Testing the lawyer-induced litigation hypothesis in Europe

ABSTRACT

Utilizing a European panel dataset, we contribute to the scant empirical literature on the lawyer-induced litigation hypothesis. To address endogeneity problems that arise when estimating the effect of the number of lawyers on civil litigation rates, we use two strategies. We first estimate our model by means of the 2SLS procedure. Second, we exploit the instrumental variable approach based on the linear GMM estimator of Arellano and Bond. The estimations result in a positive and significant effect of lawyers that is robust across the different model specifications and estimation methods in which we address endogeneity. In criminal litigation, where lawyers cannot induce demand, we find no such positive relation between lawyers and litigation.

KEYWORDS:

• Lawyers
• supplier-induced demand
• lawyer-induced litigation
• litigation

JEL CLASSIFICATION:

• K41
• J44

Acknowledgement

We thank Naci Mocan, Peter Grajzl, Valentina Dimitrova-Grajzl, Juan Mora-Sanguinetti and participants of the 3rd Economic Analysis of Litigation Workshop for helpful comments and discussions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Scholars regularly refer to a ‘litigation explosion’, ‘hyperlexis’ or ‘adversary society’.

² For instance, Germany, Slovenia and Italy list minimum fees (European Commission 2004).

³ Note, however, that the information asymmetry in the healthcare sector is somewhat different compared to the lawyer profession. That is, a patient has little information on the quality of medical care (even after treatment), whereas a litigant at least observes whether he loses the case. Therefore, we expect that if we find evidence of supplier induced demand, the magnitude of the effect will be smaller compared to the effect in the health care sector.

⁴ Albania, Armenia, Austria, Azerbaijan, Bosnia and Herzegovina, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Italy, Latvia, Lithuania, Macedonia, Moldova, Montenegro, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland and Turkey.

⁵ In the CEPEJ questionnaire, ‘civil law cases’ refer to other than criminal law cases and include for example family law cases, commercial law cases, employment dismissal cases and administrative law cases.

⁶ Because they are time-invariant, use of contingency fees and availability of legal insurance are only used in models without country fixed effects.

⁷ With exception of Growth, which has negative values, and Congestion, which is measured as cases per judge.

⁸ A list of law schools in Europe was gathered from HG.org, an online law and government information site, founded by Lex Mundi. The year of foundation of the law faculty was gathered from the official website of each university.

⁹ We grouped countries into 4 categories: North, East, South and West-Europe.

¹⁰ The first stage results are omitted because of space considerations. The results are available upon request.

¹¹ The data were gathered from the OECD (statistics on Education and Training – Graduates by field of education) and from National Statistical Offices for the non-OECD countries.

¹² A lag of two years may not seem a considerable amount of time, but note that we are looking at law graduates, and not newly registered law students. Hence, the decision to start a law school programme from graduates in year t was made in year t−x (x being the number of years of law education). Furthermore, we re-ran our regressions and lagged graduates with 4 years. Results were qualitatively the same but the sample reduced in size due to data limitations.

¹³ Of course, we also replace the number of pending civil cases by the number of pending criminal cases.

¹⁴ For example, contingency fees are allowed by law in only 37 per cent of the OECD countries (Palumbo et al. 2013).

¹⁵ Time trends are dropped because of collinearity.

¹⁶ Time trends are dropped because of collinearity.

Additional information

Funding

This work was supported by the Research Foundation Flanders [12S3117N];