Indirect Valuation and Earnings Stability: Within-Company Use of the Earnings Multiple

Abstract

This paper investigates statistical significance of earnings stability in the within-company indirect valuation method. We empirically establish superiority of a within-company earnings multiple valuation technique for the relatively most stable companies. Favorable empirical results are robust against different means of operationalization of the stability construct and valuation multiples. Results of this paper indicate that the indirect within-company price-to-earnings valuation yields the most precise and the most accurate value estimates.

Appendix

This table shows the results of the panel regression of ln(market value) on ln(earnings) using company-fixed effects and company clustered standard errors. Panel A represents the results of the regression applied on a full sample of 284,390 company-years divided into 10 earnings stability deciles based on a 5-year rolling standard deviation of the inverse hyperbolic sine of earnings. Panel B represents the results for the subsample of peer companies. We define a peer-company as one being drawn from the subsample of companies from the same year, country, industry and earnings stability quantile. In orded to include the company into analysis its peer-group has to constitute of at least 5 companies.

$$ \ln \left(\mathrm{Market}\;{\mathrm{Value}}_{\mathrm{i},\mathrm{t}}\right)=\alpha +\beta \times \ln \left({\mathrm{Earning}}_{\mathrm{i},\mathrm{T}}\right)+\varepsilon $$

We construct the confidence intervals of the regression coefficients using 95% confidence level. If the confidence interval includes 1.000 we cannot reject the general linear hypothesis of Beta coefficient being differenct from 1.000.

This table shows the results for the within-company valuation technique. We estimate the market value (hereby “MV”) of a company 4 months after its fiscal year end as a result of multiplying the last year’s price to earnings ratio of the given company by its last announced earnings. We calculate the absolute, squared and absolute log valuation error as follows:

$$ {\displaystyle \begin{array}{c}{\varepsilon}_{\mathrm{i},\mathrm{t}}=\frac{\left|\overline{{\mathrm{MV}}_{\mathrm{i},\mathrm{t}}}-{\mathrm{MV}}_{\mathrm{i},\mathrm{t}}\right|}{{\mathrm{MV}}_{\mathrm{i},\mathrm{t}}}\kern0.2em {\varepsilon}_{\mathrm{i},\mathrm{t}}={\left(\frac{{\overline{\mathrm{MV}}}_{\mathrm{i},\mathrm{t}}-{\mathrm{MV}}_{\mathrm{i},\mathrm{t}}}{{\mathrm{MV}}_{\mathrm{i},\mathrm{t}}}\right)}^2\\ {}\kern-6.23em {\varepsilon}_{\mathrm{i},\mathrm{t}}=\left|\log \left(\overline{{\mathrm{MV}}_{\mathrm{i},\mathrm{t}}}\right)-\log \left({\mathrm{MV}}_{\mathrm{i},\mathrm{t}}\right)\right|\end{array}} $$

We construct the interquartile range as value of the 75th percentile less value of the 25th percentile and Interdecile range as a value of the 90th percentile less value of the 90th percentile of absolute and squared valuation error

Panel A contains results of the valuation analysis conducted on the Final Sample, panel B contains results for the Subsample of Peer Companies.