Delistings and the costs of governance: a study of European stock exchanges 1996–2004

The recent going dark literature in the US has analyzed the impact of the Sarbanes–Oxley act on the decision to deregister (go dark). Marosi and Massoud (2007) find that higher audit costs induced by the Sarbanes–Oxley act have had a significant impact on the decision to delist. Engel et al. (2005) also find a significant effect of the Sarbanes–Oxley act on firms’ going private decision, particularly for small firms. Further Kamar et al. (2006) and also Block (2004) find that small firms tend to exit due to the Sarbanes–Oxley act. In contrast, Leuz et al. (2006) find that cost savings alone are unlikely to matter for the decision to delist, but may push more poorly performing firms to go dark. Note, however, that there is a difference between going private and going dark. When a firm goes private it becomes fully private whereas stocks of a firm going dark can still be traded in over-the-counter market.

The listing and cross-listing decision is not entirely comparable to the delisting decision. For instance corporate governance regulation is known ex ante a listing or cross-listing, however, while public new regulation serves as external shocks.

A survey of the 224 largest public firms in the USA by Financial Executives International with regard to the direct costs of complying with Section 404 of the Sarbanes–Oxley act finds that the average first-year estimate is almost $3 million for 26,000 h of internal work and 5,000 h of external work, plus additional audit fees of $823,200, or an increase of 53 % (Zhang 2005).

While the idea of colonial transplants is not directly applicable to recent changes in corporate governance regulation, the remarkable spread of quite uniform regulation (codes of best practice, EU directives, increases in investor protection measures) to countries with quite different corporate governance systems does nourish a suspicion that not all of this regulation is efficiently adapted to the local context. It is not clear whether for example smaller countries can improve their stock markets simply by adopting Anglo-American standards.

The data consists of information from the following 21 countries: Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, The Netherlands, Norway, Poland, Portugal, Slovakia, Slovenia, Spain, Sweden and United Kingdom. Data for Eastern Europe and Luxembourg is not available in the LLSVPV data set, which limited the sample to 15 countries.

The Standard Industrial Classification (SIC) codes.

The Pagano and Volpin (2005b) dataset is available at http://​www.​e-aer.​org/​data/​sept05_​data_​pagano.​zip.

Other related indices regarding securities laws such as disclosure requirements and liability standards (see La Porta et al. 2006) could also be used but they are also not time-varying. They also show to be strongly correlated (not reported) with our used measure of investor protection regulation (0.5–0.8).

EU Directive on the market for financial instruments (2004/39/21. April 2004), EU Directive on prospectus (2003/71/4. November 2003), EU Directive on Market Abuse (2003/6/28. January 2003), EU directive on Transparency (2004/109/15. December 2004), EU directive on Takeovers (2004/25/21. April 2004).

ROA is defined as net income before preferred dividends plus interest expense on debt-interest capitalized after taxes relatively to last year’s total assets.

Firm value, q, is defined as the sum of market value and debt book value to book value of total assets.

Average firm value is therefore the annual country averages of the firm-specific q-values.

An alternative strategy would be to match the subsequently listed firms to their nearest neighbor in terms of size or industry, but we maintain the comparison of the entire populations to include more information in the matched sample, for example greater country variation.

We have information of the firms’ main industry (SIC codes) affiliation from which we aggregate industry affiliation to 25 different industry groups.

Instead if the difference between firm-specific q and country average q is employed as an explanatory variable we find that when a firm is relatively ‘overvalued’ compared to country average it is less likely to delist (results not reported).

If a cumulative measure of changes in the LSSVPV index is used as an explanatory variable a significantly positive impact from this variable is found on the likelihood of going private (results not reported).

This results hold if instead of the mean, as robustness check, a similar test is performed using the 75 % percentile.

Concentrated ownership is defined as either above the mean or median ownership concentration.

The legal origin variable equals 0 when common law country and 1 if civil law country (La Porta et al. 1998). The proportionality index (Pagano and Volpin 2005b) equals 3 if 100 % of seats are assigned via a proportional rule, 2 if the majority of seats are assigned by this rule, 1 if a minority of seats are assigned proportionally, and 0 if no seats are assigned this way. For example Scandinavian countries are given the value of 3, while UK’s measure is 0. Unionization is measured as annual labor union density rates (Visser 2006).

If the electoral system is proportional winning a majority of the votes is crucial, whereas it is majoritarian when winning a majority of districts ensures victory.