Abstract
In this chapter, we discuss two alternative methods of panel data analysis. These two methods include both the fixed effects and random effects models. In addition, we discuss the dummy variable technique and the error component model. Finally, we discuss how these methods can be used to investigate alternative dividend policy hypotheses.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
The firm effect refers to the effect of factors affecting the behavior of an individual firm; it is constant over time. The time effect refers to the economic condition of particular time point; it varies over time.
- 2.
For a discussion of the existence of unobservable effects, see Friend and Puckett (1964).
- 3.
For a discussion of this sort, see, for example, Balestra and Nerlove (1968).
- 4.
Sample lists of these firms are available from the authors.
- 5.
Zarembka (1968) has employed the generalized functional form technique to determine the true functional form for money demand. The proof of this statement can also be found in his paper.
- 6.
\(\hat{\sigma }_{\tau } \left( \lambda \right)\) is obtained either from OLS, LSDV, or GLS.
- 7.
The dummy variable \(D_{i,t} \left( {g_{i,t} < c \cdot ROA_{i,t} } \right)\) used in Eq. (6.20) implies that the relationship between the payout ratio and risks is nonlinear (piecewise regression). In other words, the breakpoint of the structural change is at \(g_{i,t} = c \cdot ROA_{i,t}\). Based upon our theoretical model, we assume that c is equal to 1 in our empirical work.
- 8.
Besides merely adding an interaction dummy as indicated in Eq. (6.20), we include an intercept dummy to take care of the individual effect of two groups. We also run regressions for high-growth firms and low-growth firms separately. Results from both models are qualitatively the same as those from Eq. (6.20) and also support Hypotheses 1–3.
- 9.
For the discussion of hypotheses and data, please refer Appendix.
- 10.
Because our sample is an unbalanced panel data, the clustering computer program cannot meaningfully estimate the variance components, variance of firm \(\left( {\hat{V}_{\text{firm}} } \right)\), variance of time \(\left( {\hat{V}_{\text{time}} } \right)\), and heteroscedasticity robust OLS variance \(\left( {\hat{V}_{\text{white}} } \right)\).
- 11.
Gujarati (2009) shows that this kind of problem can be regarded as piecewise regression by using moving estimate processes.
- 12.
We filter out those financial institutions and utility firms based on the historical Standard Industrial Code (SIC) available from Compustat. When a firm’s historical SIC is unavailable for a particular year, the next available historical SIC is applied instead. When a firm’s historical SIC is unavailable for a particular year and all the years after, we use the current SIC from Compustat as a substitute.
- 13.
To avoid creating a large difference in dividend policy, on the one hand managers partially adjust firms’ payout by several years to reduce the sudden impacts of the changes in dividend policy. On the other hand, they use not only one-year firm conditions but also multiyear firm conditions to decide how much they will pay out. In examining the optimal payout policy, we use the five-year rolling averages for all variables.
Bibliography
Aivazian, V., Booth, L., & Cleary, S. (2003). Do emerging market firms follow different dividend policies from US firms? Journal of Financial Research, 26(3), 371–387.
Amemiya, T. (1972). The estimation of the variances in a variance-components model. International Economic Review, 12, 1–13.
Balestra, P., & Nerlove, M. (1968). Pooling cross-section and time-series data in the estimation of a dynamic model: The demand for natural gas. Econometrica, 34, 585–612.
Box, G. E. P., & Cox, D. R. (1964). An analysis of transformations. Journal of the Royal Statistical Society: Series B, 26, 211–243.
Boehmer, E., Jones, C. M., & Zhang, X. (2013). Shackling short sellers: The 2008 shorting ban. Review of Financial Studies, 26, 1363–1400.
Bower, R. S., & Bower, D. H. (1969). Risk and valuation of common stock. The Journal of Political Economy, 77, 349–362.
Blau, B. M., & Fuller, K. P. (2008). Flexibility and dividends. Journal of Corporate Finance, 14, 133–152.
Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2011). Robust inference with multiway clustering. Journal of Business & Economic Statistics, 29, 238–249.
Chang, H. S., & Cheng-Few Lee. (September 1977). Using pooled time-series and cross section data to test the firm and time effects in financial analysis. Journal of Financial and Quantitative Analysis.
Chow, G. C. (1960). Tests of equality between sets of coefficients in two linear regressions. Econometrica, 28, 591–605.
Chung, P. S. (1974). An investigation of the firm effects influence in the analysis of earnings to price ratios of industrial common stocks. The Journal of Financial and Quantitative Analysis, 9, 1009–1029.
Durand, D. (1959). The cost of capital, corporation finance, and the theory of investment: Comment. American Economic Review, 49, 639–654.
Fama, E. F., & French, K. R. (2001). Disappearing dividends: Changing firm characteristics or lower propensity to pay? Journal of Financial Economics, 60, 3–43.
Friend, I., & Puckett, M. (1964). Dividends and stock prices. American Economic Review, 54, 656–681.
Gordon, J. J. (1959). Dividends, earnings and stock prices. Review of Economic and Statistics, 41, 99–105.
Grullon, G., Michaely, R., & Swaminathan, B. (2002). Are dividend changes a sign of firm maturity? Journal of Business, 75, 387–424.
Gujarati, D. N. (2009). Basic econometrics. Tata McGraw-Hill Education.
Hansen, B. E. (1996). Inference when a nuisance parameter is not identified under the null hypothesis. Econometrica, 64, 413–430.
Hansen, B. E. (1999). Threshold effects in nondynamic panels: Estimation, testing, and inference. Journal of Econometrics, 93, 345–368.
Hansen, B. E. (2000). Sample splitting and threshold estimation. Econometrica, 68, 575–603.
Hausman, J. A. (1978). Specification tests in econometrics. Econometrica, 46, 1251–1276.
Higgins, R. C. (1977). How much growth can a firm afford? Financial Management, 6, 7–16.
Hsiao, C. (2014). Analysis of panel data—Econometric society monographs (3rd ed.).
Jagannathan, M., Stephens, C. P., & Weisbach, M. S. (2000). Financial flexibility and the choice between dividends and stock repurchases. Journal of financial Economics, 57(3), 355–384.
Lee, C. F., Gupta, M. C., Chen, H. Y. & Lee, A. C. (June 2011) Optimal payout ratio under uncertainty and the flexibility hypothesis: Theory and empirical evidence. Journal of Corporate Finance, 17(3), 483–501.
Maddala, G. S. (1971). The use of variance components models in pooling cross-section and time-series data. Econometrica, 39, 341–358.
Nerlove, M. (1971). Further evidence on the estimation of dynamic economic relations from a time-series of cross-section. Econometrica, 39, 359–382.
Osterrieder, D., Palia, D., & Wu, G. (2018). Evaluating panel regression models in corporate finance: Evidence from CEO pay. Working paper.
Patrick, R. H. (2018). Durbin-Wu-Hausman specification tests. In C. F. Lee & J. Lee (Eds.) Handbook of financial econometerics, mathematics, statistics, and technology. World Scientific (Forthcoming).
Petersen, M. A. (2009). Estimating standard errors in finance panel data sets: Comparing approaches. Review of Financial Studies, 22, 435–480.
Rozeff, M. S. (1982). Growth, beta and agency costs as determinants of dividend payout ratios. Journal of Financial Research, 5, 249–259.
Thompson, S. B. (2011). Simple formulas for standard errors that cluster by both firm and time. Journal of Financial Economics, 99, 1–10.
Wallace, T. D., & Hussain, A. (1969). The use of error components models in combining cross-section with time-series data. Econometrica, 37, 55–72.
Zarembka, P. (1968). Functional form in the demand for money. Journal of American Statistical Association, 63, 502–511.
Zeileis, A., Leisc, F., Hornik, K., & Kleiber, C. (2002). strucchange: An R package for testing for structural change in linear regression models. Journal of Statistical Software, 7, 1–38.
Author information
Authors and Affiliations
Corresponding author
Appendix: Optimal Payout Ratio Under Uncertainty and the Flexibility Hypothesis: Theory and Empirical Evidence
Appendix: Optimal Payout Ratio Under Uncertainty and the Flexibility Hypothesis: Theory and Empirical Evidence
By Cheng-Few Lee, Manak C. Gupta, Hong-Yi Chen and Alice C. Lee
1.1 Hypothesis Development
A growing body of literature focuses on the determinants of optimal dividend payout policy. Rozeff (1982), Jagannathan et al. (2000), Grullon et al. (2002), Aivazian et al. (2003), Blau and Fuller (2008), and others empirically investigate the determination of dividend policy, but none of them has a solid theoretical model to support their findings. Based upon the theoretical model and its implications Lee et al. (2011), three testable hypotheses are developed as follows.
-
Hypothesis 1: Firms generally reduce their dividend payouts when their growth rates increase.
The negative relationship between the payout ratio and the growth ratio in our theoretical model implies that high-growth firms need to reduce the payout ratio and retain more earnings to build up “precautionary reserves,” but low-growth firms are likely to be more mature and already build up their reserves for flexibility considerations. Rozeff (1982), Fama and French (2001), Blau and Fuller (2008), and others argue that high-growth firms will have higher investment opportunities and tend to pay out less in dividends. Based upon flexibility concerns, we predict that high-growth firms pay higher dividends. This result is obtained when risk factor is not explicitly considered.
Lee et al. (2011) theoretically find that the relationship between the payout ratio and the risk can be either negative or positive, depending upon whether the growth rate is higher or lower than the rate of return on total assets. Based upon this finding, Lee et al. (2011) develop two other hypotheses.
-
Hypothesis 2: The relationship between the firms’ dividend payouts and their risks is negative when their growth rates are higher than their rates of return on asset.
High-growth firms need to reduce the payout ratio and retain more earnings to build up “precautionary reserves,” which become more important for a firm with volatile earnings over time. For flexibility considerations, high-growth firms tend to retain more earnings when they face higher risk. This theoretical result is consistent with the flexibility hypothesis.
-
Hypothesis 3: The relationship between the firms’ dividend payouts and their risks is positive when their growth rates are lower than their rates of return on asset.
Low-growth firms are likely to be more mature and have most likely already built such reserves over time, and they probably do not need more earnings to maintain their low-growth perspective and can afford to increase the payout (see Grullon et al. 2002). Because the higher risk may involve higher cost of capital and make the free cash flow problem worse, for free cash flow considerations, low-growth firms tend to pay more dividends when they face higher risk. This theoretical result is consistent with the free cash flow hypothesis.
1.2 Sample Description
We collect the firm information, including total asset, sales, net income, and dividends payout, from Compustat. Stock price, stock returns, share codes, and exchange codes are retrieved from the Center for Research in Security Prices (CRSP) files. The sample period is from 1969 to 2009. Only common stocks (SHRCD = 10, 11) and firms listed on NYSE, AMEX, or NASDAQ (EXCE = 1, 2, 3, 31, 32, 33) are included in our sample. We exclude utility services (SICH = 4900–4999) and financial institutions (SICH = 6000–6999).Footnote 12 The sample includes those firm-years with at least five years of data available to compute average payout ratios, growth rate, return on assets, beta, total risk, size, and book-to-market ratios. The payout ratio is measured as the ratio of the dividend payout to the net income. The growth rate is the sustainable growth rate proposed by Higgins (1977). The beta coefficient and total risk are estimated by the market model over the previous 60 months. For the purpose of estimating their betas, firm-years in our sample should have at least 60 consecutive previous monthly returns. To examine the optimal payout policy, only firm-years with five consecutive dividend payouts are included in our sample.Footnote 13 Considering the fact that firm-years with no dividend payout one year before (or after) might not start (or stop) their dividend payouts in the first (fourth) quarter of the year, we exclude from our sample firm-years with no dividend payouts one year before or after to ensure the dividend payout policy reflects the firm’s full-year condition.
Table 6.7 shows the summary statistics for 2645 sample firms during the period from 1969 to 2009. Panel A of Table 6.7 lists the number of firm-year observations for all sample high-growth firms and low-growth firms, respectively. High-growth firm-years are those firm-years that have five-year average sustainable growth rates higher than their five-year average rate of return on assets. Low-growth firm-years are those firms with five-year average sustainable growth lower than their five-year average rate of return on assets. The sample size increases from 345 firms in 1969 to 1203 firms in 1982, while declining to 610 firms by 2009. A total of 28,333 dividend paying firm-years are included in the sample. When classifying high-growth firms and low-growth firms relative to their return on assets, the proportion of high-growth firms increases over time. The proportion of firm-years with a growth rate higher than return on assets increases from less than 50% during the late 1960s and early 1970s to 80% in 2008. Panel B of Table 6.7 shows the five-year moving averages of mean, median, and standard deviation values for the measures of payout ratio, growth rate, rate of return, beta coefficient, total risk, market capitalization, and market-to-book ratio across all firm-years in the sample. Among high-growth firms, the average growth rate is 12.33%, and the average payout ratio is 31.80%, but for low-growth firms, the average growth rate is 4.13%, and the average payout ratio is 57.62%. High-growth firms undertake more beta risk and total risk, indicating that high-growth firms undertake both more systematic risk and unsystematic risk to pursue a higher rate of return.
Rights and permissions
Copyright information
© 2019 Springer Science+Business Media, LLC, part of Springer Nature
About this chapter
Cite this chapter
Lee, CF., Chen, HY., Lee, J. (2019). Fixed Effects Versus Random Effects in Finance Research. In: Financial Econometrics, Mathematics and Statistics. Springer, New York, NY. https://doi.org/10.1007/978-1-4939-9429-8_6
Download citation
DOI: https://doi.org/10.1007/978-1-4939-9429-8_6
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-1-4939-9427-4
Online ISBN: 978-1-4939-9429-8
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)