Model Behavior

Over the last few days, I’ve read a number of objections to economic models (most written by non-economists, but a few by economists). These objections follow some pattern of “models are too static.  The ceteris is never paribus!  Therefore, your objection to Policy X on the basis of Model Y is invalid!  Such and such could occur!”

While true the ceteris is never paribus (in other words, not all other conditions are held the same), that objection doesn’t address the point of economic models.  Certainly anything is possible.  It’s possible that, following a minimum wage hike, there is no negative impact on low-skilled labor. It’s possible that protectionism could lift domestic well-being.  But economic models aren’t designed to capture all that’s possible.  They’re meant to indicate what is probable.  When a price floor (like minimum wage) is enacted, it is possible it has no negative effect, but it is probable it will.

To this end, models provide the good economist with insight into proposed policy solutions.  It’s difficult to predict with absolute certainty the outcome of a given policy given the Hayekian knowledge problem, but being able to make general predictions based upon probable outcomes allows us to advise against bad solutions to problems.

4 thoughts on “Model Behavior

  1. “. . . being able to make general predictions based upon probable outcomes allows us to advise against bad solutions to problems.”

    And, if government took economic advice based on probable outcomes, then the federal government would be much smaller indeed.


  2. It seems to me that models can best be used as tools for tracing out the effects of policies or changes in preferences. If we observe that the price of good X falls over time, we can use models to explain why. This can help us make predictions about the future, but predictions aren’t in and of themselves the ultimate test of the theory behind the model.

    Whenever this topic comes up, I recommend this short article. It’s probably my favorite short piece on methodology.


  3. That’s a very utilitarian view. Models let us explore our assumptions, nothing more, nothing less. Are some models helpful for thinking through real problems? Probably. But what are we to make of that? If economics is a science like biology then models will help us to categorize regularities, like a butterfly taxonomy. If economics is a science like geometry, deducting implications from the axiom of human action then models might still be interesting as thought experiments to explore relationships between different configurations of assumptions. Should we really be advising any policy action other than less government on the basis of models? I would say not. Models can potentially elucidate patterns, but in the end it’s morality that must drive “policy,” not models, and you know the correct moral answer before you even do the math: theft is wrong, taxation is not voluntary, and people who aren’t harming others should be left alone. A model can be used as a decoration or as marketing for a policy, but it’s morality that actually dictates policy. Utility is just an excuse many people confuse for a moral principle.


Comments are closed.