Национальный инвестиционный совет

Вид материала

Подобный материал:

1 ... 34 35 36 37 38 39 40 41 ... 45

Source: Ministry of Foreign affairs and European Integration.

* - first six months of 2006.

mentioned average prices and often reach 6000 euros or more (per square meter). Such purchases are particularly important for Croatian real estate market for two reasons. Firstly, it pushes up location premium on real estates in the coastal region and secondly it helps create a high end of the tourism market and contributes promoting Croatia as an exclusive tourism destination, hence raising further the price of not only exclusive real estate, but also of all real estate due to spill over effects. One might say that purchase of exclusive Croatian residences by Russians has a strong demonstration effect; since the market is rather small these kinds of transactions are well known not only to players in real estate market, but also to general public; hence influencing the expectations of the future earnings on the market and consequently changing decisions on other real estate asking prices. Finally, these kinds of transactions also spur interest of Russian investors in tourism potentials of the Croatian coastal region.

Why is foreign demand for Croatian real estate important? Given the apparent rising interest of foreigners in the Croatian property market and, within it, the demonstration effect of Russian non-residents purchases, it is likely that foreign demand has influenced real estate prices developments in Croatia. Hence, foreign demand variable, among other variables of interest, will be included in our long run econometric model designed to detect long run determinants of real estate prices.

Literature review

House price volatility can have important effects on economic activity and financial stability. Housing assets represent a substantial part of households’ portfolios, and even moderate real estate price fluctuations give rise to significant capital gains or losses, which in turn can cause changes in consumption and savings behavior of the agents. However, the actual impact of these gains and losses on the consumption, savings and borrowing decisions of households is likely to depend on the history of the housing market and the structure of credit markets. A period of large changes will inevitably concentrate households’ minds on potential capital gains in the housing market. This in turn is likely to make the housing market look more like a financial asset market. The more households take accrued or potential short-term capital gains/losses into consideration, the more the market is likely to be subject to fluctuations due to changes in consumer confidence, expectations and trading strategies. In such a situation a phenomena of “irrational exuberance” that leads firstly to bubble and than to bust may also set in.

The above mentioned effects of real estate prices volatility have prompted economic literature to examine the factors driving house price dynamics in some detail. For one, house prices in real terms have shown a tendency to move in cycles with a number of booming years followed by a sequence of busts. Hence, real house price changes have been shown to exhibit a pronounced autoregressive pattern with positive short-term autocorrelation and negative long-term autocorrelations (see e.g., Hort, 1997; Case and Shiller, 1989; 1990). In the general debate, as well as in economic research, this has raised the issue of whether short term house prices have only been driven by fundamental demand and supply factors, or whether they result also from speculative behavior in the housing market.

Since it has been found that house prices and housing returns are predictable from their own past values, it is interesting to ask what other factors might have predictive power. This is also considered by Case and Shiller (1990), who find that the level of construction costs and the percentage change of the adult population are the only two extra factors that perform significantly in a forecasting equation (both with positive sign). The importance of demographics is closely related to the claim by Mankiw and Weil (1989) about the importance of demographic factors. Similar results are obtained for the predictability of excess returns on housing over interest bearing assets. Here income growth and a measure of mortgage costs are also significant. The studies quoted above throw some light on the “efficiency” of housing markets, but they are uninformative about what types of shocks appear to drive house prices and housing returns away from predicted values. These questions have been approached in various ways.

Chinloy (1992) estimates a factor model to explain the housing returns. He finds changes in inflationary expectations to be the main macro factor to have an impact on housing returns. Hendershott and Abraham (1993, 1994), using pooled cross-section data on US metropolitan areas, find changes in the following factors to have an impact on house price changes: construction costs, employment growth, income growth, and the real after tax interest rate.

Hendershott and Abraham (1994) also find evidence of cyclical behavior. They estimate a model specification that is similar to an error correction model, which includes lagged house price changes among the explanatory variables. They find relatively slow adjustment towards equilibrium combined with a significantly positive impact from lagged price changes. These results imply a cyclical adjustment path. Further some studies, notably Koskela et al. (1992) highlight the empirical importance of household indebtedness and borrowing constraints.

Abelson et al (2005) use cointegration and error correction model to prove that, in the long run, real house prices are determined significantly and positively by real disposable income and the consumer price index. They are also determined significantly and negatively by the unemployment rate, real mortgage rates, equity prices and the housing stock. They also report rather slow adjustment towards equilibrium.

Englund and Ioannides (1997) explore the dynamics of real estate prices in 15 OECD countries. The data reveal a remarkable degree of similarity across countries and suggest rich dynamics for the first-differenced real annual house prices, with a significant importance of autocorrelation. With regards to other determinants of real estate prices, the authors find that GDP growth rate and the rate of change in real rate of interest are very significant for predicting changes in real estate prices.

Hort (1998) estimates a cointegration model and a restricted error-correction model of real house price changes on Swedish panel data. In the long-run equation, movements in income, user costs, and construction costs were found to have a significant impact on real house prices. She also shows that real house price fluctuations are well explained by the development of fundamental demand conditions in this period.

Egert and Mihaljek (2007), in so far the only paper that provides empirical evidence on determinants of real estate prices in Central and Eastern Europe, give evidence on strong positive relationship between per capita GDP and housing credit on one hand and house prices on the other. A robust negative relationship between real interest rates and house prices was also detected. Demographic factors and labor market developments also play an important role in real estate price dynamics. They seem to affect house prices more strongly in Central and Eastern Europe than in OECD countries. The authors also underscore the importance of institutional development of housing markets and housing finance and quality effects of dwellings for real estate prices in CEE.

To wrap up, real estate prices are determined by housing supply and housing demand. Housing supply, measured by the housing stock, is fairly stable in the short term, since building new housing units takes time and housing construction per year is low in relation to the total housing stock. In the short term, therefore, house prices will generally fluctuate with changes in demand. The housing stock will adapt to demand over time, however. A long term model of real estate prices should therefore, along with demand side factors, contain explanatory factors for developments in the housing supply, such as construction and building site costs and prices for new dwellings. Due to specificities of Croatian real estate market, a long term model real estate prices should also encompass variable representing foreign demand.

Econometric analysis

After we analyzed the real estate market in Croatia and the involvement of Russian investors in it and gave the literature review of the models that estimate the movement of real estate prices, we turn to econometric analysis of real estate market in Croatia. For this purpose we use a Johansen cointegration and error correction model. In the vector autoregression we included variables from the demand and supply side of the real estate market. Moreover, because of attractiveness of Croatian coastal real estate market for foreign investors, described in section 2. a variable representing foreign demand is also included. The aim of the model is to estimate the long run elasticities of real estate prices in Croatia in real terms to foreign demand, domestic demand and supply side factors. Also the estimates of the short run adjustment coefficients will be presented.

We selected the following variables to be used in the cointegration model:

Price of real estate (p_estate_cro): average price of the real estate in Croatia in real terms;
Interest rate on housing loans (ir_housing): average interest rate on housing loans in real terms – represents the real cost of acquiring real estate for Croatian households and enterprises;
Net wage (net_wage_cro): average net wage for persons employed in incorporated sector in real terms – represents purchasing power disposable for acquiring real estate for households;
Number of employed person (no_employ): total number of persons employed in incorporated sector, crafts and free lance professions – represents the number of persons receiving income from labor and are therefore eligible for obtaining housing loans;
Volume of construction works (const_vol): total volume of construction works in Croatia – represents supply of new housing units;
Gross wage in construction (g_wage_const): gross wage of employees in incorporated construction sector in real terms – represents real cost of supplying new housing units for construction companies;
Foreign demand (for_demand): Number of formal requests of foreign citizens for purchasing real estate in Croatia – represents foreign demand for Croatian real estate^³⁷².

All variables are in monthly frequencies, ranging from July 1996 to June 2006. Moreover, variables are seasonally adjusted and in logarithms. Obviously, interest rate on housing loans, net wage, housing loans and number of employed persons are the variables representing domestic demand for real estate. Higher interest rate on housing loans cools off the demand for real estate since the cost of acquiring real estate goes up. Hence we expect the elasticity of real estate prices to interest rate to be negative. Increase of net wage and housing loans boost the demand for real estate either trough the increase of disposable income or trough borrowing new funds for financing house purchase. Hence we expect a positive elasticity of real estate prices to net wage and housing loans. If the number of persons employed goes up, we can expect higher demand for real estate, since more persons will be capable of acquiring real estate from their labor income, either by taking out credit or by using savings. Hence we expect a positive elasticity of real estate prices to number of persons employed.

Volume of construction works and gross wage in construction are variables representing the supply side factors of the real estate market in Croatia. If volume of construction works goes up, we can expect more new housing units on the market, which should in turn lower down the price per unit. Thus, we expect a negative elasticity of real estate to volume of construction works. On the contrary, if gross wage in construction goes up, we can expect increase in real estate price since construction enterprises will likely transfer this cost to the end buyer.

In addition to other models for real estate prices reviewed in section 3., we used the variable representing the foreign demand for Croatian real estate. The reasons for including this variable are quite obvious. Croatia is a small open economy, situated in the geographic proximity to highly developed European countries. It has a strong tourism industry, attractive Mediterranean coastal region and agreeable climate. Such a country will naturally be exposed to strong foreign demand for second homes whose effects could thereafter easily spill over to the rest of the real estate market, hence influencing the prices. Thus we expect the elasticity of real estate prices to foreign demand to be positive. Section 2. of this paper covers foreign demand developments in more detail.

In order to use Johansen cointegration framework (Johansen, 1988; 1991; Johansen and Juselius 1992), one must pretest the selected series for unit root. Here we will use Augmented Dickey-Fuller (ADF) test for unit roots. However, before conducting unit root test it is useful to analyze the graphs of the selected series in levels and first differences in order to detect which series exhibit I(0), I(1) or I(2) process. After observing Graph 1. it is quite obvious that all eight series exhibit strong trend behavior, implying that all series will be either I(1) or I(2).

Graph 1. Series in levels and first differences

Source: Croatian National Bank, Central Bureau of Statistics and Ministry of Foreign Affaires and European Integration.

As described in section 2, price of the real estate in real term has been on the rise, just as housing loans and average net wage (although latter two have somewhat steeper trend). Construction volume and average gross wage in construction has also been rising, but they witnessed a backdrop in 2000 and 2001. Number of employed person has been on the rise until 2000, and is stagnating thereafter. Foreign demand for real estate has also been on the rise, except in 1999

Table 2. ADF unit root test - levels

Name of the variable	Deterministic trend components	Lag	t-value (ADF)	Beta	Sigma	t-value (lag)	p-value (lag)	AIC
p_estate_cro	trend and constant	11	-2.342	0.578	0.0832	2.01	0.05	-4.849
p_estate_cro	constant	12	-1.712	0.777	0.082	-1.867	0.065	-4.86
housing_loan	trend and constant	12	-3.186	0.953	0.0132	-1.74	0.09	-8.519
housing_loan	constant	12	0.956	1.002	0.0136	-2.933	0.004	-8.46
ir_housing	trend and constant	12	0.527	1.034	0.159	-3.803	0.0003	-3.536
ir_housing	constant	12	1.069	1.066	0.160	-3.80	0.0003	-3.532
net_wage_cro	trend and constant	3	-2.916	0.879	0.0109	2.376	0.019	-8.978
net_wage_cro	constant	3	-3.155*	0.9608	0.0111	2.276	0.024	-8.955
no_employ	trend and constant	9	-2.171	0.826	0.085	1.955	0.053	-4.820
no_employ	constant	9	-1.638	0.9148	0.0856	1.66	0.099	-4.817
const_vol	trend and constant	11	-2.386	0.909	0.036	1.29	0.19	-6.484
const_vol	constant	9	-0.952	0.9744	0.0377	2.19	0.03	-6.458
g_wage_const	trend and constant	12	-3.809*	0.826	0.0186	2.02	0.046	-7.829
g_wage_const	constant	11	-1.246	0.974	0.0198	1.83	0.07	-7.728
for_demand	trend and constant	3	-2.792	0.9008	0.0534	-1.557	0.12	-5.803
for_demand	constant	3	-2.139	0.9568	0.0541	-2.306	0.023	-5.786

Source: authors´ calculation.

Note: ADF - Augmented Dickey-Fuller; optimal time lag selected using Akaike information criteria; all series are seasonally adjusted and in logarithms; ** null hypothesis about existence of unit root rejected at 1 percent level, * null hypothesis about existence of unit root rejected at 5 percent level.

when NATO operation in Kosovo warned off potential buyers. Interest rate for housing loans has declined continuously during the observed period. After inspecting graphs of series in first differences it is clear that all series, except housing loans, are mean-reverting, suggesting that housing loans series is I(2) and all other series are I(1).

Blog

Национальный инвестиционный совет