Discussion Paper
No. 2010-16 | May 17, 2010
Walter Beckert
A Micro-Econometric Approach to Geographic Market Definition in Local Retail Markets: Demand Side Considerations


This paper formalizes an empirically implementable framework for the definition of local antitrust markets in retail markets. This framework rests on a demand model that captures the trade-off between distance and pecuniary cost across alternative shopping destinations within local markets. The paper develops, and presents estimation results for, an empirical demand model at the store level for groceries in the UK.

JEL Classification:

C35, C73, L11, L13, L41, L81


Cite As

[Please cite the corresponding journal article] Walter Beckert (2010). A Micro-Econometric Approach to Geographic Market Definition in Local Retail Markets: Demand Side Considerations. Economics Discussion Papers, No 2010-16, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2010-16

Comments and Questions

Sunil Sapra - Comment on Beckert
May 23, 2010 - 23:39
In this interesting paper, Beckert uses maximum simulated likelihood (MSL) estimator for estimation of parameters in a mixed multinomial logit (MMNL) model. I just want to draw attention to a limitation of the MSL estimator and relative merits and limitations of alternative simulation-based estimators for the MMNL model. The main limitation of MSL estimator is that it is inconsistent for fixed R (number of draws used in simulations). Alternative methods such as Maximum Simulated Moments (MSM) and method of Simulated Scores (MSS) estimators remain consistent for fixed R. However, MSM is less efficient than MSL if non-ideal weights are used. Depending on how the scores are simulated, MSS can differ from MSL and can attain consistency and efficiency under less restrictive conditions than MSL. Thus MSS offers the possibility of attaining consistency without a loss of efficiency. MSS however has numerical disadvantages relative to MSL: the simulator is not continuous in parameters and an “accept” may not be obtained within a given number of draws. Sunil Sapra, Department of Economics and Statistics, California State University, Los Angeles, CA 90032

Walter Beckert - Simulation assisted estimators
June 07, 2010 - 09:20
Many thanks for pointing out the relative merits of different simulation assisted estimators. These comments certainly deserve being articulated in a revised version of the paper. There may even be scope of re-estimating the model using MSM, instead of MSL, although this was not necessary for the actual competition inquiry that gave rise to this work.

Anonymous - Referee Report 1
June 15, 2010 - 09:11
see attached file

Walter Beckert - Reply to Referee Report 1
July 06, 2010 - 11:06
Many thanks for the encouraging and supportive report on my paper.

Anonymous - Reader Comment
July 06, 2010 - 08:40
see attached file

Walter Beckert - Reply to Reader Comment
July 06, 2010 - 11:14
Many thanks for this detailled and supportive report. I agree with the referee that the methodology outlined in the paper could be developed further, albeit in a separate paper (in line with the referee's own suggestion). In fact, the methodology is not limited to supermarkets, but could be adapted to choice situations involving other multiple-service points, such as tourist destinations or hospitals. In concurrent joint work with economists at the UK Department of Health, I am investigating the latter application using UK data on individual health episodes and associated elective hospital treatment. I am also sympathetic to the referee's suggestions to compress (or possibly eliminate) some sections of the paper, pending further editorial decisions. Again many thanks.

Anonymous - Referee Report 2
July 13, 2010 - 10:49
See attached file

Walter Beckert - Response to Referee 2
July 17, 2010 - 14:43
Please see the attached response. Many thanks for your comments and suggestions.