Kenya's Most Corrupt Counties Ranked - Stats Kenya

Corruption is a major hindrance to economic development everywhere. In Africa, countries with the highest levels of corruption also had lower GDP per capita. However, other than the Corruption Perception Index from Transparency International, which highlights the levels of corruption in all countries of the world, fewer tools exist for measuring subnational corruption within countries. 

For instance, despite the ubiquity of corruption in Kenya, it's difficult to determine the extent of corruption with a sufficient degree of accuracy. Furthermore, very little attention is paid to corruption in subnational administrative units such as counties. Therefore, developing such a tool would be in the public's interest.

First and foremost, here's a list of Kenya's most and least corrupt counties based on our index. Our scale runs from 0 to 100, with higher values indicating lower corruption and lower values higher corruption.

Most corrupt counties in Kenya

1. Nairobi 32.5
2. Busia 53
3. Kakamega 56.6
4. Kiambu 61.3
5. West Pokot 62.8
6. Nyandarua 64
7. Bungoma 64.8
8. Samburu 65.3
9. Nakuru 65.3
10. Kisumu 65.5
11. Baringo 67
12. Homa Bay 67.7
13. Garissa 69.2
14. Machakos 69.8
15. Turkana 70.2
16. Isiolo 70.3
17. Siaya 70.8
18. Tharaka Nithi 72
19. Mombasa 72
20. Migori 72.1
21. Kwale 72.2
22. Elgeyo Marakwet 72.4
23. Laikipia 72.8
24. Kajiado 72.8
25. Uasin Gishu 73

The least corrupt counties in Kenya

1. Embu 87
2. Nyeri 85.9
3. Mandera 85.9
4. Lamu 85.3
5. Vihiga 84.1
6. Kericho 81.9
7. Taita Taveta 80.3
8. Marsabit 80.2
9. Nandi 78.7
10. Kirinyaga 78.7
11. Kisii 77.3
12. Kilifi 76.8
13. Wajir 76.7
14. Makueni 76.6
15. Meru 76.1
16. Narok 75.9
17. Murang’a 75.3
18. Trans Nzoia 73.7
19. Tana River 73.6
20. Nyamira 73.2
21. Kitui 73.1

What’s a subnational corruption perception index

According to Transparency International, Kenya had a Corruption Perception Index (CPI) of 31 out of 100 in 2023 and was ranked 126^th globally. Our corruption levels are high compared to Denmark, which had a score of 90 and was ranked number one globally.

Our subnational corruption index works the same but uses a different methodology to arrive at scores.

Due to differences in the manifestation of corruption across countries, our corruption index recognizes forms of corruption such as bribery, misuse of office, embezzlement and diversion of public funds, state capture, and nepotism.

Furthermore, it recognizes what has been called the cultures of corruption, or behaviours that, while not formally recognized as corrupt, are inherently corrupt. These forms of corruption are highly prevalent in Africa and include reckless public expenditures on foreign and domestic travel, unnecessary allowances, donations, charities, and harambees held by elected officials, where large sums of money are publicly dished out.

What did we include in our 2023 corruption index?

Due to the scarcity of data related to corruption, it is almost impossible to measure corruption. Transparency International surveys key stakeholders across disciplines and industries to determine their scores.

We arrived at our scores by following a novel approach. First, our primary data source is the Ethics and Anti-Corruption Commission (EACC), the body investigating corruption in Kenya. We reviewed all their tweets from 1 February 2023 to 17 March 2024, looking at the names of counties and individuals that were continuously mentioned on graft-related issues.

Our assumption following this method was that by being actively investigated by the EACC, public officials must, with a very high degree of certainty, be involved in corruption. We then checked how often the EACC mentioned specific individuals or counties and isolated unique mentions. A unique mention is a code for a unique corruption case. For instance, if a county was mentioned for embezzlement and then mentioned for forgery in a different case, those were counted as two unique mentions. A higher number of unique mentions denotes a higher prevalence of corruption.

Our next source of data was corruption surveys. In 2023, the EACC conducted the National Ethics and Corruption Surveys.^[1] They asked people whether they had paid bribes before, whether they had been requested for bribes by public officials, and many more. From these, we isolated various indices that were distributed by county, including:

1. Likelihood of bribery demand
2. Prevalence of bribery
3. The average size of bribes
4. Share of national bribes

We include these four indicators as part of the variables we would use to calculate our subnational corruption index.

Our third data source included various indicators that would allow us to measure cultures of corruption. These included looking at data from the Controller of Budget to find out the following:

1. MCA sitting allowances by county
2. County expenditures on foreign travel
3. County expenditures on domestic travel

It is worth noting that these are not illegal activities. However, since they display a culture of corruption rather than corruption itself, we figured we would include it in our index. The assumption was that counties that spend conspicuously more on travel and allowances are more likely to be more corrupt than those that spend less than average.

What is our strongest indicator of corruption?

To be fair to the counties, we needed to evaluate all the above variables based on how strongly or weakly they predicted actual corruption. An EACC investigation and domestic travel are not equal predictors of corruption. As a result, we had to find a way to assign different weights to these variables based on how strongly they correlated to corruption.

We used principal component analysis to arrive at composite scores that we used to weigh our variables. An EACC mention had the highest weight, meaning it had the highest relationship to actual corrupt behavior. Other strong predictors of corruption were the likelihood of corruption in the EACC survey and the prevalence of corruption. 

We multiplied the weight with the z-score from each variable, and then all the resulting values were averaged to find a single un-normed value for corruption.

To find a more intuitive scale, we had to norm our values.

See Also

1. Gross County Product per Capita
2. Access to Electricity in Kenya by County

References