Accounting Information Free of Selection Bias: A New UK Database 1953-1999

Preprint
Publisher Website
Google Scholar
Abstract
Accounting information is indispensable in empirical finance. Unfortunately, existing databases, particularly for non-US markets, suffer from selection bias. I construct a new UK data set by supplementing existing databases with handcollected balance sheets. It contains about 100,000 firm-years of data, covers virtually all UK firms listed on the London Stock Exchange between 1953 and 1999, is fully cross-referenced to the London Share Price Database and free of selection bias. Use of previously existing UK databases is shown to introduce biases against small and high book-to-market stocks, and downward biases in sample moments of returns, a symptom of truncation of extreme observations. The new data set is particularly useful in contexts where inference is sensitive to selection bias or repeated mining of the US data is detrimental, and for any study that needs accounting measures such as book-to-market ratios for UK firms.