The Mystery of Zero-leverage Firms
--An Empirical Study on Chinese Public Firms
Supervisor: Dr. S.R. Arping
Master Thesis year ‘14/’15
Business Economics: Finance
Author: Binghong Cheng
Administration number: 10824588
Statement of Originality
This document is written by Binghong Cheng, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
The aim of this thesis is to study the zero-leverage puzzle on Chinese public firms. We present that from 1991 to 2007, 3.18% of 13434 firm-year observations have zero debt and 7.12% have book leverage below 5%. We find that zero-leverage firms and low-leverage firms have similar properties compared to levered firms. We treat zero-leverage as extreme low-zero-leverage. Our regression shows that firms which are smaller, older, more profitable, have more tangible assets, hold more cash balances and pay more dividends are more likely to be zero-leverage firms. Besides, we also find that zero-leverage firms have smaller board size and more independent board. Our results are consistent with the financial constraints explanation, but inconsistent with the management entrenchment. Besides, financial flexibility is also likely to play an important role in explaining the zero-leverage puzzle.
Keywords: Zero-leverage, low-leverage, financial constraints, entrenchment, capital
Table of contents
1. Introduction ... 1
2. Literature Review... 4
3. Methodology and hypotheses ... 8
3.1. Univariate analysis ... 8
3.2. Multivariate regression analysis ... 9
4. Data description ... 10
4.1. Data sources ... 10
4.2. Descriptive statistics ... 12
5. Results of regressions ... 14
5.1. Results of logistic regression analyses ... 14
5.2. Additional test ... 18
5.2.1. Financial constraints in a dynamic panel framework ... 18
5.2.2. Financial flexibility and the zero-leverage firms ... 19
1
1. Introduction
The modern thinking of capital structure of the firms start from the Modigliani– Miller theorem. In 1958, Franco Modigliani and Merton H. Miller states that in a perfect market1 the value of a firm is unaffected by its capital structure. However, frictions happen in real world. Taking taxes and debt into account, the capital
structures will influence the value of the firm, because the tax deductibility of interest will increase the firm value.
In the case of market frictions, trade-off theory and pecking order theory are two main theories of capital structures. Kraus and Litzenberger (1973) put forward a static trade-off model, which is a single-period valuation model using corporate taxes, bankruptcy costs and some other variables to estimate the optimal leverage level. The static trade-off theory states that both advantages (the tax benefits of debt) and
disadvantages (the bankruptcy costs and the financial distress costs of debt) exist while financing with debt. The marginal tax benefits will decrease when the debt increases. On the contrary, the marginal bankruptcy costs increase as the debt
increases. Therefore, the firms can maximize their firm values by balance the debt and equity. However, the static trade-off theory can only explain the difference of debt-to-equity ratios among industries but fail to interpret the difference of D/E ratio among the firms in the same industries.
Unlike the static model, the dynamic trade-off model introduces the time variable. Fischer et al. (1989) first developed a dynamic trade-off model in which a firm’s optimal leverage ratio level is determined by transaction costs and it changes with the value of assets over time. In the dynamic trade-off model firms still seek a balance between the tax benefits and the bankruptcy costs. However, because
transaction costs arise when the fluctuation of the capital structure happens, firms will refinance only occasionally. That is to say, firms will not refinance until the benefit of
1 perfect market refers to market with perfect information and no transaction or
2
the refinancing outweighs the cost. This fact indicates that the optimal leverage ratio is not constant at a certain time, but floating within the optimal range. Provided that the leverage ratio stays within the optimal range, the company is not necessary to adjust the capital structure. The size of the optimal leverage range is determined by the variables included in the model. The dynamic trade-off model predicts a positive relationship between the leverage and profitability of the firm. Nevertheless, empirical study shows that the relation between leverage and profitability is negative, which violate the dynamic trade-off theory.
The pecking order theory postulates that asymmetric information will increase the cost of financing, hence firms first prefer to finance using internal capital, then debt, and equity as the last choice.
In addition, agency cost theory hold the view that corporate managers are the agents of the shareholders. Conflict exists between managers and shareholders
because shareholders expect the firm to be governed so as to maximize the firm value and increase the stock price. However, managers has objectives to maximize their personal power and salary by empire building, etc. Since the managers has more information, there exists information asymmetries between managers and
shareholders. The information asymmetry lead to agency costs. The idea behind the agency cost theory is that shareholders can impose restrictions to agents by increasing the firm’s leverage, sequentially decrease the amount of free cash flow.
These capital structure models all expected that firms should use leverage to finance, which will increase the firm value because the interest is tax deductible, while dividends are not. However, a lot of empirical studies criticize these capital structure models. One of the most well-known puzzles in the capital structure study is the typical facts that there are considerable companies all over the world raise less debt than what the capital structure models predict. Gold et al (2001) take into account some possible economic factors that would force the companies to issue less debt instead of the optimal debt level.
Recent studies of capital structure have shed light on the “zero-leverage puzzle”, or “all-equity puzzle”, which refers to the stylized facts that some firms adopt extreme
3
debt conservatism policies and finance without any debt (either short-term or long-term debt) in a whole fiscal year. Many world-renown firms, such as Apple, Google, Texas Instruments, Bed Bath & Beyond and Urban Outfitters, have no debt in their balance sheet. In light of Strebulaev and Yang (2013), on average 10.2% US public firms (Including both the active and inactive firms) had no long-term debt and short-term debt in their balance sheets during the year from 1962 to 2009. Dang, V. A. (2013) do a research on the UK firms from 1980 to 2007 and concluded that on average 12.18% of the companies had zero long-term and short-term debt. Chen Xiang(2013) do the similar research on the Chinese Housing Markets and found that in 2007, there was about 7% of the Real Estate enterprises in China has no long-term and short-term debt in the balance sheet. Until now, the “zero-leverage puzzle” cannot be interpreted by the main capital structure theories. Therefore, it is necessary to acquire a better comprehension of the zero-leverage phenomenon.
Studying the zero-leverage firms is an effective way to better examine the low leverage phenomenon and further study the capital structure, even to construct a new capital structure model to better explain the puzzles. It is often hard for the scholars to reach an agreement to define the low leverage ratio. Moreover, the low leverage can result from issuing the equity instead of the low debt strategy, therefore studying the zero-leverage phenomenon instead of the low-leverage phenomenon will exclude the temporary low-leverage phenomenon caused by the issuance of equity.
The Chinese zero-leverage proportion in public firms is low compared to that of America. In 2000, only 2.99% of Chinese public firms adopt the zero-leverage policy, while the fraction in USA is 14.26%. To better understand the zero-leverage
phenomenon in China, we try to replicate part of the research of Strebulaev and Yang (2013) and Devos et al (2012). We construct the sample with the China Stock Market & Accounting Research (CSMAR). In this paper, we use the logistic regression model to study the differences between Chinese zero-leverage firms and levered firms. We also try to find the mechanisms that cause the zero-leverage phenomenon in China. Besides, we try to find why the zero-leverage puzzle is special compared to low-leverage puzzle.
4
The main contribution of our paper is that it documents empirically the properties of Chinese zero-leverage firms. As far as we know, such analyses have never been done on Chinese firms before. Therefore, our thesis extend the research on the zero-leverage puzzle.
The thesis proceeds as follows: Section 2 presents the literature reviews about the prior papers related to the zero-leverage puzzle. Section 3 shows our methodology we use in our analysis. We also give the regression model in this part. In section 4, we provide the data source and the descriptive statistics. In section 5, we report the results of the logistic regressions. Moreover, we do some additional test to further test our hypothesis. In section 6, we summarize all we get in the paper. Appendices and tables are shown at the end of the thesis.
2. Literature Review
Previous literatures mainly reported two possible theories that help explain the zero-leverage phenomenon. One is underinvestment hypothesis. Myers (1997)
considers that firms with high growth opportunities will avoid debt financing in order to ease the conflict between equity holders and debt holders, hence solving the ‘debt overhang’ problem and producing underinvestment incentives. The other is financial flexibility hypothesis. It posits that when there are market frictions such as adverse selection or transaction costs2, firms tend to take no debt and save cash to stockpile their borrowing capacity for future investment opportunities. Both the financial flexibility hypotheses and the underinvestment hypotheses point out that firms take zero leverage as a strategy in order to alleviate investment distortions (Dang, 2012).
Some prior papers tried to explain the low-leverage phenomenon from the perspective of the traditional capital structure theories. Smith and Watts (1992) and Jung et al (1996) find a negative correlation between the leverage ratio and the market-to-book ratio. Shyam-Sunders and Myers (1999) and Helwege and Liang
5
(1996) present a strong connection between the initiation/escape of debt policy and the need for external financing. Minton and Wruck(2001) state that “low-leverage firms follow a pecking order style financial policy”. They also find that the financial conservatism is a temporary phenomenon: the underleveraged firm stored their debt capacity for future borrowing. Goldstein et al (2001) propose a dynamic capital structure and considered that the real optimal leverage ratio is determined by the equity and debt prices. When the firm will increase the debt in the future, the optimal debt level at present will decrease. Also in support of the pecking order theory, Frank and Goyal (2003) state that smaller and younger firms prefer equity issues when they are in need external financing. Fama and French (2005) come to the similar
conclusions, they find that firms prefer issuing equity to debt are more likely to be small and have high growth rate. More recently, Lemmon and Zender (2010) hold the view that small firms with lower rating level will cost more to issue debt and may be constrained by limited borrowing capacity. DeAngelo and Roll (2012) consider that the booming of low leverage firms in recent years is due to IPO waves of the young growth firms.
Evidences also arose from the trade-off theory. The static trade-off view posits that there is an optimal leverage ratio that can balance the financial distress costs and tax shield benefits when use debt to finance. Based on this theory, firms abandon the debt in order to avoid incurring the financial distress costs and bankruptcy costs. For the firms perform poorly and have trouble repaying the debt, the financial distress and bankruptcy costs are likely to be the most concern. In addition, firms that have low debt tax shields and high non-debt tax shields are not willing to increase the leverage level (Graham, 2000, DeAngelo & Masulis, 1980). In contrast to static trade-off models, dynamic models emphasize that a firm’s optimal capital structural dependent on the transaction costs and the fluctuations in asset values over time. On the basis of dynamic trade-off models, Leary & Roberts (2005) find that firms may not follow their target leverage ratios as stated in the static trade-off theory and keep no debt in their capital structures, even though they should adjust the leverage ratio to the optimal level in the long run.
6
Even though some of the standard capital structure model tried to account for the low/zero leverage phenomenon, they still fail to illustrate why the proportion of the low-leverage firms (or zero-leverage firms) is so large—even some large companies, like Google, Apple etc., choose to take the low/zero leverage policy. Zero-leverage phenomenon is still a puzzle from the point of these capital structure theories. Graham (2000) holds the view that large and lucrative companies with high cash holdings and low expected distress costs could significantly improve their firm value if they made use of the optimal leverage ratio. Korteweg (2010) points out that zero-leverage firms are able to increase their firm value by 5.50% averagely, if they issued debt at the level of their optimal leverage ratios. Autore and Kovacs (2009) argues that firms with abundant internal capital are less willing to use the external financing when the
information asymmetry level is relatively low.
Recent papers make some positive steps forward. Bessler et al (2012) construct the sample using the firm data of the G7 countries and show that even though low-leverage firms are widely distributed in industries that are likely to have high financial distress costs, the adoption of conservative financial policy do not happen in a specific industry. They also report that debt conservatism is an international phenomenon and increasing firms all over the world has adopted it over time. They find that the fraction of the zero-leverage firms in the cross-country observations is only 5% in 1989, but this number increased to approximately 14% by 2010. They conclude that he trend of the zero-leverage phenomenon is partly because of the IPO waves in the sample years. They divide the firms into financially constrained and unconstrained firms and find that the most important reason why firms adopt the zero-leverage strategy is that the increased asset risk constrain their debt-raising capacity. Zero-leverage firms more likely to have small size, low profitability, and risky capital structures. Besides, they have more cash in their balance sheets than their levered counterparts to maintain the financial flexibility. At last, they find that the proportion of zero-leverage firms tend to be higher in a country with the common law system, the capital-market-oriented financial system, the classical tax system, and the high
7
On the other hand, Devos et al (2012) report that zero-leverage firms save cash from their cash flow and tend to lease their assets. They use the logistic regressions to predict the determinant factors of the zero-leverage. Their independent variable include the cash ratio to indicate the internal governance, Tobin’s Q on behalf of the external governance, and Marginal Tax Rates (MTR) referring to the financing constraints. The result shows that zero-leverage firms tend to have higher tax shields and hold more cash than levered counterpart. However, the financial flexibility variable, Tobin’s Q, is not significant. The evidences in these paper support the financial constraints explanation, but reject the hypothesis that firms are voluntarily stockpiling debt capacity. Our paper follow the hypotheses proposed by Devos et al (2012) and try to test the financial constraints hypothesis and entrenchment
hypothesis in Chinese public-traded firms.
In 2013, Dang published a paper studying the zero-leverage firms in UK over a sample period between 1980 and 2007. They find that on average 12.18% of public UK non-financial firms have zero leverage during the sample period. Similar to the study of Devos et al (2012), he also report that zero-leverage firms have lower tangibility, pay higher dividends, hold more cash, and are smaller and younger on average. Dang (2012) divide the zero-leverage firms into two distinct groups according to their financial constraints levels, that is, firms paying dividends and firms not paying dividends. They find that zero-leverage firms pay no dividends perform poorly and are small and young. These two groups adopt zero-leverage policies for different reasons. These firms adopt zero-leverage policy because of the lacking external financing method. On the contrary, zero-leverage firms that pay dividends have larger size and are more profitable. These firms are easier to issue debt to finance, and they use the zero leverage policy as a strategy that mentioned in the underinvestment hypotheses and financial flexibility hypotheses. Furthermore, Dang (2012) take the macroeconomic conditions into account and find that firms are more likely to eschew debt in a macroeconomic downturn.
Strebulaev and Yang (2013) shows that in their sample period from 1962 to 2009, on average 10.2% of the public nonfinancial US firms have zero debt and
8
approximately 22% of firms have book leverage ratio less than 5%. Strebulaev and Yang (2013) also divide the leverage firms into dividend-paying (DP) and zero-dividend (ZD) samples. Besides, they construct the proxy firm-year data according to the calendar year, industry, size and dividend-paying status. They report the
descriptive statistics for DP and ZD firms and their proxies. Strebulaev and Yang (2013) get the similar result as Dang (2012) that zero-leverage dividend-paying firms are more profitable, pay more taxes, and hold more cash than their proxies. Then they extend the regression analysis to study the zero-leverage policy entry and exit. They explain firms which are smaller and have less property tend to become zero-leverage, while firms pay dividends are less likely to adopt the zero-leverage strategy.
Moreover, Strebulaev and Yang (2013) study the chief executive offer (CEO) and governance variables. They report that firms with large CEO ownership and more CEO-friendly boards have higher possibility to exit the zero-leverage policies. Our paper try to replicate part of the results of this paper to study the zero-leverage phenomenon in USA during the sample period from 1991 to 2007, and extend the similar analyses to Chinese public-traded firms.
3. Methodology and hypotheses
3.1. Univariate analysis
In the statistic description part, we will perform the univariate analysis and compare the zero-leverage firms (ZL) with the levered firms (L) using multiple variables. To better understand the deviation between the zero-leverage firms and levered firms, we perform the T-test to determine the whether the difference of variables between zero-leverage firms and levered firms is significant. Besides, we also compare the zero-leverage firms with the almost zero-leverage firms (AZL). The almost zero-leverage firms are low-leverage firms with the book leverage ratio below 5%. The reason why we compare the ZL and AZL is to find the similarity between the ZL and AZL.
9
3.2. Multivariate regression analysis
We will conduct multivariate regression analysis to test the hypothesis. In the multivariate regression, we use the logistic regression model to examine the properties of firms with zero book debt in the balance sheet. We also try to find determinant factors that influence the decision to adopt zero-leverage strategy. The main regression model takes the following form:
Pr(ZL = 1|X) = 1
1 + 𝑒−(𝛼+𝛽𝑋) ,
Where ZL is a binary variable equals to 1 if the firm has zero debt in a given year and equals to 0 otherwise. We run the following regression:
Logit(ZL) = 𝛽0+ 𝛽1𝑆𝑖𝑧𝑒 + 𝛽2𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 + 𝛽3𝐶𝑎𝑠ℎ + 𝛽4𝐴𝑔𝑒 + 𝛽5𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦
+ 𝛽6𝑃𝑎𝑦𝑜𝑢𝑡 𝑟𝑎𝑡𝑖𝑜 + 𝜀
Where size is expressed as the natural logarithm of the total assets of the firm. The Age is the age of the firm (when the firm is first listed in the Compustat or CSMAR, the age is 0). The dependent variable ZL takes the value of one if in a firm-year observation the firm has no long-term and short-term debt. We did not include the research and development data because in our database, this item start to be recorded from 2007.
In addition, we run another regression which include the board information: Logit(ZL) = 𝛽0+ 𝛽1𝑆𝑖𝑧𝑒 + 𝛽2𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 + 𝛽3𝐶𝑎𝑠ℎ + 𝛽4𝐴𝑔𝑒 + 𝛽5𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦
+ 𝛽6𝑃𝑎𝑦𝑜𝑢𝑡 𝑟𝑎𝑡𝑖𝑜 + 𝛽7𝐵𝑜𝑎𝑟𝑑 𝑆𝑖𝑧𝑒
+ 𝛽8𝐼𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑏𝑜𝑎𝑟𝑑 𝑝𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛 + 𝜀
Where the board size refers to the number of the board of directors and the independent board proportion is calculated by:
𝐼𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑏𝑜𝑎𝑟𝑑 𝑝𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛 =𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑑𝑖𝑟𝑒𝑐𝑡𝑜𝑟𝑠 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑑𝑖𝑟𝑒𝑐𝑡𝑜𝑟𝑠
The dependent variable ZL takes the value of one if in a firm-year observation the firm has no long-term and short-term debt.
The first hypothesis in this paper is that firms with more financial constraints are
more likely to take the zero-leverage policy. Devos et al (2012) state in their paper that
10
because of the credit rating, high risk, bad reputation or moral hazard. Second of all, the financial constrained firms are more likely to utilize the bank financing.
Therefore, the hypothesis indicates that the financial constrained firms are more likely to be the small, young. These firms are tend to use the dividends payout to lower the interest expenses. Besides, financial constrained firms lack the liquidity assets which can enhance the short-term solvency.
The second hypothesis in our paper is called entrenchment hypothesis. The
entrenchment hypothesis posits that entrenched managers will use the low/zero leverage to reduce the risk of the firm when the firms are poorly managed. Therefore,
we need the variables to describe the internal governance structure. Boone et al. (2007) find that firms where management layer has great chances to acquire private benefits or firms in which managers have huge power, have larger boards and smaller independent board proportion. Devos et al (2012) also include percentage of shares owned by the directors as an independent variable in their paper. However, this data cannot be found in CSMAR, thereby we ignore this variable in our research.
4. Data description
4.1. Data sources
Following Strebulaev and Yang (2013), we construct the sample of US public-trade firms through the CRSP/Compustat Merged Data Set from the WRDS. This database provides the historical matching of CRSP market and corporate action data with Compustat fundamental data. The sample period is from 1991 to 2007. We choose this sample period because we want to compare the statistics of US zero-leverage firms with that of Chinese. We include both the active and the inactive firms listed in CRSP/Compustat database. And we excluded the firms with a
non-consolidated balance sheet in order to avoid misleading results. Following Strebulaev and Yang (2013), we drop the financial companies [Standard Industrial Classification (SIC) codes from 6000 to 6999], utilities companies [SIC codes from 4900 to 4999],
11
non-US companies [Current ISO Country Code (LOC) not equal to USA]. Nonpublic firms and subsidiaries (stock ownership variable, STKO, equal to one or two) are also excluded. There are 216,174 firm-year observations that meet the conditions. In addition, we adjust all the value with the Consumer Price Index (CPI) downloaded from OECD (base period is 2000). We then drop the firm-year data with total book value of assets (Compustat data item AT) less than $10 million after the adjustment. In this paper, the time variable DATADATE refers to the calendar date of the fiscal year-end. Therefore, we choose the fiscal year-end stock prices (Compustat item prcc_f) to keep consistency with the DATADATE.
We construct the Chinese sample from the China Stock Market & Accounting Research (CSMAR). This database offers data on the China stock markets and the financial statements of China’s listed companies. We choose the sample period from 1991 to 2007 to test our first hypothesis. We start in 1991 because Chinese public-traded firms first appeared in that year. We exclude the financial firms and the utilities. We end in 2007 because the CSMAR database have not recorded the total long-term debt data after 2007. And even though the CSMAR item total debt include all the data from the period between 1991 and 2014, it is unequal to the sum of total long-term debt and short-term debt in current liabilities. According to the references, we prefer the total long-term debt, hence we choose the sample period from 1991 to 2007. In total, there are 13,647 firm-year data satisfy the conditions. As we do to the American data, we adjust all the nominal value with the Consumer Price Index whose benchmark year is 2000. Furthermore, we choose another sample period from 2000 to 2007 to test our second hypothesis. We start in 2000 because the board of directors data is first included in CSMAR in 2000.
Strebulaev and Yang (2013) calculate the book leverage ratio of firm i in year t by:
𝐵𝐿𝑖𝑡 =𝐷𝐿𝑇𝑇𝑖𝑡+ 𝐷𝐿𝐶𝑖𝑡 𝐴𝑇𝑖𝑡 ,
Where DLTT refers to the long-term debt due after 1 year, and DLC refers to the debt in current liabilities.
12
4.2. Descriptive statistics
In the context of the existing capital structure theories, firms should issue some debt and take the advantage of the tax deductibility. However, as reported in many prior papers, zero-leverage phenomenon is very prevalent in many countries.
However, we find that in China, only a small portion of firms do not have either long-term or short-long-term debt in a specific observation year. The descriptive statistics of zero-leverage firms in China and USA from 1991 to 2007 are shown in Table 1. For instance, table 1 show that in 2000, only 2.99% public firms (30 zero-leverage firms in 1003 observations) are zero-leverage in China, while 14.26% American publicly-traded firms (660 zero-leverage firms in 4628 observations) have zero debt in balance sheet at that year. For all the firm-year observations, only 3.18% firm-year are debt-free, while the proportion of zero-leverage firms is 14.7% in total. Strebulaev and Yang (2013) reported the frequency of zero-leverage firms is 10.6% in all the firm-year observation from 1962 to 2009. The fractions data we find is lower than the fraction reported in the Strebulaev and Yang’s (2013) paper. We notice that our sample of public firms is much smaller than theirs (13434 firm-year observations vs. 73673 firm-year observations), which may partially explain the difference of the fraction. Besides, the financial systems in USA and China are quite different. The Chinese financial market is bank-oriented-based while the American market is market-oriented-based. The financial market is more mature in United States. Generally speaking, Chinese public firms face more constraints in Chinese financial markets compared to firms in United States. Therefore, on the surface the descriptive statistics seem to support financial constraints hypothesis.
In table 2, we report a range of statistics means related to the firms’ property for levered firms, zero-leverage firms and almost zero-leverage firms. For each variable, we conduct a t-test to detect whether the difference of these variables between zero-leverage firms and levered firms is significant. In addition, we also conduct the t-test for the equality of means between zero-leverage firms and almost zero-leverage firms.
13
As demonstrated in table 2, the size of zero-leverage firms is slightly smaller than that of levered firms. The average size of levered firms is 21.0, while the average size of zero-leverage firms is 20.5. The p-value show that the difference is very
significant. Besides, the zero-leverage firms tend to be younger than levered firms, the outcome of t-test is significant. While the levered firms have an average profitability of -0.14, zero-leverage firms have an average profitability of 2.10, which means that zero-leverage firms are more profitable than the levered firms. This statistic is significant at a 5% significant level. In addition, the mean of tangibility of zero-leverage firms is slightly smaller than that of levered firms, but the statistic is not significant. Moreover, zero-leverage firms have higher cash-to-assets ratio, higher payout ratio and larger board size on average. All these three statistics are significant. The independent board proportion is not significantly different between zero-leverage firms and levered firms.
According to the pecking order theory, firms may have no debt in their balance sheet because they have difficulties in issuing debt. Our results of the descriptive statistics show that the zero-leverage firms are smaller, younger, hold more cash and pay more dividends. These statistics all support the financial constraints hypothesis. Nevertheless, our statistics also show that zero-leverage firms are more profitable, less tangible, which partially violate the financial constraints explanations. On the other hand, based on the entrenchment hypothesis, zero-leverage firms have larger boards and smaller independent board proportion. However, we find a smaller board size and higher independent board proportion on average for zero-leverage firms, which is exactly opposite point of view compared to the entrenchment hypothesis. From the perspective of the descriptive statistics and T-test results, we reject entrenchment hypothesis. Meanwhile, however, it is noteworthy that the sample of zero-leverage firms is quite small compared to the whole data. We only observe 428 zero-leverage firm-year data in 13434 observations. Therefore, we need a logistic regression to further test our results.
Moreover, compared to the almost zero-leverage firms (AZL), zero-leverage firms are significantly smaller, more tangible, and zero-leverage firms have more cash
14
holdings, higher payout ratio and larger board size than AZL. We also find that means of all variables for AZL except the tangibility are between means of levered firms and zero-leverage firms. Therefore, we can treat the zero-leverage firms as extreme low-leverage firms.
5. Results of regressions
5.1. Results of logistic regression analyses
Table 3 report the result of logistic regression analysis. The first column shows the independent variables we included in the logistic regression. In regression (1), (2) and (3), the dependent variable is a binary variable which equals to 1 if a firm-year observation has zero outstanding debt and 0 otherwise. The star behind the
coefficients indicates the significance level. Three stars means the statistic is
significant at the 1% level, two stars means 5% level, and one stars means 10% level. In regression (1), the pseudo R2 of the regression is 0.1370, and the LR-squared is significant at the 1% level, which mean this regression is significant. The
coefficient of firm size is significantly negative. This result is consistent with the descriptive statistics analyze in table 1. The positive coefficient of firm size means that small firms are more likely to use the Zero-leverage policy. This result is in line with the financial constraints theory. Firms that have lower total assets are always considered to be more fragile when facing financial crisis, showing that the small firms are riskier. Therefore, small firms have to bear higher cost of debt if they want to issue debt. In other words, small firms have more severe financial constraints.
The logistic regression also report that the coefficient of age is significantly positive at the 5% level, which means that older public firms are more likely to be zero-leverage. Our result is not consistence with Strebulaev and Yang (2013) and Devos, Dhillon, Jagannathan and Krishnamurthy (2012). The outcome disagree with the descriptive statistics in table 1, which reports that zero-leverage firms are younger on average. We then run the histogram command in STATA and find that the age
15
density of the zero-leverage firms are higher to the left (as shown in figure 1), which means that many firms concentrate on the low-age group. Take the small zero-leverage sample into account, it is reasonable that the descriptive statistics report an opposite result. Furthermore, Hadlock and Pierce (2010) states that young firms face more financial constraints because they lack the reputation to issue debt. It seems that the logistic regression outcome for the age is not support the financial constraints theory, which indicates that zero-leverage firms tend to be younger. We notice that the coefficient of variable age has the smallest absolute value, which implies that the age has a weak impact on the possibility that a firm adopt the zero-leverage policy. The weak predictive power of age in the regression model can partially explain the deviation from the financial constraints theory.
In addition, the regression shows that the coefficient of profitability is positive, which is consistent with the result reported in the data description part. This
coefficient is significant at the 10% significance level. The positive coefficient means that firms that earn high profit are more likely to be zero-leverage firms. Devos, et al (2012) do not take this variable into account in their test for the financial constraints hypothesis. However, profitability are related to the financial constraints. Firms with higher profitability should have more debt capacity and face less financial constraints. On the other hand, however, firms that have high profitability can be more risky than other firms. For example, the high-technology firms, such as internet companies, take on more risk while acquire higher profit. Their operating may suddenly turn bad if their competitors announce better product. To sum up, the profitability variable is ambiguous in explaining the financial constraints hypothesis, but we find that zero-leverage firms are more profitable than levered counterparts.
The tangibility is reported to be significantly positive at 1% significance level in the logistic regression. This result is in line with the outcome of Strebulaev and Yang (2013) and is correspond to the T-test analysis. The economic meaning to this
coefficient is that firms with higher tangibility are more likely to become the zero-leverage firms. Tangibility is the fraction of tangible assets to total assets. Even though firms can use the tangible assets as the collateral when issue a debt, holding
16
too much tangible assets can be problematic because it will lower the liquidity of the firm. In other words, a firm with high tangibility will be difficult to issue debt because the illiquidity of the assets lower the solvency and increase the default risk of the firm. Therefore, compared to the low tangibility firms, financial constraints are more severe for firms with more tangible assets.
The results of regression also shows that the coefficient of cash ratio is
significantly positive at 1% level. Our result is consistent with that of Strebulaev and Yang (2013) and Devos et al (2012). Meanwhile, this outcome is correspond to the descriptive statistics in Table 1. The positive coefficient variable means that holding more cash in balance sheet increase the likelihood for firms to be zero-leverage. Firms face uncertainty about the future transactions and credit constraints. And firms hold cash and liquid assets in order to increase the flexibility that firms need in their transactions. Devos et al (2012) indicate that constrained firms put more emphasis on the financial flexibility. Firms save cash from cash flows to finance new profitable project when other financial methods are limited. Therefore, our result support the financial constraints hypothesis. Moreover, Devos et al (2012) also point out that entrenched managers will choose to lower the leverage and save more cash from interest payment in case of the shocks to entrenchment. Therefore, according to the entrenchment hypothesis, low leverage (zero leverage) always come out with the high cash holdings. According to the logistic regression model, the positive relationship between cash ratio and zero-leverage firms could support the entrenchment hypothesis to some degrees.
Both Strebulaev and Yang (2013) and Devos et al (2012) reported a higher payout ratio in zero-leverage firms than that in levered firms. Devos et al (2012) do not take the payout ratio into account when test the financial constraints hypothesis. However, we find some evidence that payout ratio is related to the financial
constraints. Jingwen Yu (2012) points out Chinese firms are tend to use the dividends payout to lower the interest expenses. We report a significantly positive coefficient of payout ratio at 1% significance level, which means that firms pay high payout ratio are more likely to be the zero-leverage firms. The result of our regression partially
17
support the financial constraints hypothesis.
We then run the regression (2) and (3) to include the board information in the sample period from 2000 to 2007, the results are shown in table 3. The regression (2) only include the board size and the independence ratio as the independent variables. The regression (3) also include other control variables. We find that the board size is significantly negative in both logistic regressions. This result is consistent with the US firms reported by Devos, Dhillon, Jagannathan and Krishnamurthy (2012). The result means that firms with larger board size are less likely to adopt the zero-leverage policy. The entrenchment hypothesis states that firms where the shocks to
entrenchment happens have larger boards. Therefore, our regression result reject the entrenchment hypothesis.
We also find that the coefficient the independence ratio is positive in both regression, but it is only significant in the second logistic regression at the 1% level. This result of Chinese public firms is not in line with that of US firms as reported in prior papers. Our regression result disagree with the entrenchment hypothesis. To sum up, our regression result show that firms whose board has higher independence
proportion are more likely to use zero leverage strategy.
In regression (4), (5) and (6), the dependent variable is dummy variable equals to 1 when the book leverage is less than 5% and 0 otherwise. In regression (4), we find that compared to other firms (both levered and leverage firms), almost zero-leverage firms are significantly more profitable, they hold more cash and pay more dividends. Regression (5) and (6) indicate that firms with lower board size and higher independent board fraction are more likely to be low-leverage firms. Except for tangibility, all variables report the same result as the leverage firms. Almost zero-leverage firms tend to have lower tangibility while the zero-zero-leverage firms tend to have higher tangibility. Therefore, zero-leverage firms are more illiquidity than the low-leverage firms. The results of the regression are consistent with the results reported in the descriptive statistics. We prove again that the zero-leverage firms can be treated as the extreme low-leverage firms.
18
age, higher profitability and tangibility, hold more cash and pay more dividends. Besides, they have smaller board size and higher independent director proportion in the board. Low-leverage firms show some similar properties to zero-leverage firms. According to our regression results, the zero-leverage/low-leverage phenomenon of Chinese public firms can be explained by financial constraints theory but not the entrenchment theory.
5.2. Additional test
In this part, we will present some additional results to further test the hypotheses. 5.2.1. Financial constraints in a dynamic panel framework
It is possible that the variables describe the characteristics may have a lag effect on the firms’ decision to adopt the zero-leverage policy. First of all, for example, the profitability and tangibility of the firm last year are likely to be the key factor that determine the degree of the financial constraints. Secondly, the operating status in current year may not be reflected to the decision-making immediately. In order to corroborate our previous findings, we use a dynamic logistic regression approach. The model can be expressed as:
Logit(ZL) = 𝛽0+ 𝛽1𝑆𝑖𝑧𝑒 + 𝛽2𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦𝑡−1+ 𝛽3𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑡−1+ 𝛽4𝐴𝑔𝑒 + 𝛽5𝐶𝑎𝑠ℎ + 𝛽6𝑃𝑎𝑦𝑜𝑢𝑡 𝑟𝑎𝑡𝑖𝑜 + 𝜀
Where profitabilityt-1 and tangibilityt-1 indicate the profitability and tangibility in year
t-1. In table 4, we report our regression result of the dynamic logistic regression in column 2 and column 3. In regression (1), we include the profitability and tangibility in current year.
We find that the coefficient profitability is significantly positive in both
regression. And this result is consistent with the result reported in table 5. This means that firms with high profitability in year t-1 are more likely to adopt the zero-leverage policy in year t.
The coefficient of tangibility is still significantly positive in both regression. The result is also in line with that of the static logistic regression reported in last section. The economic meaning to this result is that firms with high tangibility in year t-1 are
19
more likely to be zero-leverage firms in year t.
In conclusion, our results of the dynamic regression model corroborate our previous findings in section 5.1. The outcome of the dynamic regression still support the financial constraints hypothesis.
5.2.2. Financial flexibility and the zero-leverage firms
Strebulaev and Yang (2013) propose that firms adopt the zero-leverage policy to spare the financial flexibility for future investment. Based on the model raised by Titman et al. (2004), they include the abnormal capital expenditures in their regression. The definition of abnormal expenditure is:
𝐶𝐼𝑡 = 𝐶𝐸𝑡
(𝐶𝐸𝑡−1+ 𝐶𝐸𝑡−2+ 𝐶𝐸𝑡−3)/3− 1
where CE is the ratio of capital expenditure to total assets. Titman et al. (2004) report a negative relation between abnormal capital investments and future stock returns. Therefore, lower abnormal capital expenditures in current balance sheet forecast a higher stock return in the future. We include the CI in our regression and present the result in regression (3), table 4. The significantly negative coefficient of CI
demonstrate that firms with low abnormal capital investment are more likely to be zero-leverage firms. In other words, zero-leverage firms are more likely to acquire higher stock returns in the future. To sum up, financial flexibility may play an important role in the zero-leverage decision.
6. Conclusions
In this thesis, we document the zero-leverage puzzle of Chinese public firms. We collected the data of Chinese public firms from China Stock Market & Accounting Research (CSMAR) over the period 1991-2007.
We find that 3.18% of 13434 firm-year observations have zero debt and 7.12% have book leverage below 5%. The fraction of low/zero-leverage firms is low compared to that of USA. We find that zero-leverage firms and low-leverage firms
20
have similar properties compared to levered firms. We treat zero-leverage as extreme low-leverage. Our regression shows that firms which are smaller, older, more
profitable, have more tangible assets, hold more cash balances and pay more dividends are more likely to be zero-leverage firms. We find that the entrenchment hypothesis, which posits that firms take zero-leverage policy because of poor management, cannot explain the Chinese zero-leverage phenomenon. We conclude that the firms with zero book leverage ratio are more likely to be influenced by the financial constraints. Our additional logistic regression test in a dynamic framework also support the financial constraints hypothesis. Moreover, we add the abnormal capital investments in our regression model and find that firms may also save the debt capacity for future investments.
One of the most important limitations of our paper is that the Chinese database CSMAR is not as sophisticated as the WRDS used in the prior studies. Some data is omitted because the accounting system in China is not mature at the beginning of the sample period. And the sample size is small because of the data limitation. Another limitation is that there are few zero-leverage firms in China. This may makes the results of logistic regression biased.
Further studies of zero-leverage phenomenon can shed light on the financial flexibility. For example, given more time, we would like to do the Event Study to explore the zero-leverage puzzles and examine the zero-leverage policy persistence of Chinese firms. More studies in Chinese zero-leverage phenomenon will be helpful and add to the field with the corporate finance.
21
References
Autore, D., and T. Kovacs, 2009, Equity Issues and Temporal Variation in Information Asymmetry, Journal of Banking and Finance ,34, 12-23.
Bessler, W., Drobetz, W., Haller, R., & Meier, I. (2013). The international zero-leverage phenomenon. Journal of Corporate Finance, 23, 196-221.
Boone, A.L., Field, L.C., Karpoff, J.M., Raheja, C.G., 2007. The determinants of corporate board size and composition: an empirical analysis. J. Financ. Econ. 85, 66– 101.
Berger, P.G., Ofek, E., Yermack, D.L., 1997. Managerial entrenchment and capital structure decisions. J. Finance 52, 1411–1438.
Coles, J.L., Daniel, N.D., Naveen, L., 2006. Managerial incentives and risk-taking. J. Financ. Econ. 79, 431–468.
Coles, J.L., Daniel, N.D., Naveen, L., 2008. Boards: does one size fit all? J.
Financ. Econ. 87, 329–356.
Chen Xiang (2013). The Phenomenon and the influencing factors of zero-leverage firms in the Real Estate Industry (Master's thesis, Shanxi University)
Dang, V. A. (2013). An empirical analysis of zero-leverage firms: New evidence from the UK. International Review of Financial Analysis, 30, 189-202.
Frank, M. Goyal, V. 2003. Testing the Pecking Order Theory of Capital Structure. Journal of Financial Economics, vol. 67 (2), pp. 217-248.
Fama, E. French, K. 2002. Testing Trade-off and Pecking Order Predictions About Dividends and Debt. The Review of Financial Studies, vol. 15 (1), pp. 1-33.
Fama, E. French, K. 2005. Financing Decisions: Who Issues Stock? Journal of
Financial Economics, vol. 76 (3), pp. 549-582
Goldstein, R.,Ju,N.,Leland,H.,2001.An EBIT-based model of dynamic capital structure. Journal of Business, 74, 483–512.
Graham, J. R. (2000). How big are the tax benefits of debt, Journal of Finance, 55(5), 1901–1941.
Hadlock, C.J., Pierce, J.R., 2010. New evidence on measuring financial constraints: moving beyond the KZ index. Rev. Financ. Stud. 23, 1909–1940.
22
Helwege, J. Liang, N. 1996. Is There a Pecking Order? Evidence From a Panel of IPO Firms. Journal of Financial Economics, vol. 40 (3), pp. 429-458.
Jung, Kooyul, Yong-Cheol Kim, and Rene Stulz, 1996, Timing, investment opportunities, managerial discretion, and the security issuance decision, Journal of
Financial Economics, 42, 159-185.
Korteweg, A., 2010. The net benefits to leverage. J. Finance 65, 2137–2170. Linck, J.S., Netter, J.M., Yang, T., 2008. The determinants of board structure. J.
Financ. Econ. 87, 308–328.
Leary, M. T., & Roberts, M. R. (2005). Do firms rebalance their capital structures. Journal of Finance, 60(6), 2575–2619.
Lemmon, M. L. Zender, J. F. 2010. Debt Capacity and Tests of Capital Structure Theories. Journal of Financial and Quantitative Analysis, vol. 45 (5), pp. 1161-1187.
Minton, B. A., & Wruck, K. H. (2002). Financial conservatism: Evidence on capital structure from low leverage firms.
Smith, Clifford, and Ross Watts, 1992, The investment opportunity set and corporate financing, dividend and compensation policies, Journal of Financial
Economics 32, 263-292.
Shyam-Sunder, L. Myers, S. 1999. Testing Static Trade-off Against Pecking Order Models of Capital Structure. Journal of Financial Economics, vol. 51 (2), pp. 219-244.
Strebulaev, I. A., & Yang, B. (2013). The mystery of zero-leverage firms. Journal
23
Figure 1 age density distribution
This figure shows the density distribution of the firm age. Age is the numbers of years since the firm first appear in the CSMAR. On the left of this figure is the density distribution of the levered firms (ZL=0). On the right of this figure is the density distribution of zero-leverage firms (ZL=1). ZL is dummy variable equals to 1 when firms have zero debt in a firm-year and 0 otherwise.
24
Table 1 Fraction of zero/low-leverage firms
The table shows the fraction of zero-leverage and low-leverage for Chinese
companies. ZL Firms adopt the zero-leverage policy and have zero book debt. AZL firms have book leverage below 5%. We also report the ZL fraction of American public firms.
year N ZL% CN AZL% CN ZL% USA AZL%
USA 1991 13 15.38 0 9.61 21.1 1992 70 8.57 2.86 10.38 24.2 1993 201 4.97 8.46 11.6 25.6 1994 313 2.24 6.71 11.72 25.8 1995 333 0.9 5.71 11.69 26.2 1996 518 1.74 6.18 13.06 29.5 1997 698 2.87 6.30 13.19 29.6 1998 793 3.53 6.81 12.98 27.3 1999 879 3.07 8.42 13.13 28.7 2000 1003 2.99 6.58 14.26 31.7 2001 1075 3.25 6.60 15.38 31.2 2002 1126 2.57 7.99 16.37 31.1 2003 1181 3.04 6.27 18.66 33.0 2004 1263 2.92 6.49 19.89 34.9 2005 1262 3.25 7.37 20.59 35.6 2006 1322 4.39 6.96 19.83 34.7 2007 1384 3.61 8.96 19.71 35.5 Total 13434 3.18 7.11 14.7 32.4
25
Table 2 Descriptive statistics for ZL and AZL firms
This table shows the descriptive statistics for levered firms (L), almost leverage firms (AZL) and leverage firms (ZL). Almost zero-leverage firms are firms with book zero-leverage below 5%. Levered firms are firms with book zero-leverage above 5%. All variables are defined in Appendix 1. Column (1), (2) and (3) present the mean of the variables for different groups in the time period from 1991 to 2007. The last two columns show the T-statistics for each variables. The T-statistics marked with ***, ** and * are significant at the 1%, 5% and 10% level, respectively. All the variables are explained in Appendix.
Variable L (1) AZL (2) ZL (3) T-test
(1) Vs. (3) T-test (2) vs. (3) Size 21.01453 20.9969 20.52681 9.8190*** -7.4966*** Age 5.076888 4.735079 4.509346 3.0633*** -0.953 Profitability -0.1394214 0.0718471 2.096871 -2.2851** 1.522 Tangibility 0.2790656 0.2472168 0.2680959 1.2969 2.2244* Cash ratio 0.1378391 0.1998368 0.2622452 -22.7208*** 7.0033*** Payout ratio 0.0087293 0.0133815 0.0166154 -8.9822*** 2.1262** Board size 9.605521 9.524476 9.150943 3.1740*** -2.2114*
26
Table 3 Determinants for ZL and AZL firms
The table presents the results of the logistic regressions. The dependent variable in column (1), (2) and (3) is the dummy variable ZL which equals to 1 when the firm is zero-leverage and 0 otherwise. The dependent variable in column (4), (5) and (6) is the dummy variable AZL which equals to 1 when the book leverage of the firm is below 5% and 0 otherwise. Coefficients marked with ***, **, and * are significant at 1%, 5% and 10% level, respectively. All variables are explained in Appendix.
ZL AZL Variables (1) (2) (3) (4) (5) (6) Size -0.521*** -0.520*** 0.0286 0.0850* Profitability 0.037* 0.0363 0.115* 0.0803 Cash ratio 0.057* 7.396*** 3.527*** 3.500*** Age 2.047*** 0.0207 0.0996 0.008 Tangibility 6.934*** 2.200*** -0.479** -0.924*** Payout ratio 14.758*** 16.510 *** 10.683*** 15.935*** Boardsize -0.0926*** -.0429** -.0136 -0.0087
Ind. board ratio 0.473 1.440*** 0.0400 0.731**
27
Table 4 Results of additional logistic regression
The table shows the additional logistic regression results. The independent variables in the following regressions are all dummy variables which equals to 1 when the book leverage ratio is zero and 0 otherwise. Profitabilityt-1 and Tangibilityt-1 are the
Coefficients marked with ***, **, and * are significant at 1%, 5% and 10% level, respectively. All variables are explained in Appendix
Variables ZL (1) (2) (3) Size -06473007*** -0.6390837*** -0.447*** Age 0.0496042*** 0.0469429*** 0.0278 Profitability 2.151993*** 10.022*** Profitabilityt-1 4.229588*** 5.309264*** Tangibility 1.280487* 2.579*** Tangibilityt-1 1.069127* 2.177108*** Cash ratio 7.118211*** 7.049552*** Payout ratio 9.997503*** 11.1266*** 10.444 CI -0.00205* Constant 7.32462*** 7.276445*** 2.782
28
Appendix Definition of Variables This table present the definition of the variables used in this paper.
Variable Description Definition
CI Abnormal capital expenditure, CI 𝐶𝐼
𝑡 =(𝐶𝐸 𝐶𝐸𝑡
𝑡−1+𝐶𝐸𝑡−2+𝐶𝐸𝑡−3)/3− 1
Age Numbers of years since the firm first appear in the CSMAR (Age=0 for first record)
AT Total assets
BL Book Leverage (DLTT+DLC)/AT
CE Ratio of capex to total assets, CE Capital Expenditure/AT
Cash ratio Ratio of cash holdings to total assets Cash holdings/AT
CPI Annual Consumer Price
Index from CSMAR
2000 is the base year
Size Natural logarithm of total
assets adjusted to 2000
Log(AT*CPI2000/CPIt)
Profitability Ratio of EBIT to
total assets
EBIT/AT
Tangibility Ratio of fixed assets to
total assets
PPENT/AT
Board size Numbers of directors
in the board
Ind. Board ratio Ratio of independent directors to board size
Payout ratio Ratio of dividends payable
to total assets
