2008. A framework to update Hofstede’s cultural value indices: economic dynamics and institutional stability. J. Int. Bus. Stud. 39, 1045– 1063. Titman, S., Wei, J., Xie, F., 2013. Market development and the asset growth effect: international evidence. J. Financ. Quant. Anal. 48, 1405–1432. Yao, T., Yu, T., Zhang, T., Chen, S., 2011. Asset growth and stock returns: evidence from Asian financial markets. Pac.-Basin Finance J. 19, 115–139. Watanabe, A., Xu, Y., Yao, T., Yu, T., 2013. The asset growth effect: insights from international equity markets. J. Financ. Econ. 2, 529–563. Williamson, O., 1963. Managerial discretion and business behavior. Am. Econ. Rev. 53, 1032–1057. Wu, J., Zhang, L., Zhang, F., 2010. The q-theory approach to understanding the accrual anomaly. J. Account. Res. 48, 177–223. Wu, Y., Li, Y., 2011. Long-term return reversals—value and growth or tax? UK evidence. J. Int. Financ. Markets Inst. Money 21, 347–368. Zhang, Y., 2006. Net operating assets as a predictor of industry stock returns. Working paper, Chinese University of Hong Kong.

Please cite this article in press as: Papanastasopoulos, G., Thomakos, D. Managerial discretion, net operating assets and the cross-section of stock returns: Evidence from European countries. J. Int. Financ. Markets Inst. Money (2016), http://dx.doi.org/10.1016/j.intfin.2016.11.013