How to compare prices across countries and markets

Step 1: make sure it is the same product

Most false price gaps are product differences. A model sold under one name in two countries can differ in storage size, bundle, warranty or condition; a hotel “double room” can differ in board and cancellation terms; an air fare can be a different fare brand.

Eurostat's method for purchasing power parities states the principle clearly: prices are collected for products that are both comparable across countries and representative of what is bought in each. For a commercial comparison, that means matching on identifiers first and attributes second.

• Match on GTIN, manufacturer part number or the site's own product ID where it is shared across editions.
• Compare unit prices when pack sizes differ.
• For services, fix every condition: itinerary and fare brand for flights, room type, board and cancellation for hotels.

Step 2: control the point of sale

A market is not one switch. Sellers infer it from several signals, and a comparison needs each one set on purpose and written down.

Verify the exit before you trust it; testing a proxy's location shows how. For the airline version of this problem, see why flight prices differ by country and point of sale.

Step 3: synchronise the timing

Prices and exchange rates both move during the day. Collect all markets for one product within the same few minutes, store observed_at in UTC, and run the comparison on the same weekday across weeks. For travel, also hold the booking window constant — 30 days before departure in every market, not 30 days in one and 29 in another.

If a source changes prices several times a day, take several synchronised rounds and compare medians rather than single readings.

Step 4: normalise currencies with a reference rate

Use one published reference rate per day, applied after collection. The European Central Bank publishes euro reference rates for its list of currencies at around 16:00 CET on working days, based on a daily concertation between central banks at around 14:10 CET. The ECB publishes them for information purposes only — they are a measurement standard, not a rate anyone trades at.

• Keep the original amount and currency on every observation; store converted values in a derived column or view.
• Fix a rule for which day's rate applies and document it — for example, the latest rate published on or before the observation date, so weekends use Friday's rate.
• Do not use a site's own converted price as the comparison base. It carries the seller's exchange margin, which is a separate subject — see booking currency markups.
• Report the gap in both ways: in the reference currency, and in local currency against the reference rate, so readers can see how much of a gap is the exchange rate moving.

Step 5: put taxes on the same basis

Under the EU's Price Indication Directive (98/6/EC), the selling price shown to consumers is the final price including VAT and all other taxes. In the US, many prices are shown before sales tax, and even the FTC's fees rule for lodging and live-event tickets allows government taxes to be left out of the headline total as long as they are disclosed before payment.

So strip VAT from EU prices at the rate that applies to that product in that country, or estimate tax on the US side — and say which you did. For air fares in the EU, Regulation 1008/2008 requires taxes, airport charges and other charges to be specified alongside the final price, which lets you separate the fare from the tax stack.

Reading the result

• Express each gap as a percentage of the reference market, and publish the median of repeated observations, not the extreme.
• Separate effects: product mismatch, tax basis, exchange rate and genuine price difference, in that order.
• Watch for availability effects. A market that shows a higher price may simply have sold out of the cheaper variant.
• Keep the raw observations. A comparison that cannot be recomputed with a different exchange-rate rule is hard to defend.

Running this continuously is a price monitoring workload; aggregated into indices it becomes online price index data. Collecting from inside each market usually means residential proxies with country targeting.