Asset Tangibility and Cash Holdings

Transcription

1 Asset Tangibility and Cash Holdings Current Version: January 15, 2015 Abstract The paper underscores the role of financial development in shaping corporate financing policy through a collateral channel, providing cross-country evidence that asset tangibility significantly affects corporate cash holdings negatively. This paper shows that firms in countries with higher levels of financial development exhibit less sensitivity of cash holdings to asset tangibility, which suggests that the collateral role of tangible assets is substituted by higher-standard financial institutions, proxied by creditor rights protection and information sharing among creditors. The results imply that the development of financial markets and institutions helps lower borrowing costs and mitigate corporate financial constraints and precautionary savings concerns, thereby promoting corporate investments and economic growth. This study confirms the view that risky firms tend to pledge more tangible collateral and the collateral spread declines with improvements in the quality of institutions. The study also shows that asset salability enhances the collateral role of tangible assets.

2 Asset Tangibility and Cash Holdings Abstract The paper underscores the role of financial development in shaping corporate financing policy through a collateral channel, providing cross-country evidence that asset tangibility significantly affects corporate cash holdings negatively. This paper shows that firms in countries with higher levels of financial development exhibit less sensitivity of cash holdings to asset tangibility, which suggests that the collateral role of tangible assets is substituted by higher-standard financial institutions, proxied by creditor rights protection and information sharing among creditors. The results imply that the development of financial markets and institutions helps lower borrowing costs and mitigate corporate financial constraints and precautionary savings concerns, thereby promoting corporate investments and economic growth. This study confirms the view that risky firms tend to pledge more tangible collateral and the collateral spread declines with improvements in the quality of institutions. The study also shows that asset salability enhances the collateral role of tangible assets. JEL Classifications Numbers: G32, G21, G33, O16 Keywords: Asset Tangibility; Cash Holdings; Financial Development; Asset Redeployability

3 1. Introduction Considerable attention has been paid to the record-high cash holdings of U.S. firms. For instance, Bates, Kahle, and Stulz (2009) document that the average cash-to-assets ratio of U.S. firms more than doubles from 10.5% in 1980 to 23.2% in The Wall Street Journal stated in June 2010, Nonfinancial companies had socked away $1.84 trillion in cash and other liquid assets as of the end of March, up 26% from a year earlier and the largest-ever increase in records going back to Cash made up about 7% of all company assets, the highest level since Prior work on determinants of cash holdings suggests that firms accrue cash for various reasons such as the transaction cost motive, the precautionary motive, the repatriation tax motive, and the managerial agency cost motive. 2 Bates, Kahle, and Stulz (2009) also find that the average cash ratio of high-tech firms is significantly greater than the average cash ratio of manufacturing firms. This evidence provides empirical support for the precautionary motives that drive firms, for example, in computers, electrical equipment, and pharmaceutical sectors, to sit on huge amounts of unspent corporate cash. One reason is that these firms may be concerned about potential difficulties in continuously funding their costly on-going R&D projects (Brown and Petersen, 2011) and they often face higher financing and refinancing costs due to the lack of sufficient tangible assets pledged as collateral for loans. While asset tangibility is a major factor in determining capital structure, 3 much less attention has been directed to understanding the collateral channel through which asset 1 Justin Lahart, U.S. Firms Build Up Record Cash Piles, The Wall Street Journal, June 10, Studies by Kim, Mauer, and Sherman (1998), and Opler, Pinkowitz, Stulz, and Williamson (1999) show the pecking order and trade-off models of benefits and costs of cash holdings, and report that firm characteristics such as firm size, growth opportunities, and volatility of future cash flows determine the optimal investment in liquidity in the presence of capital market frictions. Almeida, Campello, and Weisbach (2011) present a model of inter-temporal investment decisions with costs of external financing and show that firms hold more cash today if they anticipate tighter financing constraints in the future. Bates, Kahle, and Stulz (2009) conduct an excellent review of the literature on cash holdings. 3 See, e.g., Kiyotaki and Moore (1997), Campello and Giambona (2013), and Rampini and Viswanathan (2013) for the positive link between fixed assets and leverage, and Rajan and Zingales (1995) for some international evidence on the interplay between asset tangibility and capital structure. 1

4 tangibility affects corporate cash holding policy. 4 This paper attempts to fill this gap in the cash holding literature by exploring in detail the association between asset tangibility and cash holdings. The argument is that since tangible assets can be used as collateral to alleviate firms financial constraints by reducing borrowing costs, firms with low asset tangibility tend to hold more cash from a precautionary motive standpoint. To motivate the argument, I first present time-series evidence from U.S. data as a point of departure. The reason is that a large body of literature on cash holdings has been devoted to explaining the evolution of cash holdings for U.S. firms over the past three decades (see, e.g., Opler, Pinkowitz, Stulz, and Williamson, 1999; Bates, Kahle, and Stulz, 2009). Figure 1 depicts two important stylized facts about the evolution of annual mean cash-to-assets ratio and asset tangibility along with net tangibility and intangibility index over fiscal years from U.S. data. 5 [Figure 1 about here] First, Figure 1 reveals that not only cash holdings have not been increasing dramatically until late 1970s, but both asset tangibility and net tangibility take on an almost reverse trend against cash over the whole sample period, a pattern that has not been documented in the cash holdings literature. To put it into perspective, in contrast to cash holdings, asset tangibility has been plummeting since the early 1980s, representing merely 29.3% of the total assets in fiscal year 2011, a 38.2% drop from 47.3% in A plausible explanation for this trend is that the 4 John (1993) shows that the liquidity ratio, measured as the ratio of cash and marketable securities to total assets, is decreasing in the ratio of inventory plus gross plant and equipment to total assets, a proxy for the liquidity costs of asset restructuring (the collateral value of the assets) suggested by Titman and Wessels (1988). Klasa, Maxwell, and Ortiz-Molina (2009) also consider the tangibility of a firm s assets, measured as the ratio of net property, plant, and equipment to book assets, as a determinant of its cash holdings. 5 Throughout the paper, I measure the degree of asset tangibility by using the ratio of 0.715*Receivables plus 0.547*Inventories plus 0.535*Fixed Capital to Book Value of Total Assets, which is developed in Berger, Ofek, and Swary (1996). Net tangibility is calculated as 0.715*Receivables plus 0.547*Inventories plus 0.535*Fixed Capital minus total current liabilities (LCT) and plus total debt in current liabilities (DLC), deflated by book assets. Intangibility index is the ratio of research and development (R&D) to capital spending. 2

5 reduction in asset tangibility lowers the overall collateralizable value of firms assets, and therefore reduces the availability of external debt finance. Consequently, firms, especially those that are prone to being financially constrained, tend to stockpile large amounts of cash to reduce potential borrowing costs, consistent with the precautionary motive for cash holdings. Second, the figure also illustrates another important fact. Specifically, the growth rate of asset intangibility index (a flow measure) starts to accelerate around early 1980s, about the same period when cash holdings start to increase. The mean capital input ratio between intangible assets and tangible assets has been sharply rising to 3.72 in fiscal year 2011, more than 10 times from 0.33 in As firms pour more funds in intangibles, the increasingly intensified precautionary demand for cash plays a more critical role than before. This stylized fact resonates with the rapid development and innovation in technology across the entire spectrum of firms over the past three decades. Undertaking a standard regression approach similar to the Tables III and V of Bates, Kahle, and Stulz (2009), I show in unreported tables that 1) asset tangibility affects corporate cash holdings negatively, which is robust to OLS regressions using variables in levels and changes, Fama-MacBeth regressions, and specifications with firm fixed effects; and 2) asset tangibility is the most important determinant of cash holdings in explaining the recent dramatic increase in cash holdings in the U.S over the 2000s among a host of well-known determinants such as industry sigma and cash flow documented in the literature. After showing a negative link between cash and asset tangibility using U.S. data, I further identify the collateral channel in a cross-country setting. The identification strategy is motivated by a recent paper by Liberti and Mian (2010) who explore how the level of financial development in a country affects the collateral cost of capital. Specifically, they show that institutions such as creditor rights and information sharing that reflect/promote 3

6 financial development alleviate borrowing constraints by lowering the difference in collateralization rates between high- and low-risk borrowers. They also document that firms in better financially developed countries pledge a less amount and a wider range of assets including firm-specific assets as collateral. Therefore, these findings imply that the collateralization rates vary inversely with the quality of institutions. It further suggests that the sensitivity of cash to asset tangibility should be smaller in countries with better institutions. Specifically, I exploit the cross-country variation in a country s characteristics such as financial development to identify the collateral channel through which asset tangibility affects cash holdings around the world. To this end, I proceed with a cross-country analysis by collecting data on a commonly-used proxy for financial development (the value of credits by financial intermediaries to the private sector, divided by GDP) from Beck and Demirgüç- Kunt (2009), and data on cash holdings and asset tangibility from Compustat for fiscal years 1993, 1998, 2003, and 2008 across 39 countries. Panel A of Figure 2 plots the annual average cash-to-assets ratio against private credit to GDP for fiscal years 1993, 1998, 2003 and Panel B plots the average asset tangibility on the horizontal axis. Two facts stand out from Figure 2: (Fact 1) there exist significant cross-national variations in cash holdings and financial development, as well as a slight positive association between the two over time in Panel A. 6 (Fact 2) the cross-sectional average of cash holdings are decreasing in asset tangibility across countries and the relationship has persisted throughout the sample period, as demonstrated in Panel B. [Figure 2 about here] 6 Studies that positively link financial development (measured by private credit to GDP) and cash holdings in a cross-country setting include Dittmar, Mahrt-Smith, and Servaes (2003) and Kalcheva and Lins (2007). Khurana, Martin, and Pereira (2006) document that the sensitivity of cash holdings to cash flows decreases with financial development. 4

7 The work of Liberti and Mian (2010) is related to a large body of literature in law and finance that has documented an important relationship between a country s legal system and the development of its financial markets. Prior work in this strand of literature by La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998; henceforth LLSV) shed light on the critical role of financial development as well as legal and institutional environment in reducing the costs of collateral evaluation, processing, and liquidation. 7 Importantly, LLSV (1998) explore the role of two financial market mechanisms in economic development. They conclude that both creditor rights and information sharing help promote capital market development, which in turn contributes to economic growth. The first mechanism is creditor s rights, which refers to the contracting environment between borrowers and lenders and the protection of creditors ability to collect the money from borrowers defaulting on a debt obligation. A strand of literature on creditor rights documents that strong creditor protection promotes credit market development (e.g., Djankov, McLiesh, and Shleifer, 2007; Haselmann, Pistor, and Vig, 2010). In line with this view, several studies also show that loans under strong creditor protection have lower interest rates, lower contracting costs of financing, and favorable terms (Qian and Strahan, 2007). Stronger creditor rights also reduce interest rate spread on loans to borrowers (Bae and Goyal, 2009). Therefore, lenders (creditors) are willing to extend credit and take risk when they are less exposed to borrower (debtors) expropriation, resulting in borrowing firms having more external finance. The collateralization of loans is also lower in financially developed markets. Moreover, LLSV (1998) construct an index aggregating the rights of secured lenders based on restriction on reorganization, no automatic stay, no management stay, and secured debt paid first. They show that countries with stronger legal protection of creditors have 7 In a similar spirit, Rajan and Zingales (1998) find that financial-sector development proxied by private credit to GDP, stock market capitalization, and accounting standards reduces the costs of external finance to firms. 5

8 deeper credit markets and that creditor rights act as a substitute for collateral in mature markets. 8 For example, restrictions on reorganization such as creditors consent or minimum dividend are imposed when a debtor decides to file for reorganization. This restriction decreases the likelihood that debtors use bankruptcy as a strategic way of avoiding debt. Similarly, the no automatic stay or asset freeze imposed by the court protects creditor s ability to seize collateral after the petition for reorganization is approved. Finally, the no management stay provides powers to an administrator rather than the incumbent management that is in control of property pending and responsible for running the business during the reorganization. Therefore, creditor rights weaken the role of tangible assets as collateral for secured credit in restricting debtors from risk-taking and risk-shifting. I expect that creditor rights protection should attenuate the cash-tangibility sensitivity. The second mechanism that LLSV (1998) discuss is public and private information sharing among lenders about the creditworthiness of loan applicants. Information sharing attenuates adverse selection and moral hazard and therefore facilitates lending. A stream of studies on creditor information sharing further shows that information exchange improves credit availability (Pagano and Jappelli, 1993; Padilla and Pagano, 1997, 2000), lowers the cost of credit to firms (Brown, Jappelli, and Pagano, 2009), motivates loan repayments (Brown and Zehnder, 2007), and reduces default rates (Jappelli and Pagano, 2002). In addition, bank lending literature suggests that collateral may be required simply to reduce information asymmetry because lenders can obtain additional information about the borrower by evaluating the quality and nature of the collateral (e.g., Picker, 1992) and assess the borrower s repayment prospects. Therefore, collateral-based lending is typically used to provide credit availability in the opaque information environment of an emerging market where information sharing and financial transparency are not available or extremely limited. 8 This is consistent with power theories of credit, based on the transfer of control rights upon default (Aghion and Bolton, 1992; Hart and Moore, 1998). 6

9 However, the improvement in the quality of creditor information through informational infrastructure such as information sharing eliminates asymmetric information between lenders and borrowers. Consequently, it directly decreases creditors effort and costs of screening firms and encourages lenders to make use of substitutes for physical collateral to provide credits. For example, the alternative types of loans may include unsecured loans or loans secured by reputation collateral or restrictive covenants which limit borrowing firms actions prior to default and therefore provide ex ante protection for creditors, as argued by Miller and Reisel (2012). Therefore, the role of collateral stipulated in the debt contract becomes less significant in countries with credit information sharing systems through public credit registries and private credit bureaus. I expect that information sharing should also attenuate the cash-tangibility sensitivity. The main results confirm the expectation that both creditor rights and information sharing independently weaken the impact of asset tangibility on cash holdings after controlling for the effects of economic development. Specifically, I estimate regressions that include the interaction effects of asset tangibility with proxies for financial development while also controlling for the interaction of asset tangibility with the log of gross domestic product (GDP) per capita, along with country, industry, and time fixed effects. The main results are robust to alternative measures of financial development, including weighted least squares regressions where each country receives equal weight in the estimation, and subsamples excluding the U.S. and Japan. Having shown that creditor right protection substitutes for collateral, lowers the importance of tangible assets in reducing a firm s financial constraint, and hence reduces the cash-tangibility sensitivity, I turn to further investigate whether the differences in laws and enforceability of contracts also matter for the effect of creditor right protection on the cashtangibility sensitivity. This approach is motivated by Bae and Goyal (2009). I show that both 7

10 the existence of creditor rights per se and the quality of their legal enforcement are important to the contracting and bank-lending process. I use four proxies including contract enforcement speed and cost, legal formalism, and judicial efficiency for law enforcement of the creditors rights to further identify the channel through which creditor rights substitute for tangible assets as collateral. The economic intuition is that strong legal protection that better ensures creditors to repossess collateral impose a credible threat and greater costs upon a borrower in the case of default. As a result, borrowers would be less willing to take on extra risk. 9 Therefore I anticipate that the expected realized value of collateral would increase with creditor protection. It then follows that stronger (weaker) legal enforcement of creditors rights makes creditor rights more (less) effective, leading to a more (less) pronounced attenuating effect of creditor rights on the cash-tangibility sensitivity. Having shown the attenuating impact of information sharing on the cash-tangibility sensitivity, I next turn to further explore how the effect of information sharing on cashtangibility sensitivity varies with the opacity of a company. To this aim, I split the sample according to firm characteristics age, size, and growth opportunities. The estimates suggest that only young, small, and high-growth firms, which arguably suffer most from financial market imperfections such as asymmetric information and hence financial constraints, benefit most from the establishment of information sharing. This result implies that information sharing reduces the role of tangible assets as collateral and loosens the borrowing constraints of informationally opaque firms. Hence, it provides additional support for the argument that creditor information sharing substitutes for collateral in countries with high transparency of credit markets. 9 Supporting to the view that borrowers are less willing to take risks when creditors are better protected, Acharya, Amihud, and Litov (2011) find international evidence that stronger creditor rights tend to reduce corporate risk taking. The right to repossess collateral gives lenders an essential threat to ensure that borrowers will not use the money borrowed unproductively. 8

11 The results also complement the findings of Liberti and Mian (2010). Specifically, I relate the average collateral spread, proxied by the differential cash-tangibility sensitivity between high- and low-risk borrowers, to improvements in the quality of institutions. I first show that firms with higher ex ante credit risk or default probability tend to have lower cashtangibility sensitivity. This implies that in order to obtain the same dollar worth of secured loans, high default risk firms have to post more tangible assets as non-cash collateral. These results are in line with the sorting-by-observed-risk paradigm which claims that observably risky low quality borrowers are required to pledge collateral while observably safe borrowers are not required or pledge less (e.g. Berger and Udell, 1990). More importantly, I demonstrate that the differential cash-tangibility sensitivity between high- and low-risk borrowers declines with improvements in the quality of institutions. This finding suggests that financial development closes the wedge in collateralization rates between high- and low-risk borrowers. Finally, this paper is also closely related to the asset salability literature (Shleifer and Vishny, 1992; Berger, Ofek, and Swary, 1996; Stromberg, 2001; Acharya, Bharath, and Srinivasan, 2007; Benmelech, 2009; Campello and Giambona, 2013). Using three industrylevel measures of asset salability (industry competitiveness, industry asset non-specificity, and industry liquidity), I first identify the cash-tangibility link through the effect of changes in the liquidation values of redeployable tangible assets on liquidity management after controlling for both institutional and economic impact. I show that higher industry-level salability enlarges the negative impact of asset tangibility on cash holdings. The result is in accordance with the argument in asset salability literature that the combination of the effects of physical attributes of an asset and the sheer number and financial strength of its potential buyers in the secondary market determines liquidation values of assets. I further show that country-level salability, proxied by log of GDP per Capita, strengthens the industry-level 9

12 salability effect on the cash-tangibility sensitivity. The argument is that tangible assets are more salable in countries with more economic activities (i.e., higher GDP per capita). 10 Intuitively, an economic boom spurs firms investments and leads to high demand for collateralizable tangible assets in the secondary market. It becomes faster and easier for economic agents to trade financial instruments in asset markets. This then causes a surge in the salability of tangible assets. Consequently, the liquidation value of the collateralizable tangible assets rises. Overall, the results support the findings of Benmelech (2009) and Campello and Giambona (2013) and complement their work by providing some new crosscountry evidence of the effects of asset tangibility on collateral value for lenders. This paper contributes to the cash holding literature. I show that asset tangibility explains the evolution of firms cash holdings across countries as a major neglected determinant of cash holdings. In addition, to the best of my knowledge, this paper is the first to explore how financial development affects a firm s cash-tangibility sensitivity. This paper contributes to the growing literature on the role of institutions in corporate finance by analyzing their importance in improving firms access to external financing and shaping corporate cash holdings across a large number of countries. Specifically, I argue and provide evidence that both creditor rights and creditor information sharing substitute the collateral role of tangible assets and exert attenuating effects on the negative cash-tangibility sensitivity, but for different reasons. Creditor rights are an effective means for imposing disciplinary restrictions on a borrowing firm s action, especially when creditor rights are well enforced by laws. In contrast, information sharing helps lenders easily gather accurate and timely information about borrowers while appraising the loans. It assists lenders to provide credit to financially constrained firms such as small technology firms that previously suffered from 10 Campello and Giambona (2013) show that the relationship between leverage and tangible assets is stronger when federal funds rate is higher, suggesting that redeployability is more important during credit contractions. Because higher FED funds rate is typically associated with economic booms, my arguments are consistent with theirs. 10

13 information asymmetry. Overall, with better institutions, firms have easy and multiple access to extensions of credit from the financial system. Therefore, both creditor rights and information sharing help promote capital market development and effectively allocate economic resources, which in turn contributes to economic growth. These findings suggest that the development of financial markets and institutions is a critical part of the growth process. It is not simply an inconsequential side show that the financial system responds automatically to demands for financial arrangements created by economic development (Levine, 1997). The paper proceeds as follows. Section 2 discusses the related literature and develops hypotheses. Section 3 briefly describes the cross-country data and econometric framework. Section 4 reports empirical evidence on the negative link between cash holdings and asset tangibility and how institutional variables as well as industry- and country-level factors that determine asset salability identify the link. Section 5 offers conclusions. Appendix contains detailed definitions and sources of variables used in the study. 2. Hypothesis Development 2.1 Asset Tangibility and Cash Holdings Economic theory and empirics suggest that collateral is commonly used in loan contracts to reduce credit risks through decreasing expected default rates and increasing expected recovery rates. There are two main functions of collateral. The first is the disciplinary role of collateral. Collateral requirements provide secured lenders the right to repossess collateral conditional on default. Therefore, collateral can be used by lenders to restrict borrowers from asset substitution (Jensen and Meckling, 1976). The second is the informational role of collateral. Lenders can obtain additional information about the borrower by evaluating the quality and nature of the collateral (Picker, 11

14 1992) and assess the borrower s repayment prospects. Moreover, collateral allows lenders to sort observationally equivalent loan applicants through signaling and help attenuate the problems of adverse selection and credit rationing (Besanko and Thakor, 1987a, 1987b; Stiglitz and Weiss, 1981). It also reduces moral hazard by aligning the interests of both lenders and borrowers (Boot, Thakor, and Udell, 1991; Holmstrom and Tirole, 1997). Tangible assets can serve as collateral in secured lending. Therefore, asset tangibility increases the recovery value for lenders in default states and is positively linked to the ease with which borrowers can obtain external financing. For example, the recent work of Campello and Giambona (2013) shows that redeployability of tangible assets is a main determinant of corporate leverage. Therefore, I hypothesize that when a firm's borrowing capacity from its existing asset base increases with asset tangibility, the firm tends to have lower precautionary demand for holding cash. I propose and empirically test the following hypothesis in alternative form: HYPOTHESIS 1: Firms with higher asset tangibility face lower borrowing costs, and therefore should hold lower levels of cash holdings from a precautionary standpoint, ceteris paribus. 2.2 The Effects of Institutional Variables on Cash-Tangibility Sensitivity Financial development reduces firms reliance on tangible assets as collateral in corporate borrowing and therefore decreases the precautionary motive of cash savings through the collateral channel. There has been a rich literature focusing on the determinants of financial development and the potential role played by both financial intermediaries and markets in the process of economic growth and development (see Levine (2005) for a useful survey). Arguing for the benefits of financial development, financial sectors improve efficiency 12

15 in financial resource allocation in the economy (Boyd and Prescott, 1986), facilitate trading and hedging in stock markets (Holmstrom and Tirole, 1993), lower liquidity risk (Diamond and Dybvig, 1983), ameliorate information asymmetries, and hence reduce contract enforcement costs, transactions frictions, and costs of external finance (Rajan and Zingales, 1998). 11 Therefore, I anticipate that with deep capital markets, firms can take advantage of easy and costless access to extensions of credit from the financial system. The collateral role of tangible assets becomes less significant. The law and finance literature also suggests that better legal systems foster the protection of creditors and shareholders legal rights, reduce contracting costs, and improve information infrastructure, such as public registries or private credit bureaus for sharing credit information across financial institutions. Better institutions therefore reinforce the liquidity provision by financial intermediaries. The strength of creditor rights affects the contracting environment constituting an essential ingredient of financial development. For example, LLSV (1997) show that countries with stronger legal protection of creditors have deeper credit markets and that creditor protection acts as a substitute for collateral in mature markets. Creditors also provide finance to firms largely because the law protects their rights (LLSV, 1998). When creditors have more bargaining power (such as being able to take control of a firm in bankruptcy), they are more willing to grant credit on favorable terms (such as longer maturities and lower interest rates). Thus, stronger creditor rights should result in an environment in which firms are relatively less financially constrained because there would be less credit rationing and lower costs of external finance. Creditor rights protection also has a number of important influences on lender s and 11 Financial development, however, does come with costs. Excessive financial liberalization may result in an unduly large expansion of credit to risky firms with unviable projects and limited monitoring of regulatory agencies (Aghion, Bacchetta, and Banerjee (2004)). 13

16 borrower s risk incentives. Better creditors rights over the collateral stipulated in the debt contract will make it easier for secured creditors to seize and liquidate assets in the event of bankruptcy. Therefore the power of secured lenders in bankruptcy is likely to have effects on over-borrowing, effective use of borrowed funds, the likelihood of bankruptcy including strategic bankruptcy, and the recovery rates in bankruptcy. Along these lines, Liberti and Mian (2010) argue that because both the expected realized value of collateral and the expected costs on a borrower for default or any deviations from the agreed upon contract increase with creditor protection, lenders can afford to reduce collateral spread in stronger legal regimes. Moreover, creditor rights protection reduces borrowers credit risks and restricts borrowers from strategic defaults, risk-taking (Acharya, Amihud, and Litov, 2011) and risk-shifting. These argument and findings are consistent with the view that creditor rights serve as a substitute for asset tangibility. In other words, creditors in poor creditor protection countries require more collateral to ensure smaller potential losses when they make loans. In financially developed countries where information sharing among creditors is available, lenders have a good knowledge of borrowers characteristics, past behavior, current debt exposure, and possible subsequent indebtedness. Therefore, credit information sharing is expected to reduce moral hazard and adverse selection in credit markets which can lead to credit rationing and underinvestment (Pagano and Jappelli, 1993). Credit information sharing also increases borrowers incentives to repay their debts as a strong disciplining device because information about defaults becomes available to all lenders (Padilla and Pagano, 2000; Brown and Zehnder, 2007; Hertzberg, Liberti, and Paravisini, 2010), and reduces overborrowing and default rates (Jappelli and Pagano, 2002; Brown, Jappelli, and Pagano, 2009; Bennardo, Pagano, and Piccolo, 2014). Arguably, creditors in countries with greater information sharing are more likely to 14

17 exploit soft information obtained from other creditors so as to accept alternative intangible collateral such as patents, trademarks, and borrower s reputation and credit scores. Creditor information sharing enables lenders to switch loan contracts from using collateral based lending technologies to using other softer lending technologies such as restrictive financial covenants which specify operating performance, balance sheet ratios, and limits to which the borrower must adhere and hence provide ex ante protection on creditors prior to default. Lenders may also underwrite unsecured loans using soft information on risk of borrower default, credit worthiness, current and future earnings prospects, etc., to opaque technology companies. Therefore, information sharing also substitutes the role of tangible assets as collateral in providing useful hard information about borrowers. Hence it reduces moral hazard and adverse selection as well as raising the discipline on borrowers. The second hypothesis is: HYPOTHESIS 2: The negative effect of asset tangibility on cash holdings is attenuated for firms operating in countries with higher-standard financial institutions, ceteris paribus; the disciplinary role of tangible assets in loan contracts is substituted by creditor right protection and the informational role of tangible assets is replaced by information sharing among creditors through alternative softer lending instruments. 2.3 The Effects of Changes in Asset Salability on Cash-Tangibility Sensitivity With liquid capital markets, firms have easy access to extensions of credit from the financial system and can easily redeploy their tangible assets at low costs. Benmelech (2009) uses the term salability to describe how the combination of the effects of physical attributes of an asset and the sheer number and financial strength of its potential buyers in the secondary 15

18 market determine liquidation values. Therefore, tangible assets are more salable in industries with more financially strong buyers and transferable tangible assets. In addition, firms tend to supply more tangible assets that are desirable to creditors in order to get favorable terms (e.g. low interest rates and long maturities on loans) if their tangible assets are more redeployable. Since asset salability is determined jointly by the redeployability of tangible assets and the liquidity of market for assets, firms can more easily redeploy their tangible assets at low costs in countries with more economic activity (i.e., higher GDP per capita). Intuitively, during economic booms, the profitability of capital and demand for funds is high and firms also tend to invest a lot. The rise of investment and demand for assets by firms is associated with high asset liquidity. Tangible assets are more salable. Therefore, it implies that economic development complements for the role of redeployable tangible collateral, thereby strengthening the cash-tangibility sensitivity. Economic agents are then less concerned about possible restrictions on future access to capital markets. Specifically, I investigate the second approach for identifying the cash-tangibility link through the effect of changes in the liquidation values of redeployable tangible assets on liquidity management by exploiting the heterogeneity in industry and economic characteristics across countries. The third hypothesis is: HYPOTHESIS 3: The negative effect of asset tangibility on cash holdings is larger for firms operating in markets with higher asset salability, ceteris paribus; the liquidation value of tangible assets is positively associated with asset salability. 16

19 3. Data and Empirical Methods This section presents the properties of the data and methods I use for empirically identifying and evaluating the proposed collateral channel through which asset tangibility affects corporate cash holdings Sample and Variable Construction I draw firm-level data for U.S. and non-u.s. firms from the Compustat North America and Compustat Global Fundamentals Annual database for the period These data include active and inactive firms that appear on Compustat at any time in the sample period. I remove the following sets of firms from the sample: 1) financial firms (SIC code ) and utility firms (SIC codes ); 2) firms missing the 48 Fama-French industry dummies constructed by using the firm's four-digit SIC industry code; 3) firms that cross-list in other markets of the world; 4) firms that do not prepare consolidated financial statements; 5) firms that have less than three years of available data over the study period; 6) firms for which cash and equivalents, asset tangibility, and/or total assets are missing; and 7) all firmyear observations with negative cash holdings, total assets and sales revenue, values for cash less than total assets, and values for the book value of total assets less than $5 million, inflation-adjusted in 2006 U.S. dollars. Finally, I further exclude countries with less than ten firms per fiscal year on average. The remaining sample consists of 29,130 unique firms representing 235,089 firm-year observations from 39 countries. In this study, the dependent variable is the cash and equivalents divided by the book value of total assets. The baseline proxy for firm-level asset tangibility is computed using the liquidation values of firm assets in discontinued operations and asset fire sales contained in Berger, Ofek, and Swary (1996). Asset tangibility is defined as 0.715*receivables plus 0.547*inventories plus 0.535*fixed capital, deflated by book value of total assets. The proxy measures the expected liquidation (resale) value of firms main categories of operating assets 17

20 such as fixed assets, accounts receivable, and inventories. Higher asset tangibility implies higher asset redeployability and liquidity, and hence higher recover/exit value for creditors. In addition, this measure allows us to examine how changes in fixed assets along with changes in account receivables (or net of payables) and inventories explain the evolution of cash holdings across the world Summary Statistics Table 1 presents country medians of some key variables employed in the analysis. In column 2, I observe that Japan has the second largest total firm-year observations and number of unique firms behind the U.S., while Columbia has the smallest. There is a wide variation in the cash ratios as displayed in column 5. The median firm in Israel and Hong Kong has a cash ratio of 20.4% and 16.0%, respectively, while the median firm in New Zealand, Chile, and Peru has a value of only 2.7%, 3.2%, and 3.2%, respectively. In contrast, as shown in column 6, the asset tangibility of the median firm in Israel and Hong Kong is merely 33.1% and 33.6%, respectively, whereas the value for the median firm in New Zealand, Chile, and Peru is 43.8%, 48.0% and 46.4%, respectively. I again observe a negative relation between cash holdings and asset tangibility in worldwide data. [Table 1 about here] I also gather aggregate country-specific data from 39 countries on private credit to GDP from Beck and Demirgüç-Kunt (2009) and on GDP per capita from the World Bank s World Development Indicators (WDI) database. The last two columns of Table 1 report the country medians for private credit creation and per capita GDP. In particular, the data show substantial variability in private credit to GDP and country wealth. Consistent with previous 18

21 literature, I use private credit to GDP, the total amount of credit by deposit money banks and other financial institutions to the private sector, divided by GDP, as the main measure of financial development or financial depth in the baseline regression model to identify the collateral channel, and I save other alternative proxies such as liquid liabilities per GDP and commercial-central bank for sensitivity tests. The median private credit measured over the period ranges from values of 159.6% in Switzerland, 153.9% in the United States, 143.6% in Hong Kong, and 142.0% in Netherlands, to values below 30% in Argentina, Turkey, Peru, and Mexico. Similarly, because the sample covers both developing and advanced countries, the median gross national income level per capita varies from well above $30,000 to as low as about $2,000 per annum, where GDP per capita is converted to 2005 international dollars using purchasing power parity (PPP) rates. This tremendous crosscountry variation in economic development helps us identify the channel through which asset tangibility affects cash holdings Empirical Strategy I look for international evidence to support the negative link between cash holdings and asset tangibility hypothesized in Section 2, and estimate a model of cash holdings by using pooled least squares regressions. The additional firm-level explanatory variables that I incorporate in the cross-country analysis are similar to those used by Dittmar, Mahrt-Smith, and Servaes (2003), and Kalcheva and Lins (2007). Specifically, the following regression model is use to test the first hypothesis:,,,,, 1 where i, c, j, and t denote firm, country, industry, and year, respectively; Cash is cash and 19

22 equivalents deflated by the book value of total assets; Asset Tangibility is defined as 0.715*receivables plus 0.547*inventories plus 0.535*fixed capital deflated by book value of total assets, following the metric introduced in Berger, Ofek, and Swary (1996). The estimated coefficient on asset tangibility delivers the prediction about the collateral channel the direct effect of asset tangibility on cash holdings through changes in a firm s collateral values of tangible assets. The higher the value of the tangible collateral, the less incentives there are for firms to hold cash. Therefore, I expect the marginal effect of asset tangibility on cash holdings to be negative and statistically significant ( 0). I also note that asset tangibility as a stock variable is a better proxy for debt capacity than capital expenditure which is considered as a very lumpy flow variable., is a set of firm-level covariates, which includes Market-to-Book, Firm Size which is measured by the natural logarithm of book value of total assets in millions of 2006 U.S. dollars, Cash Flow/Total Assets, Total Capital Expenditures/Total Assets, Total Book Leverage, R&D Expenses/Sales, and Dividend Dummy Variable., are distributed independently across firms with zero mean. The baseline regression also controls for unobservable time-invariant country level heterogeneity, industry-specific factors that capture systematic differences in liquidity management across industries, and year effects of common macroeconomic shocks that might affect firms cash decisions. Standard errors are clustered at both the firm and year levels to obtain standard-error estimates that are more conservative, as suggested by Petersen (2009) and Thompson (2011). 12 Details on the construction of all variables are provided in the Appendix. Next, I exploit the heterogeneity in financial and economic development and 12 Following Bates et al. (2009), I use the double-clustered (or Rogers) standard errors suggested by Petersen (2009), Moulton (1986), and Thompson (2011) to account for unobserved time and firm effects. Petersen (2009) finds that standard errors clustered by time are much larger than standard errors clustered by firm, and recommends clustering by time. Clustering by the higher level of aggregation (in my case, by country) is generally preferable (Cameron, Gelbach, and Miller, 2006), but it can give rise to distortions if the number of clusters is small and the cluster size is uneven, as is the case with my sample (Nichols and Shaffer, 2007). Specifically, since the number of observations for each country in my data set is not even, I use standard errors clustered by firm rather than country. 20

23 contractual environment across countries and over time to identify the channel through which asset tangibility affects corporate cash holdings. Specifically, I test hypotheses (2) and (3) using the following regression model,,,,,,,, 2,, where i, c, j, and t denote firm, country, industry, and year, respectively. Financial development is measured by private credit to GDP, a popular proxy for the development of financial intermediaries that captures the demand-side effect for collateralizable tangible assets from financial systems. Since private credit to GDP might be an outcome of financial institutions, I use creditor rights and information sharing as proxies for financial development. I anticipate a positive sign on the interaction of asset tangibility with financial development ( 0). Since a reduction in the potential liquidation costs, which stem from economic booms, increases the redeployability of tangible assets, I expect a negative sign on the interaction term between asset tangibility and economic development proxied by GDP per capita ( 0). 4. Empirical Results on Asset Tangibility and Cash Holdings In this section, I first provide international evidence on the tangibility-cash link and identify this collateral channel through cross-country variations in financial and economic development. Second, I employ alternative country-level measures the quality of institutions and the redeployability of tangible assets to investigate how these measures influence the cash-tangibility sensitivity. Third, I evaluate how a firm s default risk affects collateralization rates and how the quality of institutions affects collateral spread. Fourth, I 21

24 utilize industry-level measures of asset salability to further identify the collateral channel. Finally, I conclude with a series of robustness tests The Effect of Financial and Economic Development on Cash-Tangibility Sensitivity Table 2 presents estimates from cross-country ordinary least squares regression exploring the distinct effects of financial development and economic development on the cash holding sensitivities to asset tangibility. Financial development is proxied by Private Credit per GDP. I use the natural logarithm of country real gross domestic product per capita to measure the overall status of a country s economic development and activity. In Model 1 of Table 2, I begin the assessment of whether the redeployability measure of asset tangibility alone can explain variations in cash holdings. Only controlling for country, industry and year fixed effects, I find that the coefficient on asset tangibility is negative and statistically significant at the 1% level. This result indicates that having a high value of potential collateralizable tangible assets substantially and significantly decreases corporate cash holdings. Model 2 presents estimates of the regression model Eq. (1). The result reveals that the effect of asset tangibility on cash holdings becomes somewhat smaller after controlling for covariates. The coefficient on asset tangibility is and still significant at the 1% level. 13 The size of the effect is nontrivial and I interpret the economic meaning of the estimated coefficient on asset tangibility as follows. All else equal, a move from the 25 th percentile of the asset tangibility ratio (0.291) to the 75 th percentile (0.484) decreases cash holdings by about 12.6%, which corresponds to a decrease of 47.5% relative to the sample mean effect of In other words, if the firm decreases its tangibility ratio by one-interquartile range (IQR), it has to hold 47.5% more cash ratio relative to its mean value. Alternatively, 13 The results remain qualitatively unchanged when I use alternative definitions of the cash ratio, including cash to net assets, and log of cash to net assets, and when I replace asset tangibility by net tangibility. 22

25 interpreting the coefficient at the sample mean, for a $1.00 decrease in tangible assets, ceteris paribus, I expect to see a 65-cent increase in cash holdings. The negative connection between tangibility and cash simply reflect the redeployability of tangible assets pledged as collateral. 14 [Table 2 about here] Model 3 presents estimates of the baseline regression model Eq. (2) in which I combine the two avenues that influence the sensitivity of cash holdings to asset tangibility. The results also indicate that first, the coefficient on the interaction of financial development with asset tangibility is positive and significant at the 5% level, suggesting that the collateral role of tangible assets on cash holdings is less pronounced in financially developed countries. These results confirm the second hypothesis and provide support for the benefits of financial development. It suggests that the ease of raising money may actually lead firms to hold less cash in financially developed countries. The results are also consistent with findings in Liberti and Mian (2010) that creditors in countries with better financial development demand lower collateralization rates, implying that creditors may be able to use alternative instruments to constraint firms from risk-taking behavior. Second, I find that the coefficient on the interaction term between asset tangibility and log of GDP per capita is negative and significant. It suggests that the effect of firms supply of collateralizable tangible assets on cash-tangibility sensitivity is most pronounced in countries with high economic development where there are more economic activities and where the markets for assets are presumably more liquid. This confirms the third hypothesis that tangible assets are more salable in countries with more economic activities and the increase in salability enlarges the negative effect of asset tangibility on cash holdings. 14 Note that the negative relation between cash and the chosen measure of asset tangibility may not be mechanical, because in the regression I do not simultaneously control for Compustat balance sheet (asset side) items ACO, IVAEQ, IVAO, INTAN and AO (Other Current Assets, Investment and Advances in Equity, Other Investment and Advances, Intangible Assets, and Other Assets, respectively). Moreover, the results are robust to exclusion of total book leverage as firm-specific control variables. 23