The sample for this study was taken as the overlap between the sample of
Datastream* U K company long-term liability data, and the LSPD* 9 return sample from
1976 to 1990 This reduced the sample available for these time series empirical tests to 172 companies
" Dataslrcam contains numeric information The information relates to: (I) Company financial information from the annual repon of UK quoted companies and a selection of overseas quoted companies Information on UK companies dales back to 1968. where available (2) Slock market information (3) Economics and business statistics.
'' The London Share Price Database (LSPD) contains several different samples e g a random sample of 33% of the companies ( by market value) quoted on the London Stock Exchange in January 1 ‘>55 together with 33% of new issues in each year Since 1973, there is a complete history for all U K companies quoted in London, including those companies traded on the USM There is no survivorship bias in the LSPD after 1975
In order to reduce the bias from nontrading, the Trade-to-Trade (TT) method, which defines periods in terms of the timing of the most recent actual trades, was used for estimating risk measures Thus, returns on securities are calculated on the basis of those trades nearest the month ends, and the market returns are then measured on the same dates and over the same calendar period A daily index of market returns is required It is defined as follows in Datastream
where Rl, is the broadly-based, capitalisation weighted Financial Times-Actuaries all-
share return index (FT A)10 on day t The calculation of the return index is based on the re-investment of gross dividends and so ignores tax and re-investment charges,
PI, is the Financial Times Actuaries All-share price index on day t,
XDchange is the ex-dividend adjustment (X I)) ",
f is the grossing factor (normally 1), if dividend yield is a net figure rather than
gross / is used to gross up the yield,
Then, monthly market return, Rnx , over the same calendar period, [ s,, ( s -1), ], of
monthly return, /<„ , for security / is measured as the formula
10 The FTA is a market-value weighted arithmetic index covering 75« larger U K companies Unlike the London Share Price Database (LSPD) returns file, the FTA All Share Index excludes commodity companies, e g rubber, teas, coppers, mining and tins, and investment trusts Details about the FTA index sec "Guide to FT statistics" published by the financial limes.
" When a share goes ex-dividend, the price of the share will drop by an amount corresponding to the dividend paid (all else being equal) In the UK under the imputation tax system tax cfTccts on this ex- dividend adjustment would be expected to be less than under the classical tax system
PI, + XDchange* f
RI, ) ( PI + XDchange * f
— i— = In — :---
The leverage variable theoretically is based on market value measures of debt and equity Determining the market value of debt may seem straight forward but is troublesome as most of the events of debt cannot be valued so easily at market value However, using different procedures to estimate the market value of debt, most empirical researchers found that book value measures provided better leverage- adjusted beta estimates For example, Hamada (1972) showed that market values of equity and debt change dramatically as security prices rise and fall over time Fuller and Kerr (1981) calculated both market and book value measures of debt and equity, finding that leverage adjustments based on book values of equity and debt worked better than those based on market values of equity and debt as they presumed that book values are more stable Through careful evaluation of variables. Bowman (1980) compared the empirical applicability between the book-value and market-value measures of leverage in explaining systematic risk He concluded that a mixed measure of book-value debt and market-value equity was the most closely associated with the firm’s market risk Although Mulford (1985) indicated that employing book values as proxies for market values of debt and equity may become a severe problem in financial empirical research during periods of extreme volatility in interest rates, the book-value of debt was still used in his study. Considering the high correlation between market- value and book-value measures of debt demonstrated by Bowman (1980), the considerable costs of estimating the market values of the variety of debt instruments employed by the many companies in the sample does not appear worthwhile at this preliminary stage of testing a new model
According to these considerations, the market value of common stockholders’ equity, E , was measured as equity market value per share of common stock times the number of common shares outstanding and was adjusted every month The D o f this study was measured as the book value of long-term liabilities12 obtained from the Datastream company accounts data and was adjusted each year Therefore, the research sample was constrained in those companies for which the debt data was available