Sunday, June 12, 2005


Could the losses at BWC have been detected in advance?

In our post yesterday we asked:
Could BWC have known of the scope of the losses and done anything to lessen them as MDL's investment scheme started to blow up? Could this kind of economic catastrophe happen at the state's other funds, such as one of the five public worker pension funds. The answer is yes, yes, and yes.
As we mentioned - without explanation - there is a relatively simple method for avoiding disaster with big-time investments. It's a risk management concept called Value-At-Risk that's been around for over a decade (and arguably much longer).

Rigorous VAR discussions are usually best left for MBA numbers freaks and Finance PhDs. That's because it relies heavily on intense statistical modeling to understand conceptually and high-speed computers to actually implement. But, throw together some math whizzes, a subscription to equity & fixed income databases and some kick-ass servers, and anybody these days can have a VAR system to keep their CFOs from shooting themselves in the foot and their investment managers from running off with the money or accidentally running their organizations into the ground.

The following explanation of how VAR works and could have prevented or lessened the MCL and Noe debacles will necessarily have some flaws because we are going to try to keep it simple.

So here goes:

VAR is a method of measuring the risk at a specific point in time over a specific time period of an investment or trading portfolio. It can easily be used with portfolios that just have equities and bonds, or with some extra tricks, it can also be used for portfolios that have real estate, private equity holdings, and, hey - even hedges and coins.

Let's leave aside for now how the risk is measured. The "why" the risk is measured is shown precisely by the BWC situation. No self-respecting portfolio manager wants to wake up one day to find that he/she is suddenly, say, about $215 million short. At least they don't if they want to keep their job and stay out of jail.

No, a good portfolio manager wants to keep a finger on both his entire portfolio and each of the sub-investments he is responsible for. For example, a good portfolio manager would want to be able to have a fast answer to the question, "What's the most I could lose today (or this week, or this month)?"

(Really, that's a fair question to ask anyone - not just an investment manager - who has fiduciary responsibilities over an investment portfolio. This would include the BWC Nearsighted, sorry, Oversight Commission as well as the trustees of the five state retirement commissions. We wish someone would ask them because the answers would be fascinating. We'd bet money that half of them wouldn't even understand the question.)

Now, a real portfolio manager would actually try to answer the question with much more precision and with something that indicates how certain he is of his answer.

Let's use a somewhat clumsy example. What if someone asked us, "What are the chances today that while traveling on the interstate, a car going the other direction will cross the median and strike another car?" Let's say we also know that a car has to be traveling at least 75 mph to make it through a median barrier.

So, we might answer the question by saying that based on current traffic flow, there is a 1% chance that a car will veer across the median with a speed in excess of 75 mph. Put another way, there is a 99% probability that no car will cross the median doing less than 75 mph. That would probably make us feel pretty safe to drive.

Continuing with this same example, what if we detected that the traffic has suddenly picked up on the interstate and because of the added congestion, we now believe there is a 5% chance that a car will veer across faster than 75 mph. Given that change, we might want to change plans and take a safer route.

Back to the financial world, an investment manager using VAR might say that there is a 99% probability that the fund will lose less than $50 million that day. Inversely, there is a 1% chance that the loss will be greater than $50 million.

So far, we have been talking about investment funds and investment managers in the abstract. Let's say we are talking about the portfolio of "Big Time Bank" that uses VAR. The BTB's board of directors, or it's investment committee or maybe even a risk management committee, will have set a policy that says that the risk of the portfolio must not make things any worse that having a 99% probability of losing $50 million.

This tells the overall manager of the portfolio that if he/she detects an increase in the risk, than the portfolio must be altered to bring the entire risk back in line with BTB's policy. This might require selling an investment, moving money into another investment type, hedging an investment, or any number of ways risk can be adjusted.

The overall portfolio manager will also "budget" the portfolio's risk, i.e., allocate a specific amount of the overall risk to each large subdivision of the portfolio, and the manager or sub-manager will typically further budget the risk right down to each specific investment.

One specific purpose of having VAR and budgeting risk is to keep an eye on investment managers. From the excellent
For institutions to manage risk, they must know about risks while they are being taken. If a trader mis-hedges a portfolio, his employer needs to find out before a loss is incurred. VaR gives institutions the ability to do so.
Hmmm, "mis-hedges a portfolio." Sound familiar?

This issue raises what we believe to be a major problem at BWC and potentially a major problem at the five state pension funds: portfolio managers don't know enough or don't care enough about how their investment managers meet their goals. This is problematic because financial history is full of examples (add Tom Noe to the list) of renegade investment managers who gamble (sometimes literally) on investments that are far from what they are supposed to be investing in.

Unfortunately, portfolio managers are frequently either overjoyed when their investment managers exceed their goals or are blindsided when they start mounting losses. Both are wrong. For example, the correct reaction of a portfolio manager to a subordinate who exceeds his/her goals shouldn't be to celebrate. Instead, it should first be to find out how it happened. Did they really invest in what they were suppose to? Did they do it by exposing the entire fund to excess risk (remember, there is a relationship between risk and reward in the financial world, and if the rewards have increased there must be some place where risk was increased, too)? Who authorized the altered investments or risk?

VAR gives portfolio managers the ability to put their investment managers on a very short leash. VAR can effectively sound alarms about changes in risk at both macro and micro levels.

Okay, the one thing we haven't addressed is what makes a VAR system possible. In brief, VAR operates by applying statistics to investment prices and returns. Although financial theory says that it's impossible to predict whether a stock price will rise or fall in value the next day (the so-called "random walk" of prices) it is possible to predict with some certainty that a price will fall in a certain range.

For example, let's take a look at the price of a share of stock in the XYZ Corporation. Over the past 100 days, although the price of the share has risen slightly, the daily price changes have been less than +/- $.05 for 99 of those 100 days. Although history can't predict the future with certainty, this data suggest that tomorrow we'd have a 99% probability of losing $.05 on our share of XYZ stock. That, in a way, is the VAR for that stock

If we had shares of stock in multiple companies, we could repeat this process for each, and put all this data together to give us a VAR for our entire portfolio.

If, however, the next day, the price of the XYZ stock increases by $.09 cents, and then drops by $.08 the following day, and then $.07 the day after that, and so on for the next 20 days, this new data suggests that the XYZ stock has become more volatile and riskier. Depending on our risk tolerance, we'd have to decide if we still wanted to hang onto that share of stock.

Whew! If you've stuck with this so far, bless you. This is some pretty thick stuff. Throughout all of the above, we have had to cut some corners in order to make some of this sensible to the non-statistically inclined. However, we want the takeaway to be:
  1. That a highly-proven and relatively simple method exists to prevent fraud and excessive losses in portfolios, namely Value-At-Risk.

  2. That the big funds like BWC, OPERS and STRS could easily afford to implement a VAR system, and that the smaller funds like SERS, OPFF, and the OHP funds could probably afford it, too.

  3. That these funds lag behind private sector businesses with similarly large portfolios by 5-10 years in investment management theory and practice.
OPERS will claim that they have something of a VAR system in place. But they don't. We and others have looked at what they have and it falls far short. Some of the other funds may claim to have some VAR, too. Again, how shall we put it? Well, that's just bullshit.

Again, the test is to ask a fiduciary how much money their fund might lose this year. If they can't answer something along the lines of there being a XX% probability that the losses will be less than $ZZ millions, they don't have VAR.

[UPDATED 3:00 PM 6/13/05 with edits to make meanings clearer]


<< Home

This page is powered by Blogger. Isn't yours?