15
We further assume weak complementarity of trips with quality at the site, q. In other
words, ∂U /∂q = 0 when r = 0 (when a person does not visit the site, his or her utility is
not affected by its quality), and r is increasing in q. The individual chooses X, L and r to
maximize utility subject to the budget constraint:
(2) y + w ∙[T - L - r(t 1 + 12 )]= X + ( f + Pd • d )• r
where y is non-work income, w is the wage rate, T is total time, 11 is travel time to the
site, t2 is time spent at the site, f is the access fee (if any), Pd is the cost per kilometer,
and d is the distance to the site.10 This yields the demand function for trips:
(3) r*=r*(y,w,pr,q)
where pr = w(t1 + t2)+ f + pd ∙ d is the full price of a trip.
In this study, we assume that the demand function is log linear. Formally,
(4) r* = exp(β0 +β1w+β2pr +β3q).
In our econometric model below, r * is the expected number of trips. To estimate the
coefficients in equation (4), it is necessary to ask a sample of visitors to report the
number of trips they took in a specified period (year or season), cost per trip pr, plus w, y,
and other individual characteristics that might affect the demand for visits to the site.
Since q—the quality of the site—does not change over time, to estimate the
coefficient on q, β3 , we devised a hypothetical program that would deliver an
improvement in q, and asked our respondents to tell us how many trips they would take if
10 This model further assumes that travel time and time spent at the site are exogenous, that there is no
utility or disutility from traveling to the site, and that each trip to the site is undertaken for no other purpose
than visiting the site. It also assumes that individuals perceive and respond to changes in travel costs in the
same way they would to changes in a fee for being admitted to the site (Freeman, 2003). Finally, the model
assumes that work hours are flexible.