Skip to navigation – Site map

HomeCross-sectionsTopics199911th European Colloquium on Quant...The relationship between househol...

1999
11th European Colloquium on Quantitative and Theoretical Geography. Durham Castle, UK, September 3-7, 1999
135

The relationship between households and the housing market: A microsimulation model of household formation among the young

Les relations entre les ménages et le marché immobilier : un modèle de microsimulation de la mise en ménage chez les jeunes
Urban Fransson

Abstracts

The subject of this study is the interrelationship between household formation and changes in the housing market. The main focus will be young people's possibilities to move from their parental home. These possibilities are dependent upon a number of processes in the housing market, such as production of new dwellings or the actions of other households. This means in turn that the situation of young people is treated as part of a much larger and complex system of interaction in a local housing market. The primary thesis of the study is that interaction between individuals and structure is of crucial importance. The development of a method to describe complex interactions is an important part of the study. In order to describe many simultaneous processes on the individual level, in time and space, a micro-simulation model has been constructed. The model provides a good picture of the moves of young people and how their possibilities are restricted by both their own situation and structural conditions.

Top of page

Full text

Introduction

  • 1 Hägerstrand (1970)

1Changing financial and economic conditions affect the housing market. To answer the question on how different groups are affected by the changes the components of the housing market have to be considered. The question asked in this article is whether it is possible to model the changes of a local housing market based on individual and societal restrictions. Can individual actions restricted by the actions of other individuals as well as the rules of the environment, explain these changes?1

2For this purpose an individually based simulation model of a local housing market has been constructed. It contains the factors and processes affecting the demand of the individuals as well as the market supply of dwellings. In the model the individuals make the decision to change residence based, among other things, on the type of vacant dwellings in the market and financial resources. The aim of this article is to demonstrate how various processes co-operate to create vacancies to make the formation of new households possible and to test a method that takes into consideration household demand as well as housing market supply.

  • 2 Hägerstrand (1972)

3Housing market agents interact in a way where the actions of an individual cannot be seen independently of the actions of other individuals. The actions of each household in the local housing market effect the conditions of the decision process of all other households. Individuals competing for limited resources, such as housing, create barriers or restrictions to each other2. The relation between individuals and structure is in the model expressed as the actions of the individuals in the housing market and the continuously changing composition of vacant dwellings. The possibilities of the young to form their own households should be viewed from this perspective. The housing market consists of these agents, in a continuously ongoing interaction and of the forever-changing supply of vacant dwellings.

4In Sweden, a change of dwelling occurs on average ten times in a lifetime. Often these shifts reflect changes in the life course, such as when an individual leaves a household or moves together with another person, but also demographic changes, including migration, births and deaths. The financial situation of the household is important for their actions. The individuals make their decision and act under the restrictions and opportunities in a local housing market.

5Two different processes create vacant dwellings. The households generate a supply of vacant dwellings by their recurrent and varied moves. The supply creates the necessary conditions for the formation of new households. There is a dynamic connection between household changes and changes in the supply of vacant dwellings. Vacancies are created not only by residential mobility but also by building companies and housing managers producing new dwellings, redeveloping or changing current buildings into housing developments. However, we do know that the supply created by residential mobility and out migration are substantial.

6Using the terminology of human geography these processes are studied from a time geographical point of view. The individuals (and households) are the agents, time and space the descriptive dimensions and agents and environment influence one another. A model is used as a tool to study relations, dynamics, interdependence, parallellity and complexity in a local housing market.

Demand and supply

7The model constructed aims for a synthesis of what we already know about the different processes in housing, based on empirical studies as well as theoretical arguments. The consistent model takes into account the demands of the households as well as the supply of the housing market, i. e. the necessary conditions for an individual to act in a local housing market. Using the simulation model it is possible to judge the effects of economic as well as political changes towards different groups in society. It can be used as a tool for experimenting with such changes and to analyse the consequences.

8To produce a complete picture of an unsteady housing market the process of residential mobility needs to be included in a consistent theoretical frame. The most important terms are population, household, housing, and new construction, the match between supply and demand as well as income and housing expenses. The latter involves relations to politics, economy and labour market.

9We cannot use all the knowledge we have in the field of the process of household moves, and it is not wise to do so either. Part of what we know is therefore left out, as it is not the focus of this study. To get a complete picture of how the different parts are linked together they must be consistent in that it has to be possible to apply them on each individual at the same time, as they must be adjusted in a coherent form.

  • 3 Bysveen and Knutsen (1987)

10The number as well as the composition of households is strongly related to changes in demography. In- and out-migration to and from the geographical area of the housing market influence supply and demand3. These movements affect the change in population number and the household composition.

Figure 1. The main components deciding the moves of households in a local housing market.

11The households match their demands with the supply of the market. The households seek other housing from the supply of vacant dwellings according to their demands and from what the household budget allows them to spend on housing. In the search process the most important demands on housing can be summarised in increased dwelling size, reduced housing costs and the search for a more attractive dwelling or housing area. The search process involves gather information on different housing alternatives and the examination of these alternatives to evaluate and choose among them.

The simulation model

  • 4 Öberg (1979), Orcutt (1960)

12Demography, household formation, income variation over the life cycle, housing stock, new construction, tenure and housing costs are important factors in the process of residential mobility. They are bound together in a synthesis given the shape of an operationalised model. The model connects changes in the macro world to individual behaviour as well as individual preferences, resources and decisions to the restrictions and opportunities in the market. The model constructed is a microsimulation model that is time geographically dynamic and where each individual is represented separately rather than as an aggregate of individuals4. Actions such as unemployment thus can be linked to a specific individual and not only to a group within the population. In a context characterised by large heterogeneity it is suitable to use individually oriented models rather than transition matrices to reflect all events and actions that occur in a housing market.

13The interaction between individuals also implies, within its context, the consideration of the simultaneousnes of different processes. Time geography is particularly suitable to this study as the approach emphasises the dependency of one household’s actions in the housing market on other households’ actions.

  • 5 Clarke and Holm (1987), Clarke, Duley and Rees (1989).

14The dynamic microsimulation is particularly useful in that it is possible to change temporarily the behaviour of the simulated individual. Events can be linked directly to a single individual. The consequences of an event can then be seen immediately. When needed individuals are linked to a group. In addition it is easier to represent interdependencies on a micro level5.

15The actions of around 90 000 individuals on a local housing market during 15 years, 1976–1990 are modelled. The local housing market has 45 000 dwellings. Geographically the local housing market is delimited to the municipality of Gävle, 200 km. north Stockholm, Sweden. In the market the individuals search the type of vacant dwellings that they can afford and the ones that satisfy their demands. The individual has some specific characteristics as well as, in a technical sense, relations to other individuals but also to his or her dwelling. The relations to other individuals are first of all directed to individuals within the same household. These relations tie the household together.

16Dwellings are represented individually and as they become vacant they are directly entered in the market, i.e. available to other individuals. The actual dwelling is a single object consisting of properties representing its availability, location, tenure, size and cost (i. e. the household’s expenditure). The individual in relation to his/her preferences and resources considers these attributes when looking for a new housing among the supply of vacant dwellings. The dwelling is linked to a particular statistical area. The statistical areas represent the geographical dimension in the model. They vary in size from 0.08 sq. km. to 30 sq. km.

  • 6 Timmermans et.al. (1994).

17The behaviour of the individual is modelled by probability distributions and by rules. The rules of the models are based partly on theoretical, partly on empirical circumstances. The individuals in the model act according to deterministic rules such as children leaving school at the age of 16 (after nine years of compulsory schooling). Rules and events are also initiated based on the probabilities that certain events or actions will take place. Based on empirical observations the probability distribution is calculated and express variations in the actions of the individuals. The drawbacks are that also the restrictions and the circumstances prevailing in the housing market at the time of data collection will be implicit in the distribution6.

18A different approach is to use rules that function independently of temporary variations in the supply of vacant dwellings. Such deterministic rules do have the drawback that they do not reflect variations in behaviour due to differences in development, socio-economic class or gender. However, the actions of an individual are deterministic as well as stochastic, and are thus modelled in such a way. As a result there are reasons to combine more specific rules with stochastic influences.

19The model consists of a number of co-operating parts, modules. Important modules in this model are the demographic, the household, the housing, the income and the educational module and the search and decision module.

20Data on individuals and dwellings according to the National Census in 1975 constitute the starting population of the model. National data as well as data for the municipality under study is used for this purpose in the model. The results from the model are then compared with census data on an aggregated level.

Housing module

21In mid 1970s the production of multifamily housing was drastically reduced in the whole nation, the study area included. On the other hand the production of single family housing was substantial well into the 1980s. The housing stock in the study area has changed during the period 1975–1990, not only in that more dwellings have been produced but also in that the structure of the housing stock has changed. The number of small dwellings (1–2 rooms) has decreased whereas the number of large dwellings has increased.

22Housing expenditure on the different types of tenure are initially given exogenously to each single dwelling in the model depending on period of construction and housing size. Data on housing expenditure on all tenures is available as sample surveys. The change in number and type of dwellings is given exogenously for each year simulated. Each vacant dwelling is marked in a ‘pool’ according to the dimensions size and expenditure.

Population

  • 7 Rossi (1955)

23Demographic changes, such as the death of a single adds another dwelling to the local housing market. Even the death of one household member, such as the death of a partner may indirectly affect the availability of dwellings, as the remaining household member may have to change dwelling for financial or practical reasons. Household composition in a housing area also changes over time. The birth rate and the number of children born, affect the housing market in two ways. The parents might consider the household crowded. Because of the lack of space the household start looking for a larger dwelling7. The number of children born in a particular year also affects the demand situation in the housing market, as this group will reach nestleaving age at the same time.

24Population changes in the area of study such as out migration, number of deceased and the number of children born, is represented by known probability distributions. For example in the model the observed fertility rates are expressed as the probability for women to give birth to children due to age and previous number of children. An individual that is born is in the model population incorporated among other individuals. It is linked to the household of the mother, obtains a year of birth and is given the status of a child within this household. The process of inmigration is represented by exogenously given empirical data as in known probability distributions and given numbers.

Household module

25The reason for the population and the dwellings to be modelled more or less exogenously is that they constitute the frame within which the household and the housing market interact. The household module consists of processes that can alter the number of members of a household.

Nest leaving among the young

26The opportunities of the young to leave the parental home are, among other things, dependent on the housing market supply and the resources available. During 1991 approximately 87 000 young people aged 16–25 years left their parental home in Sweden, corresponding to around 17 percent of all young people at the same age living at the parental home. Young women to a large extent leave home earlier than young men do. Most of the young individuals leaving the parental home live as singles. To be able to study the effect of the housing market on neastleaving in the model, the intention of young individuals to move is modelled separately from the actual move.

Divorce decisions

  • 8 Noortwijk and Hooimeijer (1990)
  • 9 In the model it is assumed that all individuals commencing a process of separation will request ano (...)

27The partnership between a married or cohabiting couple is cut off at a divorce or when one partner passes away. When separating the individual moving away must find a new dwelling. Also this might be the case for the partner that is left as a low income may make it impossible to remain in the current dwelling8. In the model the separation process is divided in two. In the first part all couples are considered to have an age specific risk of separation. The result of a risk is interpreted as a decision to separate. In the second part of the separation process the actual separation takes place9. This occurs when at least one of the partners finds a new dwelling. The individual initiating the divorce will look for any new dwelling, the only restriction being the individual household budget. The economic base of the members of the household that are left is changed. As a result they may have to move to less expensive housing.

Moving together

  • 10 Clarke, Duley and Ress (1989)

28The forming of couples precedes the act of moving together to form a household. The choice of a partner is in the model considered as a sequence of actions leading up to the forming of a household of two individuals living together in a common dwelling. This process is supposed to comprise four steps: to look for a partner; find a partner; accept a partner and at last to move in together with this partner to form a household. The individual can meet potential partners on different arenas; in reality this could be at work, at the place of a friend or relative or at a more open meeting place such as in pubs or at dance locations. These arenas are in the model considered as, from a technical point of view, a partner market, where individuals look for a partner at the same time as this individual can be found by another individual10. A young man and a young woman can presumably live with their respective parents but still be considered a couple. This means that the model has to be able to reflect relations between individuals belonging to different households. If one of the partners already has a dwelling of his or her they move together.

Changing demands

  • 11 Cadwallader (1992).

29Other reasons to look for housing are related to such things as to change housing standards; to live with other individuals from a similar cultural group; to fulfil social achievements or to live in a more attractive housing area in general11. A move from a dwelling also can depend on ageing. When older individuals become disable it is assumed they will move into some institutional housing providing care for the elderly or less able.

  • 12 Household budget is here referred to as an average for each household group respectively. Data is a (...)

30In the model households on a low housing expenditure rate are assumed to request housing in more attractive areas aiming for socio-economic upgrading. This reflects some variations in the moves within a local housing market that cannot be adhered to specific events. The request for a less expensive dwelling in the model occurs when the current dwelling is considered to expensive, i. e. that expenses exceed the household’s housing budget. The budget it assumed to vary according to household type12, i. e. young households spend a larger share of their disposable income on housing than older households do.

31Household income is a financial restriction that determines the choices of the household as regard the type of housing that can be demanded. This is not only the effect of the income of an individual but rather the income of the total household. This income constitutes the base for the size of the budget that can be spent on housing. Housing as well as tax policies of the government effect the housing expenditures of a household.

Income and educational module

32In this module the income of the individuals is changed according to educational level, part time work, unemployment, income career, pension and housing allowance. The educational level constitutes the base for the income career realised. To give all individuals an income career during the simulation they are put through a simulated educational system. Changes to the income are effected by the general income development nationally that mainly can be explained by inflation and economic growth and partly by the individual income career across the life span. The income career will be expressed as changes to the individual’s salary over time. Length and type of education is assumed to affect different types of income careers over the life cycle. The income careers over the life cycle are operationalised in five different types of income careers with different starting points and maximum income.

Search and decision strategies

  • 13 Brown and Moore (1970), Flowerdew (1976)
  • 14 Barrett (1976), Huff (1982), Clark (1993)

33The households in the model use a hierarchical search strategy. First the household decides what housing area or part of the city is preferable for new housing. Secondly the household maps the vacant dwellings13. Previous studies show that the households only look for a rather limited number of objects vacant14. This is due to a limited knowledge of the housing market and to the fact that they compete with other households. In addition, it is not feasible to acquire and evaluate information on all vacant dwellings on the local housing market. They seek and evaluate only the part of the supply in which they find some interest. The requested dwelling has to fulfil certain criteria. Most often, for an already crowded household with an additional child the size of the dwelling is the dominant condition that has to be fulfilled when looking for a new place to live. Such conditions must be fulfilled if the new dwelling is to be considered an alternative when the household makes its decision. These specific conditions are matched against the structure of vacancies on the local housing market at the time for housing search.

  • 15 Montgomery (1989).

34The actual decision making is viewed as a process to find the best arguments to make a choice. The decision-maker changes the problem to acquire evident arguments. It is argued that the individual makes up the preferences only when it is time to make the actual decision. Until then the search has mainly been based on general requests and views on a future dwelling15. When the problem of decision is complex, as in the case of evaluating housing, the strategy is to reduce the information to be evaluated. The households make their decisions based on only the most important components as well as a few possible alternatives. Using such a strategy the households risk missing out on possible housing alternatives as all potential dwellings are not considered.

  • 16 Cadwallader (1992), Gärling, Garvill and Lindberg (1991)
  • 17 Montgomery (1989), Payne (1982)

35The assessment of the different housing options is based on a subjective view of the properties of the actual alternatives. Based on the concluding evaluation of the properties of housing, a choice is made16. A rule of decision that is particularly close to reality and to the actual decision making of the household is the rule of domination17. The attributes are compared across all the options in that one attribute at a time is compared on all dwellings. The rule of decision is simple and clear but has the drawback that the housing options cannot always be arranged as the rule prescribes, as one option is better than another in one sense but not in another.

The rule is expressed as follows:

For one (1) attribute must at least:

a1k > a2k , k=1,...,n

and for other attributes

a1i / a2i , i=1,...,n and i k

If a1 is to be selected, where

a = attribute of the alternative option

n = number of attributes (can vary from 1 and up)

k = dominating attribute

i = other attributes.

  • 18 Montgomery and Svensson (1976).

36If option a1 is better than option a2 in at least one attribute and a1's attributes is not considered worse as regard other attributes a1will be selected18. The rule of domination is particularly interesting when combined with rules of restructuring to find an order of dominance. The decision making process can then be seen as a search for dominance in that the properties of the different options are restructured to obtain dominance. Montgomery has shown what such a model of dominance structuring can look like. When the individual finds a new dwelling in the stock of vacant dwellings the current one is transferred to the supply of vacant dwellings, i.e. to the market. At the same time the household moves to the new dwelling.

Simulation of the period 1975-90

37Validation of a microsimulation model can be done in several ways. The most common way is to compare the actions of the individual in the model with empirical data on such particular events and actions. However, such empirical data are very rare. The correspondence between aggregated results of the model and the empirical findings available on an aggregated level thus will have to be the measure on the strength and weaknesses of the model. Aggregated data is thus used for validation in this model. Empirical data on which the model is tested is based on data from Statistics Sweden. However, a good correspondence does not necessarily indicate that the simulation model models the detailed processes realistically as deviations on a micro level can be concealed when model information is aggregated.

38A complementary method of validation is to study the sensitivity of the model for different random number orders. The probability distribution used in the model is compared to a drawn random number to determine if an event will occur or not. The starting population in this study has been randomly redistributed to make sure the order of individuals will not cause systematic faults. The results of the use of different random numbers are named model a–d. When empirical data are available these are shown together with the results of the model. Of the model runs made using different random number orders run a corresponds to something like an average.

Population and household structure

39The random variation in the modelled population is small when the random numbers are changed. The results of the different runs of the model are well clustered. The changes in population are determined by in and out migration, the number of deceased and the number of children born. The random variation of these variables is small and the effect on the total population number is only marginal.

40The position of individuals within the household is categorised according to them being children, lodgers or head of the households. In the latter case a distinction is made between couples and singles. The household consists of individuals belonging to one of these categories. These categories are close to the processes forming and altering the structure of households.

41Changes to an individual’s position within the household most often occur in connection with the dissolving or the formation of a household. The number of individuals in the categories children, lodgers, singles and couples and the simulating processes changing them in the year 1980 are shown in figure 3. The figure illustrates how individuals change their household positions. The numbers in the box for each household position state the number of individuals in each category (state). The number of children amounts to 24 401 individuals in the model by the end of 1980. The processes changing the state of an individual are illustrated in figure as ovals. Arrows mark shifts in household positions. As expected there are only a few changes in the category lodgers as they are very few in numbers. The most important changes occur when children leave the nest and when couples form a household or break it up.

42The change in number of individuals living as couples depends on a number of factors; the number of couples moving in together or breaking up and the death of a partner. The number is also dependent on the amount of couples moving in or out of the municipality. The way the model represents the different processes generating changes is discussed below.

Position in the household

  • 19 National census 1975 and 1990.

43The number of individuals living as couples has decreased in the area of study from 43 700 in 1975 to around 41 600 in 199019. On the other hand the number of single households have increased dramatically.

44 

Year

Couples

Children

Singles

Lodgers

Unknown

Total

1975

43768

25294

14137

1931

55

85185

1980

43384

23978

16754

1759

134

86009

1985

42864

22596

18963

1584

98

86105

1990

41612

20890

21187

1170

390

85249

45Table 1. The number of individuals in households distributed by year and position in Gävle. Source: Census 1975, 1980, 1985 and 1990. Statistics Sweden.

46The number of couples living together in the model corresponds comparatively well with what can be observed according to the Census data. In particular as regard the period 1975–1985. The deviation is greater 1985 - 1990. The random variation is also substantial in this period. Census data are presented in figure 4 as bars for the years 1975, 1980, 1985 and 1990. Possible variations of the real household development within the five-year periods in between cannot be seen. Thus, it is only possible to compare the number of observed households in respective category with the results of the model for the years 1975, 1980, 1985 and 1990. The differences between the model run a and d in year 1990 amount to around 700 individuals and largely depend on the fact that the number of newly formed couples is considerably fewer in the last run. The large divergence in 1990 between model and Census may depend on the number of couples in the Census data being underestimated. According to Statistics Sweden’s evaluation surveys the number of households is underestimated in Census data in 1990, also affecting the number of one and two person households.

47The representation of singles also corresponds comparatively well to the actual changes (Figure 5). The number of singles is mainly effected by the number of separations and by the number of couples dissolving through the death of a partner. Young individuals leaving home also account for a large share of the changes. Cohabitation and marriage as well as the death of a single person cause the number of single person households to decrease. The number of single person households is larger in the model than according to Census data with the exception for the year 1990. This can partly be explained by the very simple construction of the process controlling the number of lodgers. In the model individuals cannot become lodgers; they can only leave this household position.

48The random variation between the representation in the model of the number of children is small. The number of children corresponds comparatively well with Census data. Generally the number of children is somewhat higher in the model than according to the Censuses as in the model more children are born than is actually the case. This difference does not explain the lack of correspondence between model and empirical data for the year 1990.

49A shift in the type of household amongst young people in-migrating can explain some of the differences between model and Census 1990. Even if age and gender of the inmigrants can be decided, uncertainty persists regarding the household to which they belong. The young moving in to the housing market can be categorised either as children in a family or as single adults.

Mobility in the housing market

50According to the model most of the mobility within the local housing market are made by households considering the current dwelling too expensive. These are couples that have separated, landlords whose has lost their lodger or parent(s) whose children left home (children that while living at the parental home contributed to the household income). They have chosen a smaller and less expensive dwelling in the market. The other main group contributing to the movements within the housing market is households demanding larger dwellings or dwellings in more attractive housing areas.

51As can be seen from the presentations above the households have different reasons for leaving or demand a certain type of dwelling. When the households move from one dwelling to another the vacant dwelling becomes available to other households. A large part of the one room apartments on the simulated market become vacant as the person living there is deceased or has moved from the municipality or moved together with a partner.

Nest leaving of the young

52An approximation of the number of actual moves from the parental home in the area of study is based on national data. In 1976 this would be the equivalent of 1 400 individuals in the Gävle municipality. The result from the model varies between 1 200 and 1 500 young individuals. About one third of all the young leaving the parental home in the model move to a dwelling of their own, one third leave the municipality and one third move in with a partner that already have a dwelling.

53There is also a certain difference between the start and the finishing phases of the simulation. The number of 18 year olds leaving the parental home is larger towards the end of the period than in the beginning. At the same time the number of vacant dwellings is higher towards the end then at the start of the period. This makes it easier for the group of young under the age of 20 to find a dwelling. Most of them have a low income if any. To make nestleaving possible the market has to offer small rental units from the older housing stock, i.e. dwellings at low rent. A higher turn over rate and a larger number of vacant dwellings, such as during the later part of the period, provides better opportunities for the young to move out of the parental home.

54In a geographical sense the share of the young living of their own show evident differences between the Census data and the model. Individuals up to the age of 25 are treated as young in the presentation (Figure 6). The map shows the difference in the share of young persons according to the Census and to the model. Areas with very few young individuals are non-coloured in the map. In the model the young is concentrated to the centre of Gävle to a much larger extent than in the Census. Two factors explain this; the spatial distribution of one-room units as well as the way in which the young in the model compared to the ones in real life has looked for a dwelling.

55The share of one room units is high (30 percentage or higher) in the centre of Gävle but also in Brynäs; just east the city centre. The young in the model to a larger extent move into one-room apartments compared to the young in Census. This difference is probably due to the fact that the housing budget is much more diverse among the young in reality than in the model.

56The other factor influencing the concentration to the city centre of the young in the model is that primarily they look for dwellings in a 1 000 metre radii around the city centre and only secondarily on the housing market as a whole. The latter explains why the model under-estimates the number of young individuals in Brynäs and over-estimates young individuals in the Bomhus area. As the turnover-rate is low in Brynäs, young individuals turn to areas with higher vacancy-rates. Bomhus is such an area. This is the case because families with children in the model can only address alternative housing in the current neighbourhood. This limitation in search gives the effect that only specific types of household look for housing in new build-up areas like Bomhus. This example shows the importance of implementation of search rules. Specifically the necessities of take into account not only the importance to live in current neighbourhood further on but also the demand of other tenures in new build-up areas.

57It is common to differentiate between primary and secondary supply, i.e. the way in which vacant dwellings are created. The primary supply represents 60 percent of all vacant dwellings in the stock of small units during the period. Moving out of the municipality and moving together with a partner are the processes generating most of the vacant one and two room units, just below 500 and just above 300 respectively per year. The passing away of individuals that have lived as singles also creates vacant dwellings among the stock of small apartments (just above 300 annually). The addition of small apartments through new construction, reconstruction and redevelopment etc. together with the household moving because of too high housing expenses each supply about 200 units each year. Elderly leaving the ordinary housing market to move into institutions and other types of housing for seniors add another 200 dwellings per year on average during the period.

58The secondary supply in the model is created by household demand on larger housing or more attractive area as well as by households moving into lower priced housing when they regard their current dwelling as too expensive. Household moves to enlarge their dwelling size roughly generate half the secondary supply, just over 500 dwellings per year, on average. The primary and secondary supply sums up to 2500 - 3000 vacancies per year.

59The number of vacant units produced in new production as well as through the actions of other households highly affects the actions of the young in the housing market. The most important competitors are households trying to reduce their housing expenses. Among those looking for smaller and less expensive dwellings are young seniors. Also nestleavers are attracted to this part of the housing stock. A contributing reason is that the price development on newly produced housing has been faster than the change of the disposable income during parts of the 1980s. The young nestleavers compete with a bit older and more established young individuals seeking to reduce their housing expenses.

Conclusion

60The type of vacant housing is continually altered through varied new construction and by individuals leaving the local housing market. The vacancies also vary through residential mobility. The temporary supply of vacant housing limits the actions of households looking for alternative housing. Income as well as housing cost is also restrictions in that matter. The mobility of the households in turn result in new sets of vacant housing offered the market; in turn influencing the conditions of other households to act in the housing market and so on. The supply of vacant housing thus depends on the households and housing involved in residential mobility as well as migration.

61An important conclusion is that it is possible to model the dependency of household formation on the supply of vacant housing and the choices and decisions of other individuals in the local housing market. On an aggregated level the individual flows and states, like household size and household position, correspond well with empirical data. It indicates that the model is a good representation of the interaction between household formation and what happens in the local housing market. The change in the number of single person households follows the increase of the number of small households as could be seen in the Census that has taken place during the last decades. In the model the number of children is somewhat overestimated, largely due to a changing household composition among the households moving in during the later part of the simulation process.

62This means that the intentions and the decisions of the individuals in a housing market can be represented with rather simple rules and in a context where they co-operate and compete in a continuously changing environment.

Top of page

Bibliography

Barrett, F. (1976) The search process in residential relocation. Environment and Behavior, vol 8, no. 2, pp 169-198.

Bysveen, T.,Knutsen, S. (1987) Vacany Chains Initiated by Out-Migration: A Study of the Housing Market in Oslo. Housing Studies, vol 2, nr 3, pp 203-212.

Brown, L.A., Moore, E.G. (1970) The Intra-urban Migration Process: A Perspective. Geografiska Annaler, Serie B. vol 52 B, no 1, pp 1-13.

Cadwallader, M (1992) Migration and Residential Mobility. Macro and Micro Approaches. Madison:The University of Wisconsin Press.

Clark, W.A.V. (1993) Search and Choice in Urban Housing Markets. I: Gärling, T. and Golledge, R.G. (ed.), Behavior and Environment: Psycological and Geographical Approaches. Elsevier Science Publishers B.V.

Clarke, M.och Holm, E (1987) Microsimulation methods in spatial analysis and planning. Geografiska Annaler 68B, nr 2, pp 145-164.

Clarke, M.,Duley, C. and Rees, P. (1989) Micro-simulation models for updating household and individual characteristics in small areas between censuses: Demographics and mobility. Paper for the International Migration Seminar, Swedish Institute for Building Research, Gävle.

Flowerdew, R. (1976) Search strategies and stopping rules in residential mobility. Institute of British Geographers Transactions, vol 1, pp 47-57.

Gärling, T., Garvill, J., Lindberg, E. and Montgomery, H. (1991) Residential choices and beliefs about future life satisfaction. Test of a modified multiattribute utlity (MAU) framework. Byggforskningsrådet, rapport D6:1991.

Huff, J.O. (1982) Spatial aspects of residential search, I: Clark, W.A.V. (ed.), Modelling Housing Market Search. London:Croom Helm.

Hägerstrand, T. (1970b) What about people in regional science? Ninth european congress of the Regional Science Association. Papers of the Regional Science Association. Philadelphia.

Hägerstrand, T. (1972) Om en konsistent, individorienterad samhällsbeskrivning för framtidsbruk. Specialarbete S2 till SOU 1972:59, Justitiedepartementet Ds Ju 1972:25.

Montgomery, H. (1989) From cognition to action: The search for dominance in decision making. I: Montgomery,H. and Svenson, O. (eds.), Process and Structure in Human decision making. Chichester:Wiley.

Montgomery, H. and Svensson, O. (1976) On decision rules and information processing strategies for choices among multiattribute alternatives. Scandinavian Journal of Psychology, 17, pp 283-291.

Noortwijk, L. and Hooimeijer, P. (1990) Divorce and the disruption of the housing career.

Orcutt, G. (1960) Simulation of Economic System. The American Economic Review, vol 50, no 5, pp 893 - 907.

Payne, J.W. (1982) Contingent Decision Behavior. Psychological Bulletin 92, pp 382-402.

Rossi, P. H. (1955) Why Families Move. A study in the social psychology of urban residential mobility. Glencoe:The Free Press.

Timmermans, H.,Molin, E. and van Norortwijk,L. (1994) Housing choice processes:stated versus revealed modelling approaches. Netherlands Journal of Housing and the Built Environment, vol 9,no 3, pp 215-228.

Öberg, S. (1979) Matchningsprocesser på och mellan lokala arbetsmarknader. En individbaserad simulering av bl.a. flyttningar i samband med arbetsbyte. Rapport F12, Expergruppen för forskning om regional utveckling.

Top of page

Notes

1 Hägerstrand (1970)

2 Hägerstrand (1972)

3 Bysveen and Knutsen (1987)

4 Öberg (1979), Orcutt (1960)

5 Clarke and Holm (1987), Clarke, Duley and Rees (1989).

6 Timmermans et.al. (1994).

7 Rossi (1955)

8 Noortwijk and Hooimeijer (1990)

9 In the model it is assumed that all individuals commencing a process of separation will request another dwelling.

10 Clarke, Duley and Ress (1989)

11 Cadwallader (1992).

12 Household budget is here referred to as an average for each household group respectively. Data is a compilation of the Housing and rental surveys (BHU). The household groups are defined based on the age of the reference person. Age <30 years : 24%; 30≤ Age < 40 years: 22% ; 40 ≤ Age < 65 : 22% and 65 ≤ Age : 21%

13 Brown and Moore (1970), Flowerdew (1976)

14 Barrett (1976), Huff (1982), Clark (1993)

15 Montgomery (1989).

16 Cadwallader (1992), Gärling, Garvill and Lindberg (1991)

17 Montgomery (1989), Payne (1982)

18 Montgomery and Svensson (1976).

19 National census 1975 and 1990.

Top of page

References

Electronic reference

Urban Fransson, « The relationship between households and the housing market: A microsimulation model of household formation among the young », Cybergeo: European Journal of Geography [Online], Topics, document 135, Online since 13 June 2000, connection on 28 March 2024. URL : http://journals.openedition.org/cybergeo/897 ; DOI : https://doi.org/10.4000/cybergeo.897

Top of page

About the author

Urban Fransson

Urban.Fransson@ibf.uu.se
Institute for Housing Research, Uppsala University, Sweden

Top of page

Copyright

CC-BY-4.0

The text only may be used under licence CC BY 4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search