Sunday, June 23, 2013

Are models that assume linear utility useful?

I just saw a paper on how the desire of households to match with particular houses could explain housing market dynamics--in particular why house prices are more volatile than incomes.

Performing such an exercise is very difficult, and requires simplifying assumptions.  One of the most important simplifying assumptions in the paper is that utility is linear--that people value their last unit of consumption just as much as their first.  This assumption is clearly wrong--we know that marginal utility diminishes in consumption.  Yet the assumption was necessary to make the model tractable.

So do we know more about the world because of the model or not?  I really don't know.


Squarely Rooted said...

The answer is no. I'm short on time, but that's the answer. No.

Anonymous said...


I have not read the paper but a balance between unrealistic assumptions and fitting what we observe in the real world must always be struck. What the research question is and the relationship that we wish to focus on are the most important considerations when deciding if an assumption makes the model unbelievable. I think in most cases consumption is a very important part of a model so if the findings fall apart when marginal utility is diminishing then I have a problem with it.