Chart of the Day: PE or not PE

That is the question: today’s chart(s) are from Wilson HTM’s recent research on Australian equities which was prompted by Sell on News recent article on what is the fair value of the ASX200/All Ords? (h/t Damien for supplying the report)

The first chart shows the All Ords trailing Price/Earnings ratio going back to 1959:

Most brokers use forward earnings, I prefer trailing (as it is concrete – forwards are estimate only and always subject to revision) for comparison purposes.

What is obvious in this chart (and in any decent research) is that in a low inflation, “Great Moderation” environment, there is always a prospensity to bid up stocks – i.e pay more for the same earnings. The normalised trailing P/E of the last 30 years has almost always been above 14 times (the two exceptions were the bubbles in 1987 and 2007).

Now that we are approaching (or stuck within) a higher inflation, lower GDP growth, stagnant/disleveraging credit environment, what do you think is the most likely range of P/E measures? The same as the last 30 years – or perhaps a range below that, as experienced before the Baby Boomers discovered stocks?

The second chart compares the Shiller Cyclically Adjusted P/E ratio (or CAPE) of US and Aussie stocks. At Empire Investing, we use a version of this CAPE (which irons out the cycles within a market, adjusting earnings for inflation).

What is clearly evident is that our two stock markets roughly follow the same risk premia – although the US went nuts during the tech boom, whilst we caught up during the first Chinese bubble from 2003-07. The US, through fiscal and monetary stimulus and extreme cost cutting (i.e jobs) has been able to create huge margins and hence profits for its industrial companies and stands at very high CAPE. Our market in comparison has disleveraged, raising capital to pay down debt (and only getting a sub 10% return on that equity) whilst retaining (expensive) jobs and absorbing continued regulatory uncertainty amongst tightened margins.

Again notice before the 30 year megaboom, the CAPE was below 10 times – a more appropriate measure for potentially low earnings, due to margin pressure, inflation and tax constraints on industrial companies.

Latest posts by Chris Becker (see all)


  1. The question I think needs to be raised on this issue is, to what extent will the changes in the Australian sharemarket since the 60s and 70s influence this? The market back then had a much bigger industrials component, now it is dominated by resources and financials. Back then BHP was substantially an industrial stock (accounting for it appearing in the industrials share tables rather than resources).

    My feeling is that the outlook will continue to be dominated by overseas demand for our resources. If that holds up, the market will remain strong. Resources will become even more dominant in our market.

  2. There is definitely a premium paid for certainty i.e. moderate, positive stability.

    On any reasonable analysis, valuations, earnings and stock market returns during the 80s and 90s were anomalies (to the upside) in the context of financial market history. Only the post-war regeneration period generated double digit compound real returns over a decade or so (as did the 80s and 90s).

    Aided by strong tail winds of high interst rates/inflation and low valuations as a starting point, at the commencement of the 1980s (cite ‘death of equities’ headlines).

    Anyone expecting these rates of return as the ‘norm’ from equities going forward is in for a surprise. Anyone basing their retirement plans on it is in for disappointment.

    History may well show that 2003-2007 and 2009-2011 were just big bear market rallies.

  3. Prince, you should also plot 10-year average real earnings on that chart (Shiller E). It takes off around mid-1990s.

  4. What’s always interesting (and potentially the only relevant point) with any measure of market “value” is some form of empirical evidence that suggests a reasonable correlation of the measure with subsequent future returns. To put it simply, any measurement is useless if it does not draw conclusions that lead practical outcomes, outcomes that prove accurate more times than not.

    To my understanding there is virtually no statistically meaningful relationship between forward or trailing p/e’s and subsequent market returns (over any reasonable time period). The same goes for dividend vs. bond yields calculations (in fact even more so). Valuation models that attempt to adjust reported earnings for inflation and long term cyclicality are the only ones I have seen that pass the statistical relevance test, and even then, only over very long time periods.

    On that basis, any market analysis that relies upon such measures should probably fall within the same category that 90% of all market commentary belongs to popular press or investment marketing.
    As always I am happy to be proved wrong šŸ˜‰

    • With respect, you just contradicted yourself. You said there are no statistically significant forecasting models, then cited one (cyclically adjusted, seven to 10 year time horizon).

      I agree with what I think you mean, but long term cyclically and inflation adjusted models are, I think, reasonably reliably informative about future returns. Augment the Shiller PE with macroeconomic variables volatility and the findings can become rather compelling.

      Even I could get an R2 of 0.81 from an exponential trend line on average seven year returns vs starting Shiller PE. Problem is, the variability is large. You just never know if an MMT bubble is around the corner.

      I do think CAPE valuation is very informative on a longer term view, though. Valuation has to be the ultimate anchor. They tell you if it is reasonable to expect above average or below average returns, but they’re not precise.

      • Hi Ben,

        I guess I can see the confusion

        “To my understanding there is virtually no statistically meaningful relationship between forward or trailing p/eā€™s and subsequent market returns (over any reasonable time period).”

        This should say “between subsequent market returns (over any time period) and 12 month forward or 12 month trailing P/E’s.”

        and yes, my understanding is that various CAPE style ratios become somewhat useful at and beyond 10 year return marks.

        So yes, what you thought I meant was indeed correct šŸ™‚

  5. i disagree with your premise. it seems to me more likely that the higher money supply growth ushered in post-nixon-gold-window-closure put a premium on equities vis-Ć -vis bonds, dividends being a superior inflation hedge to fixed coupons.