Six Questions for a Billion-Dollar Promise

A Skeptic’s Guide to Economic Impact Claims

Six questions anyone can ask about the economic promises attached to a big development — and what I learned building the model that produced them.

When a large project comes to your county, it arrives with a number. Billions in economic impact. Thousands of jobs. Tens of millions in new revenue. That number is almost always paid for by the people who want the project approved, calculated using software whose workings are not public, and presented to a council as settled fact. Residents get to react to it. They almost never get to check it.

Last year I built an economic model of Montgomery County, Maryland, from public data. It estimates what a given amount of spending actually does to the local economy. Building it taught me something more useful than any of its own numbers: the specific places where impact claims go wrong, and how to spot them without needing a model of your own.

Here are the questions. The rest of this post explains where each one came from.

Six questions for any impact report

1. Where does this multiplier come from? Was it worked out for a place that looks like ours?
2. Is the answer a single number or a range? If a single number, how can they be that precise?
3. Which assumptions did they test, and what happened to the headline when they changed?
4. What is inside the big impact figure — and how much of each piece is actually bought here?
5. Show the revenue year by year, not just year one. Does it hold up, or does it fall away?
6. How much of the wage money goes to people who live here, and how much leaves with commuters?

None of these need a model. They need the report’s authors to answer in public, on the record, before a vote.

What a multiplier is, and why the local one matters

Every impact claim rests on a multiplier, and the idea behind it is simple. A construction firm gets paid and spends part of that with local suppliers. Those suppliers pay their staff. Those workers buy groceries and haircuts nearby. One dollar of spending sets off a chain of further spending, and the multiplier is how big that whole chain turns out to be.

The catch is that the chain is only as long as the local economy allows. If a county makes its own steel and equipment, money goes round several more times before it leaves. If it buys those things from elsewhere, the chain is short. Montgomery County has almost no heavy manufacturing, so its chains run short — and a multiplier borrowed from a state average, or from a region with factories, will promise more than the county will actually get.

Nobody had published one for Montgomery County, so I built it from federal economic data anyone can download. One rule governs every result: the answer is always a range, never a single number. Nothing here can be measured exactly, and a single figure claims a precision the data cannot support.

Question one

Where does the multiplier come from?

A published county presentation puts the multiplier for housing development at 1.7 — every dollar spent generating $1.70 of local activity. My range for the same kind of spending runs from $1.29 to $1.63. The quoted figure sits at or above the top of it.

I could have adjusted the model until the two agreed. I did not, because the gap has a real cause: the county has no local manufacturing base to supply its own construction industry, and a multiplier that assumes otherwise runs high. So when you are handed a multiplier, ask which place it was worked out for. If the answer is a state average or a national default, the local promise is probably too big — by an amount nobody has bothered to work out.

Question two

Which assumption is doing the work?

Some county employment data is hidden, because publishing it would identify individual employers. Forty-one percent of it, in fact — the sort of statistic a critic reaches for immediately. I tested a rougher way of filling those gaps. It changed the answer by a hundredth of one percent.

What did matter was something I had barely thought about: whether to count the self-employed. Standard employment data leaves them out by design, and Montgomery County has a great many of them. Putting them back in moves the construction multiplier by about 11 percent.

The lesson travels. The alarming-sounding gap was harmless; the quiet choice was decisive. When a report lists its own limitations, that list is not a ranking of what actually matters. Ask which assumptions they varied, and what happened to the headline when they did.

Question three

What is inside the big number?

To test the model on something concrete, I ran a hypothetical data centre campus, using costs published for a comparable project in a neighbouring county. Two decisions in that run matter more than any of the results.

The first is what counts as construction. The published figures gave two different costs per square foot, and the larger was twelve times the smaller, because it included the servers. Servers are bought from manufacturers far away. They are not local building work, and running them through a construction multiplier would inflate the answer enormously. So I used the smaller figure.

The second is electricity, which sits on its own line and is never added to the operations total. At this size, the facility’s annual power bill is larger than everything else it spends locally in a year, combined. And the county generates almost no electricity of its own, so most of that money leaves immediately. The largest recurring local expense is the one that stays here least. Rolled into a single impact figure, it would read as a benefit.

Question four

One year, or all ten?

During construction the county collects no property tax at all — there is nothing built to tax yet. Once the facility opens, property tax starts high, around $53 million, then drops sharply. Most of the value is computer equipment, which the state writes down quickly for tax purposes. The building itself holds its value, which is why the line flattens out instead of falling to nothing. The energy tax runs flat throughout, because it taxes power used rather than property owned.

Quote year one and you have quoted the best year of the ten. Quote a ten-year total and you have hidden where the money actually falls. Ask for the whole path. Then ask two follow-ups: are these figures adjusted for inflation, and are they on top of whatever the site already brings in today? Usually the answer to both is no.

Question five

Who actually gets the wages?

Montgomery County’s local income tax follows where you live, not where you work. A job here held by someone from Virginia or Prince George’s County generates no local income tax at all.

That turns out to be a large hole. Between 37 and 52 percent of the wages created by construction leave the county untaxed, depending on how the self-employed are counted. For the data centre’s ongoing operations it is closer to 45 percent. A jobs number is not a local income number, and a local income number is not a local tax base. Something is lost at every step, and impact reports rarely show the losses.

What this does not tell you

On the county’s books, the hypothetical facility pays its way. Revenue comfortably exceeds the cost of serving it in every one of the ten years, and it would take an enormous error in my cost estimate to change that. Nothing here supports an argument that a data centre loses the county money, and I am not going to pretend otherwise.

What the work does support is narrower and much harder to dismiss: the economic benefit is smaller and leakier than the headline promises, the revenue falls away rather than holding, and the biggest local expense mostly leaves the county. Everything else — electricity bills, emissions, water, land, the strain on the grid — is a separate argument resting on separate evidence, and this model has nothing to say about any of it. Stretching a financial model to settle an environmental question is exactly what makes the industry’s numbers untrustworthy. It would make ours untrustworthy too.

A note on the example. The data centre here is hypothetical. Its costs come from a published study of a comparable project in a neighbouring county, and no Montgomery County project is referred to or implied. The county paused approvals for large data centres for 18 months shortly before I did this work. What I have described is how such a facility’s finances would look, offered so readers can weigh it — not an argument for or against any project or policy.

This is a personal project. I am a member of Mothers Out Front, but the views here are my own and not those of any organisation I belong to or have worked for. Every source is public: federal economic and employment statistics, the county’s own tax rules, and state and federal energy data. Full method and citations available on request.