How to read a cost of living number

What our number covers

It is a monthly figure in US dollars for one person living modestly but not frugally. It assumes rent on a one-bedroom flat or a good studio in a normal residential part of the city rather than the cheapest possible room; food, with a mix of cooking and eating out; local transport; utilities and internet; and a coworking desk or the cafe spending that substitutes for one.

It does not cover flights, visa fees, health insurance, equipment, tax, debt at home, or anything you would spend in the same amount wherever you were. Those are real costs and for many people they exceed the difference between two cities on this list — which is the first reason a cost comparison between cities can be true and irrelevant at the same time.

Failure one: the number is a city, and you live in a neighbourhood

This is the largest single source of error and it is invisible in any dataset organised by city. The gap between a well-located one-bedroom and one twenty minutes further out is routinely thirty to forty percent of the rent line, which is itself over half the total. Two people can both live in the same city, both live reasonably, and differ by more than the gap between two cities on our list.

We can show you the same effect at country level, where we do have the data. These are the countries in our dataset with at least four cities and the widest internal spread:

Internal cost spread within a single country.

If a country can contain that much variation, so can a city. Use the number to compare cities against each other; do not use it to predict what you personally will spend.

Failure two: the figure was collected from people unlike you

Every cost-of-living dataset in existence is built from someone's spending, and whose spending it is determines what comes out. Crowdsourced datasets skew toward whoever fills in surveys, which in nomad circles means younger, shorter-stay, more likely to be in a serviced apartment paying a foreigner rate. Expat-focused datasets skew the other way, toward long lets negotiated in the local language at local prices.

The practical consequence is that the same city can honestly be quoted at $900 and $1,500. The first is what it costs someone who has been there eight months and signed a twelve-month contract; the second is what it costs someone arriving for six weeks on a short-let platform. Both are true. Ours sits nearer the settled end, because that is the case a nomad choosing a base is actually planning for.

Failure three: currency movement, silently

A dollar figure for a city in Turkey, Argentina or Egypt is a snapshot of an exchange rate as much as of a local price level. Where local inflation outruns the currency's fall, the dollar cost rises even though nothing about the city changed; where the currency falls faster than prices rise, it drops. Either way the number moves without anyone spending differently.

This is why year-on-year comparisons of cost figures across currencies are close to meaningless, and why a two-year-old cost table for a high-inflation economy should be treated as historical rather than current. Ours are dollar figures and carry the same caveat. For the eurozone, Japan or the Gulf states, where the currency is stable against the dollar or pegged to it, the figures age much better.

How to use the number properly

• As a ranking, not a budget. The claim “this city is cheaper than that one” is far more robust than the claim “this city costs $1,100”. The ranking survives all three failure modes above; the absolute figure survives none of them.
• With the components, not just the total. A city that is cheap because rent is cheap is a different proposition from one that is cheap because everything else is. Look at what the city page says about internet, safety and walkability alongside the cost — those are covered in what our city score measures.
• Against a visa income floor, if one applies. The most useful thing to do with a cost figure is to divide a visa's income requirement by it, which tells you how much margin the government is actually demanding. We publish that ratio for every country with both figures on the income-versus-cost page, and it ranges from under one times to over five.
• With a fifteen percent band around it. Not a point estimate. If two cities are within a hundred dollars of each other in our data, treat them as the same price and decide on something else.