Voted vs executed: where the City spends more, where it spends less than planned

Each year in December, the Conseil de Paris votes the Primary Budget (BP): the detailed forecast of expenditure and revenue for the coming year. Thirteen months later, at the start of the following year, the same Council approves the Administrative Account (CA): the actual snapshot of what was spent. Both documents cover the same items, but their amounts never coincide. For good reasons (uncertain revenue, opportunities seized, delayed worksites) — and a few less good ones (tactical over-budgeting, projects shifted with no warning).

This article quantifies the gap between voted and executed for the Ville de Paris, across the six most recent comparable years (2019-2024). It identifies the five systematically over-executed line items and the five systematically under-executed line items, and offers three non-exclusive readings to understand these gaps.

The global execution rate: between 90% and 95%

Across all expenditure, the City's overall execution rate fluctuates between 90% and 95% depending on the year. That is, 5 to 10% of voted appropriations are not actually spent — i.e. €500M to €1Bn per year, roughly.

The year 2020 shows the lowest rate (90.3%) — pandemic effect, slowed market, postponed worksites. 2019 reaches 95.4% — execution in line with forecast.

But behind this average, the gaps by line item are huge. It is this disaggregation that reveals the real texture of the City budget.

The five systematically over-executed line items

By ranking average gaps (Executed − Voted) / Voted across the six years, here are the five line items that systematically exceed their forecast:

The number 1 is striking: the "Economy" operating theme is over-executed by 35% on average over 6 years. The City votes €324M cumulative, it spends €471M. This is the most systematic gap in the Parisian budget — not a one-off overshoot but a structural forecasting bias, likely linked to economic-support schemes (exceptional grants, aid for retailers, recovery plans) that are triggered during the year and were not foreseeable at the vote.

The number 2 (Transport in investment, +23%) is almost as structural — it likely reflects multi-year commitments (tunnel renovations, IDFM contracts) that bite into the original year once invoicing is confirmed.

The five systematically under-executed line items

Conversely, here are the line items (filtered on voted volumes > €100M for relevance) that under-execute the most:

Number 2 (Loans & Debt) is numerically enormous: over 6 years, the City voted €3Bn of borrowing and debt-management operations in investment, but only carried out €2.24Bn — i.e. €743M not mobilised. This reflects a prudence policy: the City raises less in loans than it had theoretically authorised, in order not to weigh down debt servicing.

Number 5 (Urban planning) is the largest in voted volume: €4.35Bn programmed over six years, of which €350M not executed. This is the structural effect of large urban projects: between the commitment decision and the final settlement, typically 18 to 36 months go by — during which unconsumed appropriations are carried over from one year to the next.

Three readings to understand these gaps

1. The "political" reading (tactical over-budgeting)

Voting larger than what will be spent gives room for manoeuvre during the year. It is politically comfortable ("we had provisioned, we reacted") but not very transparent. In public debate, large gaps can be read as broken promises — particularly on investment, where the political visibility of a delayed project is high.

2. The "technical" reading (long worksite cycles)

Investment operations are by nature spread over time. The City records a multi-year amount in a programme authorisation (AP), and each year consumes a share of it as payment appropriations (CP). If a worksite slips by 6 months, the CPs shift to the following year — without this being a "non-decision". This is the most lenient reading, and it is generally that of the Finance Department.

3. The "accounting" reading (revenue prudence)

Since revenue is itself subject to uncertainty (local tax revenue, transfer taxes, State endowments), the City prudently over-budgets its expenditure to have a margin in case of revenue drops. This practice is broadly accepted by the regional audit chamber, within reasonable limits (Paris's overall gap of around 5-10% is one).

None of these three readings is exclusive. All three play simultaneously on any given line item. This is what makes the political reading of the administrative account so delicate: you cannot mechanically decide between "the City did not keep its promise" and "the City was prudently provisioning".

The special case of investment

If we group the figures by section (operating vs investment), one fact stands out: operating is executed almost at vote (rate 94-99%), while investment systematically under-executes (rate 72-84%).

This asymmetry is structural. Operating is rigid (salaries, contracts, social benefits — you spend what is owed), so drifts are rare and small. Investment is elastic (worksites that take 18-36 months, contracts that slip, State subsidies that do not arrive on time), so gaps are large and systematic.

This asymmetry is useful to keep in mind when reading an administrative account: an execution rate of 80-85% in investment is the norm across major French local authorities, not a Parisian exception. Reading an investment gap as a mechanical "non-delivery" of a promise abstracts away from that norm — though that does not mean the political reading should always be set aside, since it can be relevant on a case-by-case basis.

What these figures do not say

This analysis stops at what the open-public-data pipeline allows us to measure. It does not say:

• The precise reasons for each under-execution — that would require accessing the internal notes of the Finance Department (DAF) that justify CP shifts between years
• The evolution over time of a given line item — the analysis presents 6-year averages, which can mask trajectories (a line item with a growing over-execution deserves more attention than one with stable over-execution)
• The detail by operation — the reasoning is at the major-theme level, not at the individual worksite or contract level

What it does say, in one sentence: the City budget is not executed identically to the vote; some line items systematically drift upwards (Economy operating, Transport investment), others systematically downwards (Economy investment, Loans & Debt, Urban planning). Reading a voted budget as a strict promise is misleading. Reading an administrative account without referring to the original vote is too. The right tool is the systematic comparison of the two — something few observers do in the press.

Sources: open-public-data pipeline, file vote_vs_execute.json (join between editorial Primary Budget and Open Data Paris Administrative Account) · Scope: operations broken down by function (chapters 900-908 and 930-938), excluding non-broken-down operations (global debt, endowments) · 6 comparable fiscal years 2019-2024.