Reasons for Migration Reverse in Estonia

During the past 30 years migration trends in most European countries have revealed some signs of deconcentration. The aim of this paper is to analyse the reasons of such deconcentration. For that purpose we will give an overview of literature on reasons of migration turnaround and provide a deeper case analysis of one country – Estonia. In Estonia internal migration fi rst turned in favour of rural areas in 1983. The trends of deconcentration continued and rural population increased at the expense of urban population at least until the end of the 1990s. The Retrospective Estonian Living Conditions Survey of 1999 is used to analyze the possible changes in reasons for migration turnaround. That survey provides a unique opportunity for comparing reasons of migration during periods of concentration and deconcentration. In the empirical part of this paper deconcentration will be defi ned as an outfl ow of population from administrative town borders into rural areas. We will not distinguish between particular types of deconcentration for practical reasons of data limitation. We are particularly interested in migration because migration seems to be the most important factor contributing to quick population changes.


Introduction
Since the emergence of the fi rst signs of migration turnaround lots of research have been dedicated to investigation of the reasons of migration changes.In a research conducted at the University of Michigan (Frey 1988), a hypothesis that would explain the halting of the development of city cores was tested by empirical studies in 13 developed countries in 1970-1980.The results showed that a part of the data supported the hypothesis that the main reason for deconcentration is of a social nature.It is based on such conditions of the housing market that allow people to achieve all possible objectives, including socio-ecological ones.Changes in the production structure, improvement of the living standard and the development of transportation will end a situation where production determines the location of producers and consumers.This model was especially appropriate in the case of the USA, Canada, the UK, the Netherlands, Austria, and Italy.Similarly, many other authors (Long 1988;Wilson 1988) saw the reason for deconcentration mainly in a greater freedom of choosing the residential area and a diminishing infl uence of production forces.Several researchers have found that the change in migration streams in the 1970s was a change in behavior combined with a change in housing preferences (Berry 1988).
Yet in some other European countries (Belgium, Germany, Switzerland, Denmark) the main reasons for regional reorganization seemed to be outdated functions of the economy, and migration fl ows depending mainly on the restructuring of the economy (Frey 1988).Still the majority of researchers stress the mutual infl uence of several factors on migration processes.Champion (1989) found that most of the reasons for deconcentration have been relatively similar -economic, social and technological changes.Later, Champion (1992) stated that the reasons for deconcentration and reurbanization are quite similar but opposite in character.Vining and Kontuly (1978) analysed counterurbanization in 18 different countries and found that the main reasons seemed to be economic restructuring and the infl uence of government policies.Dahms and McComb (1999) gave an overview of the reasons for counterurbanization in the 1990s and listed the following factors: attraction of environment, improved communication, state policy, unique economic and demographical conditions, service industries and restructuring of production, rural amenity environment, and push factors from the cities.Others (Marksoo 1988;Stockdale et al. 2000) assumed that the change of rural environment (infrastructure) itself made the migration change possible.Johnston (2000) listed both the attractions and disadvantages of metropolitans for employers and households as factors of counterurbanization: attractiveness, land use, transport, labor force, relocation of production and the human dimension, lower costs, life quality preferred by the elderly.
After analyzing a large amount of previous research, Findlay and Rogerson (1993) found that probably all mentioned factors could be seen as one single factor under the common term "life quality".In principle they argued that migration is triggered by the infl uence of personal preferences on the factors that are considered to be the most important for migrants.Accordingly, the reasons related to the quality of life could simultaneously explain both out-and in-migration of a region.Most explanations for deconcentration and urbanization seem to lie on the dominance of reasons related to lifestyle and housing (Ainsaar 2004)."Lifestyle" factors are often also related to property, property prices, push forces out of towns because of stress, life quality preferences of different age groups and improving transportation opportunities.The second group by frequency are economic reasons.It is plausible that different circumstances prevail in different countries.
In order to understand migration processes we should bear in mind that different social groups have different sensitivity to different stimuli and motivation.Therefore, migration can depend on the demographic potential of different population groups, their level of freedom to move and certain specifi c pull and push factors specifi c to these population groups in the society.Many surveys report that the leading group among out-of-town outmigrants are retired persons (Shumway and Davis 1996;Halliday and Coombest 1995;Coll and Stillwell 1999;Kok 1999;White 1990;Beale 1977;Johnson and Salt 1992;Rees and Kupiszewski 1999) and older working age people (Cross 1990;Frey 1989;White 1990).The third group, often mentioned as urban outmigrants are young families with children and people aged between 25-34 (Fielding 1982;Borgegård et al. 1995).Consequently also the reasons behind the migration streams can vary according to the motivation of those groups.

Migration reverse in Estonia
The small population and rather simple settlement system of Estonia is well suited for migration studies.The settlement system consists of one main centre -the capital with inhabitants forming 1/3 of the total population, three bigger towns within a 200 km radius from the capital, second order county capitals, smaller towns and rural settlements.This plain settlement system, very similar to Christaller's ideal settlement landscape, makes an analysis of migration processes transparent and allows the country to be used as a model for migration studies.
After extensive urbanization (see Tammaru 2001b) the rural areas started to gain people from urban areas in 1983 in Estonia.The tendency of outmigration from towns, although unstable, remained persistent during the 1980s and 1990s (Figure 1).Studies by Marksoo (1988) revealed that the migration turnaround of the 1980s started from towns on the highest level of urban settlement hierarchy in Estonia and although the changes were due to bilateral measures of town and countryside, the decisive role in the changes of rural-urban migration patterns belonged to the labor quotas enforced in the capital.Those quotas affected fi rst and foremost internal migration from country to town.Another important factor was the development of the economic and social infrastructure of the rural areas.Some growth was observed in the number of service sector jobs as well as women's employment in such jobs in the country (see Kuddo 1988;Lõo 1987).These changes encouraged young women to remain in rural areas.The growing popularity of rural areas among young graduates was based on a better availability of housing and higher income prospects than in towns (Kuddo 1988).Also many young families left towns and tried to settle in the countryside, because of the lack of fl ats in towns and the growth of the prestige of country life.
However the offi cial internal urban negative netmigration trend remained irregular in the 1990s and revealed an obvious evidence of administrative changes infl uencing migration registration (Figure 1).Although we cannot trust the offi cial migration data since the 1990s entirely for several reasons (see Herm 1999;Tammaru and Sjöberg 1999;Ainsaar 1999), the general curve showing positive immigration into rural areas seems to refl ect the actual direction of migration trends.Also population census data from 1989-2000 proved that the overall urban-rural net migration was in favor of rural areas (Ainsaar 2003;Tammaru et al. 2003).Considering internal urban net migration we can split the period into two periods using the year 1983 as a benchmark.The period before 1983 is an urbanization stage, while the period after 1983 is a deconcentration era.

Data
To explore migration turnaround reasons in Estonia the Living Conditions Survey from 1999 is used.The survey provides a unique opportunity to compare the reasons of migration in Estonia before the start of migration reverse and after.Although there are some separate studies about the reasons of movement before and after 1983, this survey is the only one treating the reasons from two different time periods with the same methodology.The survey was conducted by FAFO, Estonian Statistical Offi ce, Ministry of Social Affairs and the University of Tartu.It contains a limited range of 1 9 8 7 1 9 8 8 1 9 8 9 1 9 9 0 1 9 9 1 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 urban net migration in thousands questions on the migration history of respondents -time of their latest migration, previous and present living place and migration reasons.
All retrospective data were collected only regarding the last move.Assuming that some people tend to change their living place more often than some others, the movement of the more frequent movers is probably underestimated in our results.It would infl uence our fi nal results only, if we could assume that "chronic" movers have an inclination towards some specifi c kind of migration.The disadvantage of the Living Conditions Survey is a rather modest set of migration reasons.Only four types of reasons -housing, change of job or getting a new job, studies, and family reasons -were identifi ed in the survey.The survey was conducted in a face to face mode; the overall individual response rate was 90.2% (Pedersen and Tyldum 2000).In order to make the sample country representative, the initial database was weighted according to the last population census results (see Oja and Tiit 2002).
From all 4,726 respondents, 70% had an experience of migration in their lives.Further only the data of internal migrants will be used.Of internal migrants 1,075 had moved during 1983-1999 and 1,367 before the year 1983.All migration moves were divided into three periods: before 1983, 1983-1989 and 1989-1999.In Table 1 the period after 1983 was divided also into two groups to catch better period changes.Because of certain limitations of the dataset we were not able to identify the socio-economic status of the respondents at the moment of migration.Therefore analysis of migration reasons by different socio-economic groups is missing in this paper and age of migration will be used as the only individual indicator.
Table 1.Reasons for migration in different time periods, %.
On a regional level we will distinguish between three different groups of settlements: capital, other towns and rural municipalities.Administrative specifi cations were used in this study in order to distinguish between urban and rural settlements.Although some rural municipalities might have tight commuting connections with their neighboring urban areas (see Tammaru et al. 2003), their environment still remains rural in Estonia.

Reasons of migration and migration reverse in Estonia
The Living Conditions Survey revealed remarkable changes in the reasons for migration during the 1990s (Table 1).The most essential change was the rising number of moves because of housing in the period of 1989-1999.Family reasons maintained their priority and employment related migration declined, in contrast to our expectations.Moves related to studies formed a minor part of all moves.
Comparing the reasons of in-migration before and after 1983 (Figure 2) we can see an increase in dwelling-related moves both in rural and urban areas.Another important change was an increase of employment-related movement into the capital, but not into other towns.On the whole, the biggest shifts in reasons of migration are visible in the case of capital related migration.The most remarkable change in out-migration was the more than double increase of the share of housing migrants among the people who left the capital in 1983-1999 (Figure 3).The share of housing migration increased in all areas, but this increase was less signifi cant outside the capital.Reasons of migration changed less in other settlement types.As a result, between 1983-1999 rural areas gained people mainly because of non-economic reasons -dwelling and family, and towns because of economic reasons.The capital attracted people also as an education center.On the whole, economic factors did not affect the migration balance essentially and therefore they did not infl uence the proportion of rural-urban population.Net migration before 1983 had a more clear and powerful direction towards towns (Figure 4).During this period towns grew mainly because of family and employment related reasons.Prevailing groups of in-movers were younger than 25.
The results after 1983 are more scattered.At fi rst, the survey data revealed a smaller net migration in numbers, which means that the netmigration direction was not so clearly determined.However, it is obvious that towns lost people mainly because of housing reasons in all age groups and gained people because of job opportunities during that period.Comparing the time periods before and after 1983 we can conclude that although changes in migration reasons occurred in all age groups, the shifts were most remarkable in the behavior of migrants who were older than 24.

Discussion
The Estonian data show remarkable changes in migration reasons during the migration turnaround of the 1990s.The main transformation was an increase in the share of housing moves and a relative decrease of employment-related reasons in the migration.This can have various explanations.The prevalence of housing factors over employmentrelated migration could partly prove some earlier assumptions according to which migration in developed societies is less economy-related (Ainsaar 1999;Bonifazi et al. 1997).In the case of Estonia, the small share of employment-related migration could result also from the short distances of migration, but this does not explain the change in employment-related migration over time, because the territory has remained the same.One possible explanation could be the replacement of employment-related migration by commuting.The commuting survey (Tammaru 2001a) shows a remarkable increase in commuting during the 1990s.According to the study, commuting increased most remarkably in the vicinity of larger towns.
The rapid rise of housing-related reasons among migration reasons in general, compared to earlier periods, could be explained with limited individual possibilities of selecting type of housing in previous periods.Namely, people had fewer opportunities to acquire their own accommodations in Soviet time and usually the employer provided fl ats for its workers.Therefore, we can assume that, even whilst the possibilities of getting a fl at were taken into account in migration, the primary objective of migration was a job place as a provider with all other amenities previously.Workplace was an agent for housing and other social benefi ts in Soviet time.The building of new private residences was strictly fi xed by municipality planning and was quite limited (Ainsaar et al. 1996).The increase of housing as a reason of migration in the 1990s can be explained with a fact that although the total volume of housing construction decreased in the 1990s, compared to the previous period (Kõre ja Hendrikson 2000), people themselves became more involved in improving their living conditions and housing became more independent from other reasons of moving.
The analysis of the change of different migration fl ows allows make the assumption that deconcentration was driven by three factors: (1) the spread of commuting, which diminished in-migration to towns, (2) the increase of housing opportunities outside towns, which supported out-migration from larger towns, and fi nally (3) towns hold a demographic potential necessary for outmigration.

Conclusions
Many developed countries have experienced a period of population deconcentration during 1970-1990s.Analyses of reasons of deconcentration in different countries show a dominance of non-economic motives among the explanations for these changes.More profound analyses of migration turnaround in Estonia have revealed the dominance of non-economic reasons related mainly to housing as the main factor of migration reverse.Analyses of the Estonian migration turnaround reasons demonstrate a diminishing infl uence of production structure on people's choices of the living place during the changes.Job and housing as determinants of net urban migration were replaced with housing and family reasons after 1983.On the whole, the biggest shifts took place in capital related migration.Chronological analysis of the change in migration reasons allows us to build assumptions that deconcentration and shifts in net migration reasons were encouraged by three simultaneous phenomena: better housing prospects in rural areas, increasing commuting and demographic potential of towns.

Figure 2 .
Figure 2. Reasons of arrival and type of destination municipality, time periods before and after 1983.

Figure 3 .
Figure 3. Reasons of out-migration and area of origin, time periods before and after 1983.

Figure 4 .Figure 5 .
Figure 4. Urban net migration by age and reasons before 1983.