Wednesday, July 16, 2014

Possible, Plausible and Probable: Big markets and Networking effects

I do not know Bill Gurley personally, but I do know of him, and I was surprised, sitting in Vienna airport waiting for a connection home on Friday morning, to get an email from him. In the email, he graciously gave me a heads-up that he was planning to post a counter to my Uber valuation and that it would not pull punches. A little while later, I started getting messages from those who had read the post, with some seeking my response and some seeming to view this as the first volley in some valuation battle. I read the post a few minutes later and the first person I wrote to after I read it was Bill Gurley and I told him that I absolutely loved his post, even though it was at complete odds with my assessment of the company, for two reasons.
  1. Like anyone else, I like being right, but I am far more interested in understanding Uber's valuation, and the post provided the vantage point of someone who not only is invested in the company but knows far more about it than I do.  Rather than berating me for not getting "it" (technology,  the new economy, progress) or abusing valuation as a tool from the middle ages, the post focused on specifics about Uber and the basis for its high value. 
  2. In this earlier post of mine, I argued that good investing/valuation is the bridge between numbers and narrative and that neither the numbers nor the narrative people have an automatic right to the high ground. Bill Gurley's post brought home that message by laying out a detailed and well-thought-through narrative, backed up by numbers. 
Mr. Gurley's narrative lends itself well to a more grounded discussion of Uber as a company and I am grateful to him for providing it. As a teacher, I am constantly on the lookout for "teachable moments", even if they come at my expense, and I plan to use his post in my classes.

Dueling Narratives
In my post on Uber's value (and in the Forbes and 538 versions of it), I laid out my narrative for Uber.  I viewed Uber as a car service company that would disrupt the existing taxi market (which I estimated to be $100 billion), expanding its growth (by attracting new users) and gaining a significant market share (10%). The Gurley Uber narrative is a more expansive one, where he sees Uber's potential market as much larger (drawing in users who have traditionally not used taxis and car services)  and much stronger networking effects for Uber, leading to a higher market share. In many ways, this is exactly the discussion I was hoping to have when I first posted on Uber, since it allows us to see how these narratives play out in the  numbers. In the table below, I contrast the narratives and the resulting values:

You can download the valuations by clicking here. (Uber (Gurley) and Uber (Damodaran)).

Given that the values delivered by the narratives are so different, the question, if you are an investor, boils down to which one has a higher probability of being closer to reality. If you had to pick one right now, I think Mr. Gurley's has the advantage over mine for at least three reasons. The first is that  as a board member and insider, he knows far more about Uber's workings than I do. Not only are his starting numbers (on revenues, operating income and other details) far more precise than mine but he has access to how Uber is performing in its test markets (with the new users that he lists). The second is that as an investor in Uber, he has skin in the game, and more at stake than I do and should therefore be given more credence. The third is that he not only has experience investing in young companies, but has been right on many of his investments.

Does that mean that I am abandoning my narrative and the valuation that goes with it? No, or at least not yet, and there are three reasons why. First, it is difficult, if not impossible, for someone on the inside not to believe the best about the company that he directs, the managers he listens to and the products that it offers. Second, an investor in a company, especially one without an easy exit route (at least at the moment), is more attached to his or her narrative than someone who has little to lose (other than pride) from abandoning or altering narratives. Third, as Kahnemann notes in his book on investor psychology, experience is not a very good teacher in investing and markets. As human beings, we often extract the wrong lessons from past successes, don't learn enough from our failures and sometimes delude ourselves into remembering things that never happened. I am not suggesting that Bill Gurley is guilty of any of these sins, but I am, by nature, a cautious convert and I will wait to buy into his narrative, compelling though it may be.

The Acid Test: Probable, Plausible and Possible
As I noted at the start of this post, I liked Bill Gurley's post because it offers a coherent narrative that leads to a higher value. The narrative has two key building blocks and I think that there is much to be gained by taking a closer look at them. The first is that Uber is pursuing a much larger market than just taxi service and that it may very well redefine the nature of car ownership. The second is that Uber will have networking effects that will allow it to capture a dominant market share of this larger market, well above the 10% that I estimated in my original value.  In the sections below, I hope to stress test these assumptions, more as a friendly observer than antagonist.

Market Breakthrough
Companies like Amazon, Google and Netflix owe their success and immense market values to their capacities to redefine markets (retail, advertising and entertainment respectively) and it is true that in these and other cases, investors and analysts have under estimated these capacities and have paid a price for doing so. Unfortunately, it is also true that there have just been many cases where managers and investors have over estimated the capacity to expand markets and lost money in the process.  The Gurley narrative for Uber makes a good case that the convenience and economics of Uber will expand the car service market initially to include light users and non-users (suburban users, rental car users, aged parents and young children), but it does have three key barriers it has to overcome:
  1. Reason to switch: Uber has to provide users with good reasons to switch from their existing services to Uber. For taxi services, the benefits from using Uber are documented well in the Gurley narrative. Uber is more convenient (an app click away), more dependable, often safer (because of the payment system) and sometimes cheaper than taxi service. However, the trade off gets murkier as you look past taxi services. Since mass transit will continue to be cheaper than Uber, it is comfort and convenience that will be the reasons for switching. With car rentals, Uber may be cheaper and more convenient in some senses (you don't have to worry about picking up a rental car, parking it or worrying about it breaking down) and less convenient in others (especially if you have multiple short trips to make). With suburban car service (the aged parents, the dating couple and school bound kids), the problem that Uber may face is that a car is usually more than just a transportation device. Any parent who has driven his or her kids to school will attest that in addition to being a driver, he or she has to play the roles of personal assistant, private investigator, therapist and mind reader. As for date nights, whether Uber succeeds will be largely a function of how much the car itself is an integral part of the date, especially with younger couples.
  2. Overcome inertia: Even when a new way of doing things offers significant benefits, it is difficult to overcome the unwillingness of human beings to change the way they act, with that inertia increasing with how set they are in their ways. It should come as little surprise that Uber has been most successful with young people, not yet set in their ways, and that it has been slower to make inroads with older users. That inertia will be an even stronger force to overcome, as you move beyond the car service market. The articles that point to young people owning fewer cars are indicative of larger changes in society, but I am not sure that they can be taken as an indication of a sea change in car ownership behavior. After all, there have been almost as many articles on how many young people are moving back in with their parents, and both phenomena may be the results of a more difficult economic environment for young people, who come out of college with massive student loans and few job prospects.
  3. Fight off the status quo: The empire, hobbled and inefficient though it may be, will fight back, since there are significant economic interests at stake. As both Uber and Lyft have discovered, taxi service providers can use regulations and other restrictions to impede the new entrants into their businesses. Those fights will get more intense as car rental and car ownership businesses get targeted.
In summary then, the difference in market size in the narratives boils down to a simple calculus of what is probable, what is plausible and what is possible, a distinction that to me is at the center of value:

Not everything that is possible is plausible, and not all plausible opportunities make the transition to the probable.  As I see it, the divergence between the my narrative and Bill Gurley's are captured in where we draw the lines between the probable and the plausible and the value that we attach to the possible. At the risk of mischaracterizing Mr. Gurley's thoughts, I have tried to contrast these differences:


Here again, Bill Gurley has two advantages to work with. The first is that as an investor and insider, he has access to information on Uber's experiences and experiments in its frontier markets (mass transit and suburban users), that may have led him to shift these markets from the plausible to the probable. The second is that as a board director and advisor to management, he is in a position to influence Uber's potential in these markets. For all we know, the Uber Momcar and the Uber Datecar have been already conceived, market tested and are ready to go.

I think Bill Gurley and I agree on the car ownership market  more than we disagree. I see it as a possibility right now and attach an option value of about $2-3 billion to it, partly because it is in the more distant future and partly because Uber's business model in this market is unformed. From Bill Gurley's description of the market, I think he sees it as a possibility as well, though I think he attaches a larger value to it than I do. The reason for the higher value is that it is a conditional possibility, with the likelihood of it happening increasing with the success that Uber has in the car service market. 

Network Benefits
The second part of the Gurley Uber narrative rests on the company having network benefits that allow it to capture a dominant market share. As Mr. Gurley notes, a networking effect shows up any time you, as a user of a product or service, benefit from other people using the same product and service. If the networking effect is strong enough, it can lead to a dominant market share for the company that creates it and potentially to a  ‘winner take all’ scenario. The arguments presented in his post for the networking effects, i.e., pick up times, coverage density and utilization, all seem to me to be point more to a local networking effect rather than a global networking one. In other words, I can see why the largest car service provider in New York may be able to leverage these advantages to get a dominant market share in New York, but these advantages will not be of much use in Miami. There are global networking advantages listed, such as stored data that can be accessed by users in a new city and partnerships with credit car, smartphone and car companies, but they seem much weaker.

In fact, if the local networking advantages dominate, this market could very quickly devolve into a city-by-city trench warfare among the different players, with different winners in different markets. Thus, it is possible that Uber becomes the dominant car service company in San Francisco, Lyft in Chicago and a yet-to-be-created company has the largest market share in London.  For the Gurley Uber narrative to hold, the global networking advantages have to become front and center and here again, it is possible that I am unaware of a management initiative designed to do exactly this. 

The Verdict Awaits

I know that this may be hard to believe but I have less of an interest in making the case that Uber is over priced than I am in understanding what it is that drives its value. I have learned a great deal about why Bill Gurley is so excited about the company but I am inherently cautious, not because I don’t find his arguments to be plausible, but because I have seen how often the plausible does not make the transition to the probable and how frequently the probable fails to show up in the actuals. That quality may make me a bad venture capitalist but I am sure that there are plenty of good ones out there to take up the slack.

Tuesday, June 24, 2014

Numbers and Narrative: Modeling, Story Telling and Investing

When I put together the outline for my very first valuation class in 1986, I was warned by a senior faculty member not to go down that path. I was told that there was really not enough theory in valuation to warrant a class and that I would end up teaching a glorified accounting class. I chose to ignore that advice and I have not regretted it since, for two reasons. The first is that I love teaching a subject where there is little theory, the questions are entirely about practice, you draw on a unique blend of skills and tools to accomplish your tasks and the market acts as your task master.  The second is that I have learned almost everything I know about valuation and more importantly, how much I don't know, in the process of teaching this class. This post is about one of the recurring themes in my class which is the interplay between narratives and numbers that makes for a good valuation.

The Numbers Game
When most people think about valuation, they generally visualize dense financial statements and elaborate excel spreadsheets, and those coming into my valuation class are no exception. They expect me to immerse them in accounting rules and the building of models and are either deeply disappointed, if their background is in accounting or banking, or relieved, if it is not, to find out that the only thing I know about accounting rules is that there lots of them and that I am not an Excel Jedi Master. Don't get me wrong! I do draw on accounting statements for my information and use Excel incessantly, but here is how I see their place in valuation:
  1. Accounting statements provide me with the raw material for my valuation, nothing more and nothing less. Like all raw material, I have to decide what I will use and what I will discard, and I discard far more than I use. As the end user of the raw material, I get to determine what makes sense for me and what does not and not GAAP, IFRS or the accounting profession. As for fair value accounting, I am sympathetic with the motives (which is to make accounting more relevant) but unimpressed with the results. To me, fair value accounting estimates are like microwave frozen dinners, quick and convenient, but you will never mistake them for the real thing.
  2. Excel (or Numbers) is a versatile and powerful multi-purpose tool, but like all tools, it can be misused or over used. My knowledge of Excel is limited to those parts of it that help me complete my valuation and I frankly have no interest in expending time and resources mastering the parts that I can get done with simpler tools or none at all.
So, why do so many appraisers and analysts emphasize their mastery (at least in their minds) of the numbers side of valuation? The answer, I think, lies in the trifecta of illusions that go with numbers-based models.
  1. The illusion of precision: For better or worse, we seem to feel better about uncertain outcomes, if we can attach numbers (expected values, risk adjusted discount rates) to them. That, by itself,  is healthy but what is unhealthy is the belief that quantifying risk somehow makes it dissipate. 
  2. The illusion of objectivity: I believe that all valuations are biased, with the only questions being how much bias and in what direction. That is because we bring in our preconceptions and beliefs about companies into our valuations and we sometimes add to the bias because we have other agendas at play. Here again, analysts point to numbers as their defense against the bias charge, with the implicit argument that numbers don't lie, when the most effective way to shade the truth is with a selective use of numbers.
  3. The illusion of control: I believe that "numbers people" often use numbers to intimidate "non-numbers people" into mute acceptance. The intimidation factor is dialed up by adding more detail  (500 line items, anyone?) and buzz words (free cash flow, a few greek alphabets and a host of acronyms) to your valuations.
In my view, there are at least three significant dangers, when numbers are used without any narrative (or story line) in constructing valuations. First, valuations become plug-and-point exercises, tools to advance sales pitches or confirm pre-conceived values. Second, if a valuation is built around line items and individual inputs, there is a strong possibility that you may be creating a business that can exist only in spreadsheet nirvana, where revenues double every year, margins expand without challenge and growth comes without significant reinvestment. Finally, discussions and debates about inputs become shallow exercises in quibbling about the "right" values to use, with no logical tie breaker.

The Narrative as Valuation
If one extreme of the numbers/narrative spectrum is inhabited by those who are slaves to the numbers, at the other extreme are those who not only don't trust numbers but don't use them. Instead, they rely entirely on narrative to justify investments and valuations. Their motivations for doing so are simple.
  1. Story telling is a powerful attention getter/keeper: Research in both psychology and business point to an undeniable fact. Human beings respond better to stories than to abstractions or numbers, and remember them for longer. After all, the Harvard Business School has taken story telling almost to an art form with its cases, tightly wound narratives that are supposed to convey larger lessons.
  2. Unrestrained creativity: "Creative" people through the ages have always fought back against any restraints on their creativity, especially those imposed by those that they view as less imaginative than they are. 
  3. The Creative Superiority Complex: Just as numbers people intimidate with mounds of numbers, good narrators can browbeat "bean counters" with superior story telling, especially if they can back their stories up with personal experience. 
Narrative-driven investing is not uncommon, especially with younger firms and start-ups, and I have been taken to task for even trying to value these companies using number-driven models. Paraphrasing some of the comments on my valuations of Twitter and Uber, the argument seems to be that while cash flow based valuations may work on Wall Street and with mature companies, they are not useful in analyzing the type of companies that venture capitalists look at. While it is true that rigid cash flow based models will not work with companies where promise and potential are what is driving value, staying with just narrative exposes you to two significant risks. The first is that, without constraints, creativity can carry you to the outer realms of reason and into fantasy. While that may be an admirable quality in a painter or a writer, it is a dangerous one for an investor. The second is that, when running a business as a manager or monitoring it as an investor, you need measures of whether you are on the right path, no matter where your business is in its life cycle. When narrative alone drives valuation and investing, there are no yard sticks to use to see whether you are on track, and if not, what you need to do to get back on the right path. 

Numbers plus Narrative
If numbers without narrative is just modeling and narrative without numbers is story telling, the solution, as I see it, is both obvious and difficult to put into practice. In a good valuation, the numbers are bound together by a coherent narrative and story telling is kept grounded with numbers. Implementing this solution does require work and I would suggest a five-step process, though I am not rigid about the sequencing.

Step 1: Develop a narrative for the business that you are valuing or considering investing in: Every business has a story line and the place to start a valuation is with that narrative. While managers and founders get to present their narrative first, and some of them are more persuasive and credible than others, you and I have to develop our own narratives, sometimes in sync with and sometimes at odds with the management story line. As an example, in my valuation of Uber, my narrative was this: Uber is an innovative car service company, with the untested potential to expand into other logistics businesses. It will expand the car service business (by attracting new users), while gaining a significant (though not dominant) market share and preserving its profitability.  The counter narrative that some of you presented is the following (and I am paraphrasing): Uber is a logistics company that will find a way to expand its profitable car service business model into the moving, car rental and electric car businesses.

Step 2: Test the narrative against history, experience and common sense: This is the stage at which you put your narrative through a reality test and examine whether it withstands multiple tests. The first is the test of history, where you look at the past to see if there have been companies that have lived the narrative that you are claiming for your company and what they share in common.  The second is the test of experience, where you draw on investments based upon similar narratives that you have made in the past and remember or recognize road bumps and barriers that you ran into in practice. The third is the test of common sense, where you draw on first principles in economics and mathematics, to evaluate your narrative's weakest links. With Uber, here is how I justified my narrative. Uber will be able to gain (10%) is that the car service (taxi and limo) business is a splintered, regulated and inefficient business that is ripe for disruption. The reason I did not assume a dominant market share for Uber (40% or 50%) is because I don't see as large a networking effect in the car service business, where the service is both physical and localized, as there are in online technology businesses (search, merchandising or advertising). At the same time, I am assuming that Uber will be able to preserve its profitability in the face of competition and overcome regulatory hurdles.

Step 3: Convert key parts of the narrative into drivers of value: Ultimately, even the most gripping narratives have to show up in the numbers. While this may seem like an insurmountable obstacle to those without a valuation background, it can be simplified by looking at the big picture. Here is my attempt to connect different narratives with key value drivers:
Narratives and Value Drivers
Step 4: Connect the drivers of value to a valuation: I use discounted cash flow models (DCF) to connect the drivers of value to value, because I am comfortable with the mechanics of these models. It is a tool that not everyone is comfortable with and you may find a different and perhaps better way to connect value drivers to value. In fact, the classic VC valuation takes a short cut by using three drivers of value: an expected earnings (or revenue) in a future period, an exit multiple (based on what others seem to be willing to pay today for similar companies) that converts that number into a future value and a target return to discount that value back to the present (and adjust for risk). To those of you who have never done valuation before, trust me when I say that valuation at its core is simple and that anyone should be able o do it. If you don't believe me, you are welcome to try my online valuation class on iTunes U. It comes with a money back guarantee.

Step 5: Keep the feedback loop open: My kids and spouse are quick to remind me that the three words that I find most difficult to say are "I was wrong" and I am sure that I am not alone in my reluctance. The biggest enemy that we (whether numbers or narrative driven) face is hubris, where we get locked into our initial points of views and view changing our minds as a sign of weakness. While it does not come easily to me, I do try to stay open to the possibility that as events unfold, my narrative will change or even shift, sometimes dramatically. With Uber, if the next few months bring evidence of tangible success of the business model in other logistics markets, I will change my story, expand the potential market and with it, the value. If, in contrast, the company gets bogged down in regulatory and legal fights in its existing car service markets or a competing service improves its offering dramatically, I will have to dial down my optimism, reduce both market share and profit margins and change value. In either case, I will view these changes as part of investing rather than as a failure in my initial valuation.

In my experience, it is easiest to play to your strengths (which, for me, are on the numbers side), but you will gain the most when you work on your weaknesses (which, for me, are on the narrative side). Consequently, I learn more from listening to those who think differently from me and disagree with me, even if they do not always do so constructively, than I do from those who agree with me. On my Uber valuation, the comments that I found most useful in fine tuning my valuation were those that I heard from those in the venture capital and technology space. After telling me that I had no idea what I was talking about and that "DCF won't work for these companies", they then proceeded to give me ideas that I incorporated into my DCF valuation. Here, for instance, are my attempts to quantify four of the most common narratives I heard about Uber, and the consequences for value.

Narrative
Total Market
Market Share
Uber Cut
Cost of capital
Failure Probability
Value for Uber
Car service company, facing significant competitive and regulatory hurdles, forced to make trade off of lower profitability for market share.
$100 billion
10%
10%
12%
10%
Car service company with potential to expand into other logistics markets, significant market share, sustained profitability (Mine)
$100 billion
10%
20%
12% ->8%
10%
$5.9 billion + $2-3 billion for disruption option
Car service company with dominant market share (from networking effects) and sustained profitability (New York Times)
$100 billion
50%
20%
12% ->8%
0%
Logistics company with expansion of car service business model into other logistics businesses, while preserving profitability.
$600 billion
5%
20%
12% ->8%
0%

There are two points I hope to make with this exercise. First, even the most imaginative and far-reaching narratives can and should be converted into numbers. So, let's retire the argument that some companies cannot be valued. Second, big differences in valuation almost always result from differing narratives about companies, not disagreements about the "small stuff".

Finally, since this is a discussion of how best to marry narrative to numbers, I cannot pass the opportunity to plug Shark Tank, one of my favorite shows, where narrative (from those pitching their businesses) meets numbers (from the venture capitalists/investors who challenge the business models while bidding on them), generating both drama and humor. 

Implications
If you view value as narrative overlaid with numbers, there are implications for both the founders/managers of businesses and the investors in these firms. To attract capital, managers need to develop coherent narratives about the firms that they run, convey these narratives to investors/markets effectively, and act consistently. To manage that capital well, they need to  identify value drivers, set yard sticks that measure how the narrative is unfolding and change in response to unforeseen events, both positive and negative.

For investors, the lessons are just as profound. They need to find companies that have compelling narratives, convert these narratives into value and make sure that they are not paying too much.  They need to spread their bets across several good narratives and be open to changes in narratives and numbers. It is true that having a great narrative and the numbers to back them up is not a guarantee of investment success. The best laid plans of mice and men can go to waste, but to not plan at all will guarantee that waste.

Attachments:
Uber: A Challenged Car Service company 
Uber: A Successful Car Service company 
Uber: A Car Service company with networking effects
Uber: A Logistics company

Monday, June 16, 2014

Bubble, Bubble, Toil and Trouble: The Costs and Benefits of Market Timing

If you believe that the stock market is in a bubble, you have lots of company. You have long-time market watchers, the New York Times and even a Nobel Prize winner in your camp. But what exactly is a bubble? How can you tell if you are in one?  And if you do believe you are in a bubble, what is your best course of action? Not only are these questions difficult to answer, but the answers can vary across markets, investors and time. 

The Bubble Machine
Every market has a bubble machine, though it is less active in some periods than others, and that machine creates an ecosystem of metrics and experts, as well as warnings about bubbles about to burst, corrections to come and actions to take to protect yourself against the consequences. In periods like the current one, when the bubble machine is in over drive and you are confronted by "bubblers" with varying credibilities, motives and methods, you may find it useful to first categorize them into the following groups.
  1. Doomsday Bubblers have been warning us that the stock market is in a bubble for as long as you have known them, and either want you to keep your entire portfolio in cash or in gold (or bitcoins). They remind me of this character from Winnie the Pooh and their theme seems to be that stocks are always over valued.
  2. Knee Jerk Bubblers go into hibernation in bear markets but become active as stocks start to rise and become increasingly agitated, the more they go up. They are the Bobblehead dolls of the bubble universe, convinced that if stocks have gone up a lot or for a long period, they are poised for a correction.
  3. Armchair Psychiatrist Bubblers use subtle or not-so-subtle psychological clues from their surroundings to make judgments about bubbles forming and bursting. Freudian in their thinking, they are convinced that any mention of stocks by shoeshine boys, cab drivers or mothers-in-law is a sure sign of a bubble.
  4. Conspiratorial Bubblers believe that bubbles are created by small group of evil people who plan to profit from them, with the Illuminati, hedge funds, Goldman Sachs and the Federal Reserve as prime suspects. Paranoid and ever-watchful, they are convinced that stocks are manipulated by larger and more powerful forces and that we are all helpless in the face of this darkness.
  5. Righteous Bubblers draw on a puritanical streak to argue that if investors are having too much fun (because stocks are going up), they have to be punished with a market crash. As the Flagellants in the bubble world, they whip themselves into a frenzy, especially during market booms.
  6. Rational Bubblers uses market metrics that are both intuitive and widely used, note their divergence from historical norms and argue for a correction back to the average. Viewing themselves as smarter than the rest of us and also as the voices of reason, they view their metrics as infallible and mean reversion in markets as immutable.
There are three things to keep in mind about bubblers. The first is that bubblers will receive disproportionate attention in the media, for the same reasons that a reality show about a dysfunctional family will have higher ratings than one about a more normal family. The second is that even the most misguided bubblers will be right at some point in time, just as a broken clock is right twice every day. The third is that being right is often the worst thing that can happen to bubblers, because it seems to feed into the conviction that they are always right and leads to increasingly bizarre predictions. It is no coincidence that every market correction in history has created its gurus (who called that correction right) and those gurus have almost always found a way to discredit themselves ahead of the next one.


Defining a Bubble
What is a bubble? The lazy definition is that any time you see a large market correction, it is the result of a bubble bursting, but that is neither a useful definition, nor is it true. To me, a bubble reflects a market disconnect from fundamentals, where prices go up steeply, with no help from the fundamentals. The best way of illustrating this is to go back to an intrinsic value model, where the value of stocks can be written as a function of three fundamentals: the base year cash flows that investors are receiving, the expected growth in these cash flows and the risk in the cash flows:


If cash flows increase, growth rates surge, risk free rates drop or macroeconomic risk subsides, stocks should go up, and sometimes steeply, and there is no bubble.  At the other extreme, if stock prices go up as cash flows decrease, growth rates become more negative and risk free rates and equity risk increase, you have a bubble. It is far more likely, though, that you will be faced with a more ambiguous combination, where shifts in one or more fundamentals (higher growth, higher cash flows, a lower risk free rate or lower macroeconomic risk) may explain the increase in stock prices and you will have to make judgments on whether the increase is larger than warranted. 

Detecting a Bubble
The benefits of being able to detect a bubble, when you are in in its midst rather than after it bursts, is that you may be able to protect yourself from its consequences. But are there any mechanisms that detect bubbles? And if they exist, how well do they work?

a. PE and variants
The most widely used metric for detecting bubbles is the price earnings (PE) ratio, with variants thereof that claim to improve its predictive power. Thus, while the conventional PE ratio is estimated by dividing the current price (or index level) by earnings in the last year or twelve months, you could consider at least three modifications. The first is to clean up earnings removing what you view as extraordinary or non-operating items to come up with a better measure of operating earnings. In 2002, in the aftermath of accounting scandals, S&P started computing core earnings for US companies which can differ from reported earnings significantly. The second is to average earnings over a longer period (say five to ten years) to remove the year-to-year volatility in earnings. The third is to adjust the earnings from prior periods for inflation to get a inflation-consistent or real PE ratio. In fact, Robert Shiller has a time series of PE ratios for US stocks stretching back to 1871, that uses normalized, inflation-adjusted earnings.

In the graph below, I report on the time trends between 1969 and 2013 in four variants of the PE ratios, a PE using trailing 12 month earnings (PE), a PE based upon the average earnings over the previous ten years (Normalized PE), a PE based upon my estimates of inflation-adjusted average earnings over the prior ten years (My CAPE) and the Shiller PE. 

Normalized PE used average earnings over last 10 years & My CAPE uses my inflation adjusted normalized earnings. Shiller PE is as reported in his datasets
While the Shiller PE has become the primary weapon wielded by those who believe that we are in a bubble, perhaps because of the pedigree of its creator,  the reality is that all four measures of PE move together much of the time, with a correlation of close to 90%. (If you are wondering why my time series starts in 1969, I use the S&P 500 and earnings on the index and I was unable to get reliable numbers for the latter prior to 1960. Since I need ten years of earnings to get my normalized values, my first estimates are therefore in 1969.)

To examine whether any of these PE measures do a good job of predicting future stock returns and thus market crashes, I computed the correlation of each PE measure with annual returns on the S&P 500 over one-year, two-year and three-year periods following the computation.
T statistics in italics below each correlation; numbers greater than 2.42 indicate significance at 2% level
First, the negative correlation values indicate that higher PE ratios today are predictive of lower stock returns in the future. Second, that correlation is weak with one-year forward returns (notice that none of the t statistics are significant), become stronger with two-year returns and strongest with three-year returns. Third, there is little in this table to indicate that normalizing or inflation adjusting the PE ratio does much in terms of improving its use in prediction, since the conventional PE ratio has the highest correlation with returns over time periods

Defenders of the PE or one its variants will undoubtedly argue that you don't make money on correlations and that the use of PE is in detecting when stocks are over or under price. For instance, one rule of thumb suggests that a Shiller PE above 15 would indicate an over valued market, but that rule would have kept you out of US equities since 1988. To create a rule that is more reflecting of the 1969-2013 time period, I computed the 25th percentile, the median and the 75th percentile of each of the PE ratio measures for this period.
PE measures: 1969-2013
I then broke my sample down into four quartile classes with each PE ratio, from lowest to highest, and computed the annual stock market returns in the years following:
One-year and Two-year stock returns
The predictive power improves for PE ratios with this test, since returns in the years following high PE ratios are consistently lower than returns following low PE ratios. Normalizing the earnings does help, but more in detecting when stocks are cheap than when they are expensive. Finally, the inflation adjustment does nothing to improve predictive returns.

Note, though, that this test is biased by the fact that the quartiles were created using data from the period on which the test is run. Thus, the conclusion that you can draw from this table is that if you had known, in 1969, what the distribution of PE ratios for the S&P 500 would look like for the next 45 years (which would suggest amazing foresight on your part), you could have made money by buying when PE ratios were in the bottom quartile of the distribution and selling in the top quartile.

b. EP Ratios and Interest Rates
One of the biggest perils of using the level of PE ratios as an indicator of stock market pricing, as we have in the last section, is that it ignores the level of interest rates. If  interest rates are lower, PE ratios should be higher and ignoring that relationship will lead us to conclude far too frequently (and erroneously) that stocks are over priced in low-interest rate environments. The link between PE ratios and interest rates is best illustrated by looking at how the EP ratio (the inverse of the PE ratio) moves with the T.Bond rate over time. In the figure below, I graph the movements of all four variants of EP ratios as the T.Bond rates changes between 1969 and 2013:

It is clear that EP ratios are high when interest rates are high and low when interest rates are low. In fact, not controlling for the level of interest rates when comparing PE ratios for a market over time is an exercise in futility.

This insight is not new and is the basis for the Fed Model, which looks at the spread between the EP ratio and the T.Bond rate. The premise of the model is that stocks are cheap when the EP ratio exceeds T.Bond rates and expensive when it is lower. To evaluate the predictive power of this spread, I classified the years between 1969 and 2013 into four quartiles, based upon the level of the spread, and computed the returns in the years after (one and two-year horizons):


The results are murkier, but for the most part, stock returns are higher when the EP ratio exceeds the T.Bond rate.

c. Intrinsic Value
Both PE ratios and EP ratio spreads (like the Fed Model) can be faulted for looking at only part of the value picture. A fuller analysis would require us to look at all of the drivers of value, and that can be done in an intrinsic value model. In the picture below, I attempt to do so on June 14, 2014:

Intrinsic valuation of S&P 500: June 2014
It is true that this intrinsic value is a function of my assumptions, including the growth rate and the implied equity risk premium. You are welcome to download the spreadsheet and try your own variations.


If your concern is that I have used too low an equity risk premium, you can solve, as I do at the start of each month, for an implied equity risk premium (by looking for that equity risk premium that will give you the current index level) and then comparing that value to historical values for that input:


The current implied ERP of 4.99% is well above the historic average and median and it clearly is much higher than the 2.05% that prevailed at the end of 1999.

Are we in a bubble?
In the table below,  I summarize where the market stands today on each of the metrics that I discussed in the last section:

If you focus on PE ratios, it is true the current levels in the market put it in the danger zone, given past history. However, bringing the level of interest rates into the measure (in the EP spreads) reverses the diagnosis, since stocks look under valued on these measures. Finally, expanding the assessment to look at growth and risk as well in the intrinsic value and ERP measures reinforces suggests that stocks are fairly valued. 

While there are some who are adamant in their belief that the market is in a bubble, I remain unconvinced, especially given the level of rates today. To those who argue that earnings could drop, growth could turn negative, interest rates could go up or that there could be another global crisis lurking around the corner, has there ever been a point in time in stock market history where these concerns have not existed? And even if they do exist, the reason we demand an equity risk premium in the first place is for the uncertainty that we feel about macroeconomic variables driving value.


Bubble Belief to Bubble Action: The Trade Off
While I believe that the risk that we are in a bubble is over stated by PE ratio comparisons, you may come to a very different conclusion. Even if you do, though, should you act on that belief? The answer is not clear cut, since there are two ways you can respond to a bubble. The first, which I will term the passive defense, is to reduce the amount of your portfolio allocated to equity to a lower number than you would normally hold (given your age, liquidity needs and risk aversion). The second which I term the active defense is to try to profit off the market correction by selling short (or buying puts). The trade off is then between the cost and the benefit of acting:
  • The cost of acting: If you decide to act on a bubble, there is a cost. With the passive defense,  the money that you take out of equities has to be invested somewhere safe (earning a risk free rate, or something close to it) and if the correction does not happen, you will lose the return premium you would have earned by investing stocks. With an active defense, the cost of being wrong about the correction is even greater since your losses will increase in direct proportion with how well stocks continue to do. (Note that using derivatives to protect yourself against market corrections or for speculation will deliver variants of these defenses.)
  • The benefit of acting: If you are right about the bubble and a correction occurs, there is a payoff to acting. With the passive defense, you protect your investment (or at least that portion that you shift out of equities) from the drop. With the active defense, you profit from the drop, with the magnitude of your profits increasing with the size of the correction.
The trade off then becomes a function of three variables: how certain you feel about the existence of a  bubble, how big a correction you see occurring as a result of the bubble bursting and how soon you see the correction coming.

To illustrate the trade off, consider a simple (perhaps simplistic) scenario, where you are fully invested in equities and believe that there is 20% probability of a  market correction (which you expect to be 40%) occurring in 2 years. In addition, let's assume that the expected return on stocks in a normal year (no bubble) is 7.51% annually and that the expected annual return if a bubble exists will be 9% annually, until the bubble bursts. In the table below, I have listed the payoffs to doing nothing (staying 100% in equities) as well as a passive defense (where you sell all your equity and go invest in a  risk free asset earning .5%) and an active defense (where you sell short on equities and invest the proceeds in a risk free asset):
Future value of portfolio in 2 years (when correction occurs)
If you remain invested in equities (do nothing), even allowing for the market correction of 40% at the end of year 2, your expected value is $1.0672 at the end of the period.  With a passive defense, you earn the risk free rate of 0.5% a year, for two years, and the end value for your portfolio is just slightly in excess of $1.01. With an active defense, where you sell short and invest int he risk free rate, your portfolio will increase to $1.3072, if a correction occurs, but the expected value of your portfolio is only $0.9528, which is $0.1144 less than your do-nothing strategy.

If you feel absolute conviction about the existence of a bubble and see a large correction coming immediately or very soon, it clearly pays to act on bubbles and to do so with an active defense. However, that trade off tilts towards inaction as uncertainty about the existence of the bubble increases, its expected magnitude decreases and the longer you will have to wait for the correction to occur. I know that I am pushing my luck here but I tried to assess the trade off in a spreadsheet, where based upon your inputs on these variables, I estimate the net benefit of acting on a bubble for the passive act of moving all of your equity investment into a risk free alternative:
Payoff to Passive Defense against Bubble (Correction of 40% in 2 years)
The net payoff to acting on a bubble generates positive returns only if your conviction that a bubble exists is high (with a 20% probability, it almost never pays to act) and even with strong convictions, only if the market correction is expected to be large and occur quickly.

On a personal note, I have never found a metric or metrics that  allow me to have the combination of conviction that a bubble exists, that the correction will be large enough and/or that the correction will happen within a reasonable time frame, to be a market timer. Hence, I don't try! You may have a better metric than I do and if it yields more conclusive results than mine, you should be a market timer.

Bubblenomics: My perspective
It is extremely dangerous to disagree with a Nobel prize winner, and even more so, to disagree with two in the same post, but I am going to risk it in this closing section:
  1. There will always be bubbles: Disagreeing with Gene Fama, I believe that bubbles are part and parcel of financial markets, because investors are human.  More data and computerized trading will not make bubbles a thing of the past because data is just as often an instrument for our behavioral foibles as it is an antidote to them and computer algorithms are created by human programmers.
  2. But bubbles  are not as common as we think they are: Parting ways with Robert Shiller, I would propose that bubbles occur infrequently and that they are not always irrational. Most market corrections are rational adjustments to real world shifts and not bubbles bursting and even the most egregious bubbles have rational cores.
  3. Bubbles are more clearly visible in the rear view mirror: While bubbles always look obvious in hindsight, it is far less obvious when you are in the midst of a bubble. 
  4. Bubbles are not all bad: Bubbles do create damage but they do create change, often for the better. I do know that the much maligned dot-com bubble changed the way we live and do business. In fact,  I agree with David Landes, an economic historian, when he asserts that  "in this world, the optimists have it, not because they are always right, but because they are positive. Even when wrong, they are positive, and that is the way of achievement, correction, improvement, and success. Educated, eyes-open optimism pays; pessimism can only offer the empty consolation of being right." In market terms, I would rather have a market that is dominated by irrationally exuberant investors than one where prices are set by actuaries. Thus, while I would not invest in Tesla, Twitter or Uber at their existing prices, I am grateful that companies like these exist.
  5. Doing nothing is often the best response to a bubble: The most rational response to a bubble is to often not change the way you invest. If you believe, as I do, that it is difficult to diagnose when you are in a bubble and if you are in one, to figure when and how it will dissipate, the most sensible response to the fear of a bubble is to not change your asset allocation or investment philosophy. Conversely, if you feel certain about both the existence of a bubble and how it will burst, you may want to see if your certitude is warranted given your metric.