The deadline everybody quotes is the wrong one for half the cities that quote it

Ask anybody in a Texas city hall when the tax rate has to be adopted and you will get September 30. It is on every calendar, it is in every checklist, and for a great many cities it is not the date that governs.

The one that actually binds arrives in the second half of August, it only applies to some cities, and which ones it applies to is not settled until the council decides how much money it wants.

What happens if you miss it

Start at the end, because the consequence here is unusual and it is the reason any of this is worth reading.

Most statutory deadlines in local government produce an argument. Somebody has to notice, somebody has to complain, and there is a process. This one does not. Under section 26.05, if the governing body has not adopted a rate by the deadline, the rate is set for it: the tax rate for that year becomes the lower of the no-new-revenue rate or the previous year’s rate. The governing body then has to ratify that rate within five days of it being established, which is a formality — the number is already decided.

No hearing. No appeal. No discretion.

In a year of rising appraised values the no-new-revenue rate is the lower of the two, sometimes substantially, and the difference between it and the rate the council intended is money that does not arrive. Not deferred — it is a different tax year now, and the levy for the one you missed is what it is.

That is the whole reason this is a deadline worth engineering around rather than remembering.

The two rates, and what each is for

A taxing unit calculates two rates once the certified appraisal roll is in hand, and they do different jobs. Conflating them is the most common misunderstanding in the whole subject.

The no-new-revenue rate is a comparison, not a limit. It is the rate that would raise roughly the same revenue this year, from the properties taxed in both years, as last year did — prior year taxes divided by current year taxable value, times one hundred to express it per $100 of value. Its purpose is to let a taxpayer see whether their bill went up because the rate rose or because their value did. A unit may adopt above it freely.

The voter-approval rate is the ceiling. It is the highest rate a unit may adopt without putting the question to voters, and it is assembled rather than calculated in one step:

no-new-revenue M&O rate × 1.035, plus the current debt rate, plus the unused increment rate

Three components, each with its own way of going wrong.

The multiplier is not 3.5 percent for everybody

The 1.035 above — the extra three and a half percent for operations — applies to most units and not to all.

Special taxing units multiply by 1.08 instead. That category covers junior college districts, hospital districts, and any taxing unit other than a school district whose maintenance and operations rate is 2.5 cents or less per $100 of taxable value. The last of those is worth reading twice, because it is a test a unit can pass or fail depending on its own rate, which means it can change between years without anybody deciding anything.

Smaller units have a second ceiling. The de minimis rate exists to give units that are small enough for a single expense to distort the arithmetic — including cities under 30,000 population — somewhere to go. It is the no-new-revenue M&O rate, plus the rate that would raise $500,000 against current total value, plus the current debt rate.

The interaction is the useful part. A qualifying unit that adopts above its voter-approval rate but at or below its de minimis rate does not trigger an automatic election. Voters may petition for one under section 26.075, which is a materially different position from having to schedule one. For a small city, the practical question in a difficult budget year is often not “can we exceed voter-approval” but “can we stay under de minimis”, and those are different numbers.

The unused increment is restraint, banked

The third component is the one most often left at zero because nobody computed it.

The unused increment rate is revenue a unit could have raised in the previous three years and chose not to: the sum of those forgone amounts divided by current taxable value. It cannot be negative. A unit that has adopted exactly at voter-approval every year has nothing banked. A unit that has held rates down has been accruing headroom, and it is entitled to use it.

The reason this gets missed is that computing it requires knowing, for each of the last three years, both what was adopted and what the voter-approval rate was — and then adjusting for any increment already applied. That is not a number sitting in the general ledger. It is a small historical reconstruction, done annually, usually by the same person from the same spreadsheet, and it is the component most likely to be quietly carried forward wrong.

The calendar, as a chain rather than a list

Every published version of this is a list of dates. It is more usefully understood as a chain, because each link determines when the next one can happen and a slip anywhere compresses everything after it.

1. By July 25, the chief appraiser certifies the appraisal roll. This is the event the rest hangs from.
2. Rates are calculated from that roll and published, including prominently on the unit’s website.
3. Notice is published — a quarter page at minimum in a standard or tabloid newspaper, with a headline in 24-point type or larger, and specifically not in the legal notices or classifieds section. The website posting must stay up continuously for at least seven days before the hearing and at least seven days before the vote.
4. The hearing is held, with a quorum of the governing body present.
5. The rate is adopted, before the later of September 30 or the 60th day after the certified roll arrived.

Two things about that last line that the reference material states and does not emphasize.

It is the later of the two. If the roll arrives on time in late July, the sixtieth day lands in late September and September 30 governs. But if certification slips — and it does — the deadline moves out with it rather than expiring. A city whose roll arrives in September has more time than the calendar on the wall says, and a finance office that does not know this loses days it actually had.

Sixty days is measured from receipt, not from July 25. Which means the date is different for different units in the same county, and it is not a date anybody can print in advance.

The trap: deciding to exceed the rate makes your deadline earlier

Here is the part that catches councils, and it is the reason for the title.

If a unit adopts a rate above voter-approval, an election is required, and elections have their own clock running backwards from the November uniform date:

• the rate must be adopted no later than the 71st day before the election date, and
• the election must be ordered no later than the 78th day before it.

Both of those fall in the second half of August. Roughly five weeks before the September 30 date everyone has in their head.

So the effective deadline depends on a decision that is usually still being argued when the deadline passes. A council working comfortably toward September 30, which reaches the conclusion in early September that the numbers do not work without exceeding voter-approval, has already lost the option for that year. Not because anybody was slow — because the deadline for the path they chose expired while they were choosing it.

The practical consequence is that the decision about whether you might exceed the rate has to be made much earlier than the decision about what the rate will be. That is an unnatural order to work in, and it is why this one is worth flagging on a calendar in July rather than discovering in September.

And if the election is held and the voters say no, the rate is limited to voter-approval for that year. The election is the ask; it is not a formality.

Why this breaks, given that none of it is hard

Nothing above is difficult. It is arithmetic with published formulas and published dates. It goes wrong anyway, and always in the same four ways.

Every input moves. Certified values are revised. Debt schedules change when something refunds. New property value and lost property levy are both derived figures. A rate computed in July from numbers that shifted in August is wrong in a way that produces no error message — it just produces a number, confidently, and the number is the one that gets published.

The components come from different places. The M&O side is the budget, the debt rate is the debt schedule, the unused increment is three years of history, and the values are the appraisal district’s. Four sources, four owners, one spreadsheet holding the join.

The deadline is conditional on a decision. As above. A calendar cannot hold “September 30, unless we exceed voter-approval, in which case late August” as a single date, so it holds September 30 and somebody is supposed to remember the rest.

It happens once a year. Whatever was learned last August is a year stale by the time it is needed, and the person who learned it may have moved on. Annual tasks are the ones institutional memory is worst at, because nobody does them often enough to get fast and everybody does them long enough after the last time to have forgotten the exceptions.

What is worth building, and what is not

Not a tax rate calculator. The Comptroller publishes worksheets, most appraisal districts provide the calculation, and a vendor’s black box that produces a rate nobody can trace is worse than a spreadsheet somebody understands.

What is worth having is narrower and unglamorous.

The deadline computed rather than typed — from the actual date the certified roll was received, branching on whether the intended rate exceeds voter-approval, so the calendar shows the date that governs rather than the date that usually governs.

The components sourced and dated, so it is visible that the debt rate came from the debt schedule as of a particular day and the values came from the appraisal district as of another. Most of the failures here are stale inputs, and a figure that carries the age of what produced it is a figure somebody can check.

The unused increment held as history rather than recomputed from memory — three years of adopted rates and voter-approval rates, kept because they were kept, not reconstructed each August.

A published number that can be traced backwards on demand. When somebody at the podium asks how the voter-approval rate was arrived at, the useful answer is not the rate. It is this M&O rate from this budget, times this multiplier because we are or are not a special taxing unit, plus this debt rate from this schedule, plus this increment from these three years.

That is the same discipline as any other reconciliation, and it is the one thing that reliably converts an annual scramble into a Tuesday.