Owned Accommodation in Consumer Prices
Introduction
This note was stimulated by the release by Statistics Canada of a paper defending their approach to measuring the rate of cost change for owned accommodation. https://www150.statcan.gc.ca/n1/en/pub/62f0014m/62f0014m2024002-eng.pdf
The authors (Sabourin and Tarkhani -ST) summarized their defence of the status quo. The purpose of this note is to raise some issues for discussion and to make a case for an alternative approach, the rental equivalence which is used in other major countries.
Owned Accommodation in Social Accounts
In the social accounts of most countries, owned accommodation is treated as a consumption of housing services that is matched by an equivalent income estimate commonly referred to as “imputed rental fees”. This treatment attempts to balance renters and owners in the analysis of income and expenditure. Owners are shown as paying for a housing service with a balancing income item. Owned accommodation is not explicitly treated as an asset in the current consumption or GDP.
In the consumption accounts, maintenance expenditure undertaken to maintain the value of a property is treated as part of current consumption and is not considered capital expenditure.
The Consumer Price Index (CPI) is intended to measure price changes experienced by all consumers. It is intended to inform consumers as well as central bankers. It is a plutocratic index that weights the expenditures of all households as a group, so it represents more than the purchases of any specific household. One standard reference, the Consumer Price Index Manual (CPIM – my text) was published jointly by a number of international agencies including the World Bank and the ILO. The link is https://webapps.ilo.org/wcmsp5/groups/public/---dgreports/---stat/documents/publication/wcms_761444.pdf
The CPIM does not specify a required approach but does discuss the options available. The discussion notes the data dependency of the choice of treatment. The r report was finalized in 2020. It is expected that the challenges of the last several years might change some of the discussion significantly.
Owned Accommodation in the CPI
Possible Approaches
ST outline a number of approaches:
· User cost – total cost of home ownership,
· Rental Equivalence – cost of home ownership if owners rent to themselves,
· Payment approach – cash costs of home ownership,
· Net Acquisitions Approach – changes in acquisition, renovation and maintenance cost of house.
StatCan Approach
ST define the StatCan approach as a variant of the user cost approach in measuring the costs to user and maintain a home but not the direct purchase. The costs measured are:
· mortgage interest cost (MIC),
· replacement cost,
· property taxes,
· home and mortgage insurance,
· maintenance and repairs,
· other owned accommodation expenses (such as real estate commissions).
ST define replacement cost as reflecting the depreciated portion of the house to reflect the cost to maintain its value. As such, there may be at least a partial explicit double counting because maintenance and repairs are separately included. The latter reflects the actual cost of living in a house along with taxes and insurance.
The estimation of mortgage interest requires an analysis of the structure of mortgages and the extent of properties without mortgages. However, the inclusion in the CPI makes the policy setting of our central bank explicit to Canadians but introduces circularity.
For owned housing, the CPI includes estimates for housing maintenance expenditures. For example, the fees paid by condo owners for the maintenance of the common elements of the property are included in maintenance expenditures. These normally support a fund which covers maintaining the value of the property, but special assessments may be used if the standard fees are not adequately estimated.
International Approaches
ST provide an excellent summary of the approaches in major international markets. The different approaches reflect available data and requirements.
One relevant point is that many European countries and the US use a rental equivalence approach. According to ST, the EU is piloting the net-acquisition approach in their Harmonized Index of Consumer Prices (HIPC). Sweden uses a user cost approach, but the central bank (Riksbank) targets a specially constructed CPI (CPIF) that excludes mortgage interest. https://www.riksbank.se/en-gb/monetary-policy/the-inflation-target/how-is-inflation-measured/
Rental Equivalence Issues
ST state that rental series may have accuracy and estimation challenges. However, these are used elsewhere in the CPI and in the income and expenditure accounts. One challenge with rental equivalents is estimation of existing rental contracts. This is successfully handled in the current economic accounts which use rental equivalence.
There is always a bigger stock of rental properties than of new homes so that it is not clear that the new home prices are not subject to similar or stronger criticism. Much of maintenance and repair costs may be embodied in the rental agreements.
The rental equivalence treatment hides the impact of the policy rate from direct view and inclusion in the inflation estimate for targeting.
ST’s appendix suggests that the rent should be a current new contract rent which is complex to obtain. The income associated with this rent is fictitious and might be unlikely to bias spending behavior. If current owners rented their places out, they would have to pay similar occupancy costs elsewhere.
The rental price series in the CPI may not reflect the structure of owned accommodation, at least according to ST. However, published rental rates on offer represent the information set that is available to owners of property. The significant increase in investor-owned properties suggest that the rental market is maturing. Most types of dwellings seem to be part of the rental stock in major cities.
The ST simulations show that the rental equivalence approach using the published rental series offers the virtues of relatively low volatility and inflation. The published series may not be perfect but the perfect is sometimes the enemy of the good.
ST discuss maintenance costs in a rental contract. Most rental contracts are inclusive of maintenance. Building insurance is a transaction which is included. Most renters are expected to insure their contents. Condo dwellers are responsible for insurance to cover renovations.
Depreciation
It is not clear that a depreciation allowance is required lo estimate homeowner costs that are easily apparent to homeowners. The actual normal expenditures to maintain the value of the house are covered in maintenance and repair. Including a depreciation cost might be considered at least partial double counting. The case is made that the expenditure represents the decline in value in the structure from wear and tear or changes in taste. It is certainly not an expenditure in the normal household budget. The StatCan methodology does not reflect the capital gain which owning a house can provide. In many periods, this substantially offsets the user-cost of depreciation. However, its inclusion might result in negative weights.
Replacement or repair in the event of specific perils is covered in the insurance costs.
Depreciation is a concept used in business to shelter a portion of income from taxation to pay for maintenance, repair, and actual replacement. For the homeowner, there is no income to shelter so the offset concept may not be relevant. It is not clear that the allowance would be recognized in the planning of households.
The weight of the replacement cost reflects estimate of the depreciated value of the housing stock. The challenge is in the price index used for replacement cost. The index is the actual cost of building a brand-new house including developers margins but excluding land. This is equivalent to the direct purchase costs of a new home. Supply shortages for lumber and trades post 2020 had a large influence on the cost of new homes and were contributors to estimated inflation and volatility. It is not clear that material costs are as critical for home maintenance and repair but may be for renovation which is not included in current expenditure.
ST claim that the new construction cost methodology does not contribute to volatility. The post-shutdown experience suggests otherwise.
Adjustments are made for changes in the type of housing currently being sold. These adjustments may not reflect the historic pattern of home purchases in terms of size, building materials and preference for single and multiple unit options. The ST paper does not go into detail on these issues.
The use of the current cost of building a new home might reflect supply-demand balances. It is subject to the vagaries of zoning and land availability. It is also not clear that this is a relevant concern for existing homeowners. If the primary purpose of the CPI is to reflect the cost of living for homeowners, it is conceptually irrelevant. More critically, these supply-demand issues have created instability in the estimates for new homes in particularly regions. It is not clear that it is appropriate for central banks to target market imperfections with their broad-based policy rate.
In summary, the current StatCan approach, utilizing new construction costs, brings in the cost of new homes into the general CPI without explicitly acknowledgement. This certainly adds to estimated inflation but as noted may not be relevant to homeowners. It also contributes to volatility. It may be a transaction cost but not necessarily one that is relevant to existing homeowners.
Net-Acquisition Costs
The Net Acquisition cost approach uses the price for new and the resale price as the key variables but includes maintenance, taxes, and insurance. The EU seems to feel that the approach is appropriate for cost-of-living analysis. The treatment of land in Canada is considered by ST to be challenging. Since land is often sold separately and banked, it should be possible to survey the value change. The use of resale prices increases volatility because of market conditions but does reflect supply-demand imbalances.
ST critique the net acquisition approach because of the requirement to split out the land costs. This is easy for new construction but possibly less easy for resales. However, one approach might be to use the same allowance as for new builds in the area.
Housing as an Asset
The consideration of an owned house as an asset may be key to some approaches. If housing is an asset, capital gains should be considered to offset the costs of the house. This is an issue if the full price change of the house is included in the estimate. The capital gain offsets the indexation in the cost of the asset but is excluded in the Canadian depreciation methodology. This biases the results up. The asset value of the house could offset the cost of renting a dwelling if the property is sold. Rental equivalence methods avoid this issue completely.
One of the big challenges is large role played by investors in the housing market. Both land and structures become involved in that investment and affect pricing. Land as an investment is clearly an issue for shelter costs.
Mortgage Interest Costs
Mortgage interest costs, as a user cost, would possibly reflect the rising resale value of houses as the market evolves. In Sweden, these are excluded to avoid circularity. The Swedish central bank targets a CPI variant (CPIF) with fixed interest costs.
In “normal” times with a stable policy rate. The inclusion of mortgage interest costs would make very little difference to the inflation rate. The chart below shows the CPI trimmed mean measure w/o mortgage interest costs. The JCI trim line duplicates the published trim estimate except for rounding issues in the source data. The other line plotted follows the same methodology but excludes mortgage interest. The lines show the annualized 3-month variant of CPI Trim.
The basic point is that its inclusion biases the perspective on inflation. Without the mortgage interest cost, the 3-month inflation rate, annualized, is only 1.1%.
Discussion and Recommendation
The current StatCan/BoC approach has several challenges. These include:
1) the exposure of mortgage costs which escalate with changes in housing values,
2) the explicit estimate of maintenance, taxes, and insurance.
3) the volatility and supply-demand imbalances of the new housing market in the depreciation estimate which may not be relevant to existing homeowners.
The rental equivalence approach offers the virtue of simplicity and comparability. There are less transactions involved and the concept matches trading partners and social accounting traditions. It avoids the new housing price indices which may been distorted by monetary policy, zoning, and market factors such as land availability.
It is strongly suggested that rental equivalence approach be further evaluated and possibly adopted.
AI not used. If it is boring, it is Jacobson. All mistakes are my own.