2.1 The Housing Market
2.1.4 Danmarks Nationalbank’s Housing Price Model
The housing price model applied in this thesis was developed in 2011 and is an extension of Danmarks Nationalbank’s housing price model, MONA, from 2003. MONA is a macroeconomic model that seeks to explain the fundamentals and relationships between several factors of demand, production, employment and wages and general price levels (Danmarks Nationalbank, 2003).
Compared to the MONA model from 2003, the applied model from 2011 is better at accounting for shocks in the economy, such as the financial crisis in 2008. When the model from 2003 was created, the housing market in Denmark had only experienced shocks of a less severe degree than what it faced in subsequent years. Hence, the 2003 model was not created in a way that made it able to capture major upheavals. Conversely, when creating the MONA model in 2011, the impact of global financial shocks to the economy was better implemented in the model formation. Additionally, the introduction of interest-only loans in October 2003 impacted the housing market demand later on, which is also taken into consideration in the model from 2011.
This specific housing model was chosen as it was developed for the Danish housing market, and hence would be the most appropriate and applicable out of the previously discussed models. One of the shortcomings of the model is that it only captures some of the extraordinary fluctuations in the housing prices in the period between 2004 and 2009. The model also has difficulties explaining the sudden “building boom” in the middle of 2000 (Dam et al., 2011). One last shortcoming is that the model does not include regional difference. However, the latter is disregarded in this thesis, thus, it does not impact the applicability of the model.
32 The housing price relation for single-family houses developed by Danmarks Nationalbank in 2011 is an error-correction model based on quarterly data. The expression “error-correction” originates from the fact that last-periods deviations, errors, in the long-run equilibrium impact the short-run dynamics (Stock and Watson, 2012, p. 693).
The model takes into consideration both short-term effects on housing prices due to shocks in demand, and long-term impacts of changes in supply. According to Dam et al. (2003) and the MONA model, housing prices are defined by interest rates, disposable income and the housing stock (Dam et al., 2003, p. 41).
The first two are reflected in the demand relation. The latter, housing stock, express the supply side and the long-term effect on housing prices. Housing stock also mirrors the investment in construction of houses, as the current housing stock is a result of the housing prices relative to the costs of construction. The relationship housing prices divided by costs are the primary influencer of the investment in housing. This relationship is also called Tobin's Q. The higher this ratio is, the more advantageous it is to build houses (Plovsing and Olsen, 2012, p. 69). An estimation for housing investment is not included in light of the thesis’ limitations. However, housing investment is reflected in the long-run relation between real income and housing stock in the housing price equation.
33
(Dam et al., 2011, p. 70).
The long-term market demand depends on the disposable income, user cost, and first-year payment, in addition to the real housing price of single-family houses. The relation is presented in the following equation.
𝐾𝐷= 𝐹(𝑌, 𝑢, 𝑦, 𝑝ℎ)
where
𝐾𝐷 reflects the demand 𝑌 is the disposable income
𝑢 is user cost, specified as 𝑢 = [(1 − 𝑡)𝑟30− 𝜋] + 𝑠 + 𝑑 𝑟30 is the 30-year bond yield
𝜋 captures inflation
𝑠 is total housing taxes as a ratio of the entire housing stock calculated at market price 𝑑 is repairs and depreciation, equal to 0 in the model
𝑦 is first-year payment expressed by 𝑦 = (1 − 𝑡)𝑟𝑚𝑖𝑛+ 𝑠 + 𝑎𝑓𝑑𝑟𝑎𝑔 𝑟𝑚𝑖𝑛 is the lowest possible bond yield
𝑎𝑓𝑑𝑟𝑎𝑔 is the lowest possible repayments on a fully leveraged home purchase 𝑝ℎ is the real housing prices
First-year payment expresses the lowest amount that can be paid on a fully leveraged home purchase. Hence, in accordance with the equation, changes in interest rates and mortgage instalments will impact demand. This variable was not included in the old MONA model from 2003 as the introduction of interest-only loans and importance of loans with adjustable rates originated after the finalisation of the model. The inclusion of this variable may be interpreted as Danmarks Nationalbank’s concern regarding these new initiatives introduced by the government. The first-year payment is adjusted for the introduction of short-term interest rates in the 4th quarter of 1999. The short-term interest rate is a weighted average of the rate of mortgage bonds with one- and two-years maturity. Prior to the 4th quarter of 1999, the interest rate applied in the equation for first-year payment is the rate of a 30-year mortgage bond.
34
(Dam et al., 2011, p. 72).
The long-term housing price relation is formed by rewriting the equation for demand. By the use of logarithms, the housing price relation is explained by the linear equation presented below.
𝑙𝑜𝑔 𝑝ℎ= 𝑎1𝑢 + 𝑎2𝑦 + 𝑎3𝜋ℎ+ 𝑎4log (𝑌/𝐾) + 𝑎5
In line with economic theory, the variables 𝑎1 and 𝑎2 are expected to be negative, while 𝑎3 and 𝑎4 are expected to impact housing prices positively. To capture the short-term effect on market demand, the model was extended to include the lagged change in real housing price and the total change in interest rate after tax and property tax. The change in interest rate after tax and property tax was only relevant until the end of the 1990s. This is taken into account by including a dummy variable.
Danmarks Nationalbank’s housing price model is not unique in the sense that it does not include numerous new variables or has a significantly higher explanation ratio compared to the other considered models. In fact, there are several similarities between the beforementioned models and the applied model. One of the similarities between ADAM and the applied model is that most of the data are estimated on national accounts data, and not separated into regions (Statistics Denmark, 2013). The main difference between the applied model and the housing price model in ADAM is the fact that the applied model is based on quarterly data, while ADAM is based on annually. This makes MONA a better model when it comes to short-term dynamics. Additionally, the ADAM model is very comprehensive and detailed, which would make it more difficult to manipulate and hence achieve the objective of this thesis.
When comparing Hviid’s model to the applied one, a similarity between them was the inclusion of minimum first-year payment as an explanatory variable. However, the inclusion of this variable might be reasoned by the fact that Hviid's model is formed on the basis of Danmarks Nationalbank.
Nevertheless, for the purpose of this thesis, the housing price model from 2011 was the best fit as it was developed for Danish housing market and was easily adjustable.
35