Repeat Sales Regression

[ThePrestige]

Mods, very sorry if this isn't the right section. I thought this would be the best place for it but if not please move to the correct area.

I have sales data for art prints and paintings dependent variable with a number of independent variables. I am wanting to run a repeat sale regression for the print set of my data. Is this possible in eViews and how would I go about doing it? Links to additional reading on how to set it up would probably be sufficient otherwise a step-by-step process would work as well.

Thanks,

[EViews Gareth]

Could you define what you mean by a "Repeat Sales Regression"?

[ThePrestige]

Sure! Probably should have done that from the beginning.

A repeat sale is when a painting is sold at auction in 2005 and then the same exact painting is sold again at another auction in say 2006.

With prints a repeat sale is when a print numbered xxx/150 is sold at auction and than another print from the same "run" but numbered differently say xxy/150 is sold at another auction at a later date.

James E. Pesando and Pauline M. Shum do this in their 1999 paper "The Returns to Picasso's Prints and to Traditional Financial Assets, 1977 to 1996"

Many others use it in the when discussing art or housing sales.

[EViews Gareth]

On the surface it sounds like a pool estimation with period effects and period-specific coefficients. However I'm not sure I follow how such a regression deals with non-simultaneity of the observations. One painting might be sold in 2005 and then again in 2008, whereas a different painting might be sold in 2004 and then 2006.

[ThePrestige]

Well into more detail I would only count a sale as a repeat sale if they sold within 6 months of each other, which is how Pesando and Shum do it.

All I really want to accomplish is remove the "noise" of art auctions. Sometimes prints will sell for say 10K in New York but an auction one month later in Paris the same print sells for 20k.

This might be getting a little bit to advanced for my skill-set but without at least attempting to take it into account I do not think I am giving the research full justification and it might undermine my results.

Originally planned on doing a straight hedonic regression approach but in the print market the same print gets sold so many times.


EDIT: Thanks so much for the very quick replies.

[EViews Gareth]

I'm not sure I know enough about the methodology to comment further.

[ThePrestige]

Well thanks for the help. Might just abandon the thought all together and just try to include it in future research.

If anyone else has any insight any direction would be appreciated.