Municipal collapse is rarely ideological, but a contractual immaturity

A municipality that could not pay Eskom in July 2026 did not begin failing in July. The failure started earlier, when governing political power proved weaker than the contest for it, and when administrative continuity collapsed with each change of executive. The Local Governance & Coalition Stability Index, launched on Wednesday, 2 September 2026 by the Pan-African Institute for Fiscal and Policy Studies, examines that earlier institutional failure.

In July 2026, National Treasury invoked section 216(2) of the Constitution and halted the equitable share transfer to 69 municipalities. The funds have since been released, the FFC questioned how the in-strument was applied, and Parliament directed that enforcement be accompanied by support rather than used as a recovery strategy.

I argued then that withholding would not repair a broken budget framework. On 2 September 2026, PA-IFS published the evid-ence. The Local Governance and Coalition Stability Index (LGCSI) is the first volume of our Municipal Intelligence Series. It exam-ines 103 of South Africa's 257 municipalities across a full term, from the 2021 elections to now, using only statutory data sources: the AGSA, Treasury, CoGTA gazettes, and IEC council records. LGCSI asks one question. Can these councils govern themselves?

Because the index chose the sample de-liberately, and deliberately over-sampled councils in distress, it cannot tell you what proportion of South Africa's municipalities are failing. It does not rank municipalities. It does not rank parties. And it forecasts no election result.

A municipality that could not pay Eskom in July 2026 did not begin failing in July 2026. It failed earlier, in the council cham-ber, when the rules governing who holds the executive proved weaker than the con-test for it. Across 8 metros in 1 term (2021-2026), we recorded anything from 1 mayor to 6. Each turnover dissolved the entire mayoral committee, and the municipal man-ager's office rotated through appointments, court removals, and acting incumbents.

That is the mechanism, and the LGCSI's central contribution. We call it the adminis-trative shield: whether, when the mayor changes, the municipality’s accounting of-ficer and Chief Finance Officer keeps their jobs. Where it holds, the budget still passes, the bulk account is paid, the audit gets done.

Where it breaks or their loss their jobs, all three are likely to also stop. The damage does not travel through the politics. It travels through the administration.

The Municipal Structures Amendment Bill (Coalition Bill) now before Parliament regu-lates the political tier. A reform silent on the accounting officer leaves that mechanism intact. The most uncomfortable finding is that instability is not a property of coalition government. Mangaung was governed by one party with an outright majority, and re-cords the same executive churn, the same

removal of senior officials and the same fiscal deterioration as the most fragmented coalition metro in the country.

Midvaal is also single-party, and among the most stable in our sample. What separ-ates them is not the number of parties, but whether the governing arrangement is writ-ten down and enforceable.

Of the 94 councils assessed on this meas-ure, 18 have some sort of governing agreement that can be read within the LGCSI sample. Those with published, bind-ing agreements and independent dispute resolution experienced limited disruption across the term. Those with pacts but no way of settling a dispute outside the cham-ber were calmer, but not stable. Tshwane signed a formal pact in December 2021; it collapsed in September 2024. Clearly, sig-nature is not architecture.

There is a second half, which councils do not control. Local government delivers roughly 46% of the frontline functions’ cit-izens experience directly and receives 9.9% of nationally raised revenue to do it.

Municipalities spend about R31 billion a year on functions assigned to other spheres, libraries, primary healthcare support, hous-ing administration. That is not a governance failure. It is an arithmetic one, and enforce-ment does not resolve arithmetic.

So, the LGCSI propose one package, in this order;

Institutionalise the politics: written, pub-lished coalition agreements lodged within 30 days, with independent dispute resolu-tion, and a constructive vote of no confidence, removal only where a successor is elected at the same sitting.

Protect the administration: decouple the tenure of the municipal manager and chief financial officer from the political executive, with caretaker delegations that keep bulk creditors paid through a transition.

Fund and enforce the mandate: withhold the equitable share from councils that adopt unfunded budgets, but with a defined cure, differentiated to the council, and paired with the duty to build capacity.

And beneath all three, overhaul the divi-sion of revenue framework. Note the order. We support withholding. We place it third. Applied first, it disciplines a council that has not been given the means to comply. Coun-cils elected on 4 November 2026 must form executives within 14 days under the current legal framework, and no law or regulations requires those terms to be written down. In December 2026, the next transfer decision falls due.

The LGCSI is not merely a record of the past municipal term. It describes the con-ditions under which the next one will begin. The question is whether South Africa will institutionalise its municipal politics, protect professional administra-tion and fund the mandate before the cycle starts again.