Discussion Paper

No. 2017-19 | April 28, 2017
Experience goods and provision of quality


Delacroix and Shi (Pricing and signaling with frictions, Journal of Economics Theory 2013) study a model featuring buyers with unit demands and sellers with unit supplies. The sellers may produce a high- or a low-quality good. The buyers get a signal about quality but the signalling technology is quite specific; the signal is either completely revealing or uninformative. The author studies the same model with a symmetric signalling technology where high and low signals are always got with positive probability. As a consequence, whenever high-quality goods are produced also low-quality goods are produced. Instead of price posting the author studies trading by auctions. There are two equilibria, and the author quantifies the efficiency loss due to asymmetric information.

JEL Classification:

D8, D82, D44


  • Downloads: 1047


Cite As

Klaus Kultti (2017). Experience goods and provision of quality. Economics Discussion Papers, No 2017-19, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2017-19

Comments and Questions

Anonymous - Referee report 1
June 02, 2017 - 12:51

This paper appears to make little scientific progress beyond the existing literature.

Specific comments.

1. The fundamental flaw of the paper is that the author did not describe the model precisely. A typical reader would understand the model in which a monopolistic seller with private information sells ...[more]

... a product to a mass of buyers who demand 1 unit of goods. This is not the model the author has in mind. The goal of the author is to improve upon Delacroix and Shi [2013] by modifying the signaling technology and the trading protocol. The model of Delacroix and Shi [2013] is much more elaborate than a simple signaling model, and requires to specify the matching and the search technologies, along the rationing rule. Unless the author completely specifies the important details of the model such as the matching and the search technolgies, the ensuing analysis would be meaningless.

2. I am sympathetic with the author that the trading protocol of Delacroix and Shi [2013] is somewhat artificial, but am not convinced of the benefit of considering more elaborate trading protocol. In fact, the auction mechanism was used by Satterthewaite and Shneyrov, to understand the asymptotic properties of the decentralized private value trading models. Satterthewaite et al. used the auction mechanism to address a specific problem arising from incomplete information. On the hand, I am not sure whether the auction mechanism offers anything new beyond what Delacroix and Shi [2013] used. Being more realistic does not mean a better modeling strategy. We have to balance the simplicity and the plausibility. By its nature, a model cannot be completely realistic. The question is what we can learn from the model, and how to choose each component of the model to deliver the main message in the most effective manner. In that sense, I believe the choice of Delacroix and Shi [2013] is justified.

3. Assumption 1 is more a condition for Proposition 1, and needs a plausible justification. What is the economic implication of Assumption 1?

4. The analysis of the paper appears to be routine, and mechanical, offering little new insight beyond Delacrois and Shi [2013].

Anonymous - response to the referee
July 11, 2017 - 13:37

1. In the revised version I describe the model more carefully; in particular I specify the timing of events, the choice sets of the agents, and the information each agent has when s/he makes a decision. I also emphasise that this model features capacity constrained sellers.

I should ...[more]

... not say that I try to improve upon Delacroix and Shi (D&S) with the different signalling technology. I just employ a different, and perhaps a more familiar, type of technology.

2. I should not depict the trading protocol more elaborate than in D&S. It is actually simpler in the sense that issues of signalling by pricing are ruled out. Auction as such is not that significant but the fact that signalling is not possible. To prevent trivial outcomes like every seller asking a very high price one has to use some mechanism of price determination such that there is at least sometimes local competition; auction is good for this purpose. I try to bring this point out more clearly.

The other two points I do not fully comprehend. First, I totally agree what is said about Satterthwaite and Shneyrov but I fail to see the relevance to the present work. Second, I do not want to hint that the modelling choices in D&S are somehow unjustified. But I think that it may be of interest to see how one can quantify the inefficiencies arising from private information in this kind of setting, and to this end one has to make some changes to D&S.

As to the role of realism I fail to disagree.

3. This is a valid complaint, and I provide heuristics for the condition by associating the level of the cost to the accuracy of the signal, trading probability and the difference between the price received under the high and low signals.

4. I think that being able to study questions of interest using routine techniques is a good thing not a bad thing. If each question requires a non-routine approach it tells that perhaps the discipline, in this case economics, is not very well developed and established. I think that the new insight is quite clear and well delineated; I am sorry that you regard it as insignificant.

Anonymous - Referee report 2
June 20, 2017 - 10:47

see attached file

Anonymous - response to the referee
July 11, 2017 - 13:39

One of the main objectives is to get some idea about the magnitudes of inefficiency. I cannot figure out how one can do this with comparative statics analysis. Comparative statics analysis, as such, is pretty straightforward just from looking at formula (6).

I pondered whether the signals should be ...[more]

... product specific or buyer specific. If they were buyer specific assuming that trades are consummated by, say, English auction would lead to updating of probabilities during the bidding phase; in general the price determination mechanism would play potentially an important role. As the price formation is not the main focus I chose the simpler option.

I have tried to improve both presentation and language throughout the article.