Vermont has the 34th healthiest economy in the United States in 2026, according to Quartz’s ranking of all 50 states. But there is a reason for that.
Remove the 10 points our ranking uses to penalize states for taxing residents and businesses, raise the weight on unemployment and poverty, and Vermont finishes 12th. Texas falls from 18th to 34th over the same change. Alaska drops out of the top 10 entirely.
That reweighting came from a Keynesian critique of our 2026 state economy ranking framework. To be precise, we asked an AI model to argue against the ranking as a proponent of Keynesian economics would and to propose how it would redistribute the 100 points.
The model produced the scheme below. The resulting ranks are real — the same 18 metrics, the same federal data, reweighted and recalculated — but no economist proposed them.
The objection
The criticism is simple, and it is the strongest single argument any of the six critiques made against this framework.
Our ranking scores taxation across 10 of 100 points, split between what residents pay and what businesses pay. In both cases, a higher burden lowers a state’s score. The framework has no corresponding metric for what the tax revenue buys. Put another way, nothing in the framework measures school quality, road condition, transit coverage, or public health capacity.
The consequence is that a low-tax state with failing schools outscores a high-tax state with good ones, and the ranking cannot distinguish between them. It measures the price of government while ignoring the product.
The Keynesian economic position is that taxes fund the public investment that makes a workforce productive in the first place, so scoring them as a pure cost does not describe a neutral measurement. It describes a thesis.
The second objection concerns inequality, which carries 4 points in our framework, split between a pre-tax Gini coefficient and the Supplemental Poverty Measure. The Keynesian argument is that this is far too low, because inequality suppresses demand. Lower-income households spend a larger share of what they earn, so concentrating income at the top reduces total consumption. On this view, inequality is not a fairness question appended to an economic ranking. It is an economic variable, and it belongs at 10 points or higher.
The reweighting
The pattern is consistent. Everything cut measures production, government cost, or people voting with their feet. Nearly everything raised measures what a household experiences: whether you can find work, what you earn, what housing costs, whether you are poor. The exception is research spending, which the scheme doubles.
Which states move
Vermont is the clearest case. It scores 1.02 of 5 on resident tax burden and zero of 5 on business tax burden, the worst business tax score in the country. Under our framework’s weighting, those two metrics cost Vermont nearly 9 points. Delete them and Vermont is left with what it does well, specifically, a labor market score of 17.07 of 25, fourth best in the nation, and low poverty. It rises 22 places on the strength of numbers it already had.
Alaska is the more interesting case, because it runs the opposite way. Alaska ranks 2nd largely on two metrics: first, the lowest resident tax burden in the country, a perfect 5 of 5, and, second, a state fiscal health score of 5.18 of 6 built on a $48,500 per-taxpayer surplus. The Keynesian scheme zeroes the first and halves the weight on the second. What remains is a labor market score of 10.4 out of 25 and the worst net migration rate in America. Alaska falls to 14th.
That is the whole argument in one state. Alaska’s position in our ranking depends heavily on the premise that lower taxes are better. Drop the premise, and the state looks ordinary.
Texas and Florida fall for a related reason. Both rank well in the Quartz ranking partly because of low taxes and strong in-migration, and both score poorly on the household metrics this scheme elevates. Texas earns 3.11 of 7 on poverty and 0.62 of 2 on the Supplemental Poverty Measure; Florida 3.95 and 0.31.
North Dakota finishes first anyway. It is the only state that holds its position under all six of the reweightings we tested, and it holds here because its strength is not concentrated in the metrics this scheme discards. It posts the best prime-age labor force participation in the country and the lowest housing cost burden, both of which the Keynesian scheme weights more heavily than we do.
What this critique would add
Four metrics it wanted and our data could not supply:
- U-6 underemployment rather than the headline unemployment rate, which excludes discouraged workers and involuntary part-time workers
- Median rather than mean wages, since the mean is pulled upward by high earners
- A public goods measure — school quality, infrastructure condition, public health capacity — to sit against the tax metrics rather than leaving only the cost side visible
- Intergenerational mobility, which measures whether a person born poor in a state is likely to escape poverty
The last of these is the gap all six AI critiques identified.
Where the argument is contested
Zeroing the tax metrics solves one problem by assuming another. The framework has no way to verify that tax revenue is well spent, and removing the cost side entirely assumes it is. A state can tax heavily and deliver little.
For example, a low-tax state can have better schools, while a high-tax state can have mediocre ones. A recent Vermont Public poll found that about half of respondents believed the state’s public schools were fair or poor despite relatively high taxes.
The supply-side response, which the companion piece in this series takes up, is that inequality measures dispersion rather than welfare. Under this weighting, a state that became uniformly poorer while compressing its income distribution would improve its score. That is a real property of the metric, not a caricature of it.
The Chicago position is narrower and harder to dismiss: income inequality substantially overstates consumption inequality, which is closer to what people actually experience. A state with many students and retirees shows elevated Gini figures for reasons that have nothing to do with economic dysfunction.
What it changes
This reweighting does not produce a more accurate ranking. It produces a different one, built on a different answer to a question the data cannot settle: whether a tax is a cost to be minimized or a purchase to be evaluated.
Our framework takes the first position by default, across 10 of 100 points, without ever arguing for it. That is the criticism worth taking seriously, and it stands whether or not you accept anything else in the Keynesian case.