Lessons from Columbia Business School Part 1: You Cannot Outsource Risk Management

The Value Investing program at Columbia Business School was a fantastic experience. I was amazed at how much information we were able to learn. Professor Bruce Greenwald is an insightful and engaging teacher who commanded our rapt attention for 3 full days.

On Day 3 we discussed risk management and portfolio construction. It was here that Professor Bruce Greenwald explained why institutional investors simply can’t outsource risk management. Here’s a summary of Professor Greenwald’s logic.

  • For a fund manager, the risk of active management is under-performing the market and consequently losing clients.
  • From the perspective of an investor who can diversify across multiple fund managers, the risk of active management is paying a fund manager for active investment and getting market returns at a high cost.
  • The two definitions of risk are irreconcilably different. Clients want fund managers to concentrate their investments in a handful of their best ideas. Meanwhile fund managers want to hold portfolios that are more diversified and more closely resemble the market.
  • Unfortunately, institutional investors often try to outsource risk management by awarding benchmark-relative investment mandates.
  • Doing so leads to poor results as this invariably leads to managers holding off-setting positions.
  • In effect, the client pays an active fee on the whole amount invested, while only a small fraction of their investment can truly be considered active.
  • Thus, the true fee paid by the client for active management (fee divided by the amount of the portfolio that differs from the benchmark) is enormous.
  • The only way to avoid this is to stop outsourcing risk management.
  •  This can be done by hiring specialist managers that each concentrate on a different set of investment opportunities (assuming you can find them), instead of multiple fund managers benchmarked against the same market index.
  • The only investors that seem to do this are sophisticated endowments and high net worth individuals.

I know from personal experience that the problems that Professor Greenwald described are much worse in markets where the market capitalization-weighted index is heavily concentrated in a few industry sectors or stocks such as Australia, Scandinavia and Canada. 

His observation about the investment habits of sophisticated endowments is corroborated by Peter Bernstein in his excellent book Capital Ideas Evolving. Bernstein writes of David Swensen, the Chief Investment Officer of the Yale Endowment.

The experience with US equities is a clear example of how Swensen operates. Yale hired no big-name managers in this market, and all of the managers of U.S. equity run specialized portfolios with relatively few holdings. One manager, for example, invests only in energy-related stocks, another only in real estate stocks, another only in biotech, and so forth. Moreover, these portfolios are highly concentrated in only a few stocks; the largest manager tends to hold only five to ten stocks, and at one point was down to three.

The result is huge tracking error against any of the major indexes like the S&P 500 or the Wilshire 5000. “There is no way you can succeed with active management if you try to control benchmark risk” Swensen declares: “You must be willing to deviate from the benchmark if you want to earn returns commensurate with the risks of owning equities. And you must be patient. These managers often lag, but they have done their homework and have no hesitation in just hanging in.”This is not a recipe for smooth returns, and the Yale U.S. equity portfolio has had a sequence of bumpy short-term rides to reach its spectacular long-term performance.

Many institutional investors have been heavily influenced by Yale’s success and have tried to obtain similar results by investing in private equity, property, real assets, hedge funds and other asset classes and strategies that are part of the “endowment model”.

I’ve always wondered why these investors were happy to follow Yale’s lead when it comes to expensive unlisted assets (where they almost certainly don’t have Yale’s competitive advantages), while at the same time ignoring Swensen’s advice on how to create a sensible multi-manager equity portfolio?

It seems to me that it’s because they just can’t bear the pain of short term under-performance in the pursuit of long-term gain (you can read a post on this subject HERE).

Another factor might be the widespread application of the Fundamental Law of Active Management.

The law states that a fund manager’s risk adjusted performance is a function of two variables. The first variable is the portfolio manager ‘skill’ in selecting securities. The second variable is breadth; the number of independent investment opportunities.

In other words, a fund manager can try to beat the market by improving their ability to select stocks, by considering more stocks as part of their investment research process, or by a combination of both. This is fairly straight-forward and makes sense.

Many institutional investors and consultants use this law as basis for recommending fund managers that cover a very wide opportunity set. For example, a consultant may recommend a global equity strategy where a fund manager is benchmarked against an index containing thousands of stocks.

But most people forget that the fundamental law of active management assumes that the quality of information that the fund manager has about each stock is constant.

Is it logical to assume that a fund manager has the same level of knowledge and understanding of each of the hundreds or thousands of stocks that make up the market? Unlikely. *

The reality is that a research analyst or a fund manager can only really cover a small number of companies in detail at a time. For example, James Valentine, former Associate Director of North American Research and Director of Global Training and Development at Morgan Stanley explains in his book Best Practices for Equity Research Analysts: Essentials for Buy-Side and Sell-Side Analysts.

One study showed that buy side analysts who covered fewer than 40 stocks, which was the mean for the sample group, did a better job in identifying the risks than those covering more than 40 stocks. In a study of sell side analysts, those in the bottom 10 per cent for forecasting accuracy covered 21 more firms than analysts in the top 10 percent.

…As a general rule, limit it to 35-50 closely covered stocks for an individual buy-side analyst and 15 to 20 stocks for a three person sell-side team. Analysts new to a sector or the investment process altogether should cut these figures in half. There are exceptions to this rule, but based on my experience, the top 10 to 20 percent of buy-side analysts who knew their stocks well enough to be ahead for the crowd and generate alpha were closely covering fewer than 50 stocks. From my perspective as a research manager, the best sell side analysts were those covering 5 to 7 stocks per team member.

Professor Greenwald seemed to think that even this was too many stocks for one person to cover successfully. He recommended that investors aim to develop specialist knowledge in 3-4 industries; offering Warren Buffett as an example of an investor who’s most successful investment are in a handful of industries (insurance, consumer non-durables, old media and banking) and in only a few stocks.

Even if you cover a handful of industries and companies, it doesn’t necessarily follow that all of them will be attractive investments at any given time. In other words, the number of stocks that a fund manager has genuine conviction in will be relatively few.

You can find out more about why focusing on a small set of stocks in my earlier blog posts.

Please follow my blog (click the button in the bottom right-had corner) if you would like to find out more about what I learned at the Value Investing program at Columbia Business School.

Finally, here’s  an extended  quote by Professor Bruce Greenwald’s taken from his lecture on risk management and portfolio construction.

Ok, what we’re going to talk about next then is building a portfolio. And here you have to understand a crucial difference between people who actually make investments and your clients. You (MF speaking of fund managers) have reputational and other risks they (MF institutional investors) have their wealth at risk.

They all invest in multiple managers, that means they are diversified across managers. That means in turn they’re not particularly interested in individual managers being diversified. They want you, especially the big smart endowments and the big smart family offices, they want to be specialized and concentrated. They are perfectly happy to have you hold 5 stock portfolios in one industry or two industries or three industries.

That is an unhappy situation for you because it exposes you to a lot of risk. There is always going to be that conflict if you’re managing money for other people and you are going to have to manager that.

On the one hand everybody says “Oh you just have to serve your clients” if you do that, that is in some sense taking on enormous risk for yourself. It’s a kind of suicide pact on your future because sooner or later you, if you’ve got a three, four or five stock portfolio in one industry it’s going to blow up. You’re going to be out of business, it’s going to be 4% of their portfolio and they’re not going to care.

You have to be aware of that conflict and you have to manage the balancing act between risk for yourself and risk for your clients.

If you are a client, the first thing you have to understand is you actually cannot delegate risk management. I’m going to tell you a story in a slightly different context but I hope you’ll see the implications for risk management.

The very first job that I did in Finance was at AT&T. I was working at Bell Labs and we were asked to look at the AT&T Pension Fund. At the time, it was a company that was one of the biggest companies in the world, they employed 1.1 million workers. They had 200 portfolio managers, each of whom operated independently.

This is back in 1979 when fees are very significant. So the first question we simply asked was: in any given week, how much of what one manager was buying was another manager selling? And what do you think that number was?

It was slightly over 90%.

That’s exactly what you’d expect with 200 managers. They are going to be replicating market sentiment and trading around it. And the reason it was so low, was that every year AT&T was putting more money into these pension portfolios, so they were net buyers as a whole so they couldn’t completely cancel each other out.

So the first thing we made them do was submit their orders to a facility, we time dated the orders so they got credit for the price when they wanted to sell or buy it, and then we only executed the net orders at the end of the week and that was an enormous saving.

Of the net orders – because we then knew what the net orders were – what percentage of it do you think was just buying the market?

93%, not quite 100%.

Because, you know, when you’ve got 200 fund managers, some of them are going to be optimistic about some sectors; some are going to be pessimistic about other sectors. When you average over 200 managers you are going to be basically buying the market.

So they were paying 50-70 points in fees for a lot of activity that cancelled out. And then just buying the market plus a 7% deviation. Well 70 bps on 7% means that for the actual investing process, they were paying fees of over 10%.

So what we wanted to do was, was buy index stocks and only pay these managers for the deviations. Needless to say that did not fly (MF the class erupts with laughter).

What does this have to do with risk management? If you decentralize risk management, if you’ve got a lot of managers, in the aggregate the positions of managers A, B and C are going to be offset consistently by the positions of managers X, Y and Z and you’re going to be paying them both for managing offsetting risks. Just like you’re paying these guys for offsetting transactions or offsetting net positions relative to the market.

Therefor this whole idea that hedge funds (MF or any other fund manager for that matter) can manage risk on a decentralized basis and they can do it so well that they’re going to charge enormous fees is just a really bad mistake.

You want to decentralize the process of investing, and you can do it both long and short, but you want to do the risk management yourself. If they do short ideas, they ought to do them because they believe they’re grossly over-valued stocks, not because they’re trying to manage the risk. Because I guarantee you one other of your managers is going to be doing that same stock long.

And it’s just going to be offsetting and you don’t want to pay for that. So you have to manage risk centrally.

* That said it is reasonable for quantitative investment managers to rely more on the breadth of the opportunity set to beat the market. This is because they generally aim to create portfolios of stocks that share one or more common attributes or factors (such as value or momentum) that have historically out-perfumed the market.

This is quite different to a fund manager employing a fundamental investment process to research and value each individual company. Therefore, the signal to noise ratio of quantitative investing is quite low, making breadth a more important variable in the performance of quantitative investment strategies.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s