reduced-form myopic demand model with habit persistence.
The contribution of this paper is at least two-fold. First, newspaper articles are differentiated
by their contents, an improvement over existing meat safety indices in the literature. Second, a
rational habit persistence model is estimated with meat data. The estimated structural parameters
can then be used to simulate demand responses to price and food safety shocks under different
expectations schemes.
The plan of this paper is as follows. In section 2, a theory of consumer response to food safety
information under rational habit persistence is described. In section 3 we discuss the data used in
our empirical analysis with special attention to the food safety data. In section 4 the econometric
technique used to obtain estimates of preference parameters is outlined, and then, empirical results
are presented and discussed. Finally, section 5 provides concluding remarks.
2 Theoretical Model
2.1 Intertemporal nonseparable preferences
The simpliest way to introduce time-nonseparable preferences is to let current consumption depend
on consumption in the previous period. It is the most common approach to consumption dynamics
in the literature on habit persistence (e.g., Becker, Grossman and Murphy, 1994; Dynan, 2000).
Under uncertainty, the representative household maximizes at period t the present value of a lifetime
utility
∞
max Et V βτ-tuτ (Xτ ,Xτ-1 ,Zτ) (1)
Xt τ=t
where uτ is the within-period utility at period τ, Xτ is a vector of N consumption goods (e.g.,
meats) at τ , Et is the expectation operator conditional on information available at time t, and β is
the discount factor. The vector Zτ contains variables that measure the quality aspects of the goods
at τ . The idea is to let Xτ-1 be the vector of habit stock variables to proxy past consumption
experience. Implicit in (1) is the assumption that other goods are weakly separable from the X
vector of commodities that are potentially habit-forming. The budget constraint is
∞
V(1 + rτ)τ-t(Yτ + PTXτ - Ут) = Wt (2)
τ=t
where rτ is the riskless interest rate between periods τ and τ + 1, Yτ represents expenditures on all
other goods at τ , Pτ is the price vector corresponding to Xτ , yτ is the household income at period