# Price dispersion in online retail

There are a number of online booksellers that are popular in India right now. The prices they charge for the same book often differ by as much as 10% (See this price comparison site to check http://www.indiabookstore.net/)

Since these are homogenous goods and anyone who can access one of the stores can as easily assess another I wonder how we might explain this price dispersion. I have not done a systematic study but it is my impression that it is not the case that there is an ordering of prices between stores which is the same for different books.

I was wondering whether there are any economic models which can explain this dispersion. In particular are there models which can be tested by using panels of price data.

[I confess I am fishing for a research topic.]

-

Yes, this has actually been quite an active area for research within the consumer search literature. As a starting point, I would recommend looking at the following:

A couple of extra references to finish up:

-
Thanks for this very detailed answer. – Jyotirmoy Bhattacharya Nov 23 '14 at 16:24
@Ubiquitous (+1). Regarding the "De Los Santos et al" paper, are they also dealing with information costs and uncertainty? Any open-ended algorithm ("continue searching until...") creates uncertainty about the length and the costs of the process. Look at businesses: Most Procurement departments determine a preset fixed number of offers (usually no less than three, rarely more than five) that they will seek to receive in order to decide on a purchase. There is a reason for that. – Alecos Papadopoulos Nov 23 '14 at 17:06
@AlecosPapadopoulos Yes, the sequential search model (which is much older than De Los Santos et al.) does take account of these costs. Still, sequential search is optimal. The way it works is that the consumer calculates a reserve price, $r$, such that they would be indifferent between buying at that price and continuing to incur costs of search. They buy as soon as they find a price below $r$. A key result of this literature is that, although the length of the search is uncertain, the reserve price is stationary throughout the whole search process. <Continued below> – Ubiquitous Nov 23 '14 at 17:21
<continued from above> An intuitive reason why the fixed-sample search is not optimal is this: suppose you decide to search five stores, but the first store is very very cheap. You would then regret having committed yourself to spending time searching four more. Conversely if all of the first five offers are very expensive you would wish that you had planned to search for longer. The sequential search model doesn't suffer this problem because the search horizon changes depending on whether you get an early offer below the reserve price or not. – Ubiquitous Nov 23 '14 at 17:23
@AlecosPapadopoulos For example, one way of modelling consumers getting worn out would be to say that the marginal cost of search is increasing. But then the consumer should use sequential search but increase their reserve price over time. Intuitively, as they get tired of searching it becomes optimal to accept a worse offer. But still, you don't want to commit ex ante to searching $x$ firms there is a change that it will only take $x-1$ searches to find an acceptable offer. – Ubiquitous Nov 23 '14 at 17:41