The last leg of our foray into deconstructing Piotroski’s F-Score is a review of the final signal which is designed to measure changes in the efficiency of the company’s operations.
The Operating Efficiency signal is derived from the following two criteria:
As Piotroski points out, these ratios are important because they reflect two key building blocks underlying a decomposition of return on assets–as they signify greater productivity from the asset base and an improvement in profitability. However, instead of using Net Profit Margin, Piotroski elects to use Gross Profit Margin.
It’s useful to mention that Piotroski uses an increase in ROA within his Profitability signal discussed in a prior post, so it would be redundant for him to use it again here. In a sense though he is double-counting an improvement in ROA–once in the Profitability signal and here, approximately, in the Operating Efficiency signal. The rationale for this is not clear to me in the paper. However, by moving up the income statement, we can better evaluate improvement in operational efficiency for the money-losing companies as well–which we see later on.
To test for the Operating Efficiency signal, I have selected the Japanese Net-Nets that show both an improvement in asset turnover and an improvement in the gross profit margin. There is no delineation on the rate or degree of improvement, it simply answers the question: is the ratio in the current period higher then ratio in the year prior? As I’ve mentioned, Piotroski uses fiscal year numbers with a 5-month lag. I use the most current reporting period available at the time of stock purchase.
Below is a summary of the number of equal-weight positions held each year. F Score-OE represents the stocks that are positive for both criteria. Conveniently, this basket of stocks is represented in every year tested over the length of the study. I’ve included the number of stocks that scored an 8 or a 9 on the F-Score as a reference.
Although the performance of Japanese Net-Nets that score well on operating efficiency do well right out the gate, they perform no better than the entire basket of stocks in the end–and drastically underperform Japanese Net-Nets that have a high F-Score. From this rudimentary study completed over multiple blog posts, it appears that using fundamental momentum with Japanese Net-Nets must incorporate multiple signals as the F-Score does. We have not seen any indication that one signal is more important than another signal; and only by incorporating all of the Piotroski signals together will lead to a reduction in noise resulting in a true signal of fundamental momentum. It might be obvious in retrospect, but the F-Score has multiple signals of fundamental momentum for a reason.
From our studies, there are two promising strategies that might improve upon the performance of net-nets: Fundamental momentum and negative earners. So far it has not been clear if there is any correlation between the two strategies; they seem to behave in isolation from each other. Both strategies significantly outperform, albeit negative earners have higher volatility of returns than the high F-Score cohort.
However, when looking at the Operating Efficiency signal it appears that it may aid in better selecting the Japanese Net-Nets with negative earnings. If we first break out all of the Japanese Net-Nets that had negative earnings during the trailing twelve months and then further separate based on positive momentum in operating efficiency, the results are enticing.
Of course it’s problematic to draw too strong of a conclusion from this exercise. Given the small sample size, your basket of stocks in any given year are going to be only 2 or 3 positions in this study. And the severe concentration of holdings magnifies the big winners and losers–particularly in 2012 and 2014, and on the downside in 2018. However, it drastically outperforms high F-Score stocks and the entire basket of negative earners. Further study is warranted.
Thanks for reading!