Wednesday, November 11, 2009

Perfect Knowledge? Perfect Peace?

People are moving back and rebuilding New Orleans, even after Katrina. It is a certainty New Orleans will suffer another catastrophic flood. The only question is when. Civil engineers have been sounding the warnings for decades. A class 4 or 5 hurricane could cause massive breaches in the levees surrounding New Orleans. A major flood on the Mississippi could cause the same sort of disaster or wipe out the dam system that keeps the Mississippi from heading down the Atchafalaya river basin. This would isolate New Orleans from the Gulf of Mexico thus ending its reason for existence. Any kind of flood could cause the failure of the system of levees that protects the city from Lake Pontchartrain. A system of levees and dams that could guarantee the safety of New Orleans is a technological impossibility. The city is built on a sea of mud. There is no bedrock. Therefore, the higher and heavier the levee, the faster it will sink into the mud. The American Society of Civil Engineers observes, “Large portions of Orleans, St. Bernard, and Jefferson parishes are currently below sea level — and continue to sink. New Orleans is built on thousands of feet of soft sand, silt, and clay. Subsidence, or settling of the ground surface, occurs naturally due to the consolidation and oxidation of organic soils (called “marsh” in New Orleans) and local groundwater pumping. In the past, flooding and deposition of sediments from the Mississippi River counterbalanced the natural subsidence, leaving southeast Louisiana at or above sea level. However, due to major flood control structures being built upstream on the Mississippi River and levees being built around New Orleans, fresh layers of sediment are not replenishing the ground lost by subsidence.” Substantial portions of the city are below sea level and they continue to sink. Technology can not change this fact. Yet, people live in New Orleans. Much to my amazement they consider it an acceptable risk.

As we recover from the collapse of the stock and real estate markets, there seems to be a renewed interest in risk analysis. Richard Young’s sustainable Portfolio Income chart shows the value of a portfolio invested 50% in fixed income and 50% in stocks rebalanced annually. Distributions are assumed to grow at the annual rate of inflation. A portfolio invested 50–50 in 1946 with a 4% withdrawal rate would have lasted for 53 years, a 5% withdraw rate would have lasted 34 years, a 6% withdraw rate would have lasted 26 years, but the past is not the future. Likewise, Schwab came up with the same 4% withdraw rule of thumb, not based on past performance but based on running sophisticated Monte Carlo simulations in order to develop a 90% confidence level in their results. In an email message to their customers, Schwab admitted that years like 2008, a once in a lifetime event (we hope), could not be covered by a 90% confidence level. If I had been an average investor and if I retired in 2008, I would be in extremely serious trouble. When the market withdraws 18% of your holdings (50% bonds 50% stocks) in a single year, the 4% rule obviously goes out the window.

Most of the predictive tools available to folks like you and me are termed deterministic. That means, as I plan my retirement, I pick a life expectancy, an average rate of inflation (often 5% is recommended), and an average rate of return (often 8% is recommended) on investments. I plug the numbers into the website and read the results. Since I can not predict my life expectancy, inflation for the next 30 years, or what the stocks I buy and sell might do or not do, these results are really not worth too much, but they are better than nothing.

A Monte Carlo simulation is more sophisticated. Most Sunday mornings I drive approximately 17 miles from Poolesville to Derwood. I make one stop to buy cheap gas at the 7-11 store. On average this odyssey takes about 35 minutes but it can take 7 or 8 minutes more or less than average based primarily on how I hit the traffic lights. I do not always start at the exact same minute. If I am running late, I drive a little faster. If I am running early, I slow down. I could develop a Monte Carlo simulation that would model my behavior and mathematically verify an average time for the Sunday morning commute. A random number generator would give a starting time, information from the computer system that controls traffic lights in Montgomery County would run a subroutine providing stop/go information to the main program, and the computer could spit out the results of 10,000 commutes in a matter of minutes. But did the program account for deer hits, accidents, and construction projects? Probably not. If get stuck in a once in 100 year event, like the time a truck flipped on I-95 releasing scores of pigs from their cages, my simulation results go out the window. In that case the road was shut down for hours while emergency personnel recaptured the surviving animals.

Systems engineering obsesses about catastrophic single point failures. If the failure of a single part can result in the complete destruction of a massively complex system, this is considered a serious design flaw. Redundant subsystems are often added to prevent an entire systems failure, like the O ring seals on the booster motors of the space shuttle. The failure of one piece of plastic destroyed a system worth tens of millions of dollars and killed 7 astronauts. These unlikely events keep system engineers awake at night. To locate these problems and correct them before they can occur, thousands of simulations involving all sorts of unlikely, “what if?” scenarios are considered and designs are modified based on the frequency of the event, the consequences of the event, and the cost of preventing the failure.

Probably 4% withdraw is about as good a model as I can hope for, as I plan my retirement. How does this work? Let us say a couple has traditional pensions totaling $20,000 a year and expects an inflation adjusted income of $25,000 a year from Social Security. If this couple wishes to retire on an income of $70,000 a year, they will need an additional $25,000 a year generated by their investments. $25,000 ÷ 0.04 = $625,000. To avoid a single point failure, diversify. At one point in time putting all your money into Enron stock looked like a pretty good idea. However, the bankruptcy of that wicked company destroyed thousands of innocent investors. A portfolio with 10% in Enron stock would have been hurt but would survive. A portfolio with 5% in Enron stock would be embarrassed but would not have suffered any grievous, irreparable harm.

Of course the problem is that we humans, the machines we construct, and the computer simulations we write all lack perfect knowledge. Chess is a game in which both players know the position of every piece on the board and the complete history of the game. Chess computers can and have been programmed so well that one of them, Deep Blue, defeated the World’s Champion in an exhibition match. Poker is a game in which no player has perfect knowledge of either the opponents’ cards or the motivations behind their betting strategies. For this reason, attempts to program a computer to play poker with human adversaries have been failures.

I bet you know by now where I am going to end this essay, with the author of perfect knowledge and perfect peace.

Psalms 23

[1] The LORD is my shepherd; I shall not want.
[2] He maketh me to lie down in green pastures: he leadeth me beside the still waters.
[3] He restoreth my soul: he leadeth me in the paths of righteousness for his name's sake.
[4] Yea, though I walk through the valley of the shadow of death, I will fear no evil: for thou art with me; thy rod and thy staff they comfort me.
[5] Thou preparest a table before me in the presence of mine enemies: thou anointest my head with oil; my cup runneth over.
[6] Surely goodness and mercy shall follow me all the days of my life: and I will dwell in the house of the LORD for ever.

No comments:

Post a Comment