Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? The Case of Transition Economy

Front. Bus. Res. China 2015, 9(2): 207 242 DOI 10.3868/s070-004-015-0010-8 RESEARCH ARTICLE Lu Dai, Qingbin Meng, Maozhu Sun Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? The Case of Transition Economy Abstract This study observes and explores a puzzle in Chinese firms whereby both cash holdings and short-term debt simultaneously account for more than 20% of total assets for at least two consecutive years over the sample period. This phenomenon conflicts with the principle of corporate value maximization, and is not clearly explained by the classical theories in corporate finance. Based on the implications in the extant literature and discussions of institutional constraints of the transition economy in China, this paper develops four hypotheses that are involved with agency conflicts between the largest shareholders and creditors and the formation of this puzzling financial structure. The empirical analyses suggest that the largest shareholders with tunneling motives seek to hold more cash to serve their private interests and/or the consequent operational deficit of the listed corporations. To the ends, these corporations tend to manage the timing of short term debt financing to increase cash reserves temporarily at the end of year. Essentially, greater cash holdings on the balance sheet of these corporations related with the puzzle become a misleading signal for potential creditors, possibly contributing to the refinancing of short-term debt of these listed firms for the following year. Hence, the puzzling financial structure is connected with the timing of debt financing and adverse selection of creditors. This study enriches the stream of literature on cash holdings and debt maturity, and provides new evidence on the impact of agency problems of the largest shareholders on the association between cash holdings and debt maturity in the context of a transition economy. Received December 30, 2014 Lu Dai a, Qingbin Meng b ( ), Maozhu Sun c School of Business, Renmin University of China, Beijing 100872, China E-mail: a dailu@rbs.org.cn, b mengqingbin@rbs.org.cn, c sunmaozhu@rbs.org.cn

208 Lu Dai, Qingbin Meng, Maozhu Sun Keywords cash holdings, short-term debt, agency conflicts, tunneling, largest shareholders, puzzle, financial policy 1 Introduction While some recent literature has noted a trend in the shortening of debt maturity in U.S. corporations over the past three decades (Custódio, Ferreira, and Laureano, 2013), the other group of studies reveal a growing trend in the increase of cash holdings in U.S. corporations (Bates, Kahle, and Stulz, 2009). Given the potential interconnection of a firm s financial policies, how debt maturity and cash holdings are jointly decided deserves academic attention (Harford, Klasa and Maxell, 2014). The extant literature in corporate finance provides plentiful evidence on the determinants of either cash reserves or debt maturity of listed corporations around the world (e.g., Faulkender and Wang, 2006; Bates, Kahle and Stulz, 2009; Denis and Sibilkov, 2010; Duchin et al., 2010; Chen, Li, Xiao and Zou, 2014; Chen, Chen, Schipper, Xu and Xue, 2012). This implies that the association between cash holdings and debt maturity across firms in different countries may be contingent on the institutional settings. However, what happens to corporations in transition economies when taking joint decisions on debt maturity and cash reserves is not clear in the existing corporate finance literature. This study observes and explores a unique phenomenon that so far has not been examined with empirical evidence in corporate finance. Specifically, using a sample of 249 Chinese listed firms between 1998 and 2013, we find that short-term debt and cash reserves simultaneously account for more than 20% of a firm s total assets for at least two consecutive years. However, classical theories in corporate finance are not able to offer much insight into this phenomenon. At first glance, a corporation may choose to hold heavy short-term debt with abundant cash reserves to relax liquidity concerns or to reduce refinancing risks (Harford, Klasa and Maxell, 2014; Berardino, 2012). However, both trade-off and pecking order theories suggest that a corporation should repay excess debt by dipping into overfull cash reserves in order to reduce the opportunity costs or agency costs of large cash reserves (Denis and Sibilkov, 2010). Therefore, the persistence of both greater cash holdings and short-term debt for a long time conflicts with the principle of corporate value maximization. Given the wedge between the classic theories and this puzzle, this paper is aimed

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 209 to explore the motives and factors behind it. If the puzzle breaches corporatelevel interest, agency problems may be an important factor. A preliminary investigation of this puzzling financial structure shows that short term debt in the firms related with the puzzle is mainly financed by commercial banks in China. As well, 95% of the firms related with the puzzle have a highly concentrated shareholding structure. Namely the financial policy of most firms related with the puzzle is controlled by their large shareholders. Thus, agency conflicts between the largest shareholders of listed companies and creditors may be a key clue for the explanation of this puzzle. Based on implications in the extant literature and discussion of institutional constraints on the transition economy in China, we develop four hypotheses. We predict that the tunneling activities of the largest shareholders are an important reason why the puzzle persists beyond the maximization of corporation value. With the tunneling motive of the largest shareholders, the formation of this financial structure of the firms related with the puzzle involves heavy financing of short term debt to serve private interest of the largest shareholders. To reduce the concern of bankers on risk of bad debt, these firms try to make an arrangement of the timing of short term debt financing to increase cash holdings temporarily to mislead potential creditors. Hence, the formation of this puzzling financial structure may be involved with adverse selection of commercial bankers to provide shot term loans. We examine these predictions empirically through regression analysis. The results confirm our hypotheses. These findings reveal the impact of agency problems on cash holdings and short-term debt, and are consistent with the perspective of agency theory on determinants of cash holdings and debt maturity. Overall, this paper belongs to the broad literature on the effects of corporate governance on financial policy and contributes specifically to the stream of literature on the association of cash holdings and debt maturity. Currently, existing studies fail to identify how agency problems between creditors and large shareholders influence debt maturity or cash holdings for a firm. Therefore, this research explores different explanations from classical theories in corporate finance on the relationship between cash holdings and debt maturity. In particular, this research reveals that greater cash reserves financed from short term debt might be a misleading signal to creditors. The management of the timing of short term debt financing of listed corporations might result in adverse selection for

210 Lu Dai, Qingbin Meng, Maozhu Sun bankers. Based on the tactic of short term debt financing, this paper finds that the tunneling of the largest shareholders produces agency conflicts with bankers. Traditionally, the tunneling of the largest shareholders is considered to provoke agency conflicts with minor shareholders. However, this paper suggests new implications beyond the existing evidence that even short term debt cannot become a governance mechanism to constrain the agency problems of the largest shareholders. The rest of this paper is organized as follows. Section 2 reviews extant literature. Section 3 discusses the institutional environment and develops hypotheses. Section 4 describes the sample selection, data collection and research design. Section 5 shows and discusses research methods and the main empirical results. Section 6 concludes with primary findings and research limitations. 2 Literature Review 2.1 Theoretical Lenses and Empirical Evidence on Cash Holdings and Debt Maturity In terms of mainstream theories of corporate finance, static trade-off theory and pecking-order theory provide different interpretations for the financial policy of corporations in general. According to static trade-off theory, cash holdings and capital structures are dependent on the trade-off between costs and benefits based on the principle of corporation value maximization. In terms of the relationship between short-term debt and cash holdings, static trade-off theory suggests that companies should accumulate more cash if they have to repay excessive short-term debt. Therefore, short-term debt and cash holdings are positively correlated (Stulz, 1990; Jensen, 1986). By contrast, according to pecking-order theory, information asymmetry between companies and investors raises the cost of external financing. Companies with growth opportunities choose to accumulate cash reserves first to finance investment projects and consider debt financing secondarily (Myers and Majluf, 1984). Numerous empirical studies have produced evidence to support either static trade-off theory or pecking order theory. In the context of emerging markets, Arslan, Florackis and Ozkan (2006) report that the relationship between cash holdings and short-term debt is non-monotonic. Specifically, they are negatively

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 211 correlated with each other when leverage is low; the relationship is reversed if short-term debt increases to higher levels. Obviously, the above findings actually support static trade-off theory. However, other empirical studies have supported pecking-order theory. Illustratively, Harford, Klasa and Maxell (2014) find that firms mitigate refinancing risks by increasing their cash holdings and saving cash from cash flows. The markedly shortened maturity of firms long-term debt explains a large fraction of the increase in cash holdings over time. Their findings imply refinancing risk as a key determinant of cash holdings and highlight the interdependence of a firm s financial policy decisions. Similarly, Guney, Ozkan and Ozkan (2007) find that cash holdings and short-term debt are positively correlated among Japanese, French, German and British companies from 1983 to 2000. Almeida, Campello and Weisbach (2004) find that companies with tighter financial constraints hold more cash for future investment opportunities. Of course, pecking-order theory fails to perfectly explain why companies hold large amounts of cash for a long time, because excessive cash reserves result in higher opportunity costs. On the other hand, agency theory offers a new lens to scrutinize the motive of excessive cash holdings for firms with agency problems. Jensen (1986) focuses on shareholder-manager agency conflicts and predicts that managers have incentives to use surplus cash flows to extract private benefits without effective external monitoring. Some empirical studies provide the evidence consistent with this argument (e.g., Dittmar and Mahrt-Smith, 2007; Harford et al., 2008; Harford, 1999; Myers and Rajan, 1998). Perks, business corruption, desire for empire building, and other such agency problems associated with executives of a firm are important drivers for greater cash holdings. Harford (1999) investigates how companies with excessive cash reserves may engage in value-detracting investments and M&A activities, where they face severe agency conflicts. Also, agency theory explains why companies maintain higher short-term leverage over a long period of time. Flannery (1986) analytically proves that companies with poor performance try to imitate those with high performance. Specifically, a mixed equilibrium emerges in bond markets if the difference between the total cost of short-term debt refinancing and the cost of long-term debt financing is near zero. In that case, two types of firms, one with better performance and one with worse performance, issue bonds with the same

212 Lu Dai, Qingbin Meng, Maozhu Sun maturity. Otherwise, there would be a separate equilibrium whereby firms with high performance would always issue short-term bonds while those with poor performance would always issue long-term bonds. This conclusion implies that short-term debt could be chosen by companies with both good and poor performance. Empirically, Hennessy (2004) notes that excessively high leverage resutl in that managers choosing inefficient or myopic investments with low returns. Overall, corporate cash holdings are mainly determined by operational determinants, governance arrangements and agency conflicts (Chen et al., 2012). With regards to operational determinants, exsiting literature indicates that fhigher transaction costs from converting noncash financial assets to cash force firm to hold more cash in order to facilitate transactions. Similarly, firms with better investment opportunities or tighter financial constraints prefer to hold more cash. These factors have been examined in extant studies (e.g. Opler, Pinkowitz, Stulz and Williamson, 1999; Bates, Kahle, and Stulz, 2009; Duchin et al., 2010). The agency considerations literature (e.g., Dittmar, Mahrt-Smith, and Servaes 2003; Pinkowitz, Stulz, and Williamson, 2006; Kalcheva and Lins, 2007; Dittmar and Mahrt-Smith, 2007; Harford, Mansi, and Maxwell, 2008) suggests that ample cash holdings facilitate overinvestment in undesirable projects or outright misappropriation. This implies that a series of governance-related determinants, misaligned incentives for corporate insiders and outsiders, can cause corporate cash holdings to deviate from levels that would be observed if only operational determinants were at work. 2.2 The Impact of Institutional Settings on Cash Holdings and Debt Maturity The literature review shows that transactional costs and precautionary motives in either static trade-off theory or pecking order theory have been identified as key drivers for greater cash holdings of firms (see Bates et al., 2009; Opler et al., 1999). Shareholder-manager agency conflicts have also been shown to be important determinants of either cash reserves or debt maturity (Jensen, 1986; Harford, 1999). In terms of the impact of institutional settings on cash reserves, in countries with weak protection of shareholder rights, Dittmar et al. (2003), Ferreira and Vilela (2004) all report that firms tend to hold more cash to mitigate tighter

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 213 financial constraints and reduce higher financing costs, if property right protection is poor and capital market is undeveloped. Kusnadi, Ullah and Wei (2011) report that investor protection has a first-order effect on corporate cash holdings and firms hold less cash in response to cash flow increases when investor protection is strong. Caprio, Faccio and McConnell (2013) find that measures of political corruption are negatively related to corporate cash holdings. Chen et al. (2014) find that firms hold less cash when local government quality is high. On the other hand, Guney, Ozkan and Ozkan (2007) report that companies hold more cash if legal protections for creditors have been strengthened. Frésard and Salva (2010) find strong evidence that cash reserves are substantially larger for foreign firms listed on the US exchanges than for their American counterparts. Furthermore, they show that excessive cash holdings stem not only from the strength of the U.S. legal rules and disclosure requirements, but also from greater pressure from investors in American capital markets. In a word, the evidence of the institutional impact on firms cash holdings is mixed in the existing literature on corporate finance. However, few studies in the existing literature identify how the agency problems between creditors and corporations influence debt maturity. In fact, agency problems are deeply influenced by institutional factors. Pinkowitz and Williamson (2001) record that the banking system in Japan puts pressure on Japanese companies to hold more cash for a longer time than American and German companies. This implies that Japanese bankers bear a monopoly advantage in debt markets and extract rents from firms with debt financing. Agarwal and Elston (2001) find similar rent-seeking behavior from commercial banks in the German banking system. Moreover, if firms rely heavily on debt financing, the influence of institutional forces on debt maturity and cash holdings has been overlooked by prior studies. In the context of transition economies, the effect of debt maturity on firms cash holdings may have different mechanisms, given the differences in law enforcement and government quality (Chen et al., 2014). 3 The Puzzle and Hypotheses Development The literature review implies that financing costs, refinancing risks, growth opportunities, agency problems and institutional constraints have some

214 Lu Dai, Qingbin Meng, Maozhu Sun explanatory power for greater cash holdings or more short-term debt. However, the reasons why they simultaneously persist over a long period of time are not addressed in the extant corporate finance literature. This phenomenon might be a breach of the interpretations and predictions of classical theories. The motives for firms to keep this peculiar structure of simultaneous greater cash holdings and short term debt deserves empirical investigation. According to pecking-order theory, a firm might maintain excessive cash reserves and short term debt financing to prepare for investment opportunities. But why would larger cash reserves not be invested for a long time, given the opportunity costs of excessive cash reserves? According to static trade-off theory, firms need to reduce financial risk by partly repaying heavier short term debt. Why have these firms not mobilized their large cash holdings to reduce short term debt? Overall, the phenomenon simply outwardly conforms to the principle of match between current assets and current liabilities. It does not comply with either static trade-off or pecking order theory. Even the agency problems between shareholders and managers cannot directly and clearly answer for this puzzle. Again, there is no empirical investigation to give insight into it. We find that 95% of the companies with large cash reserves and short-term debt have concentrated shareholding structures where the average holdings percentage of largest shareholders is at level of 37.42%. Furthermore, 51% of firms in this case are state-owned, and 49% are controlled by private shareholders. Given the role of large shareholders in corporate governance, agency conflicts between shareholders and creditors are possibly a key clue to untangling this puzzle. 3.1 China s Banking System According to China s Generally Accepted Accounting Principles (GAAP), the short term debt in firms balance sheets is completely financed from bank loans. Thus, commercial bankers are key actors in this puzzle. As a developing market, the banking system in China is an indispensable institutional influence on the financial policy of listed firms. In this era of liberalization and reform, most commercial banks in China are state-owned and endowed with monopoly operations. The monopoly advantage protects state-owned commercial banks from improving their poor risk management practices in any significant way. As

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 215 a result, these banks seem quite vulnerable to adverse selection in lending contracts with firms of higher solvency risks. At the end of last century, four big state-owned commercial banks had accumulated tremendous amounts of bad debt which was spun off to other state owned groups controlled by the central government. From the early 21 st century, China s state-owned commercial banks have been determined to avoid the resumption of a sharp rise in bad debt. The Central government does not wish to bear the notoriously expensive cost of taking over the bad debt of state-owned commercial banks again. China Banking Regulatory Commission (CBRC) has promulgated more stringent regulations on the ratio of bad debt. Aside from the change in rules and supervision, China s entry into the WTO produced huge pressure for the domestic banking system to comply with BASEL II. Emerging bad debt will be one of the biggest challenges for Chinese commercial banks when competing with foreign banks in the domestic market. Hence, it has become more and more important for state-owned commercial banks to control lending risks. One of the measures to control credit risk for state-owned commercial banks is to shorten lending periods as much as possible. As Harford, Klasa and Maxell (2014) and Barclay and Smith (1995) point out, the refinancing of short-term debt increases the transparency of corporations and reduces risk for bankers to some extent, as more frequent reviews of financial information are filed by firms with the need for debt financing. Hence, from 2000 short-term bank loans have become more and more common. This tendency has changed the debt maturity of firms, given that bank lending dominates China s capital market. But Berger, Hasan and Zhou (2009) find that state-owned commercial banks had a relatively large number of inefficient loans from 1994 to 2004. This means that shortened lending cannot screen the risk of adverse selection for state-owned commercial banks. Softened budgetary constraints arising from either political connections or their status as too big to fall, discourage Chinese commercial banks from improving their risk management systems and breaking the vested interests within the banks. During the macro-economic cycles between 1998 and 2004, 2005 and 2009, and 2010 and 2013, several rounds of liquidity surplus or quantitative easing pressured China s commercial banks to augment their lending. Under these conditions, short term lending becomes a major source of commercial bank loans for firms.

216 Lu Dai, Qingbin Meng, Maozhu Sun 3.2 The Private Interests of the Largest Shareholders of Listed Corporations in China Given the features of China s banking system noted above, other key actors in the puzzle are the large shareholders who have strong influence over the financial policy decisions of their listed corporations. As Chen et al. (2014) report, weak protection for property rights makes it possible for controlling shareholders to manage the listed company s cash and, if desired, to divert the cash to their own uses through related-party transactions and similar arrangements (see, e.g., Jian and Wong 2010; Jiang, Lee and Yue, 2010). Therefore, agency problems with the largest shareholders have been more pronounced in China than one in developed nations. 3.3 Hypotheses Development Given this puzzle, an interesting question naturally arises where the listed corporations with the puzzle accumulate such significant cash holdings. Given a preliminary investigation of operational cash flow for the treated sample, shown in Table 2, it does not appear that operational activities have been the dominant source of cash inflow for these corporations. A smaller proportion of operational cashflow to total cash inflow puts pressure on these companies to finance through the banking system. As noted in the literature review, short term lending is preferred by China s banking system than long term lending. The trend toward shortened lending periods from commercial banks compels listed corporations to refinance short term debt to support their operations. Given the dominant role of short term debt in the puzzle, the ways for listed corporations with the puzzle to maintain large cash holdings and short term debt simultaneously for a long time needs be explored. As short term debt need to be repaid within one year, firms are heavily reliant on financing and refinancing of short term debt. Thus, the liquidity of these corporations is an issue for them. As mentioned in the literature review, commercial bankers are vulnerable to adverse selection in the lending supply, due to the underdeveloped capacity for risk management and weak legal protection for creditors. Corporations with the puzzle may chose financing of short term debt to accumulate cash holdings, given the information asymmetry between firms and bankers. Therefore we summarize the first hypothesis as follows.

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 217 Hypothesis 1: Short term debt significantly contributes to the cash reserves of the listed corporations related with the puzzle. The preliminary investigation of the refinancing horizon of the listed corporations with the puzzle shows that the refinancing of short term debt in the second half of the year is significantly higher than those in the first half year, as shown in Table 3. The statistics imply that a feasible way for these corporations with the puzzle to keep greater cash reserves financed from short-term debt is that they acquire loans at the end of the year or in the second half of the year as much as possible. As discussed earlier, short term debt is one of the primary sources of debt financing in the China s banking system. Therefore, these corporations might engage in quite high short term borrowing at the end of year or in second half of the year. More importantly, they temporarily reserve these borrowed funds in bank accounts so that higher levels of cash reserves are disclosed on their balance sheets. The greater cash reserves could be a favorable signal to ease concerns of commercial bankers. Also, the deposit financed from short term borrowing could support new debt financing for the next year by virtue of being a pledge. Hence, the timing of debt financing may not only increase cash reserves, but also become a misleading signal for potential creditors to prevent their recognition of higher financial risk behind the puzzle. In a word, the core trick of these listed corporations related with the puzzle is an arrangement of timing for short term debt financing. A temporary deposit of funds financed through short term debt in the second half of the year increases the cash reserves of the listed corporations that manifest this puzzle. Rather than a precautionary motive, the ultimate ends of this financial policy are to guarantee refinancing of short term debt rather than to reduce liquidity concerns. Greater cash holdings on the balance sheet of the listed corporation may be beneficial for new borrowings for the next period. Otherwise, it is not necessary for these firms to take such financial risks of keeping both high cash reserves and short term debt. In this context, the puzzle becomes a deceptive signal to commercial bankers and traps them into lending decisions characterized by adverse selection. We develop the second and third hypotheses as follows. Hypothesis 2: Listed corporations in the study who acquire debt in the second half of the year increase their cash holdings significantly by the end of the year.

218 Lu Dai, Qingbin Meng, Maozhu Sun Hypothesis 3: Cash reserves at the end of the year significantly contribute to short-term debt refinancing for the following year. Given the tactics for the emergence and duration of the puzzle, we explore why firms prefer this financial policy. As previously discussed, the puzzle cannot be explained by corporate value maximization and is highly risky for the listed firms. This implies that agency conflict might be an important factor to explain the puzzle. In terms of excessive cash holdings, the moral hazard for managers of firms has already been verified as a key motive in the existing literature (e.g., Pinkowitz, Stulz, and Williamson, 2006; Kalcheva and Lins, 2007; Dittmar and Mahrt-Smith, 2007; Harford, Mansi, and Maxwell, 2008). As well, the agency problem of the largest shareholder has been identified as another key factor in recent studies (Liu and Mauer, 2011; Gao, Harford and Li, 2013; Kusnadi, 2011; Liu, Luo and Tian, 2015). Tactically, ample cash holdings might facilitate overinvestment in undesirable related transactions or outright misappropriation by the largest shareholders. Consequently, the largest shareholders of listed corporations related with the puzzle may prefer excessive cash reserves regardless of their sources, if they seek private interest. Furthermore, these cash holdings can support the operations of the listed corporations if the largest shareholders engage in tunneling. The puzzle means that firms simultaneously hold higher cash reserves and short term debt. Short term debt from commercial banks becomes the dominant source for greater cash holdings of these listed firms with the puzzle. Thus, the motives for more short term debt are related with the tunneling behavior of the largest shareholders and/or subsequent operation deficits. Hence, we summarize the fourth hypothesis as follows. Hypothesis 4: The amount of tunneling of the largest shareholders is significantly related with the short term debt of listed corporations related with the puzzle. 4 Sampling and Variable 4.1 Sample Selection and Data Collection In this study, our sampling criteria is that the short-term debt and cash holdings

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 219 of listed companies in China are simultaneously more than 20% of total assets for at least two consecutive years during the period from 1998 to 2013. The threshold of both higher cash holdings and short term debt are referred to in existing studies (e.g. Capiro et al., 2013; Custódio, Ferreira and Laureano, 2013; Adjei, 2013; Chen et al., 2012; Dittmar and Mahrt-Smith, 2007; Opler et al., 1999). For example, Opler et al. (1999) and Mikkelson and Partch (2003) choose a ratio of cash to assets above 15%, 20% and 25% as the criteria of greater cash holdings. Comparatively, the criteria for both higher short term debt and greater cash holdings are referred to in the study by Harford, Klasa and Maxell (2014). Furthermore, the requirement of sampling with at least two consecutive years is to reduce the temporary effect of the puzzle. Given the sampling criteria, we restrict our analysis to non-financial and non-utility firms listed on the Shanghai and Shenzhen exchanges from 1998 to 2013. We retrieved all data from the CSMAR database, the major data supplier for academic research in China. As a result, the criteria identify 249 corporations with available data. The observations account for 880 firm-years which are referred to as treated samples in following analysis. The census of observations from 1998 to 2013 is displayed in Table 1. Table 1 The Census of Observations Based on Firm-Year Data 1998 2005 Observations 2006 2013 Observations 1998 9 2006 47 1999 18 2007 43 2000 40 2008 52 2001 67 2009 64 2002 88 2010 58 2003 86 2011 63 2004 80 2012 61 2005 62 2013 42 Total 880 In addition, we identify matched samples to control for the self-selection effect. The matched samples are selected by the propensity score matching method (PSM, Rosenbaum and Rubin, 1983). We select a logistic regression with the primary factors related to both cash holdings and leverage explored in the existing literature. We add three factors into the logistic regression, including firm size (natural logarithm of total assets), operation performance (Roa), and

220 Lu Dai, Qingbin Meng, Maozhu Sun Market to Book ratio. All these factors are the major considerations for the selection of matched samples. Based on the nearest propensity scoring between the main samples and available comparison samples, we further chose industry matched firms from the set of available comparison samples as the matched samples. 4.2 Variable Definition Following Chen et al. (2012) and Mikkelson and Partch (2003), the ratio of cash holdings (Cash it ) is calculated as a percentage of firm i s reserve of cash and cash equivalents to the firm s noncash part of total assets in year t. Although companies may use lines of credit and cash as substitutes to manage liquidity (Campello et al., 2011) in China, firms access to lines of credit is in general rare, and this feature allows us to conduct a cleaner test of the cash holding decision (Harford, Klasa and Maxell, 2014; Chen et al., 2014). Similarly, the variable of short term debt (Sd it ) in this study is measured as the ratio of short term debt to the net value of total assets from short term debt (Custódio et al., 2013). According to our hypotheses, another primary measurement is the tunneling of the largest shareholders. In the literature of corporate governance, the term tunneling is commonly used to describe the transfer of assets and profits out of firms for the benefit of those who control them (Jiang, Lee and Yue, 2010). More tunneling indicates severe agency conflict and weaker governance. Thus, the variable tunneling is defined as followed a ratio of other receivables to total assets. An alternative measurement of tunneling is the connected transaction of listed firms with the largest shareholders 1, shown in Table 2. Given the impact of corporate governance on cash holdings and debt maturity, this study includes governance measures following Chen et al. (2012). The variable Largest is measured as a percentage of stock shares held by the largest shareholders to total outstanding shares of listed firms. As well, a series of control variables are considered for the next section s regression analyses. The control variables include the operational, investment, and financing determinants of cash holdings and/or debt maturity drawn from existing literature (Chen et al., 2014; Chen et al., 2012; Duchin et al., 2010; Bates, 1 The data of connected transactions of listed firms with the largest shareholders are retrieved from the CSCMAR database.

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 221 Kahle, and Stulz, 2009; Mikkelson and Partch, 2003; Opler et al., 1999). Specifically, Ld is the ratio of long-term debt to total assets. Ocash is the ratio of operational cash flow to total assets. Roa is earning before interest and tax on total assets. Growth is the growth rate of capital expenditures year by year. Div is measured as dividend payments. SEO is a dummy variable equal to 1, if the sampling companies have experienced a seasoned equity offering in year t. Size is a natural logarithm of the book value of total assets of listed companies. Age represents the listed age for the sampled firms. Other variables are defined in Table 2. Table 2 Variable Definition Variable Cash Sd Ld Tunneling Tunnel_A Roa Ocash Growth Capex Turn Net_cap P/B Age Div SEO Industry Size Definition Cash is the amount of cash and cash equivalents scaled by total assets net of cash and cash equivalents. Short-term debt is measured as the ratio of short term debt to the net value of total asset from short term debt. Ld is the ratio of long-term debt to total assets. Tunneling is calculated as the ratio of other receivables scaled by total assets to estimate tunneling activities of the largest shareholders of listed companies. An alternative measurement of tunneling is the connected transactions of listed firms with the largest shareholders. Roa is earnings before interest and tax on total assets. Ocash is the ratio of operational cash flow to total assets. Growth is a firm s sales growth. Capex is the ratio of capital expenditures to total assets. Turn is the turnover rate of accounts receivable. Net_cap is the ratio of net working capital to total assets. P/B represents market to book value. Age represents number of years since the firm s IPO. Div is measured as dividend payments. SEO is a dummy variable equal to 1 if the sampling companies have experienced seasoned equity offering in year t. These are a series of dummy variables to retrieve the first digit of industries coding with four digits coding in the CSCMAR dataset. Size is a natural logarithm of the book value of total assets of listed companies.

222 Lu Dai, Qingbin Meng, Maozhu Sun 4.3 Descriptive Statistics Given the variable definition in Table 2, we tabulate summary statistics for the most important variables. Panel A of Table 3 documents the summary statistics for the ratios of cash holdings, ratio of short term debt, leverage, Roa, Growth, and Percentage of shareholding of largest shareholders for treated samples. As is shown, the mean (median) ratio of cash holdings of treated samples is 31% (26%), while the mean (median) ratio of short-term debt of treated samples is 35% (31%). The descriptive statistics of the shareholdings structure reflect that the mean (median) percentage of shareholdings of the largest shareholders is 37.42% (35.44%). So it is proper to focus on the agency problems of the largest shareholders for exploring this puzzle. In terms of the time trends of the puzzle, Panel B of Table 3 shows that 77% of the treated samples have persisted with the puzzle for more than two years. This implies that the puzzle is characterized by the persistent financial policy of listed corporations rather than a transient event. Table 3 Financial and Ownership Characteristics of Treated Samples on Firm-Year Basis ab Variable Mean Standard Error Maximum Minimum Median Panel A Summary statistics of treated samples The ratio of cash holdings 0.31 0.14 1.09 0.20 0.26 The ratio of short-term debt 0.35 0.15 1.03 0.20 0.31 Leverage 0.59 0.21 1.49 0.07 0.58 Roa 0.05 0.05 0.48 0.22 0.05 Growth 0.20 0.45 2.21 0.43 0.06 Percentage of shareholdings of largest shareholders 37.42% 15.59% 79.77% 9.80% 35.44% Panel B Distribution of treated samples by years of the persistence of the puzzle Years of the persistence of the puzzle 2 years 3 years 4 years 5 years More than 5 years The proportion 23% 26% 21% 20% 10% Note. a all the variables are defined in Table 2. b We winsorize the variables using the 1% and 99% quintiles to eliminate outliers. We further identify the sources of cash holdings of the treated samples from 1998 to 2013. Panel A of Table 4 shows that the mean (median) ratio of cash holdings change to total assets is near zero. The mean (median) ratio of borrowing cash inflow to total assets is 36% (31%), slightly higher than the mean

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 223 (median) ratio of cash holdings shown in Panel A of Table 3. Furthermore, Panel B and Panel C of Table 4 respectively run non-parameter tests for the difference in short term debt financing in the first half of the year minus one in the second half of the year. As is presented, the mean (median) of this difference is significantly negative for treated samples. The results imply that the short term debt financing of treated samples in the second half of the year is significantly Table 4 Sources of Cash Inflow of Treated Samples on Firm Half-Year Basis Panel A Summary statistics of cash flow sources Variable Mean Standard Maximum Minimum Median Errors The ratio of cash holdings 0.00 0.11 1.25 0.51 0.01 change to total assets b The ratio of borrowing cash inflow to total assets b 0.36 0.28 2.37 0.00 0.31 Panel B Summary statistics of short term debt financing of treated samples Variable Mean Ttest Median Test Cash inflow of debt financing in the first half-year Cash inflow of debt financing in the second half-year Difference of cash inflow of debt financing in the first half-year minus that in the second half-year (Probability) 0.23 47.85 (<0.0001) 0.30 11.19 (<0.0001) 0.07 2.51 (0.0122) Probability 0.20 < 0.0001 0.22 < 0.0001 0.02 < 0.0001 Panel C Summary statistics of short term debt financing of matched samples Variable Mean T test (Probability) Median Test Probability Cash inflow of debt financing 0.21 23.38 0.1999 < 0.0001 in the first half-year Cash inflow of debt financing 0.19 19.21 0.1701 < 0.0001 in the second half-year Difference of cash inflow of debt financing in the first half-year minus refinancing in the second half-year 0.017 1.43 0.0282 0.015 Panel D Summary statistics of difference for treated samples minus one for matched samples The difference for treated samples minus one for matched samples 0.34 1.06 (0.29) 0.03 0.0515 Note. a The ratio of cash holdings and ratio of short-term debt are defined in Table 1. b There are 968 firm-half year observations from 1999 to 2013.

224 Lu Dai, Qingbin Meng, Maozhu Sun higher than in the first half of the year. In contrast, the mean of this difference is insignificant for matched samples, while the median of this difference is significantly positive for matched samples. As shown in Panel D of Table 4, the median differences for treated samples in Panel B are significantly higher than the median differences for matched samples in Panel C. Combining the non-parameter test results of both treated and matched samples, it is unusual that the financing amount of short term debt of the treated samples is higher than one of matched samples in the second half year. 5 Test and Analyses 5.1 Regression Results We construct and estimate a structural equation to test hypotheses 1, 3 and 4 below to explore the key factors of this puzzle based on the consideration of agency conflicts. The method of structural equations could address the endogenous effect existing between cash holdings and short-term debt to some extent. Furthermore, to control for the self-selection effect, our samples include the matched samples selected by the propensity score matching (PSM) as mentioned above. To differentiate treated samples from the matched samples, we design a dummy variable Match which equals 1 for the observations of treated samples and zero for the observations of matched samples. All the samples add up to 1,568 observations shown in Column 2 and Column 3 of Table 5. In equation 1 (for short Eq. (1)), the dependent variable, the ratio of cash holdings (Cash it ) is calculated as a percentage of firm i s reserve of cash and cash equivalents to the firm s noncash part of total assets in year t. For independent variables in both equations, we choose the variables to measure the factors proposed in hypotheses 1, 3, 4 and add the control variables found in the extant literature to control the normal needs of operational and investment activities for cash holdings, including the ratio of net working capital to total assets (Net_cap), turnover rate of accounts receivable (Turn), the ratio of long term debt to total assets (Ld), capital expenditure (Capex), size, listed age (Age), the shareholding percentage of largest shareholders (Largest) as well as the seasoned equity offering (SEO). All the control variables are drawn from recent studies (Chen et al., 2014; Chen et al., 2012; Mikkelson and Partch, 2003) and defined in Table 2. In equation 2 (for short Eq. (2)), dependent variable (Sd it ) is measured by the

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 225 ratio of short-term debt to the net value, that is, subtracting short-term debt from total assets of firm i in year t. To test our hypotheses, we include the variables that capture agency problems of the largest shareholders and other factors in this equation. Also, we add the control variables used in the extant literature to control the operational and investment impacts on debt maturity, including the ratio of long term debt to total asset (Ld), the shareholding percentage of largest shareholders (Largest), size, listed age (Age), as well as market to book value of net asset (P/B). All these control variables are drawn from recent studies (Chen et al., 2014; Chen et al., 2012; Mikkelson and Partch, 2003) and defined in Table 2. To control for the industry effect in both Eq. (1) and Eq. (2), we design the set of dummy variables to retrieve the first digit of industries coding with four digits in the CSMAR dataset. The series of dummy variables (Ind i ) are added into the Eq. (1) and Eq. (2). Cash it = α 0 + α 1 Sd it + α 2 Tunneling it + α 3 Match + α 4 Match*Sd it + α 5 Net_cap it + α 6 Ocash it + α 7 Turn it + α 8 Ld it + α 9 Capex it + α 10 Largest + α 11 Size it +α 12 Age + α 13 SEO + α 14 Div it + α i Ind i. (1) Sd it = β 0 + β 1 Cash it 1 + β 2 Tunneling it + β 3 Match + β 4 Match*Cash it 1 + β 5 Match*unneling it +β 6 Ocash it + β 7 P/B it + β 8 Ld it +β 9 Largest +β 10 Size it + β 11 Age + β i Ind i. (2) Based on firm-year data of treated and matched samples, this paper estimates the structural equations (Eq. (1) and Eq. (2)) by adopting Two Stage Least Squares Method (2SLS). The regression results are tabulated in columns 2 and 3 of Table 5. In Eq. (1), the coefficient for the interaction item Match*Sd it is 0.734 (t-statistic = 5.13, significant at better than 1% level). In contrast, the coefficient of Sd it (short term debt) is 0.134 (no significant). Both the results indicate that the financing of short-term debt is significantly and positively correlated with cash reserves in the same period only for all the listed firms in the sample related with the puzzle. The statistical tests of Eq. (2) display that the interaction item Match * Cash it 1 is significant and positive (the coefficient for Match * Cash it 1 is 0.191, t-statistic=3.83, significant at better than 1% level). Thus, the result is consistent with the third hypothesis that the cash reserves at the year-end significantly contribute to short-term debt financing for the following year. The estimating result of Eq. (2) reveals that the coefficient of the interaction

226 Lu Dai, Qingbin Meng, Maozhu Sun item Match * Tunneling, is significant and positive (t-statistic=3.25, significant under 1% level). The result indicates that tunneling activities of the largest shareholders are correlated with short term debt financing for the treated samples. However, the independent variable tunneling is not significantly related with the dependent variable (short term debt). Thus, the estimating result is consistent with our fourth hypothesis that the largest shareholders of listed corporations related with the puzzle take advantage of short term debt financing for their private interest. Although it is not the main focus of our paper, Table 5 also provides information about the operational and investment determinants of cash holdings and short term debt in Chinese-listed companies. These determinants are largely consistent with those identified by Chen et al. (2012) and Chen et al. (2014) Specifically, the estimating result in Eq. (2) demonstrates that operational cash flow (Ocash it ) is significantly and negatively correlated with the ratio of short-term debt. Namely, these companies utilize short-term debt to increase cash holdings in order to offset operations deficits. The variable that reflects corporate governance, the shareholding percentage of largest shareholders, is significantly and positively correlated with short term debt. The measurement of listed age is added into both Eq. (1) and Eq. (2) which represents financial constraints confronted by listed corporations (Chen, et al., 2012). The coefficient estimates of listed age ( Age ) in columns 2 and 3 of Table 5 are all significantly positive. So the estimating result of this variable is consistent with precautionary motives of cash holdings as tighter financial constraints prompt greater cash reserves of listed corporations. Furthermore, the control variables in Eq. (1), capital expenditure (Capex) is negatively significant with the dependent variable cash holdings (t-statistic = 2.1, significant under the 5% level). As well, the seasoned equity offering is positively significant with the dependent variable cash holdings (t-statistic = 6.4, significant under the 1% level). The evidence mentioned above is confirmed by the robust test shown in column 4 and 5 of Table 5. We expand the scope of treated samples by reducing the sampling criteria that both the ratio of cash holdings and ratio of short-term debt are above 15% for at least two consecutive years. The regression results in column 4 and 5 of Table 5 manifest that all the key variables, the indicator of the largest shareholders tunneling, short term debt, cash reserves in t 1 period, and part of the control variables are significant at least at the 10% level. Thus, the

Why Greater Cash Holdings and Short-Term Debt Simultaneously Persist? 227 results of key factors involved with hypotheses 1, 3 and 4 based on treated and matched samples are to some extent unconditional on the sampling criteria. In summary, the estimating results of primary variables involved with our hypotheses 1, 3, and 4 in both Eq. (1) and Eq. (2) are robust after controlling for the conventional motives of cash holdings. Table 5 2SLS Estimates of Eq. (1) and (2) abcd Main Test Variable & Specification Cash it Eq. (1) Intercept (α 0 / β 0 ) 0.200 (0.83) Sd it 0.134 ( 0.72) Sd it Eq. (2) 0.262 ( 1.48) Cash it 1 0.124 *** Tunneling it 0.012 ( 0.03) Match 0.132 * (1.89) Match *Sd it 0.734 *** (5.13) ( 4.96) 0.0129 (0.04) 0.186 *** (7.78) Match * Cash it-1 0.191 *** (3.83) Match*Tunneling it 0.525 *** (3.25) Net_cap it 0.177 ** (2.6) Ocash it 0.115 (0.95) 0.233 *** ( 3.07) P/B it 0.0027 (1.21) Turn it 0.031 (1.6) Ld it 0.349 ** (2.52) Capex it 0.124 ** ( 2.1) Largest 0.004 ** (2.08) 0.312 ( 1.29) 0.006 ** (2.34) Cash it Eq. (1) 0.322 (1.02) 0.37 1.09) 0.03 ( 1.02) 0.245 ** (1.96) 0.923 *** (3.09) 0.099 (1.56) 0.26 (1.44) 0.082 (0.73) 1.27 *** (3.01) 0.831 *** ( 4.23) 0.002 (1.03) Robust Test Sd it Eq. (2) 1.369 * (1.92) 0.473 ** ( 2.02) 0.001 (0.33) 0.346 ** (2.13) 0.943** (2.41) 0.416** (1.99) 0.178 *** ( 4.17) 0.017 ( 1.31) 0.08 ( 0.97) 0.001* (1.71) (To be continued)