Is it just me, or do Economic Policy Uncertainty indices look like they should be useful contrarian indicators?
Policy uncertainty spikes when investors are worried, headlines are dominated by political risk and the outlook feels unusually difficult to predict. That sounds exactly like the kind of environment where sentiment can become stretched and future returns improve.
But do the numbers actually back that up?
In Part 1, I looked at the different US Economic Policy Uncertainty indices, what they measure and why they can diverge.
In Part 2, I want to take the next step: how good are they actually as contrarian sentiment indicators?
Not all sentiment indicators are created equal.
Need to know
- The good: All of the indices are easily available, simple to understand and have decent histories. Several also have strong track records as contrarian indicators.
- The bad: Most of the data is only available monthly, and the pre-pandemic track record of some of the indicators was much less convincing.
Going deeper: the track records
The most important test for any contrarian sentiment indicator is whether it has shown predictive power in the past. In other words: has it consistently provided useful signals for subsequent asset returns?
For Economic Policy Uncertainty, we are looking for a pattern where the greatest policy uncertainty has been followed by the strongest equity returns, while periods of unusually low uncertainty have been followed by weaker returns.
One characteristic most EPU indicators share is that they have not produced reliable sell signals.
In principle, that is not particularly surprising given how the indices are constructed. Policy uncertainty tends to sit within a relatively tight range most of the time, interrupted by comparatively short periods when uncertainty spikes sharply.
So the more interesting question is whether those spikes have historically been useful buy signals.
And here, the track records differ quite widely across the various EPU indicators.
Best: the broad news-based indicators
The broad news-based measures have worked best as contrarian sentiment indicators.
They have strong overall track records when it comes to identifying periods of unusually high uncertainty that were subsequently followed by strong equity returns. Importantly, these results have not all been driven by one exceptional episode such as the post-pandemic recovery.
There is also an interesting result when comparing the main EPU with the broader categorical dataset.
Despite the seemingly huge difference in the size of the newspaper universe — around 10 newspapers for the main EPU versus more than 2,000 for the categorical EPU — the historical signals are remarkably similar.
A main EPU above 180 has been followed by an average S&P 500 return of 18% over the following year, with an 80% hit rate for above-average returns.