19
D. The Dependent Variable and the Choice of the Independent Variables.
The vector x includes the following regressors:
• Dummy variables for the site where the interview was conducted;11
• Dummy variables for the variant of the program the respondent was told to
consider. The dummies CULTURE, INFRASTR, and SERVICES denote the variants of
the program that, in addition to implementing conservation and restoration measures,
emphasize provision of cultural and historical information at the site, improve the
infrastructure at or near the site, and improve tourist-related services, respectively. These
dummies were always assigned a value of zero when j=1, i.e., the observation on trips
refers to actual trips.
• The total cost of the trip as reported by the respondent, divided by the number of
people for whom this cost was incurred. We label this variable PRICE. We do not include
the opportunity cost of time in our price variable.12
• Household income divided by the number of household members (PCAPPINC).
We created a companion dummy variable, INCMISS, which takes on a value of one if the
respondent fails to answer the income question. PCAPPINC is recoded to zero when the
11 We include dummies for Tatev, Garni, Haghardzin, and Khor Virap. The model does not, therefore,
contain the intercept.
12 Most theoretical models assume that the opportunity cost of time is the wage rate. Much of the empirical
literature (since Cesario, 1976) imputes a fraction (usually, about one-third) of the market wage rate as the
opportunity cost of time, but Azevedo et al. (2002) point out that doing so is likely to introduce
measurement error into the price variable, which in turn biases the coefficient on the price downward. As in
Hanley et al (2003) and Alberini et al. (2005) we prefer to enter the out-of-pocket cost of a trip and income
separately.