Abstract
This Selected Issues paper analyzes the housing prices in Australia. Housing prices in Australia have increased strongly over the past two decades, including by comparison internationally. Thus housing prices are often argued to be overvalued. Many counter-arguments have been put forward for why such measures are flawed. This paper argues that housing prices are moderately stronger than consistent with current economic fundamentals, but less than a comparison to historical or international averages would suggest. International comparisons of price-to-income ratios suggest that Australia is broadly in line with comparator countries, although significant data comparability issues make inference difficult.
Are Australia’s House Prices Overvalued?
House prices in Australia have increased strongly over the past two decades, including by comparison internationally. Thus house prices are often argued to be overvalued. Many counter-arguments have been put forward for why such measures are flawed. We argue that house prices are moderately stronger than consistent with current economic fundamentals, but less than a comparison to historical or international averages would suggest.
Argument: House prices have risen faster in Australia than in most other countries, suggesting, ceteris paribus, overvaluation
1. Australia’s real house prices have risen significantly faster than OECD average over the past two decades. After growing broadly in line with real GDP per capita from 1960-90, Australian real house price inflation picked up sharply in the mid-1990s, exceeding income growth by a wide margin (Figure 1, panel 1). As a result, the house price-to-income ratio rose sharply, despite the terms-of-trade boom that boosted Australian incomes over the past decade, also exceeding the average increase among OECD countries (Figure 1, panel 2). In view of the strong price increases, many common measures of housing valuation based on deviation from long run historical trends, such as price-to-rent and price-to-income ratios, suggest overvaluation of about 20-25 percent (see Figure 1 and OECD, 2015; IMF Housing Watch, 2014).
Rapid House Price Growth
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: RBA; APRA; ABS; OECD; Haver Analytics; and IMF Staff calculations.2. Household debt-to-income ratios have also risen. The household debt-to-income ratio tripled from 47 percent in 1990 to a historic high of 154 percent in 2014 (though high internationally, it is broadly in line with comparators such as the UK, Canada, and New Zealand). The household debt-to-income is a key variable from a financial stability and macroeconomic risk perspective as this reflects the risks borne by households and the possible amplification of house price declines to the macro economy (see paragraph 24 for further discussion of these effects).
Counter argument 1: House prices are in line on an absolute basis
3. Using differences from averages across countries of house prices changes to derive estimates of over- or under-valuation is problematic. One reason is that it assumes that the starting period was an equilibrium. It could be, for example, that house prices were particularly low in Australia at the starting point, and thus a faster increase represents “catch-up” not “overshooting”. One way around this problem is to compare on the basis of actual house prices rather than changes in house prices.
Median House Price
(In thousands of Australian dollars)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Source: RP Data-Rismark.4. House price-to-income ratios have been rising for all measures. Nationally, house prices correspond to between 4 to 6 times income, depending on which housing type is used (see chart). However, housing markets vary significantly not only across borders but also within countries.1 Prices in urban (capital cities) and coastal areas tend to be higher than the rest of the country, but the rise in the price-to-income ratio has been broadly consistent across the country. Jääskelä and Windsor (2011) argue that housing is a superior good as households have been prepared to spend proportionally more on housing as their incomes increased. In this context, house prices may rise faster than incomes, and result in a rising price-to-income ratio over time. As a result, affordability has deteriorated for some segments of the population and the proportion of first-time home buyers is shrinking (see Senate Economic References Committee, 2015).
Price-to-Income Ratios
(Various measures)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: RBA; ABS; RP Data-Rismark; REIA; CBA; and HIA.5. International comparison of absolute price-to-income ratios. Comparisons of price-to-income ratios are difficult owing to different national definitions of housing coverage and household disposable income. Fox and Finlay (2012) show that comparing equivalently defined price-to-income ratios across countries, Australia’s experience appears to be broadly in line with those of other advanced economies. Using OECD data shows a similar picture, where the Australian price-to-income ratio has risen above OECD average, but broadly comparable to comparator countries.
House Price-to-Income Ratio
(1990 = 100)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: OECD database; and IMF staff calculations.Bottom line: Price-to-income ratios have risen across all measures in Australia and are now near historic highs. However, international comparisons of price-to-income ratios suggest that Australia is broadly in line with comparator countries, although significant data comparability issues make inference difficult.
Counter argument 2: The equilibrium level of house prices has also risen sharply
6. The equilibrium levels of house price and household debt have risen owing to financial liberalization and disinflation. Another reason why comparing changes in house prices across countries is problematic, even if they were in equilibrium at the start, is that the equilibrium may have changed. Over the past two decades, financial liberalization, lower inflation and lower nominal (and real) interest rates have facilitated easier access to credit and increased the serviceability of higher levels of debt, leading to higher levels of indebtedness and higher house prices relative to incomes. As nominal (and real) interest rates have declined over a sustained period, household debt as a share of disposable income has increased (see Figure 3, panel 1). Housing demand was also boosted through lower interest margins of mortgage banks, and increased finance availability. For instance, Ellis (2006, 2013) argues that financial deregulation led to greater mortgage market competition and product innovation. OECD (2011) finds that a lower down payment requirement is associated with higher homeownership among previously credit constrained households. Moreover, OECD (2011) finds that 30 percent of the house price increases in OECD countries can be attributed to financial deregulation.
Fundamental Measures of House Valuation
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: OECD; IFS; Igan and Loungani (2012); and IMF staff calculations.Macroeconomic and Supply Factors in House Valuation
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: RBA; ABS; Haver Analytics; RBNZ; Eurostat; Federal Reserve Board; UN Demographic Yearbook 2011; and IMF Staff calculations.7. House prices and economic fundamentals. A more analytical way of looking at the equilibrium level of house prices is to model and estimate the main driving sources of house prices in a period where the transition was largely complete (e.g. 2000-14). This approach can include both fundamental economic demand and supply factors, and then calculate the gap between the actual house prices and their predicted values from the economic model. For this purpose, a broad range of models are used: time series approaches using economic fundamentals as explanatory variables, user cost of housing and statistical filters. It should be recognized, however, that analyzing equilibrium levels of house prices using economic models is still difficult and inherently imprecise and thus results should be interpreted with a large degree of caution.
8. Econometric approaches to assess the level of house prices. Two different time-series econometric approaches are used to analyze house prices in relation to economic fundamentals.
Following the approach in Caldera Sanchez and Johansson (2011), house prices are modeled with income, dwelling stock, and credit as explanatory variables, representing both demand and supply factors in driving house prices (Model A - see Appendix 1A for details). Using quarterly data from 2000-2014, the results suggest that a moderate house price gap around 5-10 percent has opened up in recent years (see chart).
Using time-series data, changes in equilibrium real house price changes are modeled as a function of real disposable income, working-age population, equity prices, and the level of short- and long-term interest rates, aiming to capture major demand factors (Model B: see Igan and Loungani, 2012, and Appendix 1B for further details). Using this approach with data going back to 1970 results in a relatively poor fit and suggests that house prices are around 24 percent stronger than economic fundamentals would support, depending on econometric specification (see Table 1). However, as discussed above, housing and financial markets have changed significantly over this period owing to structural reform in the 1980-90s, making property market developments in the 1970-80s less informative. A more appropriate time period for estimation, arguably, is using data from 2000 onwards, a period where the transition to lower interest rates and financial liberalization has been completed. Using this shorter time frame yields suggest that house prices are around 17 percent stronger than consistent with economic fundamentals (see text chart). 2
User cost approach. Another method to assess house prices is to apply the concept of user cost of housing. The user cost approach compares the relative costs of owning a home versus renting it by adding up the discounted costs of each alternative over the period for which a house is expected to be owned (see Appendix 2 for details). User costs are affected by a range of factors, including the direct cost of owning a home such as the real interest rate (after tax deductions), operating and maintenance costs, property taxes. The cost of owning a home is also affected by expectations of future house prices, but also significantly by the tax system treatment of housing. In Australia, owner-occupied housing is exempt from the capital gains tax and the tax treatment of investment in rental property, and particularly from highly geared investment, also imply significant incentives for housing investment (see SIP on Tax Reform). Following the calibration in Fox and Tulip (2014) for the Australian economy, the user cost approach results in a housing price overvaluation ranging from 0-19 percent, depending on expectations of future house price increase.3 The results of the user cost model are also sensitive to changes in the assumption of real mortgage rate variability (see Figure 3).4
Trend approach. One can also take an agnostic view on the trend level of house prices, but just assess where they are cyclically (i.e. without taking a view on whether the trend value is too high or low) based on HP-filter type cycles.5 Using this approach and applying the HP-filter to real house prices for the period 2000-14 results in an overvaluation of 4 percent (see text figure).
House Price Valuation: Fundamentals and Supply
(Index)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: OECD; ABS; RBA; Haver Analytics; IFS; and IMF staff calculations.House Price Valuation Measure Using Fundamentals
(Index)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: OECD, Igan and Loungani (2012); and IMF staff calculations.Real House Price Index and Trend
(2000Q1=100)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: Haver Analytics and IMF staff calculations.9. How can one relate household debt to house prices? The link between the real interest rate and the debt/income ratio can be illustrated in a general equilibrium framework (see, for example Walentin, 2013), where lower real rates support a higher debt ratio. One can also extend the user cost model, discussed above, to illustrate the effects of lower real interest rates on household debt (see Figure 3 and Appendix 2 for details). Using this approach, a permanent decline in real interest rates of one percent suggests a change in the debt ratio in the range of 10-20 percent although the effects are non-linear (see Figure 3). Thus, in this highly stylized context, the nominal and real interest rate assumption has considerable implications for the equilibrium level of indebtedness (see also Ellis, 2006).
10. The trends of lower interest rates, higher house prices and household debt are closely interrelated. However, the fact that higher house prices and household debt can largely be explained does not imply that it is sustainable in the long-term. Sustainability would depend on the evolution of variables such as the nominal interest rate, income growth and expectations of real appreciation.
Bottom line: Lower nominal and real interest rates and financial liberalization are key contributors to the strong increases in house prices over the past two decades. The various house price modeling approaches indicate that house prices are moderately stronger (in the range of 4-19 percent) than economic fundamentals would suggest.
Counter argument 3: High prices reflect low supply
“A country with this much land, it ought to be cheap to get a roof over your head” Gov. Stevens, February 2015
11. Housing supply has not kept up with demand. Australia’s population has grown rapidly since 2000, and much faster than the OECD average. However, residential investment has remained stable around 5 percent of GDP for much of the past two decades, only in line with OECD average (see text chart). Vacancy rates have also remained relatively low since 2006, compared with the 1990s (see Kent 2013).
12. While Australia is sparsely populated, much of the country is remote, and the population is highly urbanized. As city prices are typically higher than rural, countries with high degrees of urbanization tend to have higher house prices on average (see Figure 3). Supply of housing tends to be inelastic as geographic conditions, such as limited available land for high density housing and lack of infrastructure can restrict housing supply in certain areas, causing house prices to increase rapidly. The supply response to higher house prices has also been relatively slow in Australia (see OECD 2011), but there are signs that housing completions have increased recently.6
13. In addition to geographical constraints and household preference, housing supply has been constrained by policy factors. In a recent paper, Hsieh et al (2012) highlights the complexity of planning process; provision and funding of infrastructure; land ownership and geographical constraints; and public attitudes towards infill development as key factors in explaining the sluggish supply response. For instance, turning farmland into higher density housing typically takes about 6 years. Against this background, policy reforms can increase supply, including through designing and enforcing efficient land-use regulations; and providing complementary infrastructure and other public services.
14. Housing construction is currently at record levels, stimulated by high house prices, low interest rates, reducing the existing supply shortages in some areas. It is estimated that close to 200,000 dwelling units will be added in 2015, with high-density housing accounting for around a quarter of new approvals. Data from the National Housing Supply Center study (2011) suggest that the most significant cost in greenfield and brownfield development is construction costs, although the cost of land is also important, particularly in Sydney. At the same time, construction costs have remained relatively muted and not risen nearly as fast as house prices (see Richards, 2008).
15. A slow supply response to rising demand in some areas mitigates house price overvaluation concerns, but does not rule out large adjustments. While supply constraints do suggest that equilibrium property prices have risen, they do not rule out that demand is excessive, nor that it could fall sharply. House prices in Australia have varied by more than can be explained by the relatively stable deviation between population and housing supply. The UK, for example, had little supply response in the housing boom of the 2000s, but still saw a 20 percent fall.
Bottom line: housing supply does indeed seem to have grown significantly slower than demand, reducing (but not eliminating) concerns about overvaluation.
Counter argument 4: It is just a Sydney problem, not a national one
16. Property prices in Sydney have risen sharply in the post-GFC period. In the first half of 2015, property prices increased by 16 percent y on y in Sydney, compared with 10 percent in other capital cities. The median price over A$760,000 in Sydney corresponds to around nine times income, also the highest among capital cities. According to the Moody’s (2015), the average household now spends 35 percent of household income on monthly mortgage repayment. However, some of the increase in Sydney house prices represents a catch-up given that prices in Sydney were largely stagnant in the between 2003-09 compared to other capital cities. 7
Residential Property Price
(Annual % change, 2015 Q2)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Source: Corelogic.17. Investor activity is a key driver of house prices in Sydney. New investor loans have doubled in recent years, with low interest rates and strong competition among lenders stimulating investor lending growth, especially in Sydney. The RBA (2015) observes that investor housing loan approvals in New South Wales have increased by almost 150 per cent over the past three years and accounts for almost half the value of all housing loan approvals in that state.
18. Foreign demand for housing in Sydney is also a factor that may have supported house price increases. Gauder et al (2014) note that foreign purchases appear to be most concentrated in new, high-density dwellings in inner-city Sydney and Melbourne that tend to be relatively expensive. Foreign investment has remained broadly stable at a national level over the past decade but has picked up recently.
Bottom line: The two most populous cities, Sydney and Melbourne, have seen strong house price increase in recent years, including in the investor segment. A sharp downturn in the housing market in these cities could be expected to have real sector spillovers, pointing to the need for targeted measures –including investor lending-to reduce the risks related to a housing downturn.
Counter argument 5: There are no signs of weakening lending standards or speculation
19. Mortgage lending has grown strongly, but lending standards seem largely to have been maintained (Figure 4).
LTV ratios are stable. The proportion of high LTVs for both investor and owner occupied housing lending has remained steady (see Figure 4). The estimated average loan-to-value ratio on existing loans is around 58 percent (see Moodys 2015).
Asset quality remains strong. Non-performing loans have been historically, and remain, low. Mortgage loans are full-recourse, which implies that the mortgage holder is legally responsible for the loan amount regardless of default or repossession of the property by the lender. Non-performing loans for commercial property are higher, and peaked in 2010 at 6 percent, but have fallen back to 2 percent since then. For evidence of default behavior of Australian mortgages, see Read, Stewart and La Cava (2014). The share of low-documentation loans has declined (Figure 4) and these are essentially no longer offered due to consumer responsible lending law changes.
Mortgage buffers have increased. Australia’s mortgage structures encourage the early repayment of mortgages as lenders typically do not impose penalties on households who pay down their variable-rate mortgages. When mortgage interest rates are lowered, borrowers often continue their mortgage payment at the same level so that greater principal payments are made and buffers are therefore built automatically as interest rates are lowered. Balances in mortgage offset and redraw facilities are estimated to be 15 percent of the outstanding stock of housing loans or over two years of scheduled payments at current interest rates (Figure 5).
Debt is concentrated among high-income households. Household housing assets and financial assets are respectively around three times and twice bigger than household debt. Only a small number of indebted households (around 4 percent) currently have debt which exceeds assets. Households in the top two income deciles held around 70 percent of mortgage debt and only 10 percent of mortgage debt was held by households in the bottom two income deciles and these tended to be older households with typically smaller mortgages and higher net asset holdings (Figure 5). 8
But lending standards need to be maintained. Byres (2015) highlighted evidence of non-prudent lending standards in a recent APRA survey of lending institutions, including treatment of income, interest buffers and interest-only loans.
Financial Stability and Housing Lending Standards
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: APRA and IMF staff calculations.Households Have Large Buffers
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Sources: ABS; RBA; APRA; RP Data-Rismark; Country statistical offices; Haver Analytics; and IMF staff calculations.20. There is no sign of a generalized credit boom and estimates of credit gaps are small. Using financial gap estimates from Borio et al. (2013) methodology, yields small gaps. Following the policy tightening by APRA and RBA in 2003-04, credit growth to owner-occupied and investor lending slowed.
21. However, some specific areas of concern have emerged. Investor credit has picked up sharply lately, largely focused in Sydney, with the majority in interest-only lending (driven by tax benefits). House prices in Sydney, partly as a result, have increased sharply, and are now up by more than 30 percent since 2011. Against a backdrop of already high house prices and household debt, continued rapid house price inflation, especially in Sydney, could give rise to expectations-driven, self-reinforcing demand dynamics, accompanied by price overshooting and excessive risk taking by banks.
Financial Gap
(Percent deviation)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Source: IMF staff calculations.Bottom line: While lending standards overall seem not to have loosened, the growing share of investor and interest only loans focused on the highly-buoyant Sydney market, is a pocket of concern.
Counter argument 6: Even if they are overvalued, it doesn’t matter as banks can withstand a big fall
22. APRA recently concluded a stress test of the Australian banking system, focused on housing. APRA considered two severe tail macroeconomic stress scenarios, developed in collaboration with the Reserve Bank of Australia (RBA) and the Reserve Bank of New Zealand (RBNZ).
House price bust (Scenario A). A housing market decline, prompted by a sharp slowdown in China, where Australian GDP growth declines to -4 percent, unemployment increases to over 13 percent and house prices fall by a cumulative 40 percent.
Higher interest rates (Scenario B). In the face of strong growth and emerging inflation, the RBA raises the cash rate significantly. Global growth subsequently weakens and a sharp drop in commodity prices leads to increased uncertainty and volatility in financial markets. This leads to higher unemployment and higher borrowing in Australia and a significant fall in house prices.
GDP Growth
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Source: APRA.Unemployment
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Source: APRA.House Price Index
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Source: APRA.23. The banking sector would remain solvent, but unlikely to function well. In each scenario banks face an increase in funding costs, decline in credit quality and credit losses, with a significant adverse impact on profitability and declines in capital ratios. Losses on residential mortgages accounted for around one-third of total credit losses. These aggregate losses contributed to a material decline in the capital ratio of the banking system. Starting the scenario at 8.9 percent, the aggregate Common Equity Tier 1 (CET1) ratio of the participant banks fell by around 3 percentage points. All banks remained above the minimum CET1 capital requirement of 4.5 per cent. However, losses on housing were greater than the capital held for housing for the Internal Ratings Based (IRB)-model banks and almost all banks would use capital conservation buffers and face constraints on dividend and bonus payouts. In such circumstances banks would face funding constraints and likely curtail lending. This would likely exacerbate an already extremely difficult macroeconomic situation.
24. Even abstracting from the impact on banks, a sharp fall in house prices would likely have major macroeconomic effects. This would operate through many channels (see Debelle, 2004). For example:
Wealth effects: households would cut consumption as their housing wealth falls. Dvornak and Kohler (2003) find wealth effects of around 3 cents on the dollar.9 For example, a 10% fall in house prices would reduce household wealth by some 30 percent. Applying a typical elasticity, in an admittedly partial equilibrium framework, GDP would fall by around 1 percent, cumulatively.
Investment: Investment and employment in housing would decline. Although dwelling investment has been tightly range bound, between, 4 and 6 percent of GDP over the past decades, a slowdown would be expected to have an adverse impact economic activity.
Commercial Property Price
(Year-on-year change)
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Source: APRA.APRA Stress Tests: CET1 ratios fall sharply
Citation: IMF Staff Country Reports 2015, 275; 10.5089/9781513556499.002.A001
Source: APRA.1/ Capital conservation buffer reflects D-SIB surcharge effective January 1, 2016.Bottom line: While bank capital levels are likely sufficient to keep them solvent in the event of a major fall in house prices, they are not enough to prevent banks making an already extremely difficult macroeconomic situation worse.
References
Alsterlind, J., and U. Holmberg, K. Jonsson, B. Lagerwall, J. Winstrand, 2014, “Memorandum 6 – Risks to the Macroeconomy and financial stability arising from the households’ debts and housing prices,” Mimeo, Sveriges Riksbank.
Australia Bureau of Statistics, 2013, “4130.0 Housing Occupancy and Costs, 2011-12” http://www.abs.gov.au/ausstats/abs@.nsf/Latestproducts/4130.0Media%20Release12011-12?opendocument&tabname=Summary&prodno=4130.0&issue=2011-12&num=&view=
Borio, C. et al, 2013, “Rethinking potential output: embedding information about the financial cycle,” BIS Working Paper No 404.
Byres, W., 2015, “Sound lending standards and adequate capital: preconditions for long-term success,” Speech to COBA CEO and Director Forum, Sydney, May 13, 2015. http://www.apra.gov.au/Speeches/Pages/Sound-Lending-Standards-and-Adequate-Capital.aspx
Caldera Sanchez and Johansson, 2011, “The Price Responsiveness of Housing Supply in OECD Countries,” OECD Economics Department Working Papers No. 837.
Debelle, G., 2004, “Macroeconomic Implications of Rising Household Debt,” BIS Working Papers No 153.
Dvornak, N. and M. Kohler, 2003, “Housing Wealth, Stock Market Wealth and Consumption: A Panel Analysis for Australia,” RBA, RDP 2003-07.
Ellis, L., 2005, “Disinflation and the Dynamics of Mortgage Debt,” in Investigating the Relationship between the Financial and Real Economy, BIS Papers No 22,
Ellis, L., 2013, “Housing and Mortgage Markets: The Long Run, the Short Run and Uncertainty in Between.” RBA Address to Citibank Property Conference (4/23/2013).
Ellis, L. and C. Naughtin “Commercial Property and Financial Stability—An International Perspective,” RBA Bulletin, June Quarter 2010.
Ellis, L. and D. Andrews, 2001, “City Sizes, Housing Costs, and Wealth,” RBA Research Discussion Paper No 2001-08.
Finlay, R., 2012, “The Distribution of Household Wealth in Australia: Evidence from the 2010 HILDA Survey,” RBA Bulletin, March, pp 19–27.
Fox and Tulip, 2014, “Is Housing Overvalued?,” RBA, RDP 2014-06.
Gauder, M., C. Houssard and D. Orsmond, 2014, “Foreign Investment in Residential Real Estate,” RBA Bulletin, June, pp 11–18.
International Monetary Fund, 2014, Housing Watch, June 2014.
Hsieh W, D Norman and D Orsmond (2012), “Supply-side Issues in the Housing Sector,” RBA Bulletin, September, pp 11–19.
Igan, Deniz, and Prakash Loungani, 2012, “Global Housing Cycles,” IMF Working Paper No. 12/217 (Washington: International Monetary Fund).
Kent, C., 2013, “Recent Developments in the Australian Housing Market,” Address to the Australian Institute of Building, Sydney, 14 March.
Moodys, 2015, “Rising Housing Market Imbalances Pose Long-Term Challenges for Australian Banks,” May 13.
OECD, 2011, “Housing and the Economy, Policies for Renovation,” Chapter 4 in Economic Policy Reforms 2011: Going for Growth.
OECD, 2015, Focus on Housing, http://www.oecd.org/eco/outlook/focusonhouseprices.htm
Read, M., G. La Cava and C. Stewart, 2014, “Mortgage-related Financial Difficulties: Evidence from Australian Micro-level Data,” RBA Research Discussion Paper No 2014-13.
Reserve Bank of Australia, 2015, Financial Stability Review, March.
Richards, A., 2008, “Some Observations on the Cost of Housing in Australia,” Address to 2008 Economic and Social Outlook Conference, The Melbourne Institute, Melbourne, 27 March.
Senate Economics References Committee, 2015, “Out of reach? The Australian housing affordability challenge,” May 2015.
Urbis, 2011, ‘National Dwelling Cost Study’, Prepared for the National Housing Supply Council, May.
Walentin, K., 2013, “Housing Collateral and the Monetary Transmission Mechanism,” Forthcoming Scandinavian Journal of Economics.
Windsor, C., J. Jääskelä and R. Finlay (2013), “Home Prices and Household Spending,” RBA Research Discussion Paper No 2013-04.
Yates, J., 2011, “Housing in Australia in the 2000s: On the Agenda Too Late?,” in H. Gerard and J. Kearns (eds), The Australian Economy in the 2000s, Proceedings of a Conference, Reserve Bank of Australia, Sydney, pp 261–296.
Appendix 1A. Model A – Time-Series Model Including Supply Factors
Following the approach in Caldera Sanchez and Johansson (2011), house prices are modeled in a long-run framework, with per capita income, dwelling stock, working age population ratio, credit and long-term interest rates as explanatory variables, representing both demand and supply factors in driving house prices. The quarterly estimation period covers 2001-14.
where HPI is the log real house price index (OECD); IncomePC is the log real income per capita (ABS, national accounts); HousingStock is the lagged stock of housing (RBA); log real credit to the private sector (Haver), WorkPop is the ratio of working age population to total population (ABS), and InterestRates is the nominal long term interest rate (IFS). 10
The regression (OLS) results are as follows:
| Dependent Variable: LRHP | ||||
| Method: Least Squares | ||||
| Date: 05/28/15 Time: 13:53 | ||||
| Sample (adjusted): 2001Q1 2014Q4 | ||||
| Included observations: 56 after adjustments | ||||
| Variable | Coefficient | Std. Error | t-Statistic | Prob. |
| C | −25.1743 | 6.945491 | −3.62456 | 0.0007 |
| LRGDPPC | 1.086637 | 0.553155 | 1.964434 | 0.055 |
| LHSTOCK(-4) | −2.72587 | 0.780221 | −3.49372 | 0.001 |
| LRPCRED | 0.325403 | 0.175796 | 1.851022 | 0.0701 |
| WAPRATIO | 28.3602 | 12.5347 | 2.262535 | 0.028 |
| LTIR | 0.020073 | 0.011039 | 1.818397 | 0.075 |
| R-squared | 0.9451 | Mean dependent va | 4.514702 | |
| Adjusted R-square | 0.93961 | S.D. dependent var | 0.160498 | |
| S.E. of regression | 0.039442 | Akaike info criterion | −3.52704 | |
| Sum squared resid | 0.077782 | Schwarz criterion | −3.31004 | |
| Log likelihood | 104.7571 | Hannan-Quinn criter | −3.44291 | |
| F-statistic | 172.1482 | Durbin-Watson stat | 0.278079 | |
| Prob(F-statistic) | 0 | |||
| Dependent Variable: LRHP | ||||
| Method: Least Squares | ||||
| Date: 05/28/15 Time: 13:53 | ||||
| Sample (adjusted): 2001Q1 2014Q4 | ||||
| Included observations: 56 after adjustments | ||||
| Variable | Coefficient | Std. Error | t-Statistic | Prob. |
| C | −25.1743 | 6.945491 | −3.62456 | 0.0007 |
| LRGDPPC | 1.086637 | 0.553155 | 1.964434 | 0.055 |
| LHSTOCK(-4) | −2.72587 | 0.780221 | −3.49372 | 0.001 |
| LRPCRED | 0.325403 | 0.175796 | 1.851022 | 0.0701 |
| WAPRATIO | 28.3602 | 12.5347 | 2.262535 | 0.028 |
| LTIR | 0.020073 | 0.011039 | 1.818397 | 0.075 |
| R-squared | 0.9451 | Mean dependent va | 4.514702 | |
| Adjusted R-square | 0.93961 | S.D. dependent var | 0.160498 | |
| S.E. of regression | 0.039442 | Akaike info criterion | −3.52704 | |
| Sum squared resid | 0.077782 | Schwarz criterion | −3.31004 | |
| Log likelihood | 104.7571 | Hannan-Quinn criter | −3.44291 | |
| F-statistic | 172.1482 | Durbin-Watson stat | 0.278079 | |
| Prob(F-statistic) | 0 | |||
The explanatory variables generally have the expected sign and are statistically significant. Higher incomes have a positive impact on house prices; a higher housing supply negatively affects house prices; higher private sector credit and working age population ratio have a positive impact on house prices. A positive sign on long-term interest could signal expectations of higher income in the future, which could impact house prices positively.
Appendix 1B. Model B - Time-Series Model using Economic Fundamentals
Following Igan and Loungani (2012), real house price changes are modeled as a function of changes in affordability, real disposable income per capita, working-age population, equity prices, and the level of short- and long-term interest rates. The following quarterly regression is estimated for the period 2001-2014:
where ΔHPI is the change in real house prices over the last quarter (capital cities), A is affordability level of housing in the previous period, measured by (the log of) the ratio of house prices to income per capita; ΔYPC is the change in real income per capita over the last quarter; Δ WAP is the change in working-age population over the past year; Δsp is the change in stock prices over the year before last and
The regression equation is estimated using ordinary least squares (OLS).
| Source | SS | df | MS | Number of obs | 59 | |
|---|---|---|---|---|---|---|
| F(7, 51) | 7.38 | |||||
| Model | 0.012661793 | 7.001808828 | Prob > F | 0 | ||
| Residual | 0.012492343 | 51.000244948 | R-squared | 0.5034 | ||
| Adj R-squared | 0.4352 | |||||
| Total | 0.025154136 | 58.000433692 | Root MSE | 0.01565 | ||
| hpi | Coef. | Std. Err. t | P>t | [95% Conf. | Interval] | |
| Inafford1 | −0.0727 | .0231643 | −3.14 | 0.0030 | −0.1192 | −0.0262 |
| dlnypc | 1.0307 | .451608 | 2.28 | 0.0270 | 0.1240 | 1.9373 |
| stir | −0.0138 | .0025826 | −5.34 | 0.0000 | −0.0190 | −0.0086 |
| ltir | 0.0087 | .0045543 | 1.91 | 0.0620 | −0.0005 | 0.0178 |
| dlneq1 | 0.0352 | .0171953 | 2.05 | 0.0460 | 0.0007 | 0.0698 |
| dlnwap | 2.9587 | 1.195417 | 2.48 | 0.0170 | 0.5588 | 5.3586 |
| dlncons | 0.0964 | .0482841 | 2.00 | 0.0510 | −0.0005 | 0.1933 |
| _cons | −1.5310 | .4820586 | −3.18 | 0.0030 | −2.4988 | −0.5632 |
| Source | SS | df | MS | Number of obs | 59 | |
|---|---|---|---|---|---|---|
| F(7, 51) | 7.38 | |||||
| Model | 0.012661793 | 7.001808828 | Prob > F | 0 | ||
| Residual | 0.012492343 | 51.000244948 | R-squared | 0.5034 | ||
| Adj R-squared | 0.4352 | |||||
| Total | 0.025154136 | 58.000433692 | Root MSE | 0.01565 | ||
| hpi | Coef. | Std. Err. t | P>t | [95% Conf. | Interval] | |
| Inafford1 | −0.0727 | .0231643 | −3.14 | 0.0030 | −0.1192 | −0.0262 |
| dlnypc | 1.0307 | .451608 | 2.28 | 0.0270 | 0.1240 | 1.9373 |
| stir | −0.0138 | .0025826 | −5.34 | 0.0000 | −0.0190 | −0.0086 |
| ltir | 0.0087 | .0045543 | 1.91 | 0.0620 | −0.0005 | 0.0178 |
| dlneq1 | 0.0352 | .0171953 | 2.05 | 0.0460 | 0.0007 | 0.0698 |
| dlnwap | 2.9587 | 1.195417 | 2.48 | 0.0170 | 0.5588 | 5.3586 |
| dlncons | 0.0964 | .0482841 | 2.00 | 0.0510 | −0.0005 | 0.1933 |
| _cons | −1.5310 | .4820586 | −3.18 | 0.0030 | −2.4988 | −0.5632 |
The explanatory variables generally have the expected sign and are statistically significant. Affordability is negatively related to the change in prices and change in income per capita enters the equation with a positive sign. There is also a positive and significant relation between house price changes, equity prices, construction costs, and population growth. On the interest rates, there is a negative coefficient on short term interest rates (reflecting higher cost of mortgages) as expected, but a positive sign on long-term interest rates. A positive relationship may emerge if higher longer term rates signals an improved economic outlook which may stimulate housing markets.
To arrive at an estimate of overvaluation, it is assumed that house prices were in equilibrium in 2000 (after the transition to lower inflation and interest rates) the house price index is set to 100. Using the predicted house price changes from the regression analysis, index values are computed from that date onward. To assess whether house prices are in line with the economic fundamentals of the model, the actual index value is compared to the predicted one and the difference between the two values labeled as the estimated price gap.
Appendix 2. User Cost and Household Debt
User cost equilibrium in the housing market occurs when the expected cost of owning a house equals that of renting. In this context, overvaluation is defined by the actual price being greater than that calculated with the user cost. In equilibrium (using the definitions in Fox and Tulip, 2014):
where P* is the “fundamental” value of housing; r is the real interest rate; c is running costs such as repairs and insurance as proportion of price; s is transactions costs averaged over the period of ownership as proportion of price and π is the expectation of real appreciation on a constant quality basis.
Australia: User Cost of Housing and Household Debt
Australia: User Cost of Housing and Household Debt
| Model parameters | ||||
|---|---|---|---|---|
| M | Rental payments as a share of income | 0.2 | Source: | ABS (2013) |
| LTV | Average loan-to-value ratio | 0.578 | Source: | Moodys (2015) |
| c | Running costs such as repares insurance as a percentage of price | 0.015 | Source: | Fox and Tulip (2014) |
| d | Depreciation | 0.011 | Source: | Fox and Tulip (2014) |
| s | Average transaction costs | 0.007 | Source: | Fox and Tulip (2014) |
| change in constant quality prices | 0.017 | Source: | Fox and Tulip (2014) | |
| r | Real mortgage rate | [0.02-0.14] | ||
Australia: User Cost of Housing and Household Debt
| Model parameters | ||||
|---|---|---|---|---|
| M | Rental payments as a share of income | 0.2 | Source: | ABS (2013) |
| LTV | Average loan-to-value ratio | 0.578 | Source: | Moodys (2015) |
| c | Running costs such as repares insurance as a percentage of price | 0.015 | Source: | Fox and Tulip (2014) |
| d | Depreciation | 0.011 | Source: | Fox and Tulip (2014) |
| s | Average transaction costs | 0.007 | Source: | Fox and Tulip (2014) |
| change in constant quality prices | 0.017 | Source: | Fox and Tulip (2014) | |
| r | Real mortgage rate | [0.02-0.14] | ||
Following the approach in Alsterlind et al (2014), one can link the user cost model to household debt,
Assuming households aim to maintain a constant loan to value ratio,
where
With the user cost model, one can obtain a link between user cost and the long-term debt ratio:
The chart in Figure 3 is computed with the assumption that renters spend about 20 percent of their income on housing (see ABS, 2013), and an average loan-to-value ratio of 58 percent (see Moodys, 2015).
In the case of the US housing downturn in the Global Financial Crisis, some of the largest downturns occurred in the cheapest regions and, in that context, house price growth may be a better predictor than house price levels.
Ellis (2005) places the transition period towards higher household debt and house prices after 2000, noting that the transition was gradual.
Fox and Tulip (2014) use proprietary data on prices and rents for matched properties, whereas these estimates are based on house prices and rental yields in capital areas.
Fox and Tulip (2014) use the long-term expectations of real house price increases of 2.4 percent (average from 1955) to find owning a house is about as expensive as renting. They note, however, that using the average over the past decade of 1.7 percent results in house prices being overvalued by 19 percent.
The smoothing parameter is set at 1600.
A preference for low density housing and declining household size may also be factors in explaining house prices (see Kent, 2013).
Full analysis of housing valuation in Sydney would require analyzing fundamental determinants of house prices and beyond the scope of this note.
There are distributional and generational aspects to changes in housing prices and rents (see Richards, 2008).
Windsor et. al (2013) find that the removal of credit constraints (consumption rises with home prices due to households’ ability to borrow more, given more valuable collateral), and the related buffer-stock savings argument (higher home prices act as a form of precautionary savings for low-saving households, allowing them to increase spending.
The order of integration is the same for all variables, albeit with varying levels of significance.
Data sources are described in Igan and Loungani (2012).
