Articles

Investigators and Statisticians

January 31, 2022 | François Sicart

 I have recently been exposed to the work of what is called a “forensic accountant.” These specialists analyze financial information to investigate fraud or embezzlement. Their job is to reconstruct the entire succession of actions that led to the suspected misdeeds. I was frankly amazed at what one can uncover by assembling a puzzle of disparate and often incomplete information and trying to reconstruct the process leading to the final result.

As I recall in my recent book Luck Is Not Enough (Advantage, 2021), I was at best a mediocre student of accounting in business school. Fortunately, after graduation — and for lack of other candidates — I was selected to become an assistant to two high-level American professors of accounting, and this is when I learned most of what I still remember today of the discipline.

These inauspicious beginnings did not prevent me from (briefly) becoming a professor at an institution preparing candidates for the French equivalent of the American CPA (certified public accountant). When I discovered that my students knew more about the rules of financial reporting than I did, I pivoted to teach them the philosophy of accounting. Apparently, I was a very popular teacher, though I am not sure how my students fared at their final exams. But as Swedish essayist, Ellen Key wrote in 1891: “Culture is what subsists after we forgot everything we had learned.”

The Data Can Become Irrelevant

In the 1980s, the economic consensus was that Japan, with its bulging trade surplus and booming stock and real estate markets, was eating the lunch of the US economy. My team at the time disagreed and embarked on a research project to try and prove that American manufacturing was not dying but being reborn. One of the experts who helped us in this endeavor was Robert Kaplan, a well-known professor of accounting at Harvard.

Professor Kaplan explained to us that the statistics used to measure the performance of American manufacturing corporations were based on 19th-century accounting, whereas we were on the brink of entering the 21st century. In the earlier period, the bulk of manufacturing costs were raw materials and direct (manual) labor. For expediency, all other costs (including research, marketing, administration) were allocated to products in proportion to those direct costs.

But by the 1980s, the industry had changed radically. Labor, especially in rapidly-developing electronics, had shrunk to as low as 10% of total costs. Similarly, the industry was using less and lighter materials. So while those costs had diminished dramatically, we were still allocating corporate overhead on the basis of two much less relevant inputs. The traditional measures of corporate productivity and profitability had become highly misleading and had begun to prompt errors of judgment.

The point is that over time, the data that forms our business and economic statistics evolves. It’s necessary to analyze what goes to make up these data to avoid the traditional “garbage in – garbage out” nature of statistics.

Data vs. Process

These days, I am increasingly sensitive to the fact that statistics deal with results (usually results of measurement) without paying too much attention to how these data are obtained.

I created Tocqueville Asset Management in 1985, and I am particularly proud that, during the wild popularity of Enron Corp. among investors in the 1990s, two of my partners particularly knowledgeable in energy matters visited Enron twice and came back with an answer that is too rarely admitted: “We don’t understand how they do it.” This took some courage, as Enron’s reported results were excellent. Fortune named Enron “America’s Most Innovative Company” for six consecutive years, and management routinely humiliated skeptical analysts by labeling them as stupid or ignorant.

But in late 2001 it became clear that Enron’s apparent success was sustained by a remarkably well-constructed and sustained fraud. The company ultimately went bankrupt. And here is the simple yet crucial lesson: If you don’t know or don’t understand how a company’s profits are achieved, you should not risk your or your clients’ money on its shares.

“Performance” and The Madoff Episode

In today’s field of investments, I find a particular disconnect between two groups of professionals. On the one hand, there are portfolio managers who study the process by which companies generate profits before deciding how much they will pay for these profits. On the other hand, there are performance measures, sometimes labeled “asset allocators,” who compare companies’ statistics with little regard to how and why the performances were achieved.

Many readers will remember Bernie Madoff, the infamous American financier who ran the largest (roughly $65 billion) Ponzi scheme in history. (A Ponzi scheme is a form of fraud that lures investors by paying profits to earlier investors with funds raised from more recent investors.) Madoff claimed that his fraudulent activities began in the 1990s. However, federal investigators believe the fraud in the investment management and advisory divisions of his company may have begun in the 1970s. Be this as it may, Madoff admitted that he hadn’t actually traded since the early 1990s and that all of his returns since then had been fabricated. (Wikipedia) He was sentenced to 150 years of incarceration and died in prison in 2021.

At the peak of his pyramid, Madoff employed numerous financial organizations to raise and channel funds to his “investment” products. Surprisingly, I had never heard of him until the scandal erupted around 2000. Later I discovered that I knew a number of professionals who had contributed to his fund-raising and had sometimes invested and lost their own money, along with that of their clients.

The main problem was that these professionals only performed statistical analysis on the reported performance of Madoff’s funds without checking the process that had purportedly produced the claimed returns. I am concerned that, with today’s fixation on measuring statistical performance against widely-used benchmarks (i.e., leading indexes) without investigating the investment process, the opportunities for such catastrophic errors of judgment have multiplied.

An additional concern is that many consultants and asset allocators insist on measuring performance on ever shorter periods of time (not only yearly but monthly, or even more frequently). If only for demographic reasons, patrimonial performance only matters over long periods, but what consultant would expect to be paid for statistical analyses that he or she delivers only once every so many years?

The result of this short-term focus on performance is that advisors, to prove that they are hard at work, seem to recommend portfolio changes more often than is necessary or productive: they only wind up increasing the turnover of portfolios. The reality, as Ernest Hemingway reportedly advised, is that one should never mistake motion for action.

 

François Sicart – January 31, 2022

 

Disclosure:

The information provided in this article represents the opinions of Sicart Associates, LLC (“Sicart”) and is expressed as of the date hereof and is subject to change. Sicart assumes no obligation to update or otherwise revise our opinions or this article. The observations and views expressed herein may be changed by Sicart at any time without notice.

This article is not intended to be a client‐specific suitability analysis or recommendation, an offer to participate in any investment, or a recommendation to buy, hold or sell securities. Do not use this report as the sole basis for investment decisions. Do not select an asset class or investment product based on performance alone. Consider all relevant information, including your existing portfolio, investment objectives, risk tolerance, liquidity needs and investment time horizon. This report is for general informational purposes only and is not intended to predict or guarantee the future performance of any individual security, market sector or the markets generally.