Determinants and Consequence of Cost Stickiness Mabel D’ Costa Ph.D. Candidate 2020 Determinants and Consequence of Cost Stickiness A thesis presented in partial fulfilment of the requirements for the Degree of Doctor of Philosophy in Accountancy at Massey University Auckland, New Zealand Mabel D’ Costa 2020 i ABSTRACT This research investigates the determinants and consequence of cost stickiness using data of publicly listed U.S. firms. Understanding the determinants of cost stickiness and its implications is extremely crucial, since it affects firms’ profitability, consequently, shareholders’ wealth. Moreover, cost management has even wider repercussions for both debt and equity investors in the areas of risk assessment and the trust of customers, employees, and other stakeholders in the community. Therefore, this study is organised into three different research essays: (i) financial constraint and cost stickiness; (ii) trade credit and cost stickiness; and (iii) cost stickiness and firm value. Essay One investigates the association between financial constraints and cost stickiness. Using a large U.S. sample from 1976 to 2016, I find that financially constrained firms exhibit less cost stickiness. I document that such low-cost stickiness supports both “good” and “bad” arguments depending on the managerial motivation, namely: earnings management incentives, agency problem and value-creating potential of SG&A costs. I also investigate whether the association between financial constraints and cost stickiness varies across the economic cycle. I find that low cost stickiness is observed during both economic expansion and economic contraction periods, although it is more pronounced during contraction. As resources drive the cost of a business, and financial constraints affect resource availability, studying cost behaviour of constrained firms makes a valuable contribution to the existing cost stickiness literature. In Essay Two, I examine the relation between trade credit and cost stickiness and further investigate the moderating effects of agency problem, product market competition, and customer concentration. I find that firms using high levels of trade credit exhibit lower cost stickiness and this is prevalent in the high agency problem sub-sample. In addition, in a non- competitive market, where the agency problem arises owing to lack of competition, trade credit ii plays an external monitoring role by attenuating cost stickiness. However, high customer concentration curtails this monitoring ability of trade credit providers. Finally, in Essay Three, I investigate the association between cost stickiness and firm value, and examine whether the association, if any, is mediated by cost of equity capital and cash flows. Using a large sample of U.S. data, I find a robust negative relationship between cost stickiness and firm value. I then explore whether resource adjustment, managerial expectations, and agency theories of cost stickiness affect the negative relation, and find some support for the agency view. Furthermore, I find evidence that the detrimental impact of cost stickiness on firm value is mediated partially through the cost of equity and cash flow channels. I enrich the cost management literature by integrating cost asymmetry with corporate finance. Keywords: Cost asymmetry; Cost behaviour; Financial constraint, Trade credit, Firm value, Agency problem, Resource adjustment, Managerial expectation iii Dedicated to Baba, Ma, Shanta, Sanjib da & Shimanto iv ACKNOWLEDGEMENT “You never know God is all you need until God is all you have.” – Rick Warren I am grateful to Almighty God for keeping me safe and sane during this three-year long roller- coaster ride which is coming to an end. I have struggled with my spirituality during this phase, but I have encountered numerous of His blessings which brought me back to Him. I am thankful to my parents for having the faith in me and letting me come to this far-away land to fulfil my dreams. I am blessed that Baba and Ma always prioritized education and career and have tried to provide the best that they could afford. I am blessed for their unconditional support. My elder sister Shanta, without her support I would not have come this far in life. She has always had my back and supported all my decisions and never let my smallest wishes go unfulfilled. I am blessed to have her as my sister and giving me the best gift Shimanto, my nephew. My major sacrifice for pursuing Ph.D. has been being unable to be part of my nephew’s childhood. I am fortunate to have Sanjib as my brother-in-law who never fails to appreciate my hard work and achievements. I am truly thankful to my primary supervisor Professor Ahsan for keeping up with my ups and downs during this journey, perhaps to certain extent adjusting his supervision style and appreciating my straightforwardness. Without his guidance I would not have been able to finish my Ph.D. on time and achieved success with multiple publications. My co-supervisor Assoc. Professor Borhan has supported me with both academic and non-academic matters for which I am extremely grateful. Sean has been always very encouraging and assisted me whenever I needed help with Stata. Thank you Sean, you have been such a blessing. I am also indebted to Mostafa bhai for his support in these three years long journey. Hedy and Sophia have always v been there with their encouraging words and smile to help me face my low moments. Emeritus Professor Jill has been very welcoming and is a key person in shaping up my career. My landlord James and his wife Vivi has been such a blessing in this middle earth. They have taken care of me when I needed mental support. Thank you, chef James for feeding me such delicious food. I am thankful to Nehal bhaia who although lives in Leeds, U.K., always concerned about my well-being. Thank you Varqa for creating the “devotional group” as it is helping me to connect with my spirituality. Saqiful is like a younger brother and I am thankful to him for always being there for me by listening to my grievances and bearing with my tantrums. One person without whom I would not have applied for Ph.D. is Kamrul, my moral support. He encouraged me to prepare the proposal and apply for Ph.D. He used to suggest universities and scholarship options. I am completing this journey of Ph.D. because of you! Adnan, the logic guru has always shown faith in me and helped me with my struggles as well. Thank you, Adnan, for all the prayers. Luna, you have always highlighted my strengths to me even when I took them for granted. Thank you Kamrul, Adnan and Luna for always leaning your ears to listen to my miseries. I started off as colleagues, but you guys are some of the closest friends I have. My life in Auckland would not have been so eventful and fun without Pick. She has always pushed me to my limits and had confidence in me even when I lost it. She has changed my lifestyle as she introduced me to Muay Thai and running. Because of Pick I have completed 11km run in 2019 which is a big achievement for a person who even a year before could not think of running 1km. She has given a lot in this short time since I have known her which I would never be able to repay. Tuba, we have come a long way with our sisterhood and friendship. I do not know how you have so much confidence in me and always happy and proud with my smallest of vi achievements. Whether I was drowning professionally or personally you have looked after me even being thousands of miles away. I am not sure how I would have handled my emotional meltdowns without you. The word thank you is not enough for always being there for me. There were good, bad and ugly days but it has been a self-discovering journey for me. I believe I have evolved as a person over these three years. As Mother Teresa said “God is everywhere and in everything and without Him we cannot exist”; the few people I have named here and many unnamed who have been there for me through my thick and thin are His blessings and support to help me reach my goal and become who I am today. vii TABLE OF CONTENTS Abstract i Acknowledgement iv Table of Contents vii List of Tables x Chapter One – Introduction 1 1.1 Motivations for the Research 1 1.2 Findings of the Research 7 1.3 Contributions & Implications of the Research 8 1.4 Organization of the Research 10 Chapter Two – Financial Constraint and Cost Stickiness (Essay One) 12 2.1 Literature Review & Hypotheses Development 14 2.1.1 Financial constraints and cost behaviour during business cycles 20 2.2 Research Design 21 2.2.1 Empirical model 21 2.2.2 Measurement of the independent variable: Financial constraints 22 2.2.3 Business cycle measure 24 2.2.4 Sample selection and descriptive statistics 24 2.3 Empirical Results & Analysis 26 2.3.1 Descriptive statistics and correlation analysis 26 2.3.2 Regression results: Financial constraints and cost stickiness 29 2.3.3 Is it “Good” or “Bad” less cost stickiness? 32 2.3.4 Financial constraints and cost stickiness during the business cycle 40 2.3.5 Additional tests 44 2.3.5.1 Corporate governance, financial constraints and cost stickiness 44 2.3.5.2 Change in tax rates, financial constraints and cost stickiness 45 2.4 Chapter Summary 46 viii Chapter Three – Trade Credit and Cost Stickiness (Essay Two) 48 3.1 Literature Review & Hypotheses Development 49 3.1.1 Trade credit and cost stickiness 49 3.1.2 The agency problem, trade credit and cost stickiness 53 3.1.3 Product market competition, trade credit and cost stickiness 54 3.1.4 Customer concentration, trade credit and cost stickiness 55 3.2 Research Design 57 3.2.1 Empirical model 57 3.2.2 Measurement of the independent variable: Trade credit 59 3.2.3 Moderating variables 60 3.2.4 Sample selection and descriptive statistics 62 3.3 Empirical Results & Analysis 64 3.3.1 Descriptive statistics and correlation analysis 64 3.3.2 Regression results: Trade credit and cost stickiness 67 3.3.2.1 Trade credit and cost stickiness: baseline regressions 67 3.3.2.2 Trade credit and cost stickiness: moderating effects of agency problem, product market competition, and customer concentration 73 3.3.2.3 Robustness test 81 3.4 Chapter Summary 82 Chapter Four – Cost Stickiness and Firm Value (Essay Three) 83 4.1 Literature Review & Hypotheses Development 84 4.1.1 Cost stickiness and firm value 84 4.1.2 Mediating effect of cost of equity and cash flow on cost stickiness and firm value 86 4.2 Research Design 87 4.2.1 Empirical model 87 4.2.2 Resource adjustment cost, managerial expectation and agency problem proxies 88 4.2.3 Mediating effect variables 90 4.2.4 Sample selection and descriptive statistics 91 ix 4.3 Empirical Results & Analysis 93 4.3.1 Descriptive statistics and correlation analysis 93 4.3.2 Regression results: Cost stickiness and firm value 95 4.3.3 Additional test 104 4.3.4 Robustness test 104 4.4 Chapter Summary 108 Chapter Five – Conclusion 109 5.1 Conclusion 109 5.2 Limitations & Scope 111 References 113 Appendix A: Variable Definitions (Essay One) 122 Appendix B: Variable Definitions (Essay Two) 124 Appendix C: Variable Definitions (Essay Three) 125 Appendix D: DRC 16 (Statement of Contribution: Doctorate with publications/manuscripts) 127 x LIST OF TABLES Table 2.1 Sample Selection and Industry Distribution 26 Table 2.2 Descriptive Statistics and Correlation 27 Table 2.3 OLS and Fixed Effect Regression Results – Financial Constraints and Cost Stickiness 30 Table 2.4 Financial Constraints and Cost Stickiness – Is it good or bad? 34 Table 2.5 Business Cycle, Financial Constraint and Cost Stickiness 42 Table 3.1 Sample Selection and Industry Distribution 63 Table 3.2 Descriptive Statistics 65 Table 3.3 Correlation 66 Table 3.4 OLS and Fixed Effect Regression Results – Trade Credit and Cost Stickiness 70 Table 3.5 Agency Problem, Trade Credit and Cost Stickiness 75 Table 3.6 Product Market Competition, Trade Credit and Cost Stickiness 79 Table 3.7 Customer Concentration (CC), Trade Credit and Cost Stickiness 80 Table 3.8 Change in Trade Credit and Cost Stickiness 81 Table 4.1 Sample Selection and Industry Distribution 92 Table 4.2 Descriptive Statistics and Correlation 94 Table 4.3 OLS and Fixed Effect Regression Results – Cost Stickiness and Firm Value 97 Table 4.4 Cost Stickiness and Firm Value 100 Table 4.5 Mediating Effect of Cost of Equity and Cash Flow 103 Table 4.6 Cost Stickiness and Firm Value: Alternative cost components 105 Table 4.7 Change in Cost Stickiness and Firm Value 106 1 CHAPTER ONE INTRODUCTION 1.1 Motivations for the Research Traditional cost behaviour identifies all costs as either fixed or variable with respect to concurrent sales, or some other cost driver (e.g. Anderson et al., 2003; Banker & Byzalov, 2014). However, in reality, the relation between cost and cost driver is more complex (Noreen, 1991), in that some costs rise more with increases in activity, than they decrease with proportionate decreases in activity levels (Cooper & Kaplan, 1998): a phenomenon known as ‘cost stickiness’. For example, Anderson et al. (2003) find that selling, general, and administrative (SG&A) costs increase by 0.55% for each 1% increase in sales; however, SG&A costs decrease by only 0.35% for each 1% decrease in sales. In this thesis I examine the determinants and consequences of cost stickiness. A plethora of academic research has documented evidence of cost stickiness in the U.S. and in international contexts. Some of the firm-level determinants of stickiness include prior activity change (Banker & Byzalov, 2014), managerial incentives (e.g. Dierynck et al., 2012; Kama & Weiss, 2013; Banker & Byzalov, 2014), organizational capital (Venieris et al., 2015), and investment in corporate social responsibile activities (Habib & Hasan, 2019). All these studies on the drivers of asymmetric cost behaviour are grounded on three theories, namely, resource adjustment theory, managerial expectation theory, and agency theory (Hartlieb et al., 2020; Banker et al., 2018). Resource adjustment theory is premised on the notion that many costs arise from managers’ deliberate resource commitment decisions. Once committed, it is not easy to scale back resources without incurring some adjustment costs: defined as “economic sacrifices, social, contracting or psychological costs, which emerge during the resource-adjustment 2 process” (Venieris et al., 2015, p. 55). Thus, managers take the decision based on cost-benefit analysis. For instance, the labour adjustment cost has been found to induce stickiness (Banker et al., 2013; Golden et al., 2020). Managers are likely to enter into contract for resources, which are costly to renegotiate; thus, when demand falls, managers are bound to retain those slack resources, because discarding them would incur obligatory contractual costs, such as severance payments (Calleja et al., 2006). Thereby, owing to high adjustment costs, managers will reduce costs to a lesser extent when activity decreases than they will expand costs when activity increases, therefore, generating cost stickiness. Managerial expectation theory posits that when managers are optimistic (pessimistic) about future demand, they are likely to retain (reduce) slack resources in the event of declining demand (Banker & Byzalov, 2014; Banker et al., 2014). If the resource adjustment cost is high and the manager expects that sales will increase in future it would be unwise and expensive to dispose slack resources. However, if the resource adjustment cost is low but slack resource levels are high, managers’ expectation of future demand should have no impact on cost stickiness (Chen et al., 2019a). Finally, agency theory suggests that self-serving managers tend to engage in empire building by retaining unutilized resources in order to grow the firm beyond its optimal size (Chen et al., 2012; Hope & Thomas, 2008; Jensen, 1986; Masulis et al., 2007; Stulz, 1990). Managers’ decisions to retain unutilized resources are driven by personal motives instead of economic rationales (Hartlieb et al., 2020). Such self-serving managerial actions induce cost stickiness, as slack resources are retained when demand declines. “The relation of costs to sales is of crucial importance since it determines net profit, the maximization of which is the basic goal of organizations” (Brüggen & Zehnder, 2014, p. 170). Additionally, cost management has even wider repercussions for both debt and equity investors in the areas of risk assessment and the trust of customers, employees (with respect to job 3 security) and other stakeholders in the community. Despite the significance of ‘availability of resources’ as a driver of cost management, very little research, as of yet, has investigated the extent to which firm-level financial constraints affect cost stickiness. This is surprising, since the availability of resources affects cost behaviour, and constrained firms naturally suffer from resource shortages. Thereby, for Essay One I choose cost stickiness as the appropriate lens for understanding the effects of financial constraints. I predict that costs would be less sticky for constrained firms, because managers of financially constrained firms should decrease slack resources to control costs, to maintain profitability, and to generate internal finance for investment or expansion of their business. When sales decrease, financially constrained firms suffer a relatively greater reduction in the present value of revenue. Consequently, they decrease costs by a higher amount (Cheng et al., 2018); thus, exhibit less cost stickiness. Importantly, costs also do not increase with an increase in sales, because financially constrained firms encounter higher costs of capital from both equity and debt providers. This prediction supports the “efficiency” view of cost stickiness, which implies that firms will decrease unused resources for the right reasons i.e., for survival. On the other hand, retaining unutilised resources by managers of financially-constrained firms for ‘empire building’ reasons, is a manifestation of “bad” cost stickiness, i.e., stickiness stemming from the wrong reasons (Banker et al., 2018). Examining whether cost is sticky or less sticky provides only a partial explanation; therefore, it is crucial to examine why the cost behaves in a certain way. Thus, I use three contextual settings to examine cost behaviour by financially- constrained firms. These contexts are the earnings management context; the agency context; and the value-creating potential of SG&A costs context. In Essay Two I explore the association between trade credit and cost stickiness, and whether this association, if any, is moderated by agency problem, product market competition and customer concentration. Trade credit is the major source of external finance for many firms 4 around the globe (see, e.g. Abdulla et al., 2017; Afrifa et al., 2018; Demirgüç-Kunt & Maksimovic, 1999; Fabbri & Klapper, 2016; Wilson & Summers, 2002). Accounts payable, the major component of trade credit, comprises 25% of total liabilities, 35.36% of current liabilities and 9% of total sales in my sample of U.S. firms from 1977 to 2017. Suppliers are willing to provide trade credit because it offers them a comparative advantage over financial institutions in acquiring information (see, e.g., Ferrando & Mulier; 2013; Goto et al., 2015; Martínez‐Sola et al., 2013), evaluating the creditworthiness of buyers, and enforcing credit contracts (see, e.g. Burkart & Ellingsen, 2004; Fabbri & Menichini, 2010). However, by doing so, the suppliers put themselves at risk, because there is the possibility that the buyer might misuse resources, e.g., invest heavily in unproductive expenses including in cost of goods sold (COGS) and SG&A costs, thereby, increasing the default risk. This not only leads to significant losses for suppliers, but also increases their bankruptcy risks (Jacobson & Von Schedvin, 2015). Such concerns naturally motivate suppliers to be active monitors of resource usage by customers: a function that is facilitated by the holding of private information about their buyers. According to Anderson et al. (2003) and Brüggen and Zehnder (2014), managers retain unused resources in order to avoid adjustment costs. However, agency theory dictates that cost stickiness results from the self-serving behaviour of managers, who are likely to engage in empire building by retaining unutilised resources in order to grow the firm beyond its optimal size (see, e.g. Chen et al., 2012; Hope & Thomas, 2008; Jensen, 1986; Masulis et al., 2007; Stulz, 1990). Prior studies find that managers’ empire building tendencies lead to cost stickiness, but the relation is weaker under strong corporate governance, implying that corporate governance can mitigate the agency problem (see, e.g. Chen et al., 2012). Prior literature on trade credit has provided evidence supporting the corporate governance role of trade credit (see, e.g. Cao et al., 2018; Fisman & Love, 2003; McMillan & Woodruff, 1999; Petersen & Rajan, 1997). Thereby, I posit that, when sales decrease, trade credit will play a stronger monitoring 5 role by forcing buyers to cut slack resources in firms plagued with marked agency problem. Thus, I expect trade credit to lessen cost stickiness in firms with marked agency problem, when compared with their immaterial agency problem counterparts. Agency problem is more acute for firms operating in non-competitive industries, and existing evidence shows that market competition acts as an external governance mechanism that reduces agency problem (Giroud & Mueller, 2011; He, 2012) by disciplining managers and encouraging operational efficiency (Liu et al., 2017a). Thus, I predict that firms operating in non-competitive industries will benefit more from the monitoring role of trade credit in restraining manager’s opportunistic empire building behaviour. It is also intuitive to expect that customer size can influence the monitoring power exerted by suppliers on customers. Prior studies find that major customers exert bargaining power over suppliers by demanding lower prices and delayed payments (see, e.g. Bhattacharyya & Nain, 2011; Campello & Gao, 2017; Fee & Thomas, 2004; Murfin & Njoroge, 2014). In such a scenario, if suppliers enforced their monitoring role on customers, profits would reduce, and the volatility of suppliers’ earnings and cash flow would increase (see, e.g. Balakrishnan et al., 1996; Gosman & Kohlbeck, 2009; Huang et al., 2016; Piercy & Lane, 2006; Ravenscraft, 1983). Therefore, I would expect suppliers to play a more prominent monitoring role in the low customer concentration sub- sample, because of their customer’s relatively weak bargaining power and high switching costs. While a plethora of studies have examined the determinants of cost stickiness (see Banker et al., 2018 for a comprehensive review), surprisingly little evidence exists on its implications. Weiss (2010, p. 1442), who documents a positive relationship between cost stickiness and the analyst forecast error argues that a firm with higher cost stickiness demonstrates greater decline in earnings because “……stickier costs result in a smaller cost adjustment when activity level declines and, therefore, lower cost savings [which] result in a greater decrease in earnings. This greater decrease in earnings when the activity levels fall 6 increases the variability of the earnings distribution, resulting in less accurate earnings predictions.”. Weiss (2010) also finds that firms with stickier costs have lower analyst coverage, and that investors rely less on such firms’ realized earnings because of their lower predictive power. Ciftci et al. (2016), reveal that analysts are unable to recognize and incorporate the ‘sticky’ nature of costs in their forecasts: a feature that increases error in earnings prediction. In addition, cost stickiness has been found to increase credit risk. Firms encounter increased default and credit risk owing to higher earnings and asset volatility stemming from cost stickiness (Homburg et al., 2016). Similar to analysts, managers are also unable to incorporate asymmetric cost behaviour, as Ciftci and Salama (2018) show that cost stickiness leads to greater errors in management earnings forecasts. Thus, in Essay Three I examine the association between cost stickiness and firm value, and the mediating effect of cost of equity capital and cash flows on this relationship, if any. From resource adjustment and managerial expectation theory, I hypothesize that cost stickiness may not affect firm value adversely, as investors are likely to consider the downward resource adjustment costs (e.g., economic sacrifices, social, contracting or psychological costs) associated with disposing of slack resources when demand declines. From managerial expectation theory angle, when managers are optimistic (pessimistic) about future demand, then they are likely to retain (dispose) slack resources in the event of declining demand, to avoid future (current) adjustment costs (Venieris et al., 2015). Changes in managerial expectation, therefore, could affect the firm value differentially. However, agency theory perspective suggests that managers with empire-building tendencies are unlikely to reduce unutilised resources when sales decline (Chen et al., 2012): a phenomenon that should affect firm value adversely. 7 1.2 Findings of the Research Using a large sample consisting U.S. public listed firms, my overall findings suggest that financial constraint leads to less cost stickiness and, owing to the monitoring role of suppliers, trade credit lowers cost stickiness. From the implication perspective of cost stickiness, my evidence suggests that cost stickiness destroys firm value. Using three different measures of firm level financial constraints in Essay One, I document that financially constrained firms do, indeed, exhibit less cost stickiness, and such low cost stickiness is a manifestation of both “good” and “bad” cost stickiness. I further investigate whether the business cycle moderates low cost stickiness for constrained firms. I predict that costs would be less sticky for financially constrained firms during both expansion and contraction periods, although the effects should be more pronounced during the contraction phase. During expansion, most firms operate in a favourable business environment, owing to lower information asymmetry (Choe et al., 1993). Therefore, investors are likely to charge a higher cost for the constrained firms, and this leads to use of internal capital. Covas and Haan (2011) document the observation that internal financing is procyclical. During contractionary periods external investors have limited capital. Therefore, they are more selective in providing finance, consequently, external financing turns out be extremely expensive, or even unavailable, to constrained firms. My empirical evidence documents the existence of low stickiness for financially-constrained firms during both economic expansion and economic contraction periods. In Essay Two I use three different widely used measures of trade credit, and document that trade credit results in relatively low COGS and SG&A cost stickiness for all three measures of trade credit. I also find some evidence that trade credit-induced lower cost stickiness is prevalent in high agency problem sub-sample as compared to low agency problem sub-sample. This implies that trade credit restrains the empire building tendencies of managers in high 8 agency problem firms, by forcing them to dispose unutilized resources when sales decline. I also find that firms operating in non-competitive markets with more trade credit show low cost stickiness, as the magnitude of the coefficient on the interactive variable, trade credit and cost stickiness, is positive and statistically significant in the low, as opposed to the high, product market competition group. Finally, with respect to the moderating effects of customer concentration on the association between trade credit and cost stickiness, I find the coefficients on the interactive variable to be positive and significant for both low and high customer concentration groups. Finally, in Essay Three I document that operating cost stickiness reduces firm value. In terms of economic magnitude, a one standard deviation increase in cost stickiness decreases firm value by 2.44 percent relative to its mean. An additional test suggests that this detrimental relationship between cost stickiness and firm value exists owing to investor recognition of agency problem associated with retention of slack resources. Further, I find evidence that this negative relation between cost stickiness and firm value is partially mediated through both the cost of equity and the cash flow channels, as the direct effect of cost stickiness on firm value accounts for the bulk of the total effect. The negative relationship between cost stickiness and firm value also holds for two other cost components, namely, SG&A and COGS. 1.3 Contributions & Implications of the Research In general, one of the studies contribute to the line of research that integrates management accounting (asymmetric cost behaviour) with financial accounting and corporate finance. Specifically, through Essay One I make the following contributions: first, it fills the void in the literature on how resource availability affects cost behaviour. Since costs are driven by resources, and resource availability depends on access to capital, studying the cost behaviour of constrained firms makes a valuable contribution to the existing cost asymmetry literature. 9 Second, it describes how financial constraint impacts SG&A costs behaviour, and how this relation persists during economic cycles. Third, it is one of the first studies to test the relation between financial constraints and cost behaviour in the context of “good” versus “bad” cost stickiness (Brüggen & Zehnder, 2014). Essay Two, on the other hand, contributes firstly by reconfirming trade credit’s monitoring role from a cost management angle. To the best of my knowledge, this is the first study to examine the relation between trade credit and cost stickiness. Although a plethora of research has explored the determinants of trade credit (Seifert et al., 2013), very little research examines the monitoring role of trade credit (see, e.g. Cao et al., 2018; McMillan & Woodruff, 1999). Therefore, the findings of this study contribute to both cost management and trade credit literatures. Secondly, besides documenting that firms using high levels of trade credit exhibit less cost stickiness, I further show that such behaviour is desirable, because it disciplines opportunistic managers. Thirdly, I contribute to the product-market competition and customer concentration literature by documenting that the trade-credit and cost behaviour relationship varies conditional on the characteristics of the market in which the firms operate, as well as on the suppliers’ incentives for monitoring the cost management approaches of their buyers. Finally, the presence of trade credit is likely to assure investors, and increase other stakeholders’ confidence, regarding the operational efficiency of firms, especially when the external governance system is weak. Therefore, firms can send positive signals to investors and other stakeholders through trade credit finance. Essay Three makes a significant contribution to understanding how cost stickiness could explain firm value, because it is not clear ex-ante whether the findings documented by Weiss (2010) will mean that firms having more cost stickiness will have lower market value. Because costs are a core driver of firm profitability, consequently, firm value. Therefore, it is important to understand how cost stickiness affects firm value, since the maximization of firm value is 10 considered to be the primary objective of the firm (Jensen & Meckling, 1976). Weiss (2010) uses three-day cumulative abnormal return (CAR) surrounding the earnings announcement dates to proxy for market response to quarterly earnings announcements. However, such a short window may fail to capture managerial resource adjustment decisions. This is because companies make resource adjustment decisions throughout the year, and the chosen short window may not coincide with any significant resource adjustment decisions. Hence, taking a longer time span of one year could ensure that resource adjustment decisions during the entire year are incorporated into firm value; thereby, overcoming the limitations of the short window test to some extent. Understanding the implications of cost stickiness is important for managers as well as investors. Managers need to be aware of how their deliberate resource adjustment decisions could affect the overall financial health of the firm. Feedback from the market could enable them to be more efficient in managing resources. From the perspective of the investors, it is important that they understand the rationale behind retaining slack resources. A myopic view that fails to recognize the rationale for retaining resources: i.e., to minimize resource adjustment costs; might lead them to conclude erroneously that retention of unutilised resources is detrimental to firm value. Therefore, my study provides a timely contribution to the limited research on the implications of cost stickiness. 1.4 Organization of the Research The remainder of the thesis proceeds as follows: Chapters two, three and four review the related literature, develop the hypotheses, discuss research methods and sample selection, with analysis of test results for essay one, two and three, respectively. Essay one is titled “Financial constraint and cost stickiness”, essay two is titled “Trade credit and cost stickiness” and the third essay is 11 titled “Cost stickiness and firm value”. Chapter five concludes the thesis with its implications and limitations, and outlines the scope for future related research. 12 CHAPTER TWO FINANCIAL CONSTRAINT AND COST STICKINESS (ESSAY ONE) This study investigates the association between financial constraints and cost stickiness. Anderson et al. (2003) document that SG&A costs are sticky i.e. costs rise more when sales increase, but decrease less when sales decrease. However, costs can be anti-sticky as well, implying that the rise in costs when sales increase is less than their fall when sales decrease (e.g. Weiss, 2010; Banker & Byzalov, 2014). Despite the significance of ‘availability of resources’ as a driver of cost management, very little research as of yet has investigated the extent to which firm-level financial constraints affect cost stickiness. This is surprising since availability and accessibility to finance impacts the availability of resources, and that in turn, impacts cost behaviour. Lamont et al. (2001) define financial constraints as frictions that prevent firms from funding their desired investments1. In a frictional environment, investment and growth depend largely on the availability of internal capital, as the cost of raising outside capital can be high relative to that of internally generated funds. This is particularly true for financially constrained firms that face severe agency and transaction costs in accessing external capital markets (Korajczyk & Levy, 2003). As constrained firms have to pay high interest rate on loans, they rely heavily on other sources of finance, e.g., trade credit and internal fund to finance continuation of their operation (Mulier et al., 2016). Consequently, constrained firms with attractive growth opportunities but without access to external financing may invest less into optimal value-increasing investment projects, resulting in lower future growth and firm value. 1This study focuses on financial constraint, not financial distress. Senbet and Wang (2012) note that distressed firms are unable to keep the promises made to creditors. In other words, financial distress refers to the inability of a company to pay its financial obligations as they mature (Beaver et al., 2011). 13 Young and small firms are likely to be financially constrained because such firms suffer from high information asymmetry (Beck et al., 2006; Arslan et al., 2006; Hadlock & Pierce 2010). Firms that pay dividend (Fazzari et al., 1988), have affiliation with business groups (Hoshi et al., 1991; Kato et al., 2002), and are politically connected (Poncet et al., 2010; Shen & Lin, 2016; Cull et al., 2015) are found to be financially unconstrained. Prior research documents a number of consequences of financial constraints, including increased earnings management (Kurt, 2017), high engagement in corrupt activities (Lopatta et al., 2017), and aggressive tax planning (Edwards et al., 2016; Law & Mills, 2015). Firms incur costs even when firms are faced with financial constraints, because some of those costs are contractual, and failing to pay those could lead to bankruptcy (Chen et al., 2019b). Therefore, it is not unlikely that financially constrained firms may retain slack resources even when sales decline: an action that leads to higher cost stickiness. However, from a resource adjustment cost perspective, I posit that financial constraints will result in lower cost stickiness for SG&A costs. When sales decrease, financially constrained firms suffer a relatively greater reduction in the present value of revenue, thereby, forcing them to cut back on unutilized resources. Importantly, costs do not increase with an increase in sales, because financially constrained firms encounter higher costs of capital from both equity and debt providers. Consequently, they decrease costs by a higher amount (Cheng et al., 2018), thus, exhibiting lower cost stickiness. Said differently, financially constrained firms may put less weight on future adjustment costs, and more weight on the costs of unused capacity, thus altering the trade-offs involved in their resource allocation decision. Cheng et al. (2018) document anti-sticky cost behaviour for a sample of Chinese private firms using a regional financial development index as a proxy for access to finance. However, my paper differs from the Cheng et al. (2018) paper in a number of important ways. Cheng et al. (2018) use a sample consisting of only small and private Chinese companies. These firms, 14 compared to listed firms, have access to the debt market only. Therefore, the findings of Cheng et al. (2018) are unlikely to be generalizable to U.S. listed firms, which have access to both debt and public equity markets, and are much larger in size. In addition, prior research show that in emerging market government plays a role in allocation of financial resources (Cull et al. 2015; Chen et al., 2017). Cheng et al. (2018) use regional financial development as a proxy for financial constraints. However, such a macroeconomic variable fails to incorporate firm- specific idiosyncrasies that can affect the magnitude of financial constraints differentially. I overcome this problem by using firm-specific financial constraint measures. I also use three contextual settings, where the association between financial constraints and cost stickiness could become more, or less, pronounced. 2.1 Literature Review & Hypotheses Development In the seminal paper of Anderson et al. (2003) on cost stickiness, the authors propose two theories underlying cost stickiness: adjustment cost theory and agency theory. The former relies on the notion that many costs, arise from managers’ deliberate resource commitment decisions. Once committed, it is not easy to scale back resources without incurring some kind of adjustment costs. Therefore, to the extent that managers recognize the trade-offs arising because of adjustment costs, they will reduce costs to a lesser extent when activity decreases than they will expand costs when activity increases and, thereby, generate cost stickiness. Agency theory- based arguments for cost stickiness consider the self-serving behavior of managers, who are likely to engage in empire building by retaining unutilized resources in order to grow the firm beyond its optimal size (Chen et al., 2012; Hope & Thomas, 2008; Jensen, 1986; Masulis et al., 2007; Stulz, 1990). Such actions induce cost stickiness, as slack resources are not disposed of when sales decline. In addition, managerial expectation has also been used to explain asymmetric cost behaviour (Banker & Byzalov, 2014; Banker et al., 2014b). If managers are 15 optimistic about future demand, they are likely to retain slack resources even when sales decrease; whereas, if managers are pessimistic about future demand then they are likely to dispose slack resources when sales decline. Cost behaviour is driven by the availability of resources; and the availability of resources, in turn, depends on the availability of finance. Firms face financial constraints for various reasons, such as capital market imperfections stemming from information asymmetry, weak institutional settings (Chen et al., 2017), agency problem (Pawlin & Renneboog, 2005) and risk (Senbet & Wang, 2012); as a result, firms’ are unable to borrow or issue equity (Lamont et al., 2001). If firms operated under frictionless capital markets then managers did not have to make trade-off decisions regarding which projects to invest in and which projects to forgo, as the availability of abundant resources would have enabled managers to invest in all positive NPV projects. However, in the real-world frictional markets, investment and growth depend largely on the availability of internal capital, as the cost of raising outside capital could be high relative to that of internally generated funds. Consequently, firms with attractive growth opportunities but without access to external financing may invest less into optimal value- increasing investment projects. Financial constraints, therefore, will require managers to carefully consider retaining or disposing slack resources for optimizing future growth and firm value. From the resource adjustment cost perspective, I posit that when sales decrease it becomes more costly to maintain unutilized resources for constrained firms, because such resources incur additional costs which, in turn, place further constraints on the financial health of the firm. Maintaining unutilized resources decreases the present value of sales, and increases the opportunity cost of keeping unused resources and, thereby, decreases firm profitability. Especially, successive decreases in sales make it more costly, at a given time, for constrained firms to maintain slack resources for future periods. Moreover, constrained firms are more 16 likely to hold on to cash instead of non-cash assets, in order to fund future expansion through internal finance (Calomiris et al., 1995). Therefore, when sales decrease, financially constrained firms are more likely to dispose of unutilized resources, to reduce avoidable costs. I posit that for financially constrained firms, the option value of waiting for the arrival of new information falls short of the NPV of reduction in future costs from downward resource adjustments, since the propagation of financial constraints could lead to corporate bankruptcy. Popov and Rocholl’s (2015) study of the German market shows that firms that had a credit relation with at least one global financial crisis (GFC)-affected bank, had to decrease both the number of employees and the average compensation of the remaining employees. Fernandes and Ferreira (2017) show that, in the post-GFC era, financially constrained Portuguese firms hired more fixed-term workers compared to permanent workers, than their unconstrained counterparts.2 Fernandes and Ferreira (2017) conclude that increased proportions of fixed-term employees enable the flexibility needed to adjust future employment rates without incurring additional firing costs. Constrained firms also face more upward resource adjustment costs, as they incur higher transaction costs and have available loans that come with stricter conditions, including stringent collateral requirements. All these increase for constrained firms during an increase in sales or an expansion of the business (Cheng et al., 2018). Therefore, constrained firms rely heavily on other sources of finance, e.g., trade credit and internal fund to finance their operation (Mulier et al., 2016). However, the adjustment cost theory posits that, once committed, resources are not easy to scale down without incurring additional (future) adjustment costs; such as severance pay, search and training costs for new employees and transaction costs associated with purchasing new equipment (Cheng et al., 2018). This could be more applicable to constrained firms, who 2Fixed-term workers have a flexible fixed-term contract and, therefore, can be laid off without incurring the severance payment that is required to lay off permanent workers (Fernandes & Ferreira, 2017). 17 face difficulty in raising capital when demand increases after a slump and, thereby, are inhibited from procuring critical resources, for example, skilled employees. Therefore, managers of constrained firms might prefer to retain unused resources in the short run to minimize the adjustment costs. Banker et al. (2013a) document that at country-level, stricter employment protection regulation results in higher firing costs, which could be more onerous for financially constrained firms, as saving costs and/or conserving cash in the present are the key priorities for these firms (Caggese et al., 2019). Therefore, in a well-developed country like the U.S., adjustment costs associated with labor are high compared to those in China and other emerging economies, owing to the availability of more highly-skilled human capital and the enforcement of minimum wage regulations (Banker et al., 2013). Thus, laying off human capital, which comes with high adjustment costs, might threaten the future growth of firms. However, based on the more rational premise that financially constrained firms will reduce unutilized resources as a survival strategy, I hypothesize the following: H1: Financially-constrained firms will exhibit less cost stickiness. I subject H1 above to empirical tests using three different contextual settings as explained below, to shed light on “good” versus “bad” less cost stickiness. My first context incorporates earnings management incentives to examine variations in cost behaviour. Kama and Weiss (2013) document that managers have an incentive to cut back resources to meet earnings benchmarks. When sales decline, financially constrained firms suffer a relatively greater reduction in the present value of revenue. Thus, such firms might be tempted to cut down resources to reduce costs, hence, increase profit to meet earnings benchmarks. This should provide a positive signal about the future survival prospects of these financially constrained firms. In line with this view, Kurt (2017) predicts and finds that constrained firms engage in income-increasing earnings management more aggressively than 18 unconstrained firms, around seasoned equity offerings. Linck et al. (2013) document that constrained firms overstate earnings compared to their unconstrained counterparts during the quarters prior to investment. Earnings management by financially constrained firms designed to understate costs, therefore, could result in less cost stickiness, but for the wrong reason, i.e., “bad” less stickiness. On the other hand, the existence of less cost stickiness in the absence of earnings management would support “good” less stickiness. Thus, I develop the following hypothesis: H1EM: Less cost stickiness exhibited by financially-constrained firms without earnings management incentives will support “good” less stickiness. My second context is the existence of the agency problem manifested through actions that encourage empire building. Managers who engage in empire-building for personal benefits are likely to overinvest when sales increase but are unlikely to cut back resources when sales decline (e.g. Williamson, 1963; Banker et al., 2018). This perspective suggests that financially constrained firms plagued with pervasive agency problem (empire building incentives in my case) will exhibit cost stickiness. On the contrary, Musso and Schiavo (2008) document that financial constraints increase the possibility of firms exiting the market. Therefore, survival may become crucial for constrained firms, and might require the elimination of costs in order to generate the funds they are unable to raise from the financial market. For constrained firms suffering from agency problem, the question of whether maximizing personal benefits outweighs survival remains unclear and unexplored. Based on the aforementioned arguments, I conjecture that, even in firms having pervasive agency problem, firm survival would take priority. Therefore, if firm survival takes precedence over empire building incentives, I would expect a scaling down of resources in the event of a sales decline for financially constrained 19 firms with agency concerns (good less cost stickiness). Thus, I develop the following hypothesis: H1AP: Less cost stickiness exhibited by financially-constrained firms with high agency problem will support “good” less stickiness. My final context involves cost behaviour conditional on the value-creating abilities of SG&A costs. Empirical research shows that value-creating SG&A costs exhibit greater cost stickiness (e.g. Lev and Sougiannis 1996, Banker et al., 2011; Chen et al., 2012). Decreasing value-creating SG&A costs involves higher resource adjustment costs (Liu et al., 2017b). Chen et al. (2012) predict, and find, that SG&A costs can create greater future value, and SG&A cost stickiness is influenced by economic considerations. The latter implies that constrained firms would trade-off the benefits derived from retaining value-creating resources against the costs of maintaining those resources. On the other hand, if SG&A costs generate low future value, then managers do not have a legitimate reason to retain such resources in the event of a sales decline. However, Caggese et al. (2019) document that employee layoff decisions of constrained firms are inefficient compared to those of their unconstrained counterparts, for a sample of Swedish companies. The authors show that constrained firms fire recently-hired workers having high expected productivity growth, to take advantage of their low firing costs. Laying off long-tenured workers who are considered less productive, may expose firms to significantly larger firing costs, e.g., higher severance payments, as compared to firing recently- recruited workers. In sum, their evidence suggests that constrained firms make the wrong firing decisions in order to conserve cash in the short term. Additionally, Musso and Schiavo (2008) posit that, to alleviate financial constraints, firms are likely to shed long term investments, giving rise to detrimental implications for their long-term growth prospects. Therefore, 20 constrained firms exhibit good (bad) less cost stickiness if they reduce future value-reducing (value-creating) SG&A costs. H1VSGA: Less cost stickiness through a reduction in value-destroying (value-creating) SG&A costs will support good (bad) less stickiness for financially-constrained firms. 2.1.1 Financial constraints and cost behaviour during business cycles The business cycle adds uncertainty to economic activities and to firm performance as reflected in corporate sales, profit, cash flow, and dividends. During expansion, most firms operate in a favourable business environment, with little competitive pressure, growing sales, and rising profitability. In contrast, during recession, corporate sales decline, profits shrink, and dividends may slump, or disappear entirely. Choe et al. (1993) argue that information asymmetry is less during an expansion period. Therefore, investors are able to identify the opportunity for, and are likely to charge, a higher cost for constrained firms, and this directs those constrained firms towards the utilization of internal capital rather than external financing sources. Thus, constrained firms would face significant pressure to increase their internal resources during an economic expansion (e.g. Hennessy & White, 2007; Kurt, 2017). During contraction periods, information asymmetry increases. Investors become extremely selective while making an investment decision for several reasons, including higher information asymmetry and scarcity of resources. This creates an adverse shock for firm cash flows. Chevalier and Scharfstein (1996) find that constrained firms have low cash flows during a contraction period, and are likely to increase prices compared to their unconstrained counterparts in order to achieve short-term profit. However, such a strategy could motivate unconstrained firms to engage in predatory pricing strategies, resulting in constrained firms being driven out of the market (Liu et al., 2017b). Firms experiencing a financial downturn reduce costs, including investments in CSR and assets substantially, to stabilize business 21 operations and to maintain profitability (e.g. Robbins & Pearce, 1992; Bansal et al., 2015; Habib & Hasan, 2019). According to the seminal work of Robbins and Pearce (1992), to overcome financial downturns, managers engage in “restructuring”, “downsizing” or “downscoping”: actions that save money. Firms undergoing a declining performance will reduce costs significantly in order to stabilize operations and retain profitability (e.g. Bibeault, 1982; Hambrick & Schecter, 1983; Slatter, 1984; Finkin, 1985; Bailey & Szerdy, 1988; Grinyer et al., 1988; Dumaine, 1990; Grinyer & McKiernan, 1990; Robbins & Pearce, 1992). Banker et al. (2016) document that, owing to managerial pessimism about future sales, firms’ exhibit less cost stickiness during economic contraction. Therefore, if managers are pessimist about future demand, as would be the case during a contractionary period, they would decrease slack resources aggressively when sales decrease, resulting in low cost stickiness or high anti- stickiness (Banker et al., 2018). Following, H1, I, therefore, predict that constrained firms would exhibit a greater degree of less cost stickiness during economic recession, than during economic expansion. I hypothesize as follows: H2: For financially constrained firms, the magnitude of less cost stickiness will be greater during a contraction period than during an expansion period. 2.2 Research Design 2.2.1 Empirical model I use the following regression specification to test H1: 𝐿𝑁[ 𝑆𝐺&𝐴𝑡 𝑆𝐺&𝐴𝑡−1 ] = 𝛽0 + 𝛽1 𝐿𝑁 [ 𝑆𝐴𝐿𝐸𝑡 𝑆𝐴𝐿𝐸𝑡−1 ] + 𝛽2 𝐷𝐸𝐶𝐷𝑈𝑀 ∗ 𝐿𝑁 [ 𝑆𝐴𝐿𝐸𝑡 𝑆𝐴𝐿𝐸𝑡−1 ] + 𝛽3 𝐹𝐶 + 𝛽4 𝐹𝐶 ∗ 𝐿𝑁 [ 𝑆𝐴𝐿𝐸𝑡 𝑆𝐴𝐿𝐸𝑡−1 ] + 𝛽5 𝐹𝐶 ∗ 𝐷𝐸𝐶𝐷𝑈𝑀 ∗ 𝐿𝑁 [ 𝑆𝐴𝐿𝐸𝑡 𝑆𝐴𝐿𝐸𝑡−1 ] + ∑ 𝛽𝑚𝐸𝐶𝑂𝑁𝑉𝐴𝑅𝑚,𝑡 9 𝑚=6 ∗ 𝐷𝐸𝐶𝐷𝑈𝑀 ∗ 𝐿𝑁 [ 𝑆𝐴𝐿𝐸𝑡 𝑆𝐴𝐿𝐸𝑡−1 ] + ∑ 𝛽𝑛 13 𝑛=10 𝐸𝐶𝑂𝑁𝑉𝐴𝑅𝑛,𝑡 + 𝜀 (2.1) where SG&A (Compustat data item XSGA) is selling, general and administrative expenses, and SALE (Compustat data item SALE) is sales revenue. DECDUM takes the value 22 of 1 if sales in year t are less than sales in year t-1, and 0 otherwise. Coefficient β1 measures the percentage increase in SG&A with a 1% increase in sales revenue. The sum of coefficients β1 and β2 measures the percentage decrease in SG&A with a 1% decrease in sales. A significant positive β1 and a significant negative β2 confirm cost stickiness. FC are the financial constraint variables (detailed in subsection 2.2.2), and ECONVAR are the economic variables and include: asset intensity (AIN), measured as the total assets divided by the sales revenue for year t; employee intensity (EIN), the ratio of the total number of employees over sales; successive decrease (SUDEC), an indicator variable equal to 1 if the revenue in year t-1 is less than the revenue in t-2, and 0 otherwise; and stock performance (RET), measured as the raw stock return from the Center for Research in Security Prices (CRSP). These four economic variables are incorporated in the model as stand-alone variables and interacted with the sticky variable: β2. The coefficient of primary interest is the sign and significance of the interactive variable, β5. A positive and significant β5 will support H1. 2.2.2 Measurement of the independent variable: Financial constraints As there is no consensus on which is the best proxy for financial constraint measurement, and the majority of studies use more than one constraint measure as a proxy, I use three different measures in this study. The financial constraint measures used are: SA index (SA); WW index (WW); and the Bodnaruk, Loughran and McDonald (BLM) text-based constraint measure. (i) SA index: The index has been developed by Hadlock and Pierce (2010) using firm size and age, and higher SA indices indicate higher financial constraint. The SA index is calculated as follows: SAit = −0.737 SIZEit + 0.043 (SIZEit) 2 − 0.040 AGEit (2.2) where SIZE is the natural log of total assets, and AGE is the number of years the firm is listed on Compustat. 23 (ii) WW index: WW index has been constructed by Whited and Wu (2006), and higher WW indices mean higher financial constraint. The index is composed of six components and is calculated as follows: WWit = − 0.091 CFit − 0.062 DIVPOSit + 0.021 TLTDit − 0.044 LNTAit + 0.102 ISGit − 0.035 SGit (2.3) where CF is cash flow (Compustat data item IB plus DP) divided by total assets (Compustat data item AT), DIVPOS is a dummy variable equal to 1 if the firm pays dividends (Compustat data item DVC plus DVP) and 0 otherwise, TLTD is long-term debt (Compustat data item DLTT) divided by total assets, LNTA is the natural log of total assets, ISG is the firm’s three-digit SIC code industry annual sales growth, and SG is the firm’s annual sales growth. (iii) Bodnaruk et al. (2015) text-based financial constraint measure (BLM): Bodnaruk et al. (2015) developed a list of 184 constraining words from all 10-K filings. The commonly used constraining words from their list include required, obligations, impairment, covenants, requirements, permitted, comply, imposed, and the index uses the percentage of constraining words as a measure of financial constraint. Bodnaruk et al. (2015) show that the more managers are concerned about future financial constraints, the more they will disclose through the text of 10-K. Higher BLM values indicate higher financial constraint. The underlying notion of traditional accounting-based measures of financial constraint is that larger firms are less likely to be financially constrained; whereas, owing to financial meltdown, larger and older firms can become financially constrained (Bodnaruk et al., 2015). This inherent shortcoming of traditional measures, may result in misclassifying financially constrained firms arising from previously unconstrained firms. Thus, the text-based measure can identify the financially constrained firms more accurately. 24 2.2.3 Business cycle measure Consistent with prior studies (e.g., DeStefano, 2004; Jenkins et al., 2009; Kim et al., 2011), this study uses the well-accepted measure based on the National Bureau of Economic Research (NBER) business cycle classification in identifying different states of economic activities (i.e., expansion and contraction).3 Periods of expansion begin at the trough date and end at the peak date, and periods of recession begin at the peak date and end at the trough date. The NBER defines a recession (expansion) in terms of significant decline (increase) in economic activity spread across the economy, lasting more than a few months, normally visible in real GDP, real income, employment, industrial production, and wholesale-retail sales. Thus, unlike the employment rate and industrial production as proxies for the business cycle, the NBER’s business cycle captures more comprehensive information regarding economic activities. I also consider the OECD Composite Leading Indicators as an alternative proxy to define economic recession. According to this time series, the recession begins at the midpoint of the period of the peak and ends at the midpoint of the period of the trough. 2.2.4 Sample selection and descriptive statistics Financial data are collected from Compustat, whilst the stock return data are collected from the CRSP for the years 1976 to 2016. I deliberately choose a long sample period to provide a richer analysis of the cost stickiness behaviour. Panel A, Table 2.1, illustrates the sample selection process, which follows the process used by Anderson et al. (2003). I begin with a total sample of 462,735 firm-year observations during 1976–2016. I then exclude firm-year observations from the regulated industries (two-digit SIC code 48-49), financial institutions (two-digit SIC 3For details about NBER turning points dates of the U.S. economy http://www.nber.org/cycles/cyclesmain.html http://www.nber.org/cycles/cyclesmain.html 25 codes 60-69) and any duplicate values. This process eliminates a total of 170,800 firm-year observations. After excluding observations based on missing data on SALE and SG&A for the current and previous periods, zero SALE and SG&A values, negative SALE and SG&A values, as well as SALEmedian FC (financially constrained group), and zero otherwise (unconstrained group). I the reran the regressions for all three FC variables using OLS and FE methods using the dummy instead of the continuous FC variables. Untabulated results reveal significant and positive coefficients for β5 for WW (FE coefficient = 0.047; t statistic = 2.12; p < 0.05) and BLM (FE coefficient = 0.072; t statistic = 2.73; p < 0.01) measures in both OLS and FE: results that are consistent with those reported in Table 2.3 (Panel A and B). However, the coefficients are insignificant for SA measure. 31 PANEL A: OLS regression results SA WW BLM SA WW BLM (1) (2) (3) (4) (5) (6) [2.09] [2.03] [-0.87] [4.34] [3.06] [-3.96] Industry Yes Yes Yes Yes Yes Yes Year Yes Yes Yes Yes Yes Yes Observations 183,173 147,701 69,958 124,839 101,125 56,455 Adj. R-squared 0.47 0.45 0.50 0.52 0.50 0.54 Note: Panel A reports the results from the OLS regression of association between financial constraints and SG&A cost behaviour. Robust t-statistics are in brackets and are based on standard errors that are clustered by firm. *** p<0.01, ** p<0.05, * p<0.10. Refer to Appendix A for variable definitions. PANEL B: Fixed effect regression results SA WW BLM SA WW BLM (1) (2) (3) (4) (5) (6) β1: LN(SALEt/SALEt-1) 0.481*** 0.500*** 0.693*** 0.519*** 0.532*** 0.758*** [31.35] [40.33] [33.64] [23.12] [30.68] [35.63] β2: DECDUM * LN(SALEt/SALEt-1) 0.043 [1.63] -0.011 [-0.58] -0.274*** [-6.82] -0.058 [-1.38] -0.054* [-1.73] -0.320*** [-6.84] β3: FC 0.020*** -0.001 -2.283*** 0.010*** -0.010 -0.135 [15.68] [-0.09] [-4.18] [5.00] [-0.53] [-0.24] β4: FC * LN(SALEt/SALEt-1) -0.062*** -0.450*** -4.600 -0.054*** -0.412*** -12.159*** [-11.27] [-10.59] [-1.56] [-7.18] [-7.11] [-3.93] β5: FC * β2 0.076*** 0.492*** 11.272** 0.037*** 0.267*** 18.571*** [8.32] [7.40] [2.00] [2.97] [2.85] [3.04] β6:SUDEC * β2 - - - 0.107*** 0.108*** 0.117*** [8.19] [7.58] [6.20] β7: AIN * β2 - - - -0.046*** -0.054*** -0.039*** [-9.02] [-9.63] [-4.38] β8: EIN * β2 - - - 5.094*** 5.977*** 4.394** [5.80] [5.64] [2.49] β9: RET * β2 - - - 0.258** 0.320*** 0.163 [2.56] [2.84] [1.15] β10: SUDEC - - - -0.056*** -0.054*** -0.050*** [-36.88] [-33.36] [-24.02] β11: AIN - - - 0.018*** 0.014*** 0.027*** [10.45] [7.13] [9.97] β12: EIN - - - 2.565*** 3.178*** 2.703*** [13.38] [17.44] [5.85] β13: RET - - - -0.306*** -0.257*** -0.321*** [-20.64] [-15.22] [-16.00] Constant 0.088*** 0.033*** 0.036*** 0.033*** 0.004 -0.007 [22.59] [8.75] [9.40] [4.74] [0.74] [-1.22] Industry No No No No No No Year Yes Yes Yes Yes Yes Yes Observations 183,173 147,701 69,958 124,839 101,125 56,455 Adj. R-squared 0.42 0.40 0.43 0.47 0.46 0.48 Note: Panel B reports the results from the FE of association between financial constraints and SG&A cost behaviour. Robust t-statistics are in brackets and are based on standard errors that are clustered by firm. *** p<0.01, ** p<0.05, * p<0.10. Refer to Appendix A for variable definitions. 32 2.3.3 Is it “Good” or “Bad” less cost stickiness? I now discuss results on whether the less cost stickiness exhibited by financially constrained firms reflect “good” or “bad” less cost stickiness (test of three sub-hypotheses developed in section 2.1). To test H1EM, I split the sample observations into ‘loss avoidance’ (AVOID) and ‘earnings decrease’ (EDEC) groups. The AVOID group consists of firm-year observations with annual earnings deflated by market capitalization of shareholders’ equity at prior year end, in the interval [0, 0.01] (both inclusive). The EDEC group consists of firm-year observations with changes in annual earnings deflated by market capitalization of shareholders’ equity at prior year end in the interval [0, 0.01] (both inclusive). Panel A1 of Table 2.4 presents the results for the AVOID test. The coefficient on β5 is positive and significant for the subsample without incentives to avoid losses (AVOID=0) (e.g., the coefficient is 0.034, p<0.01 for the SA model in column 1), and insignificant for the group with incentives to avoid losses (AVOID=1). Table 2.4 Panel A2, reports the results for the EDEC test. The coefficient on β5 is positive and significant for the subsample without incentives to avoid earnings decrease (EDEC=0) (e.g., the coefficient is 0.03, p<0.01 for SA model in column 1), and insignificant for the group with incentives to avoid earnings decrease (EDEC=1). Taken together the results for the earnings management incentives context reveal that the less cost stickiness exhibited by financially-constrained firms is “good” less cost stickiness. To test H1AP, I split the sample observations into two groups, i.e., a pervasive (high) agency problem group and a minimal (low) agency problem group. I use two different proxies to measure agency problem i.e. (i) CEO fixed pay (FXP) and (ii) total asset growth (TAGROWTH). According to Chen et al. (2012) and Kanniainen (2000), a manager’s empire-building incentives can be restrained by paying fixed salaries. Banker et al. (2011) document that fixed (cash) compensation is used to penalize wasteful spending on SG&A costs. FXP is the ratio of salary plus bonus divided 33 by total compensation during the year. Firm-year observations with FXP>= median FXP constitutes the minimal agency problem group. The pervasive agency problem group consists of observations with FXP=median TAGROWTH. The ‘low’ agency problem group consists of firm-year observations with TAGROWTHMedian) High Agency (Median) High Agency (Median) High Agency (Median) Low Agency (Median) Low Agency (Median) (1) (2) (3) (4) (5) (6) β1: LN(SALEt/SALEt-1) 0.434*** 0.578*** 0.411*** 0.600*** 0.693*** 0.799*** [10.50] [23.48] [13.53] [31.85] [11.19] [33.25] β2: DECDUM * LN(SALEt/SALEt-1) 0.035 -0.260*** 0.079** -0.222*** -0.223*** -0.463*** [0.64] [-2.66] [2.00] [-3.01] [-2.69] [-4.50] β3: FC 0.007*** 0.014*** 0.043*** 0.078*** 0.547 3.659*** [4.50] [7.50] [4.19] [6.20] [0.73] [4.90] β4: FC * LN(SALEt/SALEt-1) -0.053*** -0.044*** -0.573*** -0.303*** -14.303 -15.384*** [-3.91] [-5.33] [-5.41] [-4.84] [-1.63] [-4.34] β5: FC * β2 0.034* 0.038 0.436*** 0.332 19.214* 13.557 [1.94] [1.34] [3.28] [1.53] [1.68] [0.94] ECONVAR * β2 Yes Yes Yes Yes Yes Yes ECONVAR Yes Yes Yes Yes Yes Yes Constant 0.032*** 0.035*** 0.026*** 0.022*** -0.017 -0.048*** [3.89] [4.49] [3.24] [3.05] [-1.47] [-5.04] Industry No No No No No No Year Yes Yes Yes Yes Yes Yes Observations 59,568 65,115 48,298 52,565 24,794 26,735 Adj. R-squared 0.37 0.50 0.36 0.47 0.39 0.51 Note: Panel B2 reports the results from the FE regression of association between financial constraints and SG&A cost behaviour in low and agency problem firms (proxied by total asset growth). Robust t-statistics are in brackets and are based on standard errors that are clustered by firm. *** p<0.01, ** p<0.05, * p<0.10. Refer to Appendix A for variable definitions. 38 PANEL C: SG&A_FV context SA SA WW WW BLM BLM Low SG&A_FV High SG&A_FV Low SG&A_FV High SG&A_FV Low SG&A_FV High SG&A_FV (1) (2) (3) (4) (5) (6) β1: LN(SALEt/SALEt-1) 0.526*** 0.588*** 0.569*** 0.595*** 0.807*** 0.825*** [16.65] [19.79] [25.19] [24.34] [32.00] [25.46] β2: DECDUM * LN(SALEt/SALEt-1) -0.082 -0.096* -0.094** -0.098** -0.349*** -0.425*** [-1.44] [-1.81] [-2.37] [-2.34] [-6.03] [-5.67] β3: FC 0.007*** 0.011*** 0.029** 0.073*** 2.712*** 1.809** [3.43] [6.28] [2.37] [5.89] [3.99] [2.19] β4: FC * LN(SALEt/SALEt-1) -0.049*** -0.064*** -0.323*** -0.590*** -15.293*** -12.208*** [-4.63] [-6.30] [-4.19] [-7.31] [-4.02] [-2.65] β5: FC * β2 0.027 0.066*** 0.169 0.527*** 24.417*** 20.272** [1.56] [3.86] [1.36] [3.82] [3.22] [2.06] ECONVAR * β2 Yes Yes Yes Yes Yes Yes ECONVAR Yes Yes Yes Yes Yes Yes Constant 0.030*** 0.047*** 0.017** 0.026** -0.043*** 0.002 [3.41] [6.45] [2.17] [2.18] [-4.44] [0.14] Industry No No No No No No Year Yes Yes Yes Yes Yes Yes Observations 48,473 49,481 43,557 32,999 29,599 15,490 Adj. R-squared 0.49 0.59 0.48 0.59 0.55 0.58 Note: Panel C reports the results from the FE regression of association between financial constraints and SG&A cost behaviour in low and high future value generating SG&A firms. Robust t-statistics are in brackets and are based on standard errors that are clustered by firm. *** p<0.01, ** p<0.05, * p<0.10. Refer to Appendix A for variable definitions. 39 Finally, in order to test H1VSGA, I categorise the sample into high (above median) versus low (equal and below median) future value-creating SG&A groups, based on the industry-specific future value creation potential of SG&A costs, as per the value reported in Table 2 of Banker et al. (2011). I present the results in Panel C of Table 2.4. The coefficient on β5 is positive and insignificant for the low future value subsample (coefficient = 0.027; t statistic = 1.56), but positive and significant for the high future value subsample (coefficient = 0.066; t statistic= 3.86; p<0.01). Similar results are found using the WW model. This finding might be consistent with Caggese et al. (2019), who find inefficient employee-firing decisions by financially constrained firms. This finding could also be consistent with the fact that constrained firms are exposed to high opportunity costs associated with retaining unutilised resources, compared to their non-constrained counterparts. When firms suffer from financial constraints, they may have fewer options when deciding which kind of SG&A costs to decrease, as their main objective becomes reducing the financial burden and surviving the constraint period, to bounce back later. Therefore, from a firm survival point of view, such behaviour can be a manifestation of “good” less cost stickiness; however, from an efficiency point of view, such less stickiness could be construed as “bad”. This finding also confirms empirically the theoretical conjecture by Musso and Schiavo (2008) that, to overcome financial constraint, a firm will eliminate long term investments, and this may have an adverse effect on its long-term growth prospects. Results using the BLM measure, however, show positive and significant coefficients for both low (coefficient = 24.417; p<0.01) and high (coefficient = 20.272; p<0.05) future value-creating groups, although the absolute magnitude of the coefficient is larger for the low future value-creating group. 40 2.3.4 Financial constraints and cost stickiness during the business cycle To investigate the relation between financial constraint and cost stickiness during the business cycle (expansion and contraction), the sample has been divided into expansion and contraction based on announcements by the NBER. Table 2.5, Panel A, reports regression results using the NBER business cycle measure. The SA model in columns (1) and (2) shows regression results for the expansion and contraction periods, respectively. The variable of interest is β5, which is significant during both expansion (coefficient = 0.020; t statistic = 1.68, p < 0.10) and contraction periods (coefficient = 0.064; t statistic = 2.35; p < 0.05). I find consistent results using WW and BLM. I also use the OECD composite leading indicator (CLI) as an alternative measure of the business cycle for testing H2. Reported results in Panel B of Table 2.5 show significant positive coefficients on β5 for the WW and BLM models (columns 3 to 6) for both the economic expansion and economic contraction periods. The results support the premise of H2, that during expansionary periods when information asymmetry is low, investors charge higher returns from financially constrained firms and, thus, could restrict financially constrained firms from accessing external finance. On the other hand, during contractionary periods investors have limited money to invest, and are likely to be more sceptical and selective. As a result, it becomes more difficult and time consuming for constrained firms to find external finance on favourable terms. Therefore, constrained firms minimize costs by reducing unutilised resources. This result refutes Chevalier and Scharfstein’s (1996) finding that during contraction, when firms have low cash flow and face the difficulty of raising finance externally, financially-constrained supermarkets increase prices compared to their unconstrained counterparts, to increase short-term profit. Instead my findings are aligned with Liu et al.’s (2017b) conclusion that, if financially constrained firms increase prices during economic contraction, it 41 would create the opportunity for non-constrained competitors to engage in a predatory pricing strategy, which would hinder competition, resulting in constrained firms being driven out of the market. However, although the absolute magnitude of less stickiness is higher during contraction periods compared to expansion periods in all three models, and across both business cycle measures, the difference in coefficients is not statistically different between the two sub-periods (z-test of difference in coefficients is reported in Panel C, Table 2.5). 42 TABLE 2.5: Business Cycle, Financial Constraint and Cost Stickiness PANEL A: NBER SA SA WW WW BLM BLM Expansion Contraction Expansion Contraction Expansion Contraction (1) (2) (3) (4) (5) (6) β1: LN(SALE t /SALE t-1 ) 0.572*** 0.596*** 0.584*** 0.595*** 0.791*** 0.849*** [25.98] [12.45] [35.20] [15.34] [41.68] [9.84] β2: DECDUM * LN(SALE t /SALE t-1 ) -0.112*** [-2.89] -0.022 [-0.25] -0.078*** [-2.66] -0.068 [-1.06] -0.339*** [-7.41] -0.417*** [-3.18] β3: FC 0.009*** 0.005* 0.044*** 0.025 2.267*** 0.714 [6.68] [1.83] [5.12] [1.30] [4.50] [0.50] β4: FC * LN(SALE t /SALE t-1 ) -0.047*** [-6.33] -0.029* [-1.73] -0.366*** [-6.59] -0.215 [-1.58] -12.319*** [-4.40] -25.963** [-2.13] β5: FC * β2 0.020* 0.064** 0.159* 0.452** 22.401*** 29.266* [1.68] [2.35] [1.74] [2.30] [3.66] [1.71] β6:SUDEC * β2 0.078*** 0.143*** 0.072*** 0.132*** 0.085*** 0.114*** [5.54] [5.68] [4.64] [4.73] [3.92] [3.46] β7: AIN * β2 -0.048*** -0.039*** -0.054*** -0.045*** -0.039*** -0.043*** [-9.45] [-3.67] [-9.45] [-3.91] [-4.27] [-3.00] β8: EIN * β2 3.894*** 2.812* 4.209*** 4.135** 1.916 4.648 [4.57] [1.65] [4.12] [1.99] [1.26] [1.13] β9: RET * β2 0.397*** 0.111 0.450*** 0.225 0.287* 0.095 [3.70] [0.57] [3.64] [1.07] [1.83] [0.40] β10: SUDEC -0.059*** -0.055*** -0.058*** -0.054*** -0.054*** -0.052*** [-38.47] [-14.53] [-35.08] [-13.17] [-25.27] [-8.69] β11: AIN 0.008*** 0.016*** 0.007*** 0.016*** 0.012*** 0.022*** [8.05] [6.22] [5.91] [5.38] [8.01] [6.20] β12: EIN 0.647*** 0.521*** 0.825*** 0.686** 0.637*** 0.479 [7.31] [2.61] [6.57] [2.32] [4.02] [0.85] β13: RET -0.352*** -0.280*** -0.297*** -0.248*** -0.357*** -0.322*** [-20.76] [-7.62] [-15.23] [-5.93] [-15.61] [-6.40] Constant 0.047*** 0.051 0.035*** 0.039 -0.028** -0.090** [4.82] [1.49] [3.50] [1.19] [-2.31] [-2.48] Industry No No No No No No Year Yes Yes Yes Yes Yes Yes Observations 104,294 20,545 84,120 17,005 48,615 7,840 Adj. R-squared 0.52 0.50 0.50 0.49 0.54 0.50 Note: Panel A reports the results from the FE regression of association between financial constraints and SG&A cost behaviour during expansion and contraction period using NBER. Robust t-statistics are in brackets and are based on standard errors that are clustered by firm. *** p<0.01, ** p<0.05, * p<0.10. Refer to Appendix A for variable definitions. 43 PANEL B: OECD SA SA WW WW BLM BLM Expansion Contraction Expansion Contraction Expansion Contraction (1) (2) (3) (4) (5) (6) β1: LN(SALE t /SALE t-1 ) 0.605*** 0.552*** 0.586*** 0.589*** 0.797*** 0.793*** [22.48] [19.77] [28.59] [27.41] [31.42] [29.41] β2: DECDUM * LN(SALE t /SALE t-1 ) -0.129** [-2.52] -0.052 [-1.06] -0.089** [-2.21] -0.072** [-2.02] -0.354*** [-6.20] -0.339*** [-5.72] β3: FC 0.006*** 0.010*** 0.031*** 0.051*** 2.455*** 1.586** [3.74] [6.19] [2.76] [4.57] [3.80] [2.28] β4: FC * LN(SALE t /SALE t-1 ) -0.033*** [-3.63] -0.055*** [-5.91] -0.326*** [-4.75] -0.361*** [-4.96] -12.615*** [-3.37] -13.699*** [-3.48] β5: FC * β2 0.022 0.046*** 0.229* 0.278** 16.893** 24.635*** [1.42] [2.98] [1.94] [2.41] [2.30] [3.04] β6:SUDEC * β2 0.104*** 0.094*** 0.099*** 0.091*** 0.107*** 0.099*** [6.02] [5.52] [5.22] [4.86] [4.57] [3.83] β7: AIN * β2 -0.048*** -0.046*** -0.055*** -0.051*** -0.048*** -0.035*** [-7.72] [-7.55] [-7.45] [-7.71] [-5.12] [-2.89] β8: EIN * β2 4.126*** 3.453*** 4.508*** 4.068*** 5.357*** 1.171 [3.67] [3.49] [3.08] [3.70] [2.60] [0.59] β9: RET * β2 0.003 0.487*** 0.086 0.534*** -0.287 0.651*** [0.02] [3.68] [0.54] [3.54] [-1.60] [3.51] β10: SUDEC -0.056*** -0.061*** -0.055*** -0.061*** -0.049*** -0.057*** [-28.72] [-30.72] [-25.99] [-28.24] [-18.99] [-19.05] β11: AIN 0.006*** 0.012*** 0.003** 0.013*** 0.006*** 0.022*** [4.27] [9.31] [2.18] [8.12] [3.18] [10.74] β12: EIN 0.683*** 0.592*** 0.941*** 0.698*** 0.729*** 0.544** [5.59] [5.48] [5.23] [4.36] [3.67] [2.36] β13: RET -0.363*** -0.335*** -0.329*** -0.27