How much do state-level business conditions matter?

Georgeanne M. Artz, Kevin D. Duncan, Arthur P. Hall and Peter F. Orazem have a new paper on that topic.  The answer seems to be that state-level business conditions do not matter so much, at least not in their currently measurable forms:

This study submits 11 business climate indexes to tests of their ability to predict relative economic performance on either side of state borders. Our results show that most business climate indexes have no ability to predict relative economic growth regardless of how growth is measured. Some are negatively correlated with relative growth. Many are better at reporting past growth than at predicting the future. In the end, the most predictive business climate index is the Grant Thornton Index which was discontinued in 1989.

While I don’t think this is the final word on the topic, the burden of proof clearly lies on the side which claims they do matter a good deal.


Behind a pay wall. So it hard to ask the right questions.

Did they include Puerto Rico in their study? Hawaii? Alaska?

Eleven states is a very small sample. Any reason why it is so small? The obvious question is what distinguishes these states. Are they all North-Eastern? Did they cover the whole country?

Anyone who thinks that the state-level does not matter should invest in Puerto Rico.

They are looking at relative economic performance of (I think) counties on state borders. What would be the relevant border for PR? I don't think they discount the possibility of significant regional differences.

Here is an accessible paper by one of the same authors that appears to be based on the same or a very similar analysis. NB: Koch money funded it.

Good points, but I think the overarching problem is that these are complicated and dynamic questions not easily solvable with simple statistical comparisons. I don't think this study can really confirm or disconfirm my priors, because of the complexity, so I'm free enough of having to defend my ideology to notice the following:

1. I represent a lot of businesses in various transactional matters, and federal regulations are always, always, always more pressing than state matters, although that may seem like an obvious point. It's not like you can just break state laws, but their agencies are the JV team compared to the Feds. They don't usually have the budget or sophistication to make nearly as big a nuisance of themselves as the Feds do. Also, look at the state codes and regulations -- they attempt to regulate much less than the Feds (and are in almost every area that matters are preempted by federal law anyway), and their penalties for noncompliance are often comparatively minor. So the lack of an effect at the state level could be explained by the states simply not mattering compared to the Feds, but to draw the implied conclusion that friendliness to business doesn't matter would be unwarranted.

2. When I look at state regulations, states like California and New York often appear to be able to exert more regulatory pressure than other states simply because those states are extremely desirable places to do business for purely economic reasons that exist prior to and/or independently of the regulatory environment. Does California's 10% tax rate cause prosperity in California, or is Sacramento able to stick people with huge tax bills because they know California is a desirable enough place to live that people will pay it? Is the New York Department of Financial Services the reason New York is the financial capital of the U.S., or are they able to exert disproportionate leverage over financial institutions because that's a cost firms are willing to put up with, since no other city even compares in terms of density of financial networks and knowledge? As a place becomes more desirable to do business in due to industry concentration, the host jurisdictions gain more leverage. Want to succeed in finance? Be prepared to move to New York. Film? Good luck getting a job and meeting the right people outside of Hollywood. Call it the mafia effect.

"but I think the overarching problem is that these are complicated and dynamic questions not easily solvable with simple statistical comparisons. I don’t think this study can really confirm or disconfirm my priors"

Methodologically, this paper would not pass an undergraduate level econometric class. You can rest assured of that.

Journal of Regional Science? Meh...

I think I've made my point.

Didn't confirm your priors?

Your priors are my posterior.

As a PhD student, I would have thought that you would be able to evaluate research on the basis of its own merits, and not write it off for the mere fact of who published it. A lot of stuff comes out in "lesser" journals or outside of the journal system altogether before it becomes clear that something is important.

I'm not saying that the research is necessarily good (can't get beyond the paywall), but that your method of evaluating research is piss poor. Evaluate on the basis of methodologies and whether imbued conclusions are drawn correctly with respect to context and methods, not where it is published.

I can't evaluate something I don't have access to. I'll see if I can find the paper through my university library. But 9 times out of 10 you can tell the quality by looking at where it was published.

Reading the version of the paper posted above, there isn't a methodology to critique. They're just running a regression, with apparently not even any controls, never-mind anything more sophisticated than that. This is in essence no better than a bunch of correlations. They say they're looking at cross border counties, but they're not I'm not sure what the heck they're doing if they're not matching.

It's essentially totally crap. As would be expected from such a paper.

The methodology is so bad, Table 6 is all one needs to see. Wait a minute!...they simply looked at the raw figures of economic performance on border counties and compared it to the gap in the evidence? Holy cow! That's not even undergraduate level of analysis. Yes, the border counties is where one would expect to see the bigger difference.'d be able to see this difference only if there was some external shock, or in a dynamic setting. Not through a simple cross-sectional table (not even a regression) I.e., something has to change...exogenously...across these states for you to see an effect on growth rates. If there isn't such a change, then how do you know people aren't adjusting locations and activities 5 years prior to your measurement since it's all endogenous?

So the methodology is totally crap. But even theoretically, their analysis is crap. These indexes don't change much over time, hence how exactly do the authors think they can...predict...economic GROWTH? Things that don't change much over time, aren't very good at predicting change.

Doesn't basic economic growth theory predict that business environment is correlated with past growth as it detrrmines gdp level not growth per se?

I agree, would be interesting if the eleven predictors told GDP per capita, adjusted for luck, I.e.. natural resources, proximity to beltway etc.

Of course, a favorable "business climate" is always defined as low taxes (low for business and those with high incomes), no or weak unions, little regulation, abundant low-cost labor, and business-friendly state and local governments. Those business climate indexes are honey for a race to the bottom, where every state becomes Texas and Florida. "Many are better at reporting past growth than at predicting the future." What does Texas do when the oil is gone? What does Florida do when the retirees (and the beaches) are gone?

So all states should be like New York and California eh?

I see left-of-center folk on Twitter calling, every day or two, for federal investment in education and infrastructure. As a Californian, I feel like I gave at tax time. We are high wealth, high tax, but high education and infrastructure.

Maybe two things. Maybe other states should pay for things they need. And maybe education/infrastructure calls should be more local politics.

"So all states should be like New York and California eh? "

I'm not sure about that but there is a problem with the observation we constantly hear along the lines of "Look at all the businesses that are moving from CA/NY to Texas (or insert your favorite Red State here)"

Well how did all those businesses start out in NY and CA? Was there some mythical time those states had super low taxes and regulations? When exactly was that? 1950? Roll the clock back ten, twenty, thirty even forty years. Even in 1980 those states had reputations for being more liberal rather than less. So how did those states ever get all those businesses for states like Arizona, Texas, etc to steal?

The model doesn't work but one model that might would be the scavenger state. States that can offer super low costs (not just taxes but low labor costs, low real estate costs, low cost of living etc.) would attract businesses that are declining (hence trying to find any savings they can to keep from becoming totally unviable) or in established and stagnant markets (so the only way to grow profits is by chasing lower costs rather than increased sales). The 'business friendly' state then does not see more actual economic growth than the 'unfriendly' states despite endless sound bites about "XYC Inc. Moved their Headquarters with x,xxx jobs to Red State" because they are capturing declining businesses rather than growing ones. They have to 'steal' so many companies from other states every year just to make up for the decline of previous 'stolen companies'.

I mean, I guess you can make up whatever you want.

Powerful counterargument. I'm convinced.

I can make up whatever I want, but it has to be a story that fits the facts.

The counter argument here I'm seeing is "business friendly" policies boost state economic growth but that cannot be measured because we can't measure economic growth....or we can only measure differences in growth in exceptional geographic circumstances like in counties of 'friendly' states that are touching the counties of 'unfriendly' states. I'll call that the "Phantom Fireworks Model". In PA fireworks are legal to sell to out of state residents while they are illegal in NJ and NY so as soon as you cross into PA you see huge billboards for "Phantom Fireworks". Yet despite selling lots of fireworks, it doesn't seem like the border counties of PA are doing better because of these more friendly regulations.

When defined as compared to pure communism, all variables defined as a "good business climate" are almost certainly going to show positive results. However, presumably there is some sort of happy middle. Taken too far, "improving the business climate" may ultimately hamstring the government from being able to engage in certain interventions which are critical for long-term growth. Somalia, for example, would probably be well-served by a "better business environment" which included a significantly expanded state capacity.

This is an interesting result, just taken at the surface level. Business Climate Indexes were things that people thought would shape growth. Thy don't, except for the ones that are Negatively Correlated with growth.

So people either didn't know, or got it backwards.

Or, its such a bad paper, its laughably bad.

This should be "Do state business climate indexes matter" not "Do state business climates matter."

Business climate has to matter. The issue is are we measuring it properly.

Climate has to matter, but it doesn't have to be a top factor. It can be buried, for instance if the side of the border with more business more easily supports tax - without slowing down.

"The issue is are we measuring it properly."

Blasphemy ! Economic statistics are sacred.

Casting doubt on some economic statistics puts all such statistics in question.

Essentially this is a study of the effectiveness of a scoring mechanisms as opposed to the effects of states.

No matter what, the baseline is federal regulation and taxation. Unless you are in one of the relatively lightly regulated high growth sectors where marginal differences in state jurisdictions have less an effect than accessibility to ski hills and beaches for attracting talent, the choices are not Illinois or Indiana, but rather China or Mexico.

Things like lower cost of living, housing or access to resources or transportation won't be that much different across state lines in adjacent cities. I suspect you see larger differences across county lines within large cities than you would across state lines probably due to crime more than anything else.

And any differences that you would see would probably be limited to sectors. The migration between Texas and California would be two way; low value manufacturing and service business one way, high value high tech graduates from Texas schools the other way. The latter don't use U hauls.

But the scoring systems were based on the same foundation as growth policies.

Thus growth policies are more suspect.

Yeah, it would seem to make more sense to check growth rates across all the states and evaluate which factors appear to be most conducive to the growth of business/production, not to use some ideologically-inclined "business environment indicator" and then draw conclusions from that.

Do the authors consider the hypothesis that policies that target the "business climate" might be caused by perceived negative performance? That is, suppose poor performance leads policymakers to implement reforms targeting the "business climate." We should then expect to see places with recently-improved business climate indexes have negative performance.

Analogy: Suppose you measure the temperature of my car's interior and look at the heater setting. I would guess that having a high heater setting is actually associated with a colder car, since I only turn on the heater when it's cold, and my commute is pretty short; I only get a few minutes of warm car.

You assume the heater control is connected to the heater ?


When you look at relative economic performance across states borders on a map what you immediately discern is that states in the different regions tend to have similar growth patterns. See: The study tells us less about the impact of state business regulation than about the similarities of states in various regions, eg Texas and Oklahoma have the highest rates of growth. But if TC is arguing that government is irrelevant to economic growth and state intervention should be avoided, then sure, why not.

I think he is saying that there is little evidence that government is relevant. You added that "state intervention should be avoided". The first can be true without implying the second.

"state-level business conditions do not matter so much"

The mass migration of Chicago's Black middle class out of state, primarily to to big cities in the South, says otherwise.

Most businesses are stranded. You can't move your chain of pizza restaurants across the state line to a part of the metro area that won't support so many new eateries, no matter how good the business climate there is. If you are a dentist with a 20 year old practice its hard to pick up stakes and move.

Track the things and people that are mobile, then tell me if local business conditions matter.

Business conditions caused that migration? Yes, of course, you keen observer.

But nothing says you couldn't reduce your chain of pizza places or have them open for fewer hours with fewer workers or that their receipts couldn't stagnate or decline over time.

In my experience in VT and NH, any place there are towns on both sides of the Connecticut River, the town on the NH side will have seemingly double the number of retail outlets per capita. I have not scientifically studied it.

'the burden of proof clearly lies on the side which claims they do matter a good deal'

So, something along the lines of a freedom index is just more feel good mood affiliation? I'm sure that someone at the Mercatus Center, with its fierce tradition of independent scholarship will be getting to the bottom of something like this any day now -

The oil fracking industry of California has certainly suffered thanks to state-level regulations while the recreational marijuana business in Colorado has boomed.

State level regulations do matter quite a lot for specific industries.

One additional comment: as others have already said here, this is little more than an analysis of the indexes used, rather than state policies (and its not even a good analysis, but a horrendously bad one)

But to reinforce that point, people don't look at...indexes...when they make business decisions. They look at tax rates, at specific regulations pertaining to specific issues, at overall costs of doing business etc etc. I.e., people look at the...individual components of the index...rather than the index as a whole.

Do tax rates matter? Yes. There's probably lots of papers (in good journals instead of this crap) looking at that. Taxes aggregated with tobacco and alcohol law and maternity leave law an UI law? Uhh...probably not.

Sorry but I don't think this follows. If the elements of the index individually make for a 'positive climate' then you should see results. What might make sense is if you had elements of the index that were actually positives. For example, you might cite maternity leave laws as a negative since that's regulation. But perhaps it is a plus indicating a higher quality workforce or a workforce that has a healthier work/life balance. Tax rates have to consider how taxes are spent. A state with low taxes but crumbling roads may not make it easier for a business in that state. Or taxes might simply be chasing an independent variable. For example, Detroit might be raising taxes because after losing major auto industry companies, a smaller tax base has to cover fixed costs. A boom in oil prices may fill up states with drilling and fracking industries, likewise.

But you've just made my point on how it does follow.

An index which combines 15 different individual indicators, many of which may not be important to business, many of which may be, many of which may be neutral etc...doesn't tell us as to whether more business friendly policies are better or worst. The index hides all that information by aggregating.

You're absolutely right that the tax policy in itself may not be the issue...and other issues like infrastructure, may be at play.

This paper however does not even make the most rudimentary attempts at getting to the effect of the business climate in the states. It controls for nothing, and simply correlations...between the index and economic growth. There are so many problems with this, least of which that these simple correlations tell us nothing.

And of course, there are so many taxes involved, that again you make my point on why a simple index is irrelevant.

In Texas you may not pay high taxes on some things, but you may pay higher property taxes in order to provide the infrastructure. Which means location...within Texas...matters more than location between Texas and Oklahoma. These indexes aren't designed to capture business's exposure to different taxes, which is why these indexes are pointless anyway.

But you've also made the point that this is a chicken or the egg type of argument: do you have lower taxes, and thus attract more businesses, or do you have more profitable businesses, and thus have lower taxes? Do lower taxes attract lower performing businesses, or do they attract higher performing businesses?

All of this is what's actually important here, and which a simple correlation isn't going to tell you. Especially a cross-sectional analysis like the one done here.

Clearly we have a problem here of a small sample size but possibly also a problem of understanding exactly what question we are trying to ask.

Different states will likely enjoy different rates of economic growth due to factors beyond their control and all states will be pulled towards the average growth of the entire US economy (by definition the average must be made up of the parts so any state that got exceptionally strong or bad economic growth would at some point become such an outlier that it would begin to alter the national average).

Given these facts, are there policies that individual states can implement at the state and local level to increase their growth standing relative to the overall economy? Even though there's a limited set of reasonable policies a state can implement (I doubt any state could single-handedly turn itself into a command economy nor an Ayn Rand libertarian paradise), that's a huge search space.

A lesser question is given the diverse array of state policies we see today, do we see that some set appears to move a state up or down in growth relative to the US economy as a whole?

On the second question this study seems to provide some evidence that the answer is 'no'. At least on some of what common sense suggests are the major policy variables that states might employ, it doesn't seem to hurt or help much. There might be more obscure metrics but I'm skeptical you're going to discover that marginal income tax rates have little to do with economic growth but being allowed to make a right turn on red is somehow the magic bullet for economic growth that no one expected.

I think something that is more relevant here is that there's a limited amount of variation states can really exhibit and politicians blow up small differences with rhetorical tricks. For example, declaring that your state is more business friendly because it has '20% lower income tax rates' than the next state but if the next state has a 5% rate that means your state is the true difference is just one percentage point.

Comments for this post are closed