(FORTUNE Magazine) – ''O what a tangled web we weave, when first we practice to deceive!'' Seldom has Sir Walter Scott's lament seemed more apt than in the so-called Andy Krieger affair at Bankers Trust. Though the events took place more than four years ago, some institutional investors still recall them as a reason for steering clear of Bankers Trust stock. What they remember was Bankers' surprise revision of its 1987 income statement, many months after it was originally published. But even then investors never quite caught on to what happened. The truth -- reported here for the first time -- is that Bankers Trust knowingly and intentionally, and with the acquiescence of its outside accountants, published inaccurate financial data in its 1987 annual report. Andrew Krieger's name is on this affair not because he has ever been accused -- publicly, at least -- of doing anything wrong, but because he created the profits that got restated. Now 36, he was a currency trader at Bankers in 1987, a year in which the company made phenomenal amounts of money in foreign exchange, particularly in the uncertain months after the October stock market crash. Bankers Trust disclosed its success in a January 20, 1988, press release reporting preliminary, unaudited earnings. The release said foreign exchange trading income for 1987 was a stunning $593 million, of which $338 million was made in the fourth quarter. With the help of that trading income -- which banks view as profits before expenses and taxes -- Bankers Trust managed to swallow a special $700 million addition to reserves for loans to less-developed countries and still emerge with a tiny profit for 1987: $1.2 million. Krieger's contribution to it all was made famous in a March 1988 Wall Street Journal article, which said he personally had produced $300 million of the 1987 foreign exchange trading income. For this he got $3 million in salary and bonus. That struck Krieger as insufficient. Today, from his office in New Jersey, where he runs a currency advisory company, he says he was also fed up with working ''120-hour weeks.'' So in February 1988, he left the bank. Dire financial news had by then hit the trading department, though it was unknown to the world. The news concerned the valuation of certain currency options -- that is, puts and calls -- in Krieger's portfolio. Bankers Trust, then and now, had two front-line methods of valuation. In the first, traders and their clerical staff, feeding data into automated systems, marked their portfolios to market every day, checking their own notions about value by phoning other banks and brokers to get their quotes. In the second procedure, which kicked in periodically, financial control teams independently did their own confirming valuations. The problem in early 1988 was that the earnings release went out on January 20 before the financial control people had delivered their verdict about the portfolio at year-end. And when that verdict came in -- actually later in the day on January 20 -- it reduced the value of Krieger's portfolio, and therefore also foreign exchange income, by a full $80 million. In other words, income was not the $593 million that had just been reported to the world, but $513 million. The straightforward thing for Bankers Trust to do at that point would have been to publicly correct the preliminary earnings figures, lowering them for the quarter and the year by $80 million. But that would have given the company a loss for the year, its first since the 1930s. Furthermore, a correction would have fortified the critics who contend that trading operations are habitually out-of-control, uncertainly profitable ventures. So Bankers Trust came up with an accounting ''solution'' to this very sticky problem. Unbeknownst to the world, it had not reported all the profits it * could have in 1987, but had instead accrued -- that is, charged its books -- more in compensation than it actually paid. The company says today that it thought of the overaccrual as earned compensation that it would pay its staff, particularly traders, in a future period. But it is also possible to think of the overaccrual as a hidden reserve, which would have been available to pump up profits in the future. Hidden reserves are common in many European countries. They are not, however, permitted by generally accepted accounting principles in the U.S. Nonetheless, Bankers Trust had this trove conveniently handy. Huddling internally over what to do about the vanished $80 million in foreign exchange income, management decided to offset it by reducing 1987's compensation expense by $80 million, from the $862 million reported in the January release to $782 million. That would have the effect of adding $80 million in profits to make up for the $80 million that had disappeared. As a result of this musical chairs exercise, earnings -- the bottom line -- would not be changed. And did Bankers Trust divulge all these facts and accounting convolutions in its 1987 annual report and 10-K filing with the Securities and Exchange Commission, which were published in March 1988 and which, of course, gave final, audited figures? It did not. Instead, the company and its outside accountants, Arthur Young & Co. (now part of Ernst & Young) adopted the theory that none of the above was ''material.'' Bankers Trust therefore repeated in the annual report the same foreign exchange income and compensation expense it had disclosed in January, even though it had by then internally ''adjusted'' the compensation expense and knew that the foreign exchange figure was straight-out wrong. Arthur Young & Co. simultaneously blessed the figures by issuing its standard certificate, which said that Bankers Trust's financial statements ''present fairly . . . the consolidated results of operations.'' This extraordinary episode might have ended there, with the public none the wiser. But a banana peel existed in the form of the Federal Reserve, with which Bankers is required to file periodic ''call'' reports similar to its filings with the SEC. Though the details are fuzzy, it appears that in early 1988 Bankers Trust told the Fed the truth about its foreign exchange problem and came away convinced that the Fed was willing to accept the company's unorthodox accounting solution. So the company filed a call report that once / again repeated the incorrect foreign exchange figure. This tangled web, however, was about to unravel. Sometime after the incorrect call report was filed, the Fed decreed that it would not accept a false foreign exchange figure. Bankers Trust thus came to realize that it would have to file an amended call report that corrected the foreign exchange line, reducing it by $80 million. Because the company devoutly wished not to change its net income figure, it moved to a corresponding conclusion that it would also amend the compensation figure, making it lower by the same magical and serendipitous $80 million. Call reports are public documents. Was Bankers Trust now to have in existence one public filing that said X and another -- its 10-K filing with the SEC -- that said Y? No, perhaps because the company and Arthur Young considered that an untenable position in which to put themselves. So, on July 20, 1988 -- six months after the fact -- the company startled Wall Street by suddenly announcing that, first, it would amend its call report, and second, that it would make corresponding ''adjustments'' to the 10-K, thereby changing both the fourth-quarter and full-year figures for foreign exchange and compensation. Said the company's press release: ''The corporation and . . . Arthur Young & Co. remain of the opinion that these adjustments are not material . . . under generally accepted accounting principles.'' Corporations do not lightly restate audited financial statements or implicitly acknowledge that hidden reserves may have been stashed away. Consequently, this peculiar and puzzling affair -- and the revaluation that was declared to be its cause -- sparked a brief uproar among Wall Street's banking analysts. But fundamentally they and the rest of the world remained in the dark about the deepest wrong committed. What the July press release did not in the least make clear, and what the news accounts failed to uncover, is that the bank had deliberately misstated its foreign exchange earnings in the annual report and 10-K. This is, to FORTUNE's knowledge, the first disclosure of that mind-boggling fact. The role of the SEC in this matter is not totally clear, though one fact is: The SEC did not know anything about what was going on until sometime in the summer, when it was pulled into the discussion by Bankers. At that point it obviously approved -- and perhaps even insisted on -- the restatement. What is mysterious, however, is why the SEC did not move against Bankers Trust for deliberately making an $80 million misstatement in a line item -- the foreign exchange figure -- in its 10-K. Does its lack of action imply that it is okay for companies to falsify line items? Robert E. Burns, chief counsel in the chief accountant's office at the SEC, told FORTUNE recently that the commission does not countenance deliberate misstatements in filings made with it. But he said he could not in any way comment on the Bankers Trust affair. Another ''no comment'' came from Edmund Coulson, who was the commission's chief accountant in 1988, when all this happened. Upon leaving the SEC last year, Coulson took a new job. He is today a partner in the headquarters office of Ernst & Young, which continues to handle the Bankers Trust account.