Ottawa Group 10 th Meeting Ottawa October 9

Ottawa Group 10 th Meeting Ottawa October 9 -12, 2007 The Paris OECD-IMF Workshop on Real Estate Price Indexes: Conclusions and Future Directions Erwin Diewert Department of Economics University of British Columbia

Overview • What are appropriate target indexes? • The fundamental problem • 4 methods for constructing indexes • Hedonics and the land structures decomposition problem • What is the “right” price for the services of OOH?

What are appropriate target indexes? • • Fenwick argued for a coherent conceptual framework Diewert argued that real estate price indexes should serve the SNA in the first instance and a Durables Augmented System of National Accounts could provide a coherent framework

We need two types of Real Estate index: • An asset type price index for constructing real wealth measures (and for implementing the acquisitions approach) • A service flow type measure to measure the price of the use of real estate for a period (rental equivalence or user cost type of indexes)

The next revision of the SNA will introduce capital services into the accounts as a further decomposition of operating surplus (on a noncompulsory basis) so not only will the services of OOH be required in these augmented accounts, the services of structures, land inventory will also make their appearance. These changes are driven by the need to measure productivity growth and make international comparisons of productivity

The fundamental problem The traditional matched model for constructing price indexes fails in the case of real estate price indexes because: • The structure depreciates over time (the depreciation problem) and • The owner of the structure may renovate the property and it is difficult to track these expenditures (the renovations problem)

Four suggested methods • The repeat sales method (Has many problems) • Assessment Information The problems of renovations and depreciation remain plus how are the assessments made? But may be OK • Stratification methods The Australians will explain all! Is used with sales data and may suffer from unit value bias and small samples in cells if the stratification is too fine.

Four suggested methods (cont) • Hedonic Methods • Stratification is a special case of a hedonic method • Is probably the best method from a scientific point of view • Can decompose the property value into a structures and a land component which no other method can do

Hedonic Methods (cont) • But to solve the depreciation and renovations problems, good information on renovation expenditures is needed (OK on firm side but not on household side) • The method requires a lot of information (bad) • The method is not completely reproducible (bad)

Decomposition of property price into land structure components (21) pn 0 = [ An 0 + Bn 0] n 0 Now take logs of (21) and get (22) ln pn 0 = ln[ An 0 + Bn 0] + n 0 The above applies to period 0; for t: (24) ln pnt = ln[ t. Ant + t. Bnt] + nt t is the period 0 to t price index for the type of structure and t is the period 0 to t price index for the land that is associated with this type of structure.

Equations (22) and (24) can be run as a system of nonlinear hedonic regressions and estimates can be obtained for the 4 parameters, , , t and t. Generalization to many characteristics: (25) ln pn 0 = ln{[ 0+ k=1 K Xnk 0 k]An 0 + [ 0+ m=1 M Ynm 0 m]Bn 0} + n 0 (26) ln pnt = ln{ t[ 0+ k=1 K Xnkt k]Ant + t[ 0+ m=1 M Ynmt m]Bnt} + nt

• The problem with equations (25) and (26) is that the hedonic regression is nonlinear which can be a problem with large data sets • But the method has the advantage that it can decompose the real estate value into land structures components • This is required for productivity analysis and will be required if countries decompose operating surplus into capital price and quantity components

User costs versus rental equivalence There are two ways to construct prices for the use of a durable consumer good or a capital input in production: • The rental or leasing price that the durable would fetch on the market (a type of “physical” opportunity cost) or • Its user cost, which is a financial opportunity cost.

User costs versus rental equivalence (cont) • User costs are well established in production function or productivity studies but they are often scoffed at in the household context. • But Iceland has shown that user costs can successfully be used to price the services of OOH • But which of these two prices is the “right” one?

User costs versus rental equivalence (cont) • I used to think that it did not matter very much whether we used rental equivalence or user costs to price the services of OOH; in the long run, these two concepts should give the same answer. • Why? Because the considerations that go into forming user costs must be the same considerations that landlords look at in setting rents for their properties.

User costs versus rental equivalence (cont) Two “facts” caused me to question the equivalence of user costs to rents: • User costs fluctuate much more than rents; i. e. , as a property boom occurs, the ratio of rents to asset price will typically fall dramatically (I could live with this fact); • In a cross section, as the asset price rises, the rent to asset price for OOH will fall (to about ½ the level for low asset price properties). (This is bad news).

User costs versus rental equivalence (cont) What causes these divergences of rents to asset price ratios? • High transaction costs in buying and selling houses (Randy Verbrugge) and • Market segmentation. Mainly poorer people rent; if they get rich, they buy their dwellings for security or whatever and poor people simply cannot pay high rents

User costs versus rental equivalence (cont) Back to the question then of which price is the “right” one? I now think that the “right” opportunity cost for OOH is the maximum of what the dwelling could rent for and its user cost. Both costs are opportunity costs and the “true” opportunity cost should be the maximum of the alternatives foregone.

User costs versus rental equivalence (conclusion) Note that this maximum opportunity cost approach would definitely solve the problem of negative user costs. My conclusion is that statistical agencies should pick their preferred treatment of OOH but also provide as analytical series an acquisitions price index; a rental equivalence index and a user cost index so that the needs of all users can be met.