Why Do Some Firms Go Debt Free?

Authors


  • The authors would like to thank the GAMF, and James Park for his valuable suggestions. Soku Byoun greatly appreciates the support for this project that was provided by the Hankamer School of Business at Baylor University.

Corresponding author: Soku Byoun, Hankamer School of Business, Baylor University, One Bear Place #98004, Waco, Texas 76798, USA. Tel: (254) 710-7849, Fax: (254) 710-1092, email: soku_byoun@baylor.edu.

Abstract

This paper examines debt-free firms. We find that favorable equity market valuation and borrowing constraints contribute to these firms' extreme debt conservatism. Small debt-free firms with little access to credit markets are seen to raise equity while paying high dividends. Large debt-free firms, generating more cash flows relative to their investment needs, often pay off their debt while paying high dividends. The results suggest that high dividends for small debt-free firms help them establish good reputations in equity markets, while high dividends for large debt-free firms reduce the agency costs of free cash flow.

1. Introduction

“Microsoft, Walgreen, Cisco Systems and William Wrigley have something in common that may be surprising to many readers. None of them has any debt.”1 This story is provocative, but perhaps more surprising is that the proportion of debt-free firms has steadily increased over time, with over 20% of United States firms becoming debt free in recent years. In this study, we explore the phenomenon of extreme debt conservatism and provide potential explanations for it.

The debt conservatism puzzle refers to the notion that some firms have lower leverage than that which would maximize firm value from a static trade-off perspective (Miller, 1977; Graham, 2000; Frank and Goyal, 2005). Although recent studies on dynamic trade-off models produce relative lower optimal leverage ratios (Goldstein et al., 2001; Morellec, 2004; Strebulaev, 2007), these models cannot explain why firms do not use debt. Korteweg (2010) shows that debt-free firms could increase their value by 5.5% on average if they levered up to their optimal debt ratios.

An advantage of examining debt-free firms is that the leverage ratio is not affected by the choice of denominator for zero-debt firms. As Parsons and Titman (2008) point out, there is considerable ambiguity in measuring the leverage ratio. For instance, when we consider “low-leverage” firms with less than 5% leverage ratio as used in previous studies (see Minton and Wruck, 2001), the number of low-leverage firms is significantly different, depending on whether we use the book or market value leverage ratio (see Figure 1). Scaling debt by market values may also induce a mechanical relation (Titman and Wessels, 1988) while book leverage is subject to some measurement issues (Welch, 2007). In addition, dividing dependent and independent variables by common or correlated variables as in typical cross-sectional leverage regressions induces spurious correlations (Pearson, 1897; Powell et al., 2009). Examining debt-free firms avoids these issues. Furthermore, we show that debt-free and low-leverage firms have different characteristics.

Figure 1.

Distribution of book and market debt ratios for all firms in 2000

The interaction of capital structure and dividend policies further complicates the understanding of the capital structure conservatism puzzle. For instance, either debt or dividend can be effective in reducing the agency problem of free cash flow (FCF) (Easterbrook, 1984; La Porta et al., 2000; Fama and French, 2002). However, the existing agency models have not yet fully dealt with the substituting effects of dividends and leverage in addressing agency problems. Debt-free firms allow us to examine firms' dividend policies with a common capital structure (zero debt). Strebulaev and Yang (2007, p. 17) argue that the debt conservatism puzzle is actually “an artifact of the zero-leverage puzzle.” Thus, investigating debt-free firms helps us understand debt conservatism in particular and firms' capital structure and dividend policies in general.

Previous studies show that observed leverage ratios are negatively associated with equity market valuations (Baker and Wurgler, 2002), borrowing constraints (Faulkender and Petersen, 2006), and dividend payouts (DeAngelo and DeAngelo, 2006). Even though these cross-sectional studies provide valuable insights into the directional relationship between leverage ratios and firm characteristics, we still do not understand what factors contribute to firms' debt conservatism and why some firms even go debt free.

Our findings point to borrowing constraints and equity market valuation as important explanations for firms' going debt free. Using a set of proxies for borrowing constraints and market valuation, we find that equity markets are more favorable for debt-free firms, while credit markets are unfavorable for them. Our results suggest that the comparative advantage in issuing equity versus debt is an important factor for debt-free firms' financing decisions—especially for small firms. In the presence of high market valuations and good stock performance while facing borrowing constraints, firms rely more on equity financing and become debt free.

What is intriguing is that good equity market traits of debt-free firms are accompanied by high dividend payouts. Our findings reveal different motivations for high dividend payouts between large and small debt-free firms. Large debt-free firms with more profits relative to their investment opportunities increase dividends while reducing debt. Given that dividend and debt are substitutes for controlling free-cash-flow problems (Easterbrook, 1984; Fama and French, 2002), our findings suggest that large debt-free firms with more profits relative to their investment opportunities pay large dividends to address the free-cash-flow problem and to become debt free. Small debt-free firms, while being less profitable, pay high dividends while issuing equity. For these small debt-free firms, high dividends work as a means of establishing a reputation for moderation in expropriating wealth from shareholders (La Porta et al., 2000). The reputation for fair treatment of shareholders is worth the most for firms with strong needs for external financing in order to raise external equity on attractive terms (DeAngelo and DeAngelo, 2006). Such a reputation is credibly developed only when firms rely mainly on equity (Gomes, 2000). Accordingly, small debt-free firms build good reputations through high dividend payouts and equity issuances.

Our findings are particularly important in understanding the different motivations for dividend policies of firms with differing financing needs. The dividend policies of small debt-free firms reflect their efforts to retain access to equity financing. In line with La Porta et al. (2000), by maintaining high dividend payments despite having low profitability, these firms maintain their ability to raise equity on favorable terms by moderating shareholders' concern for agency problems of expropriation. Small debt-free firms' high dividends may substitute for other disciplinary factors in order to establish good reputations in the capital markets. Such reputations are especially important for small, growing debt-free firms as they depend heavily on equity financing for their investment needs. In contrast, large debt-free firms generate more cash from operations relative to their investment opportunities. They pay out excess cash through dividends and repurchases. For large debt-free firms, high dividends substitute for leverage in addressing the agency costs of FCF. Thus, debt-free firms' large dividends allow them to maintain access to equity capital on favorable terms and to become debt free.

We further show that debt-free firms maintain persistently negligible debt, and that extreme debt conservatism cannot be explained by non-debt tax shields, future financing needs (e.g. acquisitions), off-balance-sheet liabilities, or managerial entrenchment. We also compare debt-free firms with low-leverage firms (with debt between 0% and 5% of total assets). We find that low-leverage firms are different from debt-free firms in that they have easy access to the public debt market with good credit ratings. We also find that small debt-free firms pay significantly higher dividends than small low-leverage firms. These results further suggest that debt-free firms have distinctive characteristics from low-leverage firms.

This study adds new findings to a growing literature on the debt conservatism puzzle. Our results suggest different reasons for small and large firms becoming debt free: small debt-free firms rely on external equity with little access to credit markets, whereas large debt-free firms have paid off their debt with internal equity. Our findings contrast with those of Strebulaev and Yang (2007) who find that debt-free firms issue less equity than debt firms after they become debt free. We find that debt-free firms issue more equity than debt firms before they become debt free. After they become debt free, small debt-free firms continue to finance their investments with external equity, while large debt-free firms begin to repurchase equity with surplus cash flows, which may explain the results in Strebulaev and Yang (2007).

Our study also supplements the dividend policy literature. The dividend policies of large profitable debt-free firms reflect their efforts to address the free-cash-flow problem, whereas the dividend policies of small growth firms reflect their efforts to address the expropriation problem. This is an important finding because examining the dividend policy of debt-free firms effectively controls for the effect of leverage.

The remainder of the paper is arranged as follows. Section 2 describes the data and provides summary statistics. Section 3 discusses empirical implications. Section 4 reports univariate results and Section 5 presents results from the estimation of logit regressions. Section 6 explores alternative explanations and Section 7 provides a summary and concluding remarks.

2. Data

The initial data consist of all available United States industrial firms from the annual Compustat files for the period 1971–2006. Following previous studies, we exclude financial firms (Standard Industrial Classification; SIC codes 6000–6799) and regulated utilities (SIC codes 4900–4999) from the sample. We also exclude all firms with a Company Location Code (STATE) equal to 99, which indicates that the company's principal location is in a country other than the United States. We require firms to have positive values for total assets, common equity, number of shares outstanding, and stock price at the end of the fiscal year. After these requirements are applied, the sample consists of 150 810 firm-year observations. Since we use all available observations in each analysis, the sample size varies with data availability. For example, the sample size is reduced when we combine the initial sample with the data from the Center for Research in Security Prices (CRSP).

We define a firm that has neither current nor long-term debt in a given year as a “debt-free” firm, and a firm with any amount of debt in a given year as a “debt” firm. In order to illustrate the ambiguity of book and market leverage ratios, we depict the distribution of book and market leverage ratios for year 2000 in Figure 1. The number of low-leverage firms with leverage ratios between 0% and 5% is 1179 for book leverage, while it is 1606 for market leverage. The average book leverage ratio of those non-overlapping firms (506) in the low-market-leverage ratio group is about 16%. Similar results are found for other years (not reported). Thus, there are significant differences in terms of the number and the average leverage ratio of low-leverage firms depending on how we define the leverage ratio. We avoid these issues by separating debt and debt-free firms.

Table 1 reports the yearly summary statistics of debt-free and debt firms for selected years before 2000 and consecutive years from 2001 to 2006, as well as for the full sample. The number of debt-free firms as a percentage of sample firms each year is between 5.91% and 22.90% and has steadily increased over time. Figure 2 depicts the increasing proportion of debt-free firms during the entire sample period. On average, 12.18% of the sample firm-years are debt free. Another salient feature of debt-free firms is that their size measured by total assets is much smaller than that of debt firms. To see whether debt-free firms use more equity financing as a substitute for debt financing, we also examine the common and preferred stock as percentages of total assets. The amount of common equity for debt-free firms represents 76.89% of total assets, which is significantly greater than 48.13% for debt firms. The proportion of preferred stock for debt-free firms is not significantly different from that of debt firms, implying that debt-free firms use mostly common equity as a means of financing.

Table 1. Summary statistics
YearDebtTotal assetsMV of EST debtLT debtCommonPreferred N %
  1. The data consist of 150 810 firm-year observations from Compustat for the period 1971–2006. A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. Total assets is the book value of assets in millions of dollars. MV of E is the market value of equity in millions. ST (LT) debt is the short-term (long-term) debt divided by total assets. Common (Preferred) is the book value of common (preferred) stock divided by total assets. N represents the number of observations each year. In the last column (%) is the percentage of debt or debt-free firms relative to the total number of sample firms.

1971Debt259.38236.430.07590.20300.50390.0088275293.38
Debt-free47.14113.740.00000.00000.78150.00251956.62
1976Debt382.71244.840.06150.20070.48690.0061299394.09
Debt-free54.0873.620.00000.00000.75050.00031885.91
1981Debt518.13264.740.07270.19040.47830.0068375192.21
Debt-free37.7147.770.00000.00000.79390.00233177.79
1986Debt632.84466.220.08250.19370.47210.0068399190.79
Debt-free47.8988.760.00000.00000.80040.00944059.21
1991Debt1064.98820.350.07740.18270.47590.0090353688.16
Debt-free65.04159.100.00000.00000.78420.005647511.84
1996Debt1057.941142.320.05950.17880.49880.0080498984.76
Debt-free105.94368.520.00000.00000.77630.009889715.24
2001Debt2062.072750.830.06610.17660.49200.0103388681.62
Debt-free290.751159.060.00000.00000.75850.013187518.38
2002Debt2212.301915.500.05980.17900.48040.0083354780.21
Debt-free323.84866.690.00000.00000.73420.011387519.79
2003Debt2542.082628.320.05290.18040.48320.0084331878.70
Debt-free374.571080.800.00000.00000.74850.008789821.30
2004Debt2815.713024.040.04800.17250.49290.0079322277.14
Debt-free381.901154.660.00000.00000.75010.010295522.86
2005Debt3024.363232.120.04590.17220.49270.0079306777.10
Debt-free346.781073.680.00000.00000.74940.010591122.90
2006Debt3517.334076.930.03990.18120.49500.0053258077.43
Debt-free459.271447.660.00000.00000.74790.009075222.57
AllDebt1075.881070.000.06850.18950.48130.0080132 44487.82
Debt-free177.38629.380.00000.00000.76890.007918 36612.18
Figure 2.

Number of debt and debt-free firms: 1971–2006

Table 2 includes the number of firms with various debt-free years during the sample period and the mean and median percentages of debt-free years relative to the number of years observed (number of debt-free years/total number of years observed). About 30% of firms have had at least 1 year of debt-free capital structure during our sample period. A small number of firms have operated without debt for most of the sample period. For most firms (about 95%), however, debt-free status is limited to fewer than 6 years—which is, on average, one-half of the years observed during our sample period.2 Figure 3 also shows book leverage ratios for debt and debt-free firms around the debt-free years.3 The total debt ratio of debt-free firms tends to become lower until the debt-free year and then increases thereafter. However, their leverage ratios tend to be very low relative to those of debt firms throughout the 11-year period.

Table 2. The distribution of firms across the number of debt-free years
No. of debt-free yearsNumber of firmsPercentageCumulative percentageMeanMedian
  1. The data consist of 150 810 firm-year observations (15 686 firms) from Compustat for the period 1971–2006. A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. No. of debt-free years represents the number of years during which the firm has no debt. Mean and median are the mean and median percentages of debt-free years relative to the available sample period (No. of debt-free years divided by total number of years observed) respectively.

010 90069.4969.490.00000.0000
114739.3978.880.12150.0769
29115.8184.690.22440.1429
36233.9788.660.29010.2143
44482.8691.510.35170.2857
53132.0093.510.42390.3571
62151.3794.880.47510.4000
71871.1996.070.52850.4667
81240.7996.860.55450.5000
91090.6997.560.57980.5294
10830.5398.090.62220.5882
11790.5098.590.60560.5789
12390.2598.840.65000.6316
13400.2699.090.65740.6500
14250.1699.250.55120.4667
15220.1499.390.74160.7895
16110.0799.460.67430.6957
17140.0999.550.78810.7391
18160.1099.660.65010.6207
19110.0799.730.75720.7308
2090.0699.780.64750.5882
21110.0799.850.77000.7500
2260.0499.890.78570.7586
2340.0399.920.74190.6389
2450.0399.950.72290.6667
2520.0199.960.69440.6944
2610.0199.970.72220.7222
2700.0099.970.00000.0000
2800.0099.970.00000.0000
2900.0099.970.00000.0000
3020.0199.970.83330.8333
3100.0099.980.00000.0000
3200.0099.980.00000.0000
3300.0099.980.00000.0000
3410.0199.990.94440.9444
3520.01100.000.98590.9722
3600.00100.000.00000.0000
Total15 686100.00   
Figure 3.

Book leverage ratios around debt-free year

Table 3 reports the distribution of debt-free and debt firms across two-digit SIC codes. Metal Mining, Pipelines (except Natural Gas), Business Services, and Legal Services industries have especially high numbers of debt-free firms (more than 20%). Twelve industries have more than 15% debt-free firms. Debt-free firms are not uncommon in most industries.

Table 3. Distribution of debt and debt-free firms across industries
Two-digit industry codeDebt firmsDebt-free firmsTotal
N % N % N
  1. The data consist of 150 810 firm-year observations from Compustat for the period 1971–2006. A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. N represents the number of observations. % is the percentage of firms relative to the total number of firms in each industry.

01–09Agriculture, Forestry, and Fishing68088.438911.57769
10Metal Mining91979.0924320.911162
12Coal Mining28896.64103.36298
13Oil and Gas Extraction619786.2598813.757185
14Mining and Quarrying of Nonmetallic Minerals, except Fuels32889.134010.87368
15Building Construction General Contractors and Operative Builders142096.08583.921478
16Heavy Construction Other than Building Construction Contractors51794.52305.48547
17Construction Special Trade Contractors48892.60397.40527
20Food and Kindred Products422592.923227.084547
21Tobacco Products16898.8221.18170
22Textile Mill Products165496.84543.161708
23Apparel and Other Finished Products202691.841808.162206
24Lumber and Wood Products, except Furniture113089.4713310.531263
25Furniture and Fixtures127593.68866.321361
26Paper and Allied Products197995.37964.632075
27Printing, Publishing, and Allied Industries253691.592338.412769
28Chemicals and Allied Products916082.20198417.8011 144
29Petroleum Refining and Related Industries112397.06342.941157
30Rubber and Miscellaneous Plastics Products244794.811345.192581
31Leather and Leather Products64488.838111.17725
32Stone, Clay, Glass, and Concrete Products153796.55553.451592
33Primary Metal Industries279996.85913.152890
34Metal Products, except Machinery and Transportation Equipment348093.052606.953740
35Industrial and Commercial Machinery and Computer Equipment10 01386.96150213.0411 515
36Electronic and Other Electrical Equipment and Components10 89086.20174313.8012 633
37Transportation Equipment379692.952887.054084
38Photographic, Medical, and Optical Goods; Watches and Clocks875781.85194218.1510 699
39Miscellaneous Manufacturing Industries187689.9820910.022085
40Railroad Transportation50494.38305.62534
41Interurban Highway Passenger Transportation8395.4044.6087
42Motor Freight Transportation and Warehousing134293.52936.481435
44Water Transportation48096.97153.03495
45Transportation by Air114497.44302.561174
46Pipelines, except Natural Gas5979.731520.2774
47Transportation Services45683.069316.94549
48Communications382193.842516.164072
50Wholesale Trade: Durable Goods449690.864529.144948
51Wholesale Trade: Non-durable Goods250294.311515.692653
52Building Materials Hardware/Garden Supply Mobile Home Dealers52496.15213.85545
53General Merchandise Stores167296.93533.071725
54Food Stores152794.96815.041608
55Automotive Dealers and Gasoline Service Stations55398.5781.43561
56Apparel and Accessory Stores133380.9431419.061647
57Home Furniture, Furnishings, and Equipment Stores87691.92778.08953
58Eating and Drinking Places272393.351946.652917
59Miscellaneous Retail285090.023169.983166
70Hotels, Rooming Houses, Camps, and Other Lodging Places79494.52465.48840
72Personal Services50492.99387.01542
73Business Services11 45676.90344123.1014 897
75Automotive Repair, Services, and Parking39194.67225.33413
76Miscellaneous Repair Services11191.74108.26121
78Motion Pictures119886.7518313.251381
79Amusement and Recreation Services147789.6217110.381648
80Health Services277593.661886.342963
81Legal Services1266.67633.3318
82Educational Services41284.957315.05485
83Social Services23591.44228.56257
84Museums, Art Galleries, and Botanical and Zoological Gardens14100.0000.0014
86Membership Organizations3100.0000.003
87Engineering Accounting Research Management Related Services239681.6154018.392936
88Private Households466.67233.336
99Nonclassifiable Establishments136573.1950026.811865
 Total132 44487.8218 36612.18150 810

3. Empirical Implications

Given that a significant number of firms are debt free and the number of debt-free firms has grown to over 20% of sample firms in recent years, the phenomenon we study is economically important. We incorporate the implications of existing theories and previous findings in order to motivate the empirical analyses that follow.

3.1. Borrowing Constraints

Stiglitz and Weiss (1981) suggest that market frictions may cause firms to be rationed by their lenders, leading some firms to appear under-levered relative to unconstrained firms. Thus, when estimating a firm's leverage, it is important to consider not only determinants of its desired leverage (the demand side), but also the constraints on a firm's ability to increase its leverage (the supply side). Faulkender and Petersen (2006) show that firms with good access to the public debt market use much more debt than do firms without such access. Bolton and Feixas (2000) also argue that small, growing firms prefer to reduce information dilution costs by funding their investments through bank loans or bond issues but are not able to obtain bank loans or issue bonds because of their high-risk status. In addition, small, growing firms are in the stage of reputation acquisition with no favorable borrowing track record (Diamond, 1991), and are most likely to be turned down for credit. Thus, the only option for these firms is equity financing. Barclay et al. (2006) and Byoun (2012) also suggest that, due to higher costs and lower benefits of debt, firms in the development stage will abstain from issuing risky debt. Thus, firms with significant borrowing constraints may become debt free.

3.2. Equity Market Valuations

Baker and Wurgler (2002) argue that a firm's capital structure reflects the cumulative impact of managers' attempts to time the market so that they sell shares when they are overvalued by the market and repurchase shares when they are undervalued. Welch (2004) finds that firms with underperforming stocks have high debt ratios while firms with outperforming stocks have low debt ratios. Leary and Roberts (2004), in their study of the pecking order theory of capital structure, conclude that most equity issues are undertaken by opportunistic firms attempting to take advantage of high stock prices. Helwege and Liang (2004) and Alti (2006) also suggest that hot market initial public offerings are driven by opportunistic behavior by managers taking advantage of greater investor optimism. These studies suggest that firms issue equity when investors are optimistic about firm value. Thus, we hypothesize that firms take advantage of high stock valuation and become debt free.

3.3. Profitability, Investment Opportunities and Dividends

According to the pecking order theory (Myers, 1984; Myers and Majluf, 1984), firms with sufficient profit to fund their investment outlays are more likely to become debt free as they rely solely on internal funds. As noted by Fama and French (2002), however, the pecking order prediction regarding leverage is complicated by the firm's concern for future as well as current financing costs. Dynamic capital structure models also emphasize the importance of considering future financing needs in determining the current capital structure (Goldstein et al., 2001; Hennessy and Whited, 2005). These models imply that, given the adjustment costs of capital structure or adverse selection costs, firms with large expected investments may become debt free in order to avoid either forgoing future investments or financing them with new risky securities.

The agency models of Jensen and Meckling (1976), Easterbrook (1984), and Jensen (1986) suggest that firms with greater profitability commit a larger fraction of their earnings to debt payments or dividend payouts in order to prevent managers from wasting FCF. Since dividend and debt policies help control free-cash-flow problems (Easterbrook, 1984; Fama and French, 2002), firms with large profits may pay large dividends instead of using debt (DeAngelo and DeAngelo, 2006). Thus, mature firms with greater profits relative to investment opportunities may address the agency problem of FCF with large dividends instead of debt, and may ultimately become debt free.

La Porta et al. (2000) view dividends as a means of establishing a reputation for controlling expropriation of wealth from shareholders. Crucially, this view relies on the need for firms to raise external capital. Thus, the reputation for good treatment of shareholders is worth the most for firms with significant need of external financing. This view implies that firms with better growth prospects have a stronger incentive to establish a reputation that will support future external financing (Gomes, 2000). In other words, reputation can be credibly developed for treating shareholders well only when firms are mainly dependent upon equity for their financing. Accordingly, growing firms that have built good reputations through high dividend payouts may become debt free by raising external equity on favorable terms.

4. Univariate Analyses

4.1. Borrowing Constraints

We proxy borrowing constraints by firm size, cash holdings, tangible assets, capital intensity ratio, Standard & Poor's (S&P) short-term and long-term credit ratings, and S&P stock quality ranks.

We define firm size in three different ways: (i) book value of total assets; (ii) market value of total assets; and (iii) net sales. We provide detailed variable definitions in the appendix A. Since the results are similar for all three size proxies, we report only those based on the book value of total assets. Since firm size is expected to be correlated with other variables, we divide the sample into size quintiles each year and compare other firm characteristics between debt-free and debt firms within each size quintile.

We expect that firms with large cash holdings are more constrained and are more likely to become debt free than are firms with small cash holdings. Almeida et al. (2006) suggest that constrained firms should hold more cash in their balance sheets than unconstrained firms. Calomiris et al. (1995) also classify firms with high cash holdings as relatively constrained because they accumulate cash as precautionary savings in order to avoid the high costs of being financially constrained or distressed in the future. Consistent with these arguments, Opler et al. (1999), Minton and Wruck (2001), Graham (2000), and Byoun (2012) show that cash holdings are negatively related to leverage. Cash holding is defined as the ratio of cash and marketable securities to total assets.4

The theoretical and empirical literature suggests that collateral constraints are an important factor in firms' borrowing decisions (see, Bernanke and Gertler (1989), Whited (1992); Kiyotaki and Moore (1997)). A firm's ability to post collateral determines its access to credit markets, especially bank loan markets. Tangible assets support debt financing to the extent that they serve as collateral (Fama and French, 2002; Frank and Goyal, 2005). Thus, we expect firms with fewer tangible assets are more likely to become debt free than firms with more tangible assets.

Relatively low capital intensity implies high fixed costs of employee compensation and high incentive costs of employees in cases of financial distress (Opler and Titman, 1994; Babenko, 2003). Thus, low-capital-intensive (high-labor-intensive) firms are more constrained than are high-capital-intensive (low-labor-intensive) firms. This is also consistent with MacKay and Phillips (2002) and Williams (1995), who suggest that capital-intensive firms use more leverage than do labor-intensive firms. Thus, firms with high labor intensity are more likely to become debt free. We define capital intensity as fixed assets divided by number of employees adjusted for the industry median based on the two-digit SIC.

Table 4 reports firm characteristics related to borrowing constraints and the proportion of debt-free and debt firms for each of the size quintiles. We drop observations with missing values in any of the reported variables. The data show that debt-free firms are concentrated in smaller size quintiles, with 21.90% being debt free in the smallest size quintile while only 2.56% are debt free in the largest size quintile.

Table 4. Financing constraints faced by debt and debt-free firms
Size CashTangible assetsCapital intensityST debtLT debtN (%)
  1. The data consist of 149 434 firm-year observations (debt firms = 131 267; debt-free firms = 18 167) for the period 1971–2006. Observations with missing values in any of the reported variables are deleted. Size is size quintiles based on total assets. A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. Cash is cash and marketable securities divided by total assets. Tangible assets are property, plant, and equipment divided by total assets. Capital Intensity is fixed assets divided by total number of employees. ST (LT) debt is short-term (long-term) debt divided by total assets. N (%) is the percentage of firms in each group relative to the total number of firms in each size quintile. p-value represents p-values from t-tests for difference in means with unequal variances.

1Debt0.12970.276221.96100.10480.122078.10
Debt-free0.32410.15947.11300.00000.000021.90
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000) 
2Debt0.10600.283936.06340.08510.154884.32
Debt-free0.25650.149023.01240.00000.000015.68
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000) 
3Debt0.08850.300657.51290.06570.183886.99
Debt-free0.22190.171234.82270.00000.000013.01
p-value(0.0000)(0.0000)(0.0100)(0.0000)(0.0000) 
4Debt0.06900.328982.12290.05020.228492.25
Debt-free0.17410.202049.72070.00000.00007.75
p-value(0.0000)(0.0000)(0.0100)(0.0000)(0.0000) 
5Debt0.04760.3850111.66520.04490.242097.44
Debt-free0.12280.249222.74260.00000.00002.56
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000) 
AllDebt0.08640.317764.67860.06850.189587.82
Debt-free0.25700.168423.68480.00000.000012.18
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000) 

Debt-free firms appear to be constrained using all measures. Cash holding as a proportion of total assets is negatively correlated with firm size and is significantly greater for debt-free firms (ranging from 12.28% to 32.41%) than for debt firms (ranging from 4.76% to 12.97%). The pair-wise difference in means within each size quintile is statistically significant. For all size quintiles, tangible assets—as proportions of total assets—are significantly greater for debt firms (between 27.62% and 38.50%) than for debt-free firms (between 14.90% and 24.92%), suggesting that firms with a large portion of assets in intangible forms tend to become debt free. Also, large firms have more tangible assets. The capital-intensity ratio is significantly smaller for debt-free firms than for debt firms. Thus, labor-intensive firms are more closely associated with a debt-free capital structure than are capital-intensive firms.

Following Faulkender and Petersen (2006), we use firms' long-term credit ratings and short-term commercial paper ratings as proxies for accessibility to public debt markets. Faulkender and Petersen (2006) and Lemmon and Zender (2004) find that leverage ratios of firms with credit ratings are significantly higher than are those of firms without ratings. We also examine S&P common stock quality rank as a measure of relative accessibility to equity market.5 According to S&P's description, common stock quality rank measures a stock's “relative standing based on earnings, dividends, growth and stability within long-term trend.” Firms without credit ratings but with good stock ranks are more likely to use equity exclusively and to become debt free.

In Panel A of Table 5, we classify the lowest to highest credit ratings and stock quality ranks with values from 0 to 6. For example, a long-term credit rating of AAA is assigned to 6; AA to 5; A to 4; BBB to 3; BB to 2; B to 1; and below B to 0. Panel B of Table 5 shows long- and short-term credit ratings and common stock quality ranks based on our numerical classification of the ratings/ranks for debt and debt-free firms. We report the number of observations (N) in the column next to each measure. The results show that small debt-free firms rarely have credit ratings (only 110 out of 14 814 observations), suggesting that they have little access to public debt markets. Short-term credit ratings are concentrated on firms in the largest quintile, while long-term credit ratings are concentrated on firms in quintiles 4 and 5. The results of rare credit ratings issued for debt-free firms indicate that these firms have little access to the public debt market. For a few debt-free firms with available ratings, however, their average ratings are not significantly different from those of debt firms. On the other hand, debt-free firms have significantly better stock quality ranks than do debt firms in all size quintiles except for the fifth quintile in which the difference is not statistically significant. Equity markets relative to debt markets are more favorable for debt-free firms. Financing decisions—especially for small firms—appear to be affected by their comparative advantages or the borrowing constraints that they face in the capital market.

Table 5. Credit ratings and common stock quality ranks of debt and debt-free firms
Panel A: Numerical assignments of credit rating and stock quality rank
LT creditST creditStock rankAssigned value
AAAA1A+ and A6
AAA2A−5
AA3B+4
BBBBB3
BBCB−2
BDC1
CCC and belowSuspendedD and liquidation0
  1. The data consist of 97 458 firm-year observations for the period 1985–2006. LT credit and ST credit are the S&P issuer credit rating of an issuer's overall long-term and short-term creditworthiness respectively. Prior to September 1, 1988, LT credit represents the issuer's senior debt rating that has been assigned to the company and ST credit represents the issuer's commercial paper rating that has been assigned to the company. Stock rank is S&P stock quality rank which measures a stock's relative standing based on earnings, dividends, growth, and stability within long-term trend. Each code of credit ratings and stock ranks is assigned to a number as in Panel A. A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. The averages of the assigned numerical values to ratings and ranks are reported in Panel B. N is the number of observations with a rating or rank in each size quintile. p-value represents p-values from t-tests for difference in means with unequal variances.

Panel B: Long-term and short-term credit ratings and common stock quality ranks
Size ST credit N LT creditNStock rank N Total N
1Debt  2.000011.4650283014 534
Debt-free  3.285771.595311494949
p-value    (0.0000)  
2Debt6.000011.0000121.7833580915 554
Debt-free  1.000012.021514913936
p-value    (0.0000)  
3Debt  1.01813312.2980723216 145
Debt-free  1.3000102.486313143347
p-value  (0.1000) (0.0000)  
4Debt5.5893561.459632292.8483943817 550
Debt-free  1.4800253.161410471940
p-value  (0.9100) (0.0000)  
5Debt5.481953722.946212 9183.700013 42518 861
Debt-free5.2727113.0448673.7773440642
p-value(0.1600) (0.5600) (0.3100)  
AllDebt5.483154292.614916 4912.779938 73482 644
Debt-free5.2727112.52731102.4051544114 814
p-value(0.1500) (0.4400) (0.0000)  

4.2. Firm Valuation, Stock Performance and Financing Activities

The market-to-book (MB) ratio has been used as a measure of equity market valuation in previous studies. Baker and Wurgler (2002) argue that MB has a persistent effect on capital structure. Accordingly, we examine MB for debt-free firms relative to that of debt firms over the 5 years prior to the debt-free year. We also report stock performance measured by 1- and 3-year monthly compounded stock returns above the equal-weighted NYSE/AMEX/NASDAQ returns. Additionally, we examine firms' financing activities: (i) net debt issues divided by total assets; (ii) net equity issues divided by total assets; and (iii) the change in market value of equity with split adjustment.6

Table 6 reports the results for debt and debt-free firms.7 The results show that debt-free firms have significantly higher MB than do debt firms throughout the 5 years prior to the debt-free year. One- and three-year stock returns suggest that debt-free firms experience exceptionally good stock performance prior to the debt-free year. Thus, firms' debt-free capital structure may be attributable to exceptional stock performance.8

Table 6. Firm valuation, stock performance, and financing activities for debt and debt-free firms
Years relative to debt-free year −5−4−3−2−10
  1. The data consist of 109 571 firm-year observations (debt firms = 97 308; debt-free firms = 12 263) for the period 1971–2006. A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. MB is the market-to-book asset ratio. One-year stock return is the prior 1-year stock return above the equal weighted 1-year NYSE/AMEX/NASDAQ returns. Three-year stock return is the prior 3-year stock return above the equal weighted 3-year NYSE/AMEX/NASDAQ returns. Net total debt issues are net total debt issues divided by total assets. Net equity issues are net equity issues divided by total assets. ΔMarket equity is changes in market value of equity. p-value represents p-values from t-tests for difference in means with unequal variances.

MBDebt1.64071.64291.65541.67051.70051.7162
Debt-free2.76982.71722.77602.87283.03543.2367
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)
One-year stock returnDebt0.04170.03820.03480.02890.02280.0115
Debt-free0.09430.08130.10060.09640.08690.0749
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)
Three-year stock returnDebt0.13770.12760.11970.10150.08720.0636
Debt-free0.26460.24580.27790.28860.25870.2391
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)
Net total debt issuesDebt0.01440.01420.01470.01610.01750.0196
Debt-free−0.0037−0.0054−0.0073−0.0074−0.0088−0.0150
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)
Net equity issuesDebt0.01840.01970.02120.02310.02490.0269
Debt-free0.04450.04500.04860.05230.05830.0681
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)
ΔMarket equityDebt0.05490.05890.06290.06910.07800.0738
Debt-free0.12560.12680.14320.15420.18790.2408
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)

The net issues of total debt for debt-free firms are negative prior to the debt-free year while those of debt firms are positive. This finding suggests that debt-free firms reduce their debt for several years prior to becoming debt free. The results further show that debt-free firms issue significantly more equity before the debt-free year than do debt firms. The annual changes in the market value of equity for debt-free firms significantly outpace those of debt firms prior to the debt-free year, ranging between 12.56% and 24.08% for debt-free firms versus between 5.46% and 7.80% for debt firms over the 5-year period prior to the debt-free year.

To summarize, debt-free firms experience particularly good stock performance and issue equity while reducing debt over several years prior to becoming debt-free. It appears that debt-free firms rely mainly on external equity capital in order to reduce debt when the market valuation is highly favorable.

4.3. Investment Opportunities, Profitability and Dividends

If firms become debt free by relying solely on internally generated funds, debt-free firms are likely to be more profitable than are debt firms. On the other hand, if firms become debt free in order to reduce the likelihood of having to issue risky securities or to forgo profitable future growth/investment opportunities, then debt-free firms are likely to have greater investment opportunities than are debt firms. Accordingly, we examine firms' profitability and expected growth/investment opportunities. We also examine firms' dividends in order to determine whether debt-free firms are paying high dividends as a substitute for leverage in the presence of large FCFs or as an effort to satisfy shareholders so that they can raise external equity on favorable terms without the adverse effect of the agency problem.

We measure a firm's profitability by operating cash flow (OCF) and FCF divided by total assets. We use MB asset ratio, research and development (R&D) expenses, advertising expenses (AD), and net investment (NI) as measures for growth/investment opportunities. We measure dividend payout by cash dividends divided by total assets.

Table 7 reports the results across size quintiles. For all size quintiles, debt-free firms have significantly higher MB than do debt firms. The difference in MB between debt-free and debt firms is more profound for larger quintiles. The table further shows that debt-free firms' R&D and AD are significantly greater than those of debt firms, whereas their NI is significantly less than that of debt firms. These results suggest that debt-free firms incur lower capital expenditures (tangible forms) but greater R&D and AD (intangible forms) than do debt firms.

Table 7. Profitability, investment opportunities and dividends for debt and debt-free firms
SizeDebtMBR&DADOCFFCFNINI–FCFDividend
  1. The data consist of 149 434 firm-year observations (debt firms = 131 267; debt-free firms = 18 167) for the period 1971–2006. Size is size quintiles based on total assets. A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. MB is the market-to-book asset ratio. R&D is R&D expenses divided by total assets. AD is advertising expenses divided by total assets. OCF is operating cash flow divided by total assets. FCF is free cash flow divided by total assets. NI is net investment divided by total assets. NI–FCF is the difference between NI and FCF. Dividend is cash dividends divided by total assets. p-value represents p-values from t-tests for difference in means with unequal variances.

1Debt2.53750.11990.0494−0.1224−0.09060.07880.16900.0033
Debt-free3.29510.22330.0547−0.3745−0.24800.01820.26490.0236
p-value(0.0000)(0.0000)(0.3800)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)
2Debt1.71320.08610.04400.05720.04440.08950.04530.0060
Debt-free2.50830.13840.05300.01200.01820.05920.04170.0163
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.5300)(0.0000)
3Debt1.59760.06000.04370.10560.08260.10220.01970.0085
Debt-free2.61840.10190.05480.10020.09030.10040.01010.0145
p-value(0.0000)(0.0000)(0.0000)(0.1000)(0.0100)(0.5900)(0.0100)(0.0000)
4Debt1.53660.04050.04170.13280.10270.10590.00360.0112
Debt-free2.68610.06920.05710.16370.13830.0974−0.04150.0216
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0100)(0.0000)(0.0000)
5Debt1.54090.03020.03970.14430.11280.1026−0.01050.0172
Debt-free2.98080.07200.05370.17590.14950.0994−0.05060.0204
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.4900)(0.0000)(0.0200)
AllDebt1.76150.06630.04350.07010.05550.09650.04100.0096
Debt-free2.85700.14740.0546−0.0814−0.04100.05990.10050.0194
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)

Interestingly, profitability measured by OCF and FCF shows an unexpected pattern: cash flows are lower for small debt-free firms than for small debt firms (in quintiles 1 and 2), whereas cash flows are higher for large debt-free firms than for large debt firms (in quintiles 4 and 5). We also observe that cash flow deficit (NI–FCF) is much larger for debt-free firms than for debt firms in the first size quintile, which suggests that small debt-free firms invest more than internally generated funds. In contrast, the cash flow deficits for debt-free firms in size quintiles 4 and 5 are negative and significantly lower than those for debt firms in the same size quintiles, suggesting that large debt-free firms generate more cash flows than their investment needs. These results imply that small debt-free firms become debt-free by solely relying on external equity, whereas large debt-free firms replace debt with internal equity.

The table also shows that across all size quintiles, debt-free firms pay out more cash dividends than do debt firms. Small debt-free firms pay exceptionally large dividends relative to debt firms of similar size. Moreover, the percentage of debt-free firms among dividend-paying firms increases from 18% in the 1970s to 33% in the 2000s. This is puzzling since small firms do not generate large cash flows from their investments.

The results suggest that debt-free firms' high dividend payouts and investment activities breed favorable equity valuations. High dividend payouts by debt-free firms should be well received by shareholders as they address shareholders' concerns for agency problems. High dividend payments by small debt-free firms work to build their reputations for addressing the agency problem of expropriating outside shareholders (La Porta et al., 2000), which allows them to raise external equity on favorable terms. Such efforts are especially credible for small, growing debt-free firms as they become exclusively equity-dependent (Gomes, 2000). Large debt-free firms' high dividend payments, on the other hand, mitigate the need for leverage to reduce the agency costs of FCF. Accordingly, large debt-free firms use surplus cash flow to pay large dividends while reducing existing debt to become debt-free—without deteriorating their equity valuation due to shareholders' concern for agency costs of FCF.

4.4. Financial Activities Around the Debt-Free Year

We examine the history of debt-free firms' equity issuances and dividend payouts in order to understand how debt-free firms replace debt with equity and whether there are changes in payout policies around the debt-free year.

Figures 4 and 5 show the net new equity issues and dividend payouts around the debt-free year, for small and large debt-free and debt firms respectively.9 Net equity issues after the debt-free year for large debt-free firms are negative and significantly lower than are those for large debt firms, suggesting that large debt-free firms repurchase shares after becoming debt-free (Figure 4B). In contrast, small debt-free firms continue to use more external equity than do small debt firms—even after becoming debt-free (Figure 4A). The repurchasing activities of large debt-free firms make it difficult to explain the debt-free capital structure solely in terms of the market timing effect. Figure 5 also shows that regardless of firm size, debt-free firms pay significantly higher dividends prior to and following the debt-free year. Interestingly, they pay more dividends after becoming debt-free. These findings suggest that large, profitable debt-free firms address the agency costs of FCF by repurchasing their shares as well as increasing regular dividends after they become debt-free, whereas small growing debt-free firms address the agency costs of external equity by paying large dividends.

Figure 4.

Net equity issues (as percentage of total assets) around debt-free year (a) small firms (b) large firms

Figure 5.

Cash dividends (as percentage of total assets) around debt-free year (a) small firms (b) large firms

Coupled with evidence that debt-free firms face harsher borrowing constraints than do debt firms, debt-free firms' high dividend payouts appear to reflect efforts to keep the accessibility of the equity market on favorable terms by mitigating the agency costs of FCF, the success of which is illustrated by high stock returns and high market valuations. We turn to logit analysis in the next section in order to examine the relative importance of our proxy variables in explaining the probability of firms becoming debt free.

4.5. Debt-Free versus Low-Leverage Firms

In this section we investigate whether debt-free firms show characteristics significantly different from those of low-debt firms. We also separate firms with zero debt for more than 2 years (persistently debt-free firms) in order to examine if they are different from firms with zero debt for 1 year (transitorily debt-free firms).10 We report the results here without providing tables in order to conserve space.

Most characteristics for persistent debt-free firms are less favorable for debt financing than are those of transitory debt-free firms. Also, dividend payouts are significantly higher for persistent debt-free firms than for transitory debt-free firms. Thus, the high dividend payout is a distinctive characteristic of debt-free firms that rely solely on equity for their capital.

When we compare low-leverage (between 0% and 5% of total assets) firms with debt-free and other debt firms, low-leverage firms' cash holdings, debt financing, tangible-to-total asset ratios, and MB ratios are between those of debt-free and debt firms. However, the dividend payouts of small low-leverage firms are significantly lower than those of debt-free firms and are similar to those of other debt firms, whereas the dividend payouts of large low-leverage firms are very similar to those of debt-free firms and significantly higher than those of other debt firms. Also, there are more credit ratings available for low-leverage firms and their credit ratings are significantly higher than those of both debt-free and other debt firms. We also find that low-leverage firms are largely transitory, consistent with the findings of Minton and Wruck (2001). Thus, low-leverage firms are different from debt-free firms in several respects.

5. Regression Analysis

5.1. Logit Regression

In order to determine relative importance of the factors related to the debt-free capital structure, we estimate the following logit model in which the dependent variable is zero for debt firms and one for debt-free firms:

display math(1)

where X is a vector of independent variables including: size (Size), cash (Cash), tangible assets (TA), capital intensity ratio (CI), OCF, dividends (Div), dividend dummy (Div_D), research and development expenditures (R&D), R&D dummy (R&D_D), AD, market-to-book ratio (MB), 3-year stock return (Ret3), credit rating dummy (CRating), common stock quality rank (SRank), 5-year lease commitment (Lease5), and non-debt tax shields (TShields). Variable definitions are provided in the appendix A.

In order to assess the effects on probability (rather than on the odds ratio) of changing a predictor from one level to another, we estimate the marginal probability corresponding to one standard deviation change around the mean (i.e. change from one-half standard deviation below the mean to one-half standard deviation above the mean) of each explanatory variable except for dummy variables, holding all other variables at their respective means.11

Table 8 presents the estimation results with industry and year effects included. Regressions with various combinations of independent variables yield similar results. Accordingly, we report the estimates of three regressions with and without variables related to credit ratings, stock quality ranks, 5-year lease commitments, and non-debt tax shields.12 The negative intercepts simply suggest that debt-free firms are less common than debt firms. The effects of other variables are consistent with the results in the previous section, as shown by the significant coefficient estimates.

Table 8. Logit regressions for the effects of the firm characteristics on the type of debt policy
 Parameter estimateMarginal probabilityParameter estimateMarginal probabilityParameter estimateMarginal probability
  1. A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. The dependent variable equals 1 for debt-free firms and 0 for debt firms. Independent variables are as follows: Size = logarithm of total assets adjusted by GDP deflator with base year 2000; Cash = cash and marketable securities divided by total assets; TA = tangible assets measured by property, plant and, equipment divided by total assets; CI = capital intensity measured by fixed assets divided by the number of employees; OCF = operating cash flow divided by total assets; Div = common stock cash dividends divided by total assets; Div_D = dummy variable equal to 1 if a firm's dividend is missing, 0 otherwise; R&D = research and development expenditures divided by total assets; R&D_D = dummy variable equal to 1 for missing R&D and 0 otherwise; AD = advertising expenses divided by total assets; MB = market-to-book ratio of assets; Ret3 = prior 3-year stock return above the equal weighted 3-year NYSE/AMEX/NASDAQ returns; CRating = dummy variable equal to 1 if the firm has S&P long-term credit rating, 0 otherwise; SRank = common stock quality rank as scaled in Table V; Lease5 = lease commitments for the next 5 years divided by total assets; and TShield = non-debt tax shields measured by deprecation, amortization, deferred tax, and investment tax credit divided by total assets. Industry and year effects are included. **, *denotes significance at the 1% and 5% levels respectively.

Size−0.3429**−0.0218−0.2697**−0.0124−0.2329**−0.0126
Cash3.7102**0.23574.0052**0.18403.9698**0.2146
TA−2.5996**−0.1651−2.5500**−0.1171−2.6459**−0.1430
CI0.0001**0.00000.0001**0.00000.0002**0.0000
OCF1.5234**0.09681.3321**0.06121.3483**0.0729
Div6.4511**0.40983.2681**0.15012.4687**0.1334
Div_D−0.2150**−0.0138−0.1001*−0.0046−0.1008*−0.0055
R&D1.0426**0.06621.3429**0.06171.5123**0.0817
R&D_D−0.0992**−0.0063−0.0430−0.0020−0.0391−0.0021
AD1.3757**0.08741.7509**0.08041.7642**0.0954
MB0.1417**0.00900.1332**0.00610.1290**0.0070
Ret3−0.0376**−0.0024−0.0475**−0.0022−0.0467**−0.0025
CRating  −2.6939**−0.0871−2.7466**−0.1044
SRank  0.2203**0.01010.2158**0.0117
Lease5    0.13080.0071
Tshield    −0.1139−0.0062
Intercept−1.5041** −2.0463** −1.9948** 
Observations81 530 39 565 31 155 
Pseudo R20.2152 0.2457 0.2425 

The marginal probability is reported beside each parameter estimate. For example, from the first regression, controlling for other firm characteristics, an increase from one-half standard deviation below to one-half standard deviation above the mean cash balance increases the probability of firms becoming debt-free by 23.57%. The results show that the likelihood of being classified as debt free relative to in debt is particularly greater for firms with greater cash balances, fewer tangible assets, and higher dividend payments. OCFs, R&D expenses, and AD are also associated with relatively high marginal probabilities. The long-term credit rating dummy variable (CRating) is also highly significant and negative, which suggests that the existence of a long-term credit rating decreases the probability of a firm becoming debt-free by 9% to 10%. On the other hand, lease commitments and non-debt tax shields do not significantly affect the likelihood of a firm becoming debt free.

An unexpected result is the negative coefficient estimate for 3-year stock returns (Ret3) prior to the debt-free year.13 When we estimate the logit regression without MB and OCF (not reported), the coefficient estimates on stock returns become positive, suggesting that prior stock returns have limited ability in explaining the likelihood of firms becoming debt free. Overall, our logit regressions confirm the results from the univariate analyses.

5.2. Other Specifications and Robustness Checks

In this section, we address potential concerns about model specification and other estimation issues. First, in the logit regressions, we use White's (1980) method to correct for the standard errors for cross-sectional heteroskedasticity; we also use Fama and MacBeth's (1973) approach, in which the coefficients are estimated by the averages from year-by-year cross-section regressions, and the time series standard errors of the estimated coefficients are used to draw inferences. Second, we estimate the regressions with different combinations of variables in order to include the different sets of variables available for different time periods and to mitigate multicollinearity in the variables. Third, we run probit regressions as alternatives to logit regressions. Fourth, we estimate the regressions with lagged variables in order to address the issue of endogeneity. The results are qualitatively similar and we are confident that the results are not driven by any misspecification or peculiarity in the data.

For an additional robustness check, we estimate the regressions with different panels divided into various periods. Separate regressions allow us to examine whether the determinants of debt-free policy change over time. Some firm characteristics that are significant in the pooled cross-sectional logit regressions are not significant across all panels with subperiods. Specifically, the effects of R&D on the likelihood of following debt-free policy are less significant in the pre-1990 period. The effects of tax shields are significant and positive in the pre-1990 period, whereas their effects are weaker and reversed in the post-1990 period.

Lastly, we estimate the regressions for samples divided into small and large firms based on the median of total assets. We find that the effects of the capital intensity ratio on the likelihood of becoming debt free are significant and negative for large firms, whereas they are not significant for small firms. This finding suggests that large firms with high capital intensity are less likely to become debt free. Also, the effect of OCF on the likelihood of becoming debt free is much more significant for large firms than it is for small firms. This is consistent with our univariate results that large firms' debt-free capital structures result from the replacement of debt with internal funds, while small firms' debt-free policies result from heavy reliance on external equity.

In summary, although there are some differences over time and across firm size, our primary findings from logit analysis are robust and consistent with the univariate results in the previous section.

6. Alternative Explanations

In this section, we explore alternative explanations as to why firms become debt free. First, dynamic capital structure models emphasize the importance of considering future financing needs in determining the current capital structure (Goldstein et al., 2001; Hennessy and Whited, 2005). These models suggest that firms may lower leverage in order to prepare for large acquisitions or capital expenditures in the near future. This implies that the debt-free capital structure is transitory. Accordingly, we examine leverage ratios, acquisitions and capital expenditures around the debt-free year. Our results (available on request) show that debt-free firms increase debt on a very small scale following the debt-free year, and they operate with negligible debt even 5 years after the debt-free year (see also Figure 3). This is in contrast to low-leverage firms that keep low leverage transitorily (Minton and Wruck, 2001). Also, we find no evidence that debt-free firms significantly increase their spending on acquisitions and capital investments following the debt-free year.

According to the DeAngelo and Masulis (1980) model, firms with ample non-debt tax shields may become debt free. There is also evidence that United States corporations have been taking a host of other (newer) non-debt tax shelter alternatives (Desai, 2003; Desai and Dharmapala, 2006; Graham and Tucker, 2006). Graham and Tucker (2006) find that firms engaging in tax shelter activities use less debt. Stefanescu (2005) also points out that pension plans have the features of debt in that pension contributions are tax deductible and that failure to make mandatory contributions leads to bankruptcy. She documents that including pension assets and liabilities significantly increases firms' leverage. If debt-free firms are over-burdened with other off-balance-sheet liabilities, then their debt-free financial policy could be justified from the perspective of the trade-off theory. Thus, observing more non-debt tax shields and off-balance-sheet liabilities among debt-free firms is consistent with the argument that these firms become debt-free because they achieve their tax deductions from non-debt sources.

Table 9 shows that non-debt tax shields (depreciation, amortization and tax credit) of debt-free firms are significantly lower than those of debt firms. Surprisingly, though, debt-free firms pay no more taxes than do debt firms. Debt-free firms appear to engage in a variety of transactions that minimize taxes.14 The table also shows that the overall 5-year lease commitments are significantly lower for debt-free firms than they are for debt firms, while the 1-year lease commitments are not significantly different between debt-free and debt firms at the 5% level. The table further shows that overall pension liabilities of debt-free firms are significantly smaller than are those of debt firms.15 If off-balance-sheet liabilities are substituted for debt, debt-free-firms are expected to have more—not less—of these liabilities. Overall, non-debt tax shields and off-balance-sheet liabilities cannot explain why firms are debt free.

Table 9. Non-debt tax shields, pension obligations and lease commitments for debt and debt-free firms
SizeDebtNon-debt tax shieldsTax rateOne-year leaseFive-year leasePension net liabilitiesPension net worthPension expenses
  1. The data consist of 149 434 (76 043) firm-year observations for the period 1971–2006 (1991–2006 for pension information). A firm is defined as a debt-free firm if it has no debt, and a debt firm otherwise. Non-debt tax shields are depreciation, amortization, deferred tax, and investment tax credit divided by total assets. Tax rate is income tax divided by before-tax earnings. One- and five-year leases are lease commitments for 1 and 5 years, respectively, divided by total assets. Pension net liabilities are the sum of the pension projected benefit obligation of overfunded plans and the pension projected benefit obligation of underfunded plans divided by total assets. Pension net worth is the sum of pension plan assets of overfunded plans and pension plan assets of underfunded plans minus pension liabilities divided by total assets. Pension expenses are pension expenses divided by total assets. p-value represents p-values from t-tests for difference in means with unequal variances.

1Debt0.06390.21790.04810.16450.02890.02680.0058
Debt-free0.05140.24710.05740.15130.02340.02220.0139
p-value(0.0000)(0.6800)(0.5300)(0.4100)(0.4800)(0.5400)(0.3500)
2Debt0.06130.30190.03470.12720.07010.06920.0056
Debt-free0.04750.27730.03050.10720.03840.03470.0070
p-value(0.0000)(0.4100)(0.0000)(0.0000)(0.0000)(0.0000)(0.4900)
3Debt0.06330.39130.03170.12000.10680.10090.0058
Debt-free0.05370.36310.03220.12870.09610.11330.0046
p-value(0.0000)(0.4500)(0.6200)(0.1100)(0.5800)(0.6800)(0.0100)
4Debt0.06950.42510.02970.10370.13150.12470.0056
Debt-free0.05600.35400.03600.14140.10070.10270.0055
p-value(0.0000)(0.0500)(0.1000)(0.0000)(0.0000)(0.0400)(0.5900)
5Debt0.08440.41270.01880.07240.15110.15080.0059
Debt-free0.06550.36670.02340.09220.08030.08830.0078
p-value(0.0000)(0.1200)(0.0000)(0.0000)(0.0000)(0.0000)(0.0700)
AllDebt0.06900.35560.03150.11400.11870.11570.0058
Debt-free0.05210.29890.03980.13020.05790.06070.0077
p-value(0.0000)(0.0400)(0.0800)(0.0000)(0.0000)(0.0000)(0.3400)

Next, we examine debt-free firms' governance structures compared to those of debt firms. If managers prefer less than the value-maximizing level of debt (because they personally suffer bankruptcy costs, or have less discretion in more highly levered firms), then they are expected to take actions that reduce debt when their control increases. Jung et al. (1996) find evidence consistent with the presumption that managerial discretion causes some firms to issue equity when they should issue debt. Berger et al. (1997) find that managers prefer to use debt conservatively. In general, managers with greater equity holdings, in terms of common stock ownership or options, are more likely to make capital structure decisions to maximize shareholder wealth (Berger et al., 1997). Accordingly, if a debt-free capital structure is the result of non-value-maximizing managerial discretion, then we expect managers of debt-free firms to hold fewer shares and options than those of debt firms. In order to examine this possibility, we obtain data from the Compustat Execucomp database, which contains demographic and compensation data for all of the Chief Executive Officers (CEOs) of firms in the S&P 500, S&P MidCap 400 and S&P SmallCap 600 since 1992.

For managerial incentives, we examine the percentage of common shares, vested options held by the CEO, and the number of years the CEO has been in office. The CEO's control over internal monitoring mechanisms increases as her tenure lengthens (Berger et al., 1997). Following Malmidier and Tate (2005), we examine whether the CEO also holds the title of Chairman of the Board. In addition, we examine the proportion of executives who are also directors (inside directors), the number of board meetings per year, and the percentage of shares and shares entitled for options held by other executives. These additional measures are intended to assess the CEO's autonomy and the monitoring of the CEO by other high-ranking company executives.

Table 10 shows the results for firms in the Execucomp universe starting from 1992. On average, CEOs and other managers of debt-free firms hold more shares as well as more options of their firms (CEO Shares and Inside Shares) than do those of debt firms. This result contrasts with that of Lewellen (2006), who finds that leverage is positively associated with CEOs' option ownership and negatively associated with CEOs' stock ownership. The CEO tenures of debt-free firms—especially large firms—are typically longer than those of debt firms. The proportion of executives who are also directors (Inside Director) is smaller for debt-free firms than for debt firms. Small debt-free firms tend to have more board meetings than do small debt firms, whereas large debt-free firms tend to have fewer board meetings than large debt firms. The proportion of firms in which CEOs also hold chairmanship of the board tends to be lower for debt-free firms. Overall, the results in Table 10 do not support the hypothesis that extreme debt conservatism results from managerial entrenchment.

Table 10. Governance structures of debt, debt-free and long-term debt-free firms
SizeDebtCEO sharesCEO optionsInside sharesInside optionsCEO tenureInside directorBoard meetingCEO chair
  1. The data consist of 100 082 firm-year observations (debt firms = 89 323; debt-free firms = 10 759) for the period 1992–2006. Debt firms are firms with any level of debt and debt-free firms are firms with no debt. CEO Shares (options) is the proportion of shares (shares entitled to options) owned by the CEO relative to outstanding shares. Inside shares (options) is the proportion of shares (shares entitled to options) owned by executives relative to outstanding shares. CEO Tenure is the number of years the CEO has been in office. Inside Director is the proportion of executives who are directors. Board Meeting is the number of meetings per year. CEO Chair is a dummy variable equal to 1 if the CEO is also the chairperson of the board of directors and 0 otherwise. p-value represents p-values from t-tests for difference in means with unequal variances.

1Debt0.03780.01130.03640.01865.66670.26847.51270.0735
Debt-free0.01590.01850.02510.03187.25000.24738.40340.0466
p-value(0.0500)(0.1300)(0.0100)(0.0000)(0.3000)(0.0500)(0.0000)(0.1000)
2Debt0.05930.01370.07400.02585.41570.28077.69910.0594
Debt-free0.05560.02090.07100.03785.06760.28686.95750.0511
p-value(0.7800)(0.0000)(0.6100)(0.0000)(0.5100)(0.4600)(0.0000)(0.4200)
3Debt0.03710.01250.06970.02586.68200.28326.72400.0636
Debt-free0.04010.01310.06420.03087.15440.28636.82340.0716
p-value(0.4300)(0.4500)(0.2700)(0.0100)(0.2000)(0.3800)(0.2400)(0.1400)
4Debt0.03860.01040.05770.02127.87990.30356.70310.0839
Debt-free0.05690.00870.08480.01929.33410.29636.44970.0884
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0100)(0.0000)(0.3100)
5Debt0.02140.00570.03260.01297.38590.30797.40160.1088
Debt-free0.03310.00800.05400.01569.17510.28026.81290.0917
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0100)
AllDebt0.02840.00770.04330.01657.42350.30407.16360.0972
Debt-free0.04610.01100.07060.02348.26400.28856.70410.0807
p-value(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)(0.0000)

We also examine whether or not debt-free firms are more likely to be delisted from exchanges due to acquisitions. If the managers of debt-free firms make non-value-maximizing decisions, then these firms are more likely to be taken over by other firms. We find that there is no significant difference between debt-free and debt firms in the proportion of firms delisted due to acquisitions.

7. Summary and Conclusion

Our results suggest that borrowing constraints and equity market valuations have important bearings on firms' debt-free capital structure, and hence, on their financing decisions. However, we do not seek to distinguish between whether debt-free firms do not actively tap into other capital markets in favor of one type of capital or if they choose to rely exclusively on one type of capital due to market constraints. A firm might use more external equity after having first established a good reputation about its stock or it might resort to equity financing after discovering that it is unable to issue debt at attractive terms. We find evidence for both perspectives. On the one hand, debt-free firms face significant borrowing constraints. On the other hand, the dividend policies of debt-free firms appear to reflect their efforts to retain the accessibility of equity financing.

Exceptionally high dividend payouts by debt-free firms are particularly intriguing. By paying high dividends, debt-free firms address shareholders' concerns for the agency problem and maintain their ability to raise equity capital on favorable terms without excessive adverse effects. Such efforts are especially credible for growing debt-free firms since these firms are exclusively equity-dependent and have many borrowing constraints. These firms may also use dividends to bid up their stock prices and thus decrease the dilution effect of raising capital (John and Williams, 1985). Also, high dividends for large profitable debt-free firms appear to substitute for leverage as a means of addressing the agency problem of FCF. Thus, an important issue for future study is whether the correlation between market valuation and capital structure stems from managers' attempts to take advantage of overvalued stock, as suggested by the market timing hypothesis, or their efforts to keep the accessibility to external equity on favorable terms.

  1. 1

    Krantz, M., 2002, “Companies with no debt fly high”, USA TODAY, August 21.

  2. 2

    To address the potential problem associated with consecutive debt-free years, we conduct unreported year-by-year analysis and find similar results to the tabulated results in Sections 4–6.

  3. 3

    In each of the years relative to the debt-free year, we include all available firms that survived from the prior years to the debt-free year or from the debt-free year to the post years. We apply the same criteria for debt firms. Accordingly, the sample size varies across years. We also conduct the same analysis for firms that survive for the entire 11-year period, but the results are very similar and hence are not reported.

  4. 4

    Including accounts receivable in addition to cash and marketable securities or using short-term investments instead of marketable securities produces almost identical results. We also examine the current ratio and the quick ratio as broad measures of financial constraint. The results are similar and are therefore not reported.

  5. 5

    We also measure firms' degrees of equity dependence by the K Z score as used by Baker et al. (2003). Based on parameter estimates from Kaplan and Zingales (1997), Baker et al. (2003) construct a K Z score as follows: math formula where C F is cash flow, DI V is cash dividends (item 19 + item 21), C is cash balance (item 1), and LEV is leverage ratio ([item 9 + item 34]/[item 9 + item 34 + item 216]). Higher values of K Z indicate more constraints to equity financing. However, we find no systematic differences in K Z among debt-free and debt firms.

  6. 6

    We follow Fama and French (2005) in using the change in the market value of equity as a measure of equity issuance.

  7. 7

    In each of the years relative to the debt-free year, we include all available firms that survive from the prior years to the debt-free year. We apply the same criteria for debt firms. Accordingly, the sample size varies across years. The number of observations reported is based on year 0. For firms with consecutive debt-free years, we examine 5 years prior to the first debt-free year. We exclude firms that are debt free for their entire sample periods. Whether or not we require firms to have at least 5 years of data prior to the first debt-free year does not alter the result.

  8. 8

    Note that significant positive stock returns above the equal-weighted NYSE/AMEX/NASDAQ returns may reflect survivorship bias—as we require 1- or 3-year returns for the sample firms. However, our focus is not on the abnormal return itself, but rather on the difference in stock returns between debt firms and debt-free firms. Our examination of the number of years covered in the Compustat database indicates that there is no systematic difference between debt and debt-free firms' survival.

  9. 9

    Small firms are those in the first two smallest quintiles each year, while large firms are those in the largest two quintiles each year; the firms in the third quintile are dropped. We also find similar figures when we divide small and large firms based on the median.

  10. 10

    We conduct this analysis with the sample of firms with a minimum of 3 years of observations. We obtain similar results when we separate the sample between 1-year debt-free and more-than-1-year debt-free firms.

  11. 11

    For a given value of X and the parameter estimate, the probability is computed as follows:

    display math(2)
  12. 12

    The sample period is limited to 1985–2006 when rating information is included.

  13. 13

    We find a similar result when we use a 1-year prior return instead of a 3-year prior return.

  14. 14

    Desai (2003) reports that—according to a review of typical tax shelter transactions and the policy issues provided by the United States General Accounting Office from 1996 to 2000—approximately one-third of large United States corporations reported zero tax liability and, by 2000, 53% of large United States corporations (a minimum of either $250 million in assets or $50 million in gross receipts) reported tax liabilities lower than $100 000.

  15. 15

    Pension information is available in Compustat from 1991 on.

Appendix: A: Definitions of variables

The following are variable definitions used in this study. Item numbers in parentheses are as used in the Compustat database.

Accounts receivable = item 2

Advertising expenses (AD= item 45/item 6

Book value of total assets = item 6

Cash (Cash= cash and marketable securities/total assets

Cash and marketable securities = item 162/item 238

Capital intensity ratio (CI) = fixed assets divided by number of employees (item 8/item 29)

Change in the market value of equity with split adjustment = [item 25(t× item 27(t item 25(t  1) × item 27(t  1)] × [item 199(t)/item 27(t) + item 199(t  1)/item 27(t  1)]/2, for given year t

Common stock = item 60

Common stock quality rank (SRank= common stock quality rank (item 282) as scaled in Table V and all non-ranked firm-year observations are set to zero

Credit rating dummy (C Rating= dummy variable equal to 1 for firms with a long-term credit rating and 0 otherwise

Current debt = item 34

Current ratio = item 4/item 5

Deferred tax and investment tax credit = item 35

Deprecation and amortization = item 14

Dividends (Div= common stock cash dividends (item 127)/total assets

Dividend dummy (Div D= dummy variable that equals 1 for firms with missing Div and 0 otherwise

Five-year lease commitment (Lease5= five-year lease commitments (item 95) divided by total assets

Free cash flow (FCF) = we follow Frank and Goyal (2003) and Byoun (2008) in defining FCF: for firms reporting format codes (item 318) 1–3, FCF equals Income Before Extra Items (item 123) + Discontinued Operation (item 124) + Depreciation and Amortization (item 125) + Deferred Taxes (item 126) + Equity in Net Loss (item 106) + Gain/Loss from Property, Plant & Equipment Sales (item 213) + Other Funds from Operations (item 217) + Other Sources of Funds (item 218). For firms reporting format code 7, item 218 is replaced by Exchange Rate Effect (item 314)

Long-term credit rating = item 280

Long-term debt = item 9

Market leverage ratio = [current debt + long-term debt]/market value of total assets

Market-to-book ratio (M B= Market value of total assets/book value of total assets

Market value of total assets = book value of total assets minus book value of equity plus market value of equity (item 6  item 60 + item 199 × item 25)

Net debt issues = item 111  item 114 + item 301

Net equity issues = item 108  item 115

Net investment (NI) = we follow Frank and Goyal (2003) and Byoun (2008) in defining net investment: for firms reporting format codes (item 318) 1–3, NI equals Capital Expenditures (item 128) + Increase in Investments (item 113) + Acquisitions (item 129) + Use of Funds (item 219)  Sale of Property, Plant & Equipment (item 107)  Sale of Investments (item 109). For firms reporting format code 7, NI equals item 128 + item 113 + item 129  item 107  item 109  Short-Term Investments (item 309)  Other Investment (item 310)

Net sales = item 12

Non-debt tax shields (T Shields) = depreciation and amortization (item 14) and deferred tax and investment tax credit (item 35) divided by total assets

Operating cash flow (OCF) = item 13/item 6 operating lease (one-year) = item 96/item 6 operating lease (five-year) = item 95/item 6 pension expenses = item 43

Pension plan assets of overfunded plans = item 28

Pension plan assets of underfunded plans = item 296

Pension projected benefit obligation of overfunded plans = item 286

Pension projected benefit obligation of underfunded plans = item 294

Preferred stock = item 130

Quick ratio = [item 4  item 3]/item 5

Research and development expenditures (R&D) = item 46

R&D dummy (R&D D) = dummy variable that equals 1 for firms with missing R&D and 0 otherwise

Short-term commercial paper rating = item 283

Short-term investments = item 193

Size (Size) = logarithm of total assets adjusted by GDP deflator with 2000 as a base year

Tangible assets (TA) = property, plant and equipment (item 8)/total assets

Tax rate = item 16/[item 16 + item 53 × item 54], 0 if negative

Three-year stock return (Ret3) = prior three-year stock return minus the corresponding equal weighted NYSE/AMEX/NASDAQ returns

Ancillary