1. IMPUTED RENT IN HOUSEHOLD BUDGET SURVEY Inga Masiulaitytė, August 24-28, Ventspils, Latvia 2006

2. Content  Definition of the imputed rent  Methods  Results  Conclusion and proposals

3. Imputed rent The imputed rent refers to the value that shall be imputed for all households that do not report paying full rent, either because they are owner- occupiers or they live in accommodation rented at a lower price than the market price, or because the accommodation is provided rent free.

4. Methods Self – assessment method Homogeneity groups method Heckman method Log-linear method

5. Self assessment method Self – assessment method is based on the owner- occupiers answers about potential rent for their dwellings. This method is subjective method, and can not show real situation in the rental market.

6. Homogeneity groups method The dwelling are disjointed in quantity groups so that the rental value of dwelling within each group is a homogenous as possible. Data on actual rents paid within each group are obtained. The average rent paid in the group to which dwelling belongs is imputed to the owner-occupier.

7. Heckman method 1 The sample selection model consists of two equations The first equation indicates if target variable is observed, i.e. if there is a response or a nonresponse: here z k is a value of a vector of auxiliary variables,  is a vector of parameters, k is the error term, a random component. We only observe whether it exceeds a certain threshold, say 0, because that results in the response or nonresponse:

8. Heckman method 2 The second equation assumes a linear relationship between the target variable y and a vector of auxiliary variables here is a random component, is a vector of parameters that have to be estimated using the sample data. is also a latent variable. The study variable y is defined by:

9. Log-linear regression Lets us introduce a variable l with the values: Then the log-linear model can be defined as: here – is vector of coefficients, – are values of known auxiliary variables, – are random errors.

10. Population and sample design Nr.StrataPopulationSample 1Vilnius1997206 2Kaunas1854191 3Klaipėda76479 4Šiauliai62564 5Panevėžys62865 6Other cities6262645 7Rural areas7000750 Total191302000 Stratified random sample is used. Replication number is 1000.

11. Auxiliary variables Location of dwelling (strata) Type of dwelling Number of rooms Water Hot water Electricity Heating Gas Kitchen Kabel TV Garage Sewerage Phone

12. Results 1 Empirical average of estimates of mean rent using different imputation methods by strata (5 per cent of tenants)

13. Results 2 Empirical average of estimates of mean rent using different imputation methods by number of rooms (5 per cent of tenants)

14. Results 3 Empirical average of estimates of mean rent using different imputation methods by strata (10 per cent of tenants)

15. Results 4 Empirical average of estimates of mean rent using different imputation methods by number of rooms (10 per cent of tenants)

16. Results 5 Empirical average of estimates of mean rent using different imputation methods by strata (30 per cent of tenants)

17. Results 6 Empirical average of estimates of mean rent using different imputation methods by number of rooms (30 per cent of tenants)

18. Results 7 Empirical average of estimates of coefficient of variation of rent using different imputation methods

19. Conclusion The imputed rest estimated by Self-assessment method is better to use when we have 5 per cent of tenants. The imputed rent estimated by Homogeneity groups method has a smaller bias than estimated by Heckman or Log-linear methods. Proposals Test these methods on EU-SILC data. Test influence of new auxiliary variables on the estimation of rent.

20. Reference Jeroen Smits, (2003) Estimating the Heckman two-step procedure to control for selection bias with SPSS. http://home.planet.nl/~smits.jeroen/selbias/Heckman- SPSS.doc.http://home.planet.nl/~smits.jeroen/selbias/Heckman- SPSS.doc WEN Hai-zhen, JIA Sheng-hua. GUO Xiao-yu, (2004) Hedonic price analysis of urban housing: An empirical research on Hangzhou, China, Journal of Zhejiang University SCIENCE, 6A(8), p. 907-914. EU–SILC DOC TFMC-12/06. Third Meeting of the EU-SOLC task force on methodological issues. Imputed rent.. Eurostat-Luxembourg, April 2006, EU-SILC DOC TFMC-12/06. DOC HBS/161/2006/EN, DOC EU-SILC/162/06/EN. Meeting of the working group on living conditions (HBS, EU-SILC and IPSE). HBS and EU-SILC Imputed rent.. Eurostat- Luxembourg, May 2006. Bethlehem J., Cobben F., Schouten B., May 2006, Nonreponse in Household Surveys. Version 1, Statistics Netherlands, Methods and Informatics Department, p. 141-142. Bradley C. Martin, Rahuk Ganguly, SAS program for Heckman 2-stage estimation. University of Georgia, College of Pharmacy, Athens, GA 30602, http://www.rx.uga.edu/main/home/cas/faculty/heckman.pdf http://www.rx.uga.edu/main/home/cas/faculty/heckman.pdf Eurostat Phare MultiCountry Programme, Organise by ICON-Institute GmbH/Germany of behalf of Eurostat. Report on the Methodological Study: “Possible data sourses and methods to calculate the target variable “imputed rent” for Lithuania”. Action code: 1138-ILC-SerCon-Lithuania-1

21. THANK YOU FOR YOUR ATTENTION!