Wednesday, March 10, 2021

PepsiCo Valuation, Part 1

This will be the first of a multi-part analysis of PepsiCo. All source data is from Macro Trends. Computed values are in columns with italicized headers.

YearRevenueOper Incomeoper marginNet Incomenet marginShares
2020$70,372$10,08014%$7,12010.1%1,392
2019$67,161$10,29115%$7,31410.9%1,407
2018$64,661$10,11016%$12,51319.4%1,425
2017$63,525$10,27616%$4,8577.6%1,438
2016$62,799$9,80416%$6,32910.1%1,452
2015$63,056$8,35313%$5,4528.6%1,485
2014$66,683$9,58114%$6,5039.8%1,527
2013$66,415$9,70515%$6,74010.1%1,560
2012$65,492$9,11214%$6,1719.4%1,575
2011$66,504$9,63314%$6,4369.7%1,597
2010$57,838$8,33214%$6,31410.9%1,614
2009$43,232$8,04419%$5,94013.7%1,577
2008$43,251$6,95916%$5,13411.9%1,602
2007$39,474$7,18218%$5,64614.3%1,658
2006$35,137$6,50219%$5,63116.0%1,687
2005$32,562$5,98418%$4,06012.5%1,706

Operating margin is very stable. You can see the effect of the acquisitions of two bottlers, The Pepsi Bottling Group, Inc. ("PBG") and PepsiAmericas, Inc. ("PAS"), in 2009 had the effect that one would have expected, to lower operating margins. Net margin is also stable, and also dropped as one would expect from the bottler acquisitions. Net margin is full. There is likely little slack or room for improvement, at least not at a level that would fundamentally alter a valuation.

Revenues are not growing. I'm ignoring any comparison to 2005-2009, which is before the bottler acquisitions. Revenue was $67K in 2011 and $67K in 2019. Operating income is also mostly unchanged over that same period. Net income, however, appears to have benefitted from some improvements, of roughly $1B per year.

The one area where PEP is seeing steady improvement is share count. PepsiCo has done a good job in shrinking the share count since 2005, as can be seen in the following table, with an annual compounded shrinkage rate of 1.35%.

YearEPSShare Holder EquityROEROE w/ outliers removedshare shrink age
2020$5.12$13,55252.5%52.5%1.07%
2019$5.20$14,86849.2%49.2%1.26%
2018$8.78$14,60285.7%0.90%
2017$3.38$10,98144.2%44.2%0.96%
2016$4.36$11,19956.5%56.5%2.22%
2015$3.67$12,03045.3%45.3%2.75%
2014$4.27$17,54837.1%37.1%2.12%
2013$4.32$24,38927.6%27.6%0.95%
2012$3.92$22,39927.6%27.6%1.38%
2011$4.03$20,89930.8%30.8%1.05%
2010$3.91$21,47629.4%29.4%-2.35%
2009$3.77$17,44234.1%34.1%1.56%
2008$3.21$12,58240.8%40.8%3.38%
2007$3.41$17,23432.8%32.8%1.72%
2006$3.34$15,36836.6%36.6%1.11%
2005$2.39$14,25128.5%28.5%

Average ROE is 41.2%, 38.2% if you remove outlier year 2018 from the calculation. Though this has been raised by years of constant share buybacks, ROE was already excellent in 2005 and the overall record indicates that the business is a frugal user of capital assets, despite the inclusion of bottling and distribution operations.

PEP's use of its generated cash is reasonable, though there are some worrisome signs. Dividend policy 15 years ago might have been compatible with a payout of 40% of earnings as dividends, but in recent years payouts have been more in the neighborhood of 75%. 

YearCommon Stock Dividends Paid% of income pd as divid% income avail for reinvestROE * reinvestbookpotential add on EPS
2020$5,50977%23%12%$10$0.61
2019$5,30473%27%14%$11$0.70
2018$4,93039%61%52%$10$4.56
2017$4,47292%8%4%$8$0.12
2016$4,22767%33%19%$8$0.82
2015$4,04074%26%12%$8$0.43
2014$3,73057%43%16%$11$0.67
2013$3,43451%49%14%$16$0.59
2012$3,30554%46%13%$14$0.50
2011$3,15749%51%16%$13$0.63
2010$2,97847%53%16%$13$0.61
2009$2,73246%54%18%$11$0.69
2008$2,54149%51%21%$8$0.66
2007$2,20439%61%20%$10$0.68
2006$1,85433%67%25%$9$0.82
2005$1,64240%60%17%$8$0.41

That leaves a lot less cash available for reinvestment, whether it is for upgrade and expansion, acquisition, or buying back shares. As a check, I've calculated what the retained earnings can support in terms of increments in EPS, assuming that ROE is representative of incremental investment. Of course, acquisitions like Sodastream are very unlikely to generate a 38% return on equity, but internal developments like snack innovation might, in limited cases. The creation of the Flamin' Hot snack line is one example. All is well there, and cash flow can support reinvestment.

PEP is clearly showing signs of maturity. It's stable, and based on my quick look at Coca-Cola Co. financials it is in better shape than KO, but it's hardly posting 8% to 10% gains that you want to see from a major dominant food company. 

Valuation then falls to estimates of revenue growth rate and duration. If we set growth at 3-4%, for 25 years, and a gradual decline of 4% annually thereafter, then using pre-COVID discount rates of about 5%, then we get company valuations of $131 to $150, which are pretty close to its recent trading range. If we use a COVID-era (or extended malaise) discount rate of 3%, then PEP is worth $188 to $217.

The hazard of using these valuations is that so far I've treated PEP as a single monolithic business. Sodas may not do so well as snacks. If we analyze the snack, beverage, and food portions of the business separately, we may get a different answer. We will do that analysis in the next essay.

Tuesday, March 9, 2021

Social Media Toxicity and the Equity Premium

Back in the 1980s, when the efficient market hypothesis was fully believed, Mehra & Prescott in their paper "The Equity Premium: A Puzzle" (1985) analyzed the origins of the rate of return premium that equities command over other asset classes. Their mathematical model predicted a premium of only 0.35%, which was clearly much smaller than the empirically observed 6% to 7% for U.S. stocks in the 20th century. Later, behavioral economists explored psychology-based explanations for the risk premium. They studied the effect that seeing losses frequently, versus seeing them less often, had on experimental subjects. Their findings were that people who see frequent updates of portfolio results tended to be more risk averse. Some these results were published in papers, including for example "The Effect of Myopia and Loss Aversion on Risk Taking: An Experimental Test" (1997) by Thaler, Tversky, Kahneman, Schwartz. Sensing pain from immediate losses has evolutionary survival value. Acting to cut the loss provides a benefit. This is normal if baking a pie, but if investing, it's risk aversion.

The lesson might be applied to other areas of human experience that involve repeated interactions, possible deferral of benefits, and gains and losses. For example, social media. Social media is an investment of reputation, and "clan values". Negative feedback is sensed as a loss, and crisis needing action. "Negative feedback" in this case is not necessarily directly negative feedback, like someone insulting you. It could simply be postings by others that tend to injure your position politically, economically, or religiously. A popular post by an "opponent" then results in a threat to your future return on your investment in society as a whole. In the past, before the advent of social media, interactions with strangers and their opinions about politics and religion was much slower and it took much longer to get feedback from the world. Instead of instant and hourly feedback, checkable at any time, people would get news daily, and they would have to wait for it.

Hence, if we believe Thaler et al, then social media likely causes risk aversion among people who engage heavily with it.

What does that mean? It's known that the excesses of the recent decade, including intoleration of speech on college campuses, claims that people are "injured" by words they don't agree with, and similar social phenomena, may have their origin in social media and the wide availability of the smart phone. John Haidt and Greg Lukianoff explored these ideas in their book Coddling of the American Mind. The degree of damage and its perniciousness was highlighted by the Netflix movie "The Social Dilemma." Risk aversion is manifested by the unwillingness to tolerate the speech or view of others. People view their original commitments and feelings as a baseline that should not be changed.

If this article were about social justice, conformity, and political endangerment of the innocent, it would be much longer. You can go on Twitter and YouTube and find many stories and instances of cancellation that fully justify the reticence of those desiring to avoid losses.

Instead, this article is more about embarking on a cross-disciplinary exploration: If myopic loss aversion explains the equity risk premium, which is a very powerful phenomenon, what other aspects of mass human behavior also fit that psychological model? Social media may just one of several. Although some have viewed social media as a corporate problem, a legal problem, or a technological problem, the problem may be inherent to people. Social media just "wakes up" the perniciousness that was always lurking.

If such is the case, then we already have some political economy theory already waiting to help frame how we can distribute remedies. I return again to Thomas Sowell's A Conflict of Visions (review here). The constrained vision see mankind as, at best, only slightly changeable, and subject to a number of fairly predictable limitations. Political beliefs that go with the constrained vision tend to be those that use long-established rules of thumb and consequent laws; these are usually labeled conservative. Mankind has a long history of living with clans, families, and towns, observing religion, caring for one another. We have a great deal of wisdom (including internal, built-in moral sentiments) for handling these. We have no history of handling high-speed electronic communications avalanches and contentious conversations with strangers. 

Hence, if you have more of the constrained vision as your beliefs about what people can do, you will advocate for limitations on social media until we figure out how to tone it all down and make it less toxic. If you are of the unconstrained vision, and believe people can be changed, then you will likely have faith that people will quickly adapt to our brave new electronic world. That makes the battle to solve social media a political battle as well as psychological, emotional, and technological. The medium itself might even be used to lengthen the arguments so that one side can win.

Since this is a large topic, it will not be solved in a single article. I will likely post other articles later that address parts of what I covered here, including other resolutions to the equity premium puzzle, and other applications of myopic loss aversion. I welcome your comments and thoughts.