ABSTRACT
Introduction This chapter sets out to explore the determinants of population growth in Europeancitiesandhow those forcesdifferandcoincidewithfindings for the USA. The results illuminate both the extent to which European integration has had an impact on population movements and also suggest important differences between the situation in Europe and that in the USA. In Europe, urban population growth seems likely to be a rather imperfect signal of changes in welfare in cities. This is apparently much less true within countries within Europe despite the notorious geographical stickiness of people in Europe. Once one has controlledforothersystematicinfluences,urbanpopulationgrowthisnotaffectedatall by a city having a better climate than the European Union (EU)-12 as a whole – that is the 12 countries that formed the European Union until its enlargement in 1996toincludeAustria,FinlandandSweden–butisstronglyinfluencedbyit having a better climate than its country. The results strongly suggest that the model widely used in the USA – both in the quality-of-life literature (Roback, 1982; Blomquist et al., 1988; Gyourro and Tracey, 1991; Gyourro et al., 1999) and in the analysis of urban growth (Glaeser et al., 1995; Rappaport, 1999) – has application, but it has to be fundamentally modified for a European context. The evidence presented here suggests that labour in Europe is geographically immobile and, in as far as there is mobility in search of quality of life, it is a within-country phenomenon. The central assumption of perfectly mobile factors and the equalization of real marginal returns across cities explicit in the US models, therefore, cannot reasonably be maintainedintheEuropeancontext.Thatsaid,however,itisinterestinghowsignificant climatic differences between cities within Europe appear to have been in determining differential patterns of urban population growth over the past 20 years. In this sense, the results reported here can be seen as complementing those reported for Germany by Rehdanz and Maddison (2004). The impact of climatic variables is analysed within a wider examination of ratesofurbanpopulationgrowth.Otherinfluentialregionalfixedeffectshaveto be allowed for. These have been modelled explicitly, however, since some of themareofinterestintheirownright.Inthecontextofourfindingswithrespect
to quality-of-life differences, the most interesting is the impact of systematic regional differences in a measure of the economic gains from European integration.Hereonefindsthatchangesinacity’seconomicpotential–ameasureof the accessibility of incomes at any point – associated with the integration of the EUand falling transport costs, havehada statistically significant influenceon differential rates of population growth across the whole space of the EU-12. This influenceonthespatialpatternofincomes,therefore,doesappeartooperateat anEUlevel,althoughinstatisticaltermsitisnotsoinfluentialorsosignificant as climatic variation within countries. The idea that differences in climate might influence long-run population movementhasalonghistory.OneofthefirsttoinvestigateitwasGraves(1976, 1980,1983,2003a,b;GravesandLinneman,1979).Hisinterestwasintheinfluenceof climaticdifferencesonmigrationflows,however, and itwasnotuntil the work of Roback (1982) that climatic differences were put into the more general framework of a compensating differentials model from which inferences might hope to be made about regional and urban differences in the quality of life. If people are perfectly mobile and vote with their feet, then it follows that in equilibrium no one could move without being worse off. In such a model, differences in climate will be just one source of differences in quality of life. Other features of the natural environment such as topography, scenery, and major natural attractions including ski slopes and access to large bodies of water will alsohaveaninfluence.Sotoowillfeaturesofthemanmadeenvironmentsuchas conserved open space, the quality of the housing stock, the quality of local public goods, and, of course, jobs, wages and job availability. Such differences will be capitalized into land values (and hence house prices) and reflected in labour markets. There they may appear as differences in wages for given occupations, jobs and qualities of labour; or to a varying extent, depending on the way in which wages are set or the ease with which they adjust to changes in the balance of supply and demand, such differences may appear as differences in expected earnings. Observed real wages at any time may not vary for a given job and quality of labour but expected real earnings can vary allowing for the probability that an individual with a given stock of human capital is actually employed. Put more simply one may chose to live in a nicer place accepting that one will have lower expected money earnings – whether because real wages are loweroritishardertofindajobforgivenskillssoonewillbeemployedfora lower proportion of the time. IntheUSA,thereseemstobesignificantpowerinthismodel.Variousstudies (e.g. Hoehn et al., 1987; Blomquist et al., 1988; Gyourro and Tracey, 1991) have found convincing evidence that interregional environmental differences are valued and capitalized in the way predicted by the compensating differentials model. However, as Gyourko et al. (1999, p. 1415) conclude:
recentwork…reportsthepresenceoflargecity-specificerrorcomponents in the underlying … estimated … local trait prices … it turns out that … the level of imprecision is such that much better descriptions of local amenity
and fiscal conditions, plus superior controls for housing, worker and job quality are needed… .
