Rents in private social housing

Transcription

1 Rents in private social housing Mary Ann Stamsø Department of Built Environment and Social Science Norwegian Building Research Institute P.O. Box 123 Blindern, NO-0314 Oslo, Norway Summary This paper discuss if the rents are averagely above market price, in the private rent sector for tenants receiving social security. The empirical analysis uses data from a proceeding inquiry where registration scheme with questions about new tenancy agreements had been filled out by social security offices in two cities, Oslo and Trondheim. In the analyses this data are compared with data from Norwegian surveys on living conditions and data on rents in the private market in Oslo and Trondheim. Besides a direct comparison, hedonic regression is used to show the price on different properties in the two markets. The main question is if there exist different average rents in these two markets (submarket), and if there exist price discrimination. The conclusion is that price difference exists and that the rents are above market price for small dwellings in Oslo, but not in Trondheim. 1. Introduction The Norwegian housing market differs from other countries that natural compares with Norway, when it comes to housing policy. The rented sector is in Norway 24 % where public rented housing is only 4 %. Also the private rented sector differs from other countries. In Oslo professional landlords only cover for 29 % of the private rented sector, which then means that single persons cover for 72 % of the private rented sector. Low income groups then have to rent in the private housing sector, from both professional landlords and single private persons. Social security is also a common mean in the housing policy in Norway. In 1998 housing allowance towards low income groups was in sum 1, 4 billion NOK and social security to cover housing expenses (mostly rents) was in sum 2, 2 billion NOK (Stamsø and Østerby 2000). Compared with other countries social assistance in Norway is residual and very local. The level of benefit is generous, but the means test is very tough and therefore social controlling regime and stigma probably relatively high (Bradshaw, 1996). In Norway receivers of social security are in some extent been discriminated in the private rented housing sector. Some of the landlords don t want to rent dwelling to anyone that receive social security. They argued that the shut out are due to risk differentials, because this group in a larger degree damaged the dwellings or don t pay the rent. In additional they are also afraid of renters having drug problems or psychiatric problems among this group. Anyway, people with that kind of problems will also easier get a public rented dwelling,

2 since the claim to get that is social problems in addition to economic problems. The social security offices are also careful about getting dwellings on the private market to claimers with such problems since they are dependent of a good relationship to the landlords. On the other hand some landlords don t fear these problems because the social security offices give a social guarantee that cover for such risk. Some landlords even prefer renters that receive social security and offer the social security offices to rent dwellings, also above market price. (Stamsø and Østerby 2000, Stamsø 2002). It is the social guarantee that shows that the tenants are receiving social security, and then open the possibility for the landlords to either refuse the tenants to rent or demand rent above market price. The social guaranties cover four months rent and last for the period that the tenancy agreements are lasting. In Oslo 84 % of the tenants receiving social security where using social guaranties and 29 % in Trondheim. In some cases the social security offices gives guarantee as ordinary deposit. In Oslo 6 % was given deposit and 32 % in Trondheim. Deposit doesn t necessarily show that the tenant is receiving social security. In Oslo 10 % have no kind of guarantees and 39 % in Trondheim. Both the social guaranties and the deposit that the social security uses cover for a higher amount, and the use of guaranties is also more common then in the ordinary market (Stamsø and Medbye ). There have been speculations and assumptions that receivers of social security are paying rents above market price, because they have problems finding a dwelling and since social security offices have a responsibility to help them finding a dwelling (according to the social law), and then also may be willing to pay more if necessary. These assumptions were the reason why an ongoing project examines this, and the paper is based on result from this project. A possible prise difference will be examined. If those who receiving social security have to pay more and this cant be explained by higher risk the difference in price is to be considering as price discrimination. Price discrimination is selling the same goods to different price to different demanders, and where it is an advantage to the seller. There are tree conditions that must be fulfilled to accomplish price discrimination (Ringstad, 2000): i) Different demanders or groups of demanders, with different willingness to pay ii) The groups must bee possible to separate iii) Sale between groups must not be possible Condition number tree is automatically fulfilled because this is not allowed in Norway (and not possible for this group). According to the Landlord and Tenant (Rent Restriction) Act, the rent should not exceed the level of rents at the same dwellings at the same region (market rent). 1 Stamsø, M. and Medbye, P. (2004): Privat rent towards low income groups. Norwegian Building Research Institute not published yet 2

3 2. Data Low income groups are hard to reach through surveys. Registration schemes with question about rent, dwelling 2, tenancy agreement, landlords and guaranties, then has been filled out by social security offices. I Oslo five offices from different geographical regions where selected, and all six in Trondheim. Data where collected in a period from the 20 of January to the 20 of July in The numbers of observations in the samples are 284 in sum, 171 in Oslo and 113 in Trondheim. Only new tenancy were registered because it should represent market rents and compared with other new tenancy. This data is compared directly with data on market rents based on advertisements 3. Data on market rents for tenants dos not exist. This means that these data are approximations since the demand rent does not equal actual rent in all the tenancy. When market prices are declining the demand prise in the advertisements usually will bee above the actual price. In the period the data was collected the market price was declining. The demand prices are in average above market price since the landlords are testing the market price and advertising several times, with higher prices the first time and then lower (Oslo County, 2003). The data from the registration schemes are also compared with data from Norwegian survey on living condition. Because the rented sector is small in Norway, also the numbers of observations in this survey are small, specially when consider only Oslo and Trondheim. Data on unpaid rents and damages on the dwellings where also collected. But these were to insufficient to use. But for those offices that had the knowledge of that, it was for the most unpaid rents. The one office (in Trondheim) with most such problems, all the loss where covered for by the social guarantee, which means that the landlords didn t loose money. 3. Method Data on rents for a sample of tenants receiving social security is compared with data on rents for all tenants. As noted above it is not possible to direct compare the market price on rented dwellings for the group receiving social security to the market price for all tenants, due to the lack of data of market price for all tenants. Nevertheless, it is possible to observe approximation on market price for all tenants are based on demand rents from advertising. If the rents in average are at a higher level for the group receiving social security it is then possible to identify difference since we know that the demand rents in advertisements in average will be at a higher level then actual rents. Hedonic regression is an important means to analyzing commodities that are heterogeneous, or are complex. It is then proper to use to predict the affect that various attributes (region, number of room etc.) have on the price. In this paper hedonic regression is used to predict the affect on the rents that various attributes have in two different markets. One market is a sample of tenants receiving social security and one market is a sample of all tenants. 2 Hospice and block of bed-sitters with temporary tenancy agreements and payment per night and week are not included. 3 Report on demand rents in the private market 2003, made by Oslo and Trondheim County. 3

4 4. Results Different in rents One problem with the direct comparison between data on rents from the social security offices and the market price is that the data over market price have a category for lodgings. For the comparison the data from the social securities are then categorised in two ways one where lodgings are categorised as dwellings below 25 square s and one where lodgings are categorised as housing where the landlord lives in the same house. The problem with the first categorising is that only 30 % of the observations on rents also included the size in square s and mostly of the observations will then bee lost. The other comparison is not suitable for Trondheim, since those dwellings are more what we in Norway calls lodgingsg dwelling, which means that they filled the claims to be a dwelling even if the landlord lives in the same house. Those dwellings are common in Trondheim, but not in Oslo. Table 1 and 3 provides the result of average market prise on rents for a sample of tenants receiving social security in Oslo and Trondheim. Table 2 and 4 provides the demand price based on advertisements as a proxy on the market price of rents in the private market in Oslo and Trondheim. Table 1. Rent, square and price per square. Tenants receiving social security in Oslo. Lodgingss classified as dwelling below 25 square Lodgingss classified as housing where the landlord lives in the same house Lodgings 1-room 2-rooms 3-rooms Lodgings 1-room 2-rooms 3-rooms Rent (1015) N=18 Kvadrat Kvadrat pris (432) N= (1357) N= (2009) N= (1322) N= (1468) N= (1251) N= (794) (364) Source: registration shemes (328) N= (186) (1246) N= 4 Table 2. Rent, square and price per square. All tenants in Oslo Advertisements Lodgings 1-room 2-rooms 3-rooms Rent (703) N=682 Square Price per square (889) N= (1264) N= (704) N= (875) N=332 Source: Oslo County (1353) (1898) N= (2023) N= (694) N= (350) (1543) N= (186) 4

5 From table 1 and 2 we can see that the rents are at an averagely higher level for small dwellings for the group receiving social security, and the opposite for dwellings with 3 rooms. T-test shows that the difference is significant on 5 % level for lodgings and 1- room dwelling (when lodgings are classified as housing where the landlord lives in the same house) and for square price for 1- room. Since the rents in the total market (table 2) are advertised price and not actual price, the difference for small dwellings will bee higher. For the same reason rents in dwellings with 2 rooms probably will be almost equal. It is small dwelling that matters most, because mostly in the group with social security lives in small dwellings. In Oslo 62 % lives in one room and 39 % in Trondheim, 25 % lives in two room in Oslo and 37 % in Trondheim. In the total private rented housing market 30 % lives in one room in Oslo and 27 % in Trondheim, and 42 % in two room in Oslo and 49 % in Trondheim. Table 3. Rent, square and price per square. Tenants receiving social security in Trondheim Lodgingss classified as dwelling below 25 square Lodgingss classified as housing where the landlord lives in the same house Lodgings 1-room 2-rooms 3-rooms Lodgings 1-room 2-rooms 3-rooms Rent N=37 Square Price per square Source: registration shemes N=37 N=18 N=49 N=26 N= N= N= N= N= N=9 Table 4. Rent, square and price per square. All tenants in Trondheim Advertisements Lodgings 1-room 2-rooms 3-rooms Rent N= N= N=419 Square Price per square N= Source: Trondheim County From table 3 and 4 we can see that the tenants receiving social security don t pay a higher rent then other tenants in Trondheim. Mostly data shows that they pay less. But mainly this difference can be explained by the difference between demand rent trough advertisements and the actual rent. The difference in rents between the groups of tenants will then be smaller. The reason why averagely rents in lodgings are at a higher level when lodgings are classified as housing where the landlords lives in the same house, is that many of these are bachelor flat and not lodgings. This is common in Trondheim, but not in Oslo in the same degree, and especially not in the centre of the city where this data are collected from. 5

6 Table 5 provides hedonic regression on observations for all tenants and for the tenants receiving social security. The objective is to determine the role that various housing characteristics play to value the rents. In table 5 model 1 represent all tenants and model 2 those with social security. Table 5. Hedonic rent regression Dwellings hired by all tenants and the those who receive social security model Model 1 4 Model 2 variable Constant 2124,49 (1,03) 5062,73 (15.09) Numbers of room 1060,44 (3,99)* 635,86 (6.01)* D Trondheim -1843,96( -2,29)* -2449,39 (-8.52)* D Region Oslo1-250,41 (-0,31) -167,16 (-0.65) D Region Oslo2-661,03 ( -0,82) (0.41) D Region Oslo4 196,66 (0,20) (3.68)* Bath 1243,59 (0,59) (-0.28) Length of tenants 3,98 (0,29) (-0.76) agreements R 2 0, N Notes: T-value is reported in parentheses. * Reports if the value is significant on 5 % level. The signs of the coefficients are as expected. For both models numbers of rooms are positive and significant at 5 % level. The dummy variable for Trondheim is negative and significant at 5 % level. Also one region in Oslo has a significant effect on price at 5 % level in model 2. There is some difference between those two markets. The coefficients are different in the variable of Trondheim and number of room. The rent increases in a larger degree by an increase in number of room in model 1 then in model 2. This indicates that the market in model 2 is less flexible, where this variable does not affect the price in the same degree as in model 1. The differences in rents between tenants receiving social security and all tenants that only occurs for the small dwellings from table 1 and 2 is then supported by table 5. The reason may be that the social security offices are willing to pay more for small rooms, and that they have an upper limit for what they will pay in rents that reduces the price on the larger dwellings. The variable of Trondheim affects the price in a larger degree in model 2 then in model 1. The geographical differences are more significant in model 2. This means that tenants receiving social security pay a relatively higher price/level of rent in Oslo then in Trondheim, compared with all tenants. The differences in rents between tenants receiving social security and all tenants that only occurs in Oslo from table 1-4 is then supported by table 5. Submarkets are usually defined in terms of geographical areas or physical characteristics of the dwelling. Since the market for private rents are spread over different geographical aeries, this submarket also will cover different geographical aeries. Submarket may be defined by structure type, e.g. single families detached (Goodman and Thibodeau, 2003). The large share of private single persons as landlords also has the effect that the housing market for poor household being less segregated. The landlords are spread in different aeries and since they 4 Model 1 is based on data from Norwegian survey of living conditions which means that the rents not only represent new tenancy, and then market price, but also older tenancy. This only affects the rent level, but probably not coefficient value. 6

7 hired out usually one dwelling. The submarket is then defined as private rented housing market that hired out dwellings to tenants receiving social security, and which is possible to identify. The typical approach to analyzing housing market segmentation involves estimating hedonic equations for various assumed or defined submarket. Submarket should be defined so that the accuracy of hedonic predictions will be optimized (S.C. Bourassa et al, 2003). These data are not appropriate for direct tests of submarket, since the data are from two different surveys. But we may also in this case se a submarket. The adjusted R 2 is higher in model 2 which indicate a more homogeneous market. And also the differences between the coefficients indicate a submarket. 5. Conclusions Data shows that tenants receiving social security pay rents above market price for small dwellings in Oslo. In Trondheim they don t. If this is to bee defined as price discrimination, we have to consider the risk. The data on risk are not sufficient to be considered. Anyway, other factors indicate that price discrimination is present. It is difficult to explain rent above market price according to risk, since it only occurs in Oslo, and only for small dwellings. One possible explanation is that in Trondheim the conditions for price discrimination are not fulfilled, since it is not possible to identify the group who is discriminated in the same degree. This is because social guaranties only are used in 29 % of all the tenancy. Another possible explanation is that the supply of dwellings for tenants receiving social security is not reduced because of discrimination in Trondheim as in Oslo. The housing markets in these two cities are quit different, where the market in Oslo is more expensive and the share of rented dwellings are smaller. Then the condition of different willingness to pay is not fulfilled, since tenants receiving social security then easier get a rented dwelling. 6. Acnowledgements I will thank my colleague Per Medby who has been working on the project and made the regression based on Norwegian surveys on living conditions. References Bourassa, S.C., Hoesli, M. and Peng, S. (2003): Do housing submarket really matter? Journal of Housing Economics 12, Bradshaw J. (1996): Sosialhjelp I Norge. Spesiell I et internasjonalt perspektiv? Article in Terum, L. Sosialhjelp endring og variasjon. Norwegian research council. Goodman, A.C. and Thibodeau, T.G. (2003): Housing market segmentation and hedonic prediction accuracy. Journal of Housing Economics 12, Norwegian Statistics (2001): Norwegian survey of living conditions. Oslo County (2003): Leiekrav I det private markedet Ringstad, V. (2000): Samfunnsøkonomi og økonomisk politikk. Cappelen akademisk forlag. Stamsø, Mary Ann og Østerby, Steinar (2000): Forholdet mellom bostøtte og sosialhjelp. Norwegian Building Research Institute Report 288. Stamsø, Mary Ann (2002): Privat profesjonell utleie til sosialhjelpsmottakere. Norwegian Building Research Institute Report 52. Trondheim County (2003): Leiekrav I det private markedet