Expense reports is a novel way of measuring cost of travel, but it is still prone outlier effects and will be skewed because of the volume and nature of travel. Some cities see a lot of conferences. Others get visits mostly from sales or relationship managers out to show someone a good time.

Most surveys of "most expensive" cities are pretty useless. There are so many variables and assumptions that have to be built in— currency (PPP results often vary from dollar or euro results), quality of life, duration of stay, location, local lifestyle/culture— that they are unreliable for planning.

In its defense, it does account somewhat for substitutions, where most other such surveys fail. If milk is expensive in Hawaii, I just don't buy it, so the higher price won't be reflected in my spending. And I'm not exactly concerned about budgeting for omiyage when doing rounds in Austin. But for an example more applicable to business travel, consider this:

If my company sent me to a trade show in New York, I'd probably be taking the subway a lot because, like many IT operations, we are lean on travel perqs (the big boss flies AirTran). In Santa Clara, though, relying on VTA buses to get from my hotel to the conference is not really realistic, so I'd end up renting a car. It might be a cheap car booked on Hotwire, but with gas, taxes, and insurance it would still be $50/day whereas I might have spent $25/day on subways and taxis in NYC.

Most surveys, on the other hand, try to compare costs based on a certain standard of living, so of course you're going to see Libreville and Fort MacMurray up there with Geneva. After all, replicating New York or London is going to require a pretty exclusive neighborhood, imported food and goods, and perhaps private provision of what are taken as public goods, like the electrical grid or potable running water.

That's still better than dealing with the cost of living as a local, of course: http://www.theatlantic.com/business/...locals/255741/