Dynamic macroeconomic modelling of housing market fluctuations in Australia

The thesis investigates the housing market from four perspectives. First, the relationship between the economy and the housing market is examined using empirical analysis to reassess stylised facts. Second, the link between consumption and different wealth measures is evaluated involving house prices to quantify the wealth effect. Third, the role of the Australian housing market in credit instruments’ fluctuations and business cycles is explored. Fourth, housing affordability policies are analysed in a calibrated medium-scale DSGE model.

The first study investigates the role of the housing market in aggregate fluctuations at a business-cycle frequency based on the identified stylised facts using a quarterly data set of the main macroeconomic and housing market indicators between 1982 to 2021. An investigation of the average volatilities finds that non-residential investment is almost seven times higher, while residential investment is six times higher volatility than output. The construction activity (10.77) and house prices (3.63) show a higher relative standard volatility in the sample. The total credit and housing mortgage credit have almost the same relative volatility (2.2 and 2.9, respectively). The Australian data do not support the theoretical claim that business investment should lead output whereas residential investment should lag it. The calculations show the opposite. Their correlation with the output gap is the same (around 0.50). Both housing price and housing material costs indexes are weakly procyclical with the output gap of around the same magnitude (around 0.30 and 0.32, respectively). The former is contemporaneous, the latter lags the cycle by two quarters. The total value of new building approvals is weakly procyclical (0.29) and it leads output, thus tending to peak one quarter before GDP. The real variable mortgage lending rate is negatively correlated with the cyclical component of output (–0.14), and it is a lagging variable. The results show that house prices become countercyclical, at least in the last decade. There are consistent findings with the transmission mechanism of monetary policy on house prices and residential investment by the variable lending rates. The results also indicate a sharp rise in total credit-to-output and mortgage credit-to-output ratios. Averaging over the sample period, the housing debtto- total assets ratio is 18%, the housing debt-to-housing assets ratio is higher, 22%, and the debt-to-income ratio is 88%, which has increased sharply and now is above 140%. The annual growth rate of real housing prices exceeded the household’s real disposable income in the last two decades. There is also an increase in households with private rental agreements, particularly in the under 35 age-group. These imply that during this period housing became, almost continuously, less affordable for households.

The second study examines the long-run relationships between household consumption, different wealth measures, housing equity withdrawal and house prices using the DOLS method on Australian data from 1988 to 2021. Overall, the estimations find that the elasticity of housing wealth is higher than stock market wealth in both measures of financial wealth. These results suggest that changes in housing wealth play a more important role in changes in consumption expenditures in the long run. Based on the estimation, on average, a 10% higher positive cash flow from housing transactions could increase per capita consumption by 0.66%. During the sample period, the household sector in aggregate increased its mortgage debt by more than its net spending on housing assets; therefore housing-related transactions generated a net positive cash flow for households that is available for other uses. The estimated elasticity of house prices on consumption is 0.125 (stock market wealth includes share equity) and 0.1593 (stock market assets involve all types of equity) which are both smaller compared to the case when financial wealth is not placed in the estimation (0.18). The elasticity of stock market wealth, if it incorporates neither share equity nor all equity holding, is much smaller than the elasticity of house prices. That comparison demonstrates that final consumption expenditure is more sensitive to changes in house prices than stock prices. In the last experiment that includes all variables, changes in house price and housing equity withdrawal have a greater impact on consumption than other wealth components. However, the presence of house prices tends to soften the power of housing equity withdrawal on consumption.

The third study estimates two SVAR models. First, I evaluate the mutual dependence between housing prices and both owner-occupier and investor credits using monthly Australian data from 1990 to 2021. Second, I examine the role of the housing market in business cycles in Australia using a quarterly dataset between 1982 and 2021. To investigate the mutual relationship between house prices and housing mortgage credits, the results find: (1) A high speed of mortgage credit acceleration leads house prices to rise in the short run; however, the effect of mortgage credit shock on house prices is very short-lived in the case of a shock to both credit aggregates. It is about three months. The results suggest that both increase house prices to the same extent, although a shock to investor credit keeps prices above their steady state in the full sample period. In the case of a shock to owner-occupier credit, house prices decline below their long-run equilibrium value after six months. (2) A house price acceleration has a role in high debt accumulation, the reaction of owner-occupier mortgages is stronger, while investor credit is weaker to a shock in housing demand. In the short run, there is an existing bi-directional relationship between mortgage credits and housing prices. The results from the extended SVAR model suggest that foreign shocks influence the Australian housing market and credit market, particularly the shock to the foreign interest rate. The exchange rate shock has an impact only on the credit market variables. The effects of domestic economic activity on housing shows a mixed picture: (1) A positive shock to output has a (weakly significant) negative impact on new houses built and house prices. (2) Due to a shock to the inflation rate both credit and housing market variables decline, thus the inflationary shock can be identified as an adverse supply shock. (3) The structural shock associated with the cash rate is not an unanticipated monetary policy shock, although the real value of building approvals significantly diminishes in the long run and house prices slightly rise. The credit supply shock can be identified but compared to the simple SVAR model, shocks to mortgage credit demand do not appear significant under this specification. Consequently, the mortgage credit acceleration does not lead to an increase in house prices in the short run. Housing supply through residential investment mildly affects output, inflation and cash rate, and thus business cycles. In addition, a positive shock in construction activity negatively impacts house prices. The extended SVAR model can produce only a one-way relationship between mortgage credit and prices; specifically, the causality goes from house prices to mortgage credits. The empirical results show that the housing demand shock has a significant role in Australian economic fluctuations in the short run, although not as strong as generally believed.

The last study models housing affordability policies on the Australian housing market in a calibrated medium-scale DSGE model. The results suggest that a positive technology shock in the non-housing production sector does not have a significant role in delivering greater housing affordability in terms of house prices and rent, although the aggregate housing stock improves. A positive housing-specific productivity shock can improve housing affordability, not just in terms of declining house prices and rents, but also by raising owner-occupied, government and aggregate housing stock. If the government provides public capital to the private construction sector, the program creates a shelter for households who cannot afford to purchase dwellings. In addition, the policy can lead to a higher level of housing affordability in terms of declining house prices and increasing housing supply, and stocks of available owner-occupied and government housing. Although, the rise is due to the higher demand for rental housing, it can be supported by a well-designed subsidy program. Without public capital involved, the government spending shock has a crowding out effect, that is both types of housing and residential investments decline. Real house prices decrease, but real rents increase moderately. Overall, a fiscal policy shock does not have a significant impact on housing affordability: house prices may show a contrary declining response in the model because they should increase instead as empirical evidence suggests. Both a tightening monetary policy and a positive inflationary shock raise the nominal interest rate under both specifications. An inflationary shock impacts the variables less than a monetary shock. A monetary policy contraction can have a positive impact on housing affordability in terms of house prices, that is, they are declining. However, without government housing in the model, residential investment housing activity for investment housing and demand for rental properties fall. Due to a monetary shock owner-occupied housing and residential investment rise and then sharply decline instead of an expected immediate fall. ...... [for the full abstract text, see the attached file abstract.pdf]