Wall Street loves financial models. They bring with them the aura of objectivity and science, and the sweet smell of high profit margins. I’ve spent a career building models and purchasing sophisticated financial systems. Although these tools can be helpful, they seldom provide definitive answers to the key questions facing investors.
At their best, models provide you with an educated guess, and at their worst, they come up with nothing more than statistical noise. Models are especially dangerous to your investment health, because they appear to be extremely precise.
This week I want to focus on how models drive the two most important decisions for any investor. How should you allocate your financial assets? And, will you run out of money?
Meet the optimizer
Anyone with a financial adviser or the curiosity to surf financial websites will have encountered an optimizer. The optimizer is simply a model that figures out the most efficient trade-off between return and risk among asset classes.
In theory, the optimizer is a rational way to find the greatest possible return for the least possible amount of risk. Since the calculation is performed on a computer, the optimizer will give you precise-looking results. But the numbers are only rough estimates.
For example, the computer might tell you that you should be in 18.3 percent international stocks, 38.2 percent domestic stocks, 32.8 percent municipal bonds and 11.7 percent real estate given your tolerance for risk. The optimization has given you a sense of direction, not an exact recipe for your asset allocation.
The results, however, aren’t the key elements in an optimization model. Rather, the assumptions are the critical components.
Those exact-looking results aren’t merely driven by a couple of sophisticated formulas. Rather, the results depend on what your broker, adviser or financial plan entered as assumptions for the future returns and risk of equities, bonds and any other asset classes.
In addition, there’s a third important assumption buried in these models, known as correlation. Correlation is simply the relationship between any two asset classes.
In other words, correlation measures how closely or loosely stock and bond prices track one another, or how precisely or poorly international equities track domestic equities.
In the first weeks of my tenure as North Carolina’s chief investment officer, I reviewed the asset allocation recommendations made by the state’s money managers. Each of our money managers performed an optimization of the state’s pension assets, and lo and behold, each of their results recommended that we add the exact product they were trying to market.
When I asked the managers to furnish their assumptions, instead of just the results, I discovered they had tweaked the inputs to nudge their products to the forefront.
I’m not suggesting that you ignore the optimizer as a financial tool. Instead I’m urging you to get a cogent, common-sense explanation of the assumptions and how they were derived before accepting the results of optimization as the basis for allocating your assets.
Tools project retirement needs
In order to determine whether an investor has amassed enough savings to last a lifetime, advisers typically use two-interrelated tools: a cash flow model and a Monte Carlo simulation.
A cash flow model simply projects your income and expenses and credits and debits your savings accounts. The model is designed to determine whether Social Security, tax-deferred savings, a defined benefit plan and after-tax savings less living expenses will be sufficient.
The cash flow model has a major flaw in that it can only accommodate one scenario at a time and produces a single answer. Clearly, over several decades or more a range of outcomes is possible depending on your compensation and spending as well as the market’s returns.
The Monte Carlo simulation allows you to see a range of potential outcomes and also to weigh the odds that you’ll outlive your assets. Thanks to the computer, a Monte Carlo simulation can run thousands of scenarios in which each of the critical assumptions in a cash flow model are varied.
As a result, the simulation can predict the odds of running out of money at any given age. For example, the analysis might show that there’s a 50 percent chance of running out of money at age 95. If your family’s genes suggest that you’ll live into your tenth decade, you might want to increase your savings rate or take a bit more market risk in order to decrease the odds of going broke.
Monte Carlo simulation works well in the realm of science. However, when it is exported into the world of investments, it should be approached cautiously.
In a scientific context, the key assumptions that drive the Monte Carlo simulation meet rigorous statistical standards. In the investment context, those statistical standards are severely compromised. As with optimization, you should ask your financial planner or broker to lay out the key assumptions behind their work.
If the model uses reasonable assumptions about the financial markets and reflects your income and spending habits, then a Monte Carlo simulation may be a useful guide.
While investing is chock-full of numbers, it is not science. So when financial advisers apply scientific tools such as optimizers and Monte Carlo simulation, you should proceed with caution.
These techniques can offer some useful guidance, but they are far from precise. Moreover, a financial adviser can use the assumptions to steer you toward favored products instead of giving you objective advice.
In general, I’ve found it’s most helpful to use conservative assumptions about financial returns and rely fairly heavily on saving money. When I’ve let computers convince me that I can generate big investment returns, I’ve often been disappointed.
Andrew Silton’s Meditations on Money columns can be found twice a month in The N&O’s Work&Money section. He is a retired money manager living in Chapel Hill. He was CIO for the North Carolina Retirement System from 2002-2005. He writes the blog http://meditationonmoneymanagement.blogspot.com/