Walter Beckert

Lecturer in Economics
PhD (UC Berkeley)

Phone: +44 (0) 20 7631 6414
Email: w.beckert@bbk.ac.uk
Fax: +44 (0) 20 7631 6416
Room: 724
Office Hours: email for appointment

 

Research Interests

My research examines intrinsic market uncertainty, arising from stochastic demand as a consequence of incomplete information about consumers' valuations or preferences. In my econometric work, I develop a structural discrete-continuous econometric choice model to estimate heterogeneity in Internet users' valuations of bandwidth (transmission speed) and its utilization for online activity. The empirical assessment of the distribution of consumer preferences in network services is important since quality of service is determined by aggregate usage of the network by all contemporaneous users, subject to the network capacity constraint. Quality of service can be improved through efficient capacity allocation, which, in turn, can be achieved through non-linear prices. The optimal design of non-linear prices critically hinges on the empirical assessment of consumers' preference heterogeneity.

In my theoretical work, I examine strategic intertemporal interactions between the monopolistic seller of a good and potential buyers whose valuations of the good are private information. It is shown that competition among buyers in sequential take-it-or-leave-it sales induces intertemporal competitive externalities that can increase the seller's expected profits beyond the expected profit of a one-period offering, contrary to Coase-type conjectures, and beyond the expected profit of sequential auctions. This work helps explaining why sellers sometimes prefer take-it-or-leave-it deals over auctions, and vice versa. My current research extends this line of work, examining firms' capacity choices under demand uncertainty and different sales mechanisms; and it compares induced welfare under these mechanisms. Preliminary results suggest circumstances in which auctions lead to strategic choices that result in outcomes which are dominated in terms of expected welfare by the outcomes under take-it-or-leave-it sales mechanisms. This work has important policy implications, for instance for real estate markets or the evaluation of proposed landing slot auctions, as opposed to traditional pricing schemes.

My recent research in mathematical statistics, joint with Daniel McFadden at U.C. Berkeley, develops a parametric estimation methodology that can be applied when the classical maximum likelihood theory breaks down, either because classical regularity conditions fail, or because the maximum likelihood estimator, while existing in theory, is analytically intractable in practice. Such situations arise frequently in Applied Econometrics. For example, the absence of classical dominance conditions, as in the switching regression model, lead to the former case, while structural random utility models for continuous choices induce the latter. The methodology proposed in this work replaces the full information likelihood function by a limited information multinomial approximation, where the approximation refinement increases with sample size. Building on the theory of locally asymptotically quadratic families (Hayek and LeCam), it is shown that, as sample size increases, the limited information estimator converges at the fastest possible rate, is consistent, has a stochastically bounded limiting distribution, and is asymptotically fully efficient.

Auxiliary research, joint with Daniel McFadden, determines maximal uniform convergence rates in parametric estimation problems. It focuses on rate efficiency of parametric estimators, next to conventional asymptotic variance-covariance efficiency in the sense of the Gauss-Markov and Cramer-Rao Theorems. This work provides an almost universally applicable approach, based on the Hellinger metric on the space of parametric probability measures, to establish maximal uniform convergence rates. It extends related results by Ibragimov and Has'minskii (1981).

Selected Publications:

(1) Theoretical and Applied Econometrics

Beckert, W. (2005): Estimation of Stochastic Preferences: An Empirical Analysis of Demand for Internet Services, Review of Economics and Statistics, 87(3), 495-502

Beckert, W. (2007): Specification and Identification of Stochastic Demand Models, Econometric Reviews, 26(6), 669-683

Beckert, W. and R. Blundell (2008): Heterogeneity and the Nonparametric Analysis of Consumer Choice: Conditions for Invertibility, The Review of Economic Studies, 75(4), 1069-1080

Beckert, W. and D.L. McFadden (2008): Maximum Uniform Convergence Rates in Parametric Estimation Problems, forthcoming in: Econometric Theory

Beckert, W. (2010): Choice of NHS-funded Hospital Services in England, forthcoming in: Economic Journal (joint with M. Christensen and K. Collyer)

(2) Economic Theory

Beckert, W. (2006): "Competitive Externalities in Dynamic Monopolies with Stochastic Demand", The B.E. Journals in Theoretical Economics, 6(1), Article 17

Recent Working Papers

CV

Department of Economics, Mathematics and Statistics, Birkbeck, University of London, Malet St, London WC1E 7HX.