Recently, I was researching on an economic modeling company called Regional Economic Modeling, Inc., better known as REMI. REMI is one of several companies that produce economic models used by state, regional, and local economic planners, primarily to measure the impact of government programs. REMI and other companies, the best-known being Impact Analysis for Planning (IMPLAN), make their money by providing proprietrary economic models to private businesses, trade associations, government agencies and consulting firms that use them to estimate GDP growth, job creation, and other impacts associated with public policies that are usually (but not always) funded through state and local government programs. I have made several criticisms of these models, arguing that they, in fact, shouldn’t even be considered to be serious economic analyses and are simply a tool of special interest politics. I will not review those arguments here.

After reviewing promotional materials produced by REMI and posted on the internet, I came across this webinar produced in September 2018 that introduces users to the features of one of their models, REMI 2.2. (As the name implies, they have developed several models.) Representatives from REMI state that the purpose of the webinar, which is part one of a longer series, is to guide “users through the updated interface and [to offer] insight into how to make the most of the new experience.” What the webinar actually does is to demonstrate how their model can be used to generate any result that the analyst would like to produce simply by choosing their “storyline” correctly.

The webinar features a PowerPoint presented by two REMI “economic associates.” I think it’s smart that they don’t call them economists because they are clearly not conducting actual economic analysis. Their goal is to show their audience of users and potential users what the model can do when projecting economic outcomes.  The presenters decide to demonstrate this by looking at the effects of implementing a guaranteed “living wage” for Uber drivers, a policy proposed by Alexandria Ocasio-Cortez.

Perhaps to avoid offending some of their audience, they tell two “economic stories,” neither of which, by the way, represent sound economic theory about the ways that a living wage mandate for Uber drivers will affect the economy of Massachusetts. One story starts with the hypothesis that the wage increase for Uber drivers will lead to increased worker productivity due to its impact on traffic congestion. The other story starts with the assumption that the wage increase will increase prices for transportation services, which the presenter erroneously refers to as inflation. (It is what economists call a relative price change.)  I will not go down the road of criticizing the specifics of the “economic stories” presented, other than to say that neither of their assessments of the mandatory wage hike for Uber drivers would pass a test on the effects of minimum wage in an undergraduate-level microeconomics course. The point that they seem to make is that no matter what your storyline might be, the model can process it and give some sort of empirical projection of its impact on the economy. Of course, the impact on the economy is predetermined by the story.

Ultimately, what they demonstrate is that the model is not actually set up to provide an accurate answer based on sound economic principles but, instead, will provide whatever answer the analyst wanted from the start.

The first model processed through REMI 2.2 considers the story that starts with the idea that all higher labor costs will be passed on to consumers, increasing the overall price level. After running this scenario through the model, a series of results are generated. Employment will fall, GDP will fall, government revenues will fall, and the population will decline. Pretty bad right? Well, it doesn’t have to be. Just change your storyline.

The second model simulates the impact on the Massachusetts economy if we change, not the policy, but the storyline about the probable effects of the policy. In fact, the “economic associate” presenting this storyline in the webinar suggests that this could be an analysis presented by someone who is already in favor of the policy, such as Ocasio-Cortez and her supporters. It assumes that this policy leads to an overall increase in worker productivity. And if you like the policy going in, this is the storyline you would want to adopt. All of the bad news generated by the previous storyline is reversed. Employment goes up, GDP goes up, immigration into the area goes up, and tax revenues get a boost.

This means that if you are a policymaker searching for information on the impact of a proposed policy change such as this one, you could actually encounter two analyses using the exact same REMI model that are giving you the exact opposite answer to how the policy would impact the economy. The analysts presenting this webinar inadvertently tell the public that the REMI model is totally driven by the results that the analysts want before the model is even put into use. While this makes analysis coming from this model totally useless for policymakers who are searching for real results based on sound economic analysis, the REMI model appears to be the perfect tool for special interest groups pushing a rent-seeking or ideological agenda.