RBA recommends you buy shares

Via Damien Boey at Credit Suisse:

The RBA has just released a paper about the history of the Australian equities. It contains some very useful and unique historical data on the Australian market. For context, earnings, dividends and total returns data for the Australian market has been quite sparse, and of questionable quality going back in time. Many use the Macquarie/AGSM 50-leaders data as a benchmark prior to 1974. But of course, these data are limited by naïve weighting and narrow sample size. The new RBA database has been constructed by the analyst (Matthews) sifting through decades of newspaper data going back to the early 1900s, covering as many index constituents as possible for the ASX 100, and properly cap-weighting their metrics. It fills in much of the missing data (quality) we have for the Australian market.


From Matthews’ database, we can reconstruct valuation and total return measures, and re-estimate the linkage between the two. We can run the usual regressions, of 10-year forward returns against today’s valuation, across a variety of different metrics. As we should expect, there is a negative correlation between future returns and today’s valuation. To an approximation, it is always the same uncertain future we are buying. The only question that matters is how much we are paying for that future. A higher (lower) valuation starting point should be consistent with lower (future) returns.


We run a battery of tests to see which valuation measure has the best predictive power. Candidates include 12-month trailing price-to-earnings and the so-called Shiller cyclically-adjusted price-to-earnings ratio, which uses a prior 10-year window to normalize the earnings and inflation cycle, thereby giving us a rough “through-the-cycle” valuation ratio. The Shiller measure clearly outperforms the 12-month trailing measure. And currently, it points to 10-year annualized returns of roughly 7.5%.

But we can go even further. In Australia, 10 years is not a long enough period to cover a cycle. After all, housing and commodities experience multi-decade cycles. If we calculate the Shiller valuation using a rolling 20-year window, the fit of the expected return equation improves even more, although the mean estimate of future returns remains around 7.5%.

Overall, equities look much better than bonds on a 10-year horizon, using the new RBA data. But there is one major limitation that we need to be aware of with the RBA data. More recent earnings data excludes loss making companies. Indeed, the ASX officially excluded loss makers from the calculation of price-to-earnings ratios since April 1991. Therefore, any valuation measure which does not correct for this structural break will implicitly overestimate the trajectory of earnings growth. Our preferred measure of valuation, which focuses on dividends rather than earnings, and makes an adjustment to payout ratios from dividend imputation, points to lower expected returns of around 4.3% annualized. For further discussion, please see our earlier article “The Fed model for Australia” dated 22 November 2018.

Latest posts by David Llewellyn-Smith (see all)

Comments are hidden for Membership Subscribers only.