Sunday, October 26, 2008

Somebody needs to do a new hedonic regression

I came back to Washington this weekend to help my wife prepare for putting our house on the market. Per the instructions of the Realtor, we are working on "staging" it.

I wonder how much this really matters. Once one gets beyond structural and neighborhood characteristics, does "staging" matter to the ultimate sale price? One would need to collect data on sold houses (including some index of how well presented they were) and then run a hedonic regression that included the presentation index in order to find out.

Paul Carrillo at GW has a nice working paper where he finds that houses that are marketed on line with pictures get better outcomes than those that don't. But the pictures could just reflect the fact that Realtors are more likely to present pictures of houses that are better looking (and therefore more valuable) in the first place.


Anonymous said...

The implied argument for staging is that it affects the slope of the demand curve and therefore the price people are willing to pay. But I don't think that small effect is going to be captured in most econometric models. You'd really have to have some fantastic data, especially about competing offers.

Wouldn't another argument for staging be to decrease the time on the market? If so, a researcher may have better luck teasing out an effect with survival analysis.

Thomas Galvin said...

I think a hedonic model would be a little complex, since no two houses are the same capturing the marginal effects of a staging dummy variable and then trying to pry that out of what the selling price of the house would be difficult.

And you would probably have a selection problem. Presumably, the houses that are staged can afford realtors who can afford to stage a house. It is not a random sample.

What about a difference-in- difference model to capture the effects of a staged house?

The information may be a little difficult to obtain, you would need a large group of houses that all need to be sold (like bank foreclosures) and some of these houses would be sold at auction (no staging) and some would be sold by realtors representing the banks (presumably with staging).

Again this would have some selection problems, since we cannot randomly assign which houses were to be auctioned and which were to be sold by realtors, and we cannot guarantee that realtors would stage each and every house, but it is a start.

Anonymous said...

That is an interesting response. I can see where you are coming from, good input.