D. Allegations that agents were ordered to turn back aliens rather than apprehend them

In his August 9, 1996, Congressional testimony, Bonner alleged that illegal aliens were turned back into Mexico without being apprehended so as to keep apprehension numbers low. This tactic is known as "TBS" or "Turn Back South," and involves chasing aliens towards the border until they cross back into Mexico. Bonner claimed that supervisors ordered agents to use this technique because apprehension figures were too high, but offered no evidence as to how often the TBS tactic was used or how many aliens had been turned back south.

Three Imperial Beach agents testified that supervisors advocated the use of a TBS strategy to keep apprehension numbers low. A Union official at Imperial Beach claimed that supervisors had told him, "[w]e don't need the apprehensions. Chase them back if you can." A former Imperial Beach agent testified that "at one time" agents were encouraged to chase aliens south rather than apprehend them. He said supervisors told the agents that "Washington" wanted deterrence, not apprehensions. Another Imperial Beach agent testified that TBS was the preferred strategy to keep apprehensions low. Although, he believed that supervisors wanted agents to use the TBS tactic to keep apprehensions low, he conceded, however, that no supervisor had ever issued any specific instruction to use TBS.

Two Chula Vista agents also said they believed TBS was being used to keep apprehension numbers low. One stated that agents were told to "play defense" and to push the aliens back into Mexico. The result was lower apprehension numbers, which made the FOSs happy. He conceded, however, that agents sometimes use TBS for safety reasons - for example, to avoid being assaulted by aliens.

Thirty-four other witnesses told us that turning aliens back south was a legitimate strategy. When aliens far outnumber available agents, for example, chasing the aliens south instead of trying to apprehend them makes sense. If agents in such circumstances attempt apprehension, many of the aliens will inevitably escape. In contrast, herding the aliens towards the border will prevent successful entry.

These witnesses also stressed the importance of agent safety near the border. When agents try to apprehend aliens close to the border fence, they face a significantly greater risk of assault and injury. Because the Gatekeeper strategy requires agents to patrol close to the border, the need to turn aliens south instead of trying to apprehend them has increased.

Most of the witnesses we interviewed emphasized that the TBS strategy was used - for both safety and containment reasons - long before Operation Gatekeeper. They testified that the decision to use this strategy in any given situation is made by the agents involved, not because of a predetermined policy announced by supervisors. Finally, they emphasized that agents do not physically force aliens over the fence; instead, aliens are maneuvered into a position where they must choose between apprehension and retreat. In such circumstances, most aliens choose retreat.

One Union official alleged that use of TBS was a "radical departure" from the Border Patrol's previous practice. According to this witness, prior to Gatekeeper, if an alien set foot in the United States the objective was "to grab that person, apprehend them, write them up, [and] create statistics for the Border Patrol." This witness believed that any change from this approach constituted an effort to mislead.

While this witness conceded that agents used this tactic prior to Gatekeeper, he argued that it was not then condoned by management. No other witness, including any of the other Union officials whom we interviewed, shared this witness's view on this subject. Indeed, other agents told us that TBS was viewed as a legitimate, useful tactic prior to Gatekeeper.

We discovered no formal written documentation or set of rules regarding TBS. We did see references in station daily reports, however, indicating that groups of aliens had been turned back south. This information generally came from scope reports that noted how many individuals had been observed, how many apprehended, how many turned back south, and how many "gotaway." It is clear to us that the use of this tactic was openly recorded in reports sent to the Sector.

Based on the testimony of a broad range of supervisory and line agents, we conclude that TBS is a legitimate strategy to prevent entry close to the border. The problem of agents being assaulted close to the border is well-documented. If it is safer for agents to force aliens to retreat to Mexico rather than to apprehend them, agents should be free to use this tactic. Similarly, if agents can contain a large group of aliens and force them to return to Mexico, agents should be free to force the aliens to retreat. The primary goal of Gatekeeper and the Border Patrol's operations everywhere, of course, is to prevent illegal entry. Apprehension should be the fallback option when deterrence and prevention have not succeeded. If TBS permits agents to achieve the primary objective more successfully and safely than apprehension, then agents should use TBS.

Although a few agents believed that supervisors wanted them to use TBS as a way to keep apprehension numbers low, they provided no evidence that supported this suspicion. No agents, for example, showed us that supervisors had ordered them to use TBS in inappropriate situations, such as a long distance from the border. There is no question that if aliens are turned back south rather than apprehended, apprehension numbers will be lower. The issue is whether supervisors were ordering agents to TBS aliens in inappropriate circumstances as a means of keeping apprehension numbers low. We found no evidence of any organized effort by supervisors to encourage agents to use the TBS tactic and no evidence that supervisors had ordered its use in inappropriate circumstances.113

Given that TBS is a legitimate strategy along the border and that there is no evidence that supervisors ordered agents to abuse this strategy, we reject as unsubstantiated Bonner's claim that agents were told to TBS aliens so that apprehension statistics would be lower.

E. Allegations that roving patrols were eliminated to keep apprehension figures low

In the summer 1996 State Assembly and Congressional hearings, Bonner alleged that traffic checkpoints were understaffed and that agents were prevented from apprehending aliens who circumvented the checkpoints. As a result - Bonner charged - the effectiveness of the checkpoints has been "greatly diminished," and they no longer accurately reflect how many aliens are evading agents at the border. On July 25, 1996, similar allegations were made by other agents in a radio talk show and in The Californian, a North County newspaper.

Most of these allegations concerned the Temecula Station, which is responsible for inland checkpoints on several northbound routes leading out of the San Diego area. The Station operates a primary checkpoint on northbound I-15, a second permanent checkpoint on Highway 79 in Oak Grove, and several temporary checkpoints on other roads that may be used to circumvent I-15. Prior to Gatekeeper, this station also conducted patrols for both foot and vehicular traffic attempting to circumvent the I-15 checkpoint.114  These random patrols are known as "roving patrols." If an agent observes a vehicle that he believes contains undocumented aliens, he stops the vehicle to investigate further.

During the hearings, Bonner complained that agents were no longer allowed to conduct roving patrols and that smugglers could now avoid the I-15 checkpoint without risk of apprehension. He claimed that management eliminated these patrols in order to reduce apprehensions. Moreover, because the two northbound checkpoints - San Clemente and Temecula - were considered barometers of the amount of traffic that successfully eluded the border defenses, undermining the effectiveness of these checkpoints would reduce apprehensions and make it appear that the border defenses had become more effective.

On a July 25, 1996, radio program, Ralph Boubel - a Temecula agent, Union official, and Bonner's assistant at the National Border Patrol Council - claimed that agents were being instructed not to stop vehicles except at checkpoints.115   He claimed that the checkpoints are "the only barometer we have to show what is actually getting through the San Diego line," that the roving patrols were an important complement to the checkpoint system, and that management had banned these patrols to lower apprehension numbers. He further contended that Temecula Station only had enough manpower to staff two or three of six proposed checkpoints, and that several checkpoints would be left unmanned. He also claimed that most commercial smugglers circumvent checkpoints by choosing alternative roads. An unnamed agent quoted in the July 25, 1996, edition of The Californian echoed these concerns.

The OIG interviewed 19 Temecula agents and 18 other witnesses regarding the use of roving patrols at Temecula. Two Temecula agents told us that they believed that roving patrols were curtailed to reduce apprehensions and make Gatekeeper appear effective. The first claimed that most of Temecula's apprehensions stemmed from roving patrols, so that curtailing these patrols was likely to dramatically reduce apprehensions. He acknowledged, however, that accidents resulting from vehicle pursuits - and extensive critical media coverage of these accidents -had also likely played a role in the decision to curtail roving patrols. The second agent said that while roving patrols were never explicitly forbidden, after Gatekeeper agents were criticized for the grounds they relied on to make each vehicle stop and the paperwork required to justify vehicle stops became "overwhelming." He said he felt that the Border Patrol had overreacted to recent court decisions regarding vehicle stops. He contended that agents "in general" assumed that the criticism of their vehicle stops and their reports was due to Gatekeeper and management's desire to make the new program appear effective.

Our investigation revealed that three factors significantly influenced Temecula's use of roving patrols. First, a series of serious accidents during vehicle pursuits led to debate over vehicle pursuit policies. Second, recent decisions issued by the U.S. Court of Appeals for the Ninth Circuit had announced stricter guidelines for vehicle stops. Third, certain Congressmen had become interested in the checkpoints and begun to demand changes in their operations. We address each of these factors in turn.

Agents conducting roving patrols sometimes encounter drivers who refuse to stop their vehicles upon demand. A refusal to stop has occasionally led to high-speed pursuits and fatal accidents. In June 1992, five U.S. citizens - four teenagers and one adult - and one undocumented alien were killed in an accident caused by a vehicle driven by undocumented aliens fleeing from Border Patrol agents. This accident received extensive media coverage, and the Border Patrol was strongly criticized for its vehicle pursuit policies and forced to defend a lawsuit. Although this was not the first such accident, the death of the teenagers represented a turning point for Border Patrol managers. One witness testified that the then DCPA was badly shaken by the accident, and would have "done anything" to avoid a repetition. The witness said that the DCPA had been working on a revised vehicle pursuit policy at the time of the 1992 accident and felt that the accident could have been prevented if he had fought harder for a new policy.

After additional accidents took place in 1994 and 1995 during vehicle pursuits, the PAIC at Temecula announced new rules governing when agents must terminate vehicle pursuits. An April 20, 1995, memorandum containing the policy revisions stated that the changes were necessary "to diminish agent and agency liability." The new rules at Temecula were identical to a vehicle pursuit policy implemented at El Cajon Station two months earlier.116

A Temecula Union official immediately objected to the new Temecula policy, contending that it changed working conditions and therefore could not be implemented without Union approval. The policy was subsequently rescinded. A similar policy was reissued in June 1995 as a "supervisory guideline," however, and the Union responded with an unfair labor practice charge against the Sector. In early October 1995, the PAIC formed a working group to evaluate roving patrol stops. The group was directed to develop methods to ensure officer safety and compliance with the legal criteria for a stop. This group discussed potential civil liability and the possible impact of an adverse court decision on the viability of the checkpoints. It also announced two new rules: all traffic observation units had to be staffed with at least two agents, and all vehicle stops had to be radioed in.

Numerous witnesses, including the Commissioner, Sector management, and Temecula agents, mentioned the 1992 fatalities as the crucial factor in the Sector's decision to curtail roving patrols. William Veal, the DCPA at the time of the 1992 accident, testified that he disapproved of roving patrols because they frequently led to vehicle pursuits that resulted in accidents. He stated that the priority of the station was to operate the I-15 checkpoint and respond to calls from other law enforcement agencies that had apprehended undocumented aliens. CPA Williams told the OIG that he likewise disapproved of roving patrols because they were more dangerous than checkpoints and not as effective; while checkpoint personnel can stop and inspect all traffic, roving patrols can only stop traffic based on an articulable suspicion.

We conclude that concern about potential vehicle pursuits and resulting accidents was a major factor in Sector management's efforts to change Temecula's roving patrol policies. We also conclude that this concern began to influence the Station's operations long before Gatekeeper was initiated.

While concerns about accident injuries and fatalities were rising, several court decisions also began to affect the use of roving patrols. On April 28, 1994, the Ninth Circuit issued an opinion declaring that a particular vehicle stop during a roving patrol by San Diego Sector Border Patrol agents was made in bad faith and violated the Fourth Amendment.117 In responding to this decision, then CPA Gus de la Viņa announced a policy requiring that all I-44 forms (Record of Apprehension or Seizure) from roving patrol stops be submitted to the appropriate ACPA for a factual basis review. In April 1995, one ACPA sent a memorandum to the PAIC at Temecula indicating that a review of I-44 forms had revealed "numerous very questionable narratives." The memorandum reported that supervisors were not properly reviewing I-44 forms and were "oblivious to legal guidelines necessary in these cases." It noted that the Deputy Chief was "looking hard" at all cases for prosecution submitted by the checkpoint stations. Later that same month, the Ninth Circuit held that a vehicle stop by Livermore Sector agents was unconstitutional. These two cases were part of a series of adverse court decisions that raised the level of suspicion required to justify a vehicle stop.

In May 1995, CPA Williams sent a memorandum to the INS Western Regional Counsel indicating that the Sector had curtailed roving patrols and reassigned agents to checkpoints. He noted that although roving patrols have "historically proven to be highly effective in the interdiction of undocumented aliens and contraband, we felt restricted by Ninth Circuit decisions to the point of curtailing this method of operation." While noting that checkpoints were more manpower intensive, Williams stated that they were legally justifiable and appeared more acceptable to the Ninth Circuit. Rather than relying on roving patrols near the checkpoints, Williams directed that operational efforts should be focused on intercepting undocumented aliens in areas near the border, before they reached smugglers' vehicles. Finally, CPA Williams noted that roving patrols continued to be used where necessary to control illegal traffic, and that the Border Patrol viewed these patrols as an important law enforcement tool. Williams commented, however, that agents had difficulty applying the Ninth Circuit's standard for reasonable suspicion.

This and other documentary evidence demonstrates that the more intensive review of I-44 forms was designed not to reduce apprehension numbers but to ensure that the Sector complied with the legal requirements imposed on roving patrols by the Ninth Circuit. Because mobile checkpoints did not present the same legal uncertainties as roving stops, management's decision to emphasize the former enforcement tool was logical. We further find that this factor also began to influence Border Patrol operations prior to Gatekeeper.

Congressional action was the third factor that strongly influenced Temecula Station's use of roving patrols. In September 1994, Congressman Ron Packard proposed transferring agents and equipment then in use at checkpoint stations to positions along the border. He subsequently argued that if the checkpoints were not in operation 24 hours a day, they should be closed permanently. Congresswoman Lynn Schenk also began to lobby for the closure of the checkpoints, arguing that the agents would be more effective if stationed at the border. In response, INS conducted reviews of checkpoint effectiveness, including closing the I-5 checkpoint and temporarily transferring its resources to the border. After conducting these studies, INS concluded that the checkpoints were effective and should be kept in operation.

In 1995, Congressman Packard introduced legislation that mandated closure of the San Clemente and Temecula checkpoints which were not in constant operation. By mid-1996, INS reached an agreement with Packard that the checkpoints would operate full-time except when conditions became unsafe. This compromise meant that the checkpoint stations would have first call on personnel; only after the checkpoints' staffing needs were satisfied could remaining personnel be assigned to alternative enforcement efforts. In October 1996, CPA Williams sent a memorandum to the PAICs at San Clemente and Temecula announcing that because of an agreement with Congress the checkpoints must operate on a continuous, 24 hour-a-day basis. Williams noted that other enforcement efforts should be curtailed or suspended as necessary to ensure that adequate personnel were available at all times to conduct lane inspections.

Witness testimony was generally consistent with the documents described above. The PAIC and others testified that roving patrols had been curtailed but not completely eliminated. The witnesses made clear that CPA Williams wanted Temecula to deploy mobile checkpoints on the roads used to circumvent the I-15 and Oak Grove checkpoints rather than conduct roving patrols. He and other supervisors believed that the mobile checkpoints were safer than roving patrols because they did not involve vehicle pursuits. Furthermore, mobile checkpoint operations were far less likely to be successfully challenged in court. Although several agents told us that the Border Patrol could not operate enough checkpoints to be effective, management countered that mobile checkpoints could be moved from road to road to keep smugglers off-guard. Although managers conceded that roving patrols were effective, they contended that safety and legal concerns made mobile checkpoints preferable.

Memoranda from the second half of 1995 and 1996 indicate that Temecula was working on alternatives to roving patrols. Sector Counsel and station management met to develop policies and in January 1996 Temecula supervisors were instructed to set up mobile checkpoints instead of assigning agents to roving patrols. By May 1996, Sector management had issued clear directives that Temecula personnel should not conduct roving patrols. All of the relevant documents we reviewed linked these directives to the safety and legal issues discussed above.

The Union, however, continued to adamantly oppose efforts to curtail roving patrols, contending that management could not prevent agents from making vehicle stops. The Union filed a series of unfair labor practice claims and announced that its members would not cooperate with any efforts to curtail roving patrols. In October 1995, a Union steward told Temecula management that agents were conducting vehicle stops without notifying the station, and were not terminating pursuits as instructed. He contended that management could do nothing about such conduct. In May 1996, a Union steward sent an electronic message to Temecula's APAIC asking who was saying that agents should not make vehicle stops; the steward commented, "Why do they think that this time we will comply?" In opposing the new policy, the Union never mentioned the impact it would have on apprehension numbers or on perceptions of Gatekeeper's success. The Union did not link these concepts until allegations of fraud had become public.

While overwhelming evidence supported management's claims that the elimination of roving patrols was based on safety and legal concerns, Bonner and Boubel alleged that managers curtailed these patrols in an effort to reduce Temecula's apprehension numbers. They assumed that roving patrols were responsible for most of Temecula's apprehensions, and that eliminating these patrols would significantly decrease apprehensions at the station and make Gatekeeper appear successful.

We could not confirm that roving patrols had historically accounted for the overwhelming majority of Temecula's apprehensions. The figures submitted by the station to Sector Intelligence did not differentiate checkpoint apprehensions from roving patrol apprehensions. What information we were able to obtain did not suggest that roving patrols were responsible for most apprehensions at Temecula. For example, a September 1995 electronic message regarding two days of apprehensions at Temecula provided separate statistics for checkpoint apprehensions and non-checkpoint apprehensions. The total number of apprehensions for each day was similar, but on one day most of the apprehensions came from roving patrols, while on the other two checkpoints were responsible for most of the apprehensions. On the day that roving patrols provided most of the apprehensions, one checkpoint was never open and the other was open for only a half-day. The next day, when both checkpoints were fully operational, most apprehensions came from the checkpoints. Although the data in this document is too limited to provide a basis for generalizations, it does not suggest that checkpoints are less effective than roving patrols. The first extended report we discovered that separated checkpoint apprehensions from other apprehensions was for August 1996. That report indicated that no more than 27 percent of Temecula's apprehensions came from sources other than the checkpoint, other agencies, or sensors. It did not specify how much of the 27 percent was due to roving patrols.

We also reviewed a series of Gatekeeper Updates, monthly reports prepared by Sector Intelligence. These reports repeatedly stated that most of Temecula's apprehensions were produced by roving patrols. The reports, however, contained no statistics that supported this claim. Moreover, when we asked Sector Intelligence personnel for any statistics regarding how many of Temecula's apprehensions stemmed from checkpoints and how many were produced by roving patrols, they could not provide any such information.

Although we could not determine the percentage of apprehensions produced by roving patrols, witnesses generally conceded that if roving patrols were expanded to previous levels, apprehension numbers would increase. That, of course, does not necessarily mean that these patrols were curtailed to reduce apprehensions. While we concluded that the curtailment of roving patrols was fully justified by safety and legal concerns - regardless of the impact of this action on apprehension numbers - we sought to determine whether a concern about apprehension numbers played a role in the decision to make this policy change. We found no such evidence. Indeed, we found no evidence that management expressed any concern about Temecula's apprehension levels during Gatekeeper.

Bonner and Boubel argued that apprehensions at San Clemente and Temecula were the sole barometer of Gatekeeper's success at the border, and that accordingly management felt pressure to keep apprehensions low at these stations. CPA Williams conceded that apprehension statistics at the checkpoints were one barometer of success at the border, but contended that there were many others, including the number of undocumented aliens in Imperial Beach neighborhoods, sensor hits, information from scope reports, citizen calls, and apprehensions at the border stations. Furthermore, because the checkpoints are inland, illegal alien traffic that entered in East County or even further east might be apprehended at the checkpoints. As a result, apprehensions at the checkpoints might increase even though traffic had decreased in the westernmost stations.

Although apprehensions at Temecula have declined each year since 1994, this decline could result from improvements in border operations; it is not necessarily evidence of an effort to stop Temecula agents from making apprehensions. Furthermore, if witnesses are correct that roving patrols accounted for as much as 80 percent of apprehensions, eliminating these patrols should have resulted in an immediate and sustained decline in apprehensions.

We saw no evidence of such a sudden and sustained decline in apprehension numbers at Temecula. In fact, the largest percentage change in apprehension figures from a previous month was an increase, not a decrease. More importantly, at the time that the station was allegedly attempting to reduce apprehensions, it created mobile checkpoint positions on each of the alternative roads and used them randomly in an effort to catch smugglers. Such efforts would undermine any plan to decrease apprehensions and they contradict any inference that management curtailed roving patrols in order to reduce apprehensions.118  Finally, because Temecula accounted for only about three percent of Sector apprehensions, fine-tuning apprehensions there would have little measurable effect on the Sector's total apprehension numbers.

In sum, we found no evidence indicating that Sector management eliminated roving patrols in order to reduce apprehensions and make Gatekeeper appear successful. Overwhelming evidence demonstrated that legal and safety concerns, and Congressional pressure, motivated the shift in operating strategy at Temecula.119

F. Allegations that backup positions were eliminated to keep apprehension numbers low

When Gatekeeper was first implemented at Imperial Beach, agents were deployed in three parallel, half-mile wide tiers, with each successive tier allowed increasing mobility to apprehend alien traffic that had successfully penetrated the previous tier. The use of tiers of agents rather than a single line along the border was what distinguished Operation Gatekeeper from the "Hold the Line" program. In the initial Imperial Beach plan under Gatekeeper, 20 agents were assigned to positions in the first tier; 30 agents were assigned to the second tier; and 26 agents were assigned to the third tier. Everyone we interviewed agreed that the second and third tier positions were not always fully staffed, but the witnesses disagreed as to when these positions were not staffed and why.

In the June 23, 1996, North County Times article, Joseph Dassaro, a Union official at Imperial Beach, alleged that "[t]he first tier is often left without backup. This in itself would reduce apprehension statistics since illegal crossers who get through are not very likely to get caught." In his summer 1996 hearing testimony, Bonner echoed these comments, alleging that the three tiers have "for the most part degenerated into a single line of defense on the immediate border."

One Union official from another station told the OIG that agents had told him that the third tier at Imperial Beach was eliminated to decrease their effectiveness, and that Imperial Beach had enough manpower to staff the third tier but had chosen to transfer third tier agents to first-tier positions along the border. This witness limited this particular complaint to Imperial Beach, which he claimed was the only station that had ever employed a three-tier strategy.120

Thirty-three witnesses assigned to Imperial Beach acknowledged that at some point in time the Station had used a three-tier strategy. One witness claimed that this deployment had continued for only two days at the beginning of Gatekeeper, while others said it lasted several weeks. Still other witnesses said that the three-tier deployment had continued for five months. Nine witnesses claimed that the three-tier approach was never implemented at Imperial Beach.

After reviewing all of the evidence, we conclude that all three tiers at Imperial Beach were manned when Gatekeeper began. G-481 forms - which list each agent's daily assigned position - indicate that agents were assigned to second and third-tier positions at Imperial Beach in October 1994, the first month of Gatekeeper. During October and for several additional months, agents from other stations and sectors were detailed to Imperial Beach, giving the station significantly more manpower and permitting it to staff the third tier. On occasion the third tier was staffed by specialty units such as the Bike Patrol or the Anti-Smuggling Instant Strike Team (ASIST).

Beginning on February 27, 1995, the G-481 forms at Imperial Beach no longer list third-tier assignments. A number of factors led to the demise of the third tier. The Gatekeeper strategy was implemented at Chula Vista at this time, a development that required additional personnel to staff border positions at that station. A February 1995 memorandum from CPA Williams to William Slattery, INS Assistant Commissioner for Operations, indicates that on February 19, 1995, Chula Vista and Brown Field agents who had been working on the third tier at Imperial Beach were transferred to positions closer to the border. While checkpoint station agents were transferred to Imperial Beach for a two-week detail to compensate for this loss, at the end of this detail Imperial Beach Station probably did not have sufficient manpower to staff the third tier.

It was also at this time that CPA Williams concluded that Imperial Beach had been brought under control. From November 1994 through March 1995, Imperial Beach's apprehensions were 40 to 50 percent below the apprehensions recorded in the same months in the previous year. Meanwhile, by January 1995, Chula Vista's apprehensions had begun to rise to levels well above numbers posted the previous year. Brown Field had begun experiencing a similar increase in November 1994. Williams' decision to begin transferring agents from Imperial Beach to Chula Vista and Brown Field was motivated by this change in traffic patterns.

The inability of the Imperial Beach Station to staff a third tier was also related to Williams' decision to temporarily expand the Station's area of responsibility. Under Williams' program of "flexible corridors," boundaries could be shifted "at a moment's notice" to address shifts in alien traffic patterns. In an effort to assist Chula Vista in implementing Gatekeeper, Williams shifted a portion of that Station's area of responsibility to Imperial Beach. As a result, Imperial Beach had a broader area to patrol, and fewer agents to perform these duties.

At this same time, additional fixed positions were created directly along the border in an effort to boost the level of deterrence at Imperial Beach. As these new fixed positions were filled, fewer agents were available for second or third-tier positions. The strategic plan for fiscal year 1996 indicates that during the first year of Gatekeeper the number of first-tier positions increased by six, while the third tier lost an equal number of agent positions. As a result, there were days when Imperial Beach no longer had sufficient manpower to staff three tiers. Indeed, on occasion, the second tier was not fully staffed.121

By April 1995, apprehensions at Imperial Beach had increased by nearly 4,000 aliens per month, exceeding 10,000 per month for the first time since October 1994. More than 10,000 aliens were apprehended at Imperial Beach each month from April through September 1995. As discussed previously, the higher apprehension numbers at Imperial Beach caused the Commissioner some consternation, and the DOJ OPA employee discussed above wrote a memorandum to the Attorney General about this increase. In May 1995, the REACT team was brought back to Imperial Beach and worked on the third tier for a week. The evidence suggests, however, that other than this week, the third tier at Imperial Beach was probably not staffed during the period from March 1995 through August 1995.

Beginning in approximately September 1995, the third tier was again staffed to some extent. The September 13, 1995, Southwest Border Operations Weekly Situation Report notes that a detail of 26 agents was sent to Imperial Beach on September 3 so that all three tiers could be manned.122 The detail was scheduled to last a month.

In October 1995, another specialty unit was detailed to Imperial Beach and this unit patrolled the third tier from October 1, 1995, through December 23, 1995. This unit was not, however, deployed seven days a week. In short, while the third tier was patrolled to some extent in 1995, it was not fully or consistently staffed.

Sensor Intelligence Reports and Tier Apprehension Reports prepared during the period from January through May 1996 refer to apprehensions made in the third tier, indicating that this tier was patrolled to some extent during this time.

We conclude that initially Imperial Beach deployed three fully-staffed tiers of agents, but that over time this deployment was not maintained. From March through August 1995 there was little patrolling of the third tier as manpower levels decreased and agents were transferred to positions closer to the border. Contrary to the Union's complaints, however, we find no misconduct in the deployment of agents within the tiers. We find that the elimination of the third tier was motivated by a lack of manpower, not by a desire to reduce apprehension numbers.

Numerous supervisors and line agents testified that the third tier positions were eliminated because of insufficient manpower. In the June 23, 1996, North County Times article reporting the initial fraud allegations, an INS spokesperson acknowledged that the third-tier system had been reduced and placed the blame on a lack of manpower. As agents were transferred to border positions, other deployments had to be sacrificed. Given that approximately 80 percent of apprehensions were made in the first tier and nearly 95 percent of apprehensions were made in the first two tiers, the most logical positions to eliminate were those in the third tier. Third tier positions were the most expendable because they had the smallest impact on station effectiveness. The third tier had no deterrent effect and apprehended the fewest aliens among the three tiers.

One Union official argued, however, that the transfer of agents to the first tier border positions proved that managers wanted to reduce agent effectiveness. He testified that "holding down an X [does not] qualif[y] as something all that important." He complained that "the whole focus nowadays is to engage in this fiction of deterrence. It simply doesn't work, it [has] turned the Border Patrol into the Keystone Cops of the law enforcement world because the aliens laugh at us." This witness believes that fixed positions along the border are not effective, and that deploying agents in third tier positions is a better strategy. He, however, offered no evidence supporting this view.

The Union's views appear to be guided primarily by agent preferences, rather than operational effectiveness. Bonner's testimony in the summer of 1996 makes quite clear that agents disliked the fixed border positions and preferred assignments - such as those in the third tier - that offered more flexibility and freedom of movement. He argued that the fixed border positions should be eliminated and that agents should be allowed to return to their former patrol practices.

Substantial statistical evidence, however, shows that increased staffing of positions at the border has a significant impact on the level of traffic attempting entry. The most dramatic evidence of this is at Chula Vista. When it rains, Chula Vista agents are forced to abandon their positions along the border and retreat to paved roads some distance away. When agents retreat from the border, however, entries and apprehensions soar, sometimes increasing by more than 100 percent over the previous day. When the agents return to their border positions, alien traffic and apprehensions fall to their previous levels. Likewise, the significant shift in traffic from Imperial Beach to more easterly locations - after additional agents were placed in high visibility positions at Imperial Beach - demonstrates that the Gatekeeper program is more effective than a strategy that permits aliens to enter and requires agents to then attempt to apprehend as many as possible.

In addition to the statistical evidence supporting the use of border positions, we reviewed contemporaneous memoranda documenting that supervisors frequently created additional first-tier positions in areas that were experiencing high levels of alien traffic. These memoranda support the testimony of the supervisors that the additional positions were necessary to effectively plug holes in the first line of defense.

Although Bonner argued to the Congressional subcommittee that agent resources should be directed at apprehending aliens who have entered the country, his position is completely inconsistent with the underlying strategy of Gatekeeper, which emphasizes deterrence first and permits apprehensions only after deterrence has proven ineffective. Bonner's position is also contrary to that of numerous members of Congress who have repeatedly pressed INS to deploy more agents directly on the border to deter aliens from attempting entry, rather than allowing aliens to enter and then attempting to apprehend them.

No. 1922 also alleged that third-tier agents were transferred to first-tier positions because they were apprehending too many aliens. While third-tier agents did initially apprehend large numbers of aliens, this indicated that there was insufficient deterrence at the first tier. More agents were needed in the first tier to deter aliens from attempting entry. We found no evidence, however, that additional first-tier positions were created so that no agents would be available to patrol the third tier. As noted above, the evidence clearly demonstrates that increasing deterrence at the first tier results in fewer aliens attempting entry. As a consequence, fewer aliens will successfully penetrate the first and second lines of defense. As noted above, apprehension records show that the overwhelming majority of apprehensions - in early 1996 between 70 and 80 percent - are made in the first tier. During that same time, between 90 and 95 percent of apprehensions were made in the first two tiers. Given that virtually all witnesses agreed that Imperial Beach neighborhoods experienced a sharp decline in alien traffic after the initiation of Gatekeeper, transferring agents from third-tier to first-tier positions was clearly an effective strategy.

Finally, Union official Joseph Dassaro's suggestion in the North County Times article, that the Border Patrol eliminated third-tier positions in order to reduce apprehensions, was based on the assumption that a weakening of the third tier would lead to more "gotaways" and fewer apprehensions. Dassaro's reasoning suffers from several defects. First, some third-tier agents were transferred to first-tier positions, thereby increasing deterrence along the border and reducing traffic for all three tiers. Such a reduction in traffic would yield fewer rather than more "gotaways." Although it is impossible to calculate precisely whether the number of aliens deterred or apprehended by additional first-tier agents was greater than, less than, or equal to the number of persons who would have been apprehended in the third tier if agent deployment had remained static, the fact that border apprehensions declined over time - without evidence of more traffic penetrating the third tier - suggests that the new strategy was effective. Furthermore, we find no evidence that supervisors acted in bad faith in creating additional deterrence-oriented positions along the border.

Dassaro's complaint also assumes that the number of aliens apprehended in the third tier is substantial enough to significantly affect apprehension and "gotaway" statistics. All of the documentary evidence, however, indicates that even when the third tier was fully staffed, apprehensions made in that tier were not substantial enough to materially affect the Station's overall performance.

Moreover, it was not seriously disputed that the level of alien traffic at Imperial Beach was significantly lower post-Gatekeeper. Thus, if supervisors weakened the third tier in March 1995 in order to reduce apprehensions and increase "gotaways," their plan failed miserably. Apprehension levels jumped dramatically and remained high from March until September 1995 when additional agents were transferred to Imperial Beach.

Although Bonner and Dassaro are correct that Imperial Beach Station did not employ a full three-tier deployment throughout Gatekeeper, we found no evidence that agents were deliberately transferred from the third tier to positions closer to the border, or to other stations, to decrease operational effectiveness or to artificially limit apprehensions. The evidence indicates that supervisors attempted to maximize the effectiveness of available manpower by assigning agents where they were needed most. The evidence also demonstrates that when additional resources - in the form of specialty units - became available, they were assigned to patrol the third tier. Furthermore, although in retrospect CPA Williams prematurely transferred agents from Imperial Beach, we found no evidence that he acted in bad faith. Moreover, the resulting increase in apprehensions at Imperial Beach negates any inference that the transfer of agents was intended to artificially reduce apprehensions at Imperial Beach so as to make Gatekeeper appear more effective.

In sum, we reject as unsubstantiated the claim that the third tier at Imperial Beach was dismantled to decrease agent effectiveness and to improve the public's perception of Gatekeeper.123

113 Even if some supervisors did suggest that agents use the TBS tactic, such a suggestion would not be evidence of fraud. The Border Patrol's goal is to prevent illegal entry, and we see no impropriety in a supervisor suggesting that agents use an alternative, legal means to accomplish that goal. Indeed, using TBS where appropriate could substantially enhance the ability of the Border Patrol to achieve its mission. Apprehending, processing, and transporting aliens consumes agent resources and costs money. If an alien can be prevented from entering without incurring such costs, that would appear to be the better approach. We do not suggest that TBS is appropriate away from the immediate border area or that agents should physically push aliens over the fence once they are apprehended. TBS is to be used only when physical apprehension is dangerous or containment is a more appropriate tactic.

114 The San Clemente checkpoint is bounded on the west by a coastal area with no through roads and to the east by Camp Pendleton, a U.S. Marine facility. As there are no proximate alternative roads for circumventing this checkpoint, roving patrols for vehicles were not used as they were at Temecula.

115 On October 13, 1996, Boubel was promoted to a Supervisory Border Patrol Agent position and thereafter was no longer a Union official.

116 The El Cajon memorandum said that the new policy was not intended to "adversely impact" agents' enforcement efforts, but rather to reduce the potential for "unnecessary and adverse civil and/or administrative action."

117 Gonzalez-Rivera v. INS, 22 F.3d 1441 (9th Cir. 1994).

118 For example, in June 1996, after Sector management's edict prohibiting roving patrols, the Anti-Smuggling Unit was sent to Temecula to operate checkpoints on secondary roads to intercept smugglers attempting to circumvent the I-15 checkpoint. Sending in reinforcements to conduct additional traffic checks and intercept smugglers is hardly consistent with an effort to reduce apprehensions.

119 In addition to complaining about the elimination of roving patrols, Bonner had alleged - at the summer 1996 hearings - that checkpoint stations were understaffed. A Union official from another station likewise contended that the checkpoints were deliberately kept understaffed to reduce apprehensions and make Gatekeeper appear successful. He acknowledged to the OIG, however, that this suspicion was pure speculation. Neither witness could suggest where additional agents for the checkpoints could be obtained. Bonner complained that there were not enough agents in the backup tiers at Imperial Beach or enough agents at the East County Station. He complained about agents being detailed to San Diego from other sectors and about mandatory sixth-day overtime for San Diego agents. All new hires were being assigned to San Diego. The fact remained that the Sector had a finite number of agents and every additional agent assigned to a checkpoint station meant that one less agent was assigned along the border. The border positions are an integral and effective part of a deterrence strategy, however, and must necessarily be given first priority. Accordingly, the Border Patrol's failure to abandon border positions - so that staffing at checkpoints over 60 miles from the border could be increased - does not demonstrate a desire to avoid apprehensions. There is simply no evidence that Sector management deliberately understaffed the checkpoint stations in order to reduce apprehensions and make Gatekeeper appear effective.

120 This witness was incorrect on this point. Chula Vista and Brown Field also used three-tier deployments at various times during Gatekeeper. Like Imperial Beach, however, these stations could not consistently staff these positions due to a lack of manpower. The OIG, however, received no complaints about the failure to maintain the three-tier strategy at Chula Vista and Brown Field.

121 By mid-1995, INS press releases stop making reference to the three-tier strategy and begin to speak more generically about "backup positions."

122 These agents were sent to Imperial Beach after Commissioner Meissner expressed concern to WRD de la Viņa about the high level of apprehensions at Imperial Beach and a perceived loss of control in the area. De la Viņa contacted CPA Williams and ordered him to reinforce Imperial Beach and to regain the level of control enjoyed there prior to the transfer of personnel to Chula Vista and Brown Field.

123 Several Imperial Beach agents complained that they were no longer allowed to patrol downtown San Diego. They could provide no evidence to support their belief that city patrols were eliminated in order to reduce apprehensions, however. The evidence shows that these patrols - like the third-tier positions - were eliminated because of limited manpower and operational priorities. As we discussed above, the border positions appropriately received first priority.