objectives.
The analysis is undertaken in two steps. As a preliminary step, historical price data are divided
into two periods (before and after liberalization) and descriptive statistics are generated for
various crops and markets. Price correlation analysis, though with well established weaknesses,
is also used to provide preliminary insights into the changes in market integration as a result of
liberalization. In the second step, econometric analysis is undertaken to estimate the effect of
liberalization on equilibrium market prices, after controlling for other factors such as rainfall, food
aid, and seasonality.
The model
One approach to modeling price effects would be to build a structural econometric model
consisting of behavioral equations to explain the supply and demand decisions of all participants
in the market, including producers, consumers, traders, and state agencies involved in food
marketing. However, this would require a large model which embodies many over identifying
restrictions drawn from economic theory. These restrictions usually take the form of excluding
variables from particular equations in order to motivate a particular economic interpretation for
the model. Of course, it is not necessary to work with large systems because there are methods
for estimating individual structural equations embedded within a larger system. However,
estimating price effects in individual equations only provides information on the effects of price
on the behavior of the particular agent being modeled (e.g. on producers if a supply equation is
being estimated). A structural approach to estimating the effects of market reform on equilibrium
prices would require structural equations for all market participants at each stage in the system,
from production to marketing to consumption.
A potential problem with large-scale structural models is that the restrictions used to identify the
model may not be valid. A multi-market structural model of a vertical marketing chain is
complicated, particularly when it involves international trade. But economic theory often only
provides weak guidelines on how identification can be achieved. For example, Sims (1980) has
shown that if expectation variables enter an equation then it is almost impossible to exclude any
relevant variable which is known at the time expectations are formed, because these variables will
enter through the expectations term. If incorrect identification restrictions are imposed then the
model can provide misleading results (Tomek and Myers 1993; Sims 1982; Jayne and Myers
1994).
An alternative is to directly specify a reduced form model for equilibrium food price levels. Such
a model would include variables that might be included in structural models drawn from economic
theory, but otherwise the model is left relatively unrestricted. Data availability will also affect
what can be feasibly estimated. Historical price correlations are summarized by including lagged
variables, and statistical criteria are used to determine how many lags to include (Judge et. al.
1985, Chapter 16). The advantage of this approach is that the minimal restrictions applied to the
reduced form provide flexibility which allows the model to be consistent with a wide range of
alternative economic structures (Tomek and Myers 1993). The disadvantage is that structural
information on the effects of price on the supply or demand decisions of particular market
participants is not available. Nevertheless, the main goal of the present study is confined to
estimating the net effect on CPI-adjusted price levels during the pre- and post-reform periods,
-4-