Persistence of Margins

Unlike Growth, Margins Persist
 

By: Greg Obenshain

In a recent piece, "Persistence of Growth," we showed that growth is far less persistent than many might suppose. In fact, it is not persistent at all. As a result, valuation multiples tend to mean revert. This turns out to be useful because it is the mechanism of action that powers returns in a value strategy. But multiples do not mean revert fully, and in "Fair Multiples," we argued that the best indicators of whether a future multiple should be (slightly) higher or lower than the average market multiple are profit margins and free cash flow conversion.

We wanted to take a closer look at the persistence of margins to get a sense of the base rates of how they change over time. If they are important, we should have a good visceral understanding of how they have behaved in the past.

We’ll start with a measure of persistence that we used to judge the persistence of growth and ask, “What is the probability that a company has above-median margins every single year for five years in a row?” If margins persist perfectly, the 50% of companies that are above median in year 1 will also be above median in the following years, resulting in the straight black line across the top of Figure 1 below. If margins follow random chance, we’ll get the curved black line at the bottom of Figure 1, meaning only 3.125% of companies will have above-median margins every single year.

Figure 1: Persistence of Margins

Source: Capital IQ, Verdad Analysis. Uses annual data as of June 30 each year from 1996 to 2022 for US-listed companies with market cap greater than $100M.

Relative to the persistence of EBITDA growth, which hugs the random chance line, the results for margin persistence are remarkable. Gross profit margins are extremely persistent, EBITDA/GP is also persistent, and even free cash flow conversion, which we’d expect to be more volatile, is relatively persistent.

This also holds true for measures of returns on assets, the preferred metric for many measuring the quality of a business and that we believe works well when combined with value strategies.

Figure 2: Persistence of Return on Assets

Source: S&P Capital IQ, Verdad Analysis. Uses annual data as of June 30 each year from 1996 to 2022 for US-listed companies with market cap greater than $100M.

So, margins and return on assets persist, but the above charts don’t really give us the visceral sense of the actual levels of margins and how they change over time. To get to this, we divided the universe into quintiles for each valuation metric and then looked at the average starting value for each metric in year 1 and then how it changes in each subsequent year. Here are the results for gross profit margin.

Figure 3: Persistence of Gross Profit / Sales by Quintile

Source: S&P Capital IQ. Uses annual data as of June 30 each year from 1996 to 2022 for US-listed companies with market cap greater than $100M. Quintiles are formed each year. The median company value is calculated for each quintile and then averaged over all years.
  
This table is boring. Nothing happens. Gross profit margin is very sticky, and it seems that today’s gross profit margin is a very good estimate of next year’s gross profit margin. The story is a little more interesting for EBITDA/GP and free cash flow conversion (FCF/EBITDA). Shown below.

Figure 4: Persistence of EBITDA/GP and FCF/EBITDA by Quintile

Source: S&P Capital IQ, Verdad Analysis. Uses annual data as of June 30 each year from 1996 to 2022 for US-listed companies with market cap greater than $100M. Quintiles are formed each year. The median company value is calculated for each quintile and then averaged over all years.

Here we have some movement. Looking at the top few quintiles first, it appears that there is some mean reversion for the highest margins, but for the most part they are persistent. The lowest margins seem to mean revert upward dramatically, but there is a caveat to this. The worst quintile includes companies that either improved or went out of business. Almost by definition, the margins must get better in that quintile because only survivors are included in the out years.

To correct this, we can also look at survival-adjusted returns for each of these segments. This will capture the poor return from companies that went out of business. Below we show market-weighted returns.

Figure 5: Returns by Margin Quintile

Source: S&P Capital IQ, Verdad Analysis. Uses annual data as of June 30 each year from 1996 to 2022 for US-listed companies with market cap greater than $100M. Returns are market-capitalization weighted compound annual returns.

Here we can see a few effects that conform with our expectations. First, the further down we go on the income statement and cash flow statement, the more the margins spread returns, especially in the lower quintiles. Second, low-margin companies do worse, which conforms with what we’d expect from the academic literature on quality. Profits matter. And third, mean reversion in the top quintiles does in fact weigh on returns. This last point is important. Very high margins do not persist as well as “normal” margins, and so margins at the extreme high should be mean reverted to some extent.

This holds on an industry basis, as well. Below we show EBITDA/GP and FCF/EBTIDA persistence along with return spreads for excess margin over the industry median.

Figure 6: Persistence of EBITDA/GP and FCF/EBITDA Minus Industry Median EBITDA/GP and FCF/EBITDA

Source: S&P Capital IQ, Verdad Analysis. Uses annual data as of June 30 each year from 1996 to 2022 for US-listed companies with market cap greater than $100M. Margins are calculated as company margin minus industry median in each year. Quintiles are formed each year. The median company value is calculated for each quintile and then averaged over all years.

Using industry does lead to slightly greater mean reversion in the extremes and does spread survival-adjusted returns better, so looking at margins relative to an industry comp set is a valid approach, which is especially useful for sector analysts.

Finally, we can look at the data for return on assets measures. Below we show GP/assets and EBIT/assets, this time with the measures relative to industry.

Figure 7: Persistence of GP/Assets and EBIT/Assets vs. Industry Median

Source: S&P Capital IQ, Verdad Analysis. Uses annual data as of June 30 each year from 1996 to 2022 for US-listed companies with market cap greater than $100M. Margins are calculated as company margin minus industry median in each year. Quintiles are formed each year. The median company value is calculated for each quintile and then averaged over all years.

Here the message is even clearer. According to the data above, high return on capital persists, and higher profitability is better. The message to the analyst modeling company financials is somewhat heartening. While we’ve been adamant that growth is not persistent and that high growth rates should not be carried out for many years, margins and returns on assets are relatively sticky. Very high margins and returns on capital do mean revert, but slowly. While growth may be ephemeral, profitability is not.