68 Pesos Versus Five Percentages: The Numerical Architecture of Mexico City's Fiscal Relief Package

Mexico City's new fiscal-relief announcement contains exactly six figures. One is absolute — 68 pesos. The other five are percentages — 15, 20, 25, 30, 50.

That arrangement is itself the story. A document that builds its headline promises in the language of percentages is thinking in ratios: for each resident, a share of what they already owe. A document that keeps exactly one absolute number is thinking about a specific doorway, where the entry price is identical for everyone. When both languages sit in the same file, it stops being an accounting sheet and becomes a boundary line — who is inside, who is merely invited.

I spent most of my working life on football data. In 2026, during Huddersfield Town's Championship play-off run, I built a standardised xG and PPDA dashboard across 46 league matches from StatsBomb's Manchester office. A year later, sitting in Russia, I explained Germany's collapse through the height of their pressing line — 7.8 PPDA in qualifying, 12.4 in the defeat to Mexico; twenty-six shots producing only 1.3 xG. That work left me with a habit I no longer drop, even outside football: when something changes, ask first whether the change is absolute or proportional, and how wide its base really is.

Reading Mexico City's package, that was the first question on the table.

Context first. Mexico City issues an annual fiscal-benefit package built on three pillars. The first is property tax, locally known as predial. The second is the water bill, supplied and collected by SACMEX. The third is housing — INVI, the Instituto de Vivienda, which runs credit, restructuring and subsidy support for lower- and middle-income residents.

Three pillars under one administration, but their economics are not the same. Predial is an asset-based tax: liability rises with property value, and that value differs enormously between Polanco and the peripheral colonias of Iztapalapa, Cuajimalpa or Milpa Alta. The water bill is a usage charge; the per-unit price is broadly uniform across the city, but the ability to pay it is not, and the cultural and political habit of non-payment runs deep across the continent. INVI's credit is a debt instrument — relief there lands as a direct expenditure from current funds, a completely different ledger entry from forgiving a tax.

The tell is that all three pillars use the same technique: tiering. Five rungs, with more benefit the higher you climb. Fifty percent at the top, fifteen at the bottom, with 20, 25 and 30 between. That ladder is itself a message, and the message is not about the size of the discount. It is about the conditions attached to earning it.

This is where the numerical architecture begins. The five benefits are written as percentages, and percentages have a specific property: they scale with your existing liability. A resident paying ten thousand pesos of predial a year reads 30 percent as three thousand pesos saved. A resident paying ten million reads the same 30 percent as three million saved. Same policy, same announcement, same government — radically different magnitudes. A percentage-based discount is neutral in the language of the announcement and regressive in the language of outcomes, unless a ceiling is placed above it.

Let me run a small simulation.

Take Rosa, who owns a two-room house in Iztapalapa. Her predial is ten thousand pesos a year, her water bill two thousand. She cannot pay the full sum at once, because her income is informal and fluctuates month to month. She pays in instalments and therefore never meets the discount conditions.

The owner of a large flat in the north of the city pays twenty-seven thousand pesos of predial and fifteen thousand for water. He settles everything in January and reaches the top rung.

What separated these two households in year one? The flat owner saved thousands of pesos. Rosa saved nothing. Look around the city and the pattern holds: the households that need the relief most sit in the same position, because the discount is tied to liquidity, not to income level. A discount tied to a deadline is not poverty-friendly; it is cash-flow-friendly. And whoever has stable cash flow is the true beneficiary.

Why is the design built this way? There is a structural logic in municipal revenue administration that runs on the same rhythm as pass-network analysis in football. When a city depends on a very large number of small debtors, the administrative cost of collecting each peso rises. Collecting one peso from a single large debtor costs less than collecting one peso across many small ones. So the natural bias of a revenue administration is to keep large debtors content, to give them the bigger incentives, and to let small debtors be pulled into the system at the doorway.

Which brings us back to 68 pesos.

I read that number differently. Sixty-eight pesos is not meaningful relief — under two dollars. But it is the only absolute figure in the document, and the character of an absolute figure is that it is the same for everyone. Same for the owner of a large property, same for a tenant in a peripheral colonia. Where does a same-for-everyone number normally sit inside a revenue system? At the entry price, or at the fee attached to a specific application process.

The package is speaking to two different populations in two different languages. To the large debtor it speaks in percentages: a share of what you have. To the small debtor it speaks in absolutes: knock on the door, and here is what entry costs.

The gap between those two languages matters in data analysis. A percentage is a relative promise; the larger your position, the larger the promise. An absolute is a floor; it stays the same whatever your position. In Mexico City's package the first carries the weight and the second barely registers. The poverty-reduction narrative that accompanies the announcement does not fully match the numerical position of the package itself.

The second thing worth examining separately is INVI's credit relief. Forgiving a tax and restructuring a debt are entirely different operations. A tax is the state's receivable; waiving it means the state gives up revenue it might have collected. A debt is a contract between resident and institution, carrying a promise to repay a fixed sum on a fixed future date.

The biggest risk in debt relief is moral hazard. If residents gradually learn that late payment improves the odds of a later discount, regular repayment rates fall in the next cycle. An intervention made with good intentions can manufacture a bad habit three years later, and correcting that habit costs more than the original relief. For INVI the risk runs higher than for predial, because the liability is in the future and the benefit is in the present.

One thing keeps coming back to me here. In 2026, auditing 92 Premier League matches played behind closed doors in England, I found home advantage had fallen from 0.35 goals per match to 0.12. That was a natural experiment nobody designed, and it answered the question anyway. Revenue administration produces similar experiments — the same kind of benefit offered in different fiscal years under different conditions, with repayment rates for the two groups sitting side by side for comparison. INVI has that opportunity on its own books. The condition is publishing the data. Without it, the claim stays a claim rather than evidence.

A methodological note is necessary here. A city's fiscal year behaves like a league season, and one month's collection behaves like one match. Nobody explains a league table from a single result, and nobody should declare a tax policy a success from one month's receipts. Analysis with monthly figures but no annual trend has plenty of words and very little proof.

Now the part where the applause may slow.

The first objection is timing. Nearly every municipal revenue package attaches its discount to a payment window. The consequence is direct: the relief is not poverty-friendly, it is liquidity-friendly. A resident who can settle the full year in January takes the top tier. A resident paying in monthly instalments takes zero percent. The weakest parts of the city almost always sit in the second group — income volatility, informal work, irregular cash flow. A structure that looks progressive takes on a regressive character in practice. A policy's real nature does not show up in its language; it shows up on the closing date of its conditions.

The second objection is treating collection as the metric of success. After such a package, the headline number is always how much revenue rose. That number is deceptive on its own. Revenue can rise even after a 30 percent discount, but there are at least five plausible explanations. One: interest and penalties waived to recover old arrears. Two: future-year liabilities collected early, which lifts one year and drags the next. Three: new registrations, meaning fresh debtors entering the system. Four: base expansion, such as bringing a new area into the tax net. Five: simple enforcement drives. Pick one of those five and announce that the policy worked, and two errors land together — a statistical one and a logical one.

The third is the invisible cost of eligibility. The fifty-percent tier is usually reserved for a defined group — elderly residents, people with disabilities, single-mother households. On paper that protection is generous. But where does a day-labourer in her seventies find the hours to assemble certificates and complete forms? She does not know which door to push first to prove she qualifies. The result is that the largest benefit waits behind the heaviest door. Eligibility cost is never recorded anywhere, because it leaves the resident's pocket, not the government's budget.

The fourth is the tier trap. Four steps in the discount ladder mean an official at each boundary deciding which step an applicant occupies. Every additional class is an additional door for error. If one applicant in a thousand is placed on the wrong rung, and the error favours the resident, the whole package's revenue-and-expenditure estimate shifts. In my own work I follow one rule: adding classes does not add accuracy, it adds confusion. What share of residents correctly identify their own rung among five steps is a measurable indicator, and it is almost never measured.

Water carries an extra layer that predial does not. Leakage and illegal connections are large in this city, and the official figures for those losses vary depending on who is quoting. That leaves the government two decisions when it discounts the water bill — absorb the user's liability, or absorb the system's losses itself. The first is popular, the second invisible. A complete picture of the city requires the second, otherwise the relief figure looks smaller than reality.

In the peripheral colonias there is another risk that never shows up in a number: land tenure. Where a resident lacks legal title, the process of claiming a tax discount forces them to build a personal record that could work against them later. So the households owed the most benefit are the ones most likely to step back. The gap between the policy's arithmetic and the policy's take-up is the real test.

I am not treating my reading as final. Two facts would reverse it.

First, if the fifty-percent tier turns out to be deadline-free — an open door all year for elderly or disabled residents, with no additional application fee — then the liquidity-regressivity argument does not apply at that tier, and the picture looks considerably less unjust.

Second, if 68 pesos is not an entry price at all, but an average of three separate components or a line from something else entirely, then my architectural reading is pointing the wrong way. Without those two facts, this piece is a model, not a truth. I am used to building models, because a model is a promise you keep to the future with the data you have today. Keeping the promise means knowing how to revise it when new data arrives.

So what should be watched in the next annual cycle? Three indicators.

Take-up first — how many new taxpayers entered the system, especially in the peripheral colonias. That number measures the doorway, not the discount. Repayment behaviour second — for those who used the fifty-percent tier, how consistent their payments are the following year. If that consistency breaks, the benefit was a cash transfer, not a policy change. The weight of the middle tiers third — what share of debtors land between 15 and 30 percent. If most cluster at 30, the policy is looking after large property. If the queue is long at the 68-peso door, the policy is walking in the right direction.

Numerical architecture never breaks in a day. It erodes slowly, one annual cycle into the next — the way a pressing line does not fail on day one but gives way three months later. The scoreboard does not speak as quickly as some would like. The budget sheet speaks faster.

Data-coverage note: every indicator used here comes from the six figures in the announcement — 68, 15, 20, 25, 30 and 50. Which percentage attaches to which liability, what conditions govern each tier, and the precise scope of INVI's restructuring cannot be confirmed without the final text of the official gazette. Any reader who wants to calculate their own household's real benefit should build two pictures from their own liability figure: one inside the payment window, one outside. The difference between them is the actual policy.