Effects of EU-funded tourism promotion in the Cap of the North

Abstract Information: The Cap of the North, also known as Lapland, covers the northernmost parts of Finland, Sweden, and Norway. It is the home of the Sami, a Finno-Ugric-speaking indigenous people with an estimated population of 80,000. The area is very sparsely populated: the population density is below two inhabitants per square kilometer, roughly comparable to Alaska. Besides traditional reindeer herding, the livelihoods are few and far between. In order to rectify this, the European Union has since 2015 funded a multi-million project aiming to increase the number of international tourists in Lapland. We evaluated the effects of the program using detailed panel data on registered guest nights in hotels and other accommodations. Overall, the results are ambiguous: using other regions (and in Finland, also municipalities) as a control group, the number of overnight stays has increased in the treatment regions after the project started, but the trend has been similar for several decades. That is, we cannot disentangle the effect of the program from the long-run trend of increasing international tourism in Lapland. A further complication is that Lapland's international tourism promotion campaigns originated already in the 1950s. However, we carried out a re-evaluation of the data in 2023 using a new method, synthetic difference-in-differences, which yields results that are different from the original audit report. The routine follow-up study will have some actual content in this case.

Relevance Statement: This case highlights the inherent difficulties in evaluating regional development programs. If the treatment and control regions' outcome variables have differing trends ex ante, it may be impossible to separate the effects of a particular project from the long-term trend ex post. This issue remains despite the fact that we have high-quality data available and use modern econometric methods (in this case, fixed effects regression and event study methods). Possible discussion points include: Could we find better control groups? Where exactly? Could qualitative methods ("stories") produce more useful results? How? Note that after submission, we tried out a new method, synthetic difference-in-differences, which looks very promising.