Early in the spring of 1972 I saw an announcement on the bulletin board near the main entrance to Bridge Laboratory. "Attention Math Students" it said in big bold letters. The poster went on to say that the Society of Actuaries was administering tests in May of 1972, that it only cost $10 per test to participate, and that the top-scoring paper on test #1 (Basic Algebra and Calculus) would receive a cash prize of $100. So I did a little research on the actuarial profession, and decided that it sounded like a pretty good way to earn some money over the summer. I sent in $20 to take parts I and II of the actuarial exams, and wrote my dad a letter explaining that I was planning to land a job in Los Angeles, and wouldn't be coming home for the summer. He said OK.
Since I had already completed the course requirements for a BS in mathematics, I petitioned the Caltech faculty committee in charge of academic affairs to receive my degree in June of 1972, one year ahead of schedule. They sent me a very nice letter that said "Well, you just barely have enough credits, and we haven't let a math student get out in just three years since 1949. Plus you're not as smart as that guy was. So no." I was a little disappointed, but not terribly surprised.
If I were going to spend the summer in Los Angeles, I would need a set of wheels. You just cannot survive in L.A. without wheels. I found a nice little 4-door Chevy II Nova for sale by a grad student at Caltech -- $350. After forking over my cash (plus another $100 for insurance) I was on the road. The Chevy II would serve me well for about two and a half years. By the time it bit the dust, I had spent more money to insure it than I laid out to buy it. Insurance is ridiculously expensive for young male drivers.
Because of my part-time job at Bridge Laboratory, I had developed a solid friendship with Nancy Durland (who served as Carl Anderson's secretary) and her husband Les, who was a grad student at Caltech. I talked to Les about the actuarial profession; he had a job working for Occidental Life in downtown L.A., and said he'd put in a good word for me. Occidental was planning to hire a couple of college students over the summer. One day late in April Les gave me a ride to Occidental's home office on Olive Street in downtown L.A. I had three interviews that day, and thought I might have a real shot at a job. But a few weeks later I got a "Dear David" letter explaining that, in the spirit of diversity, Occi would be hiring a black guy and a woman from back east into their internship program. So much for "white privilege".
In mid May I sat for the actuarial exams. Part I was a three hour affair with 150 multiple choice questions (mostly advanced algebra and calculus of a single variable, with a smattering of trigonometry thrown in just to keep it interesting). Part II was a two hour test on Probability and Statistics. I finished both exams with time to spare. I had just finished a course in probability and statistics, and calculus was second nature to me after three years at Tech.
I've never believed in putting all my eggs in one basket. Besides the interview at Occi, I had sent my resumé to all the insurance companies in Los Angeles that were sponsoring interns that summer. In early June I was invited to interview Al Lewis and Jim Smith, the two head actuaries at Equity Funding Life in Century City. A week later I had a job. A couple of weeks after that I received the news from the Society of Actuaries: I had received a score of 10 (the highest possible score) on both the actuarial exams I took in May.
Equity Funding's office was on Avenue of the Stars in Century City. It was about 20 miles from my apartment in Pasadena to my new job. The best route I finally settled on was out the CA-134 freeway to Coldwater Canyon Drive, and thence to Santa Monica Boulevard and Century City. It's hot in Pasadena in the summertime. And Chris Goldstein and David Kazan were gone, leaving me to pay the rent on a two bedroom apartment. I started looking for a new place to live that was a bit closer to work, smaller, and nearer the ocean, too. I soon found a nice apartment at 417 Ocean Front Walk in Venice. I had to give 30 days notice to my old landlord. But on August 1, 1972 I moved all my stuff (not much, really) to my new digs, just eight miles from work.
I wasn't the only Caltech student working for Equity Funding that summer. Albert K. Christians, a Rudd who was a year older than I, was also working there. The first project to which we were assigned was to consolidate three ratebooks. Equity Funding Corp, Equity Funding Life's parent, had recently purchased two more life insurance companies (Northern Life, of Seattle, and Bankers National, from New Jersey), and Albert and I were supposed to come up with premiums, cash values, and reserves for a new consolidated ratebook that would serve the agents of all three companies. So that's how I was first exposed to PL/I: by writing programs to compute and print tables of cash values and schedules of premiums.
PL/I was a bit more complicated than Fortran. There were a lot more data types, including data structures (allowing one to define record layouts sort of like COBOL), and pointer variables (which work like indexes into arrays, except that they could be used to index one's way through a character string or a data structure, one byte at a time). The syntax in PL/I is very free-form. Statements could start and end anywhere on an 80-column punched card. There were a couple of hard and fast rules. Every program statement ended with a semi-colon, and comments were delineated by an opening /* and a closing */. After testing the limits of this free-form format, I settled on the simple expedient of putting just one statement on each line. It made the program listing much easier to read.
The new ratebook project didn't last long. I'm not entirely certain, but I think some political battles were started by the various agencies, none of which wanted to give up its own favorite flavor of life insurance. Anyway, about mid-August Albert and I were yanked off the ratebook project and told to start working on a new program. Our goal was to set a price for a new "variable life" product that Equity Funding hoped to bring to market. This was cutting edge stuff. Congress was considering legislation that would allow the SEC and the various state insurance commissioners to jointly supervise the marketing and maintenance of a new kind of life insurance. Instead of having guaranteed cash values and a constant death benefit, a variable life policy would have a cash value that depended on the investment performance of an underlying basket of common stocks, and a death benefit that could increase if the stocks did well, but would never decrease below a guaranteed floor. There were two competing designs: a New York Life design (that aimed to maintain a death benefit proportional to the increase in cash value over the original "guaranteed" level) and an Equitable Life design (that was similar to buying paid-up additions with dividends).
This was a project we could really sink our teeth into. Albert read up on building computer models of the stock market, and I read up on methods for projecting the profitability of a life insurance policy. We settled on a fairly standard "asset share" approach to the profitability question. I wrote that program in PL/I. Albert devised a "random walk" model of the Standard & Poor's 500 index. We fitted the first four moments of the random-walk model to actual S&P data for the period 1918 - 1971, then used Albert's program to generate 10,000 different 30-year simulations of the market. We saved these simulations as a data file, then fed them into my profit-testing program and summarized the resulting distribution of profits and losses. The technique we used is known as Monte Carlo simulation. Running all 10,000 scenarios for a single issue age soaked up about two hours of compute time on the company's S/370-145; elapsed time tended to run from 3 - 6 hours, depending on what was running in other partitions. So we had to schedule a bunch of our simulations to run at nighttime.
About this time Equity Funding offered me a full-time job, at a salary of $1,000 a month. Since I only had a few credit hours to go to fulfill the degree requirements at Caltech, and since my little sister Betty was about to enter Boston College, I took the job. During my senior year I worked full time and took a smattering of electives (Industrial Relations at Caltech; Buddhism and a course entitled "Ideology and Moral Philosophy" at UCLA) by attending evening classes. It was also about this time that the Society of Actuaries informed me that I had won second prize on the Part I exam, and that they'd be sending me a chack for $75 "when the president of the Society returns from his European vacation". I began to wonder if that check would ever arrive; it finally did, along about Thanksgiving.
The 370-145 was not the only computer in the building. The actuarial department had a System 3 Model 6, with a card reader, a printer, a combination keyboard / video terminal, and a disk drive. The card reader was unusual; it read data from a 96-column card (actually three sets of 32 columns, stacked in rows). It didn't use the same Hollerith code that the bigger 80-column punched cards did; I think it was a straight binary encoding (eight rows in each of the "columns"; a hole was a "1", and a not-hole was a zero). The holes were round, not rectangular like the holes in the 80-column cards. The disk drive had one removable 2.5 MB platter. I think we had three or four of those. The operating system and the BASIC interpreter program resided on a fixed disk; the removable disks used the same drive hardware, which was inside a drawer.
I got a lot of practice writing BASIC programs on that old System 3. BASIC is a lot like FORTRAN, with a few minor differences (for instance, a loop in BASIC starts off "FOR i = 1 to 100 BY 1" -- a similar loop in FORTRAN starts off "DO 500 i = 1, 100"). The biggest difference is that FORTRAN is a compiled language (meaning that the FORTRAN compiler produces an object program, in machine language) while BASIC is an interpreted language (the BASIC interpreter processes the source program one line at a time, and never does translate the entire program into one machine language module). There really weren't many actuarial applications for the System 3. The most complicated program I wrote in BASIC calculated the odds for all the possible distributions of cards in a set of four bridge hands (4, 3, 3, 3; 5, 3, 3, 2; 5, 4, 2, 2; etc.) In other words, If my hand is (4, 3, 3, 3), what is the chance my partner also has balanced distribution? How about our opponents? If I have 5 cards in one suit and my partner has 4, what is the chance of a (2, 2) split? A (3, 1) split? A (4, 0) split? Etc. (Albert and I were both avid bridge players; we met at Jeff Hartman's house every Monday night to play bridge for a couple of hours. Jeff was another of the actuaries at Equity Funding Life.)