Hedonic Price Models for Dynamic Markets*
| DOI | http://doi.org/10.1111/j.1468-0084.2006.00441.x |
| Author | Dadi Kristofersson,Kyrre Rickertsen |
| Published date | 01 June 2007 |
| Date | 01 June 2007 |
Hedonic Price Models for Dynamic Markets*
Dadi Kristofersson and Kyrre Rickertsen
Department of Economics and Resource Management, Norwegian University of Life
Sciences, Aas, Norway
(e-mail: dadi.kristofersson@umb.no; kyrre.rickertsen@umb.no)
Abstract
The price of a product depends on its characteristics and will vary in dynamic
markets. The model describes a processing firm that bids in an auction for
a heterogeneous and perishable input. The reduced form of this model is estimated as
an expanded random parameter model that combines a nonlinear hedonic bid
function and inverse input demand functions for characteristics. The model was
estimated by using 289,405 transactions from the Icelandic fish auctions. Total catch
and gut ratio were the main determinants of marginal prices of characteristics, while
the price of cod mainly depended on size, gutting and storage.
I. Introduction
The price of a heterogeneous good is determined by the bundle of characteristics of
which it is composed. The hedonic price function relates the observed price to these
characteristics. However, the marginal characteristics’ prices are not constant over
time but rather the result of interactions between demand and supply of
characteristics. Changes in demand and supply result in changes in marginal prices
of characteristics (Rosen, 1974). To estimate these variations in marginal prices is
not straightforward. Brown and Rosen (1982) suggested a two-stage method using
data from multiple markets and assuming identical preferences across the markets. In
the first stage, the parameters of the hedonic price function are estimated for each
market to calculate the marginal prices of characteristics. In the second stage, the
*We would like to thank Subal Kumbhakar, Ragnar Tvetera˚s, Atle Guttormsen, and an anonymous referee
for their useful comments. The Research Council of Norway, grant no. 144496/110 providedfinancial support
for this research.
JEL Classification numbers: C49, C51, Q21, Q22.
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 69, 3 (2007) 0305-9049
doi: 10.1111/j.1468-0084.2006.00441.x
387
ÓBlackwell Publishing Ltd and the Department of Economics, University of Oxford, 2006. Published by Blackwell Publishing Ltd,
9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
supply and demand functions of characteristics are estimated by using the shadow
prices of the characteristics from the first stage as variables. The two-stage method
has been used to estimate the demand for heterogeneous goods such as air quality
(Zabel and Kiel, 2000), cotton fibres (Bowman and Ethridge, 1992), US automobiles
(Bajic, 1993), baseball players (Stewart and Jones, 1998) and coal (Kolstad and
Turnovsky, 1998).
We use 289,405 transactions over the 1996–2000 period obtained from the
Icelandic auction market for cod to study how prices are determined by quality
characteristics. Fresh fish is perishable and the short-run supply is highly inelastic,
resulting in substantial variations in marginal prices of characteristics. These
variations imply that the parameters of the hedonic price function are likely to vary
significantly in the short run and suggest that a random parameter model that allows
for such variations is appropriate. Following Barten and Bettendorf (1989), we
assume that the supply of fish is given at the beginning of each trading day. This
assumption implies that the daily supplies of characteristics are predetermined, which
in turn implies that the prices of characteristics are determined by the quantities of
characteristics demanded by the fish processing industry. Consequently, the problem
is reduced to identifying the characteristic input demand functions. In this case, the
efficiency of Brown and Rosen’s (1982) two-stage method can be improved by
estimating both stages as a reduced form model consisting of a hedonic bid function,
which describes the bids in the auctions as functions of quality characteristics, and the
characteristics’ input demand functions. Following Kristofersson and Rickertsen
(2004), we use an expanded random parameter model. In this model, the estimated
parameters for each day are treated as coming from a separate market with identical
technology and preferences. The shifts that caused the marginal prices of charac-
teristics to change are used to identify the characteristic demand functions.
For several reasons, cod is an interesting example of a heterogeneous good. First,
cod is the most important whitefish species, with an annual catch of about one
million metric tonnes. Whitefish is the second largest group of fish traded in the
world market after pelagic species such as anchovies and sardines. Secondly, cod is a
highly priced species of fish and a large proportion of traded cod is sold fresh.
Thirdly, cod has a large potential as an aquaculture product because of the high price,
the size of the cod market and recent advances in cod farming. Fourthly, cod has a
seasonal pattern in quality characteristics. Seasonal changes in characteristic supply
indicate that marginal characteristic prices for cod are not stable over time.
Information about the seasonal pattern of marginal prices of characteristics is of
considerable interest for the emerging production of cod in aquaculture. Fish farmers
can deliver cod of relatively constant quality throughout the year and, thereby, profit
from any seasonality in the prices of cod and of the various characteristics of the cod.
Seasonality of the quality of cod has been studied by, for example, Huss (1995),
Love (1975, 1976), Botta, Bonnel and Squires (1987), and Botta, Kennedy and
Squires (1987), and for Icelandic cod by Ingo´lfsdo´ttir (1996), Birgisson and
‡orsteinsson (1997) and Eyjo´lfsson et al. (2001).
ÓBlackwell Publishing Ltd and the Department of Economics, University of Oxford, 2006
388 Bulletin
To continue reading
Request your trial