CFA sans BA: The Case for a Barbell Approach to Education

Every summer, roughly two hundred thousand people around the world will take a six-hour multiple-choice and essay-question test organized by the CFA Institute. At a recommended 300 hours of study per test, we’re talking about roughly 60 million person-hours devoted to studying the ins and outs of discounted cash

Every summer, roughly two hundred thousand people around the world will take a six-hour multiple-choice and essay-question test organized by the CFA Institute. At a recommended 300 hours of study per test, we’re talking about roughly 60 million person-hours devoted to studying the ins and outs of discounted cash flows, LIFO and FIFO accounting, the capital asset pricing model, and other such topics. The end goal of most test-takers is to pass all three levels and get designated as a Chartered Financial Analyst. Based on Google Autocomplete, the single best window into the safe-for-work portion of the zeitgeist, people consider the CFA designation a possible alternative or complement to the MBA: it’s something you do after a couple years of school and a couple years of work. However, unlike business school, you don’t strictly need your undergraduate degree to get the designation, so, given my odd background, it was the perfect choice for an academic credential. Low overhead, given the absence of a BA requirement. And a good story, since, based on my cursory Googling, there aren’t any examples out there.

It’s also a way to apply both Tim Ferriss’ “minimum effective dose” framework and Nassim Taleb’s “Barbell Portfolio,” of owning lots of low-risk assets and a small smattering of asymmetric bets. A small dose of extremes is better than a constant trickle of barely-anything.

The CFA exams quiz you on many topics, but one they’ve somehow missed is the economic concept of signaling. A question they might have asked: does an academic credential like a bachelor’s degree or a CFA designation:

a) have no effect
b) confer valuable skills
c) signal valuable traits

Using our standard test-taking skills, we can start by eliminating the obviously wrong answer: clearly college grads and CFA charterholders earn more than other people, so there’s something going on. But which way does the causation run: does studying teach you income-producing skills, or do the sorts of people who have income-producing skills also accumulate credentials?

There’s a lively debate on that topic as it pertains to higher education, with Bryan Caplan playing the role of the 300 Spartans, every college administrator and high school guidance counselor taking on the role of Xerxes’ army. (Thermopylae, the battleground, is the last couple decades of labor market and economics-of-education reserach. The other city-states, waiting in the wings to see how things shake out, are employers.)

Usually, people who write about this will say that Caplan takes an “extreme” position when he argues that the returns to education are mostly from signaling, rather than from skill creation. That’s backwards: the “moderate” view is that education’s measurable impact on incomes is entirely a human capital story — in their view, college makes people qualified for jobs; it doesn’t redistribute jobs to people who signal quality. “Extremists” argue that education is mostly signaling and a little human capital. They only start to sound extreme when they take this very logical notion to its very logical conclusion: we shouldn’t subsidize education so much.

In fact, if anything, we should tax it. In a world where bartenders need degrees, getting a degree redistributes a bartending job; it doesn’t create one. In general, while we should tolerate zero-sum activities, we shouldn’t subsidize them.

The case for the signaling model is beyond the scope of this piece, so I’ll just note that Caplan makes a strong case and his critics make a weaker one. He’s cheating a little bit by saying that his critics are partly-right: if you go to school, you will develop new skills. But his critics are really shooting themselves in the foot by arguing that the signaling component is zero, or at least that we ought to round it down to zero when we decide how much money we as a country should spend on schools.

The “Sheepskin Effect” is the most powerful restatement of the signaling model: students who stop attending school one semester short of their degree have half the wage premium they’d get if they finished it. (Do schools just save the most lucrative lessons for the last couple months?) And students who were wait-listed for elite schools earn about as much if they’re accepted as if they’re rejected. If you go to Harvard Business School, you will read thousands of pages for school, but just two pages account for the vast majority of the wage premium: your acceptance letter and your diploma.

From that perspective, the CFA may be the most virtuous educational credential available today. It’s not subsidized. It doesn’t send mixed signals — you don’t have to learn or burn the great works of the Western Canon. There isn’t a CFA football team (the Closet Put Writers? The Fightin’ Asset Accumulators?)

The way the CFA works is that you sign up, they send you a stack of books roughly the same dimensions and weight as a cinderblock, and then one day in June there’s a test. The CFA institute suggests about 300 hours of studying per test. I recommend dividing 300 hours by the number of days until the test very early.

There are two personality types when it comes to studying: architects and iterators. Architects build a study plan with steady pacing: week 1, read the Fixed Income section. Week 2, equities. Etc. This approach is great if you’re a disciplined planner with no prior exposure to the material, but that doesn’t describe the typical test-taker.

The approach I took was an iterative one: take a practice test. Write down all the concepts you got wrong. Study those. Repeat. If you’ve been working with bonds for the last three years, you will (I hope) score easy 80–90%s on the bond section, so why waste time on it? On the other hand, if you’ve been doing the wrong sort of bond math — maybe you do distressed and the test asks a lot of questions about risk-free rates — you’ll want to know early.

Iterating is only possible with test-prep materials, which is why I bought a bunch of practice tests and flashcards (I chose Kaplan-Schweser and thought it was great, but I’m told the other programs are comparable.)

The cool thing about iterating is the constant feedback. The scary thing about iterating is taking the practice version of a test you need a 70% grade, on, and scoring in the 50s. The week before the test is a blur. Your leisure time suddenly vanished into a whirlwind of last-minute review. Lots of people take off a day or two to study, which I did — but realistically, that’s just twelve hours of thinking about how twelve hours spread out over the last three months would have done a lot more good.

And then: test day! Don’t bother setting an alarm. I got up at a little before 4, slammed the first of the day’s three Venti red-eyes, and I was off. If studying measures chronic time management skills, the test itself is all about acute time-management: how quickly can you identify which questions will take you fifteen seconds and which will be a ten-minute slog? How do you balance between solving more hard questions and reviewing what you’ve already done?

The material itself is not challenging. I wouldn’t feel bad assigning any chunk of the CFA curriculum to an AP high school class. What is challenging is the sheer volume. The CFA is a test of your ability to manage time and memorize.

And you might argue that these are not useful skills. Wrong! Memorization is essential because finance relies so heavily on pattern recognition: your ability to distinguish between things that remind you of 2007 and things that remind you of 2002 is a major factor in whether or not you’ll make good decisions, and the size of your analogy-library is a serious limitation. Have you noticed how many great investors keep investing into their 80s, 90s, and 100s? They stay good at their jobs because they never stop accumulating data.

Time-management is also essential. If you’re modeling a company, you might have a couple dozen variables that you could estimate in each quarter, but being able to drill down to the two or three that matter, and figuring out how to estimate them better, is what counts. (When people talk about “the story” for a particular stock, they mean: what real world uncertainty is going to collapse into certainty in the next few quarters?)

Some study techniques work well for building a sustainable foundation of knowledge: learn a topic, take a test, do some of your own research, present results. Some study techniques work better if you’re optimizing for a test score: read, test, read, test, flip through flash cards for hours the night before and morning of the test. Ironically, both of these modes are, in fact, useful in finance. You do need to develop deep knowledge of the company and the industry you’re investing in. And sometimes, there’s an acquisition — or a lawsuit — or a joint venture — or a sea change in strategy — or a natural disaster — and your job is to become an Instant Expert on something you hadn’t even heard of the week before. And, whaddyaknow, learning enough real estate terminology to tell your AFFO from your elbow so you can pass that section of the exam builds/tests that exact skill.

One bank-shot argument in favor of schools is that part of the capital they build is social capital. The theory goes that if you develop a group of close friends, you have a trusted peer group who you can do business with later, and since the economy runs on trust, it’s net beneficial. It’s not that college is a bad investment because it’s a four- to six-year party — it’s only a good investment because of the parties! (I haven’t seen anyone suggest that we fire the professors and subsidize fraternities directly, but maybe I’ve just missed it.)

The CFA offers something like that, in a surprising way. Most people study alone, so you don’t have a bunch of directly-shared experiences the way you would if you took a challenging class together. But even though you don’t share the experience, you do have a shared experience. When I showed up for the first CFA test, at first I couldn’t find the entrance. Then a saw a really tired-looking guy in a Dartmouth t-shirt, carrying a sheaf of notes. Behind him was a woman rummaging through a Barclays banker bag, looking for a calculator. My tribe!

One topic the CFA covers in great depth is asset allocation. The two key insights here are that:

  1. Within a given asset class, you get paid for taking on non-diversifiable risk. If you move some of your portfolio from cash to equities, the variance and return both go up. You don’t get paid for taking on un-diversifiable risk. If you switch from an index fund to a single stock, your variance goes up, but your expected return stays the same. Assuming markets are efficient, this makes diversification an economic free lunch. Assuming markets aren’t efficient, it gives you a benchmark: you have to ask yourself if you’re a sufficiently talented stock picker to overcome the volatility drag from picking fewer stocks than John Bogle.
  2. When you’re looking at multiple asset classes, it’s possible to achieve a higher risk-adjusted return by diversifying into two assets that don’t correlate with one another, even if it means diversifying into an asset with a worse risk-adjusted return. A levered portfolio of stocks, treasury bonds, and gold outperforms a portfolio of just stocks with the same volatility, since treasuries and gold tend to move in the opposite direction from stocks.

You could defensibly view the CFA-sans-BA strategy as either imprudent un-diversification or prudent diversification across asset classes. On the imprudent side, you’re betting that the traits the CFA signals (intelligence, work ethic, time management) are more important than the ones the BA signals (intelligence, work ethic, time management, conformity). On the other hand, there are a lot of people with BAs, or at least parts-of-BAs. Competing with all of them sounds daunting.

One result of the popularity of this asset-allocation model is the rise of indexing: rather than pay someone to pick stocks, you can pay someone a lot less to pick all the stocks. If you want diversification, you can narrow down your decision — instead of buying an index fund, you can buy an industry-specific fund, or a country-specific fund instead.

While this is smart in theory, indexing leads to some odd distortions. Essentially, it overvalues assets in proportion to how indexable they are. We saw a test case of this in the 1990s — the investor Michael Green argues that the tech boom was, in fact, driven by the rise of index funds: at the time, indices were weighted based on the total shares outstanding for a given company. A sensible choice, which made them better represent the market’s aggregate performance.

This becomes a problem when some companies’ shares are held by a small number of insiders who are unwilling or unable to sell. A recently IPO’d tech company, for example. If index funds are 1% of the market, they’ll buy 1% of the shares outstanding of every company. But if 90% of a given company’s stock is owned by insiders, that 1% is 10% of the freely-trading shares. This caused a healthy run-up in tech stocks, which raised the demand for more tech IPOs — which, since they were newer and younger companies, tended to have more concentrated ownership.

S&P quietly fixed that issue a couple years later, and now indices are weighted by freely-tradable stock, not total shares. But that’s a sketch of the sort of problems that can develop when indexing is a small part of the market.

You can treat hiring on the basis of conventional credentials as a form of indexing: rather than pick out specifically talented employees, an employer can invest in a diversified portfolio of The Boston 2000 or The Ithaca 5000. They won’t do too well buying known quantities at the market price, but they’ll have a good idea of what they’re getting. The strategy performs fine in backtests, at least.

Strategies have a tendency to unwind painfully, especially when people don’t know they’re putting them on. (Witness the cab driver who thought he was buying a biotech stock when he was actually shorting volatility futures.)

If you already have your credentials, market distortions that exaggerate their return suit you juts fine. And in a worst-case scenario, you go from earning a premium to earning a reasonable wage; that’s not a bad deal at all. But if you’re considering which credentials to pursue, it’s incumbent on you to think about which credentials might be overvalued now (in terms of time and effort required to attain them) and to consider your other options. Education is chunky; you can’t construct a portfolio that’s half long Harvard, half long coding bootcamps, and short State U as a hedge.

When you compare a BA to the CFA, you’ll notice the CFA gets you similar exposure to the upside you care about — potentially developing skills and definitely signaling traits — at a lower cost in time and money. Sure, the BA is more conventional, but as the economics of equity asset management demonstrate, the closer you are to conventional the more you’re competing on price.

Education is a long-term bet: do you want to be long an asset that people are systematically encouraged to overpay for (for the moment)? Or do you want to go all-in on something else?