Value during inflation

A natural question to our presentation arose: which factors offer protection under inflation? Value.

A quick thank you to those who participated in the London Quant Group discussion of the paper we presented on the 10th. The question of inflation and its mitigation with overlays came up, and a simple follow up is in order. There is a duration argument to make, Value firms are providing lower duration cash flows than Growth. When those dollars are suffering from inflation, a dollar today buys more than a promised dollar tomorrow. Another consideration is that Value (without a proxy for quality) embeds some distress risk. They are cheap for a reason. Inflation correlates to an expanding economy, and there is less concern with a firm starving in times of bounty.

To look at this, I went to Ken French's website to grab the returns to book to market portfolios going back to 1926. The value weighted large cap tercile data was used to construct the HML ( high minus low) Value returns monthly. This represents the cheapest and easiest to realize implementation. As Stefano Cavaglia pointed out in our presentation, Dimensional Advisors has replicated the Fama French results for the past 37 years with about 15 basis points of slippage.

Next to Robert Shiller's page to get the CPI data. One could go to Fred, but Shiller replicated the data going back to 1871. Ken French goes back to 1927, and Fred starts in 1960. Using 12 month rolling changes in log CPI, the HML data was grouped by inflation quartiles. Here is the chart.

Figure 1: Box plots of the distribution of Value HML returns by CPI quartile. Highest inflation quartile is on the right. Most of the months fall within +/- 2.0%

Overall, the mean monthly return to HML was 25 basis points. We have 1,138 months of data to review. For the lowest inflation quartile, the mean was 1/4 of one basis point. The [.25, .50, .75]tile values within the Q1 group were [-2.4, -.39, 1.65], so half of the 285 observations were below -39 bips. For quartile 2, the mean was 37 bips and [-1.4, .12, 2.0]. The median of the 285 observation in this group were 12 bips. For Q3 these rise to a mean of 20 basis points and [ -1.79, .01, 1.85]. Finally under the highest inflation group, the mean return to HML is 43 basis points, and the quartile values were [ -1.59, .11, 2.36].

This is for US data, and the usual caveats of "this is not professional advice" it's just two hours of curiosity satisfied. The results are consistent with the presentation and earlier work on the protection offered by factor overlays. The simple analysis can be easily improved upon. I have not made use of commodities for instance. But it should be clear, Value helps ride out the storm of inflation.

Thanks again to all who participated, to our host Rupert Goodwin and big shout out to my co-authors, Stefano Cavaglia, Ken Blay and Scott Hixon.