What do Rick Perry and pro sports teams have in common?

They use the same shady economic methodology to promote their policies.

If you follow the news, you’re familiar with “IMPLAN”, albeit indirectly. It’s the software package underlying the studies that pro sports teams, among others clamoring for public favors, use to claim that each new stadium will generate several gazillion dollars for the local economy—supposedly justifying a massive public outlay. Here’s a study using IMPLAN to justify a new Sacramento Kings stadium; here’s another that looks at the proposed Santa Clara stadium for the 49ers and another that attempts to justify a new stadium for the A’s. There are studies looking at the impact of the Mavericks’ American Airlines Center, the Packers’ Lambeau Field, and Oriole Park. And, of course, there are countless others: whenever someone wants to make preposterous claims about the benefits of his pet project, he’ll inevitably turn to IMPLAN or a similar package.

Artist's rendition of the new 49ers stadium proposed for Santa Clara, whose economic impact has been studied using the same highly reliable methodology now applied by Rick Perry's campaign.

There’s an obvious element of pseudoscience to these studies. They use “input-output” models that painstakingly track the path of spending through the economy—a worthy goal, though perhaps an overambitious one. But they fail entirely to model the supply side of the economy, effectively assuming that there is unlimited capacity, and that each additional dollar of “spending” (magically generated by the new stadium) will become an additional dollar of economic activity—even more, in fact, after you account for the multiplier.

Strangely enough, Rick Perry’s campaign is using the same model to analyze his tax plan, in a context where it makes even less sense.

As James Pethokoukis explains, the Rick Perry presidential campaign has contracted with John Dunham and Associates to run a revenue analysis of Perry’s new tax plan. The impact of the plan depends on your choice of baseline policy: it raises $4.7 trillion less than the CBO baseline for 2014-2020 under conventional, static scoring, and $1.7 trillion less under “dynamic scoring”.  Relative to the CBO’s more arguably realistic alternate baseline, the plan does better. But regardless of your preferred baseline, it’s clear that the plausibility of Dunham’s “dynamic scoring” model is key: it provides an additional $3 trillion over only 6 years!

It’s troubling, then, to learn that the Perry campaign’s idea of “dynamic scoring” bears absolutely no relation to what most economists mean by the term. In fact, the Dunham model more closely resembles the shady estimates for the 49ers stadium than any accepted methodology in public finance.

The idea behind dynamic scoring—as economists generally understand the concept—is that we should account for how the incentives created by the tax system affect the economy, and how those effects might feed back into revenue. A income tax cut, for instance, might lead to higher taxable income and a new stream of tax revenue—though certainly not by enough to fully offset the initial revenue loss, as Art Laffer once claimed. In theory, capital tax cuts may to even larger offsetting movements in revenue, though still not enough to recover the loss completely. Greg Mankiw and Matthew Weinzierl provide a short guide here.

Dynamic scoring is controversial: many Democrats believe that in practice it’s a gimmick that obscures the revenue losses from tax cuts. But in principle, it’s hard to deny that dynamic scoring would be the ideal way to evaluate the effects of tax policy: taxes do have real effects, and those effects eventually find their way back into the tax base. The challenge is that the relevant magnitudes are extremely uncertain, and it’s hard to calibrate a model that realistically accounts for the effects of new policy. Moreover, if the overall effect on revenue is negative, to be complete you need a model of how other tax and spending policies will eventually adjust to close the additional deficit—a very difficult task indeed, one that practitioners usually ignore. Some proceed nevertheless; some think it’s better to avoid the issue until we have more accurate models.

But none of this matters to the Perry analysis, because it’s completely unrelated. It doesn’t look at the effect of taxes on incentives at all. Instead, it simply feeds the increased personal income from tax cuts into the IMPLAN model and churns out the same kinds of estimates we typically see for sports stadiums. If you think I’m kidding, read the document:

In order to better understand the effects of the Perry tax proposal on the national economy, a dynamic scoring exercise was conducted by JDA. JDA used an input-output model of the US economy to estimate the true revenue effect of personal income tax proposals, including the feedback effects of taxes on national income.

The dynamic analysis used in this model was based on tax savings (or tax increases) for various income groups in each of the 7 years between 2014 and 2020. These savings were run through the IMPLAN input-output model as increases to income for each group and the resulting change in GDP was fed back through the model for subsequent years. This led to higher GDP growth estimates for each year beginning with 2014 (see Table 6). Based on this analysis, GDP is expected to grow faster than forecast by the CBO, reaching $26.5 trillion by FY 2020 – a 16 percent increase.

Needless to say, this description is a little hazy, but it’s pretty clear what’s going on: they look at tax savings as increases in “income” to each group and feed them through the IMPLAN input-output model, which vastly multiplies the initial impulse and leaves us with an utterly implausible estimate for improvement in GDP. (Sixteen percent? Are they kidding?) There’s no recognition that long-run output is determined by supply constraints, not demand; in fact, this is a completely demand-side analysis trying to pass itself off as supply-side dynamic scoring. Rather bizarre for a Republican candidate, particularly one as hostile to demand-side policy as Rick Perry!

Now, to be clear, there is a place for demand modeling and multipliers like those in the IMPLAN model: when we’re in a demand-constrained recession and monetary policy has reached its limits, tax cuts may provide economic stimulus by boosting aggregate demand, not just improving supply-side incentives. (Though there’s a debate about that.) But this is explicitly a short-to-medium term phenomenon, one that only matters (if at all) in a zero lower bound recession. No one—not even the most fanatical Keynesian—claims that such multipliers provide a foundation for long-term analysis of public finance. And certainly no one is crazy enough to think that the demand-side effects of a tax cut can boost GDP by sixteen percent, as the Perry analysis claims.

In fact, this model makes even less sense in the context of federal tax policy than in its usual, already dubious applications. When we’re looking at a stadium, at least we’re confining ourselves to a particular region: consumers flocking to a stadium can’t boost the productive capacity of the US economy as a whole, but they might encourage labor and capital to relocate around the stadium, delivering economic expansion to the region in question. But this doesn’t apply to the US as a whole: we only have so much labor and capital. Granted, if the model looked at the supply side of Perry’s plan, it might demonstrate how improved incentives lead to an expansion in labor and capital supply, thus increasing potential economic output. That is, however, what the model explicitly does not do: it ignores supply considerations completely, instead assuming that supply constraints are irrelevant and that the income from tax cuts will forever ripple throughout the economy and prompt a demand-led expansion that would put the Clinton era to shame.

I never thought I’d see the day when I had to lecture a Republican presidential candidate on the importance of supply-side analysis, or the dangers of overexuberant demand-side logic. Apparently that day has come!

The truth, of course, is that neither Rick Perry nor his staff have any idea of the analysis behind their numbers. Instead, they hired a consulting firm that specializes in using IMPLAN to create exaggerated estimates for the effect of particular industries (“Meat! Responsible for 5 trillion jobs!”) in order to please its lobbyist clients. The firm evidently knows nothing about tax analysis; it has no credentialed public finance economists on its staff and no experience in analyzing tax policy. When asked to conduct a study, it turned to the only game it knew: IMPLAN, which just happens to be a absurd way to analyze national fiscal policy.

But hey—cut them some slack! It’s not like they’re evaluating the key economic proposal from a major presidential candidate or anything.

Update. Lifted from the comment section, from economics professor and sports researcher Donald Coffin:

Nice to see a good take-down of the IMPLAN modelling approach. Those of us who do sports economics and urban economics seriously are almost constantly having to push back against those kinds of studies. The single most disturbing aspect of the IMPLAN model for local economic analysis is the wildly unreasonable values that have for multiplier effects (compared, for example, with the BEA’s Regional Input-Output Modeling System). IMPLAN is exactly what you describe it as, a “model” designed to generate large impact numbers to please a client who wants to lobby someone.



Filed under fiscal, policy

23 responses to “What do Rick Perry and pro sports teams have in common?

  1. Gerald McCracken

    You’re a 23-year-old with a bachelor’s degree (congratulations on accomplishing something that is neither impressive nor unique). Let’s take it easy on the attacks against other people for not being “credentialed public finance economists.”

    • Not sure what your point is. I didn’t say that *I* was qualified to run a dynamic analysis of a new tax proposal. I’m not. But I do know enough about public finance to say that this analysis is clearly nonsense, of a kind you never see from people who do this professionally. I’m not an ardent credentialist—if Dunham and Associates put out a credible study using even marginally accepted methodology to analyze the effects of tax policy, I wouldn’t be harping on their lack of “credentialed public finance economists”. But unfortunately, their study was not credible, and I think the absence on their staff of any professional expertise in this area is clearly part of the story.

      By the way, I am a Ph.D candidate in economics at MIT, which hopefully makes me just a little more knowledgeable that some guy “with a bachelor’s degree”. (Not sure how you could have possibly know the latter without knowing the former—the old resume is on the page that also discusses my more recent background. I guess you just wanted to take a cheap shot?) Public finance is not my area of specialization, which means that I certainly shouldn’t be your top source on the precise details of fiscal policy. But I have done some work in public finance, helped with the research of a leading professor in the field, and so on. And again, it doesn’t take a lot of background to understand that the Perry study is not credible and uses methodology that is never (and should never be) otherwise applied to the study of tax policy.

      • Donald A. Coffin

        Nice to see a good take-down of the IMPLAN modellng approach. Those of us who do sports economics and urban economics seriously are almost constantly having to push back against those kinds of studies. The single most disturbing aspect of the IMPLAN model for local economic analysis is the wildly unreasonable values that have for multiplier effects (conpared, for example, with the BEA’s Regional Input-Output Modeling System. IMPLAN is exactly what you describe it as, a :model” designed to generate large impact numbers to please a client who wants to lobby someone.

  2. Yep. How to do an economic impact assessment.
    1. Start with a standard Cost-Benefit Analysis.
    2. Throw away all the non-pecuniary stuff.
    3. Delete the benefits.
    4. Change the sign on the costs, so the costs become benefits.
    5. Multiply by two.

  3. Hondo

    Glad to see the pajama’s taken off IMPLAN but that could be said about many (most) economic models. By the way having a Phd. is not necessarily impressive especially if one has been taught and still adheres to the old dogma’s of economic modeling that have proved worthless but are still very much in use.

  4. Pingback: Consider Magazine » Blog Archive » Endpoint

  5. Benjamin Cole

    Excellent blogging…and I am an old fart. I can remember using Fortran cards, and I bet Matt Rognlie will have to look up that word on Wikipedia.

    As for models….they always seem to support the partisan, ideological or commercial biases of the modelers. Gee, how does that happen?

    Another thing about models: Remember Long Term Capital Management? Smart guys. Had no biases or axes to grind, just wanted to make money, Had plenty of money to make high-quality investment models. And they blew up.

    Economics is not physics. The latter is easier.

  6. K

    Matt: OT, sorry. Would it be possible to add a list of recent comments in the right column on the home page? If you aren’t using RSS it’s tough to remember which threads you should be checking.

  7. As Steven Benner wrote in his book (“Life, the Universe, and the Sientific Method”), “modeling is doomed to succeed”. Which means that if you control the model and its parameters, and if you know the “right” answer, your model will give you the “right answer”. Why? Because you stop parameterizing the model when it does. And if you do not, then your thesis adviser will send you back to the lab to continue parameterizing until it does. And if the adviser does not, then the journal editor will …
    And Benner was talking about the “hard” sciences (chemistry,physics, biology), Imagine how much worse this is in sciences that have no access to controlled experimentation, like economics and (should I dare) climate “science”.

  8. Pingback: Romney Might Be a Pretzel, But We Should Be Bent Out Of Shape » Postmodern Conservative | A First Things Blog

  9. It’s great to read something that’s both enjoyable and provides praagimtsdc solutions.

  10. I see you have died once again.


  11. Pingback: Obama’s brain mapping project is a good idea. But is he selling it like a sports stadium? | AEIdeas

  12. Going to start blogging again? Really enjoyed the discussion of yours with Miles Kimball.

  13. Wheat Thin

    Wow, maybe you should actually contact IMPLAN and thier staff. Blaming IMPLAN for bad reports is like blaming Ronald MacDonald for getting a bad cheeseburger or Microsoft word for for the Unibombers manifesto. However, people like you never write or report on anything positive, all negative… great way to slog through life. Also good way to burn bridges being a soon-to-be unemployed college graduate with a PhD In Economics… kind of slims your field down a little.

  14. Great post. Rick Perry has been a (willing?) victim of shoddy math throughout his tenure, e.g. grossly oversized tax incentives to attract companies to TX. Nearly a year after you wrote this, I investigated his $250M tax break granted to Amazon and found it around 12X what it should be for the state to not suffer tax revenue shortfalls in the future: https://jdrch.wordpress.com/2012/12/03/spending-and-tax-cuts-are-the-same-thing/

    There’s also the fact that Texas deliberately operates as a low-cost brain drain for talent developed by other states’ investment in their own education and public services, but that’s another story.

  15. Andrew

    Keynesian multiplier = broken window fallacy

    advanced mathematics can’t replace universal laws of economics (Mises)

  16. Matt Molewski

    I like how you’ve been getting comments on this one post for years now: Your readers are clearly starved for content.

    Completely unrelated, but I just read your two papers critiquing Piketty’s Capital, and want you to know how much I and others were impressed (and that’s me speaking as a fan of the French economist). You undoubtedly have a bright career ahead of you, and I look forward to your future work: I also really hope Piketty answers you, as I’m not sure how he can maintain his central conclusions in light of some of your objections.

  17. Lowering the marginal tax rate on additional taxable activity need not have any effect at all on the growth of real GDP to make the lower tax rate “pay for itself.” All that is required is a high elasticity of taxable income with respect to the net-of-tax rate.

    To take an obvious example, a recent JEC-CBO study cannot rule out a sufficiently high elasticity for capital gains realizations to make, say, the 20% capital gains tax rate of 1997-2002 bring in far more revenue than the 28% rate of 1987-1996 or the 38.9% rate of 1976-77 (as, in fact, it did).

    Looking only at top 1% tax data, it is arguably plausbile that a much lower tax rate on qualified dividends in 2003 also raised more revenue http://www.cato.org/publications/working-paper/misuse-top-1-percent-income-shares-measure-inequality Firms responded by paying more dividends; investors responded by holding more dividend-paying stock in taxable accounts.

    All but one ETI estimate for top 1% reported income in general suggest that reducing the top-bracket rate could easily bring in more revenue over time. http://www.cato.org/publications/commentary/course-70-tax-rates-are-counterproductive

    If lower marginal tax rates on highly responsive activities also contribute to more rapid growth of real GDP (not “per capita” because migration matters) then the odds improve that U.S. marginal rate reductions in 1925, 1964 and 1983-88 could raise more real revenue over time (as, in fact, they did).

  18. Matt, this blog is fantastic, I hope you’ll find the time and energy to resume it.


    Kenneth Duda
    Menlo Park, CA

  19. Sami Karim

    Thanks for the discussion. I am slightly confused as to whether we should be criticizing the model or its use? Thanks for this post. Really interesting discussion!

  20. Pingback: “$9.1 billion economic impact”—really? – Dragonpipe Diary

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s