2. Monetary Policy and Real Estate Bubbles
2.4. Data
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
25
In the empirical part, VAR and VEC models are applied to analyze dynamics between the bubble, Euribor and the lending for house purchase-to-GDP ratio.
2.4. Data
For Ireland, Portugal and Spain, the analysis covers the time period from the inception of the single monetary policy in the Eurozone in 1999 to the third quarter of 2012. As Greece entered the Eurozone two years later, the analysis starts in 2001.
As for the house price, price indices on residential property from the property price database of the ECB and the Bank for International Settlements (BIS) are used. In the case of Ireland, there is no complete time series on house prices for the entire period available. In order to cover the full time period, two overlapping time series are consolidated into a single item. The rent index is sourced from the ECB and available for the entire period. For the calculation of the WACC, due to data availability, data from two separate data sets is used. Historical quarterly data on retail interest rates is sourced from Eurostat and covers the time period from 1999 to 2003. The second data set is sourced from the Monetary Financial Institute (MFI) database from the ECB and covers the period from 2003 to 2012. In order to allow the analysis of the full time period, both data sets are consolidated. As for the cost of debt, the average interest rate for housing loans is used. Regarding the cost of equity, the average interest rate on deposits of up to one-year maturity is used. For Ireland, the average rate for overnight deposits is used as a proxy for the cost of equity as the previously mentioned interest is not available for this country. The data on the Euribor and the lending for house purchase-to-GDP is also sourced from Eurostat and available for the entire period.
‧
definition in (4). Ireland, Portugal and Spain experienced an increase in the bubble at the beginning of the 2000s, followed by a decrease up to 2005 in Portugal and a returning upward trend in Ireland and Spain in 2003. After a sharp drop in 2001, the bubble in Greece developed steadily before increasing at the end of 2005.Figure 2.1: Residential Property Bubbles with LTV Ratio of 70%
Figure 2.2 shows the simulation of the bubble in the four countries under five different scenarios assuming LTV ratios of 0.5, 0.6, 0.7, 0.8, and 0.9. Two interesting observations can be made from this simulation. First, it can be observed that the higher the LTV ratio the bigger the bubble. This is because the interest rate on equity is usually smaller than the interest rate on debt, implying that an increase in the loan-to-value ratio gives more weight to cost of debt, resulting in a higher WACC.
Reverting back to formula (3) the fundamental value of residential real estate decreases when the WACC increases. Hence the bubble increases with an increase in the WACC.
9 The starting value of the rent has been adjusted so that fundamental value equals the market value in the base period. The base period is set according to the point when the respective country entered the Eurozone. Therefore, the base period is 2001Q1 in the case of Greece and 1999Q1 for the other countries. This adjustment does not imply that the fundamental value is equal to the fundamental value in the base period; it rather defines the respective year as the reference point. What matters in the subsequent analysis is the movement or the trend of the bubble.
-100
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Greece Ireland Portugal Spain
‧
Figure 2.2: Residential Property Bubble Simulation under Varying LTV Ratios
-60 -40 -20 0 20
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
0.5 Greece
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 0.5 Ireland
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
0.5 Portugal
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 0.5 Spain 0.6 Spain 0.7 Spain 0.8 Spain 0.9 Spain
‧
movement or trend of the bubble rather than its absolute value.Figure 2.3 compares the minimum bid rate for main refinancing operations set by the ECB with the 3-month Euribor money market interest rate. It is obvious that the money market rate is closely related to the key interest rate of the ECB. In the subsequent analysis the 3-month Euribor is used as proxy for the interest rate on the main refinancing operations of the ECB.
Figure 2.3: 3-Month Euribor and Key Policy Rate of the ECB
Figure 2.4 shows the ratio of lending volume for house purchase-to-quarterly GDP in the Eurozone. Except for a drop at the end of 2000, the ratio increased strongly in the period from 1999 to 2007. From the year 2007 onward, the ratio increased much slower.
Figure 2.4: Lending for House Purchase-to-Quarterly GDP in the Eurozone
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
3-month Euribor
ECB rate on main refinancing operations
100
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
‧
Dickey and Fuller (1979) (ADF) test is applied to determine the order of integration of the variables. A variable is said to be integrated of order n when it achieves stationarity after taking its n-th difference. As shown in Table 2.5, the ADF test statistic indicates at a significance level of 10% that all of the variables in its level contain a unit root.Table 2.5: ADF Unit Root Tests
Level Difference 2001Q1 to 2012Q3 and 1999Q1 to 2012Q3 respectively. The number of lags included in the ADF test is decided by the automatic lag length selection criteria based on SIC with maximum lag length of 10. c indicates that a constant term and ti indicates that a constant term as well as a linear time trend have been included in the model.
The test further indicates that the first difference of all variables is stationary at the same level of significance. Thus it can be concluded that all of the variables are integrated of order one, or I(1).