Deconcentration, Counter-Urbanization, or Trend Reversal? The Population Distribution of Switzerland,

Transcription

1 Socio-Econ. Plann. Sri. Vol. 26, No. 2. pp , 1992 Printed in Great Britain. All rights reserved /92 $ Copyright Pergamon Press Ltd Deconcentration, Counter-Urbanization, or Trend Reversal? The Population Distribution of Switzerland, PETER V. SCHAEFFER School of Architecture and Planning, University of Colorado at Denver, Denver, CO , U.S.A. Abstract-This paper provides a detailed description of changes in the population distribution of Switzerland from 1900 through Switzerland is a fine candidate for such a study since it has excellent population data and has not suffered major disruptions from wars as have other European nations. The results show that Switzerland has experienced deconcentration from 197OG1980. There is also weak evidence for counter-urbanization. The data are not sufficient, however, to conclude that this most recent development is atypical. The Hoover index and the measure for counter-urbanization both seem to be sensitive to external developments in the economy and in politics. INTRODUCTION The trend away from large urban centers during the 1970s in industrialized countries has been labelled deconcentration, counter-urbanization or trend reversal. Many authors have used these terms interchangeably, though each can mean something different, as Frey [14] recognizes. Yet, the link between regional redistribution and the urbanization dynamic would appear to be more tenuous for the new post-1980 shifts. Can we necessarily conclude that the recent growth in Vining et al.-defined core regions is associated with the return to more traditional urbanization patterns? Or can the recent core region growth coexist with the less concentrated settlement patterns that emerged during the 197Os? ([14], p. 261) It is possible that continued urbanization coincides with deconcentration; the presence of deconcentration does not imply counter-urbanization and the two terms should therefore not be used interchangeably. If one accepts the difference between deconcentration and counter-urbanization, the meaning of trend reversal becomes ambiguous. Counter-urbanization is the reversal of the long trend towards more and larger urban settlements, and deconcentration is the reversal of the long trend towards greater population concentrations. Since counter-urbanization and deconcentration are not the same, the two trend reversals are different. In this paper, trend reversal will be defined to mean neither deconcentration nor counter-urbanization, but have a meaning of its own. The purpose of this paper is to define the three terms and to apply their definitions to the study of the population distribution of Switzerland from 1900 through We will pay particular attention to the changes between 1970 and Switzerland is an excellent candidate for study over a long period of time. The quality of its census data is high and the country has not suffered major population dislocations from wars. We offer a more detailed description of the Swiss case than is at present available in the literature. This will be particularly useful to those engaged in comparative studies but should be of interest on its own merit. Switzerland is an important industrialized country, has one of Europe s largest immigrant populations, and ranks among the largest trading nations. Many authors have carefully described changes in population distributions in other industrialized countries (e.g. Casetti [8]; Champion [9]; Fielding [ 121; Frey and Speare [16]; Jones et al. [20]; Kontuly and Vogelsang [23]; Kontuly et al. [24]; McCarthy and Morrison [27]; Rogerson [28]; Vining and Kontuly [30,31]; Vining and Strauss [32]). However, the small size of Switzerland permits an overview of the population distribution relative to major transportation links, 89

2 90 PETERV. SCHAE~R topography and other characteristics. This helps the derivation of specific hypotheses (see Frey [15, 131, particularly pp of Ref. [13]). The empirical component of the current study is descriptive and exploratory. Rigorous tests of hypotheses for Switzerland require additional information and analysis. Deconcentration DEFINITIONS For the definition of concentration/deconcentration, we follow the literature and use the Hoover index. The index ranges from 0 (perfectly even distribution) to 100 (perfect concentration). Deconcentration results in a decrease in the value of the Hoover index. N is the number of regions. P, denotes the proportion of the population in region i at time t, and Ai is the proportion of the country s area covered by region i. Counter-urbanization Since a major characteristic of an urban area is high density, counter-urbanization exists if there is a negative correlation between population density and a region s increases in population per km*. This definition is compatible with that proposed by Champion [lo] and Fielding [12]. Fielding uses almost the same definition in his Table 3 [12], where a negative relationship between net migration and population density is defined as counter-urbanization. The relationship between the size of a settlement or region and net migration has also been employed in the literature [12,23]. This seems less satisfactory. Urbanization is not necessarily a progression towards ever larger regions or settlements but is a social and economic phenomenon poorly characterized by that definition. We prefer the use of density. Trend reversal Trend reversal exists if the population distribution becomes more similar to that of the past. Our measure compares the percent of a country s population living in a region in a base year and an ending year, taking the average of their absolute differences. Trend reversal exists if the sum of this average over all regions, TR,,,, declines. For our purposes, Population in region i in base year (0) TR,,, = $ L i= I N Population of Switzerland in base year (0) Population in region i in ending year (1) - Population of Switzerland in ending year (1) where, as before, N stands for the number of regions. Accommodating quickly changing population distributions can be expensive. Governments may therefore wish to control changes. In the Swiss Federal Planning Law [7], one of the stated goals is the promotion of a reasonable decentralized population and economy (Article 1.c). When towns decline, a considerable infrastructure is often left underutilized. The critical size for the economical provision of important public services may no longer be reached. The Swiss federal government has designed programs to help such areas, at considerable cost. Trend reversal could mean either that such policies have worked or have never been needed. The political pressure on governments to do something for regions who have recently declined is usually larger than the pressure to help those who have never prospered. DATA The population data used in our study are from the decennial censuses of Switzerland from 1900 to 1980 [2,5]. The war-time census of 1940 was delayed and taken in Population figures after

3 Population distribution of Switzerland are available from other sources but will not be used. They are less reliable than the census information and have been collected differently. In particular, they include seasonal workers which the censuses do not. Since Switzerland has a large number of such workers who are permitted to stay for, at most, nine months per year, the two sources of population information are not compatible. The geographic subdivisions used in our analysis are the cantons and the districts (Bezirke), respectively. Cantons are the functional equivalent of states, and the districts functions are comparable to those of a county in the United States. By using cantons and districts we get some sense of the possible effects of aggregation on the results [22]. The Swiss census does not define commuting or economic functional regions and the information available does not permit the construction of such areas by the author. The use of administrative regions has precedent in the literature [17]. For a comparison of population distributions over a period of eighty years, administrative regions are preferred over other functional regions that have expanded greatly with improved transportation and communication networks. Comparisons based on such regions are therefore difficult to interpret. To overcome weaknesses associated with using administrative regions, we use different levels of aggregation and different measures. This reduces the possibility of errors of interpretation. The choice of the appropriate geographical subdivision is not just a technical problem but depends on the purpose of the study. In a multi-cultural federal country like Switzerland, the nature of changes in the population distribution among political regions is of great practical interest. The goal in the planning law, mentioned above, is motivated by the political need to maintain a reasonable balance between different parts of the country. Switzerland is 41,293 km2 (15,734 sq. miles) in size (Table Al). The average size of the cantons is 1,588 km* (605 sq. miles). With an area of 7,106 km2 (2,708 sq. miles), Grisons is the largest canton. Basel-City, with an area of only 37 km* (14 sq. miles), is the smallest. The average size of the districts is 224 km* (86 sq. miles). The Canton of Uri also fulfills the functions of a district. It is the largest, with an area of 1076 km*. With an area of 6.2 km*, Solothurn is the smallest district. For this study, the Canton of Geneva has been divided into the City of Geneva and the rest of the canton, though the canton also serves as the district. At the district level we therefore have 185 regions. A map of Switzerland is included in the appendix (Fig. Al). Deconcentration TRENDS IN POPULATION DISTRIBUTIONS: CANTONS The values of the Hoover index show that the concentration of the population of Switzerland increased steadily between 1900 and During the ten years from 1970 to 1980, however, the Hoover index decreased (column 1 in Table 1). In a small country there is a greater danger that one or a few regions have a disproportionate weight in the calculation of the Hoover index. Zurich was the most populous canton in 1980 (1,122,839 inhabitants, 17.64% of the Swiss population). Bern was second with 912,922 inhabitants (14.33% of population). But while Bern s share of 14.65% of Switzerland s area was almost the same as its population share, the Canton of Zurich occupies only 4.19% of the land. Because of Table I. Hoover indexes for cantons. 190&1980 I Hoover index, Hoover index, Hoover index, Hoover index, Year all cantons all cantons but ZH all cantons but ZH, BE all cantons but ZH, BS I Source: Based in information from [2,5]. Note: BE = Bern, BS = Basel-City, ZH = Zurich.

4 92 PETER V. SCHAEFFER the weight Zurich exercises on the Hoover index, we recalculated the Hoover index for all cantons except Zurich (Table 1, column 2). Without Zurich, the Hoover index increased from 1900 to During and , however, the rest of the country experienced deconcentration. Several cities lost population between 1920 and 1930, including Geneva and Neuchatel. St Gall lost population from 1910 (80,072) until 1941 (66,590). From 1941 to 1970 the Hoover index increased, and from 1970 to 1980 we again observe deconcentration with a larger decrease in the Hoover index than with Zurich included. The deconcentration results for and were more pronounced when we subtracted both Zurich and Bern, and Zurich and Basel-City, respectively (Table 1, columns 3 and 4). Bern and Basel-City were considered along with Zurich because Bern was the most populous canton in 1900 and was still second in It is also the second largest canton with an area of 6,069 km2 (2,305 sq. miles). The Canton of Basel-City consists of little more than the city itself. Its population share of 3.45% in 1980 far exceeds its share of the Swiss territory (O.O9O/). Both cantons have, therefore, a potentially large impact on the value of the Hoover index. For the period , however, the results are very similar, with and without the deletions. To illustrate how individual cantons fared, we calculated the difference between the cantonal and national growth rates for seven relevant decades/periods (Table 2). Of the large cantons, Zurich grew at a rate consistently above the national average until Only during the last period ( ) did its growth rate fall below the national average. While the country grew at a rate of 1.53%, the population of the Canton of Zurich increased by 1.36%. The Canton of Bern did not do as well. From a population of 587,100 inhabitants in 1900, when it was the most populous canton, it grew to 912,022 in 1980 (+55.34%), well below the national growth for of 92.01%. It achieved growth above the national average only from 1920 to Of the other populous cantons, Vaud (third most populous in 1900, Table Al), experienced growth below the national average until During 196&70 and , the canton grew quickly. The increase of 87.91% for the eighty years from 1900 to 1980 is still below the national average, however. Basel-Country achieved the most spectacular growth ( % between 1900 and 1980), even though it experienced two periods of below average growth. Zug also grew at a rapid pace ( %). Population growth in both cantons continued at rates far above the national average during 197&80, although Basel-Country s growth had slowed considerably compared with the previous twenty years. Neither canton has a large city but both are located next to cantons that do. Basel-Country s growth is, at least in part, a consequence of the suburbanization Canton Zurich 3.70 Bern LIKerIle 0.92 Uri Schwyz Obwalden Nidwalden Glarus Zug Fribourg Solothurn 2.95 Basel-City 7.90 Basel-Country SchatTbausen Appenzell-IR Appenzell-AR Saint Gall 7.81 Grisons - I.20 Aargau Thurgau 5.96 Ticino Vaud Valais Neuchatel Geneva 3.61 Jura - IO.91 National growth rate Source: Calculated from [5]. Table 2. Deviations from natmnal growth rates by canton, (in percent) 19Ol&lO 191&20 192&30 193&4l I95&60 196& I I I I I ~ I IO IO IO I I.53

5 Population distribution of Switzerland 93 of the population and industry of Basel-City. Two national highways link the canton with the rest of the country on the one hand, and to the important export point of the city of Base1 on the other. Base1 is the major export point to Germany, Switzerland s largest trading partner. Together with Geneva, it is also the major point for exports to France. It is Switzerland s only port city, connecting it to the Atlantic Ocean via the Rhine River. Zug is a neighbor to both Zurich and Lucerne. Though its largest city had only 21,609 inhabitants in 1980 (down 5.9% from 1970) it had the second-highest per capita income. Its favorable tax rates have attracted a large number of corporate headquarters. The trip to Zurich by train takes less than 30 minutes and there are frequent connections. Zug-Lucerne can be traveled by train in about 20 minutes. A national highway links Lucerne and Zug and gives the canton convenient access to the very important St Gotthard road which connects northern and southern Europe. The planned national highway between Zug and Zurich had not been built by The two Appenzells had smaller populations in 1980 than in 1900 (Table Al). Their growth was below the national average in every period (Table 2). Both cantons are rural agricultural cantons with small farms and without large industries. The population of the Canton of Jura increased at a rate of only 13.10% from , far below the national average. Its population was 57,460 in 1900 and 64,986 in Other cantons with low growth rates are Neuchatel(+25.41% growth from ) and Glarus ( %). Neuchatel is a manufacturing canton. Glarus is more rural/agricultural than Neuchatel, but it also has important manufacturing industries. Table 3 shows that some urban cantons with high densities (Aargau, Basel-Country, Geneva, Zug) experienced strong growth during the 1970s. Of cantons with lower than average densities, only Nidwalden realized any such change. On a per km2 basis, Zurich s growth still ranked sixth in the country. Table 3. Density and population growth per square kilometer, 197&1980 (1) (2) (3) Density Deviation of density Growth Canton in 1970 from national average per km Zurich Bern Lucerne Uri Schwyz Obwalden Nidwalden Glarus Zag Fribourg Solothurn Basel-City Basel-Country SchaRIausen Appenzell IR Appenzell AR Saint Gall Grisons Aargau Thurgau Ticino Vaud Valais Neuchatel Geneva Jura National average 152 Correlation between (2) and (3) &20 192& & All cantons Standard deviation of (2) Standard deviation of (3) BS, BL combined Standard deviation of (2) Standard deviation of (3) Note: BS = Base&City, BL = Basel-Country. Source: Calculated from [2,5].

6 94 PETER V. SCHAEFFER Counter-urbanization For a check of the existence of counter-urbanization, we calculated the correlation coefficient between a canton s density, and population growth between 197&80 per km2 (Table 3). The correlation coefficient is very highly positive until the period During that period it was only It was again positive (0.727), however, during 196&70. The next period, 197&80, brought a dramatic reversal as the correlation coefficient fell to , seeming to strongly support the counter-urbanization hypothesis. Basel-City (1980 population share of 3.45% and share of the Swiss territory of 0.09%), however, is an exceptional case. The standard deviation in 1980, across the cantons, of the deviation of the density from the national average was persons/km2. The population density of Basel-City was more than five times that number. A reliable rule of thumb identifies an outlier as a case more than three to four standard deviations from the mean [1 11. Basel-City s density was much larger than the standard deviation in each of the seven periods. If we, therefore, regard it as an outlier and delete it from the observations, we obtain a correlation coefficient of Deletion of this outlier is not a satisfactory solution here, but it shows how a single case can obscure what is happening in the rest of the country. For a better alternative, we added Basel-City to its natural (and historical; the two Basels once were one canton) hinterland, Basel-Country. The correlation coefficient then was If we look at the results with the two Basels combined, we notice that large variations in the correlation coefficient were not uncommon during the study period. Although the value for is the smallest, it is not possible to say that it signals a new development. These results thus do not support the counter-urbanization hypothesis. Trend reversal The calculations of the trend reversal indexes, TR, show that the population distribution changes of the 1970s were unusual in historical comparison only because they were the second smallest in this century. Except for the nine years , which include most of World War II, every other decade saw greater change. The slowdown in the magnitude of change compared with the previous decade may also already reflect the war that started in The available data do not permit us to check this contention. Surprisingly, the decade of World War I (191418) does not show the same effect. If we consider the districts, however, this changes (see analysis of districts, below). A comparison of the index, TR, from census to census, and those measured against the base year, 1900, also show no trend reversal as defined in this paper. In every decade, the population distribution departed farther from the pattern present in the 1900 census, although the difference between T&x,,,,,, and T&M).1980 is not very large. As the numbers in Table 4 show, the greatest changes occurred in the decades and 195&60. Deconcentration TRENDS IN POPULATION DISTRIBUTIONS: DISTRICTS The analysis for districts supplements that for the cantons and pinpoints more accurately the areas where growth was strongest from 1970 to For that purpose, let us first look at the locations of the ten districts that experienced the largest absolute and relative population gains , respectively (Table 5). The ten districts gaining the largest number of inhabitants include either a medium size or large city, or they are located next to a district that does. They account for all districts that gained more Table 4. Trend reversal, TR,., + IO cantons By decade 190&10 191& &4l &60 196&70 197&80 TR I I II92 Base year &10 190&20 190&30 l90&41 190&50 I 90@60 190&70 I90&80 TR Source: Calculated from [5].

7 Population distribution of Switzerland 95 Table 5. Winners and losers, 197&1980 (Districts) The top ten winners (in absolute numbers) The top ten losers (in absolute numbers) Canton of Geneva w/o City 34,554 Zurich -47,862 Uster (Zurich) 14,487 Basel-City -31,030 Dielsdorf (Zurich) 12,555 Geneva (City) -17,113 Arlesheim (Basel-Country) 11,477 Biel/Bienne (Bern) - 10,179 Lugano (Ticino) 11,074 Lausanne (Vaud) - 8,029 Nyon (Vaud) 10,613 La Chaux-de-Fonds (NE) -5,137 Morges (Vaud) 9,719 Courtelary (Bern) -3,836 Horgen (Zurich) 9,206 Schaffhausen -3,377 Zug (Canton = District) 7,934 Saint Gall -3,079 Biilach fzurichl Le Locle (Neuchatel) -2,958 The top ten winners (in percent) The top ten losers (in percent) Nyon (Vaud) Gersau (Schwyz) Dielsdorf (Zurich) LaVaIke (Vaud) Echallens (Vaud) Val-de-Travers (NE) Fraubrunnen (Bern) Le Locle (Neuchatel) Merges (Vaude) Biel/Bienne (Bern) Affoltern (Zurich) Courtelary (Bern) Canton of Geneva w/o City Base&City Rheinfelden (Aargau) Grandson (Vaud) Uster (Zurich) La Chaux-de-Fonds (NE) - II.80 Lavaux (Vaud) Solothurn Notes: Name of canton in parentheses unless self-explanatory (e.g. Solothurn). NE = Neuchatel. Source: [7]. than 7,000 inhabitants between 1970 and Four of the ten districts are from the Canton of Zurich, all adjacent to the district that includes Zurich, Switzerland s largest city. The district Biilach is adjacent to both Zurich and the district containing Winterthur, the sixth largest city. Two districts are from the Canton of Vaud. Morges is adjacent to Lausanne (fifth largest city), while Nyon shares a border with the fast-growing Canton of Geneva. Arlesheim, the district from Basel-Country, is next to the Canton of Basel-City (second largest city). The district of Lugano includes the city of Lugano, with 27,815 inhabitants, the largest in the Canton of Ticino. The Canton of Geneva, without the city, and the Canton of Zug, as a single district, are also included. The twenty top districts include three more districts from the Canton of Zurich, three from the Canton of Bern, three from the Canton of Aargau, and one from the Canton of Lucerne. This illustrates that growth during continued to be highly concentrated and occurred primarily in the already densely populated Mittelland (Swiss plateau, approximately 30% of total area) and in those regions with a strong industrial or service base. Next, we look at the top ten districts in terms of growth rates 1970 to Eight are next to a district containing a large or medium sized city. The two exceptions are Nyon (Vaud) and Rheinfelden (Aargau). Nyon is crossed by the national highway between Lausanne and Geneva. Its largest city, Nyon (1980 population of 12,842, up 12.4% from 1970), is located on the highway. It is less than 25 km (15.4 miles) from the center of the city of Geneva. By train it takes 1.5 minutes from Nyon to Geneva and less than 30 minutes to Lausanne. Travel times by car are similar. The whole district had a 1980 population of 37,998. Rheinfelden s largest town, Rheinfelden (1980 population of 9456 or about one-third of the district s 29,374 inhabitants), is only 15 km (9.5 miles) from the center of the city of Basel. It is on the national highway leading into Basel. By train it is a 20 minute ride to Base1 with frequent service. The district s second largest town, Mdhlin (1980 population of 6,360), is less than 20 km (12 miles) from Basel. The town of Kaiseraugst is the third largest community in the district of Rheinfelden. Its 1980 population was 3,044. It is only some 10 km (67 miles) away from the center of the city of Basel. From 1970 to 1980 it experienced the most dramatic growth of all communities in the district, increasing by 132.2%. Miihlin and Kaiseraugst are also served by train. From Miihlin, it takes about 30 minutes, and from Kaiseraugst less than 15 to ride the train into the center of Basel. Hence, all ten fastest-growing districts are within convenient commuting distances to an important urban center. The ten districts that experienced the largest absolute population losses include most of the largest cities: Zurich (largest city), Basel-City (second largest city), Geneva (third largest city) and Lausanne (fifth largest city). Also included are districts with medium size cities: Biel/Bienne (ninth largest city), St Gall (seventh largest city), Schaflhausen (fourteenth largest city) and La

8 96 PETER V. SCHAEFFER Chaux-de-Fonds (eleventh largest city). Cities in this second group have populations between 35,000 and 75,000 (Table 9). Only two of these ten districts do not include a large or medium sized city. Le Locle (Canton of Neuchatel) had a population of 15,902 in 1980, down from 18,860 in Most of its inhabitants live in the watch-making town of Le Locle which had 12,039 inhabitants in Though small, it is an urban district with important manufacturing activities. All but one of the seven towns in the district suffered population losses between 1970 and Courtelary (Canton of Bern), also located in the watch-making region of the Jura mountains, has a more dispersed population than the other districts in this group. Its 1970 population was 26,442. It decreased to 22,606 inhabitants in Its largest town is St Imier, which had 6,740 inhabitants in 1970 and 5,430 in 1980 (- 19.4%). All but two of the district s eighteen towns experienced population losses. Two towns gained 2 (+ 1.6%) and 25 (+2.4%) inhabitants, respectively. Of a total of 185 districts, only four lost more than 10,000 inhabitants from 1970 to 1980, six lost more than 5,000 inhabitants, and twenty-nine more than 1,000 inhabitants. We expected that the largest districts with the old central cities would be the biggest losers in absolute terms. For this reason, we also looked at the districts that experienced the fastest rate of population decline. They are much more mixed than those with the largest absolute population losses. Of the districts with large and medium sized cities, only Basel, Biel/Bienne and La Chaux-de-Fonds are included. The last two of these three districts are located in the watch-making region of the Jura mountains, which also includes Courtelary, Le Locle, Val-de-Travers, La Vallee, and Grandson (Table 5). Thus, seven of the ten fastest-declining communities are located there. An eighth district, Solothurn, at the foot of the Jura mountains, is at the periphery of this region. The six districts Grandson, Val-de-Travers, Le Locle, La Chaux-de-Fonds, Courtelary and Biel/Bienne form a contiguous territory of decline. Table 6 lists the relevant Hoover indexes. They show a higher degree of concentration than at the level of cantons. The index shows weak deconcentration between 1970 and If one removes the ten largest districts in 1900 and recalculates the Hoover indexes, they show continuing concentration. Similarly, if one removes the five districts that lost the highest number of inhabitants between 1970 and 1980, the index again shows continuing concentration. The trend towards a more concentrated population among the remaining districts is much stronger than the reversal for all districts combined. Hence, a small number of districts determined the decrease of the Hoover index from 1970 to These districts are concentrated in a few cantons, as Table 5 shows. Counter-urbanization Calculation of the correlation coefficient between population density and growth per km gave a result similar to that for the cantons. The correlation was positive except for the period The two districts Basel-City and the City of Geneva were outliers for the whole study period, however. The coefficient was recalculated with these two observations deleted. This led to a smoothing of the differences from period to period but not to qualitatively different results. In 1970, the districts Zurich and Solothurn were also outliers. Even with these additional districts removed, the correlation coefficient remained negative with a value of The district of Biel/Bienne had a density 2.84 times the standard deviation in Only if this district was also deleted did the correlation coefficient turn positive (+0.15). Thus, the rest of the country continued the trend Table 6. Hoover indexes for districts, All districts except All districts except the ten lareest the five tod losers Year All districts districts in T900 of population, 197& I Source: Calculated from [2,5].

9 Population distribution of Switzerland 97 Table 7. Correlation between growth/km* and population density G &80 All districts Standard deviation density Standard deviation growth I Without BS, GE (city) Standard deviation density Standard deviation growth I Source: Calculated from [2,5]. Note: BS = Basel-City, GE = Geneva. towards more densely populated areas. Taken together, the results for the cantons and the districts reflect the population losses of the largest urban districts and the dispersal of populations within the most populous cantons. Trend reversal The procedure to test for the presence of trend reversal was the same as for the cantons (Table 8). The changes measured are more modest than for the cantons. This is because the largest districts are concentrated among a few regions. Since there are only 26 cantons, but 185 districts, these few regions have less weight in the case of the districts. The trends for the districts are the same as for the cantons, except that the change from 1970 to 1980 is more in line with those of the preceding two decades. The period experienced the smallest change, though that from is not much smaller. As mentioned above, we expected this as these are the decades of World Wars I and II, respectively. As was the case for the cantons, calculation of the measure TR for the districts showed no sign of trend reversal. DISCUSSION OF RESULTS The results show that deconcentration did occur during the period Deconcentration was much stronger for cantons than districts. At the district level, the Hoover index showed only weak deconcentration. Some evidence of counter-urbanization was also obtained. What caused these changes? Frey [13] presents three possible explanations. They are the period explanations, the regional restructuring perspective, and the deconcentration perspective. The period explanations hold the events specific to the period responsible for the observed changes. The oil crisis and the economic recession resulting from it, followed by stagnation, are thought to be important (see [25] for an empirical test). Manufacturing industries, in particular, were hurt by increased international competition. This seems a likely explanation for changes in the Swiss watch-making region. This important Swiss industry lost significant market share to Japanese firms during the 1970s. Employment in Neuchatel, the canton most dominated by watch making, dropped from 77,712 in 1965 to 70,281 in 1975 (-9.56%) [3,4]. The regional restructuring perspective attributes the changes in the population distribution to changes in the organization of the production of goods and services that has occurred. Losses are expected to be particularly severe in regions that are economically advanced with labour-intensive economies. The deconcentration perspective implies that better transportation and communication links, combined with a higher standard of living, have relaxed previously existing constraints on movement to more removed regions where densities are lower. The deconcentration perspective has Table 8. Trend reversal, districts TR,.,. ID By decade &20 192G C C TR TRm Base Year & &40 190&50 190&60 190&70 190&80 TR Source: Calculated from [S].

10 98 PETER V. SCHAEFFER considerable appeal in Switzerland, as most of the national highway system was built during the twenty years The evidence presented here does not provide clear answers regarding the merits of the different hypotheses. Table 9 lists Switzerland s fifteen largest cities, ranging from 369,522 to 33,441 inhabitants in Only two of these fifteen cities experienced growth during , and only Koniz, Bern s largest suburb, grew at a rate above the national average. The last two columns in Table 9 suggest that the growth or decline of cities may be related to the availability of developable land. Switzerland s land prices are very high and homeownership rates very low. In 1980, slightly less than 30% of households owned the dwelling they inhabited. By comparison, the figure was over 50% in France, Great Britain, Italy, Norway and Spain. It was more than 40% in Sweden, and about 37% in the Federal Republic of Germany [6]. A population redistribution towards less crowded regions with cheaper land, particularly as transportation services improved, could be the result of economic incentives. The correlation coefficient r between %-change and developed and developable land is This finding is compatible with the deconcentration perspective. We also ran an exploratory regression. The population growth rate 197&80 was the dependent variable, and developable and developed land per capita (as defined in Table 9), employed growth , and the growth in the foreign population , were the dependent variables. The results are summarized in Table 10. The coefficients have the expected sizes and signs. They are larger and more highly significant for the employment growth rate and the growth of the foreign population than for land availability. The change in the foreign population was used because of the dependence of Swiss manufacturing and service industries on guestworkers for low skilled jobs. During the 1970s almost 20% of the Swiss labor force consisted of foreign workers, with a significant share of workers holding only an annually renewable visa (see also [26]). According to the regional restructuring hypothesis, and because of visa constraints, the jobs of this group (foreign workers) are the most vulnerable. Foreign workers are also more likely than natives to occupy older dwellings in central cities. A recession-induced loss of foreign residents would affect these cities more than other districts (see [21] for a recent study). The empirical results are compatible with the period and the restructuring explanations. The low R shows that population redistribution cannot be explained satisfactorily by a simplistic regression. The Canton Basel-City had stable employment from 1965 to 1975, but still lost more Citv 1970 Zurich (Zurich) 422,640 Base1 (Basel-City) 212,857 Geneva (Geneva) 173,638 Bern (Bern) 162,405 Lausanne (Vaud) 137,383 Winterthur (Zurich) 92,122 Saint Gall (Saint Gall) 80,852 Lucerne (Lucerne) 69,879 Biel/Bienne (Bern) 64,333 Fribourg (Fribourg) 39,695 La Chaux-de-Fonds (Neuch.) 42,341 Thun (Bern) 36,523 Neuchatel (Neuchatel) 38,784 Schaffhausen (SchafThausen) 37,053 Kiiniz (Bern) Table 9. The fifteen largest Swiss cities in 1980 Developed and 1980 %chawe developable land 369, I , , , I.62 75, I.64 63, , I.61 37, , , , , Developed and developable land is defined in the Swiss Federal Planning Act of 1979 by the term Bauzone. This definition includes land that is already developed or is suitable for development and is likely to be needed withm the next fifteen years (Art. 14, Bundesgesetz iiber die Raumplanung). Land not so defined may not be included in the Bauzone. Unfortunately, the information about developed and developable land is available only for the early 1980s. Large changes in the amount of Bauland do occur slowly, however, so that the figures can be used as an approximation for the 1970s. In this study, we took all land in the Bauzone not used for industrial or for traffic purposes, and divided It by the population in The dimension is one of Are /Person. The typical amount of developed and developable land per inhabitant in fast-growing districts was approximately 4.0. One Are = acres. Note: Neuch. = Neuchatel. Sources: [1.2,5].

11 Population distribution of Switzerland 99 Table IO. Population growth as a function of land availability, change of employment, and the growth rate of the foreign resident population Dependent variable: population growth 197&80 in Swiss Districts constant Land in Bauzone per capita Employment growth rate Growth rate 197&80 of Toreign population R*: Note: I statistics in parentheses (1.541) (1.510) (3.013) (2.714) than 31,000 inhabitants. By contrast, the population of the Canton Aargau increased 4.7% from 433,284 in 1970 to 453,442 in 1980 while its employment decreased from 184,264 in 1965 to 178,924 in 1975 (-2.9%). The population of the Canton Nidwalden grew even faster, at 11.6%; employn-.snt increased by only 2.05% (+208 people employed) from Other factors, not accounted for here, have obviously played an important role. The results show that employment and industry-related factors are more important than land availability. It is too early to conclude that land availability is unimportant, however. The choice of the size of the Bauzone is as much a political one as one dictated by natural constraints. The review of zoning, which is done by the communities, is under the control of the cantons, not the federal government. Some cantons, for example Fribourg and Valais, have much more developed and developable land per capita than the average, while Neuchatel seems to be quite conservative. Large amounts of such land in communities which, for other reasons, have little growth potential, tend to distort the results. The regression identifies eight observations as outliers or having great leverage on the results. Not unexpectedly, these observations include the quickly growing districts Dielsdorf (Zurich: population change , +33.3%), Fraubrunnen (Bern: +23.4%), Echallens (Vaud: +28.4%), the declining district La VallCe (Vaud: -20.5%) the two districts Herens and Goms in the Valais, both with a very high ratio of developed and developable land per capita, and the City of Geneva. Low homeownership rates and very low vacancy rates for dwellings provide casual support for the hypothesis that land availability poses a constraint to growth. Data about vacancies have been collected by the Bundesamt fur Statistik since The vacancy rate (national average) reached its highest level, 2%) in 1975 and Construction of new dwellings peaked in the early 1970s. It decreased dramatically during the recession of the mid 1970s and did not recover by Vacancy rates decreased to only 0.74% in 1980 [6]. A large share of the vacant dwellings was in buildings that had been constructed within the last two years of the period These carried rents far above the average ([6], p. 192). In 1979, an unfurnished two bedroom apartment in the City of Zurich that was built after 1960 for example, had a medium rent of US $540. The normal local rent was US $300. In Geneva, the rent was US $510, with the normal rent reported at US $390 [18]. During the same period, the average household size continued to decline from 3.4 in 1960 to 3.0 in 1970, and 2.6 in 1980 [6]. Tight housing markets discourage people from moving. Older households, in particular, stay in the dwellings in which they raised their children, as few better choices are available that do not cost much more. In 1970, the number of dwellings occupied by households with children was almost the same as the number of dwellings occupied by households without children. The latter grew by 32.0% between 1970 and 1980, and the number of persons in such households decreased by 11.4%. Dwellings occupied by households with children increased by only 0.4% during the same period, and the number of persons per household decreased by 6.6%. Hence, some of the large declines of central city populations can be attributed to demographic change. Figure 1 shows the dramatic increase in the share of the population 65 years of age and older in the City of Zurich between 1960 and The city s increase far exceeded that experienced by the canton. While the city s total population declined from 440,170 in 1960 to 369,522 in 1980 (- 16%), its elderly population (65 years and older) increased from 47,384 to 73,256 (+ 55%). We would, therefore, expect the City of Zurich to grow again in the future, as younger and larger households replace elderly households. The city s higher relative share of young adults (2&24 years of age) is likely due to the presence of two universities and other major educational institutions in the city.

12 100 PETER V. SCHAEFFER f $ to 5 ty 1980 : 8 o ro-, J9 4o v e5+ Age cohorts Fig. I, Age distribution 1960, 1970 and 1980, City and Canton of Zurich. Source: [29]. CONCLUSIONS Our results show evidence for the occurrence of deconcentration and weak evidence for counter-urbanization in Switzerland for the period There was no evidence of trend reversal, as defined here. Closer examination of where the strongest growth occurred shows that suburbanization continues to be an important force. Fast-growing districts were all located in the most industrialized cantons, close to large- and medium-sized cities. Growth is very unevenly distributed. In an analysis that uses data for different geographic scales it is possible that there may be a blending of intra- and inter-regional processes. This may be the case here. To explore this possibility, an extension of this research to look separately at deconcentration at the intra- and inter-urban levels would be useful?. This paper points to the importance of moving from description to more rigorous analysis. The next step should be to test the hypotheses offered by Frey [13] and others. Kontuly and Bierens [25], Hoenack et al. [19], and Wilson [33], are among those who have begun this process. In the case of Switzerland, some factors must be included that are of less importance elsewhere. These include the presence and rights of a large foreign worker population and their dependants, constraints imposed by land and housing markets that are tighter than in most other indust~alized countries, and the fact that the functional region of Basel, Geneva and the southern part of the Ticino extend beyond the national borders. Some of the international comparisons in the literature have stressed similarities between countries without concern for differences. While the information they provide is useful and has stimulated others, it is now time to provide more detailed country studies to remedy their shortcomings. Acknan~ledgements-I gratefully acknowledge the kind help I received from Ernst Reinhardt, Wiliiam Frey, Tom Kontuly and staff of the Bundesamt fur Statistik of the Swiss government, particularly Mr Dominick Ullman. The comments of the anonymous referees and of the editor of the journal also helped improve this paper. Any errors are my responsibility, however. tthis point has been suggested to me by Thomas Kontuly of the University of Utah.

13 Population distribution of Switzerland REFERENCES Bundesamt filr Raumplanung. Eauzonenerhebung. Eidgenossisches Justiz und Polizeidepartment, Bern, Switzerland (1989). Bundesamt fiir Statistik. Arealstatistik der Schweiz Statisfische Queiienwerke der Schweiz, Heft 488. Publikationsdienst, Bundesamt fib Statistik, Bern, Switzerland (1972). Bundesamt fiir Statistik. Eidgeniissische Befriebsztihlung 1965: Beschtiftigte in den Gemeinden. Reihe Af2. Publikationsdienst, Bundesamt fur Statistik, Bern, Switzerland (1968). Bundesamt fiir Statistik. Eidgentissische Berriebsztihlung 1975: Arbeitsstiitren, Hauptergebnisse fir die Gemeinden. Band 4, Heft 608. Publikationsdienst, Bundesamt fur Statistik, Bern, Switzerland (1977). Bundesamt fur Statistik. EidgenGssische Volksziihlung Wohnbevdlkerung der Gemeinden, Band I. Statistische Quellenwerke der Schweiz, Heft 710. Publikationsdienst, Bundesamt fiir Statistik, Bern, Switzerland. (1981). Bundesamt fiir Statistik. Stafisfisches Jahrbuch der Schweiz Verlag Neue Ziircher Zeitung, Zurich, Switzerland (1989). Bundesgesetz iiber die Raumplanung (RPG) vom 22. Juni Swiss Federal Government, Bern, Switzerland (1979). E. Casetti. Peripheral growth in mature economies. Econ. Geogr. 60, (1984). A. G. Champion (editor). Counterurbanization: The Changing Pace and Nature of Population Deconcentrarion. Edward Arnold, London (1989). A. G. Champion. Counterurbanization: the conceptual and methodological challenge. op. cit. [9], pp N. Draper and H. Smith. Applied Regression Analysis. Wiley, New York (1966). A. J. Fielding. Counterurbanization in Western Europe. Prog. Plann. 17, 1-52 (1982). W. H. Frey. United States: counterurbanization and metropolis depopulation. op. cif. [9], pp (1989). W. H. Frey. The re-emergence of core region growth: a return to the metropolis? Inf. Reg. Sci. Rev. 11, (1988). W. H. Frey. Migration and metropolitan decline in developed countries: a comparative study. Popul. Develop. Rev. 14, (1988). W. H. Frey and A. Speare Jr. Regional and Metropolitan Growth and Decline in the United Stares. Russell Sage Foundation, New York (1988). G. V. Fuguitt, T. B. Heaton and D. T. Lichter. Monitoring the metropolization process. Demogr (1988). M. Guttmann and A. Kruck. Prices and Earnings Around rhe Globe. Economic Research Department, Union Bank of Switzerland, Zurich, Switzerland (1979). S. A. Hoenack, J. A. Peris and W. C. Weiler. Can economic incentives explain the recent population movements to nonmetropolitan areas? Ann. Reg. Sci. 18, (1984). H. Jones, N. Ford, J. Caird and W. Berry. Counter-urbanization in societal context: long-distance migration to the Highlands and Islands of Scotland. Prof. Geogr. 36, (1984). P. N. Jones. The declining guestworker population in West German cities: the case of Nuremberg. Urban Stud. 27, (1990). G. Kephart. Heterogeneity and the implied dynamics of regional growth rates: was the nonmetropolitan turnaround an artifact of aggregation? Demopr. 25, 99-l 13 (1988). T. Kontuly and-r. Vogelsang. Explanations of the intensification of counterurbanization in the Federal Republic of Germany. Prof: Geogr. 40, (1988). T. Kontuly, S. Wiard and R. Vogelsang. Counterurbanization in the Federal Republic of Germany. Prof: Geogr. 38, (1986). T. Kontuly and H. J. Bierens. Testing the recession theory as an explanation for the migration turnaround. Envir. Plann. A 22, (1990). W. E. Kuhn. Guest workers as an automatic stabilizer of cyclical unemployment in Switzerland and Germany. Int. Migration Rev. 12, (1978). K. F. McCarthy and P. A. Morrison. The changing demographic and economic structure of nonmetropolitan areas in the United States. Int. Reg. Sci. Rev. 2, (1977). P. A. Rogerson. The demographic consequences of metropolitan population deconcentration in the U.S. Prof: Geogr. 36, (1984). 29 Statistisches Amt der Stadt Zurich: Data provided to author based on the population censuses of 1960, 1970 and D. R. Vining Jr and T. Kontuly. Increasing returns to city size in the face of an impending decline in the size of large cities: which is the bogus fact? Envir. Plann. A 9, 5942 (1977). 31. D. R. Vining Jr and T. Kontuly. Population dispersal from major metropolitan regions: an international comparison. In?. Reg. Sci. Rev. 3, (1978). 32. D. Vining Jr and A. Strauss. A demonstration that the current deconcentration of population in the United States is a clean break with the past. Envir. Plann. A 9, (1977). 33. F. D. Wilson. Components of change in migration and destination-propensity rates for metropolitan and nonmetropolitan areas: Demog. 25, (1988). APPENDIX--See overleaf