Asif R. Anik, Gunnar Breustedt, Siegfried Bauer:
An Inquiry into Causes of Corruption: Instrumental Variable (IV) Approach


1Justus-Liebig University Giessen, Institute of Farm and Agribusiness Management, Germany
2Christian-Albrechts-University of Kiel, Institute of Agricultural Economics, Germany

Corruption as a research topic is relatively new. Most of the literature on corruption is at macro level and deals with cross country analysis. This gives little insight about the relationship between corruption and individual agents. The present work is an attempt to reduce this gap by analysing experiences of 210 farm households belonging to six villages of Bangladesh. Instrumental variable (IVPROBIT and IVTOBIT) approaches are employed to identify farmer's socio-economic determinants for corruption and bribe experiences. This study focuses on two questions: which farm households are more likely to experience corruption; and why amount of bribe varies among households for the same service. In the IVPROBIT model the dependent variable is the dummy of corruption experience(s), whereas it is amount of bribe in the IVTOBIT model.

Among service recipients, 70.4% experienced corruption during the period July 08-June 09 while interacting with different service delivery organisations. Bribery is most the common (64.7%) form of corruption; followed by negligence of duties (19.3%) and nepotism (9.2%). In the IVPROBIT model, household's relation with sub-district executive officer's office (UNO office) and location dummy has significant impact. The associated signs explain that households having relation with UNO office and living in urban areas are more likely not to experience corruption than their counterparts. The UNO office has immense influence by being the prime authority for implementing and monitoring different government programmes and projects in respective locality. These influences are helpful while facing corruption. Several factors are responsible behind notable difference in rural and urban experiences. Compared to urban areas services in rural areas are limited. Besides media, civil society organisations, and people's awareness are relatively less in number and, thus, maybe less effective in rural areas.

The IVTOBIT model shows that household's living expenditure (per head) has positive significant impact on bribe amount. This suggests that service recipients' socio-economic status except economic capability to pay bribe does not get consideration whenever they fall on bribery. The rent seekers only discriminate service recipients depending on their economic status; as they find rich households more suitable to devour higher sum of bribe.

Keywords: Bribe, corruption, instrumental variable, Probit, Tobit

Full paper: Poster (pdf-Format):


Contact Address: Asif R. Anik, Justus-Liebig University Giessen, Institute of Farm and Agribusiness ManagementSenckenbergstr.3, 35390 Gießen, Germany, e-mail:
Andreas Deininger, October 2010