Performance-based Incentives and the Behavior of Accounting Academics: Responding to Changes

Abstract

When laws change the rules of the game, it is important to observe the effects on the players' behavior. Some effects can be anticipated while others are difficult to enunciate before the law comes into force. In this paper we have analyzed articles authored by Spanish accounting academics between 1996 and 2005 to assess the impact of a change in the Spanish university regulation. Results indicate a switch from publishing professional papers to academic ones and also to change research methodologies in order to meet the new requirements. This was to be expected due to the explicit mention of the law in favor of academic journals. However, we have also detected a significant decrease in the publication of professional papers. These side-effects could have a negative impact on the transmission of knowledge from university to society putting the relationship between accounting research and professional practice in jeopardy.

Key Words:

• Accounting research
• performance-based incentives
• scientific production
• accounting academics
• research and practice gap

Acknowledgements

The authors thank M. Angeles Morales for assistance with data collection and database construction. They also thank Stefan Van Hemmen and David Urbano for their valuable comments.

Funding

This study was financially supported by the Spanish Department of Science and Innovation (Plan Nacional de Investigación Científica, 2010-2013, Code ECO 2010-18967).

Notes

1 According to the new Spanish regulation, the quality of journals is determined by their presence in several internationally recognized indexes. Particularly, the regulation notes Science Citation Index, Social Science Citation Index, Econlit and Latindex, leaving the list open to other similar possibilities and without specifying any ranking in the indexes, although high quality papers can have a special consideration. While this view is an interesting and debatable issue in itself, it is far beyond our scope of discussion and therefore not addressed here.

2 Research assessments have been very scantily used in the field of accounting and finance. From 1989 to 2005, only 30% of scholars applied for voluntary assessment. Of those, only 27% received positive evaluations. In 2005, only 31 scholars in the field were evaluated, with a success rate of 65%.

3 Findings of a 2008 study commissioned by HEFCE (The Higher Education Funding Council for England) states that the RAE encouraged researchers to concentrate on publishing more papers, particularly in peer-reviewed journals (HEFCE 2008).

4 This database is freely accessible at the following URL: http://ec3.ugr.es/in-recs/. To our knowledge no Spanish accounting scholars published in English only during this period.

5 The EAA taxonomy is more complete than Sundem's classification, because it identifies in great detail the numerous categories of financial accounting and does not consider ‘research method’ and ‘professional papers’ to be topics, as Sundem does.

6 Although of course the categorization will not be completely free of bias.

7 Publish or Perish is a free software program that retrieves and analyzes academic citations. It uses Google Scholar to obtain raw citations, then analyzes these and calculates a series of citation metrics. The webpage http://www.harzing.com/pop_gs.htm explains the advantages and limitations of using Publish or Perish as opposed to other databases, such as the Thomson ISI Web of Science. From the accounting point of view, the obvious advantage of Google Scholar over the ISI Web of Science as a source of information is that accounting journals are well represented in the former and extremely scarce in the latter.

8 In this article we work with categorical variables (orientation, year, country and research area and method). Thus, a bi-variable analysis based on contingency tables and Pearson χ² is the most appropriate statistical tool to apply. We could have employed a multivariate analysis, such as logistic or probit regression with article orientation (professional or academic) as the dependent variable; however, the innumerable independent variables and their interactions (e.g., interaction of year with area and empirical/non-empirical methods yields 260 variables: 10 years × 13 areas × 2 methods) with increasing multicollinearity would lead to problems that are difficult to manage. Even so, we have performed this analysis, but by only regressing the dependent variables parsimoniously: in the first case with years, in the second with research method, and finally with research areas. The findings are similar to those we obtained, although less significant.

9 In a separate analysis, not included in Table 4, we found that changes between 2004 and the earlier years emerged as significant in the year 2004 for both deductive and statistical methods.