Our latest rental forecast report will be published via News Ltd mid-February 2024.
In 2024, the rental market in Australia is anticipated to undergo significant strain, with widespread surges in rental costs placing considerable pressure on household budgets. Our analysis forecasts a notable 11% increase in house rents across the average Australian suburb. More pronounced, however, is the expected leap in unit rents, which are predicted to soar by an extraordinary 27%.
Links to the story will be added here. The full PDF version of the report will also appear once the results have been published by News Ltd.
In our study, we've grouped rental markets by broader suburban areas, aligning with the Australian Bureau of Statistics' Statistical Area 2 (SA2) standard. This approach enhances our modelling by providing a more stable and meaningful analysis base. However, to improve our forecast's precision and address the volatility often seen at the SA2 or suburb level, we shifted our focus to the Statistical Area 3 level (SA3). This strategic decision allowed us to capture broader market trends without the noise of short-term fluctuations.
Our Approach with the Gradient Boost Model:
Data Preparation: Utilising suburbtrends rental market data, our model harnessed detailed information across various SA3 regions. This data spanned different property types (houses and units) and included historical rental prices and market dynamics, offering a comprehensive view of the real estate landscape. This also extended to a simple 20% indexation adjustment applied to household income data from census 2021 data to estimate December 2024 income levels, helping us with affordability measures.
Model Training: We employed a Gradient Boosting model, a sophisticated machine learning algorithm renowned for its accuracy in handling complex datasets. The model operates by creating a series of decision trees, each one refining the predictions made by the previous, thereby improving accuracy incrementally.
Prediction Process: The model was tasked with projecting rental prices 12 months into the future for over 300 SA3 regions and more than 2,200 suburb areas (SA2s) . This involved analysing patterns and trends from historical data to predict future market movements.
Performance and Interpretation: Our model achieved an R-squared value of approximately 0.943. Put simply, this means it could explain about 94.3% of future rental price changes based on the factors it analysed. This high level of accuracy indicates that our model is adept at identifying key market drivers.
Sample Size and Coverage: The model drew on a diverse and extensive range of data covering numerous SA2 and SA3 regions, ensuring a robust and representative forecast. To enter the final report, only SA2 markets with more than 100 private rentals have been selected (one for houses and another for units).
Additional Processing for Accuracy: To refine our results, we averaged the last 12-month rental changes at the SA2 level with the 12-month forecasts at the SA3 level. To counter extremely high forecasts in some SA3 unit markets, we applied a maximum cap of 50% on increases.
In summary, our Gradient Boost model leverages historical data to deliver highly accurate forecasts for future rental prices. It stands as a vital tool for understanding market trends and making informed decisions in the real estate sector. The model's impressive explanatory power underscores its reliability and effectiveness.