It’s talking about cost mainly in terms of land value. If the parking spot was lost and had to be rebuilt in a similar location, what would that cost? In the centre of a city the cost of tarmac isn't going to be the dominant issue!
Using city population for a per capita calculation seems perfectly reasonable as long as that’s understood, even in most tourism cases. For instance it’d be a perfectly reasonable economic calculation to look at the tourism income brought into a city per head of the resident population, it gives an idea of how valuable that is to an ordinary person, even if they don’t work in the industry themselves. The calculation being made here is similar.
This only dramatically falls down where the land value is dominated by car-borne tourism value, for instance for a stop off on a long distance road route, reducing parking spaces by 50% might well reduce income by 50%. But that doesn’t apply in the central areas of most cities.
Perhaps the largest issue is with the statistic in the study named "parking cost per household". This figure suggests strongly this cost is somehow a cost which is a burden to each family's taxes or own money. However, this study includes all parking spots. This leads me to believe the majority of these spots are in parking garages which in almost all cases are privately funded (and those public may still have fees which pay for themselves).
First, it tries to use city population as a metric to get per capita cost. It seems to suggest that households are on the hook for a significant portion of these spots. First, a city like Jackson Wyoming is mostly tourists and therefore the spots pay for themselves easily. Second, most spots are privately funded.
Second, the estimate of "cost of replacement" is pretty silly. It used 7100 as an estimate per spot. I doubt the large parking spots in a ski lodge base are going to cost anywhere close to 7100. And even if a spot in a city actually cost 7100 to pave, if it then lasted 20 years is the cost anywhere close to what the article suggests? Doubtful.