Statistics Iceland has published the working paper "Owner occupied housing in the Icelandic Consumer Price Index". The paper describes details about the method for calculation of owner occupied housing in the period 1922-2019 and the CPI results.
The expenditure weights for housing in the Icelandic Consumer Price Index have been estimated in three different ways over the years. In the period 1922-1968 the weights were estimated from payed rent for a house, between 1968 and 1992 the expenditure weights were based on the estimated cost of owning a house using Household Expenditure Survey data where the weights included financial cost and reflected a payment method, and from 1992 the weights were calculated as user cost based on the real estate value of a house with the aim to measure flow of services.
The main use of the CPI in the period 1922-1984 was for indexation of wages. In the period 1922-1992, two different input price indexes were used for price updating of housing. Rent index that was input price index including wages (45 per cent) and material (55 per cent) or Building cost index. In the years 1939-1968 only part of the price change measured by the rent index was used (15 per cent) in the compilation of owner-occupied housing. Even if the weights were decided by estimated rent payed, the price change of rents was not used in the CPI compilation. These measurement methods underestimated systematically the CPI price change.
Various use of the CPI through the years
From 1992 user cost method was applied for measuring owner occupied housing and the price change measured by a house price index and real interest rates. The index was to measure the flow of services as imputed rent. The CPI was no longer mainly a wage compensation index but general price indicator and the CPI measured more price changes than for the CPI less housing. According to the 1995 CPI act the scope for the CPI was to measure prices for private consumption and a flow of services method chosen for owner-occupied housing in line with national accounts.
In 1997, actual rents and imputed rents for owner-occupied housing were separated in the compilation of the CPI with own weight and price measurement of actual rent. Rent is the payment for using a property over given time period. In Iceland the rental market is small as only 20 per cent of properties are rented and only half of tenants pay market rent as the other half pays subsidized rent.
Actual rent and imputed rent should in theory move in similar fashion. During 1992-2019 these two series measured approximately the same price change. Over this period there have been different phases of price movements. In the years 2005-2008, imputed rent increased more than market rent and in the years 2009-2016 market rent changed more than imputed rent. The latter was a volatile economic period which is often considered to begin with the banks entering the property mortgage market in 2004, to peak with the bank crises in 2008 and to conclude with the economic recovery that followed.
Even if imputed rent both rose and fell during the period 2004-2016 the CPI did not change steeply. Imputed rent is counter-cyclical to the CPI less housing and stabilised the CPI total results. The exchange rate has strongly affected the changes in the CPI less housing as approximately one third of expenditures were imported goods. Owner occupied housing is not directly influenced by the exchange rate but is decided by the prices of properties through actual rent and imputed rent. The fact that these factors balance each other out stabilises the CPI and is a very important characteristic of the Icelandic CPI.
Different aims when using the CPI
Three countries (Iceland, Sweden and Canada) calculate their owner-occupied housing as a user cost. Iceland aims at calculating general price indicator where all prices used in the CPI price measurement reflect the current price level. Sweden and Canada use payment methods and the aim of the calculation is compensation. The prices of the capital stock is always on average 12-15 years old and the price level reflects prices from earlier time periods.
Simulation of the Swedish user cost method with Icelandic data show that over the period 2007-2019 the overall price change of the CPI is similar by both methods. The Swedish method is though much more volatile using Icelandic data, partly reflecting higher rate of price change in Iceland compared to Sweden.