A statistical system for. Residential Property Price Indices. David Fenwick

Transcription

1 A statistical system for Residential Property Price Indices Eurostat IAOS IFC Conference on Residential Property Price Indices Hosted by the Bank for International Settlements November 2009, Basle Panel Discussion David Fenwick

2 Some pertinent questions What do users need? Which techniques are most widely used? Can international best practice be identified or is a flexible approach appropriate to reflect local circumstances? Can existing data sources be improved? What criteria should guide the production of Residential Property Price Indices? User driven? How far should and can the handbook go in answering these questions?

3 User needs: some pertinent observations - aggregation Different aggregation procedures can produce different results But indices may not necessarily move differently

4 User needs: some pertinent observations - disaggregation (1) The change in house prices can vary between different regions (Academtrics Ltd)

5 User needs: some pertinent observations - disaggregation (2) House price inflation can vary at a localised level (Academtrics Ltd)

6 Can existing data sources be improved? Enhancing data sources. Collecting more house characteristics on forms recording transfer of ownership. Making better use of data. Integrating data sets & modelling E.g. Identifying relationship between offer prices for houses purchased with a loan and final transaction prices for all properties including cash sales (cyclical relationship?) Sampling, electronic transmission. Early (revisable) estimates based on partial data.

7 Which techniques are most widely used? No clear picture: data-led methodologies Country Data source All residential properties Australia (ABS) Belgium (National Bank) Canada (Statistics Canada) France (INSEE) United Kingdom DCLG) USA (Federal Housing Finance Agency) Administrative data from Values General Office Sales registration offices Sample of builders Notaries, registrations of change of ownership Valuations relating to completed mortgage approvals from Council of Mortgage Lenders Acquisitions index based on secondary mortgage providers Fannie Mae & Freddie Mac Excludes apartments & new houses Existing dwellings only New property only Flats in Paris, flats & houses in rest of France Only properties bought with a mortgage New homes excluded Geographical coverage Quality/mixadjustment Urban areas only Mix-adjusted Whole country Average price per sq metre Urban areas only Contractors selling price repeat sales? All areas Hedonic regression UK all regions Whole country urban & rural but only transactions involving Fannie Mae & Freddie Mac Hedonic regression Regression at sub-national level. Repeat sales? National index estimated using weighted. Aggregation /Weighting Stock of housing from 5 year Population Census Transactions at district level Average prices, weighted Transaction prices weighted by value of housing stock in reference period Average valuations, expenditure weighted Weighted average of price changes in nine subareas. Stock weights interpolated from 10- yearly survey Frequency Time-lag from reference period Quarterly days Yes Quarterly? Yes Monthly? Yes Quarterly 3-4 months Yes Monthly 6 weeks Yes Monthly 60 days based on 70% of transaction s, subsequent revisions possible Indices produc ed by others Yes

8 Criteria for future index production: user needs - comparison of recent trends in UK - need to look at the cost-benefits

9 Can international best practice be identified or is a flexible approach appropriate to reflect local circumstances? Two main options under data-led methodologies Create best achievable index in each country based on available data Or family of indices Compromise full comparability Identify the best target index i.e. conceptual Cost-benefits of practical options Agree appropriate methodology Consensus = Comparability Bear the cost of data collection Some compromises extensive data set approximately correct/comparable, fully comparable but few indices

10 Way forward Build on earlier work: Strategic approach Family of house price indices Agreed conceptual framework Greater international harmonisation Costs versus benefits Better and more systematic metadata Needs: More investment by Statistical Offices International effort/driver (one international agency to take lead responsibility similar to ILO resolution on CPIs) International consensus Implications for handbook

11 A statistical system for Residential Property Price Indices Panel Discussion End of presentation David Fenwick