Foreign Investment and Residential Property Price Growth

Wokker, Chris; Swieringa, John Working Paper Foreign Investment and Residential Property Price Growth Treasury Working Paper, No. 2016-03 Provided in Cooperation with: The Treasury, The Australian Government Suggested Citation: Wokker, Chris; Swieringa, John (2016) : Foreign Investment and Residential Property Price Growth, Treasury Working Paper, No. 2016-03, ISBN 978-1-925504-16-3, The Australian Government, The Treasury, Canberra This Version is available at: https://hdl.handle.net/10419/210391 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/3.0/au/legalcode FOREIGN INVESTMENT AND RESIDENTIAL PROPERTY PRICE GROWTH Chris Wokker and John Swieringa 1 Treasury Working Paper 2 2016-03 Date created: December 2016 1 The authors work in the Macroeconomic Modelling and Policy Division of Macroeconomic Group at the Australian Treasury. Correspondence to: The Australian Treasury, Langton Crescent, Parkes ACT 2600, Australia. Email: [email protected]. 2 This paper has benefited from the assistance of Tanuja Doss, Sam Hill, Michael Kouparitsas, Yi Yong Cai, Angelia Grant, Linus Gustafsson, Luke Willard, Hamish McDonald, Nigel Ray, John Lonsdale, Roger Brake, Adam McKissack, Sina Grasmann, Nicole Merrilees, Derrick Calder and David Earl. We also thank Glenn Otto at the University of New South Wales and Emma Schultz and Markus Brueckner at the Australian National University. © Commonwealth of Australia 2016 ISBN 978-1-925504-16-3 This publication is available for your use under a Creative Commons BY Attribution 3.0 Australia licence, with the exception of the Commonwealth Coat of Arms, the Treasury logo, photographs, images, signatures and where otherwise stated. The full licence terms are available from http://creativecommons.org/licenses/by/3.0/au/legalcode. Use of Treasury material under a Creative Commons BY Attribution 3.0 Australia licence requires you to attribute the work (but not in any way that suggests that the Treasury endorses you or your use of the work). Treasury material used ‘as supplied’ Provided you have not modified or transformed Treasury material in any way including, for example, by changing the Treasury text; calculating percentage changes; graphing or charting data; or deriving new statistics from published Treasury statistics — then Treasury prefers the following attribution: Source: The Australian Government the Treasury Derivative material If you have modified or transformed Treasury material, or derived new material from those of the Treasury in any way, then Treasury prefers the following attribution: Based on The Australian Government the Treasury data Use of the Coat of Arms The terms under which the Coat of Arms can be used are set out on the It’s an Honour website (see www.itsanhonour.gov.au) Other uses Enquiries regarding this licence and any other use of this document are welcome at: Manager Media Unit The Treasury Langton Crescent Parkes ACT 2600 Email: [email protected] Foreign Investment and Residential Property Price Growth Chris Wokker and John Swieringa 2016-03 2 December 2016 ABSTRACT This study uses fixed effects panel regression techniques to estimate the impact of foreign demand for Australian residential real estate on property prices. All model specifications find a positive relationship between foreign investment approvals and price growth at the postcode level, but the majority of price growth experienced in recent times does not appear to be attributable to increased foreign demand. This is unsurprising given that in the short run the supply of residential property is relatively fixed so any increase in demand, whether domestic or foreign, should result in higher prices. Indeed, there have been many other significant domestic drivers of property prices over the period examined. The majority of foreign investment approvals are for new as opposed to established dwellings. This provides some indication that, in the longer-term, foreign demand is increasing property supply consistent with Australia’s foreign investment framework. JEL Classification Numbers: F21, R31 Keywords: Foreign Investment, Residential Real Estate, House Prices Chris Wokker Macroeconomic Modelling and Policy Division Macroeconomic Group The Treasury Langton Crescent Parkes ACT 2600 John Swieringa Macroeconomic Modelling and Policy Division Macroeconomic Group The Treasury Langton Crescent Parkes ACT 2600 1 1. INTRODUCTION Australia has run a current account deficit almost continuously since the Australian Bureau of Statistics began compiling balance of payments statistics in 1959. As such, foreign investment has been critical to the economic welfare of Australian households by financing investment beyond what would be possible from domestic saving alone. But the effects of foreign investment are often poorly understood, particularly with regards to residential real estate. This paper explores the relationship between foreign investment in Australian residential real estate and property prices. The number of foreign investment approvals has trended up in recent years, which has coincided with strong property price growth in many parts of Australia. While domestic buyers make up the vast majority of demand for property, it may be the case that, at the margin, foreign buyers are affecting property prices. This is because the stock of dwellings is relatively fixed in the short run so any increase in demand, whether from domestic or foreign sources, would be expected to result in higher prices, at least until increased prices have provided an incentive for the construction of additional supply. In the longer term, the high level of house prices in Australian capital cities, relative to those in other countries, likely reflects supply constraints. These constraints include state government land release and zoning policies, infrastructure provision and local government development approval processes. There is a large literature discussing these issues (see, for example, Hsieh, Norman and Orsmond, 2012). As outlined in Treasury (2014), Australia’s policy for foreign investment in residential real estate aims to increase Australia’s housing stock. As such, applications from non-residents to purchase new properties are usually approved without conditions, but non-residents are prohibited from purchasing established dwellings. 3 Temporary residents can apply to purchase one established property to use as a residence while they live in Australia. The majority of approvals have been granted for investment into new, as opposed to existing dwellings. This suggests that foreign demand is being channelled into increasing the property supply as intended. While some commentators have argued that foreign demand is pricing out first home buyers, it is not clear that this is the case. The number of foreign investment approvals granted for new properties is especially noteworthy given new properties make up a very small proportion of the total number of properties in Australia and because first home buyers tend to buy established properties (Gauder, Houssard and Orsmond, 2014). In recent years the level of foreign demand for Australian property has increased strongly. This has been driven largely by increasing applications from Chinese nationals, which rose from around 50 per cent of total foreign investment approvals in mid-2010 to around 70 per cent in early 2015. The increased importance of Chinese demand to Australian real estate increases Australia’s exposure to factors affecting the Chinese economy. Further, any change to the relative attractiveness of holding assets outside of China or ability to do so will likely affect foreign demand for Australian property, which may have domestic economic and financial implications. Over the period of this study, foreign investment in residential real estate has been concentrated in Melbourne and Sydney (Chart 1). But despite Melbourne receiving more foreign investment approvals than Sydney, price growth in Sydney has been much stronger than in Melbourne over the period. As such, it is difficult to directly attribute price growth in Sydney to foreign investors alone. Other factors, 3 Broadly speaking, a new dwelling is a dwelling built on residential land that has not been previously sold as a dwelling and has not been previously occupied. New dwellings do not include established residential real estate that has been refurbished or renovated, or a single dwelling that has been built to replace one or more demolished established dwellings. An established dwelling is a dwelling on residential land that is not a new dwelling. More detail is available at: www.firb.gov.au/real-estate. 2 such as the relatively low number of building approvals, commencements and completions in the late 2000s are potential longer term drivers of the recent price growth in Sydney. Chart 1 — Location of foreign investment approvals (1 July 2010 — 31 March 2015) Melbourne postcodes between 3000 and 3207; Sydney postcodes between 2000 and 2234. Source: Treasury; Authors’ calculations To estimate the sensitivity of property prices to changes in foreign demand we develop a fixed effects model of postcode level price growth using foreign investment approvals data from the Foreign Investment Division of the Treasury as the main explanatory variable. Under almost all model specifications there is a statistically significant and economically meaningful relationship between foreign investment approvals and property price growth, but the majority of price growth experienced in recent times does not appear to be attributable to increased foreign demand. Instead, the fact that property price growth has been strong over an extended period is likely to have been primarily driven by other factors such as impediments to supply, especially in some regions where natural and human-imposed constraints on supply are especially limiting. The increase in prices attributable to foreign investors is small when compared to the average quarterly increase in property prices of around $12,800 in Sydney and Melbourne during the study period. Across Sydney and Melbourne, the models which we consider to be the best specified indicate that, for a typical postcode, foreign demand increases prices by between $80 and $122 on average in each quarter. This is based on the average postcode in these two cities receiving around 0.6 more foreign investment approvals each quarter over time. Further, for each additional foreign investment approval beyond this typical increase of 0.6, median property prices are estimated to rise by between $145 and $222. Given that the typical increase in the number of foreign investment approvals from one quarter to the next in Sydney and Melbourne is only around 0.6, one additional foreign investment approval beyond this trend increase would be a relatively large spike in the number of approvals. As such, it can be seen that foreign demand has accounted for only a small proportion of the increase in property prices in recent years. While the results of this study show a consistent, but small positive relationship between foreign investment approvals and property price growth, there are some limitations. This includes the data limitations set out in Section 3, particularly around compliance and that the data reflects intentions to purchase and not actual purchases. The foreign investment data also may not pick up purchases by a citizen or permanent resident on behalf of family members overseas. Quantifying the effect of these limitations is difficult. It is also important to note that while the results suggest the impact across Australia and the capital cities is small, the impacts in certain areas or at particular times may be more intense. 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 Melbourne Sydney Rest of Australia Existing Existing Existing New New New 3 2. LITERATURE REVIEW Property prices are a function of demand and supply. Domestic demand for properties is affected by interest rates, financial regulation, incomes, demographics as well as the prices of alternative investments such as equities. Foreign demand may also contribute to property price growth. But the impact of changes in demand on prices depends on how supply evolves through time. This, in turn, depends upon both human imposed constraints, such as planning restrictions, and natural constraints imposed by challenging geography. In the longer run, the costs of new property as a close substitute for established property should set the marginal price in the market. If prices remain stubbornly high over time, impediments that increase the cost of bringing new properties to market need to be considered. In addition to these ‘fundamental’ drivers of property prices some have argued that less rational ‘psychological’ factors can explain property price movements. Interest rates affect property prices by affecting the size of the loan that can be serviced at a given level of income. When interest rates are lower, larger loans can be serviced and prices are bid up, because in the short term the quantity of housing is fixed. This has been witnessed in Australia (Tumbarello and Wang, 2010; Williams, 2009; and Otto, 2007). Successful inflation targeting by the Reserve Bank of Australia since the early 1990s has helped to lower nominal interest rates structurally, which has supported property price growth. The impact of interest rates on house prices is generally considered to be cyclical because interest rates typically move to help manage fluctuations in the economic cycle. While some have argued that greater net foreign capital inflows can lower interest rates by reducing bond yields (see Favikulis, Kohn, Ludvigso and Van Nieuwerburgh, 2012 for a discussion of this in the U.S. context), our study does not consider net capital inflows at the national level. As such, we do not consider the effect of capital flows on interest rates. The responsiveness of house prices to interest rates increased after financial market liberalisation in Australia in the 1980s and 1990s because quantity-based controls on lending were relaxed (Williams, 2009). Financial market deregulation has consistently been found to have supported property price growth (Andrews, 2010; Kohler and Van der Merwe, 2015; and Williams, 2009). Financial market deregulation can structurally lower interest rates by increasing the quantity of funds able to be lent, as well as by increasing competition for borrowers. Lower interest rates can relax credit constraints on households, therefore increasing the demand for dwellings. Additionally, financial market deregulation can loosen borrowing requirements, for instance by increasing the permitted loan-to-valuation ratio, and result in more people being eligible to borrow, again increasing the demand for dwellings. This can be thought of as an outward shift in the supply of credit (Favikulis et al., 2012). Our study does not explicitly address changes in financial market regulation as our data cover a relatively short period of time during which there were few substantial regulatory changes. Rising incomes enable more to be spent on properties and more people to enter the market for dwellings. This increase in the quantity of funds competing for the same number of properties should result in higher prices, at least in the short term before supply can respond to higher prices. This has been found to be the case internationally as well as in Australia (Andrews, 2010; and Bourassa and Henderschott, 1995). Similarly, higher terms of trade lead to increased national income and, as a result, are associated with house price growth (Tumbarello and Wang, 2010). In addition, changes in the rate of unemployment can impact on incomes and therefore house prices in a number of ways (Andrews, 2010; Hanewald and Sherris, 2013). Firstly, higher rates of unemployment create slack in the labour market and are associated with lower income growth, thus resulting in relatively fewer funds competing for the same quantity of dwellings. This puts less upward pressure on house prices. Secondly, higher rates of unemployment make servicing mortgages more difficult, resulting in pressure to sell, which puts downward pressure on prices. This applies to both owner occupiers and investors. While there is a positive correlation between property prices and female labour force participation in some markets, it is unclear whether greater female labour force participation leads to higher property 4 prices as a result of increased household income, or whether higher property prices necessitate higher household incomes (Johnson, 2014). 4 Population growth resulting from natural change and/or net overseas migration can increase the demand for dwellings, as can a reduction in the average number of people per household. 5 Such factors have been found to increase property prices (Andrews, 2009; Kohler and van der Merwe, 2015; Bourassa and Henderschott, 1995). But others have found a negative relationship between house prices and population growth in some Australian cities (Gitelman and Otto, 2012; and Otto, 2007). This may be because, to the extent that population growth is predictable, it should be factored into the planned supply of dwellings. While the natural rate of increase in population in Australia is relatively constant in the short and medium term, net overseas migration can be volatile and is the main contributor to unanticipated population growth. While migrants would likely rent when they arrive in Australia, net overseas migration would have to consistently surpass expectations for it to be a driver of sustained high property prices 6 . To the extent that recent migrants do purchase properties this will be partly captured in the foreign investment approval data for those on a temporary visa. Equity market returns may affect the demand for properties as some people view them as substitute investments. If investors use past performance as an indicator of an asset class’s future performance then a fall in equity returns would decrease expected future returns and thereby reduce the attractiveness of equities vis-a-vis property. This would imply a negative correlation between equity and property returns, as is found by Otto (2007) and Glindro, Subhanij, Szeto and Zhu (2010). But the effect of equity markets on house prices is complicated by wealth effects, whereby increases in equity values flow through to higher property prices (Fry, Martin and Voukelatos, 2010). Our study considers growth in the ASX200 total return index as a control variable under some regression specifications. Foreign demand for properties is not captured by the factors listed above, but likely contributes to overall demand for properties and therefore price changes. It appears reasonable that an increase in foreign demand for properties will result in price increases, at least in the short run before supply can adjust. Indeed, testing this proposition is at the heart of this study. Previous analysis of foreign investment in residential real estate in Australia at the national level has found that it has probably resulted in a greater stock of housing, but that the delays in the supply response may have at least temporarily resulted in higher prices (Gauder et al., 2014). At the capital city level, it has been suggested that foreign buyers have contributed substantially to price increases in Sydney and Melbourne, but may have resulted in lower prices for new properties in Brisbane and Perth (Guest and Rohde, forthcoming 2017). Our work differs from that of Guest and Rohde (forthcoming 2017) in a number of ways, including that we consider monthly and quarterly property data, mostly at the postcode level, as opposed to yearly and state level. We are unaware of other analysis of the 4 This study considers married women in the United States. It does not assess the relationship between house prices and female labour force participation in other contexts. 5 However, the average household size may be affected by an increase in the cost of housing. See Richards (2009) for a discussion on average household size in Australia. In addition to changes in population and the average number of people per household, the number of vacant properties and houses being replaced impacts on the overall demand for housing (Kohler and van der Merwe, 2015). Our study does not consider these factors as data disaggregated through time and by postcode does not appear to be available. However, the number of vacant properties and properties being replaced is small and relatively constant through time. 6 We note the strong correlation between net overseas migration and rental price growth. 5 relationship between foreign investment approvals and property prices at this level of disaggregation either internationally or in Australia. 7 Importantly, property prices are also affected by changes in the supply of dwellings. If land supply is unrestricted, then in the long run prices will not deviate much from construction costs, even with large increases in demand. When the additional supply of properties is less than the additional demand in a period, prices can be expected to rise. Higher property prices incentivise additional supply, but supply can be slow to respond to price signals. In general, this delayed response results from development processes and the time needed to construct properties (Ellis, Kulish and Wallace, 2012). With most of Australia’s population concentrated in major capital cities, the supply of freestanding houses is less responsive to price changes than the supply of higher density properties (Liu and Otto, 2015), probably owing to the limited availability of land as well as zoning constraints. Further, in postcodes which have both houses and units, house prices have grown faster than unit prices (Kulish, Richards and Gillitzer, 2012). There is also evidence that supply is less responsive to prices in postcodes closer to the central business district and that supply has become less responsive to price over time (Gitelman and Otto, 2012). While it may not be surprising that the supply of freestanding houses is price inelastic in Sydney, it could be expected that higher prices would lead to increased supply in regional areas where land is much less scarce. However, it appears that the supply of houses is relatively unresponsive to price in all parts of regional New South Wales (Liu and Otto, 2014). This suggests that state-wide impediments to supply, likely resulting from state and local government regulation, are resulting in higher house prices. These findings are consistent with the view that there is a relatively fixed quantity of land suitable for properties because of a combination of planning restrictions and geographic characteristics and that this puts upward pressure on property prices (Kulish et al., 2012; Saiz, 2014). In Australia, the pipeline of new properties is measured at three points: building approvals, commencements and completions. It is unclear at what point this additional supply impacts on property prices. This is because properties can be purchased but not lived in prior to being completed and buyers and sellers can be expected to be somewhat forward looking. However, evidence of previous housing cycles suggests a degree of myopia among market participants. We use completions instead of approvals or commencements as our measure of property supply. This is because a varying proportion of approvals are never commenced or completed, and there can be an indeterminate lead time between commencements and completions. While much of the literature has focused on demand and supply to understand property prices, some argue that prices are not exclusively a function of economic fundamentals. For instance, Shiller (2007) proposes a psychological theory whereby expectations about future price growth explain property prices. Varying degrees of ‘herd mentality’ have been found in Australian property markets, with Sydney being the most ‘excitable’ housing market (Valadkhani and Smyth, 2015). Additionally, Williams (2009) has found that when there is very strong property price growth in a quarter, this is followed by a higher than otherwise rate of price growth in the following quarter. 7 Internationally, this likely results from the absence of relevant data in jurisdictions which allow foreign investment in residential real estate, and the prohibition of foreign investment in residential real estate in many other jurisdictions. Domestically, data on foreign investment approvals for residential real estate are released on an annual basis, and is disaggregated at the state level only. 12 4. METHODOLOGY Chart 3 demonstrated a positive correlation between property price growth and the number of foreign investment approvals in a postcode. We now outline a framework to examine whether this relationship persists under a range of regression specifications. We employ fixed effects regression techniques to take advantage of repeated observations at the postcode level. This allows us to disregard known and unknown factors which are time-invariant and may contribute to changes in postcode level property prices, such as distance to the central business district or the quality of amenities (Allison, 2009). That is, there is no need for us to consider the characteristics of postcodes as is done by Abelson, Joyeux and Mahuteau (2013). Because of this, any change in our dependent variable can be attributed to factors which change through time, such as the number of foreign investment approvals. The main alternative to the fixed effects specification for panel data regressions is the random effects specification. 20 Random effects can be used when some omitted variables vary through time but are constant between postcodes and would allow us to estimate the effect of time invariant characteristics on property prices. But this would require measurement of postcode level characteristics which affect property prices. This would be extremely difficult to do accurately, especially because we consider over 2,000 postcodes. Because we are primarily interested in the relationship between property prices and factors which vary through time, namely foreign investment approvals, we do not pursue a random effects specification. 21 20 Another specification, between effects, is used when all omitted variables vary through time but are constant between postcodes. This is equivalent to taking an average of each variable through time for each postcode. This results in a loss of information regarding variation through time the estimates are more likely to be affected by factors not controlled for. For these reasons we do not consider the between effects specification. 21 Hausman tests indicate that for some regression specifications random effects could be used. On such occasions there is little difference between the results of fixed and random effects specifications. We only present the results of fixed effects regressions as we consider it reasonable that postcode specific intercepts are correlated with our foreign investment variable. 13 The fixed effects regressions estimated are of the form: (1) where: • is the percentage point change in price for postcode at time ; • is the intercept 22 ; • is the foreign investment approvals variable, which varies by postcode and through time; • are independent variables which vary through time but not by postcode for example, interest rates 23 , though some variables, such as income, vary by state; • and ρ are coefficients; and • is the error term which varies by postcode and through time. – Note that = ui + eit , where ui are unobserved predictors of that are postcode specific and time constant, and eit are unobserved predictors of that are specific to the postcode and point in time. We assume that errors are normally distributed with a mean of 0 and a variance of and that the error term is not correlated through space or time. These assumptions are formalised in equations 2, 3 and 4. (2) E( (3) E( (4) We begin by estimating a fixed effects regression with the level change in foreign investment approvals relative to the number of properties in the postcode and time fixed effects being the only independent variables. Lags and leads are then considered. As a check of robustness additional explanatory variables are added in successively. 24 With the exception of the foreign investment variable the control variables are not available at the postcode level at a monthly or quarterly frequency. As such, we do not consider 22 This is the average value of the fixed effects. 23 Time fixed effects are also included under some specifications. This allows factors which are constant across postcodes at specific points to be considered. 24 While we aim to estimate the impact that foreign investment approvals have on property prices, property price growth could also affect the number of foreign investment applications and therefore approvals. However, we consider that most foreign persons looking to purchase properties in Australia are flexible regarding property characteristics, so long as their budget constraints are not breached. For example, if property prices rise we consider that many foreign persons are likely to purchase smaller properties which remain within their budget constraint, as opposed to giving up on acquisitions entirely. This is because many foreign persons are motivated by a desire to facilitate potential future migration (CBRE, 2015) or diversify personal asset allocations outside of their home country, particularly in jurisdictions like Australia with strong property rights regimes. An instrumental variable approach could also address potential endogeneity problems but we are unable to identify a variable which affects the number of foreign investment approvals but not prices and has frequent data at the postcode level. This is a potential area of future research. 14 them to be particularly informative. Given this limitation, we consider these variables to be primarily a check on the robustness of the results. We do not consider that emphasis should be placed on the coefficients of these other control variables. After estimating regressions for all postcodes we restrict analysis to capital cities, then to Sydney and Melbourne. Because of the use of fixed effects regression techniques, this study does not measure postcode level determinants of property prices such as distance to the central business district or average property size. These postcode level characteristics are often correlated with one another. Instead, explanatory variables used in our study for robustness are national and state level and in level change or percentage change form, thus reducing the level of correlation between variables. As such, multicollinearity is less likely to be a problem in this study than in studies which utilise hedonic price models where postcode and property characteristics can be highly correlated. Time fixed effects are not employed in models which include control variables other than the foreign investment variable. This is because the time fixed effects are likely to capture the same factors as the other control variables, namely changes in income, the standard variable mortgage rate, dwelling completions and equity returns. 15 5. RESULTS Across Sydney and Melbourne, the models which we consider to be the best specified indicate that foreign demand typically increased prices by between $80 and $122 on average in each quarter. 25 This is based on the average postcode in these two cities receiving around 0.6 more foreign investment approvals each quarter through time. It is important to note that this is very small when compared with the average quarterly increase in Sydney and Melbourne property prices over the period studied of around $12,800. Table 4 records, for a range of models, the increase in median property prices in a postcode that occurs if the number of foreign investment approvals in a period increases by one more than usual. 26 This represents the impact of an additional foreign investment approval at the margin, as opposed to the effect of the average foreign investment approval. But once again, it is important to realise that the price increases indicated in Table 4 are small when compared to the average quarterly increase in property prices over the period of the study, which range from around $3,800 for Australia overall to around $12,800 for postcodes in Sydney and Melbourne. Note that these figures relate to foreign investment approvals at the margin. As such, multiplying these dollar figures by the total number of foreign investment approvals received in a postcode would be nonsensical and would not give the overall dollar impact of foreign demand on property prices. While this particular study focuses on a measure of foreign demand, there isn’t any inherent reason why an increase in demand for one additional property from a foreign investor should be considered as any different in its impact on prices from an increase in some forms of new domestic demand, for example a first home buyer moving out of the family home to purchase a property for the first time. Table 4: Estimated price increase attributable to one additional foreign investment approval beyond the trend increase (selected models). Model 1 (monthly, all postcodes) Model 9 (quarterly, all postcodes) Model 11 (quarterly, all capital cities) Model 13 (quarterly, Syd and Mel only) Model 14 (quarterly, Syd and Mel only, leads and lags) Model 15 (quarterly, Syd and Mel only, pre-inquiry) $117 $127 $195 $145 $222 $155 25 Models 13, 14, 15 and 16 are included in this range. Model 16 is the only model estimated for which the contemporaneous foreign investment variable is not statistically significant. As such, a dollar value is not calculated for Model 16. We note that Model 16 includes somewhat fewer observations than Model 15. Zero lies below the 95 per cent confidence interval of the contemporaneous foreign investment variable for all monthly and quarterly models, excluding Models 10 and 16. 26 These dollar values do not align with coefficients because the different regression specifications include different postcodes. For instance, Model 9 includes all postcodes while Model 13 excludes postcodes outside of Sydney and Melbourne. The average level of the foreign investment variable changes because the average number of properties in a postcode (the denominator of the foreign investment variable) varies depending on the range of postcodes included in the regression. Also, the monthly coefficient is comparably large because one additional approval is more substantial in a monthly context where the typical increase in approvals through time is around one third of that of the quarterly specification. 16 Interpreting the coefficient of the foreign investment variable The following example of the price effect of a marginal foreign investment approval is based on the results of Model 1. ( ) 27 The numerator of the foreign investment variable is the level change in the number of foreign investment approvals between periods. Most postcodes receive very few foreign investment approvals. Indeed, as noted earlier, the average postcode receives one foreign investment approval roughly every two months. The level change in the number of foreign investment approvals between months is smaller again, with an average of around 0.06. 28 In other words, for the postcodes included in Model 1, each month the average postcode receives 0.06 more foreign investment approvals than it did in the preceding month. 29 The denominator is the number of properties in a postcode, which is around 5,700 on average for the postcodes which are included in Model 1. The value of the numerator is small and the value of the denominator is large. On average the foreign investment variable has a value of around 0.00001, as noted in Table 7. Given that on average the trend increase in the level of foreign investment approvals is 0.06, a ‘shock’ of one additional foreign investment approval is large — over 15 times the usual increase in the number of foreign investment approvals from one period to the next. The method to estimate the dollar value of the additional foreign investment approval is outlined below. • Instead of the numerator being 0.06 it will now be 1.06. • The denominator will remain at 5,700 as the number of properties in the postcode remains unchanged. • The value of the variable increases from (0.06/5,700≈0.00001) to (1.06/5,700≈0.00019). The difference between these two values is around 0.00018. • Given the coefficient value of 144 and the level change in the value of the foreign investment variable of 0.00018, the percentage point contribution of the one additional foreign investment approval to property price growth in that postcode is 144 x 0.00018 ≈ 0.026 percentage points. • The average property price is around $461,000 across the whole sample. So the increase in prices attributable to the additional foreign investment approval is $461,000 x 0.026 percentage points ≈ $120. 30 27 This is the time fixed effects coefficient corresponding to the last month of the study. 28 This figure excludes observations for which price change data is not available because these observations are not included in regressions. 29 Note that the level change in the number of foreign investment approvals is negative in many instances. I.e., a postcode receives less foreign investment approvals in the current period than it did in the previous period. 30 This differs slightly (by $3) form the result in Table 4 because we round the numerator and denominator for ease of exposition in this example. 17 One standard deviation in the foreign investment variable (0.00014) is roughly equivalent to the increase in the foreign investment variable which occurs when one additional foreign investment approval is received in a postcode (0.00018). Graphically, the increase in monthly price growth which occurs when the number of foreign investment approvals increases more than usual is displayed in Chart 4. Chart 4: Property price increase attributable to a beyond trend increase in foreign investment approvals in a postcode From Chart 4 it can be seen that property prices in the average postcode in Model 1 typically increase by around $1,200 in a month (this is the blue section of the chart). 31 This $1,200 monthly increase includes the trend increase in the level of foreign investment approvals of 0.06. But if this average postcode receives one standard deviation more foreign investment approvals then prices will increase by an additional $92 (this is the red part of the chart). While the slope of the red area of the chart would be steeper under some regression specifications, the blue portion of the chart would be much higher because of the greater property price rises, for example in Sydney and Melbourne. As such, the impact of a one standard deviation increase in foreign investment approvals would still result in only a small increase in prices relative to the price rises that would have occurred otherwise. Since a one standard deviation increase in the number of foreign investment approvals beyond the trend level of foreign investment approvals is uncommon, and its effect on price growth is small, it can be seen that little of the property price growth witnessed in recent times can be attributed to the activity of foreign investors (noting the previous caveats around purchases for which Foreign Investment Review Board approval was not sought and that certain locations might be impacted more than others). 31 This value of a product of the average property price and the average quarterly percentage point price increase. $0 $300 $600 $900 $1,200 $1,500 $0 $300 $600 $900 $1,200 $1,500 0 One St.Dev. Two St.Dev. Usual monthly level of property price growth over the sample period, including trend growth of foreign investent Price effect of additional (beyond trend) foreign investment 18 Detailed monthly Considering prices and foreign investment on a monthly basis maximises the number of observations included in the regression. Under all monthly specifications the coefficient on the contemporaneous foreign investment variable is positive and zero is below of the 95 per cent confidence interval. Chart 3 demonstrated that there is a positive relationship between price growth and the number of foreign investment approvals in a postcode. 32 Model 1 estimates a regression with contemporaneous foreign investment approvals and time fixed effects as the only explanatory variables and finds that the coefficient on the foreign investment variable is statistically significant at the one per cent level. This is detailed in Table 5. 33 , 34 It may be the case that the relationship between property price growth and the number of foreign investment approvals changes once leading and lagging relationships are factored in. That is, depending on the time it takes to find a suitable property or understand the foreign investment regime requirements, foreign investors may ‘create’ demand in months slightly before or after the foreign investment approval is granted. Model 2 sees the addition of two lags and one lead of the foreign investment variable. These additional foreign investment variables are not statistically significant. Most postcodes receive no foreign investment approvals or very few foreign investment approvals in a period. The postcodes that receive most approvals are in capital cities, particularly Melbourne and Sydney. It is these cities where property price growth has been most pronounced. Areas with relatively few approvals, such as rural and regional areas, will have less variation in the foreign investment variable, which can make interpreting regression results that include such postcodes more problematic. For this reason, and because we are principally interested in the effects of approvals in areas where approvals have been concentrated, Models 3 and 4 employ the same explanatory variables as Models 1 and 2 but restrict data to capital cities. The coefficients on the contemporaneous foreign investment variables are slightly larger in Models 3 and 4 than in Models 1 and 2. Models 5 and 6 restrict analysis further and only include postcodes in Sydney and Melbourne. The regression results are little changed from those which include the postcodes from all capital cities. There is an increase in the level of foreign investment approvals at the end of the study period. This could be attributable to applications being brought forward following the release of the Standing Committee on Economics report into Foreign Investment in Residential Real Estate, which recommended the introduction of application fees and tighter enforcement. 35 This potential bring-forward could reflect foreigners seeking the option to purchase a property in Australia in the following 12 months without paying an application fee. So the level of foreign demand for properties immediately after the release of the report may have been lower than the number of approvals 32 It may be the case that higher numbers of foreign investment approvals in a postcode increase prices in that postcode, and that this price growth forces some domestic buyers to consider properties in adjacent postcodes. We do not consider this potential flow-on impact in our analysis. 33 Observations where postcode level monthly price growth is greater than ±20 per cent are considered to be outliers and are excluded. These outliers may result from a change in the quality of properties transacted in a postcode in a period. 34 The coefficients of the time fixed effects variables are not presented in Table 5 as this would require over 50 additional rows. F-tests indicate that time fixed effects are jointly significant for all monthly models. Full details regarding monthly time fixed effects are available from the authors on request. 35 Note that the actual application of fees and tighter enforcement occurs after the end of our sample regardless. 19 indicates. It could also be the case that this higher level of approvals more accurately reflects the true level of foreign investment activity and that prior to the release of the report a portion of foreign investors were failing to seek approvals for purchases. If the increase in approvals is because of applications being brought forward artificially, or increased compliance, this constitutes a structural break in the foreign investment series. Alternatively, the increase in approvals could be driven by a genuine increase in intentions to purchase properties ahead of the introduction of the fees which were recommended in the report (but which were not the policy at that time), or greater ‘push factors’ in foreign jurisdictions leading to an increase in foreign demand. 36 Because of the possibility of a structural break the results of Models 5 and 6 may miss-estimate the impact of foreign investment on price growth. To correct for this, Models 7 and 8 replicate Models 5 and 6 but exclude the months after the Standing Committee on Economics report was released. While it may be expected that the coefficients on the foreign investment variables would be larger than the previous specifications, they are little changed from Models 5 and 6. This may suggest that the increase in applications towards the end of the period was not the result of an increase in compliance, but more likely a genuine increase in the level of foreign demand, potentially spurred by applications being brought forward following the release of the House of Representatives report or greater push factors from overseas. These monthly regressions have a number of important limitations. Because the property price variable covers a three month period which rolls forward monthly, we cannot be sure whether price changes through time reflect movement in the month falling into the rolling three month period, or movement in the month falling out of the rolling three month period. The models presented thus far do not control for possible determinants of price growth apart from the change in foreign investment approvals and time fixed effects. Other controls, namely the standard variable mortgage rate, dwelling completions, equity market returns and incomes are considered. The results of regressions which incorporate these variables are presented at Appendix C. We consider the addition of these controls to be useful as a robustness check but we do not interpret the coefficients. 37 This is because the short time series of this study features relatively little variation in these control variables. Additionally, given that these variables are national or state level they are insufficiently disaggregated to be considered alongside the price or foreign investment variables, which are postcode level. The inclusion of such variables on a monthly or quarterly basis and at a more disaggregated level is a potential area of future research. We note that under all such specifications the contemporaneous foreign investment variable remains positive and statistically significant. As mentioned previously, we do not consider time fixed effects and other non-foreign investment control variables simultaneously because there is a high degree of overlap between these additional controls and time fixed effects. 36 Such evidence may include pressure on pegged currencies in foreign investment source countries. 37 As an additional robustness check we examine whether, when using an OLS specification for specific postcodes, a robust standard errors approach affects the statistical significance of the foreign investment variable. The robust standard error specification shows little effect. 20 Table 5: Regression output — monthly price growth, postcode level 38 Model 1 Model 2 Model 3 (capital city only) Model 4 (capital city only) Model 5 (Syd and Mel only) Model 6 (Syd and Mel only) Model 7 (Syd and Mel only, pre-inquir y) Model 8 (Syd and Mel only, pre-inquir y) (FIA3mtFIA3mt-1) /dwellings 144.14*** (3.10) 107.71** (2.14) 180.68*** (3.37) 159.29** (2.53) 129.52** (2.41) 152.86** (2.38) 138.00** (2.26) 137.46** (2.04) (FIA3mtFIA3mt-1) / no. properties [lag 1 month] 28.00 (0.50) 15.78 (0.24) 5.25 (0.08) 30.73 (0.45) (FIA3mtFIA3mt-1) / no. properties [lag 2 month] 41.97 (0.65) -21.25 (-0.30) -1.65 (-0.02) -10.01 (-0.13) (FIA3mtFIA3mt-1) / no. properties [lead 1 month] 63.10 (1.10) 18.21 (0.27) -24.11 (-0.36) -25.74 (-0.35) Constant 0.34* 0.22 0.26 -0.06 0.38 -0.16 0.39 -0.16 Observations 46,155 43,643 26,282 24,865 15,016 14,209 13,954 13,386 Within R2 0.0085 0.0088 0.0164 0.0166 0.0375 0.0358 0.0351 0.0366 Between R2 0.0008 0.0009 0.0074 0.0072 0.0819 0.1095 0.0745 0.0663 Overall R2 0.0079 0.0083 0.0158 0.0159 0.0370 0.0355 0.0348 0.0362 Notes: ***, ** and * indicate significance at the 1, 5 and 10 per cent level, respectively. t statistics in parentheses. 38 Observations where postcode level monthly price growth is greater than ±20 per cent are considered to be outliers and are excluded. 21 Detailed quarterly Given that the measure of property prices is a median price over a three month period, considering prices on a quarterly as opposed to a rolling three-monthly basis removes uncertainty regarding which period contributed to property price growth. 39 Model 9 is a quarterly regression with contemporaneous foreign investment and time fixed effects as the only variables. The foreign investment variable is positive and statistically significant. Quarterly results are detailed in Table 6. The monthly regressions found that lags and leads of the foreign investment variable did not help to explain price growth. To test whether this is the case in quarterly regressions also, Model 10 includes two lags and one lead of the quarterly foreign investment variable. As was the case with the monthly regressions, the results of Model 10 indicate that there is no clear relationship between lags or leads of the foreign investment variable and property price growth. This suggests that foreign investors generally create demand in the same quarter as an approval is received. 40 Models 11 through 16 consider successively more restricted geographic and temporal ranges. Limiting the postcodes considered to capital cities, then exclusively Melbourne and Sydney does not greatly change the coefficient on the contemporaneous foreign investment variable. There is relatively little variation in the coefficient of the contemporaneous foreign investment variable in the quarterly models. The coefficients range from a minimum of around 166 in Model 9, which is Australia-wide, to a maximum of around 280 in Model 12, which only includes capital cities. The coefficients for the models which only consider postcodes in Sydney and Melbourne (Models 13 through 16), are between 163 and 257. This indicates that the price impact, measured in percentage points, of a foreign investment approval beyond the trend increase in foreign investment is broadly consistent across geographic areas. The maximum and minimum coefficients correspond to an increase in postcode level prices of between $127 and $234 for each marginal foreign investment approval. As discussed earlier, one additional foreign investment approval in a quarter is a considerable spike in approvals. Overall, we consider Models 13 through 16 to be the best specified. The reason for this is twofold. Firstly, and as detailed earlier, the quarterly price data utilised in these models means that fluctuations in price through time can be attributed to a specific period, unlike the monthly price data. Secondly, the results which are specific to Sydney and Melbourne are of most interest because this is where price growth and foreign investment were concentrated over the period of this study. 39 Although there is some autocorrelation in price changes in around a third of postcodes observed at the monthly frequency, this is not unexpected, given the fairly consistent positive price growth observed over the sample period. There is little autocorrelation in the quarterly series. Inasmuch as the coefficients are roughly equivalent across the monthly and quarterly regressions, autocorrelation in the monthly series does not appear to be having a substantial impact on the results. 40 As noted earlier, foreign investment approvals are valid for 12 months, so it could have been the case that a considerable proportion of foreign investors purchased properties later in this 12 month window. But there is no evidence of this. Similarly, there is no evidence of foreign investors creating demand shortly before receiving foreign investment approval. 28 Shiller, R. J. (2007). Understanding Recent Trends in House Prices and Home Ownership . National Bureau of Economic Research Working Paper No. 13553. Treasury (2014). Submission to the House of Representatives Inquiry into Foreign Investment in Residential Real Estate. Tumbarello, P., & Wang, S. (2010). What Drives House Prices in Australia? A cross-country approach. IMF Working Papers, 1-24. Valadkhani, A., & Smyth, R. (2016). Self-exciting effects of house prices on unit prices in Australian capital cities. Urban Studies. Williams, D. (2009). House prices and financial liberalisation in Australia. University of Oxford Working Paper No. 432. Windsor, C., Finlay, R., & Jääskelä, J. (2013). Home Prices and Household Spending. Reserve Bank of Australia Research Discussion Paper 2013-04. 29 APPENDIX A Other control variables Contemporaneous values as well as a range of leads and lags are used under different regression specifications for all control variables. Standard variable mortgage rate (SVMR) This is a measure of the cost of borrowing to purchase dwellings. To avoid problems of stationarity we use the level change in the standard variable mortgage rate over rolling three month periods in our regression analysis. As can be seen in Chart 5, the SVMR has trended lower over the period of this study, broadly in line with movements in the cash rate. However, the regressions which employ lags will include the tail end of the tightening cycle which finished in mid-2010. The quarterly regressions with lags will cover a period which include three interest rate increases and correspond to a period of very high property price growth. Chart 5: Standard variable mortgage rate and the cash rate Source: RBA Income This is an index of the total remuneration of all employees in a state in a quarter. This is a function of the number of employees and the average level of remuneration in a state and is referred to as compensation of employees (COE). To avoid problems of stationarity we use the percentage change in COE by state in our regression analysis. As can be seen in Chart 6, COE increased in all states and territories over the study period, with the greatest increase occurring in Western Australia and the smallest increase occurring in Tasmania. 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 Jul-10 Sep-11 Nov-12 Jan-14 Mar-15 Per cent Per cent Standard variable mortgage rate Cash rate 30 Chart 6: Quarterly COE by state Source: ABS, Authors’ calculations Australian equity market We use a market capitalisation weighted index of the 200 largest stocks listed on the Australian Stock Exchange with cash dividends reinvested (ASX200 Total Return Index). To avoid problems of stationarity we use the percentage change in the ASX200 Total Return Index in the regression analysis. Chart 7 shows that the level of the ASX200 Total Return Index increased over the period of this study, though there were fluctuations over this time. Chart 7: ASX200 Total Return Index Source: Bloomberg Property supply This measures the total number of houses and units completed in each state in a period. To avoid problems of stationarity, we use the percentage change in property completions in the regression analysis. 90 100 110 120 130 140 150 90 100 110 120 130 140 150 Sep-10 Mar-12 Sep-13 Mar-15 Sep-2010=100 Sep-2010=100 NT NSW QLD ACT 90 100 110 120 130 140 150 90 100 110 120 130 140 150 Sep-10 Mar-12 Sep-13 Mar-15 Sep-2010=100 Sep-2010=100 WA SA VIC TAS 20 30 40 50 60 20 30 40 50 60 Jul-10 Sep-11 Nov-12 Jan-14 Mar-15 Points (thousands) Thousands Points (thousands) Thousands 31 Chart 8 shows that Victoria recorded more completions than any other state over the period, in part reflecting strong migration from other states and overseas as well as the relative availability of greenfield and brownfield sites in close proximity of the Melbourne central business district. New South Wales also experienced considerable growth in property completions over the period, but Tasmania saw a decline. The number of completions in other states and territories did not change substantially over the period of the study. Chart 8: Quarterly property completions (houses and units) Source: ABS 0 3 6 9 12 15 18 0 3 6 9 12 15 18 Sep-10 Mar-12 Sep-13 Mar-15 '000 Thousands '000 VIC NSW QLD WA 0 1 2 3 4 0 1 2 3 4 Sep-10 Mar-12 Sep-13 Mar-15 '000 '000 SA ACT TAS NT 32 APPENDIX B Table 7: Summary statistics of control variables — monthly Summary statistics for all postcodes with price change data Mean Median Std dev Percentile 25th 75th Percentage point price change 0.13 0.22 2.26 -0.20 0.67 (FIA3mtFIA3mt-1)/ no. properties 0.0000095 0 0.00014 0 0.00001 (FIA3mtFIA3mt-1)/properties [lag 1 month] 0.0000054 0 0.00015 0 0.00001 (FIA3mtFIA3mt-1)/ no. properties [lag 2 month] 0.00001 0 0.00016 0 0.00001 (FIA3mtFIA3mt-1)/ no. properties [lead 1 month] 0.0000086 0 0.000176 0 0.00001 Table 8: Summary statistics of control variables — quarterly Summary statistics for all postcodes with price change data Mean Median Std dev Percentile 25th 75th Percentage point price change 0.78 0.87 4.30 0.03 2.04 (FIAqtrtFIAqtrt-1)/ no. properties 0.000037 0 0.0003 0 0.000038 (FIAqtrtFIAqtrt-1)/properties [lag 1 qtr] 0.000033 0 0.0003 0 0.000038 (FIAqtrtFIAqtrt-1)/properties [lag 2 qtr] 0.000037 0 0.0003 0 0.000031 (FIAqtrtFIAqtrt-1)/properties [lead 1 qtr] 0.000023 0 0.0003 0 0.000042 33 APPENDIX C: Results of monthly regressions excluding time fixed effects but including other control variables Model 17 Model 18 Model 19 (capital city only) Model 20 (Syd and Mel only) Model 21 (Syd and Mel only, pre-inquiry) (FIA3mtFIA3mt-1)/ no. properties 166.39*** (3.59) 130.81*** (2.60) 198.56*** (3.15) 210.87*** (3.27) 192.78*** (2.84) (FIA3mtFIA3mt-1)/properties [lag 1 month] 39.94 (0.72) 39.05 (0.61) 21.91 (0.33) 49.79 (0.73) (FIA3mtFIA3mt-1)/ no. properties [lag 2 month] 37.96 (0.59) -29.26 (-0.41) -16.50 (-0.23) -26.17 (-0.35) (FIA3mtFIA3mt-1)/ no. properties [lead 1 month] 65.63 (1.15) 26.54 (0.40) 11.16 (0.17) 4.71 (0.06) SVMRtSVMRt-3 0.04 (0.29) -0.04 (-0.34) -0.14 (-0.89) -0.03 (-0.18) 0.06 (0.33) SVMRt-3SVMRt-6 -0.41*** (-3.17) -0.50*** (-3.65) -0.70*** (-4.28) -0.93*** (-4.69) -0.93*** (-4.58) %∆ (ASX200T/ASX200t-3) 0.03*** (6.95) 0.03*** (6.14) 0.03*** (5.03) 0.04*** (5.11) 0.04*** (4.45) %∆ (ASX200T-3/ASX200t-6) 0.02*** (3.26) 0.02*** (3.65) 0.02*** (3.32) 0.02** (2.23) 0.01** (2.00) Constant 0.09*** 0.07** 0.16*** 0.23*** 0.27*** Observations 46,155 43,643 24,865 14,209 13,386 Within R2 0.0023 0.0023 0.0038 0.0068 0.0065 Between R2 0.0008 0.0008 0.0298 0.0305 0.0386 Overall R2 0.0022 0.0022 0.0036 0.0065 0.0062 Notes: ***, ** and * indicate significance at the 1, 5 and 10 per cent level, respectively. t statistics in parentheses. 34 Results of quarterly regressions excluding time fixed effects but including other control variables Model 22 Model 23 Model 24 (capital cities only) Model 25 (Syd and Mel only) Model 26 (Syd and Mel only, pre-inquiry) (FIAqtrtFIAqtrt-1)/ no. properties 175.75** (2.42) 246.76** (2.22) 342.45*** (2.85) 341.20*** (2.83) 271.59** (2.05) (FIAqtrtFIAqtrt-1)/properties [lag 1 qtr] 145.10 (1.18) 114.21 (0.91) 154.06 (1.22) -16.00 (0.10) (FIAqtrtFIAqtrt-1)/properties [lag 2 qtr] 4.47 (0.03) 198.83 (1.49) 232.59* (1.70) 244.53* (1.65) (FIAqtrtFIAqtrt-1)/properties [lead 1 qtr] 71.40 (0.81) 86.08 (0.92) 115.59 (1.25) 98.21 (0.91) SVMRqtrtSVMRqtrt-1 1.59** (2.50) 2.63*** (3.55) 4.37*** (5.08) 4.30*** (4.05) 4.53*** (3.96) SVMRqtrtSVMRqtrt-1 [lag 1 qtr] -0.02 (-0.04) 2.92*** (3.87) 2.18** (2.49) 5.12*** (5.00) 5.34*** (4.87) SVMRqtrtSVMRqtrt-1 [lag 2 qtr] -1.85*** (-3.24) -1.64** (-2.37) -2.74*** (-3.38) -1.88* (-1.93) -1.89* (-1.93) %∆ (COEqtrt/COEqtrt-1) *100-100 0.12*** (3.15) 0.13*** (2.86) 0.18*** (3.38) 0.30*** (3.63) 0.30*** (3.55) %∆ (COEqtrt/COEqtrt-1) *100-100 [lag 1 qtr] -0.04 (-0.95) 0.07 (1.43) 0.04 (0.66) 0.11 (1.17) 0.11 (1.08) %∆ (COEqtrt/COEqtrt-1) *100-100 [lag 2 qtr] -0.05 (-1.50) 0.05 (1.12) 0.01 (0.29) -0.10 (-1.19) -0.09 (-1.13) %∆ (Compl.qtrTCompl.qtrt-1)*10 0-100 0.01 (1.05) 0.00 (0.74) -0.01 (-0.95) 0.00 (-0.28) -0.01 (-0.44) %∆ (Compl.qtrTCompl.qtrt-1)*10 0-100 [lag 1 qtr] 0.01*** (2.64) 0.00 (0.42) 0.00 (-0.66) 0.01 (0.46) 0.01 (0.44) %∆ (Compl.qtrTCompl.qtrt-1)*10 0-100 [lag 2 qtr] 0.01*** (2.57) 0.01** (2.20) 0.00 (0.40) 0.01 (0.88) 0.02 (0.99) %∆ (ASX200qtrt/ASX200qtrt-1) 0.13*** (6.48) 0.19*** (8.04) 0.21*** (7.70) 0.29*** (8.50) 0.30*** (8.40) %∆ (ASX200qtrt/ASX200qtrt-1) [lag 1 qtr] 0.04** (2.12) 0.07*** (2.97) 0.05 (1.64) 0.10*** (2.86) 0.10*** (2.79) %∆ (ASX200qtrt/ASX200qtrt-1) [lag 2 qtr] 0.00 (0.24) 0.02 (1.01) 0.05** (2.43) 0.05** (1.99) 0.06** (2.09) Constant 0.35* 0.47** 0.86*** 1.15*** 1.21*** Observations 14,315 12,021 6,978 4,066 3,784 Within R2 0.0120 0.0155 0.0255 0.0480 0.0504 Between R2 0.0091 0.0092 0.0384 0.1000 0.0518 Overall R2 0.0115 0.0149 0.0254 0.0494 0.0517 Notes: ***, ** and * indicate significance at the 1, 5 and 10 per cent level, respectively. t statistics in parentheses. 35 COPYRIGHT AND DISCLAIMER NOTES Property price data used in this publication are sourced from CoreLogic. In compiling this data, RP Data Pty Ltd trading as CoreLogic has relied upon information supplied by a number of external sources. CoreLogic does not warrant its accuracy or completeness and to the full extent allowed by law excludes liability in contract, tort or otherwise, for any loss or damage sustained by subscribers, or by any other person or body corporate arising from or in connection with the supply or use of the whole or any part of the information in this publication through any cause whatsoever and limits any liability it may have to the amount paid to CoreLogic for the supply of such information.