Pecking order or trade-off hypothesis? Evidence on the capital structure of Chinese companies

Abstract

This study tests the pecking order and trade-off hypotheses of corporate financing decisions using a cross-section of the largest Chinese listed companies. The study is built on Allen (1993), Baskin (1989) and Adedeji (1998) to set up three models in which trade-off and pecking order theories give distinctively different predictions: (1) the determinants of leverage; (2) the relationship between leverage and dividends; and (3) the determinants of corporate investment. In model 1, a significant negative correlation is found between leverage and profitability; in model 2 a significant positive correlation between current leverage and past dividends is found. These results broadly support the pecking order hypothesis over trade-off theory. However, model 3 is inconclusive. Overall, the results provide tentative support for the pecking order hypothesis and demonstrate that a conventional model of corporate capital structure can explain the financing behaviour of Chinese companies.

Acknowledgements

This paper is based on and extends Guanqun Tong's MSc Dissertation at Loughborough University (Tong 2003). We thank participants in the 2004 EEFS Conference in Gdansk for their comments on an earlier draft of this paper, and two referees for their useful comments, all of which have improved the paper. Any remaining errors are our own responsibility.

Notes

¹ The origins of trade-off theory pre-date Modigliani and Miller (1958), but modern versions, based on trade-offs among agency costs, were stimulated by the seminal paper of Jensen and Meckling (1976). Graham (1996) estimates the tax advantages of debt. Pecking order theory was first set out by Myers and Majluf (1984).

² The sample is a panel, but because of the volatility in the data, the authors estimate cross-section regressions of year 2000 leverage on the time-average of each of the explanatory variables, with the averaging done over the available data for each company.

³ This finding needs to come with a health warning: Green et al. (2003) have shown that cross-country comparisons of leverage are even more difficult than Rajan and Zingales (1995) originally suggested.

⁴ According to the Ministry of Finance of China, the ‘Random check of the quality of accounting information’ of 2001 showed that 18.8% of 192 randomly chosen listed companies provided dishonest information in assets, 53.6% of them were dishonest in profits and 11.5% of them prepared ‘internal accounts’ which were different from those disclosed to the market. (Ministry of Finance of People's Republic of China, 2003). The names of the dishonest companies were not disclosed, but we would argue that the largest firms would be more reliable than others, based on the government's emphasis on the regulation of key large companies due to their significance to the stability of Chinese capital markets.

⁵ There are insufficient reliable observations to permit estimation of a panel, particularly as China's accounting standards were only unified in 2001 (Ng et al., 2002).

⁶ Watson and Wilson (2002) test trade-off against pecking order theories by distinguishing anticipated and unanticipated changes in firms’ debt ratios. The difficulty with this approach is that the distinction is based on a fixed debt-to-assets ratio, and therefore does not allow for any conditioning of debt targets on the characteristics of individual firms which is an important feature of trade-off theory.

⁷ ‘Financial slack’ is the term used by Myers and Majluf (1984) to refer to available internal funds in a firm.

⁸ This issue is discussed by Chirinko and Singha (2000) in their comment on the pecking order tests of Shyam-Sunder and Myers (1999), and by Allen (1993).

⁹ As some data are needed with a one-year lag, this enables models to be estimated for 2002 and 2003. Accounting statements were extracted from the website of the Shenzhen Securities Information Company, Ltd (2003).

¹⁰ The company names are available from the authors on request. Singh and Hamid (1992) study manufacturing companies only, but more recent studies include all non-financial companies, for example: Prasad et al. (2001b).

¹¹ For a comparison of IFRS and PRC GAAP see Ng et al. (2002).

¹² Effective 2003, cash dividends decided after the balance sheet date but before the authorization of the financial report were no longer included in ‘dividends payable’ in the balance sheet, but in ‘shareholders’ equity’. Data for 2002 used in the 2003 regressions were restated to 2003 standards.

¹³ An anonymous referee is thanked for this point.

¹⁴ Adequate smoothing of the share price would simply create a measure similar to that obtained from book equity.

¹⁵ Adedeji (1998) adopts a broadly similar approach to distinguishing pecking order and trade-off theories in the UK, but estimates a simultaneous model of leverage, dividends and investment using three stage least squares. In principle, this generates more consistent parameter estimates, but only if the model is specified consistently. Since leverage is a stock and dividends a flow they clearly cannot be determined simultaneously in calendar time unless further assumptions are made about firms’ expectations, and Adedeji does not explicitly do this.

¹⁶ Also, we found four firms with values for INVGROW in 2002 of 725.73, 10.01, 155.50 and 53.80, compared to a maximum for the rest of 3.65, and a mean of 1.38. Similarly high values were observed for 2003. Therefore the INVGROW regressions were re-estimated after dropping these four firms.

¹⁷ The proportion of variance explained is obviously lower when the outlier dummies are excluded.