# Travel Cost Method - how to treat people who live at a site

We are about to do a survey using the individual travel cost method to value a harbour, located in a fishing society, which is visited by tourists each season. My question is - should we include in the analysis people who live in this society and who, therefore, have zero travel costs to go there? Or should we skip them in the econometric analysis?

• Welcome to the site. Could you clarify "people who live in this society"? If you don't include them, who would you include? Or did you mean people who live at the harbour, as contrasted with other members of the fishing society who use the harbour but live some distance from it? Jun 4, 2020 at 11:28
• I mean people who live in the (small fishing) community where the harbour is situated and who can therefore get to the harbour by foot or by bike. Jun 4, 2020 at 20:58
• Is it just the small fishing community who use the harbour, or are there other people who travel longer distances to use it? Jun 4, 2020 at 21:09
• There are plenty of tourists coming each season, otherwise we wouldnt be able to use the travel cost method. The idea of dropping people who walk to the harbour is that they have no travel costs to use in the analysis. Jun 6, 2020 at 6:26
• I have edited your question to include details from your comments, but please feel free to edit further if not appropriate. I'm now preparing an answer. Jun 6, 2020 at 10:19

Short answer. It is perfectly correct, and probably simplest, to exclude from the analysis the members of the fishing society, provided that you present your result as “the tourist value of the harbour” or “the value of the harbour to tourists” (and not as the total value of the harbour).

Long answer. The individual travel cost method involves two stages. In the first stage, the data collected is used to estimate a trip-generating function, something like:

$$V_i = f(TC_i,\pmb{A}_i)$$

where $$V_i$$ is the number of visits by individual $$i$$, $$TC_i$$ is the round-trip travel cost from $$i$$’s location of origin, and $$\pmb{A}_i$$ is a vector of characteristics judged to be relevant, eg age, income. In the second stage, a demand curve for visits to the site is estimated on the assumption (call this the key assumption) that individuals would respond to different hypothetical prices added to their actual travel costs in the same way that, according to the trip-generating function, they respond to different travel costs. The site value is then taken to be the consumer surplus, which if access to the site is free is the whole area under the demand curve.

The fact that travel cost for some individuals is zero is not in itself a reason for excluding them from the analysis. One reason is that their data is relevant in estimating the trip-generating function: a data point with $$TC_i = 0$$ is still a data point, and to exclude such points would shift the line of best fit. Also, the second stage methodology means that, other things being equal, individuals with zero travel costs make a disproportionately large contribution to the total consumer surplus (because the sum of a hypothetical price and zero is less than the sum of that price and a positive travel cost, so that the number of visits associated with the former will be greater).

However, there are other reasons for excluding members of the fishing society. The origin locations of the tourists, it can probably be assumed, are to a large extent randomly spaced. It seems unlikely however that this is so for the fishing society. Very likely they live at the harbour because, given their dependence on fishing, they have chosen to do so. If so, one of the assumptions of the travel cost method is undermined. Freeman, Herriges & Kling (2014) list as one of the assumptions of the method (for the case of recreation but the point is more general):

… it is assumed that the individual’s choice of where to live (which is one determinant of the cost of a trip to a recreation site) is independent of preferences for recreation visits. If people choose residential locations so as to be near preferred recreation sites, then the price of a visit is endogenous. (p 273)

As a consequence, including the fishing society would tend to result in a biased valuation.

A second reason for excluding members of the fishing society (which would apply even if their travel costs were positive) relates to the key assumption above. The assumption may be reasonable for a fairly homogeneous group such as tourists, or indeed for the fishing society considered on its own. But if one puts together two such very different groups, the assumption becomes much more dubious. Suppose, for example, that the incomes of the tourists are typically much larger than those of members of the fishing society. In that case the former might respond to a certain price by visiting less frequently, while for the latter the price might be so large in relation to their incomes that they could not afford even one trip. Thus a ‘hybrid’ demand curve, based on data for the two groups combined, would be unlikely to give a valid consumer surplus estimate.

• Thanks, this was really helpful Sep 18, 2020 at 16:08
• @Kjisti If so you might consider upvoting and/or accepting my answer, as is normal practice on this site. Sep 18, 2020 at 21:53
• I did upvote it, but aince I have made too few comments on the site, the upvote is not publicly displayed. (I was told) I couldn' find where to accept the snswer, but I will look for it again. Edit: just found it. Sep 20, 2020 at 11:27
• I overlooked that you need at least 15 rep to be able to upvote. Sep 20, 2020 at 11:52