The study investigates the impact of urban population growth on housing costs in Australia. While urban agglomeration boosts productivity, it also increases social costs, especially housing expenses. Balancing these is crucial for urban planners and policymakers. Existing research emphasizes the economic benefits of large cities but lacks quantified evidence on the housing cost implications of population growth. Estimating this effect is complex due to several challenges: omitted variable bias, measurement error (leading to attenuation bias), and the potential bidirectional relationship between population and housing costs (where high housing costs might discourage population growth). This paper addresses these challenges using a novel approach.

This paper contributes to the literature on housing affordability, a global issue. Existing research examines various drivers, including supply-side factors (construction costs, regulations) and demand-side factors (income inequality). The study connects to the literature on the effects of urban amenities; rising demand for urban amenities explains why rents might outpace wage growth. Cities with high productivity for skilled workers attract these workers, increasing local productivity, wages, and amenities, further fueling housing demand. This work also relates to studies on urban inequality and sustainability, acknowledging that smart city technologies may not necessarily reduce inequalities, and that low-income households face disproportionate risks of energy and transport poverty. Intrinsic inequalities are also more prominent in larger cities.

The study uses a panel dataset of 513 Australian cities (Local Government Areas or LGAs) from 2003 to 2016. Data on housing costs (average home and rental prices), housing supply (number of houses listed), and employment rates are from the Australian Urban Research Infrastructure Network (AURIN) and the Australian Bureau of Statistics (ABS). Visa issuance data (permanent and temporary skilled migrants, international students, long-stay businessmen) are from the Department of Home Affairs. City climate data are from the Bureau of Meteorology (BOM). Monthly data are aggregated to yearly frequencies. The main model examines the relationship between the logarithm of housing costs and the logarithm of city population, controlling for housing supply, employment rates, city fixed effects (time-invariant city characteristics), and state-year fixed effects (state-level and macroeconomic factors). To address challenges like measurement error, reverse causality, and omitted variable bias, the study employs a two-stage least squares (2SLS) instrumental variable (IV) approach within a panel data framework. The IV is constructed by interacting two variables: a dummy variable for ‘favorable climate’ (based on BOM’s climate zone classification, considering Zones 2, 5, 6, and 7 as favorable) and the lagged log of visas issued (log(visas_t-j), where j=1 or 2). The first stage regression models the city population as a function of this IV, control variables, city fixed effects, and time fixed effects. The second stage regresses housing costs on the city population (instrumented in the first stage) and control variables. A reduced-form regression analyzes the direct impact of the instrument on housing costs. The key identifying assumption is that visa issuance affects housing costs solely through its impact on city population, and this assumption is supported by the use of lagged visa issuance and the exogeneity of climate and the visa issuance process from a city perspective. The study compares the 2SLS estimates to ordinary least squares (OLS) estimates to demonstrate the potential bias in OLS estimates due to omitted variables and reverse causality.

OLS estimates show a statistically significant positive relationship between city population and housing costs (home and rental prices). However, the 2SLS estimates, which address endogeneity and measurement error, reveal considerably larger elasticities. Specifically, a 1% increase in city population is associated with a 1.16% to 1.59% increase in home prices and a 1.84% to 1.97% increase in rental prices, according to the 2SLS estimates. This highlights the downward bias in OLS estimates. Reduced-form estimates show that a 1% increase in visa issuance leads to an additional increase in housing costs, particularly in cities with favorable climates. The Kleibergen-Paap F-statistics confirm the validity of the IV. This implies that housing costs increase at a faster rate than population growth.

The findings strongly support a significant impact of city population growth on housing costs in Australia. The elasticity estimates suggest that rental costs rise more rapidly than population growth. This reinforces the concern that population growth exacerbates income inequality after housing expenses, as lower-income households allocate a larger portion of their income to housing. The study's methodological contribution lies in its use of a novel IV approach within a panel data framework, addressing reverse causality and unobserved heterogeneity, unlike previous studies that used pooled cross-sectional data or potentially endogenous instruments.

The study demonstrates a strong causal link between urban population growth and escalating housing costs in Australia, particularly rental costs. This highlights the need for proactive policies to address housing affordability challenges. Future research could explore the effectiveness of specific policy interventions in mitigating the impact of population growth on housing costs, potentially disaggregating the analysis at a finer geographic scale or examining the differential impacts across income groups.

The study focuses solely on Australia, limiting generalizability. The use of a national-level visa issuance as an instrument assumes uniform migration patterns across climates; variations in migration preferences among different visa categories might affect results. Furthermore, unobserved factors that correlate with climate and population could affect results, although the IV approach partially mitigates such concerns. Finally, the housing supply proxy might not capture all facets of housing supply dynamics.