Financial factors and firm growth: evidence from financial data on Taiwanese firms

Abstract

We investigate the possible predictability of firm growth in Taiwan using cross-sectional data of financial factors for the years 1997 and 2003 via principal component analysis. Our results reveal that the 18 financial variables (sales growth rate, total assets, total sales, return on assets, return on equity, gross margin, operating cost minus depreciation divided by sales plus other trading income, acid test ratio, debt–equity ratio, time interest earned, average receivables per average daily sales, inventory, average payables per average daily sales, working capital, working capital as a fraction of total assets, long-term liabilities as a fraction of total assets, and sales as a fraction of net worth of the firm) that we employ bunch together into five different financial ratios for the years 1997 and 2003 that are stable between these years. These financial factors are short-term liquidity, return on investment, long-term liquidity, firm size and capital turnover. Regressing these ratio groups (extracted principal components) on firm growth, we find return on investment in the year 1997 was positively and significantly related to firm growth, while long-term solvency was negatively related to firm growth. In addition, smaller firms tended to grow faster. By 2003, larger firms grew faster than smaller ones and short-term liquidity was positively and significantly related to firm growth, while return on investment was no longer a significant determining factor. Our findings suggest that firms that finance internally or do not rely too heavily on indebtedness may end up growing slower during boom periods but they are the ones that survive and outperform after the bust.

Keywords:

• Applied econometrics
• Control and optimization
• Corporate finance
• Empirical finance

Acknowledgements

The authors wish to thank the three anonymous referees who significantly improved the quality of this paper. All errors and omissions remain the responsibility of the authors.

Notes

¶Modigliani and Miller (1958, 1963), Miller and Modigliani (1961), and Miller (1977).

†See Chen et al. (2007a) for a discussion of widely held and closely held firms with special emphasis on Taiwan.

‡Sales growth rates, total assets, net sales, return on assets, return on equity, gross margin, operating cost minus depreciation divided by sales plus other trading income (OI %), acid test, debt equity ratio, time interest earned, average receivables per average daily sales, inventory, average payable per average daily sales, working capital, working capital as a fraction of total assets, long-term liabilities as a fraction of total assets, and sales as a fraction of net worth of the firm.

†The explanatory variables of this study were public information that could be collected and computed from published annual financial statements produced for the Taiwan Stock Exchange Committee (TSEC).

‡Sales growth rates, total assets, net sales, return on assets (ROA), return on equity (ROE), gross margin, OI, acid test, debt equity ratio, time interest earned, average receivables per average daily sales, inventory, average payable per average daily sales, working capital, working capital as a fraction of total assets, long-term liabilities as a fraction of total assets, and sales as a fraction of the net worth of the firm.

§Chen and Shimerda (1981) used principal component analysis in conjunction with more than 100 variables and dozens of financial ratios that were employed in 26 prior studies for bankruptcy prediction and showed that the theory does not provide any basis for selection of these ratios. Likewise, LeClare (2005) noted that the selection of variables is an ad hoc process and the unavailability of economic theory as a means of selecting such variables means that although studies cannot be generalized, their specificity leads to a high predictive power for the task at hand.

†The total combinations of variables are referred to as vectors, components, factors or eigenvectors.

†The amount of variance explained by each rotated component is the sum of the squared loadings shown in the column.