Free Cash Flow yield – value lost in estimation

The idea that we can use free cash flow (FCF) yield as a way to value a company is an extremely attractive one. The fundamental value of the business is, after all, derived from the cash that it generates and returns to its shareholders, be that in dividends, buyback or sale of the shares.

The way that free cashflow yield is promoted as stripping out the noise of nasty accounting adjustments is also attractive, particularly I suspect for non-accountants. It has been ‘adopted’ by a number of investment firms that like to talk about it in their marketing and sermonize about how they use it.  But as with PEGs, we should we wary of applying what the soundbites describe and perhaps give ourselves time to understand how they might actually use it.

FCF is generally defined as cash from operations less capex.  This FCF is examined against the enterprise value to give an ungeared FCF yield.  Normally interest is deducted, in which case the FCF is geared and examined against market capitalisation to find the geared cash flow yield.

FCF yield seems a nice single number answer that can be easily compared to investment instruments other than equity.

This is not to dismiss FCF yield as clickbait analysis.  FCF is one of the best indicators of value we have, and it is all too often ignored.

The problem is that having established that it bests EPS and EBITDA in a number of ways, and having heard the sermons, investors are wont to apply it without restraint or consideration.

Look before you leap

We are not going to dwell on how FCF should be used and what consideration must be given to issues such as cash flow timings, financial structures and share-based remuneration. The issue we are concerned about comes before that.  FCF as a valuation or screening technique needs FCF estimates, and it needs reliable estimates.

Particularly with smaller companies, the FCF metric is prone to significant distortion from cash flow events (or non-events) around the year-end – for example large contract payments.

Even though the likes of Refinitiv, Bloomberg, FactSet and S&P are systematically inhaling and exhaling the forecasts from sell-side analysts, it does not necessarily follow that these forecasts will be accurate or that the analysts producing them have given the metric much attention.

Company managements generally guide to revenue and then (adjusted) EBITDA and/or (adjusted) PBT and/or (adjusted) EPS, and then, generally in the broadest of terms, year-end cash.  They do not guide to FCF, geared or ungeared.

Some might suggest that this lack of guidance proves the value of FCF as reflecting true cash flow. It cannot be manipulated.  Others might suggest that this lack of guidance is because of FCF’s volatility, its unpredictability and its inherent limited value as a single period measure that does not reflect growth or returns.  We are comfortable sitting on the fence on this; on the cushion of ‘it depends’.

In the large-cap world, these issues tend to be less significant. They are perhaps better identified by the companies themselves and are considered in more detail by both the sell-side and buy-side communities.

In the small-cap world, it is always dangerous to make assumptions about the quality of estimates.

Looking at errors in context

We have looked at the errors in revenues, EBITDA, PBT and FCF forecasts in terms of percentage of the actual revenues – giving us figures for mean errors and std deviations.

By doing this, we can perhaps better understand how reliable FCF is going to be versus other metrics and where the borderline(s) may lie in terms of market cap and other parameters. Although the results are not going to be 100% accurate, without this basic test of the quality of the data we can’t say that FCF analysis is even 1% reliable.

Starting with UK listed companies with market caps above £100m provided a starting point of 1,032 companies. We regard this as a suitable cut-off point that avoids much of the seemingly random forecasting often found in micro-cap.

We removed the following ICB sectors from the original data: Oil, Gas & Coal, Open and Closed End Investments, Banks, Industrial Metals & Mining, Precious Metals & Mining, Finance & Credit Services, Investment Banking & Brokerage Services, Real Estate Investment and Services Development, Real Estate Investment Trusts, Life and Non-life Insurance and any stocks for which no ICB sector was returned. This narrowed our field down to 393 stocks, of which 359 had forecasts in the system (Refinitiv in this case) for revenue, EBITDA, profit before tax and FCF (ungeared).

A review of the data showed a number of anomalous sets of forecasts, so we applied with the criteria of limiting errors on estimate as a percentage of revenues: 5% for revenue, and 40% for EBITDA and PBT.  This reduced the number of companies with data to 242, and applying the 40% limit to FCF forecasts took it to 161 companies.  However, even within this sanitised sample set, the results are not that encouraging.

FCF worst of a bad(ish) bunch?

The data suggests clearly that FCF is not as reliable a forecast variable as some of the more-guided-to figures. The average error is 2-3 times that of the guided-to multiples.  It is important to remember that it is not simply about the error in the metric; it is also about the multiple we are placing on the error in the metric. A FCF yield of 5% is the same as paying 20x for that ‘free’ cash flow and 20x for that error – a similar multiple to that applied to EPS errors but significantly higher than is usually applied to the typically smaller errors in EBITDA.

Splitting the results above and below the £2,000m threshold does not suggest a considerable difference between large and small cap, but there is slightly greater accuracy with the larger companies. That is not to deny, however, that within tighter bounds or specific sectors significantly greater predictability may be present; perhaps a subject for another blog on forecast accuracy.

From this, it would appear foolhardy to blithely follow single-period FCF yield as a primary valuation tool, appealing as the rationale might be.

Judge the metric by its own yardstick

However, the proponents of FCF yield as a valuation tool (or screen) typically talk in terms of mid-single-figure yields as being appropriate for the kind of companies they are looking to invest in.  It would therefore be unfair to judge it (and them) on a set of data that they would not envision using. As can be seen in the table below for FCF yield of 4%-8%, which is where its proponents suggest its use, FCF yield is a somewhat more reliable metric (particularly for the largest stocks), but it remains significantly less reliable than the traditional guided to metrics.

However, we only have a limited number to choose from: 19 with a market cap of >£2,000m and still only 71 when we go down to a market cap of >£100m.

This hasn’t really got us far – other than to increase our focus and/or restrict our options.  FCF is looking far more like a crude screen than a precise valuation metric.  Perhaps the best term is ‘net’ rather than ‘screen’ – it catches some stuff but lets a lot of things through, some of which you might want, much of which you won’t.

Some advocates of the FCF, understanding its potential instability, often counsel not only targeting FCF yield range but also ensuring that the healthy FCF is consistently generated.  If we apply a threshold of a minimum of 4% FCF yield across Y-1, Y0 and Y+1, we are down to 46 stocks.  This falls to 32 stocks if we exclude those where the FCF forecasting error in Y0 was 20% or greater. A one-in-three chance of an error of over 20% is, after all, not great odds.

Of these 32 stocks, 10 are in the FTSE 100 and nine show growth in FCF in Y0 and Y+1.  All, needless to say, have FCF forecasting errors in our base year (Y0), but encouragingly for 24 of the 32 the error is in the right direction – something that suggests this may merit revisiting should we consider upside surprises and where to find them.

Looking at the list of stocks, it is difficult not to be impressed by the output of what is a relatively simple filtering process. However, this is 32 stocks to look at from a starting point of 359.  There are a host of other factors that need to be considered in making an informed investment decision, and if we are to look at 32 stocks and compare each to the other on a FCF yield basis, it does not provide much of a basis for granularity – we have created a tiny universe of investible stocks and (importantly) comparators.

While the healthy, multi-year FCF yield is often talked about, it is not as common as the fund manager rhetoric/marketing might suggest – certainly not enough to build an entire balanced portfolio around.  It is a fundamentally solid investment criteria that improves with consideration over multiple time periods, but it would be wrong to follow it as a valid and reliable valuation methodology as set out in the soundbites alone.

Ian Robertson

irobertson@progressive-research.com