Housing Standards 2004/2005
Financing Housing and Refurbishing Housing Estates

1. Monitoring and analysis of the prices of owner-occupied housing

1.2 An Analysis of the Prices of Owner-Occupied Housing

Housing prices can essentially be analysed from two perspectives. First, statically, for a particular moment in time, where we ask what the main factors of differentiation and variability are, that is, the main factors in the disparity of prices between individual types of housing, regions, locations, etc. In other words, we are looking for a response to the question of what is the main factor in the disparity of prices at a given moment in time, what factor has the biggest influence on housing price differences, why some flats are more expensive and others less expensive, what influences most the price of a flat or home at a given moment in time? Second, we can analyse prices in a time series, in terms of how they evolved over time, and rather than looking for the causes of the differentiation we ask what is the main "driver" of price developments over time, what are the main factors behind the general rise or fall in prices of owner-occupied housing?

In the first type of analysis, for the purpose of determining the factors of price variability, usually regression models are used to explain the variability of prices on the basis of the variability of their individual attributes - these are hedonic price models, that is, models similar to those used in the simple measurement of housing price developments and real estate value assessments.

The majority of models in the field of economic housing research deal with the second type of housing price analysis: the dynamics of housing price developments over time. They trace the main factors that influence average housing prices over time. This kind of knowledge could, among other things, make it possible to develop price predictions.

Models explaining the rise and fall of average prices of owner-occupied housing are based on the assumption that in the long term the price of housing reflects demand and supply in a given national housing market. On the demand-side influential factors include the decisions of future actors looking to acquire housing as to whether they will live in rented housing or own their own (their decisions influence in particular the interest rate level) and fundamental economic factors - real household incomes, the level of unemployment, demographic factors like an increase in the number of households or the number of inhabitants, and the introduction or abolition of state interventions and regulations.

On the supply side in the long term what is mainly decisive is the ratio of the real market value of existing flats and the cost of acquiring new flats (Tobin's Q) - a rise in the price of existing flats or a fall in construction costs can increase the supply of flats by stimulating housing construction.

In a dynamic analysis of prices in the long term many analysts ascribe the greatest significance to real household incomes (household incomes adjusted for inflation) and claim that in the long term a stable relationship forms between real household incomes and housing prices. This relationship is measured with the aid of the price-to-income ratio indicator (P/I) - the share of the average (median) price of a flat/house in the average (median) total annual net household income, measured in real values and adjusted for inflation (CPI).

Muellbauer and Murphy (1997) have demonstrated that the prices of owner-occupied housing are also sensitive to changes in real interest rates, as these directly affect the decisions of future actors looking to acquire housing about whether to seed rental housing or own their own housing (by comparing net rent and the so-called user costs of owner-occupied housing). According to Rooij (2003), a study by Henley and Morley (1999) demonstrated the strong negative dependency between the level of real interest rates and the real price of real estate in six countries of the OECD (France, Germany, Italy, Japan, United Kingdom, and the United States). Similarly, Aoki et al. (2001) showed that an unexpected fall in interest rates leads to an increase in the real prices of real estate. Rooij (2003) also mentions the study by Miles (2003), in which the decline in real and nominal interest rates is considered to be the main factor behind the rapid rise in the price of real estate in the United Kingdom in recent years. It is clear that interest rates do affect the prices of real estate, particularly in countries where mortgage loans with variable interest rates predominate. Rooij (2003) claims that changes in interest rate levels have a strong effect on housing prices particularly in those segments of the market where the demand exhibits relatively little price elasticity. He cites the Netherlands as an example of such a market.

One of the most important demographic factors affecting the demand for housing and indirectly also the price of owner-occupied housing over the long term is the rate at which new households are forming. Ortalo-Magné and Rady (1999) note in this connection that the incomes of young households in particular (aged 20 - 29 years), which are the most frequent first-time acquirers of owner-occupied housing in the United Kingdom and the United States, are a critical factor in explaining price variations. The expected growth in real incomes leads to a higher demand for housing on the part of those households, which is projected into a growth in housing prices in the sector with the cheapest "start-up" owner-occupied housing. Owners of "start-up" housing make a capital profit owing to price appreciation, which enables them to move into the sector of higher quality homes/flats. This results in a chain-effect, where the rise in demand and prices in the cheapest sector gradually overflows into other sectors of the housing market. The authors indicate that therefore housing prices are affected by both the size of individual age cohorts of the population and by the distribution of wealth among these cohorts.

Empirical studies looking at the volatility of prices in the housing market also stress the significance of microeconomic factors, which affect the price elasticity of the housing supply.

Taxes, subsidies, and other instruments by means of which the state intervenes in the housing market have a significant impact on housing prices particularly during periods of more substantial reforms, and more in the short term. Malpezzi (1996), for example, has studied the impact of regulatory measures on the prices of owner-occupied housing and rent levels in urban areas of the United States. The factors described above by no means represent an exhaustive account of the variables that researchers use to explain housing price developments in the long term, and is rather a selection of the most commonly cited among them. Kenny (1998) presents an overview of the factors that are usually presented as arguments in supply-side and demand-side functions in the simulation of housing prices, along with their impact on housing supply and demand (Table 1).

According to Meen (Sullivan, Gibb 2003) the most frequently used method of estimateing long-term price developments is the empirical rule combined with the price-to-income ratio (P/I). Although the majority of studies (Garratt 2001, Case, Shiller 2003, PricewaterhouseCoopers 2002, PricewaterhouseCoopers 2004) point to household income as one of the main factors influencing housing prices in the long term, it is far from the only factor. Meen notes that, particularly after 1990, estimates of price developments in the United Kingdom, especially in the short term, based only on the value of the P/I, were unreliable.

Another indicator used to determine national housing prices and estimate price developments (e.g. Holt 2003) is the price-to-earnings ratio (P/E), which is similar to the indicator regularly used in stock market in the assessment of shares. As with shares, in the case of real estate this approach is based on the assumption that the price of real estate is a function of the expected future returns on this real estate. However, in the case of real estate these returns are more difficult to quantify than, for example, in the case of stocks and bonds. The reverse value of the P/E indicator is the also used rent-to-price ratio indicator (used by McCarthy and Peach 2004).

Weeken (2004) explains housing price increases in the United Kingdom in recent years in an alternative way also using the model of capital asset pricing, which he applies to the housing market. He bases this approach on the Dividend Discount Model, which is used to value publicly traded stocks. A simple dividend discount model assumes that the correct value of a stock can be determined as the sum of the expected returns from this stock (i.e. future paid dividends) discounted to the present point in time. In the case of housing it is assumed that the national value of a house or flat can be determined as the sum of future discounted net returns in the form of rent. In the case of housing the dividend is that part of the net rent that is not re-invested into housing.

Meen (Sullivan, Gibb 2003) notes that analysts focus mainly on models drawn from neoclassical economics. The real "correct" price of a house/flat is determined as a quotient, where the nominator is the expected implicit income from rent and the denominator is the costs of owner-occupied housing. The relationship is formally expressed with the following equation:

where:

g(t) - the real purchasing price of a house/flat over time t,
R(t) - the real income from implicit rent over time t,
- the marginal household tax rate,
i(t) - market interest rate on loans,
- inflation rate,
- expected rate of real estate depreciation (costs of repair, maintenance and running of the house/flat),
g^e/g(t) - expected real capital valorisation (the real increase in real estate prices).

Housing is considered equivalent to any other kind of financial asset. No lags are included in the equation (e.g. inflation rate from the previous period ), in other words, it is assumed that the housing market is efficient and that prices respond to price-forming information. With only few exception there is not empirical model that estimates the amount of the cited equation directly, especially as the implicit rent R(t) cannot be measured directly; usually it is replaced by a function:

R(t) = h [RY(t), W(t), HH(t), H(t)]

where

RY - the real disposable income per person per household,
W - real household wealth,
HH - number of households,
H - housing stock size.

The models based on the equations presented above and derived from neo-classical economics are substantially more comprehensive than using just one single indicator (e.g. P/I), but again in the case of the United Kingdom they produced results that were not very reliable, especially over the course of the 1990s and especially for predictions of price developments in the short term (for the next several years).

The main reason why the neoclassical models based on fundamental economic factors, in particular from the level of real household income, lost strength in the 1990s was that they did not explain the short-term price fluctuations, i.e. short-term price deviations from the price equilibrium (price bubbles). Today three explanations are considered to account for this: first, it is caused by the speculative behaviour of investors in the housing market; second, too much state or local intervention and regulation in the field of housing construction; third, the existence of substantial transaction costs associated with the acquisition of housing.

In recent years talk has centred mostly on the influence of the speculations of investors, the first of the above-mentioned reasons why the rise or fall in prices cannot be predicted in the short term purely on the basis of fundamental economic factors. A more detailed analysis of the efficiency of the housing market has shown that residential real-estate price developments are to a large degree a self-determining process. What prices will be like next years has been shown to depend, not only on fundamental economic factors, but also on what prices were like last year, the years before, and even three years earlier. If housing prices can be predicted on the basis of past price developments, then it is possible with just some knowledge of price changes to attain abnormally high profits, afflicting the housing market with considerable inefficiency - among other things, it can no longer satisfy a basic precondition of neo-liberal economic models.

If the behaviour of buyers in the housing market is motivated mainly by an optimistic outlook, which means a quick and stable price increase in the future, then the level the prices is currently at is logically unsustainable in the long term. Prices cannot rise interminably, and the moment investors get the impression that price growth has stopped, and that prices could even begin to fall again, the bubble bursts. Price bubbles are to a considerable degree the result of human psychology - of expectations about future price developments, individual theories about potential risks from a fall in prices, concerns about being ousted from the market in the future, and so on. Some researchers (Case, Shiller 2003), therefore, believe that it is important to study how people think when they are making fundamental economic decisions such as purchasing a house or flat.

Current housing research has succeeded in incorporating even short-term price fluctuations into its models and has thus managed to make its predictions somewhat more precise. In this case the simulation method is the error correction model, which was used, for example, by Meen (2002), Hort (1998), Pagés and Maza (2003) and many others in their analyses of prices. The model is divided into two sub-models, two equations: a sub-model dealing with long-term developments of "equilibrium" housing prices, predicted in a manner similar to that used in classic neo-liberal models, with the aid of fundamental economic factors; and a sub-model dealing with short-term housing price developments predicted on the basis of housing price changes in recent years (reflecting the auto-correlation of prices in the short term) and with the aid of a correction coefficient.