Economics - General

When do Consumers Search?

In classical economics models, supply and demand curves create a unique market price. Anyone who has shopped around for a good deal, however, knows that there is often significant price dispersion, even for homogeneous goods. For instance, gas prices can often vary greatly within a single neighborhood.

On Monday I attended a seminar by Matthew Lewis regarding his paper titled “When do Consumers Search?” He claimed that most search models are able to create an equilibrium which includes price dispersion by balancing two opposing forces:

  1. Increased price dispersion causes consumers to search more for better deals.
  2. Increased consumer search leads to reduced price dispersion.

One research problem encounter by those who model consumer search behavior is that it is very difficult to measure search empirically. Mr. Lewis solves this problem by using internet traffic data for, a website dedicated to providing information on local gas prices throughout the U.S. and Canada.


Lewis uses the following econometric specification.

  • ln(Reach)t = α1 + α1date + α2{k=0 to 4} Δpt-k ξt-k + α3*Σ{k=0 to 4} Δpt-k (1-ξt-k) + …

The equation measures the impact of retail gasoline price changes (i.e.: Δpt-k) on the GasBuddy traffic (i.e.: ln(Reach)). The term ξt-k is equal to unity if the price change is positive and equal to zero if the price change is negative. Using this method, Lewis can separately analyze the impact of positive and negative price impacts. Lags of 5-19 and 20-49 days are also included in the full specification.


Lewis found that increased gas prices lead to increased consumer search. Decreasing prices, however, had no effect on consumer search. Lewis explains this by citing his observation that when prices increase, price dispersion increases, but when prices decrease, there is no change or a decrease in price dispersion. Thus, the asymmetrical pricing strategies by the retail gasoline firms may explain these search results.

On the other hand, a prospect theory may explain these results more intuitively. If individuals are loss averse, an increase in gas prices will be more painful to consumers and thus search will increase. A decrease in gas prices means and increase in utility relative to the reference point and thus consumers may not be as motivated to search. An income effect could also explain this finding. This micro-theoretic explanation could help to explain why retail gas station price dispersion occurs asymmetrically with respect to price increases compared to decreases.