Australia’s central bank has made hawkish noises of late.
“The board has a low tolerance for a slower return of inflation to target than currently expected,” Reserve Bank of Australia policymakers said earlier this month.
They’re especially concerned about oil prices, given the turn of unrest in the Middle East.
To generate a prediction for the Oct. 25 third-quarter consumer price index, we deployed univariate and multivariate forecasting techniques using our partnership with Indicio Technologies.
This chart tracks the FactSet consensus estimate on CPI from the third quarter of 2023 to the second half of next year – and compares it with our models from Indicio.
Our models forecast that inflation will soon start falling significantly below the estimate. Even for third-quarter CPI, our model sees the print coming in between 5 and 15 basis points below the FactSet forecast.
If that trajectory continues, we could return to the RBA’s inflation target range of 2 to 3 percent in 2024.
Should dovish inflation data prompt the RBA to “pause” instead of raising rates, this could be great news for the performance of various risk assets.
However, this would create a scenario where the Aussie dollar might continue to suffer.
Risks for the currency
As our following chart shows, the AUD (in blue), once at parity with the USD, has been on a downward trend for a decade.
The Aussie dollar is usually known as a “commodity currency” due to its sensitivity to prices for key exports, such as iron ore and natural gas. While the nation is not a major exporter of oil, this chart uses crude prices (in orange) as a proxy for commodities, given oil’s historic correlation with the AUD.
However, that historical correlation is not what it used to be. Since the start of 2021(and arguably in 2018-19), AUD/USD is much more correlated with the spread between US and Australian short-term bond yield curves (in green).
In summary, Australians should watch out for not only slowing inflation, but its knock-on effects.
More about how we leveraged Indicio x Macrobond
To produce the most accurate forecast, Indicio can calculate a summary of various models weighted according to their historical performance and accuracy.
In the model we constructed, the machine-learning models GRU (gated recurrent unit), LSTM (long short-term memory), and artificial neural networks scored the highest, and were thus assigned the highest weights in the final summary.
We ran two iterations of our forecast. The first one included all econometric and machine learning models that were statistically significant. The second selected only AI and machine-learning models.
In our data-driven approach, more than 20 variables were initially selected. Indicio’s advanced indicator analysis tool used a lasso structure to penalise variables that detracted from the forecast. Ultimately 13 real-time input parameters were chosen; the most influential included data series such as monetary aggregates, import prices, and personal deposit rates.
After picking out the series, the tool can then run more than 25+ advanced statistical models to create a a forecast on the Aussie CPI:
The Macrobond community can now access Indicio’s technology through a direct API. Indicio’s latest release allows Macrobond users to export outputs from their models and store them in Macrobond – using our front-end for visualisation purposes. Here’s an example of an Indicio dashboard that demonstrates its functionality for statistical modeling: