Understanding the three-tier government

Raila Odinga favours a shift from two levels of government — the national and the county — to three levels, the third being between the other two.

This is part of his push for bringing back ‘Bomas’, the draft Constitution produced in 2004 by the National Constitutional Conference.

The 1963 Majimbo constitution (designed to a considerable degree to protect minority communities) created seven regions (similar to provinces until 2013 ), with Nairobi as a capital city, not a region. Each region had an elected assembly, which elected a president from among its members. Each regional government took the form of a committee of the regional assembly, with the regional president as a member and a public servant as chief executive officer. It was a four- tier system. Within each region there were local authorities: The top level being either municipalities or counties (rural), and the second level townships, divisions or local councils. In fact, divisions were themselves to be divided into local council areas. All councils were to have at least 75 per cent elected members, with possibly some appointed ones, also.

Regions could make law on a wider range of matters than the modern counties. They had far more powers to tax than current counties. And although there was a national police force, each region had a contingent, for which it paid.

It was clearly envisaged that these various assemblies and councils would be part-time.

This system was strangled soon after birth by Jomo Kenyatta, who had plans for a highly centralised government, with himself at its head.



Yash Ghai had a vision of a country building from the bottom. Below the national government would be four levels: Village, location, district, and province (the village symbolic of people’s sovereignty). Village and district councils would be directly elected by the people, and at the district level, a governor would also be directly elected.

Location councils would be composed of members of village councils. Districts would be the main level of government, partly because some provinces were so large that their headquarters were as remote from most people as Nairobi itself, and also because of concerns around identity and potential conflict.

Under time pressure, the CKRC envisaged that many details of the new system would be worked out in legislation to be passed by Parliament, including on the role of provinces.

A number of NCC delegates recommended that provinces cease to exist or be broken up. “That system has made us think in terms of tribes and we know tribalism is the biggest cancer in this country today,” a Moyale MP said. But the former Vice-President of a region in the majimbo period argued: “Had we been allowed to continue, the story today, in terms of development, would have been different.”

Much discussion centred on having larger regional units as the principal level of devolution: Perhaps 10, 13, 18 or even 27 units. By mid-September 2003, the conference had decided on 18 ‘zones’, but, for reasons that are not really clear, the 18 regions were reduced to 14 in February 2004. At the end of January 2004, many issues were still ‘contentious’, including whether there should be three or four levels of devolution; whether the constituency rather than the district should be a unit of devolution; and distribution of powers between various levels of government.

Only days before the winding up of the NCC, Ghai produced a ‘Compromise Proposal for Devolution’, making it clear that there was still a good deal of confusion on the topic. He suggested the root of the problem lay with the question of what was the principal unit of devolution: “One group favours the district, the other the region. The compromise the committee struck (which reflects CKRC’s proposal) is to vest significant devolved powers in the district but to find a not unimportant role for the region. … This produces somewhat confused lines of authority as well as adding greatly to the cost of devolution.”

Ghai proposed about 20 units at the intermediate level. This would have enabled greater participation of people than would be possible in a smaller number of large units. The smaller regions would be less threatening to national unity and more protective of the interests of minorities.



However, in March 2004 the NCC adopted 14 regions, and 70 districts (the latter as the main recipient of powers). The regions were to “coordinate the implementation, within the districts forming the region, of programmes and projects that extend across two or more districts of the region”. On these issues they would have the power to make law, which would prevail over district laws.

Each region would have had an assembly composed of four delegates from each district, elected by the district council but not members of the district council. The regional executive included a premier, elected by all the district councils, with other members chosen by the premier and approved by the regional assembly.

This obviously would have been very expensive: 70 districts and four Nairobi boroughs, plus 14 regions, and locational government (elected by the residents) too. However, although this was not spelled out, there was still an assumption that assemblies and councils would be part-time, like the old local authorities.



The CoE’s first draft in late 2009 drew heavily from the Bomas draft. And the weakness of the Bomas provisions about the role of regions was pointed out by some commentators on the CoE draft: That the role of the regions was not very clearly explained. So, was the existence of an intermediate level of government actually justified?

The CoE responded by removing this element of their draft. They reduced the number of districts (now called counties) to 47, thus making them rather larger on average. They also significantly increased the size of county assemblies by the provision on top-up members to ensure gender balance. The result of this (however desirable from the gender perspective) is currently to increase the number of MCAs from 1, 450 (for wards) to 2, 086 (including 542 extra women and 94 for marginalised groups — of whom about half are women).

The CoE second draft was laid on the table of the Parliamentary Select Committee in Naivasha, and duly hacked about, but they did not do much to devolution.



The South African Constitution does have three tiers. We could earn a good deal from them. There are, in fact, many issues that would have to be thought about and resolved if the Bomas model was to be adopted, including on the legal position of the capital.

Bomas was clear about neither the powers nor the institutions of regional government (again due to time pressure). These we would need to clarify in any revision of the constitution that adopted three tiers.

The Constitution now allows counties to set up joint committees and authorities, and several grouping of counties are emerging. Would establishing intermediate governments work better than this? Or is it possible that county groupings might be different for different purposes?

Thirdly, below county level administration does not have to be democratically elected. Is this a good thing?

Fourthly, the nature of county government now — with full-time MCAs, and so many of them, and with little presidents as governors (with motorcades, mansions and flags) — is much more elaborate and expensive than the drafters imagined. But can they ever be scaled down (fewer counties or fewer MCAs)?

Would a referendum support this? If not, can we afford an intermediate level of government? Or can we scale down Parliament to some reasonable number (perhaps 94 MPs, one woman and one man, for each county as has been suggested)? Would the turkeys — in this case MPs — vote for Christmas?

Alternatively — or additionally — is there some way we can constitutionally guarantee more modest payments for those elected? The drafters tried, with the Salaries and Remuneration Commission. The SRC tried but the burden of paying politicians remains enormous.



 Jill Ghai is a director of the Katiba Institute


Stay in the Know!

We respect your privacy.