Methodology

The Rosés-Wolf database provides GDP and population data for 173 regions in 16 European countries. It is the closest available match for this project. This sheet supplies the GDP figures used to calculate regional GDP per head in 2011 international dollars. I use the Spanish regions and retain each observed value as an anchor point.

This table supplies the population figures for the Rosés-Wolf calculation. GDP is reported in millions and population in thousands; after adjusting for that difference in scale, dividing one by the other gives dollars per person.

The Maddison Project compiles long-run growth and income data for 169 countries. Rosés-Wolf uses it for national figures. Here it provides the baseline for EU member states through 2022, as well as Spain’s national figure, which is used to scale Eurostat’s regional data.

I use this table only to calculate a population-weighted average for today’s 27 EU member states. Maddison does not provide that average as a ready-made series. Coverage becomes complete in 1985; before then, the countries included vary with data availability and are identified in the generated output.

Eurostat publishes this series as part of its GDP and main components dataset. I use its real growth rates per head, not its absolute levels. Data for 2015–22 are used to calibrate the recent difference between Eurostat and Maddison; figures for 2023–25 extend the country series, the European average and Spain’s national series, which anchors the regional estimates.

Eurostat’s nama_10r_2gdp table reports regional GDP at current prices. Its Spanish figures come from the national statistics office, INE, so the two sources are effectively the same. I use the PPS figures to track each autonomous community relative to Spain and to extend the regional series through 2023–24, rather than treating them as the final GDP series.

Funcas publishes GDP-per-head forecasts for every Spanish autonomous community. I use it only for 2025, because Eurostat’s regional series ends in 2024. Each community’s index is applied to the projected figure for Spain.

I compared each region’s consumer price index with the national index to see whether separate regional deflators were justified. Between 2002 and 2024, the Balearic index remained within 1% of Spain’s. Given that small difference, the method assumes uniform purchasing power across Spain.

These are estimates by an academic author, not official counts. They cover the period before systematic tourism statistics existed. I use only the estimates for 1900 and 1920 as early anchor points.

These estimates were reconstructed in the mid-1960s from contemporary hotel registers, port records and regional statistics. They cover 1930–65. Some refer to Mallorca alone rather than the Balearic Islands as a whole, an important difference when comparing the figures.

Valdivielso and Moranta combine arrival estimates from earlier research and regional statistics, updated with AETIB data. Where their figures overlap with AETIB/FRONTUR, the reconstruction gives priority to AETIB/FRONTUR. Their study therefore supplies most of the key observations from the start of the tourism boom to 1993, while later overlaps provide a useful cross-check.

This is the only modern primary statistical source in the arrivals series, but its methodology is not consistent throughout. Figures for 1998–2000 cover air arrivals; 2001–09 combines air and sea; 2010 marks the move to annual FRONTUR data; and INE/IBESTAT redesigned FRONTUR in 2016. AETIB/FRONTUR is the preferred source from 1998 onwards whenever another source also covers the same year.