I sat for Part III of the Actuarial Exams in November of 1972. In January I learned that I had passed; not with a 10, but with a solid 8. I was studying to take Part IV (Life Contingencies) of the actuarial exams in May. I had passed the finals in my Industrial Relations class with flying colors, I was doing well in my Buddhism class at UCLA, and things were going swimmingly well at work -- the world was my oyster. And then my world turned upside-down.
I think it was on Monday, April 2, 1973 that I got the bad news. I drove to work as usual and walked to the elevators, where I was met by a guy in a uniform that said "FBI". He was packing a .45 pistol, and he asked for my driver's license. "Why?" I inquired. "I'm not driving." He gave me a hard stare and patted his holster. I gave him my driver's license. After checking it against a list he told me I could get on the elevator. Thus began a surreal ordeal that lasted about two months.
As the world eventually learned, Equity Funding Corp had orchestrated a massive fraud upon Wall Street and American investors in general. They had been padding their actual book of life insurance business with thousands of fictitious policies that had been sold (for a 200% first year commission) to a number of unsuspecting reinsurance companies under a modified coinsurance arrangement. ("Modco" was popular in the '70s and '80s because it allowed both parties to the contract to lower their income tax liability, when done correctly.) Somebody had finally blown the whistle, and the Feds had descended on EFCA in force. Nobody was getting within a hundred yards of the company's computers except for some special investigators hired by the SEC and the California Department of Insurance. The variable life project came to a screeching halt. There wasn't much work to do; for a couple of weeks I prepared corporate income tax returns for some of Equity Funding's real (mostly subchapter S) clients, but that petered out in about two weeks. My days settled into a dull routine. I would buy the Wall Street Journal and the Los Angeles Times, settle into a chair in a coffee shop in the basement, and read both papers from cover to cover while consuming copious quantities of coffee.
I suppose some people would have considered this a dream job. I was getting paid $1,000 per month for doing nothing. But I was young, and ambitious, so I didn't see it that way. I wanted to be doing things, and moving up in the world. So I started lining up interviews with actuarial departments all around town. The result was always the same. The interview would go very well. But a few days later I would get a call. "We're sorry. We really like you. You're extremely well qualified. But I spoke to the president of the company, and he says we can't offer you a job because you're associated with Equity Funding Life. Good luck."
I asked a few of the people I'd interviewed what I might do to break out of this rut, and they all gave me the same advice: "You need a letter of recommendation from the California Department of Insurance, signed by the Commissioner, if you can swing it." I knew a couple of the guys who were working for the Insurance Department, so I drew up a letter that said something like "Mr. Bryant's conduct throughout this entire unfortunate affair has been exemplary, and his integrity has never been called into question." I gave it to one of the Insurance Department guys I knew, and he said he'd pass it up the chain of command. About three weeks later I got a letter signed by the Deputy Commissioner that said "Our investigation to date has not yet implicated Mr. Bryant in the wrongdoing at Equity Funding Life." Oh, well. Half a loaf is better than no bread at all. I sent the letter around to half a dozen companies and a couple of weeks later, right after I had graduated from Caltech, one of them bit. I had a job working for Roger Johnson at Great American Life, just off San Vicente Boulevard in mid-town Los Angeles.
Great American Life is a wholly-owned subsidiary of Great American Insurance Company, an old-line property and casualty insurance company from Cincinatti, Ohio. They had moved out to California about 1965, but have since moved back to Cincinatti. When I worked there, both the casualty company and the life insurance company were housed in one building. Total staff was about 150 people. GALIC did not have any computers in house. Their daily cycle (batch processing for policy administration) was being run at a service bureau in West L.A., and they were using an on-line service provided by the Service Bureau Corp (SBC) for their product development work. SBC was a spin-off from IBM that provided dial-up access to mainframe computers. Programs could be written in "QUIKTRAN" (which was sort of like CITRAN at CalTech) or in BASIC. We were charged for connect time, CPU time, and storage space on disk.
One of my first projects was to calculate cash values and reserves for a whole raft of "mortgage protection" decreasing term policies, so called because the face amount decreased each year to match the outstanding principal balance of a mortgage loan at a particular interest rate and for a certain term of years. The plans for which I had to calculate values ran for either 15, 20, 25, or 30 years, and corresponded to mortgage interest rates running from 6% to 12%, in increments of ½%. That made 52 insurance plans all together: 4 terms of years and 13 mortgage interest rates. I set this all up in BASIC. Once I had the schedule of death benefits, the rest was fairly simple. Writing the actuarial memorandum that explained how the values were calculated took longer than getting that program written and debugged. Roger was very happy with the way I handled that first assignment. He gave me a pretty free rein to do things my way from then on.
It was about this time I learned that I had failed the actuarial exam I had taken in May. This was an annoyance. I reviewed the material I had studied, and decided that I had most likely slipped up on the questions about "compound contingent functions". These relate to special forms of insurance involving two or more people. For instance, a business might want to buy an insurance policy on one of its four owners that pays a full benefit if he is the third of them to die, half the full amount if he is the first or second to die, and nothing if he is the last of the four to buy the farm. I concentrated on that part of the material, and was rewarded with a passing grade on Part IV in November of 1973.
There are a couple of other projects at PFLIC that stick in my memory. Late in 1973 the company decided to enter a modified coinsurance agreement with a big company back east; I think it was Connecticut General. I helped prepare a profit and loss model for the reinsured policies. The chief accountant came to me with a question. "For GAAP accounting purposes, I want to amortize the big commission we're receiving up front in proportion to the payments we'll make in years 2 through 10. What interest rate should I use?" I set up a polynomial equation of the 10th degree that equated all the cash flows to zero, and used Newton's method to find a real positive root that was roughly 1.07, as I recall. I worked out the value to eight or nine decimal places, and gave the number to the accountant. He was back the next day. "Wow! It worked out to the penny! How did you find that interest rate?" I started to explain, but as soon as I said "calculus" his eyes glazed over. This is one of the very few instances in which I was actually able to use mathematics -- not simple stuff like plus and minus, but a real derivative! -- to solve an actuarial problem. The actuaries talk a good game, but when you come right down to it, most actuarial work is fifth or sixth grade math.
Roger gave me one other assignment that made me think pretty hard. He wanted an estimate of the pure risk premium for a stop-loss reinsurance agreement. Stop-loss reinsurance is not generally available. It covers an insurance company's entire book of business and promises to pay some percentage (typically 90% - 100%) of the company's total claims in excess of a specified threshold within some period of time (usually one year). Most reinsurance is on single risks. Setting the right price for a stop-loss agreement is a lot more difficult than pricing individual risks. I started by analyzing GALIC's business by size of policy: under $10,000, $10,000 t0 $24,999, $25,000 to 49,999, etc. I then found a few probabilities of loss for each size of risk in the model: I used broad age ranges like under 35, 35 - 49, 50 - 64, 65 and older to pick initial values of qx, then fine-tuned the values by reference to an expected value of total claims estimated independently. I ended up with something like 40 "cells" to represent GALIC's book of business.
Now the rest of the problem was relatively straightforward. All I had to do was evaluate the distribution of claims for each cell in the model (a simple binomial distribution in each case), then perform a convolution integration across the 40 some-odd cells. I wrote the program and ran it. The results seemed OK, except for one thing: the total probability of claims being some amount or another came to 96%. That was an error -- the total had to be 100%. Clearly my results were being poisoned by round-off errors. After thinking about this for a while, I decided that the problem was caused by significance errors: I was tryng to add very small quantities to much larger quantities, and losing many digits of significant information in the process. I re-ordered the calculation so that all the smallest probabilities were evaluated first, and added together before evaluating the more likely combinations of individual claim amounts. Voila! The revised probability of total claims in some amount came out to 99.9999999%, close enough to 100% to satisfy just about anyone.
I really liked Roger Johnson and the other people at GALIC, and I might have stayed there a lot longer than I did if there had been a better path open for advancement within the company. In November of 1973 I passed Part IV of the actuarial exams. Roger called me into his office. "I would like to give you a raise for passing the exam, but my budget is limited. Bill Peterson has worked here a lot longer than you have, and I can't in good conscience give you a raise unless he gets one too. Maybe when you pass Part V and become an Associate of the Society." I wasn't real happy about this. The market value of Part IV was $150 a month or more, which is about what I was paying to rent an apartment in Venice. I started looking around town for a new opportunity and soon found one at Pacific Fidelity Life, conveniently located in Century City, just two blocks from my old stomping grounds at Equity Funding. Steve Eisenberg offered me a $2,000 raise; by March of 1974 I was working at PFLIC.