Tax Collection in the Next Decade
The revolution could be just a few years off.
Sometime early in the nineties, according to a feasibility study by the Internal Revenue Service, the IRS could make the first mailings of a special new form to as many as 20 million American taxpayers. On the business part of this document would be printed the exact earnings of the selected taxpayers and precisely how much each owed the government in federal taxes, all calculated by the giant computers of the IRS. If the individual’s withholding for the year did not cover the tax that the IRS had decided was due, the form would serve as a bill. If withholding exceeded the individual’s legal obligation to the government, the form would serve as a refund notice.
By 1993, the IRS study estimated that up to half of all American taxpayers could choose to let the government take over the tiresome annual chore of figuring out how much they earned and how much they owed the government.
The concept of having the Internal Revenue Service assume responsibility for virtually all aspects of tax collection—including those traditionally undertaken by the individual taxpayer—has been a gleam in the eye of the IRS for more than three decades. That may explain why the agency gave the latest version of this proposal such a modest, positive, and marketable name: the Return-Free Tax System. But for a nation whose politicians have constantly invoked the unique glories of voluntary compliance, the full-scale adoption of the “return-free” process would represent a significant shift in the function of the IRS and an alteration in the basic responsibilities of a U.S. citizen.
Benjamin R. Barber, professor of political science at Rutgers University in New Jersey, has spent a great deal of his academic career writing and thinking about the importance of representative democracy. “If the IRS moves ahead with the return-free system—which seems almost inevitable—we will see a profound change in one of the basic roles of the citizen,” he said in an interview.
“The citizen will become even more passive than he is today, the government more active. Instead of the citizen telling the government what taxes he or she owes, the government, like Bloomingdale’s, will just submit its bill to the citizen.”
Barber noted that the act of filling out the tax form served a positive civic function. “It makes you wonder about your government. Where is the money going? Who are getting the special deductions? You need active, engaged citizens if representative democracy is to survive and flourish. The government’s desire to take over a key responsibility of every citizen is just one more step in the continued pacification of the American people.”
Interestingly enough, Barber’s concerns are partly shared by some officials in the IRS, although they emphasize how the process might affect tax collection rather than representative democracy. In March 1988, tax experts from all over the United States met for several days with Lawrence Gibbs, then the commissioner of the IRS. The members of the commissioner’s advisory group had come to Washington for one of their regular policy meetings. During a discussion about possible changes in IRS operations, the conversation turned to the return-free tax system.
“This involves policy questions of genuine significance,” Gibbs told the group. “It is important that the individual taxpayer identify with his or her return. If we ever get to the point where taxpayers came to see tax returns as government returns, rather than their returns, it would have vast implications for the whole tax collection process.”
Ever since the adoption of the modern income tax just before World War I, government and business leaders have joined in celebrating the American people, who, they claimed, paid the taxes they owed on a voluntary basis. The virtuous nature of the home-grown taxpayer was often compared with the devious nature of French and Italian taxpayers. Never mind that systematic withholding made this patriotic boast a somewhat doubtful proposition. Voluntary compliance has long held an elevated position as one of the small but important myths of the nation.
Whatever the current reality of taxpaying, however, the public acceptance of the return-free tax system would mark an important development in the evolution of the passive citizen. At the same time, the IRS’s implementation of the plan would provide concrete evidence that technology could radically alter the shape, purpose, and even power of this tradition-bound agency.
Despite the enthusiastic lip service paid to the spirit of voluntary compliance, the IRS has long yearned for a more intrusive, all-knowing kind of tax system. As far back as July 1953, the agency’s planning staff completed work on a secret study “to eliminate the need for filing tax returns by wage earners when tax is withheld by employers.” The first benefit of the 1953 proposal, the planning document said, would be to reduce “the burden of the wage earner.” But the plan gave equal weight to a second goal: The project would “safeguard the revenue” by “reaching more people for annual tax accounting.”
The IRS’s 1953 plan to collect sufficient information about the citizenry to enable it to calculate the tax obligations of a large part of the population did not fly, probably because of the impossible bureaucratic burden of handling the necessary paperwork in the precomputer age.
But three decades later, in November 1984, the proposal was resurrected by Donald Regan, then secretary of the treasury, in a report to President Reagan. The report bore the title “Tax Reform for Fairness, Simplicity and Economic Growth.” By this time, Regan and the IRS had come up with the catchy soft-sell advertising phrase for their revolutionary plan. “The return-free system” was now possible, the report said, because the IRS’s improved information-processing equipment would allow it to calculate the tax liabilities of millions of Americans. As a result, Regan called for the initiation of a system “under which many individual taxpayers will be relieved of the obligation of filing an income tax return.” While the 1953 Treasury Department documents show that increasing the IRS’s power to collect the revenue was an important factor behind the government’s interest in the project, the 1984 report skips this essential element and only mentions the burdens it would lift from the shoulders of the nation’s taxpayers.
In late 1984 and 1985, Roscoe L. Egger, Jr., the IRS commissioner at that time, continued to beat the drums for the return-free system, which he called a “very exciting proposal.” By 1990, he predicted, two out of three taxpayers “would never have to wrestle with a tax return again.”
The Reagan administration’s enthusiasm for the return-free system is interesting. Remember, President Reagan was elected partly because of his promise to get the government off the citizen’s back. Here his treasury secretary and tax commissioner were boosting a plan that would vastly broaden the supervisory role of government over tens of millions of taxpayers.
Congress has displayed little concern about the system one way or another. While it did not buy the full project as proposed by the Treasury Department, the final version of the 1986 Tax Reform Act contained a brief section authorizing the IRS to initiate an experimental project to test the “return-free” concept. It was under this authorization that the agency completed the 1987 feasibility study suggesting that the return-free system could be operating by 1991.
But the IRS, with no specific authorization from Congress, already had cranked up a smaller computerized tracking operation that in many ways resembles the return-free tax system. In a brief press release handed out in January 1987, the agency said that this more modest operation had two apparently simple goals: to improve the detection of those individuals who didn’t file any income tax returns and then to automatically calculate what they owed the government.
Probably because the announced targets of the new system have always been considered to be members of a loathsome class, genus tax cheat, the new IRS plan received almost no public notice.
At first glance, the new Internal Revenue Service operation does seem relatively harmless. But the already functioning “automated assessment system” is of interest because it provides a rough preview of the return-free tax system.
Here is how this increasingly effective process works. First, at relatively little cost to the government, IRS computers have been programed to look through the 1 billion third-party financial reports the agency now routinely receives each year from a large range of private and public organizations. Next, the computers commit all these randomly recorded payments to their memory, eliminate the duplicate names, and prepare a list of everyone in the United States who received a salary from an employer or interest from a bank or alimony from a former husband or dividends from a corporation or a royalty from a publisher or rent from a tenant or a refund from a state tax agency or one of the many other kinds of payments. Then the computers match the list of Americans receiving money from all these sources with the list of those who paid their taxes. The final product: a third list made up of individuals who were on the first list but not on the second, people who reportedly had earned money from one or more sources but then failed to file an individual tax return.
The second step of the automated assessment procedure is equally challenging. The computers take the list of all the miscreants who apparently have failed to file an income tax return, add up all the income they reportedly received during the past year, and prepare an income tax return complete with the taxes they allegedly owe. The machines then mail the returns they have prepared to the individuals who failed to prepare them and dispatches a series of follow-up notices if they are required.
The 1986 start-up of this complex automated detection system by the nation’s largest and most powerful civilian agency clearly represented the culmination of a major series of technological advances in the administration of the decades-old federal income tax. At the same time, however, the establishment of this sophisticated computer system targeting a relatively small number of deadbeats was an interim step. First an automated system to calculate the tax for the relatively small number of Americans who don’t file any return at all. Then the far more ambitious system designed to calculate the returns of half the taxpayers in the United States.
In the first two and a half years of operation, the IRS automated assessment system identified 827,319 individuals and businesses that apparently failed to file any tax return at all. Furthermore, the system then calculated that these nonfilers owed the government nearly $3 billion in federal taxes. Because mailing a bill is a lot easier than persuading a debtor to cough up the taxes, however, the actual collections for the period, $220,799,000, were far less that the original assessments.
“If the program works the way we want,” said William Wauben, assistant IRS commissioner for collection, “the next few years will see a significant drop in the number of people who previously had just walked away from their obligations and paid no taxes. We think the new system will push people toward voluntarily complying with the tax laws.”
The system is a technical marvel.
But the system behind the system is just plain astounding. Beginning about fifteen years ago, several congressional committees began to sense that developments in computer technology had moved along to a point where the government could seriously consider launching a tax enforcement technique that it had long dreamed about. Although IRS officials initially disagreed with this judgment, they slowly began to change their minds, eventually becoming committed converts.
The legal requirement that every employer report how much it paid its regular employees and withhold a portion of the paycheck for taxes has been in place since 1943. From the very beginning, IRS statistics show that this reporting and withholding law has guaranteed that most of the country’s working stiffs, those who have no source of income other than what they earn from their employers, pretty well meet their federal tax obligations.
But for those who received part or all of their income in the form of dividends, interest, or royalties, it was a different story. It is true that in 1939 Congress had approved a law requiring banks and corporations to report to the government the interest and dividend payments they were making to their customers and stockholders. But except for two very brief periods, Congress has never required the banks and corporations to withhold the federal tax that was due on these kinds of payments.
The law thus has always tended to punish lower-income salaried workers and reward upper-income investors. For without the bite of withholding, the bark of the so-called 1099 reporting forms had minimal impact until the age of the computer. This was because the small number of banks and corporations that bothered to meet their legal requirements initially submitted the 1099 reports on little pieces of paper. And even that massive agency the IRS did not have sufficient funds to hire the thousands of clerks it would have needed to match the hundreds of millions of interest and dividend reports with the correct income tax returns.
But continuing developments in computer technologies have changed that. Now, thirty items of information contained on every one of the 102 million individual tax returns filed each year are immediately transferred to an electronic master file. Now, a significant proportion of all the 1099s and other third-party income reports is submitted to the IRS on tape or computer disk.
This transition from paper to computer has had a significant impact on the IRS’s enforcement efforts. In 1986, agency computers compared information found on 818.6 million of the 1 billion third-party reporting documents with information on the 102 million tax returns filed that year by individual taxpayers.
Not only did this ability to read and compare massive amounts of data with extraordinary speed enable the IRS’s automated assessment system in one thirty-month period to identify nearly 1 million Americans who did not file any tax return at all. It also allowed the IRS to spot an additional 3.8 million taxpayers who had filed returns but had failed to include one or more of these third-party payments as part of their total income.
Aside from the concerns of Barber and former Commissioner Gibbs about altering the general attitude of the public toward its government, the IRS’s headlong rush into the computer age raises many other genuinely serious worries.
GARBAGE IN, GOSPEL OUT
First is the apparently simple question of accuracy.
The massive size of the agency’s programs means that a very small problem can very easily result in wrongful government actions against hundreds, and in some cases even millions, of innocent people. One recurring hazard within the IRS, and other large institutions, is the computerized procedures that inadvertently lead to the generation and easy distribution of incorrect information. Depending upon the number of affected taxpayers and the government action actually triggered by the false information, such projects can sometimes lead to serious violations of constitutionally guaranteed rights, such as the Fifth Amendment’s fundamental promise of due process.
A few years ago, for example, the IRS sent a notice to a woman in Austin, Texas, asserting that she had failed to report $4,000 of her income and demanding $800 in back taxes, penalties, and interest. It turned out, however, that the IRS’s computer-generated assertion was dead wrong. Nor was she the only person who was falsely accused.
The problem began when a local bank purchased some software, or computer instructions, to help it comply with an administrative reporting requirement of the IRS. The new software was supposed to guide the bank’s computer through the time-consuming chore of preparing a list of the names and addresses of all the customers who received interest and the amount they had received. The bank then gave the list to the IRS. Because it was on computer tape, and thus did not require error-prone clerks to retype the required information, the list was assumed to be correct. But sad to say, the software was faulty, and the bank had inadvertently informed the IRS that two hundred of its customers had received interest when they had not. These incorrect reports, of course, triggered the IRS’s computers to kick out two hundred incorrect dunning notices.
The rash of incorrect notices generated by the faulty software was not discovered by the Austin bank or the IRS but by an enterprising newspaper reporter. The trail, however, did not end with the single Texas bank and its two hundred customers. In fact, further investigation revealed that the faulty list-making software had been sold to at least fifty other companies all over the United States. These companies, in turn, also had sent incorrect income reports to the IRS.
In fact, by the time the computer glitch had been identified, the companies using the faulty software had sent the IRS incorrect income information concerning approximately 1 million Americans. Because of the wide differences in the income levels of the taxpayers incorrectly identified as having received payments, the IRS to this day insists that it is unable to calculate how many wrongful dunning notices were generated as a result of the false information it had been provided.
It seems likely, however, that the agency’s bum notices went to a minimum of tens of thousands of mystified citizens. It also seems likely that these tens of thousands of men and women spent hundreds of thousands of hours searching their financial records, made tens of thousands of angry telephone calls, and wrote tens of thousands of confused letters trying to collect the information they needed to rebut the false information in the IRS documents. Finally, it probably can also be assumed that some unknown proportion just went ahead and wrote a check for the additional taxes that the IRS had incorrectly said they owed. After all, a lot of people feel it doesn’t make any sense to fight city hall.
(The question of who should be held accountable in such situations actually is one of the more interesting problems of the computer age, a period when a growing number of large bureaucracies seem to be actively involved in more and more of the intimate decisions of each of our lives. Should the company that wrote the bad software be held liable? Or the companies that bought the bad software and sent the incorrect information to the IRS? Or the IRS, which dispatched the inaccurate dunning notices? How can anyone ever be held accountable in this very common kind of mess?)
But when inaccurate information entered into the IRS system triggers more serious enforcement activities such as levying a taxpayer’s bank account or seizing a home, even the most obtuse federal judge may begin to worry. The IRS, of course, insists that adequate protection is provided taxpayers by its elaborate notice system. Given the huge numbers of notices involved, the high mobility of the American people, and the somewhat quirky performance of the U.S. Postal Service, however, it is obviously not possible to assume that the notices will always get through.
Because inaccurate records are extremely costly in terms of the efficiency of the agency and the goodwill of citizens, it is nice to hope that the IRS would work very hard to avoid all kinds of errors. A long string of reports by the agency’s own internal audit team and Congress’s General Accounting Office suggests that such hopes may be naive. The worry that senior IRS administrators view the accuracy of records as a secondary problem is reinforced by the lackadaisical attitude of another large federal agency: the Federal Bureau of Investigation. In 1981, about twenty years after the FBI created a national computerized network to link tens of thousands of local, state, and federal law enforcement agencies, a congressional research office conducted a rigorous first-of-its-kind audit of the computerized entries that the FBI system was zipping about the nation. The finding: 54.1 percent of the entries transmitted were inaccurate, incomplete, or misleading. This was an astonishing finding because of the serious consequences of the flawed records. Innocent people were being arrested, guilty people were being set free, patrol officers were unnecessarily endangered by genuinely dangerous people who were not correctly identified. Individual victims living in Massachusetts, New Jersey, Michigan, Louisiana, and California have brought suits charging that their constitutional rights had been violated when incorrect information in the computers led to their false arrest. Although at least one local police department, Los Angeles, has been ordered to pay damages because of such incorrect information, the FBI has so far managed to avoid any direct liability. The bureau argues that it is not responsible because the inaccurate entries it transmits were created by state and local groups, not by the FBI. By this logic, of course, it certainly would be unfair to hold the IRS accountable for any incorrect information it picks up from other federal agencies, the states, or any other source.
No similar audits of the accuracy and completeness of the data that trigger IRS enforcement actions have ever been made public. But remember the GAO report that almost half of all the official letters the IRS dispatched to taxpayers contained inaccurate or misleading information. This finding is worrisome.
THE COST OF PROGRESS
Hank Philcox is in charge of what the IRS calls information systems development. As part of this overall effort, he and his staff are hard at work on Tax System Redesign (TSR), a twelve-year, multibillion-dollar project intended to fundamentally alter the IRS’s tax collection process by the year 1998. It is a massive job, involving the analysis and reorganization of hundreds of different procedures that will affect every taxpayer and IRS employee in the nation. The basic goals of TSR sound prosaic enough. Minimize the IRS’s reliance on paper documents. Develop modern data bases through which almost all information collected by the IRS about every taxpayer can be instantly examined by an agent anywhere in the United States. Provide expert systems to help agents better understand tax laws. “We want to modernize the processing and support systems so that IRS employees have the information they need to better serve the public,” Philcox said in a lengthy interview.
“With the full-blown system, a tax assister anywhere in the country will be able to look up an account, and say, whoops, there’s been a mistake; we’ve got to get this corrected, and do it right then and there. And the correction would be instantly posted to the national file.”
Philcox views in a completely positive way the massive increase in the IRS’s computer power, the improved linking of its data bases by secure lines of communication, and the steady growth in the use of computers. More specifically, Philcox believes that the technical revolution he is quietly masterminding will at the same time reduce the cost of collecting taxes, improve the ability of the IRS to investigate tax fraud, and provide the public with better service.
“We’re drowning in paper now,” he said. “The heart of what we’re trying to do is modernize the system so it does not require an excruciating effort for our people to get at the information we routinely collect. We’re focusing on the elimination of many of the manual and paper processes we now have so we can avoid a lot of the labor-intensive procedures that are extremely costly and inevitably lead to a certain percentage of errors.”
Philcox outlined the complex steps the agency now follows when a revenue agent in one of the IRS’s local offices decides it is necessary to obtain detailed information from a taxpayer’s return that has not been recorded in the computerized summary statement maintained at the National Computer Center in West Virginia. “The opportunity for error in that process is enormous,” the official said. “Right now it’s all we can do to make a ten percent error rate.”
It is worth noting that Philcox, the official responsible for convincing Congress and the public that the massive new system is needed, is one of the few senior IRS officials who openly discuss the serious inadequacies of current procedures that generate an astonishingly large number of errors.
The IRS’s initiation of the Tax System Redesign project in the mid-1980s was not the first time that the agency has sought to systematically enhance its computer operation. More than a decade before, in the early 1970s, the IRS informed Congress that it would require between $750 million and $1 billion for a proposed new computer project called the Tax Administration System (TAS).
But that time, in contrast with the eighties, strong concerns among a number of House and Senate members that the revolutionary new computer system might develop into “a system of harassment, surveillance and political manipulation” led Congress to delay its funding.
The general congressional worry about the potential of TAS to affect the rights of citizens was fueled in part by the rash of disclosures about the various ways the Nixon administration had marshaled the powers of a number of federal agencies, including the IRS, for political purposes. But the specific doubts about TAS were formally articulated by a critical analysis of the system published in March 1977 by the Office of Technology Assessment (OTA), a research arm of Congress. Shortly after a brilliant OTA analyst named Marcia J. MacNaughton completed the critical study of TAS, the incoming Carter administration killed the IRS computer project.
As described by IRS officials today, the computerized procedures envisaged by TSR are very similar to those of TAS. But as the IRS in the last few years has methodically gone about the task of developing the somewhat updated version of TAS, not a single House or Senate member has questioned its potential for mischief. One explanation for this silence may be the general belief that growing federal deficits represent such a serious threat to national security that all questions must be suspended. Another explanation may be that many of the computer applications that together make up the TSR package have already been adopted by a number of large private institutions, such as American Express.
In response to a specific question, Philcox said that the IRS has not developed an overall estimate of what the twelve-year TSR project will cost. “TSR is being constructed in incremental stages during a period of rapid technological change and thus we haven’t attempted to estimate the total dollars that will be required.”
But in late 1988 the IRS’s planning division developed a five-year strategy for only the first half of the 1990s. The plan identified a wish list of fifty-six specific actions that IRS managers said they hoped to accomplish during the period. Twenty-one items were computer and telecommunication projects that appeared to be related to TSR. The estimated cost of the twenty-one items was about $11 billion, ten times more than the TAS system rejected by the Carter administration.
One defense of large government computer programs sometimes offered by technology buffs is that, if many major companies in the private sector have not misused the information they collect, is it fair to be suspicious of agencies like the IRS? Because the IRS has a range of powers unavailable to American Express, TRW, and other major information-hungry companies, the logic behind this argument is obviously specious. On the other hand, it must be acknowledged that such projects as the return-free tax system and TSR do have the potential of providing the American people more efficient, more accurate, and fairer administration of the tax laws.
Another flaw in the “American Express defense” is that the return-free system and TSR are not the only technical changes the agency is developing to discipline society. In fact, both are only small discrete parts of a far broader revolution that affects almost every aspect of our lives.
FRANKENSTEIN’S MONSTER?
Since the 1920s the IRS has provided state tax agencies with various kinds of detailed information about the tax payments made by their citizens to the federal government. The states used the federal information to identify individuals and corporations that might not be paying their state taxes. Then, in the early 1980s, it dawned on the IRS that a great deal of the information held by a variety of state agencies might help its enforcement efforts.
Now, with the gradual computerization of all levels of government, this systematic sharing of information has exploded. In 1988, for example, a Treasury Department report listed more than two hundred systems of records, most of them computerized, that the IRS was regularly sharing with the states.1 The lists contained an astonishing range of information about hundreds of millions of individual and corporate taxpayers. The same report listed 172 systems of records that various states were routinely sharing with the IRS.
Although the IRS had been giving the states tax information for many decades, it was not until 1976 that Congress passed the first law specifically authorizing what until that time had been an informal process. The law contains some curious provisions. One section, for example, specifically prohibits the IRS from passing the data to the governors of the states. As a matter of state law, of course, governors direct the agencies receiving the tax information that they are excluded from obtaining. One does not have to be overly skeptical to wonder how the directors of many state tax agencies who have been appointed by a governor would respond to a gubernatorial order for a peek at a particular tax return?
Another curious and perhaps well-meaning aspect of the law is the broad power it gives the IRS to establish and enforce the computer security standards of all state tax agencies.
For many years, the IRS has maintained a single office with responsibility for supervising the disclosure of tax information. At the time I was exploring the questions involved in its sharing of federal returns with the states, this office was headed by Arnold Gordon. “The federal tax information is terrifically helpful to the states,” Gordon explained. “This is because a large majority of them lag far behind the IRS in income tax enforcement. Some states have no income tax enforcement at all. So our tapes are wonderful for them, really a system of enforcement that doesn’t cost them a penny.”
Every year, for example, all but one or two states now receive a computer tape containing all the information on the IRS’s Individual Master File (IMF). This tape gives state tax agencies more than thirty items of information about every federal-tax payer living in their states, including the taxpayer’s name and Social Security number; the number of exemptions claimed; an indication of the child-care credit claimed; total itemized deductions claimed; wages reported; interest income earned; total income tax paid; gross pensions and annuities reported; gross income from small businesses, partnerships, estates, and trusts; and the amount of taxes withheld.
But as we know, that’s just one of the more than two hundred tax files that the IRS shares with the states. A second tape is called the Individual Returns Transaction File (IRTF). Among many other matters, it tells the states how the IRS’s computer system ranked each individual return for audit potential, the state income tax each individual reported, alimony that has been paid or received, capital gains distributions, employee business expenses, and IRA and Keogh payments.
Then there’s the Business Master File (BMF) tape. It includes the receipts, deductions, income, and reported taxes of every business in each state; the number of states to which each business made unemployment insurance contributions; the total income each business withheld from its employees; and all sorts of useful identifying numbers. A second computer file pulled from the agency’s Audit Information Management System (AIMS) allows the states to obtain the names and federal identification numbers of out-of-state corporate taxpayers that have operations in a state and have been audited by the IRS.
Another source of information is the Revenue Agent Reports (RARs), summary statements that are completed every time an IRS agent audits an individual or corporation. RARs have for many years been available on paper. Now, they too go to the states on easy-to-match computer tapes.
The details of information sharing are worked out by state tax officials and an IRS district disclosure officer. Because some states have more than one agency that collects taxes, the IRS has completed arrangements with more than ninety different state agencies concerning exactly what information will be shared and how it will be protected.
Over the objections of the IRS, Congress recently passed legislation authorizing the agency to provide federal tax information to New York City. Currently, a number of other large cities desire the same data. Gordon’s explanation of why the IRS felt that information should be shared with the states, but not the cities, is less than convincing. It also suggests the flimsy nature of the privacy safeguards ostensibly in place to prevent the misuse of the sensitive information. “We’ve always been uneasy about the cities because they are so political,” he said. “The states, most of them, have long histories of professional tax administration.” Everyone knows, of course, how sneaky political calculations never undermine the actions of governors.
Possibly to assure me about the IRS’s vigilant concern, Gordon also acknowledged that the IRS had uncovered a number of cases where states have not met the federal security requirements. He declined, however, to name the offending jurisdictions. “In one state,” he recalled, “we discovered that all computerized information was stored in a single data base which could be accessed by all of the different agencies of the state. At least in theory, this meant that the state police could have obtained all the tax material that we provided that state.” Gordon did not explain how the IRS had allowed the state to even receive the data without spotting such an obvious security problem.
The easy availability of federal tax information, as well as the growing sophistication of computer technology, has encouraged the states to develop programs designed to identify nonfilers and other kinds of potential enforcement targets by comparing names, addresses, and a variety of other information recorded on the computer tapes of a number of different government agencies.
The rapid growth of this investigative technique is illustrated by a research paper presented to the National Association of Tax Administrators by the Pennsylvania Department of Revenue.2 In 1975, the report said, Pennsylvania ran one match that compared the names of those paying state income taxes with those paying federal income taxes. The goal, of course, was to identify those not paying state taxes. A decade later, Pennsylvania was regularly conducting twenty-six separate matching applications and was planning to start at least a half dozen more.
Pennsylvania’s overall effort is indeed impressive. The granddaddy of the state’s matching programs, the annual computerized cross-check of all the individuals listed in the federal and state income tax files “results in 140,000 new accounts [investigations] and approximately $700,000 in additional revenue each year.” Many of the programs, however, are much more narrowly focused. “Using medicaid payment tracking tapes provided by the Department of Public Welfare, the Department of Revenue has reviewed about 600 physicians, dentists, osteopaths, etc. for income tax compliance. Approximately $163,000 has been collected or billed for three different years.”
A pilot program comparing the names of a sample of four hundred licensed attorneys with the state’s income tax files “collected about $8,400 in delinquent collections.”
Under an old Pennsylvania law, all state contractors, consultants, and vendors are supposed to be in compliance with state tax laws. Before the computer age, it was impossible to enforce this law. The report showed how the times they are a-changing. “In order to determine compliance, we obtained computerized listings of contractors and vendors from the individual agencies. Both letter dunning and field contact of a small group of sample accounts resulted in the collection of $150,000. The extension of this program will result in a procedure to ‘set off’ delinquent liabilities against current contract amounts.”
The IRS, working through the Federation of Tax Administrators, hopes to get a somewhat modified version of the Pennsylvania project going in all the other states that collect income taxes. Under the IRS plan, each state would require any person who provides goods or services to the state or applies for a license to certify under penalty of perjury that the individual has met the state’s tax obligations. With this certification system in place, the IRS believes, states could spot their delinquent or nonfiling taxpayers and then make those names available to the federal tax agency for its enforcement efforts. To further enhance the value of the certificates, the IRS hopes to persuade the states to maintain their lists of certifying contractors and license seekers in a uniform computer format that could be easily read by the federal machines.
Sometimes federal agencies having nothing to do with taxation collect information that can be useful to state tax collectors. The Pennsylvania Department of Revenue, for example, obtained aircraft registration tapes from the Federal Aviation Administration in Oklahoma City and compared the names of owners with the list of individuals paying state income taxes. This match “has resulted in collections of about $1.1 million in unpaid Sales and Use taxes over the last year.”
Despite the reach of Pennsylvania’s current computer-sharing efforts, however, matching in the state is still very much a growth stock. The Revenue Department’s report said that it recently had arranged for the Pennsylvania Department of State to redesign the forms that all applicants must fill out either to obtain or to renew a state license. There currently are 600,000 professionals holding various kinds of licenses from Pennsylvania. The new forms will include the Social Security numbers of all such professionals and will allow easy use of the names for a range of other matching projects.
Among those that the Pennsylvania report said would soon be examined on a regular basis were all state employees, licensed insurance brokers, and licensed security brokers. Can barbers be far behind?
In the past, the IRS has tended to look down on state tax agencies, often viewing them as unsophisticated country bumpkins. But in recent years the IRS’s attitude has changed as it gradually began to see how it might use all kinds of data, even those gathered by the states, to identify taxpayers who might be trying to bend the law. The agency has not limited its collection efforts to tax information. Just as Pennsylvania has moved to obtain federal data not directly connected to taxes—the tape of licensed aircraft from the Federal Aviation Administration—so the IRS has begun to collect federal, state, and county information concerning matters outside its direct responsibilities.
The IRS’s systematic drive to collect additional information from the states and a variety of other institutions was launched in 1984 with the adoption of the agency’s 204-page Strategic Plan.
One part of the plan, for example, called for the creation of an indexed directory that could help IRS agents access information gathered by the investigative activities of other federal agencies.
“We are aware that the Board of Governors of the Federal Reserve System, the Federal Deposit Insurance Corporation, the Federal Home Loan Bank Board, the National Credit Union Administration and the Office of the Comptroller of the Currency conduct examinations” that warrant “consideration by Service officials for possible tax implications.”
Another part of the action plan noted that the agency in the past had relied too much on its own information to spot taxpayers who apparently were not meeting their obligations. The plan called upon the agency’s managers to “develop additional resources of information to detect non-compliance through cooperative arrangements with the states.”
The federal-state arrangements have not been limited to the process of exchanging computer tapes. As described in part in the chapter on collecting taxes, the IRS has sought to create permanent communication links with a variety of local, state, and private organizations with no obvious connection to tax collection.
IRS offices in several parts of the United States, for example, have developed on-line links with the computerized directories now used by the telephone companies. Although the information in an electronic directory appears to be similar to that in a printed telephone book, the computer has wrought extremely useful, and surprising, changes in the process of locating an individual. Several years ago, Pat Callahan, then an IRS special projects manager, described the benefits of electronic directories to a conference of tax administrators. The specific subject of her talk was called Scantel, a service then offered by Mountain Bell.3
“Benefits resulting from Scantel are its increased speed, its increased success rate and the more valuable data it provides,” Callahan said. With the new system, IRS agents were two times more likely to locate the individual they were looking for than with the old one. “At the same time, one hour of Scantel work replaces ten hours of city directory reach,” she stated.
Callahan said that Scantel, unlike a printed telephone book or reverse directory, was updated on a weekly basis; allowed for on-line access searches by name, specific address or range of addresses, and telephone number; and conducted the searches over a multistate area. One unique feature was a signal that flashed on when the person the IRS was seeking had an unlisted telephone number. While Mountain Bell had a policy of not providing unlisted numbers to the IRS without a warrant, the signal definitely showed that the targeted individual still resided in the region. Another useful feature was Scantel’s ability to give IRS investigators the names and telephone numbers of the immediate neighbors of a particular target.
The computer-linking process is a gradual one, with many advances and occasional retreats. The IRS district in Dallas, Texas, for example, was one of the early pioneers in scooping up local records, seeking to establish direct electronic links to the computers of eighty counties within its boundaries. Dallas County, for example, an area with more than 1.6 million residents, was one of the first to sign up. Under this contract, about two thousand IRS agents were currently able to make nearly instantaneous checks about the property owned and the property taxes paid by every person in the county; the names and addresses of all persons with registered vehicles; the make, year, and weight of those vehicles; and the name and address of every registered voter. Marlene Gaysek, at that time an IRS public affairs officer in Dallas, said that the computer links would save the agency about $200,000 a year because lower-paid clerks, rather than field agents, would be able to gather the required information.
Gaysek said that the IRS did not intend to compile new federal lists from county information. “We’re not getting this information for general matching programs; we are using the direct access to track specific taxpayers. We need this information when we file a lien against someone or want to check to make sure a taxpayer’s financial statement is correct.”
On this particular occasion, however, outspoken public opposition forced the IRS to back away from its ambitious plans to form an intricate areawide data base. While officials in Dallas County had welcomed the direct links with the federal government, neighboring Tarrant County, the area around Fort Worth, refused to take part. “This was just another extension by the federal government to increase its power over local government,” said B. D. Griffin, then a Tarrant County commissioner. Probably for the first time in Texas history, Tarrant County found itself walking down the same side of the street as the local office of the American Civil Liberties Union. James C. Harrington, an ACLU lawyer in Austin, said that even though the information the IRS was receiving by computer was public, its use of the county data conflicted with one of the prime goals of the Privacy Act, guaranteeing that the information an individual provides for one purpose will not be used for a second purpose without the individual’s permission. “We generally oppose this kind of cross-computerization,” Harrington said, “because despite what the IRS says, history tells us that information collected by the IRS will be compiled into a giant centralized data base.”
The strong negative views expressed by both ends of the political spectrum must have been persuasive because some months later the local office of the IRS actually unplugged some of the computer links to the county data bases.
But the growing intimacy between federal and state tax administrators is not limited to the sharing of income tax information and the imposition of uniform federal security standards on the hundreds of state computers that process federal tax information. On May 7, 1985, the IRS and the National Association of Tax Administrators met to discuss the extent to which fuel suppliers around the country were not paying both federal and state excise taxes. As a result of these discussions, the IRS and five major states—California, Florida, Louisiana, New Jersey, and New York—launched a series of experimental investigations of suspect dealers to test the effectiveness of joint enforcement efforts.
Two years later, the IRS issued a highly favorable report on the experiment. “State results were generally better than the average historical results. IRS results were significantly better than the historical totals.” For the states, the primary benefits occurred through an exchange of examination leads and audit results. For the IRS, the advantages included immediate access to the abundance of information contained in state excise tax returns and various state registration and license records.
“While logistical difficulties existed, the consensus was that the joint examinations were highly desirable,” the report concluded. One key reason, in addition to the collection of substantially increased revenues, was that the joint audits generally were found to be of higher quality and to take less time to complete than separate audits.
Many are the ties that bind. Joint investigations of suspected tax cheats by the IRS and the state revenue departments. Regular exchanges of IRS and state computer tapes frequently containing highly private information about millions of taxpayers. Systematic development of direct computer links between IRS collection offices and a variety of state and county agencies. Congressional approval of a law requiring the IRS to establish federal computer-security standards for the state tax agencies and then to ensure that the agencies are meeting these standards.
What is one to make of these four developments? One possible conclusion is that federal, state, and local tax agencies, taking advantage of the tide of technology, have improved their ability to identify the individuals and businesses that are not paying their dues to society.
A second conclusion is that the tax laws are today enforced by an increasingly unified army made up of 123,000 federal and more than 60,000 state tax agents. The technology allows it. The revenue needs at every level of government require it. The law-and-order psychology of tax administrators encourages it.
A unified army, of course, is almost always more effective than one that is disorganized. But the Founding Fathers of this nation, aware of the abuses of power that had originally kindled their revolution against England, believed that the agencies of government could have too much unity. As everyone knows, they sought to avoid the hazards of concentrated power by deliberately dividing the new federal government into three competing units; the legislative branch, the executive branch, and the judicial branch. In addition to this horizontal separation of powers at the federal level, the founders also envisaged a strict vertical separation: a society where the federal government was the creature of the states and where powers not explicitly given to the federal government were reserved for the states and the people.
In late 1986, a special panel created by President Reagan’s White House Council on Domestic Policy issued a report on the status of federalism in the United States. The panel noted that the founding generations of Americans, and those that followed, were “acutely aware of danger to liberty posed by the concentration of government powers in a central government.”4
Despite a general awareness of this danger to freedom, the Reagan administration study said, the “framers’ vision of a limited government of enumerated powers has gradually given way to a national government with virtually unlimited power to direct the public policy choices of the states in almost any area. The States, once the hub of political activity and the very source of our political tradition, have been reduced—in significant part—to administrative units of the national government, their independent sovereign powers usurped by almost two centuries of centralization.”5
Over the years, however, the American interest in local control has remained a potent force for at least some parts of government. Consider, for example, our more or less continuous debate about crime. While a long line of presidents and presidential candidates has denounced crime and talked about issues like capital punishment, the Mafia, and drug dealers, it is the locally elected sheriffs and locally appointed police chiefs who remain in command of the war on crime. One product of this consensus on local control has been Congress’s explicit decision to limit the jurisdiction of the FBI to a relatively narrow area of crime control. No one, left, right, or middle, openly supports the development of a national police force.
But with almost no public debate, and very little congressional guidance, a somewhat amorphous but still powerful federal/state organization appears to be slowly co-opting the enforcement of the nation’s tax laws, thereby undermining the traditional notion of federalism. There is no claim here that the IRS is seeking to assume control of the state tax agencies. Even if this were a secret IRS goal, the tax collectors in such powerful states as California, New York, and Texas are not likely to agree placidly to such an arrangement. What does appear to be emerging, however, is a subtle meshing of federal and state agencies into a unique enforcement alliance with a formidable range of legal powers and resources of information.
Arnold Gordon, in charge of improving the flow of data between the IRS and the states, does not believe this steadily expanding process has reduced the independence of state organizations or diluted the structure of federalism. “The proof of this,” Gordon explained, “is that under the law the IRS for many years has had the legal authority to collect and process the income taxes of any state who wanted to use our services. Not one state has picked up on this option. The reason is obvious. Any state that signed up would lose some of its sovereignty. The information sharing we now do increases everyone’s efficiency. It has no impact on sovereignty of the states.”
Gordon exaggerates. The federal enforcement of computer security requirements on the states obviously has some impact on their sovereignty. So does the steady pressure on the states to store their tax information in forms that are compatible with IRS computers. Because the IRS will naturally be the dominant partner every time the agency and a particular state decide to initiate a joint tax-enforcement project, the balance of power in tax enforcement seems to be gradually drifting from the state capitals to Washington.
There is also the question of legislative oversight, an important part of the checks and balances essential to American representative democracy. How could the House Ways and Means Committee and the Senate Finance Committee which historically have failed to supervise tax collection at the federal level, ever keep track of tax collection managed by a confusing consortium of more than fifty agencies? And how about the tax committees of New York or California? Would the IRS, which often resists answering questions put to it by both the president of the United States and the U.S. Congress, be willing to meet with tax committees in Albany or Sacramento to answer questions about its misuse of state tax data? It seems highly unlikely.
A SUSPECT IN EVERY KITCHEN
In the tax enforcement game, it is not only the lineup of players that seems to be gradually shifting. Also in flux are the basic tactics of the game. During the last two decades, for example, the IRS has been properly engaged in a variety of efforts to identify the fairly small proportion of Americans who seek to avoid any taxes by dropping out of the economy. But in the spring of 1983, Walter E. Bergman, then the agency’s deputy assistant commissioner for planning, finance, and research, had an idea that carried to an entirely new level this search for improved techniques for detecting those who do not file tax returns.
Computerized mailing lists are now a major tool of American commerce. Many lists contain far more detailed information about the living habits, political views, and health of the American people than is generally understood.
The first source of information for many lists is the telephone book. The directory, of course, gives the researcher the names and addresses of most of the people living in a given city. But it can also be the key to other insights.
Several marketing companies, for example, have computer programs that automatically assign each name and address in the directory to the correct census tract, the Census Bureau’s basic geographical unit. Because American neighborhoods tend to be homogeneous, this matching allows the researcher to make a well-informed guess about the income, age, race, and family makeup of the individuals who live in a given tract.
Researchers use the telephone directories for a second not-so-obvious purpose. In most cities, directories are published every few years. By checking previous editions, the listmaker determines how long a given family has lived at the same address.
The automobile registration lists maintained by state departments of motor vehicles are another excellent source of information. By combining the information drawn from the telephone directory with that provided by the departments of motor vehicles of thirty-six states, the commercial listmaker learns that the Jones family living at 55 Elm Street has two cars. (Some states for either technical or privacy reasons do not sell the lists.) They also learn the age, type, cost, and model of the two cars. If the Joneses happen to own a recent-model station wagon and a luxury convertible, for example, the marketing company can infer that they are wealthier than a neighboring family which only owns a low-cost hatchback.
State departments of motor vehicles also maintain lists of licensed drivers. From this source, the marketing company can learn the age and race of Mr. and Mrs. Jones and their two teenage children. (In some states, the company may also be able to determine that Mr. Jones has diabetes and that Mrs. Jones wears glasses, two conditions that might affect their driving and also can be useful to advertisers.)
Other computerized public records available in various states allow the marketers to determine whether the Joneses are registered voters, which party they belong to, whether they live in an apartment or a house, the size of their living space, and whether they have a dog or a cat.
On the basis of all this information about the life-style of the Jones family—where they live, how long they have lived there, how many children they have, and what kinds of cars they drive—the marketing company develops a detailed computerized portrait of the family, including an estimate of their annual income.
One of the largest and best listmakers in America is the Donnelley Company. In 1988 the company was merging up-to-date information about 75 million specific families or households. The information was drawn from a number of sources, including 4,700 telephone directories, Postal Service change-of-address cards, and the motor vehicle departments of thirty-six states and the District of Columbia. The telephone information is updated on a daily basis. The automobile-related information is updated once or twice a year.
The managers of the Donnelley Company residential data base try to develop thirty-five sets of information about each of the 75 million families. One of the most interesting sets is called Interest Categories. It shows whether the members of a particular family hold a bank card or a travel card, their level of response to past mailings, and whether they have contributed to religious organizations, politicians, or groups such as the American Cancer Society.
It was this life-style information that Walter Bergman, the IRS researcher, thought might help the government track down people who had not paid their taxes. At first glance, Bergman’s surveillance project sounds quite reasonable. Using the agency’s computers, the IRS would compare the names and addresses of all taxpayers with the names and addresses of all the households identified by the marketing companies.
“If the cross-check suggests a family hasn’t paid,” Bergman told me in an interview, “we’ll make an inquiry to find out why. This is no big deal.”
But the matching project was not so innocuous as Bergman suggested. One of those who saw problems was Robert Ellis Smith, a lawyer and the publisher of a specialized newsletter called the Privacy Journal. “The IRS experiment is very troublesome,” he said in an interview. “While I am quite sure it does not violate the law, it graphically demonstrates the growing links between government and private computers. National lists of households and their incomes are sufficiently accurate for soliciting business, but that doesn’t mean they are precise enough to trigger investigations.”
Interestingly, the three major companies that prepare such national lists agreed with Smith’s concern. “It is inappropriate for the IRS to use the kind of lists we produce to identify errant taxpayers,” said Richard Vincent, then director of marketing for the Donnelley Marketing Service. “This IRS experiment is ill conceived because such lists are not accurate on an individual basis, but only in the aggregate. If a company wants to send a mailer to all American families with incomes over $40,000, we rent it a list for onetime use. Depending somewhat on the group that the company wants to target, we guarantee that 75 to 80 percent of those receiving the material will have the correct characteristics.”
Because of these reservations, and concerns that IRS use of the mailing lists might ultimately damage their business, Donnelley and two other companies, R. L. Polk and Metromail, refused to give the agency any names. Several months later, however, a small independent list broker in Washington provided the IRS with a list of 2 million names of households in Brooklyn, New York, Wisconsin, and Indiana, the areas Bergman had selected to test the feasibility of his project.
But the problem of accuracy identified by Robert Ellis Smith and the three major marketing companies is only the first level of concern. Under the American system of law, a policeman may obtain a search warrant from a judge after he has presented evidence that there is “reasonable cause” to begin an investigation. But because of the speed of the computer, this time-honored tradition has been subtly modified.
In fact, the IRS began with its normal assumption that everyone was guilty and the computer allowed it to conduct a general search of the records of every taxpayer in the three districts. Because the courts have held that such records belong to the government, and not to the citizen, they have quite consistently ruled that broad matching projects do not violate the Fourth Amendment prohibition against unreasonable searches. Despite the legal logic of these decisions, however, the steadily growing ability of the IRS to access and compare the computerized files of millions of Americans in hopes of spotting inconsistencies represents a profound change in law-enforcement strategy.
Appropriately enough, Bergman’s experimental research project was conducted in the year Orwell made famous, 1984. This time it did not work. Although the IRS never issued a report explaining why, it appears that industry experts were correct: The 1984 version of the marketing lists contained too many inaccuracies to serve as a cost-effective surveillance device.
Some experienced marketing experts, however, think that the project may have failed because of the notoriously inaccurate information contained in IRS files. “Well over 95 percent of the names and addresses on the current lists are accurate,” said a senior New York marketing executive with wide experience in using the lists to target customers. “There’s no reason why the IRS couldn’t use something like the Donnelley list to reduce the U.S. population to mush except that the agency can’t control the accuracy of the information in its own files.”
However, the improved information technologies resulting from the Tax System Redesign project, parallel developments within the commercial marketing companies, and theoretical work by a number of academics on a computerized analytic technique called “block modeling” suggest that Bergman’s dream may well come true in the near future.
Of course, the IRS has already begun to work toward this goal. For more than two decades, the agency’s Taxpayer Compliance Measurement Program has collected detailed tax information from a representative sample of taxpayers to develop statistical models predicting the behavior of different groups of taxpayers.
But with the increasing amounts and kinds of information being collected and stored on computers, research to produce far more sophisticated models of expected behavior is inevitable. The possibilities for such research have long fascinated a small circle of law-enforcement officials and academic experts. In December 1981, Roger H. Davis, a special agent in the Behavioral Science Unit at the FBI Academy in Quantico, Virginia, wrote a paper describing how “social network analysis” can be used by police to understand the nature of complex conspiracies, determine the leaders of secret organizations, and ultimately “predict criminal behavior.”6
Davis described a case where the police in an unnamed western city obtained information about an organized fencing operation that was selling guns to criminals. The operation was being run by the members of a local gang. As the first step in its investigation, the police established a secret surveillance of a tavern known to be popular with the gang. The surveillance was continued for four days.
“From this surveillance, 18 people believed to be connected with the group were identified,” Davis reported. “Using social networking techniques, the officers converted their observations of people arriving at and departing from the tavern into a network diagram showing the structure of interpersonal relations within the group. From this picture, police determined connections between group members and began to focus logically on those they considered suspects and who would be most knowledgeable about the crime.”
Davis added that social network analysis could be particularly useful for investigating cases involving organized crime, the distribution and sale of illicit narcotics, illegal gambling services, and business fraud.
Although network analysis is hardly routine among either federal or local law-enforcement agencies, its potential has long fascinated a small group of sociologists and mathematicians. In 1975, Scott A. Boorman, then at the University of Pennsylvania; Ronald A. Breiger, then at Harvard University; and Phipps Arabie, then at the University of Minnesota, published a highly technical paper in the Journal of Mathematical Psychology describing an advanced method for the hierarchical clustering of relational data that could give researchers a new tool for analyzing the behavior of various groups of individuals. Research for the paper was supported by grants from the National Science Foundation and the National Institute of Mental Health.
On November 20, 1983, after a two-year stint at the IRS, Boorman, by then a professor of sociology at Yale University, and Paul R. Levitt, a research mathematician at Harvard, published a brief article in The New York Times about what they now called “block modeling,” a computer programing technique to organize various bits of seemingly innocuous data in ways that allow researchers to predict the behavior of individuals with improved accuracy. Boorman and Levitt said that block modeling did not require particularly sensitive information such as medical histories. “Rather, it exploits the unexpected, even uncanny synergy of large masses of ‘relational’ data buried in organizational files. Examples of relevant data: Whom you talk with in your company, whose phone calls you do not return, whom you eat lunch with, to whom do you send carbon copies of memos.” They noted that the increasing use of the computer meant that more and more such data were being routinely collected in ways that made the information readily available for analysis.
The researchers said that the output of block modeling was simple to understand. “As Justice William O. Douglas once observed, a person is defined by the checks he writes. Blocks in block modeling generalize this principle. They are discrete sets of people occupying similar positions in the relational networks, and who thus are likely to behave similarly in ways important to the organization, and be candidates for receiving similar ‘treatment.’”
Boorman and Levitt argued that, with proper safeguards, block modeling could provide managers valuable insights about the dynamics of large groups of individuals in a sound, ethical, and constructive way. They warned, however, that “strides in the new ‘guilt by association’ technologies are easily outstripping the vastly slower evolution of protective legal and administrative responses.”
Remember Hank Philcox, the assistant IRS commissioner for Tax System Redesign? Remember Philcox’s overall plan to reduce sharply the IRS’s reliance on paper? Remember the return-free tax system under which the IRS will calculate the returns for half the taxpayers in the nation? Remember the shared concerns expressed by Benjamin Barber and Commissioner Gibbs about how these new processes might affect the basic role of individual citizens? Remember all the matches with which the IRS is routinely comparing its information against the information in the files of the states? Remember Walter Bergman’s 1984 experiment to use commercial marketing lists to spot nonfilers? Remember all the computerized information the IRS is collecting from the states and counties?
Where do these developments take the IRS during the next decade? While no one can predict the future, it is always tempting to try. In 1985, the Office of Technology Assessment, a research arm of Congress, undertook a broad study of how information technology was changing the federal government. In connection with this study, the OTA requested a small Washington research group headed by Joseph F. Coates to extrapolate how five federal agencies would be functioning in the year 1995. One of the agencies was the IRS. Although Coates and his research team assumed full responsibility, the scenarios were all based on interviews with federal officials, and each agency was given an opportunity to comment before the report was published.7
The IRS mission in 1995, Coates said, would still be to collect taxes. “Although ‘voluntary compliance’ will remain the byword, by 1995 the potential for complete internal matching of returns from source information, electronic funds transfer networks, and near complete computerization of even small-scale financial transaction records could make compliance a matter of automatic processing of taxpayer and IRS electronic records rather than a separate voluntary filing.”
Electronic filing, return-free filing, optical data storage, and automated examination will have become a routine part of American life. “As these technologies become commonplace they will change the flow of information and responsibility among taxpayer, employer and financial institution and government,” Coates continued.
In the government’s continuing effort to collect additional taxes, Coates predicted, increasing attention would be focused on tapping the incomes of those who, to various degrees, try to live outside the regular economy. He estimated that by 1989 various illegal entrepreneurs, primarily drug dealers, would be earning as much as $45 billion a year in unreported income.
To capture the tax from this elusive group, Coates foresaw that the government could adopt a drastic law-enforcement strategy with far-reaching implications for the entire society: the printing of machine-readable money. Combined with optical scanners located at every bank, Coates said, such money would enable the IRS to monitor citizen transactions of almost any size.
Legal and technical developments already are carrying the IRS down this road. In 1970, Congress approved legislation requiring banks to report every deposit or withdrawal of more than $10,000 in cash. For many years, this law was largely ignored. Recently, however, computerized record keeping has made compliance less onerous, and in 1987 the banks sent 5.4 million detailed reports to the IRS showing the name, Social Security number, and other details about those involved in large cash transactions. In recent years, IRS reporting requirements have been extended to anyone—a lawyer, auto dealer, or jeweler—who is paid $10,000 or more in cash.
While such reporting requirements obviously are useful to the IRS, Coates believes that they would be far more effective when combined with a monetary system that could constantly monitor the movement of individual bills above a certain denomination in the same fashion as the government now can follow a personal check. During the 1988 celebration of Australia’s bicentennial, a special $10 note was issued that might provide the technical answer to how such monitoring might be achieved. The note is printed on a special plastic that includes a hologram portrait of Captain James Cook. While the Reserve Bank of Australia said that this “optically variable device” with Cook’s likeness would make counterfeiting almost impossible, Coates said that the experimental bills easily could be adopted for an automated system that tracked cash.
Coates also predicted that as the IRS continued to enlarge its national data bases with new kinds of information the data bases themselves would become increasingly useful to all kinds of institutions, public and private. As their value increased, IRS efforts to protect them would require such stringent security measures as more or less continuous surveillance of all IRS employees, biological identification systems such as voiceprints or retina prints, and increased psychological evaluation and screening.
“The need for privacy and security of IRS data must continually be traded off against the usefulness of that data for non-IRS purposes,” Coates observed. “Data sharing can range from the tracking of specific, known criminals, to searching for a small set of suspected child support evaders, to wholesale matching of IRS and Social Security Administration data bases. Data sharing within the government is more restrictive than in the private sector. What is the individual’s right to privacy of his own data when that data may help provide information on criminals?” he asked.
Assume, for argument’s sake, that we live in a world where every bit of information in every one of the IRS’s computers is accurate and that every IRS notice is correct. Do the American people want a world where much of the personal information collected by the IRS is routinely shared with state and local tax agencies? Do the American people want a world where the IRS, to improve its enforcement efforts, has initiated a long-range plan to gain direct access to a broad range of information collected by the states during the registration of automobiles and voters and the licensing of drivers, doctors, and other groups? The American people have long opposed the idea of the federal government creating a single national data base containing a computerized portrait of every citizen. But it is clear that a de facto system is already in place.
Because this growing electronic net is being tossed over the country in the name of such worthy causes as fighting crime and catching tax cheats, the nation’s current leaders have not responded to the challenge of men like Professor Emeritus Joseph Weizenbaum of MIT or the late Sam Ervin, the constitutional scholar and Democratic senator from North Carolina. Senator Ervin believed that computer technology threatened to erode liberty and the creative, spontaneous spirit of the American people.
Unless new legislative controls can be devised to limit the reach of the computer and unless federal officials can be persuaded to limit their hunger for more and more information about the activities of the American people, Ervin once warned, individuals will gradually stop exercising “the freedoms that are calculated to make their minds and spirits free. And that, in the long run, the government is going to suffer from the effects of this as much as the citizens are to suffer the loss of their freedom.”8
Almost 150 years before the worrisome thoughts of Ervin and Coates, a Frenchman named Alexis de Tocqueville came to the United States and expressed some of the same concerns. Tocqueville, of course, celebrated the great American experiment. But he also predicted that the egalitarian spirit of its people might someday lead to the growth of a despotic centralized government. Such a government, Tocqueville wrote in Democracy in America, would be very different from the despotisms of old, where the savage power of the emperors was extremely onerous to a few, but largely ignored the many. The despotism of the new democratic nations, he believed, unlike the off-with-their-heads style of the ancients, would be “more extensive and more mild; it would degrade men without tormenting them.”9
Tocqueville anticipated that the new government’s power over its subjects would be minute, regular, and provident. “For their happiness, such a government willingly labors, but it chooses to be the sole agent and only arbiter of that happiness, it provides for their security, foresees and supplies their necessities, facilitates their pleasures, manages their principal concerns, directs their industry, regulates the descent of property, and subdivides their inheritances; what remains but to spare them all the care of thinking and the trouble of living? Thus it every day renders the free exercise of the free agency of man less useful and less frequent, it circumscribes the will within a narrower range and gradually robs a man of all the uses of himself.”10
Having subverted the individual citizen, the government then extends its arm over the entire community. “It covers the surface of society with a network of small complicated rules, minute and uniform, through which the most original minds and the most energetic characters cannot penetrate, to rise above the crowd. The will of man is not shattered, but softened, bent and guided; men are seldom forced by it to act, but they are constantly restrained from acting. Such a power does not destroy, but it prevents existence; it does not tyrannize but it compresses, enervates, extinguishes, and stupefies a people, till each nation is reduced to nothing better than a flock of timid and industrious animals, of which the government is the shepherd.”11