CEO Turnover Database

I maintain a dataset of forced CEO turnovers together with my co-author Alex Wagner (University of Zurich).

The dataset contains the announcement dates of forced CEO turnovers of all firms recorded in the Execucomp database between 1993 and 2020. It currently contains 1,438 forced turnovers and is therefore the largest dataset of forced CEO turnovers with complete announcement dates.

The file comes from the data used in Peters and Wagner (Journal of Finance, vol. 69, no. 4, pp. 1529-1563, 2014) and Jenter and Kanaan (Journal of Finance, Vol. 70, no. 5, pp. 2155-2184, 2015) and it has been extended to 2020 by Peters and Wagner since publication. The construction of the dataset has also benefited from turnover data generously provided to us by Greg Nini, Luke Taylor, Cami Kuhnen, Andrea Eisfeldt, and Ofer Eldar. In particular, discrepancies in classification between these datasets and ours have been resolved.

The original criteria for classifying a CEO turnover as forced are from Parrino (Journal of Financial Economics, 1997) and are described in detail in Peters and Wagner (Journal of Finance, vol. 69, no. 4, pp. 1529-1563, 2014). We have slightly modified these criteria to better reflect the circumstances of CEO departure and to better exploit the increasing availability of information. (E.g., to reflect changes in the typical retirement age over time, we do not use a strict age cutoff of 65 which, under the Parrino (1997) and Jenter-Kanaan (2015) classification, classifies very few departures of CEOs over 65 as forced. We do not strictly classify departures of CEOs who remain on the board of directors as voluntary. We do not restrict the set of publications in which we search for information on the CEO’s departure as does Parrino (1997).)

An important feature of the dataset is that, since 2001, it records forced turnovers also for the cases where a company is dropped from Execucomp from year t to year t+1. This means that all firm-year observations in Execucomp which are not coded as “forced” in this dataset are cases where either no turnover occurred or where the turnover was voluntary. There are very few exceptions to this rule. Since Execucomp regularly backfills or modifies some observations of earlier years, our dataset may miss a few turnovers or assign a turnover to a different fiscal year than a current vintage of Execucomp although we do regularly account for such changes.

The data can be easily merged with the full Execucomp CEO database (i.e. applying condition CEOANN == CEO) on the variables (gvkey, year, execid). On a current Execucomp vintage, the merge should produce a perfect match. All successfully merged observations within the 1993-2020 period are forced turnovers, all other observations are either voluntary turnovers or cases where no turnover occurred.

We are updating the dataset periodically to include more recent years and to account for observations in earlier years that, over time, are added or modified by Execucomp.

Please cite this data as follows (Peters and Wagner for the 2001-2020 period, Jenter and Kanaan for the 1993-2000 period):

“The Executive Turnover Risk Premium” by Florian Peters and Alexander Wagner, Journal of Finance, vol. 69, no. 4, pp. 1529-1563, 2014

“CEO Turnover and Relative Performance Evaluation” by Dirk Jenter and Fadi Kanaan, Journal of Finance, Vol. 70 (5), pp. 2155-2184, 2015