VII. Allegations that the INS and the Border Patrol Overstated Gatekeeper's Success in Internal Documents and to the Public
One of the underlying threads of the Union's allegations of fraud was that the claims of Gatekeeper's success were overstated. There were two distinct aspects to these claims. The first was directed at the Border Patrol's internal figures estimating its rate of effectiveness - referring to the percentage of the aliens attempting entry in a defined area who are apprehended. Union President T. J. Bonner contended that the supervisors' estimates of effectiveness were substantially higher than the estimates of experienced line agents. He concluded from this alleged disparity that the supervisors were intentionally providing inflated estimates of effectiveness. The second aspect of these allegations related to the INS's public pronouncements regarding Gatekeeper's success. These complaints contended that these statements gave the impression that the results achieved at Imperial Beach were similarly attained across the Sector. We address these two contentions in turn.
A. Allegations that the Border Patrol's internal estimates of effectiveness overstated Gatekeeper's actual results
One of the most challenging tasks facing the INS and the Border Patrol was measuring Gatekeeper's effectiveness. Although the numbers of apprehended aliens were well-documented, the numbers of illegal aliens who successfully eluded capture were unknown, and significant effort went into devising ways to measure them. In the hearings before the State Assembly and Congressional subcommittees, Bonner alleged that official station estimates of effectiveness significantly overstated the station's actual performance. He testified that although "experienced agents at one station," later identified as Campo, estimated that they were apprehending 1 out of every 10 or 20 illegal aliens who crossed, "the official reports reflected that they were apprehending two out of three."
Because the issue of effectiveness went to the heart of the Union's allegations - that significant numbers of illegal aliens were successfully entering the United States because agents were not being allowed to apprehend them - the OIG did not limit its inquiry to Campo. We interviewed collateral intelligence agents at each station who were responsible for calculating effectiveness rates; the station managers, who reviewed reports containing these rates; individuals at the stations and at INS Headquarters who devised the methods used to calculate these rates; and officials in Sector, INS, and DOJ management - 103 interviews in all. We also reviewed hundreds of documents reporting effectiveness estimates and/or discussing estimation methods.
1. Efforts to measure effectiveness
The problem of measuring the effectiveness of the new operation was debated from its very beginning. Management clearly recognized that apprehensions were only one part of the Gatekeeper performance equation; the more telling measure of performance was how many illegal aliens were still getting through. Teams of statisticians and analysts were formed at INS Headquarters to develop methods of measuring effectiveness. Outside consultants were hired to tackle the problem. Mathematical formulas were debated. No easy solutions were found, and to date, no statistically reliable means of calculating entries along the San Diego border, or indeed any portion of the border, have been found. As a result, proclamations of success out of INS Headquarters and DOJ have talked about achieving control of various segments of the border and have reported increases in apprehensions, but have refrained from explicit claims about gotaways or effectiveness.
While INS Headquarters was trying to solve the effectiveness puzzle, the individual stations were looking to scope reports, drag roads, sensor data, and other anecdotal information to determine just how well they controlled their respective territories. In October 1995, Sector Intel asked stations to begin reporting estimates of their effectiveness rate, figures that it soon characterized as "guesstimates." The stations complied. Because no guidelines had been offered, each station devised its own methods. For example, Imperial Beach agents and managers initially developed a formula that resulted in a figure significantly lower than their "gut feel" for the station's effectiveness rate. Attributing this to the double counting of gotaways detected by both scopes and drag roads, they simply divided the figure in half.215 Although this effort was clearly well-intentioned and not intentionally devised to overstate the station's performance, it had no statistical validity and was no better than a guess. That the agents were sometimes confused about the process was evident in some of the station reports. On occasion, a report indicated that a station was 100 percent effective even though the same report indicated that there were gotaways - suggesting a methodology that asked only whether aliens actually seen were apprehended. On one occasion, Imperial Beach reported a 106 percent effectiveness rate.
Brown Field's attempt at a formula threw in a factor keyed to whether the station was experiencing high or low levels of traffic that day, which the agents gauged by looking at apprehensions. Because this factor essentially assumed one aspect of what the formula was supposed to measure it undermined the whole exercise.216 Other stations were even less methodical. One agent at Chula Vista said the desk officer responsible for estimating effectiveness would say something like, "Hmmm, I didn't hear that much traffic was getting away on the radio. We must have been about 95 percent effective."
The methods Campo used to estimate the total number of entrants varied by agent. One simply assumed a rate of effectiveness somewhere between 10 and 20 percent. Generally, each agent considered signcutting and scope reports, and sensor activity, and then added a number to cover unmonitored areas. The reported numbers suggest that each agent's estimate of total entries was significantly affected by, if not wholly driven by, his belief of how effective the station was. In short, the exercise easily devolved into translating the agent's general sense of the effectiveness of operations into seemingly precise numerical terms.
2. Allegations that the stations intentionally inflated effectiveness rates
Bonner claimed that Campo was officially reporting a 66 percent effectiveness rate while its experienced agents were estimating it was only 5-10 percent effective. He did not specify when this happened, and did not produce any reports of 66 percent effectiveness.
The OIG's review of the Campo daily reports found no such estimates prior to Bonner's State Assembly testimony in July 1996 when he first made this claim.217 For most days the station reported a range of estimated effectiveness. For the first six months of 1996, the average daily low estimate of effectiveness was 17 percent and the average daily high estimate was 27 percent. The lowest estimate was 10 percent, the highest, 41 percent. One station report estimated that the station was 30 percent effective in the 25 percent of its territory that was patrolled. Even after the station began reporting "detected entries" and "known" gotaways, it continued to report its effectiveness as in the high 20s and low 30s, well below Bonner's claims.218 These estimates were prepared by various line agents, not supervisors. There was no evidence that supervisors ever changed line agents' figures or ordered such a change, and the estimates in the reports were consistent with estimates that agents gave the OIG. None of the witnesses we interviewed at Campo contended that the station was apprehending less than 10 percent of the attempted entries through their station boundaries in the summer of 1996, although several thought the station might have been apprehending only that percentage at the time. Even CPA Johnny Williams was aware that hundreds of aliens were crossing through Campo on a daily basis in the summer of 1996 and that the station was nowhere near 60 percent effective. His estimates put the station at 20-30 percent effective, within the range estimated by a number of the Campo line agents and managers.
In addition to giving low estimates of effectiveness, Campo's reports made no effort to paint a rosy picture of operations. They consistently reported evidence of large groups of aliens traversing the station's boundaries and complained about the insufficient manpower assigned. Similar statements were made in Sector weekly reports, including those sent to the Western region. Bonner's allegation that Campo's reports materially overstated effectiveness compared to its agents' estimates is thus unfounded.
One Sector Intel witness testified, however, that he believed that other stations inflated their effectiveness estimates to please Williams. He claimed to have been at a meeting where James Bradshaw, then head of Sector Intel, approached the APAIC at Brown Field and said, "Catching `em all?" The APAIC allegedly replied, "If we are 20 percent effective we are lucky, but the Chief wants to hear 75 or 80 so that's what we are, because I'm not stupid." In a subsequent interview, the witness said the PAIC from another station said he knew he was 15 to 20 percent effective but allegedly said, "If the Chief corrects me and says I am 70 percent effective, then I am 70 percent effective."
Neither of these claims could be substantiated. The APAIC involved denied ever making such a statement. He noted that, after trying to estimate effectiveness at Sector Intel's request, his station had abandoned the effort when it found it impossible to make accurate estimates. This is confirmed by a sample of the station's reports.219 The second individual identified as making such a statement had left the Sector before the stations began reporting daily effectiveness rates. Moreover, when his former station began calculating effectiveness rates, they were well below 70 percent.
No line agent at Campo testified to having been told of a particular, preferred rate. One ACPA recalled that Attorney General Reno had announced at a meeting approximately ten days after Gatekeeper began that she would be happy if the Sector could be 80 percent effective. The Attorney General said she did not recall ever making such a statement. WRD Gus De la Viņa testified that the Attorney General had never spoken about particular numerical goals, and that it was he who had set a goal of 80 percent effectiveness, though he "knew nothing" about how to actually measure effectiveness. De la Viņa's testimony about having set such a goal was corroborated by another witness and by notes from a September 1994 meeting at INS Headquarters.
Others may have had a similar goal in mind. According to a report from an agent in the Tucson Sector, when a Time magazine reporter asked that sector's chief what would be an "acceptable" level of measured success, the reporter was told 80-85 percent. Indeed, there seems to have been a widely-held belief in INS and the Border Patrol that such a rate of effectiveness would be needed to deter would-be crossers.
There was no evidence, however, that anyone who actually formulated the stations' estimates was aware of this goal or, if they were aware, felt constrained to conform to such an estimate. As late as spring 1996, Imperial Beach, the station considered the most controlled in the Sector, was regularly estimating its effectiveness as less than 80 percent, and the other stations were reporting even lower rates.
The OIG did find an occasion when Williams questioned a reported effectiveness rate. Upon learning, in May 1996, that San Clemente had estimated that it was apprehending only 20-25 percent of the illegal crossers within its boundaries, Williams sent an electronic message to the ACPA with oversight for the station asking, "What do you suppose they use to describe or to arrive at the fact that they are only 20 percent effective. If true, we have big problems." The ACPA forwarded Williams' message to the station, asking the basis for the station's conclusions. The acting PAIC replied that there was a mathematical error and based on the reported estimate of entrants, the figure should have been 66 percent - which in fact was closer to the number one would get by using the apprehension and total traffic figures given by the report itself (63 percent). He noted his doubts about this figure, however, because the checkpoint had been closed for two shifts, and opined that the figures are "pure supposition and not based on any facts, just an educated guess." The ACPA responded that the station should "take a long look" at what it was reporting because if it "produces numbers that are of no value and can be misconstrued we will take a hit from the critics if they get their hands on them and we/you can't explain them." No suggestion was ever made that there was an acceptable figure that should have been reported. Indeed, the acting PAIC's statement that he thought the figures were too high belies claims that station managers were inflating figures to please Williams.
We also note that when testifying before a Congressional subcommittee in June 1995, Williams declined to provide an exact effectiveness figure for the Sector, noting that others may have different figures and none could be proven. He simply indicated his belief that effectiveness had "risen dramatically," and had "at least doubled" since Gatekeeper.
In sum, we find no evidence to support the allegation that station managers were improving their effectiveness estimates to please Williams.
3. Conclusions regarding reports of effectiveness
Estimating effectiveness to any degree of certainty remains an impossible task. Despite that, the stations made an effort to report their sense of how well they were doing. While everyone did not agree with the reported estimates, agreement could not be expected. As Williams testified, regardless of what rate he offered as an estimate, someone else would suggest a different one. This highlights the difficulty of arguing that anyone was falsifying effectiveness estimates.
Sector Intel asked for precise guesses of effectiveness rates, and it received them. Although some of the guesses may have been rosier than others, we found no credible evidence that any of the guesses, high or low, were made in bad faith, or that anyone was pressured to make estimates intended to mislead anyone into believing the station was more effective than the evidence suggested.
215 The assumption that there was some double counting was probably a good one. The formula, however, assumed that every individual was counted twice - not a fair assumption. A more sophisticated analysis of the information was needed to determine how many of the persons were actually counted on both reports. It could not be assumed that this ratio was constant over time. Finally, this system did not provide for aliens who eluded detection by either the scopes or the drag road. As a result, the reported effectiveness rate probably overstated performance. It is impossible to say with certainty the extent of the overstatement. At Imperial Beach in 1996, when these figures were being calculated, however, the overstatement may not have been terribly significant because its rate of effectiveness based on objective factors such as scope and drag road reports, sensor activity, information from citizens living in the area, and other information was relatively high.
216 One problem with this methodology was that if the station was very effective and consequently had a high number of apprehensions, the formula assumed there was more traffic and concluded that the station was less effective; if the station was not very effective and reported fewer apprehensions, the formula made the station seem more effective. Another problem was that the ratio between effectiveness and traffic levels was almost certainly not linear. The station's effectiveness depended on number of agents, weather conditions, size of groups, points of entry, and other factors that had no relation to the number of individuals who attempted entry on any given day. Because there was no measurable or constant correlation between traffic levels and effectiveness, this factor increased the likelihood that the resulting figure did not accurately reflect actual performance. It should be noted, however, that one cannot determine on any given day whether this factor caused an over- or understatement of actual performance.
217 There was one day, January 25, 1996, when the station did not provide an estimate of effectiveness but when apprehensions were divided by estimated entrants resulted in a calculated effectiveness rate of 61 percent. This was the only date prior to Bonner's July testimony when the calculated rate was above 50 percent.
218 In July and August, after Bonner made this claim, there were five days when the upper range of estimated effectiveness was above 60 percent. Estimates of effectiveness later rose to this range on a steadier basis, after the station received a large infusion of agents and other resources, began patrolling its entire territory, and began a program of constantly patrolling the border and using mobile teams to intercept detected traffic.
219 Due to Bradshaw's untimely death, the OIG could not interview him.