![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
To alleviate poverty at the grass-roots level the Kenyan government has devised the Constituency Development Fund (CDF) which allocates resources to constituencies for MPs and residents to decide how to spend. This article assesses how the CDF has been spent since its implementation in 2003 and whether MPs' re-election chances were affected by how they managed the fund. Using administrative data on the CDF management and two surveys conducted before and after the 2007 MP election, it is found that the use of the CDF intensified nearer the elections. However, the probability of the MP being re-elected was not affected by the amount of reported funds spent, but by how the CDF was spent and by residents' ethnicity. MPs that ran the most projects on education and the least on other projects such as on health or water were less likely to be re-elected.
Acknowledgements
This article was funded by the UK Department for International Development (DFID) as part of the Improving Institutions for Pro-Poor Growth (iiG), a research consortium aimed at studying how to improve institutions in Africa and South Asia. The views expressed are not necessarily those of DFID. The author also acknowledges financial funding from the Spanish Ministry of Science and Innovation (MICINN) with reference ECO2010-21668-C03-02. Thanks are due to the staff members of the Steadman group, with whom the pre- and post-election surveys were conducted, Stefan Dercon, Mwangi Kimenyi, Adam Pepelasis, the participants of the seminars at the iiG research consortium and the University of Oxford and two anonymous referees for their suggestions and comments.
Notes
1. Examples of past decentralisation programmes are the Majimbo system (1963); the District Development Grant Programme (1966); the Special Rural Development Programme (1969); the Rural Development Fund; the District Development Planning (1971) and the District Focus for Rural Development (1983).
2. Among the Rural Electrification Programme Levy, Local Authority Transfer Fund, Roads Maintenance Funds, Secondary School Education Bursary Fund, HIV/AIDS Fund and Free Primary Education.
3. Examples of these reports can be found at http://www.nation.co.ke/blogs/-/446696/463474/-/12sply2z/-/index.html and http://www.nation.co.ke/News/-/1056/214276/-/se1flv/-/index.html.
4. The primary sampling units (PSUs) were selected using the most updated Kenyan Census Enumeration Areas available from the Kenyan National Bureau of Statistics. Within each PSU, field teams travelled to a randomly selected sampling start point and selecting respondents using the same method as the surveys conducted by the Afrobarometer, explained in detail at http://tinyurl.com/3p8c966.
5. Respondents were replaced in this way, as the pilot study conducted in the worst areas affected by violence (Rift Valley, Nyanza and Nairobi) showed that the great majority of displaced people moved within the same constituency to areas where there were more clusters of people from their ethnicity. Respondents of Meru, Kikuyu, Luo or Luhya were found as the worst affected by violence and above all by displacement.
6. Although it would be interesting to assess whether respondents' ethnic attachment occurs on the basis of the ethnicity of the MP, this information is not available. The parliament and electoral commission do not provide information on the MPs' and candidates' ethnicity. Anecdotal evidence (and by analysing the surnames of contending MP candidates which are suggestive of one's ethnicity) suggests that political parties might select candidates taking into account their ethnicity. Since there is no exact evidence on the candidate's ethnicity, the regression results should be interpreted as the impact of respondents' ethnic attachment to a political party rather than a candidate's specific ethnicity.
7. Since respondents that could not be traced in 2008 were replaced by new ones, the number of observations remains 1026 in the analysed responses in both the pre- and post-electoral surveys.
8. The presented probit marginal effects of continuous variables should be interpreted at the mean of the variable, in this case at the average spending share in education, which is 53 per cent.
9. The respondent's wealth level was estimated with an asset index based on the respondent's answer to ‘Does your household own the following items?’ The items asked about were: book, radio, television, motorcycle, motor vehicle, house, oven, bicycle, fridge, telephone, washing machine, computer, mobile phone, land and cattle.