Monday, 5 October 2015

Procter & Gamble, circa 1999/2000. DDM vs. Klarman DCF, and Thoughts on Risk

I seem to keep picking on poor P&G, but there's a method to my madness.  To me, P&G is the prototypical dividend stock.  Dow component, around for almost 200 years, predictable annuity type brands, and one would hope, predictable cash flows.

I recall March 2000 pretty vividly. I was a reconciliation clerk on the equity trading floor at CIBC World Markets in Toronto. Nasdaq 5,000 was the flavour of of day, and Jim Cramer was pounding the table on Amazon (unlike now). And there was P&G, a victim of 1998-1999 through early 2000 euphoria.   At one point in early 2000, P&G got up to a pre-split of almost $120 per share, so I thought it would would be an interesting exercise to look at the numbers behind the market cap.

For the year ended June 30, 1999, PG earned "core diluted" eps of $2.85, and "core" net earnings of $4.15B on revenues of $38.1B.  PG hit a pre-split high of $118.38 on January 12, 2000.  This would put the PE at 41.5x "core" eps.  PG expected to pay annual dividends of $1.28 per share in 2000, so the forward dividend yield was pretty low, about 1%.  The market cap at the January 12, 2000 high was about $169B, on revenues of $38.1B, equivalent to a price sales ratio of 4.4x.  Expectations were obviously pretty lofty for a 160 year old consumer products company (In defense to P&G, everything was bid).

I'm going to digress slightly and bring in Klarman (again).

In Margin of Safety, Klarman gives an example of a spin-off called Esco.  Here's the background from the book:

"Esco Electronics: An Exercise in Securities Valuation 

Let me offer a specific example of the security valuation process. Esco Electronics Corporation is a defense company that was spun off from Emerson Electric Company in October 1990; the shares were distributed free to shareholders of Emerson. Esco competes in a variety of defense-related businesses, including electronics, armaments, test equipment, and mobile tactical systems. Holders of Emerson received Esco shares on a one-for-twenty basis; that is, a holder of one thousand Emerson shares received fifty shares of Esco. 

Esco first traded at around $5 per share and quickly declined to $3; the spinoff valued at market prices was worth only fifteen cents per Emerson share (which itself traded around $40). Needless to say, many holders of Emerson quickly sold their trivial Esco holdings. 

What was Esco worth at the time of spinoff? Was it undervalued in the marketplace, and if so, why? Was it an attractive value-investment opportunity? The way to answer these questions is to evaluate Esco using each of the methods that value investors employ...

In order to value Esco using NPV analysis, investors would need to forecast Esco's likely future cash flows. Goodwill amortization of forty-five cents a year, as stated, was free cash. Beginning in 1996 there would be an additional forty-five cents of after-tax earnings per year as the $7.4 million guaranty fee ended. Investors would have to make some assumptions regarding future earnings. One reasonable assumption, perhaps the most likely case, was that earnings, currently zero, would gradually increase over time. Unprofitable contracts would eventually be completed, and interest would be earned on accumulated cash flow. An alternative possibility was that results would remain at current depressed levels indefinitely.

What was Esco worth if it never did better than its current depressed level of results? Cash flow would equal forty-five cents per share for five years and ninety cents thereafter when the guaranty payments to Emerson had ceased. The present value of these cash flows is $5.87 and $4.70 per share, calculated at 12 percent and 15 percent discount rates, respectively, which themselves reflect considerable uncertainty. If cash flow proved to be higher, the value would, of course, be greater."

So, with this in mind, I built a DCF matrix to test these calculations.

Notice that Klarman focused his analysis on two very interesting sources of "free" cash, 1) goodwill amortization, and 2) a payment commitment that Esco would need to pay Emerson until 1996.  He didn't consider accounting income, accounting EPS, or as reported free cash flows.  He focused in on only two aspects of free cash which he considered forecast-able.

Klarman makes the case for the NPV of just goodwill amortization and just the guarantee payment at a zero growth assumption, and at discount rates of between 12% and 15% in order to ensure that he had a truly sufficient margin of safety in the valuation.

I've run my own #'s as follows using a range of possible growth rates and a range of discount rates:

Assumptions:
- Two stage DCF model, start with FCF as calculated on Klarman FCF tab
- I want a large margin of safety, therefore, grow yrs 1-5 at low conservative growth rates
- Model considers years 1 - 5, then 6 through 10. Terminal val is perpetual growth discounted at 6-10 growth
- Create a matrix of resulting DCF's
- Margin of safety likely if resulting IV/share at low growth and Re b/ween 12% and 15% is greater than current market
Long Term Discount Rate10
2-Stage DCFYrs 1-5 Yrs 6+
5,00010,000
TWO STAGE DCFDep'nGuaranty
Free Cash Flow5,000.00
SharesOut11,111.00
Net Long Term Debt0.00
DCF-IV0.45
Sum of DFC'sRe =
G = 8%9%10%12%15%Average, growth, blended Re
0,09.458.477.716.665.737.60
1,09.518.527.766.715.787.65
2,110.739.468.507.216.108.40
3,1.511.5110.048.967.526.308.87
4,212.4110.719.477.856.529.39
5,2.513.4711.4610.058.226.749.99
6,314.7312.3410.698.626.9910.67
7,3.516.2613.3711.439.067.2511.47
Average, Re, blended growth12.2610.549.327.736.42
Current33333
Prem/disc to avg-75.53%-71.55%-67.82%-61.19%-53.31%


Now back to P&G.

I ran the same numbers through my valuation matrix circa 1999/2000.


Assumptions:
- Two stage DCF model, start with FCF as calculated on Klarman FCF tab
- I want a large margin of safety, therefore, grow yrs 1-5 at low conservative growth rates
- Model considers years 1 - 5, then 6 through 10. Terminal val is perpetual growth discounted at 6-10 growth
- Create a matrix of resulting DCF's
- Margin of safety likely if resulting IV/share at low growth and Re b/ween 12% and 15% is greater than current market
Long Term Discount Rate10
2-Stage DCF
TWO STAGE DCF
Free Cash Flow3,468.00
SharesOut1,446.00
Net Long Term Debt6,671.00
DCF-IV-2.22
Sum of DFC'sRe =
G = 8%9%10%12%15%Average, growth, blended Re
0,020.7517.4214.7610.766.7614.09
1,022.0717.6915.0211.006.9814.55
2,126.9121.3317.9513.048.3017.51
3,1.530.7523.6519.8214.349.1519.54
4,235.2626.2921.9015.7510.0621.85
5,2.540.6329.3224.2517.3111.0424.51
6,347.1132.8326.9219.0312.0927.60
7,3.555.0936.9629.9820.9313.2331.24
Average, Re, blended growth34.8225.6921.3215.279.70
Current118118118118118
Prem/disc to avg238.88%359.40%453.34%672.75%1116.17%


Even at the lowest possible Re = 8% and the highest two stage FCF growth rate of 7% and 3.5%, P&G seemed to be trading at double the theoretical FCF DCF valuation.

Obviously this is an exercise in over-simplification, and I wonder if the analysis is really relevant because I'm comparing apples to oranges.  I do think the analysis is relevant in order to hammer home the theoretical concept of margin of safety, and resulting risk.

P&G will liklely always trade at a premium valuation relative to unheard of spin-offs with complexities inherent in determining their future valuation, vs. P&G's predictable arsenal of brands and expected resulting future cash flows, and, due to the fact that it's a large cap Dow component, index funds, dividend, funds, index ETF's etc.,  have to own it in market weight proportions.

For the most part, P&G shouldn't stray too far from theoretical value (unless we're in the middle of Nasdaq 5000, everything's bid, and everyone loses their collective minds in terms of valuation - but this only happens, what, once every 10 years?), but the tiny little spin-off that institutions got from holding parentco's shares and don't really want, is too complicated to value, or cannot be owned due to institutional constraints? Herein lies an opportunity.

Inherent in P&G's valuation at the time was an expectation of high growth.  I interpolated the required two stage growth rates at required returns of between 8% and 10% in order to get close to a fair value of $120.

At Re = 8%, two stage FCF growth would need to be 11% for yrs 1-5, and 5.5% thereafter
At Re = 9%, two stage FCF growth would need to be 14% for yrs 1-5, and 7% thereafter
At Re = 10%, two stage FCF growth would need to be 16% for yrs 1-5, and 8% thereafter

The higher Re gets, the more ridiculous the FCF parameters become.

I find it interesting that while Klarman managed to prove a theoeretical margin of safety on Esco using the lowest possible FCF growth (0, 0) and highest possible Re (10% and 12%), I couldn't do the same for P&G.  My matrix at the highest FCF growth (7,3.5) and lowest Re (8%), only produced a theoretical value of $55 per share.

My layperson's guess is that this is symptomatic of  P&G prestige held by the market.  The expectations leading up to 2000 were that this stalwart dividend payer only  required an Re of around 8%.  i.e., the risk premium afforded to holding stocks was too low (and we know how this ended up).

And now DDM;

I'm going to assume that in late 1999, early 2000, investors expected to discount future dividends at Re = 8%, and, as noted above, expected FCF growth of 11%, 5.5% (leading up to PG $118, expectations surrounding dividends paid out of FCF should have, in theory, mirrored expectations surrounding FCF growth).

Base year dividends were $1.28.

Using a two stage model, I get PV of years 1-5 dividends of $6.95, and a terminal dividend $91, for a DDM "intrinsic" value of about $98, using Re = 8%, g yrs 1-5 of 11%, and perpetual g of 6%.

An investor focused on DDM only, using market-based assumptions, would probably have concluded that P&G was over-valued at $118, but under-valued below $98.

And then, in March of 2000, P&G warned.  Here's a link to an article describing the warning at the time.  I remember the day well, P&G lost almost 30% of it's value in one session.  Here's the chart:




So the question comes back to risk, and expectation of future growth.  How is it that investors were tickled pink to pay $118 for P&G in January 2000, and three months later, didn't want to pay more than $54?

Here's the DCF matrix again at $54:

Sum of DFC'sRe =
G = 8%9%10%12%15%Average, growth, blended Re
0,020.7517.4214.7610.766.7614.09
1,022.0717.6915.0211.006.9814.55
2,126.9121.3317.9513.048.3017.51
3,1.530.7523.6519.8214.349.1519.54
4,235.2626.2921.9015.7510.0621.85
5,2.540.6329.3224.2517.3111.0424.51
6,347.1132.8326.9219.0312.0927.60
7,3.555.0936.9629.9820.9313.2331.24
Average, Re, blended growth34.8225.6921.3215.279.70
Current5454545454
Prem/disc to avg55.08%110.24%153.22%253.63%456.55%


A few closing thoughts:

  • In my layperson's opinion, the concept of risk in approaching large caps is a function of investor expectations surrounding both future growth and required return.  In retrospect, most stocks leading up to March 2000 were bid, and P&G was no exception.  Investors had exceedingly high expectations regarding future growth and low expectations regarding required return.  This was a recipe for disappointment.
  • I wonder if investor expectations re: required returns = Re, surrounding widely held large caps ever really gets above 10%.  Professor Aswath Damadoran has an interesting study on historical risk premiums, which put the risk premium on stocks at around 6% back in 1999.  Here's a link.
  • The historical 10 year treasury yield in 1999 was around 4.72%, so in theory, the expected required discount rate should have been close to 11%
  • I personally feel more comfortable modelling and looking for "real time" margin of safety at the top right hand side of my DCF matrix than I do at the bottom left hand side of my DCF matrix.  If I can satisfy myself that there is a theoretical margin of safety under the most conservative growth and discount rate assumptions, then I may be on to something.  Truth be told, I haven't found much yet, but this may be more a function of my only recently starting to think along these lines
  • Even after P&G slid more than 50% between January and April 2000, it still only got down a theoeretical DFC of $55 at G = 7,3.5, and Re = 8%.  It never got to be a $.50 on the $1 situation, and perhaps this is telling me that large caps, unless severerly distressed, don't typically sell much below 1:1 (perhaps a function of large institutional support / money flow) 
  • Looking for a theoertical margin of safety has nothing to do with running a DCF in order to justify current valuation, and everything to do with thinking out of the box in order to try and see if there's a theoertical margin  of safety under the most dire circumstances (i.e., expectations of zero growth, and Re = 12% - 15%)






















No comments:

Post a Comment