The choice of private versus public debt in an emerging economy: does information asymmetry play a role?

Rajagopal, Sanjay

INTRODUCTION

More concentrated ownership and greater flexibility to renegotiation in the private debt market is argued to result in superior ex-ante information among private lenders, and in an increased creditor control over the borrower's investment and liquidation decisions. This mitigates in private markets the problems of adverse selection and moral hazard associated with credit generation (see, for example, Smith & Warner (1979), Diamond (1984), Blackwell & Kidwell (1988), and Berlin & Loeys (1988)). However, there are also costs to private debt in the form of monitoring costs, bank regulatory taxes, premiums for reduced liquidity, and agency costs of delegated monitoring (see Berlin & Loeys (1988), Diamond (1991), and Zwick (1980)), and the tradeoff of benefits and costs suggests the existence of an "optimal" level of private debt. This line of argument suggests that for firms with larger information asymmetries, the benefits of private debt will tend to outweigh the costs of private lender monitoring, ceteris paribus.

A growing body of literature has addressed, from both a theoretical and an empirical perspective, the precise factors influencing this private/public debt choice (e.g., Rajan (1992), Diamond (1993), Houston & James (1996), Johnson (1997), and Krishnaswami et al. (1999)). While the results of this research generally suggest a unique role for private debt markets in the presence of information asymmetry-and consistently indicate the relevance of a well-defined set of firm-specific factors to the choice of debt source-the empirical work remains largely limited to the context of a developed economy such as the US.

In contrast to the bulk of the extant research, the current paper focuses on the firm's private/public debt choice in the context of an emerging economy, India. This country represents an interesting case because, while its traditionally over-regulated domestic capital markets and commercial banking system have increasingly been exposed to the discipline of market forces, the process of reform has by no means brought its financial system yet to quite the levels of competition, efficiency and relative transparency found in developed countries (see indications of this in, for example, the Guha-Khasnobis & Bhaduri (2000) study). In this state of "partial adjustment" towards a competitive financial system, does private debt play a significant monitoring and certification role? Or is a predominantly publicly-owned banking and financial sector-the principal lenders in the private debt market-so constrained by regulation, so shielded from the discipline of market forces, or so affected by inefficiencies, nepotism and corruption that the existence of private debt relations with its institutions do not convey any valuable information to the investor? By using a logistic regression model, this paper examines whether the likelihood of a publicly traded Indian firm's exclusive dependence on private debt is systematically related to firm-level measures of information asymmetry. If there is a systematic positive relationship between that likelihood and information asymmetry, the results would be consistent with private lenders having a special monitoring role similar to the one documented in developed economies.

The primary purpose of this study, then, is to ascertain whether the observed role of lenders in the private debt market in a developed economy such as the US can be generalized to an emerging economy like India, in which the financial institutions have traditionally been largely under state control and shielded from market discipline. Its findings are potentially significant in assessing any progress the country has made in reforming its financial sector. The paper is organized as follows. The section below provides an overview of the financial sector reforms undertaken by India during the 1990s. Following that, the literature on the firm's private/public debt choice is reviewed. The end of that section also presents the hypothesized relationships for the current study. The next section discusses the data and variables used in the study, along with criteria for sample selection, and a univariate differences-in-means analysis of the group of firms relying solely on private debt versus the group of firms with public debt in their capital structure. In the section following, multivariate logistic regressions results on the private/public debt choice are provided and discussed. The final section provides conclusions and implications, along with suggestions for further research.

A BRIEF OVERVIEW OF FINANCIAL SECTOR REFORM IN INDIA

Financial sector reform has been an integral part of the new economic policy adopted by India beginning in 1991 (see, for example, Guha-Khasnobis & Bhaduri (2000)). This process of financial reform has entailed, among other things, a significant reduction in the cash reserve requirement (CRR) and the statutory liquidity ratio (SLR). Between 1991 and 1998, for instance, the SLR declined from 38.5% to 25%, and the CRR has declined from about 25% to 10.5% (for a more detailed discussion of the reforms, see Ahluwalia (1999); Beim & Calomiris (2001); and Laeven (2003), from which much of the discussion in this section is adapted, and which are also the source for the data quoted herein). Thus, the proportion of incremental resources to banks (from deposits) that was pre-empted by the government was roughly 65% prior to the reforms; that number now stands at about 36%. Put another way, the "tax" on financial intermediation has significantly been reduced over the 1990s.

In addition, interest rate controls have been eased progressively, moving the loan market away from a regime of subsidized rates and towards a more rational, market-based system. Thus, banks are relatively more free to price loans on the basis of fund costs and credit risk. The easing of entry restrictions to a traditionally state-controlled banking sector, which occurred in 1993, is reflected in the fact that between 1991 and 1997, the market share of private and foreign banks increased from roughly 11% to approximately 18%.

Another significant development has been the introduction of capital adequacy standards for banks. Prudential norms somewhat similar to the ones recommended by the Basle Committee were phased in by 1996, which, in addition to lending a certain transparency to the balance sheets of banks, lean on the institutions to improve asset quality. Bank supervision has also been strengthened, with the establishment of the Board for Financial Supervision within the Reserve Bank of India (RBI), India's central bank. The role of internal controls and audit, and that of external auditors have been shored up, and the time taken for the inspection and follow-up cycle has been cut in half.

Despite the improvements described above, though, the question which is of particular relevance to the current paper is how the new, reformed standards compare with international practice. As Ahluwalia (1999) notes, the government-appointed Committee on Banking Sector Reforms (CBSR) has reported several deficiencies with regard to the comparability of the new Indian banking norms with international standards. For instance, capital to "risk-weighted" assets for banks continue to be below international standards. Indian standards are more lax with regard to the reclassification of "substandard" and "doubtful" assets; a greater period of delinquency is allowed by Indian banks before such assets are downgraded. "Directed credit policies", which require banks to earmark 40% of their commercial loans to "priority" sectors identified by the government, remain in place. Significantly, the government still maintains a majority ownership of public sector banks (which account for a significant share of the market). This public ownership "involves 'politicization' and 'bureaucratization' of banking" (Ahluwalia (1999); p.44). Thus, it is quite possible that Indian banks, despite the recent progress in liberalization, are beset by "cronyism" in loan making, and an impaired ability to respond to commercial and customer needs (or dictates of the market).

It is this "transitional" nature of the Indian financial sector that provides the context for the current paper, and motivates the question as to whether the private/public debt choice determinants identified for firms in developed economies also play a significant explanatory role in India. For example, if banks are operating in an environment divorced from the pressures of the market, are making credit decisions not on the basis of a rational assessment of risk but for the sake of political expediency, and have little incentive to monitor the borrower, or generate information that is economically valuable to loan making, then the roles of variables such as the "age", "size", "risk", or "collateral asset value" of the borrowing firm, which appear to systematically explain the private/public debt choice in developed economies, might not be significant factors in the Indian context. Put another way, if private lending practices are not "fair" and "economically rational", then we are unlikely to observe economically sensible patterns in the private/public debt choice of Indian firms. This point will be elaborated upon at the end of the following section.

THE PRIVATE VS PUBLIC DEBT CHOICE AND INFORMATION ASYMMETRY

Since the theoretical and empirical research on the monitoring role of banks and private lenders is well established and abounds in the literature (the theoretical more so than the empirical), this section will provide only a brief overview of a representative sample of these studies. The main concern of this part of the paper will be to provide a justification for the inclusion of the (six) explanatory variables in our study, and to provide a basis for its a priori expectations. These explanatory variables, along with their expected influence on the private/public debt choice, are summarized at the end of the section.

Financial policy research has sought to explain the cross-sectional variation in the "placement" choice of debt. Most of the work on this subject has demonstrated either theoretically or empirically a special "monitoring" role for lenders in the private debt market. Since in private markets the ownership of debt is more concentrated than in the public debt markets, and there is also greater flexibility to renegotiation in the former, it is argued that private lenders have superior ex-ante information and increased control over the borrower's investment and liquidation decisions. This mitigates in private markets the problems of adverse selection and moral hazard associated with credit generation (see, for example, Smith & Warner (1979), Diamond (1984), Blackwell & Kidwell (1988), and Berlin & Loeys (1988)). Interest in the subject has been sustained through the past decade, and the more recent studies on the issue include Diamond (1991), and Rajan (1992). The first study stresses a "reputational capital" effect in the debt source choice, wherein larger firms, with established reputations (to maintain) are more likely to access the public debt market since they do not need to rely now on the monitoring role of banks. The study by Rajan (1992) stresses possible "hold up" problems when firms borrow from a single bank (which might have an informational monopoly), which makes reliance on bank debt less likely for firms with high quality projects. Recent empirical work on the subject has been conducted by Easterwood & Kadapakkam (1991), Houston & James (1996), Johnson (1997a), and Krishnaswami et al. (1999), among others. On balance, the vast majority of these studies have confirmed the presence of a special monitoring function for privately negotiated debt.

Easterwood & Kadapakkam (1991), who study the effect of transactions costs and leverage-related costs on the private/public debt choice, find that larger firms, with their larger issue sizes, save on transactions costs (the large "fixed" cost portion being spread over a large issue size), and tend to lean on public debt more than medium-sized firms do. While it is their conjecture that information asymmetry are also likely to be important to the private/public debt choice, they do not directly test this proposition. Houston & James (1996) find evidence consistent with Rajan's (1992) "hold up" problems model, a problem mitigated by having multiple-banking relationships. Johnson (1997a) finds, among other things, that firms have a greater reliance on public debt if they face lower information and monitoring costs. Krishnaswami et al. (1999) also find evidence consistent with the hypothesis that younger firms, and those with greater potential information asymmetries, will tend to rely more on private debt.

It should be noted that some theoretical studies distinguish between bank and non-bank private lenders; and they ascribe the unique monitoring role to the former. But most theoretical and empirical studies of debt ownership do not. And when the distinction is made, the evidence is not consistent (see Johnson (1997a), who considers the distinction between bank and non-bank private debt to be an empirical issue.)

Several studies employ the natural logarithm of a firm's total assets as a proxy for information availability and/or economies of scale in flotation costs (as found in the Easterwood & Kadapakkam (1991) study). Some of the studies employing this proxy include Krishnaswami et al. (1999), and Hadlock & James (2002). This "size" variable is therefore expected to have a negative effect on the likelihood of a firm issuing private debt.

Barclay & Smith (1995), Krishnaswami et al. (1999), and Hadlock & James (2002) use the market-to-book ratio as a proxy for a firm's growth options. To the extent that these options are difficult to value, the market value-to-book ratio can be used as a proxy for asymmetric information. Given the special monitoring role of private lenders, it is expected that the market value-to-book ratio will have a positive relationship with the likelihood of a firm choosing private debt.

Krishnaswami et al. (1999) employ the residual volatility in a firm's stock returns as a proxy for information asymmetry. Hadlock & James (2002) argue that firms with greater stock return volatility are more likely to possess large information asymmetry between insiders and outsiders. The latter measure this volatility as the standard deviation of the firm's daily stock returns. It is expected that there will be positive relationship between the volatility measure and the likelihood of privately placed debt. Note that this is the variable of primary interest in the present study, which focuses on information asymmetry; while other variables such as "size" or "market value-to-book" ratio might also proxy for asymmetric information, they are also related to other factors that affect the private/public debt choice-such as economies of scale in flotation costs, or growth options, as noted earlier.

Johnson (1997a) uses "age", or the number of years since incorporation, as a proxy for reputation to test the Diamond (1991) hypothesis regarding the role of reputational capital and choice of public debt. Since the reputational capital of older firms ensures that they will avoid actions harmful to creditors even if unmonitored (since their reputation is valuable and represents a large investment at risk if tarnished), it is expected that the "age" variable will be negatively related to the likelihood of choosing private debt.

Johnson (1997a) and Johnson (1997b) employ the ratio of fixed assets to total assets as a proxy for the assets that can be used as collateral to reduce the asset substitution problem. The reduction in asset substitution problems should increase access to public debt. Hoshi et al. (1993) argue that firms with more valuable "assets in place" will use public debt since the implied "collateral at risk" will bond the firm's investment decisions and mitigate the asset substitution problem. The fixed asset ratio is therefore expected to be negatively related to the likelihood of use of private debt.

Diamond (1993) suggests that the more leveraged firms may limit the use of bank (private) debt so as to frequent firm liquidation (by private/bank lenders). Hadlock & James (2002) suggest that leverage might be a proxy for lower agency costs of debt. Consequently, the leverage variable is expected to be negatively related to the likelihood of choosing private debt.

In the present study, therefore, the standard deviation of stock returns and market value-to-book ratio are used as proxies for information asymmetry. The size of the firm also serves as a proxy for asymmetric information (with a smaller firm size suggesting greater informational opacity), but, as has already been pointed out, the size measure could also proxy for the firm's access to the public debt market, which is characterized by economies of scale in issuance. The age of the firm proxies for the entity's reputational capital, which has been argued by researchers to influence the choice of debt source. The fixed asset ratio measures the value of the firm's assets that might be used as collateral in order to mitigate asset substitution problems. Finally, the firm's leverage is also included as a negative indicator of the firm's agency costs of debt.

DATA AND UNIVARIATE ANALYSIS

The data used in this study are collected from a publicly available database compiled by the Center for Monitoring the Indian Economy (CMIE). This dataset provides a comprehensive coverage of firms operating in India's industrial sector, and carries substantial financial statement information and some market data. All the income statement data in this study pertain to the year ending December 2002, and all the balance sheet information is as of December 31, 2002. Market data are averaged over the 365 days ending December 31, 2002.

This study employs the following seven variables, using the acronyms indicated:

 Variables: Acronyms and Definition

PVTDEBT This is a binary variable which takes a value of 1 if the
 firm has only private debt, and 0 otherwise. This serves
 as the dependent variable in the logistic regression.

AGE Age is the number of years since the firm's incorporation,
 as of the year 2002.

FAR The fixed asset ratio (FAR) is calculated as the ratio of
 the firm's net fixed assets to total assets.

MVB The market value-to-book ratio (MVB) is the firm's the
 sum of the book value of liabilities plus market
 capitalization, divided by the book value of total assets.

SD This measure of stock volatility is the standard deviation
 of the firm's daily returns average over a 365-day period.

LNASS This is the natural log of the firm's total assets, and is
 used as a proxy for firm size.

LEV Leverage (LEV) is calculated as the ratio of the book
 value of total liabilities over the book value of total
 assets.

In order to enter the sample, a firm had to meet the following the criteria:

Criteria for Entry into the Sample

1. The firm is categorized as a manufacturing firm in the
 CMIE dataset.

2. The firm is publicly traded.

3. The firm has some debt in its capital structure.

4. There is firm-specific data available on all seven variables
 for the firm for the time period of interest.

These four restrictions yield a final sample of 949 firms, of whom 633 (or about 67%) borrow solely in the private debt market. The table below provides the overall average for each variable, along with each variable's average for the sample partitioned into two groups-one that relies exclusively on private debt, and the other that has some public debt in its capital structure. The table also provides a simple difference-in-means test for each variable between the two groups.

The variable means in the table above reveal clear differences between firms that depend solely on private debt, and those that carry public debt. Firms that borrow solely in the private debt market are, on average, younger, smaller, have a smaller fixed assets-to-total assets ratio, have greater stock return volatility, and carry less leverage. These differences are statistically significant at the 1% level. However, there is no significant difference in the average market value-to-book ratios for the two groups of firms.

It should be noted that these univariate comparisons of firms in the Indian manufacturing sector yield results that closely parallel the ones reported by Hadlock & James (2002) for firms in the US. They too find firms without public debt to be, on average, smaller, younger, less levered, and characterized by greater stock return volatility. The market value-to-book ratio does not differ significantly between the two groups of firms in their study either. Unlike their study, however, we find that there is a distinct difference in the proportion of fixed assets between the two groups. The following section supplements these univariate comparisons with the results of a multivariate logit regression which estimates the likelihood that a firm will rely solely on private debt.

MULTIVARIATE LOGIT REGRESSION RESULTS

The data summary indicates that approximately two-thirds of the 949 firms in the sample rely solely on private debt for their borrowing needs, while the others carry public debt in their capital structures. Also, it is apparent that there are considerable differences between the two groups of firms on almost all the explanatory variables included in study. The focus of the our analysis now turns to ascertaining whether the likelihood of a firm's exclusive dependence on private debt is systematically related to proxies for information asymmetry. Again, the presumption is that private lenders can more closely monitor the borrower, and more accurately price a firm's claims, thereby mitigating adverse selection problems. The benefit of private debt financing, if it indeed exists in the case of India, should be greater for those firms that are characterized by greater information asymmetry or opacity.

As noted previously, the standard deviation of stock returns (SD) and market value-to-book ratio (MVB) are used as proxies for information asymmetry. The size of the firm (LNASS) also serves as a proxy for asymmetric information (with a smaller firm size suggesting greater informational opacity), but, as has already been pointed out, the size measure could also proxy for the firm's access to the public debt market, which is characterized by economies of scale in issuance. The age of the firm (AGE) proxies for the entity's reputational capital, which has been argued by researchers to influence the choice of debt source. The fixed asset ratio (FAR) measures the value of the firm's assets that might be used as collateral in order to mitigate asset substitution problems. Finally, the firm's leverage (LEV) is also included as a negative indicator of the firm's agency costs of debt. These proxies have been employed by previous studies, which have been cited above and are not repeated here. The following table provides the results of three estimated models; the motivation for estimating three models is provided below.

The empirical analysis in this section is built around the model of debt source choice estimated by Hadley & James (2002). Model 3, which is the most "inclusive" in that it incorporates all the six regressors in the equation, conforms to the relationship defined by Hadley & James. However, both theoretical and practical considerations compel us to begin with a model with just AGE, FAR, MVB, and SD. Recall that our main focus is on the signs and significance of the two proxies included purely to represent information asymmetry-namely MVB and SD; the other regressors enter as control variables. Yet, as the list of variables grows, the chance of multicollinearity among any pair of variables rises considerably. As the correlation matrix below indicates, there is a high degree of collinearity between the size variable (LNASS) and several of the variables included in Model 1. In addition to the issue of multicollinearity, one has to contend with the theoretical basis for expecting a relationship between leverage (LEV) and several of the other regressors in the model. Indeed, in his empirical study of the ownership structure of corporate debt in the US, Johnson (1997a) confronts this issue, noting that the inclusion of the leverage variable in the equation renders the interpretation of the regression coefficients problematic, since theory suggests that this debt measure is related to such variables as asset collateral value and firm size.

Given that we need to contend with these two methodological issues, we begin by estimating a simplified model employing only AGE, FAR, MVB, and SD as the explanatory variables, and recognizing the possibility of the confounding effects of omitted variables. However, note that the omitted variable LNASS plays a dual role, in that it represents the accessibility that larger firms have to the public debt market, but it also proxies for information availability. This fact might mitigate the "omitted variable" problem, given the main focus of this study. This, least inclusive, logit regression (Model 1) yields results that are entirely in keeping with our expectations: all the variables have the hypothesized signs, and are significant at the 1% level. While the "pseudo R-square" is reported, it should be interpreted with caution as in any logistic regression (see, for example, Aldrich & Nelson (1984)); we include the information purely to facilitate a comparison across the three models.

Model 2 attempts to address the issue of omitting LNASS by including an instrumental variable (LNASSr) in place of the missing size variable. This instrumental variable represents the residuals from the regression of LNASS on AGE, FAR, MVB, and SD. LNASSr is highly correlated with LNASS (rho=0.901, as the correlation matrix below indicates), but is orthogonal to the other regressors, and does not affect their coefficient estimates. The inclusion of LNASSr in the model leaves the sign and significance of all variables unaffected, except for MVB; the coefficient for this variable maintains its sign, but is no longer significant even at the 10% level. Significantly, the other measure of information asymmetry, SD, maintains its sign and significance. While the results are not reported here, we used an instrumental variable for SD (constructed in a manner similar to LNASSr), and found that the general nature of the results were identical to the those reported for Model 2; only the precise magnitudes of coefficients were slightly different, but their signs and significance were maintained.

Model 3 attempts to address the issue of the simultaneous inclusion of leverage and other variables (age, size, asset collateral value, etc.) with which it is theoretically related. As in Johnson (1997a), we partition leverage (LEV) into an "endogenous" ascribed to the other regressors, and an "exogenous" part that is independent of those variables. The latter, LEVr, represents the residuals from the regression of LEV on the other explanatory variables. This "exogenous" portion of LEV, which is highly and significantly correlated with LEV (rho=0.91) is then used as an instrumental variable in the logistic regression. The results of Model 3 indicate considerable stability in the magnitude of the coefficients, all of which-with the exception of MVB-maintain their signs and significance. The coefficient of the market value-to-book ratio maintains its negative sign, but is now significant at the 10% level. It should be noted that our inclusion of "instrumental variables" to try and circumvent multicollinearity or simultaneity problems is similar to the approach taken by other studies pertaining to corporate financial policy, such as Johnson (1997a) mentioned earlier and Bah & Dumontier (2001).

Overall, the results of the logistic regressions confirm the stylized facts that emerged from the univariate analysis of the previous section. The likelihood that an Indian manufacturing firm will have an exclusive dependence on private debt is found to be positively related to stock price volatility, and negatively related to size, age, fixed asset ratio and leverage. The market value-to-book ratio appears not to play a significant role, though it does enter as a mildly significant variable in Model 3, with the hypothesized sign. These results closely conform to the findings in the recent private/public debt choice studies conducted on US firms (e.g., Houston & James (1996), Johnson (1997a), and Hadlock & James (2002)).

CONCLUSION

The results of this study are entirely in keeping with the hypothesized relationships which were based upon observed relationships in developed economies, and predicated upon banks and other private lenders playing a unique monitoring function. Since traditionally the role of the private debt market has been studied in the context of these developed economies, with their well-functioning, competitive, and market-oriented financial institutions (which accounts for the bulk of private lending), the present study, with its contrasting look at an emerging economy, constitutes a useful addition to the study of private lending.

India serves as a useful context within which to study the role of private debt markets, and the choice-of-debt source question because its commercial banking sector and other financial institutions have traditionally been over-regulated, but are now undergoing a process of liberalization. In such a situation of "partial adjustment" towards market discipline, the current study considered whether private debt played a role similar to the one observed in developed economies. The results of this paper suggest that that question may be answered in the affirmative, at least to the extent that the debt source choice appears to be related to information asymmetry in the case of Indian firms as well. This finding is consistent with the idea that the process of financial sector reform which gained momentum in the latter part of the 1990s has had a positive impact on a traditionally state-dominated industry. In that regard, these results counter some of the more bleak conclusions about the liberalization process in India that might be drawn from other studies, such as that by Guha-Khasnobis & Bhaduri (2000).

Since the process of financial sector reform is continuing in India, it might be informative to extend the time frame of this study in either direction of 2002, which was our time period of interest. For instance, a comparison might be made between the relationships observed for 2002 in this study, and another estimated model for, say 1992, which would pre-date a bulk of the reforms that were put into place by the Indian government during the 1990s. This would document the changing role of the financial institutions and intermediaries as monitors and information producers across different regimes of "financial repression". Finally, the study can also be made more comprehensive by including firms from the growing services sector within India.

REFERENCES

Ahluwalia, M. S. (1999). Reforming India's Financial Sector: An Overview. In J.A. Hanson & S. Kathuria (Eds.), India: A Financial Sector for the Twenty-First Century (pp. 29-54). OUP: India.

Aldrich, J. H. & F. D. Nelson (1984). Linear Probability, Logit, and Probit Models. Sage Publications.

Bah, R.& P. Dumontier (2001). R&D Intensity and Corporate Financial Policy: Some International Evidence. Journal of Business Finance & Accounting, 28, 671-692.

Barclay, M. & C. Smith (1995). The Maturity Structure of Debt. Journal of Finance, 50, 609-631.

Beim, D. O. & C. W. Calomiris (2001). Emerging Financial Markets. McGraw-Hill.

Berlin, M. & J. Loeys (1988). Bond Covenants and Delegated Monitoring. Journal of Finance, 43, 397-412.

Blackwell, D. & D. S. Kidwell (1988). An Investigation of Cost Differences Between Public Sales and Private Placements of Debt. Journal of Financial Economics, 22, 253-278.

Diamond, D. W. (1984). Financial Intermediation and Delegated Monitoring. Review of Economic Studies, 51, 393-414.

Diamond, D.W. (1991). Monitoring and Reputation: The Choice Between Bank Loans and Directly Placed Debt. Journal of Political Economy, 99, 689-721.

Diamond, D.W. (1993). Seniority and the Maturity of Debt Contracts. Journal of Financial Economics, 33, 341-368.

Easterwood, J. C. & P. Kadapakkam (1991). The Role of Private and Public Debt in Corporate Capital Structures. Financial Management, Autumn 1991, 49-57.

Guha-Khasnobis, B. & S. N. Bhaduri (2000). A Hallmark of India's New Economic Policy: Deregulation and Liberalization of the Financial Sector. Journal of Asian Economics, 11, 333-346.

Hoshi, T., A. Kashyap & D. Scharfstein (1993). The Choice Between Public and Private Debt: An Analysis of Post-Deregulation Corporate Financing in Japan. Working Paper, NBER (1993).

Houston, J. & C. James (1996). Bank Information Monopolies and the Mix of Private and Public Debt Claims. Journal of Finance, 51, 1863-1889.

Hadlock, C. & C. James (2002). Do Banks Provide Financial Slack? Journal of Finance, 57 (3), 1383-1419.

Johnson, S. (1997). An Empirical Analysis of the Determinants of Corporate Debt Ownership Structure. Journal of Financial and Quantitative Analysis, 32, 47-69.

Krishnaswami, S., P.A. Spindt & V. Subramaniam (1999). Information Asymmetry, Monitoring, and the Placement Structure of Corporate Debt. Journal of Financial Economics, 51, 407-434.

Laeven, L. (2003). Does Financial Liberalization Reduce Financing Constraints? Financial Management, 32, 5-34.

Rajan, R. (1992). Insiders and Outsiders: The Choice Between Informed and Arm's Length Debt. Journal of Finance, 47, 1367-1400.

Smith, C. & J. Warner (1979). On Financial Contracting: An Analysis of Bond Covenants. Journal of Financial Economics, 7, 117-161.

Sanjay Rajagopal, Montreat College

Variables: Hypothesized Direction of Relationship With
Likelihood of Private Debt

 Variable Hypothesized Influence on Private Debt Choice

Age Negative
Fixed Asset Ratio Negative
Market Value-to-Book Positive
Volatility Positive
Size Negative
Leverage Negative

Data Summary and Univariate Analysis

Variable All Firms With With t-statistic
 (N=949) Private Public
 Debt Only Debt
 (N=633) (N=316)

AGE 26.45 23.63 32.12 -6.01 ***
LNASS 4.93 4.32 6.16 -19.75 ***
MVB 0.94 0.95 0.92 0.91
FAR 0.40 0.38 0.46 -6.10 ***
SD 0.07 0.08 0.05 5.16 ***
LEV 0.59 0.54 0.69 -11.01 ***

*** The difference in means between the "no public debt" and
public debt groups is significant at the .01 level.

Logit Models of Private Debt Choice

 Variable Model 1 Model 2 Model 3

AGE -0.0196 -0.033 -0.035
 (-5.23) *** (-7.60) *** (-7.57) ***
FAR -2.73 -3.28 -3.52
 (-6.86) *** (-7.01) *** (-7.22) ***
MVB 0.5275 0.2717 0.492
 (2.84) *** -1.38 (1.88) *
SD 0.2149 0.036 0.096
 (7.9) *** (2.71) *** (2.92) ***

Logit Models of Private Debt Choice

Variable Model 1 Model 2 Model 3

LNASS -- -1.09 -1.103
 (-12.99) *** (-11.59) ***
LEV -- -- -3.23
 (-5.89) ***
Intercept 0.5731 2.7488 2.424
 (1.73) * (7.87) *** (5.24) ***
N 949 949 949
Ps. R-Sq 0.14 0.28 0.30

Notes:
Each cell shows the estimated coefficient, with the asymptotic
t-values in parentheses.

*** indicates that the coefficient is significant at the .01
level; * indicates significance at the 0.10 level. The size
proxy in Models 2 & 3 is the residual from the regression of
LNASS on AGE, FAR, MVB, SD. Leverage in Model 3 is the residual
from the regression of LEV on the other explanatory variables.

Correlation Matrix for Regressors

 AGE FAR MVB SD

FAR -0.080
MVB 0.042 -0.063
SD -0.077 -0.021 -0.100
LNASS 0.301 0.095 0.181
LEV 0.209 0.291 0.021 0.009
LNASSr 0.000 0.000 0.000
LEVr 0.000 0.000 0.000 0.000

 LNASS LEV LNASSr

FAR
MVB
SD
LNASS -0.271
LEV 0.269
LNASSr 0.000 0.901 0.191
LEVr 0.000 0.907 0.000