27
How do we extrapolate these figures to the population of domestic visitors? We
wish to emphasize once again that by intercepting people on-site, we end up observing
only people who have taken at least one trip for this year, and visitors that tend to be
more avid than the average. First, we need to calculate the appropriate welfare measure
for the average visitor in the population, which must then be multiplied by the number of
people that visit each of the selected sites.
For illustrative purposes, we compute out-of-sample predictions for Garni using
specification I in table 8, the monthly household income of the average Armenian—
which we conservatively estimate to be 21.28 US dollars, or a little less than half that in
our sample22--and further assume that the cost of a trip to Garni for the representative
domestic visitor is the same as the average in our sample. This time, since the prediction
is for the population of domestic visitors, and not for the sample we intercepted on site,
the appropriate formula for the consumer surplus is λi /(-β2 ) = exp(xβ1 + pβ2)∕(-β2),
which pegs the value of access for the representative domestic visitor at 11,742 AMD a
year.
We know from a companion CV survey of Armenian households that 11.80% of
the respondents have visited Garni in the last year. Assuming that this percentage mirrors
the population share, then the total consumer surplus is 3.1 million (population) × 0.72
(percentage adults in the population) × 0.1180 × 11,742 AMD=3,093 million AMD (=6
million US dollars per year).
22 To arrive at this estimate, since the average annual income of an Armenian household was 1045 US
dollars in 2001, we increased this figure by 10% to conservatively account for growth, divided it by 12, and
further divided it by 4.5, the number of household members of the average Armenian households.