Countering Bloated Economic Impact Reports

(Thanks to Dave Swenson for this post. Journalists should stop quoting self-serving, industry-generated economic impact numbers. - promoted by desmoinesdem)

Iowa was recently informed with quite a bit of media hoopla that the proposed Dakota Access pipeline conveying oil from the Bakken formation in North Dakota to a refinery in Illinois will give a “$1.1 billion boost” to the Iowa economy and support 7,600 jobs.
 
Both numbers are hooey.
 
Before I get to the hooey, however, I need to talk a little bit about economic impact studies.  These studies usually utilize an input-output model of the study region.  These models are initially constructed properly, and they provide reasonable and reliable estimates of the multiplied-through consequences of economic change. Those consequences are often called “the ripple effect” because a change in activity in one industry affects all of the industries that business relies on for inputs; hence, the economic impact.  Done properly, they are useful tools for economic development planning.
 
There are times, though, that the modeling activity produces information that is not reliable.  The first instance arises where the modeler does not specifiy the study situation properly. And a second problem arises when the modeler misstates the result of the modeling process because of inexperience, ignorance, or cynicism.
 
Let's start with the second problem. The Dakota Access pipeline analysis from Strategic Economic Partners of West Des Moines reports the economic impact information as if the activity occurred in one year.  And in the main, that is how the results were reported by the press. But the project, they tell us will take 2 years. The $1.1 in total economic activity supported and the 7,600 jobs should therefore be divided by 2.  When we report economic impact activity, we properly do so on an annualized basis — how much activity was supported in year 1, and how much activity was supported in year 2. If there is a figure to be conveyed to the press, then, it is the average annual value of the project. So, at a minimum, both numbers should have been divided by 2 and reported as such.

I have two simple examples: suppose I work for a firm for 20 years. That equates to 20 job years, the kind of number alluded to in the consultants' report. But I only count as one jobholder working for 20 years. That's how the government counts me, that's how the IRS counts me, and that is how a competent social scientist would count me. Now let's assume that I made an average of $50,000 a year over all of those years. That would sum to $1 million, but I certainly am not a millionaire. In my example I was one job making an average of $50,000 a year. That is the true statement. I was not 20 job years making $1 million. That would be a distortion.

Now let’s go back to the first problem, the one of properly specifying the model.  This study assumes that a substantial fraction of the critical durable goods into the project — the pipes and pumps, for example — will be manufactured within the study region. They provide no justification for that assertion other than the default steel pipe manufacturing values that are contained within their modeling system.  But the probability the pipes and compressors are in fact produced in the region are zero or close to zero. One would have only needed to do a little research to demonstrate that. For one, there are no oil pipeline manufacturers in the region traversed the by the project according to all of the searches that I did,but which I doubt they did.  There are pipe and tube manufacturers of all kinds, but not pipeline manufacturers.  And secondly, and to my amusement, the documentation for the much more notorious Keystone XL pipeline, upon which the methods for the Dakota Access study were based, stated that they would purchase only half of their pipe from a U.S. supplier, and that supplier was in Arkansas.  I contend that the same situation also applies to pump manufacturers.  By assuming a default linkage to regional pipe and pump manufacturing, the study authors inflated the secondary economic impacts substantially.

Finally, the study assumes there is a native pipeline construction industry that will ramp up to accommodate the new demand. In 2013, Iowa had 34 establishments and 245 payroll employees doing that work. It is really hard to believe these, on average, very small firms will be the successful bidders on the pipeline project and able to ramp up in employment, machinery, and other operating scales to accomplish the task.  It is much more likely, and my research on the Keystone Phase 1 pipeline confirms this, that state pipeline construction employment may increase from between 150 jobs to about 650 jobs annually, but that a very healthy fraction of the construction jobs will go to firms from out of state who bring their most skilled workers with them on the project (and who, of course, will send their weekly paychecks back home).  

There are many, many more quibbles that I have with the Dakota Access study, the Keystone XL analysis, and the Keystone Gulf study that was released last year. I believe all engage in substantial double counting of activities and didn't go through the hard work of properly specifying the pipeline construction industry within their models.  I will write more about this in the coming weeks.  All certainly suffer from the two problems that I described, all overstate or mis-measure the likely impacts, and all, I contend, were conducted in a manner so as to paint the rosiest, albeit distorting, job picture possible to policy makers and regulators who are easily wowed by job growth declarations.

Everyone does this. Corn ethanol did this during the last decade and succeeded famously,if not notoriously. Industry sponsored job impact claims for Iowa were in the neighborhood of nine-times greater than I could honestly report.  I have also critically reviewed bloated wind energy and solar economic impact claims.  Some were naïve misstatements, and the proponents were open to corrections. But others were not, and in the case of the Dakota Access study I know this for certain: the authors would not fare well in presenting that study before a panel of regional scientists.  However, in light of most private consulting studies purporting economic impacts, it is a state of the art study.

And now you know what it takes to be state of the art in private sector economic impact forecasting. 

Dave Swenson, Ames, Iowa 

  • Thanks, Dave

    Keep writing here.  We need to use numbers more and we need to have them explained.  Otherwise it’s just lies, damned lies and statistics.  Don’t let the propaganda carry the day.

  • But still.....

    Great piece, Dave.  That is indeed how it is done.  When working for the Farm Bureau in the late 90s, I was asked to prepare an economic impact analysis of the Kyoto treaty.  I asked the boss how he thought I should estimate the base case in which Iowa becomes the wheat belt.  I was quickly on to my next project.

    But I’m also guessing you would not want your readers to assume you believe the pipeline is a bad idea, at least not from this piece.  Pipelines move oil more cheaply and safely than other forms of transportation.  While economic modelling is usually pure sophistry, it’s nice when they at least get the sign right, no?  Or are you suggesting there is not a net benefit from the project?

You need to signin or signup to post a comment.