Equity valuations uncovered (part 1): Fathoming murky waters

Valuations and their role in forecasting returns is a fascinating yet complex subject. In fixed income, the connection is clear: for example, for a zero-coupon gilt with negligible default risk, the yield is the return if you hold it to maturity. Easy. 

But when it comes to equities, matters become far murkier. Future cashflows are highly uncertain (do high valuations signal high growth?), dividends may be economically irrelevant[1], accounting standards introduce nuances in measured earnings and equities have no maturity date.

And then there’s the statistical side of things – regressions on valuation ratios can introduce all sorts of complications, often undermining confidence in their usefulness. 

In a series of blogs, I take a deep dive into the role of valuations in forecasting equity returns, exploring the following questions:

Part 2: How can we avoid common pitfalls when interpreting valuation regressions and scatter plots? And are equity returns ultimately forecastable?

I start by examining the challenges of interpreting regressions of historic returns against starting valuations. There are plenty of pitfalls – statistical quirks, sample biases, and issues with how valuation metrics themselves evolve over time. The good news? Once these are carefully accounted for, it appears that valuations can still show meaningful forecasting power.

Part 3: Do valuations matter in the short term?

A common refrain is that valuations are irrelevant over short horizons. I investigate if that’s a fair assessment and in what ways valuations might actually be most relevant in the short term.

Part 4: Is tilting based on valuations a high-information-ratio trade?

It’s often assumed that understanding how expected returns change over time should offer a strong edge, but I investigate how much you can plausibly seek to boost risk-adjusted returns by market timing. We find that while traditional market timing can struggle, allowing for changes in risk at the same time may improve matters. 

Part 5: Do valuations only matter in extremes?

Another widespread claim is that valuations only matter in extremes, so I investigate this question from various perspectives, such as if extremes are removed. I also investigate how the precise form of the valuation metric used can lead to different conclusions, and why we might need to be particularly mindful of this choice when valuations are rich.

By the end of this series, I hope to have given a clearer picture of the role valuations play in forecasting equity returns – when they matter, when they don’t, and when their signals are worth acting on. Please stay tuned for part 2.

 

Assumptions, opinions, and estimates are provided for illustrative purposes only. There is no guarantee that any forecasts made will come to pass.

[1] Miller & Modigliani’s dividend irrelevance theorem says that in a perfect market the decision to pay dividends should not affect a company’s value since investors can ‘create their own dividends.’