The Air Traffic Controller Workforce Imperative: Staffing Models and Their Implementation to Ensure Safe and Efficient Airspace Operations (2025)
Chapter: 5 Facility Strength Assessment, Attrition Estimates, and Annual Hiring Goals
5
Facility Strength Assessment, Attrition Estimates, and Annual Hiring Goals
As part of the request for this report, Congress asked for an assessment of the “attrition rates at each air traffic control facility operated by the administration.” This chapter reviews the models and procedures the Federal Aviation Administration (FAA) uses to estimate attrition in the broader context of the various steps FAA takes to assess the adequacy of its facility workforces, anticipate the losses they will experience in future years, and estimate the hiring FAA needs to make each year to maintain a staffing level that meets the staffing standards reviewed in the previous chapter. Facility “strength” refers to how well a facility is staffed. FAA describes these steps as the A−B+C=D modeling process (see Figure 1-1).
Estimates of facility staffing standards or targets using the facility staffing models described in Chapter 4 generate the modeled input into estimating the value of “A.” The calculation of “B” assesses the current facility workforce strength. Estimates of “C” rely on a set of models and calculations to estimate future attrition that has to be accounted for in establishing the staffing gap. D is the result of subtracting the sum of current facilities’ workforce strength from the sum of their staffing standards and adding the sum of their forecast attrition. The aggregate estimate of “D,” if not further adjusted by other, nonmodel considerations, provides an unconstrained model-based estimate of the number of hires that need to be made to ultimately close the staffing gap. Closing the gap requires accounting for the time new hires require for training and certifying, which is a multiyear process. This chapter assesses FAA’s Office of Finance and Management’s (AFN’s) estimates of A, B, C, and D. The final section offers the committee’s findings, conclusions, and recommendations.
ILLUSTRATIVE EXAMPLE OF A−B+C=D CALCULATIONS
A very simplified example using the AFN approach for a hypothetical facility follows in Table 5-1 of the A−B+C=D calculations that would done at the beginning of a Fiscal Year (FY) for an individual facility. One of the many simplifications in this example is that it does not include Developmental Controllers (DEVs) on the staff who might certify to Certified Professional Controller (CPC) in the coming year. Another simplification is that training failures of new hires are assumed to occur in the coming year even though some will occur in following years. Thus, a facility with a staffing standard (A) of 204 with 207 CPCs and Certified Professional Controllers-In Training (CPC-ITs) on hand (B) is increased by 17 (C) to account for attrition to estimate a staffing gap of 14 (D). The staffing gap is then increased by six more to account for anticipated training losses among the required new hires, or (204–207+17+6=20). The total hires to be made in the forecast year modeled in this process represents the sum of such calculations made for all 313 facilities. It should be noted that with an expected time to certify of multiple years, the estimated hires shown in the final column of Table 5-1 will not close the staffing gap in the year in which hires are made.
In practice, the estimation process that leads to the hiring goal is considerably more complicated to account for forecast changes such as those that will occur over the course of a year due to (a) the actual time that existing trainees will need to certify to CPC, (b) the point in time when the attrition and hires occur, and (c) the ongoing annual adjustments needed to close the staffing gap in future years.
CALCULATIONS FOR A: SETTING OF FACILITY TARGETS
The models described in the preceding chapter generate an estimate of the appropriate level of staffing at each facility to safely and efficiently manage traffic as approximated by the task load models used by AFN and the survey-based responses used by the Collaborative Resource Working Group (CRWG). Presently, with exceptions noted in the previous chapter,
TABLE 5-1 Illustrative Example for Single Facility of the A−B+C=D Modeling Process
| A: Staffing Standard | B: CPCs + CPC-ITs on Hand at End of Prior FY | C: Forecast Attrition from Current Workforce | D: Staffing Gap | Estimated Hires Including New-Hire Training Failures |
|---|---|---|---|---|
| 204 | 207 | 17 | 14 | 14+6=20 |
SOURCE: Committee generated from FAA presentation.
the targets or staffing standards for each facility are the modeled estimates. As noted in previous chapters, only the AFN modeled estimates, as adjusted upward for minimum watch staffing and the 5-hour Time on Position (TOP) policy, are used in setting the targets at the time of this writing. The Collaborative Resource Working Group (CRWG) estimates, to date, have only been used for placing Academy Graduates (AGs) or new hires with previous experience in individual facilities and for influencing the transfer process (see Chapter 6). Exceptions are made to AFN’s or the CRWG’s staffing standards for individual cases when the Air Traffic Organization (ATO) and AFN agree that the staffing levels estimated are not working out in actual practice. These adjustments are made as such problems occur.
Before 2016, AFN and the ATO had engaged in a negotiation about appropriate facility staffing standards or targets based partly on input from the field about the adequacy of the proposed targets (NRC 2014). This was referred to as “Service Unit Input (SUI),” which reflected the expert judgment of facility managers and others in ATO. ATO relied on surveys of facility managers to gather their perceptions, although it is unclear whether this was done annually. SUI was sometimes supplemented with comparison of the staffing levels of similar facilities and their productivity. The committee for the National Academies of Sciences, Engineering, and Medicine (NASEM) 2014 report believed that comparing the modeled estimates with SUI was appropriate. The committee for the 2014 report concluded that a set of models meant to estimate staffing levels at 313 facilities, even though differentiated by facility type, will be unable to account for unique operational aspects of individual facilities that should be considered in setting staffing targets. These might include a variety of conditions such as airline plans to add a new hub or otherwise expand traffic at a particular airport, rapid population growth around small airports in the periphery of a metropolitan area or unusually high peaks and valleys in traffic operations. Although the 2014 NASEM report committee supported AFN and ATO combining modeled outputs with SUI, it also observed that the processes by which SUI was generated were not well documented and appeared to change over time for unexplained reasons.
In its Recommendation 3-3, the 2014 NASEM report committee (NRC 2014, 78) stated that “FAA should ensure that the field understands the staffing process by providing greater clarity and transparency, and it should continue collaborative efforts ensuring that local facility considerations are properly addressed in—and continuously fed back to—its generation of the staffing standards.”
The 2014 NASEM report committee’s recommended practice of combining modeled estimates with SUI reflected current practice at the time and continued to be followed by FAA until 2016, at which point collaboration between ATO and AFN broke down for reasons the committee is not fully
cognizant of. From 2016 through 2022 onward, AFN continued to generate modeled staffing standards for each facility and report staffing standard ranges for each facility, which were listed in the appendix of the annual Controller Workforce Plans (CWPs).1
By 2016, the CRWG had developed its own process for estimating appropriate facility targets and used these estimates to set priorities for placements and transfers:
- 12,895 was the target set for FY 2016 through FY 2018
- 13,014 was the target set for FY 2019 through FY 2022
- 14,633 was the target set for FY 2023 and FY 2024
As described in Chapter 4, these targets are 30% higher than AFN’s in FY 2023–2024,2 but the difference was not as great beforehand. Although the CRWG targets were not used in setting facility staffing targets, beginning in FY 2023 the annual CWP dropped reference to staffing standard ranges and referred to a single target for each facility based on AFN’s model and also provided a target for each facility as estimated by the CRWG.
The CRWG method is effectively a much more elaborate process for generating SUI than was used in the past. As described in Chapter 4, it begins with the judgment of each facility manager about how each facility should be staffed to cover the positions that would need to be open on busy days that are akin to the 90th percentile day of traffic used by AFN in its models. Prior to the first CRWG estimates, ATO estimated SUI with values increasing from 12,176 in FY 2010 to 12,838 in FY 2015, thus the early CRWG targets were consistent with SUI estimates made in previous years. The FY 2023–2024 targets are the first time that the procedures used by the CRWG were made fully traceable and transparent (MITRE 2023). The reason for the 30% increase in the target as estimated by the CRWG for FY 2023–2024 compared with FY 2019–2022 was not explained to the committee; it was apparently related to making the process of generating an estimate more rigorous and traceable.
CALCULATIONS FOR B: CONTROLLERS ON HAND
The number of CPCs, CPC-ITs, DEVs, and newly arrived AGs on the rolls of facilities at the end of a FY preceding the annual A−B+C=D process
___________________
1 This paragraph was revised after release of the report to clarify FAA practice of reporting the staffing standard range in annual CWPs.
2 The 30% estimate applies to CPCs only. CRWG assesses workforce strength with CPCs, and AFN with CPCs and CPC-ITs. Hence, an estimate must be made of CPC-ITs in the AFN case in order to be comparable.
identifies the existing workforces at facilities (“B”). (Counting the different types of controllers on board is important for estimating training failures in subsequent years in C.) AFN refers to “B” as the “contribution from the existing workforce,” which is a better description of headcount than of the productive work the existing workforce is capable of.
Facility Strength Assessment
The calculations to derive B in the A−B+C=D modeling process, indicates the strength of the staffing level of each facility. As described in Chapter 4, the AFN and CRWG facility staffing models estimate the positions that need to be staffed by qualified controllers, but the models themselves do not specify what type of controller these qualified controllers need to be other than being qualified, which implies that they include being certified for at least one position. For the AFN facility staffing models, the adequacy of staffing at a facility is based on the number of CPCs and CPC-ITs at facilities relative to its staffing standard, even though certified DEVs are able to carry out independent TOPs for which they are certified. For the CRWG model, the assumption is that only the number of CPCs qualifies for assessing the adequacy of facility staffing even though CPC-ITs and DEVs who are qualified for at least one position carry out independent TOP.
One of the CRWG arguments for using CPCs only is that it wants to free CPC-ITs and DEVs to devote their time to training and progress to CPC status as quickly as possible, a concern addressed in the following subsection.
The committee for the NASEM 2014 report observed that relying only on CPCs and CPC-ITs in assessing a facility has the following two disadvantages in addition to not reflecting the productive work that DEVs perform:
- The progression of DEVs through the certification process in the assessment year and their likely progression in the subsequent year is not accounted for. A facility with adequate DEVs in the pipeline may appear understaffed in the assessment year but be adequately staffed in the next year without a change in staffing due to continued levels of certification achieved by DEVs in the interim.
- The actual certification Levels of CPC-ITs at a facility are not accounted for but assumed. A newly arrived CPC-IT is unable to work TOP at any Level, and an assumption about the extent to which facility CPC-ITs are certified for TOP could be misleading in the short term as these CPC-ITs require 1.5 years, on average, to attain full certification.
- The extent to which the number of CPCs within the staffing range may change once staff qualify at a facility would not be reflected. A facility that appears to be below range in one year may appear to be within or above range the next, not because its staffing levels have changed but because of successful training of the staff already at the facility.3
In the NASEM 2014 report committee developed a CPC-Equivalent Workforce (CEW) concept and made the following Recommendation 4-1: “FAA should examine the merits of using a more appropriate analysis of facility demographics (e.g., the CEW concept) in place of the CPC + CPC-IT metric when it assesses facility staffing status and develops its annual staffing plans” (NRC 2014, 105).
The vast majority of the workforce consists of CPCs. Individuals in the remainder of the workforce (non-CPCs) have yet to achieve full certification status but can and do contribute to the primary controller mission of serving TOP. The underlying concept of the CEW is to quantify the effective contribution of all members of the workforce.
At each facility and FY, five segments of the workforce according to the Level of certification are identified, call it L, defined as follows:
- L=4, CPCs
- L=3, CPC-IT3s and DEV3s
- L=2, CPC-IT2s and DEV2s
- L=1, CPC-IT1s and DEV1s
- L=0, CPC-IT0s, AGs, and those with Prior Experience (PEs) hired and assigned directly to an operational facility
CPCs (L=4), by definition, are qualified to work on every position in the facility. Non-CPCs (Levels L=0 through 3) progress through training and become qualified at an increasing number of positions at each Level L and can, accordingly, serve more TOP.
The committee constructs the CEW in the following manner:
- For each Level of certification L:
- Identify the number of TOP hours reported by segment L, call it TOP(L)
- Identify the number of Full-Time Equivalents (FTEs) in segment L, call it FTE(L)
- Calculate the average TOP per FTE, call it TOP per FTE (L)
___________________
3 This text was revised after release of the report to more accurately reflect the NASEM 2014 report.
-
- Calculate the contribution to the workforce of each Level L controller relative to a CPC,
- Then the CPC-Equivalent Workforce is:
- Calculate the contribution to the workforce of each Level L controller relative to a CPC,
In general terms, the CEW concept can be operationalized by accounting for the actual number and certification Levels of DEVs and CPC-ITs at facilities throughout the assessment year and the actual independent TOP they carry out relative to that of CPCs at the facility. See Box 5-1 for an example using the Austin Tower (AUS), which is a facility of average size.
Depending on the number and mix of DEVs, CPC-ITs, and CPCs at each facility and the certification Levels of the DEVs and CPC-ITs, the
BOX 5-1
CPC Equivalent Workforce Construction – Numerical Example AUS – Austin Tower FY 2024
The calculations shown below illustrate how the application of the CEW concept would measure the staffing level at the Austin Tower for FY 2024.a
| Level L | Level Description | TOP(L) | FTE (L) | TOP per FTE(L) | Equiv(L) | Equiv(L)*FTE(L) |
|---|---|---|---|---|---|---|
| 4 | CPC | 36,429 | 32.8 | 1,111 | 1.00 | 32.8 |
| 3 | CPC-IT3/DEV3 | 608 | 0.8 | 724 | 0.65 | 0.5 |
| 2 | CPC-IT2/DEV2 | 1,815 | 3.0 | 5,99 | 0.54 | 1.6 |
| 1 | CPC-IT1/DEV1 | 1,430 | 2.6 | 542 | 0.49 | 1.3 |
| 0 | CPC-IT0/AG/PE | 557 | 2.7 | 206 | 0.19 | 0.5 |
| 42.0 | CEW | 36.8 |
__________________
a This text was revised after release of the report to more accurately reflect the purpose of the CEW concept.
actual strength of the workforce to carry out TOP at facilities may be higher or lower than assessing facility strength based on the number of CPCs and CPC-ITs, but it would be more accurate. Because FAA records and closely monitors the number of controllers by type, individual controllers’ Levels of certification, and the points in time that their certification Levels change, it has all the data needed in hand to make these calculations.
With the data provided by FAA, the committee calculated the CEWs for all facilities; it accounts for the number and mix of staff at each facility and their certification Levels and compares them to the AFN assumptions about how the adequacy, or strength, of a facility should be assessed for FY 2024. The results of these calculations are presented by number of facilities for FY 2024 in Figure 5-1. (The CRWG method of assessment would place 92% of facilities at more than 15% understaffed in FY 2024, so is not shown in Figure 5-1, but is included in a similar figure [Figure B-9] in Appendix B.) As can be seen in Figure 5-1, assessment of facility strength is similar in FY 2024 between the AFN and CEW meth=ods. Facilities rated as within +/−10% of the AFN staffing standard by the two methods results in a difference of only three facilities. For those outside +/−15% the differences are larger: the AFN method would rate 15 more facilities below 15% than the CEW, and the CEW would estimate 12 more facilities above 15% than the AFN method would. Comparison of the number of FTEs within +/−10% of the AFN standard by the AFN or CEW methods show similar patterns: the CEW method would estimate 418 FTEs as being within +/−10% of the AFN standard that the ANF method would estimate as outside the +/−10% range. Thus, the assessment of facility strength by either the AFN or CEW
SOURCE: Generated by the committee using data provided by FAA.
method for facilities or FTEs would make a modest but valuable improvement as an input to setting the hiring goal and would also indicate that the level of understaffing below 85% is not as severe as use of the AFN method indicates. In addition, the CEW method is a more accurate measurement of individual facility strength.
Time to Certification in Understaffed Facilities
The assumption made by the CRWG in assessing facility strength is that it should be measured in terms of CPCs only so that trainees could focus their time on training and progress to CPC status as quickly as possible. The assumption here is that facilities that are short of CPCs will not have adequate time to devote to On-the-Job Training for trainees in these facilities, which will result in a longer time to certification.
To test this assumption, the committee scored all 312 or 313 facilities (depending on the year) on their share of CPCs, divided into equally sized quintiles as follows:
- Q1: 0.00 < CPC Content ≤ 0.671
- Q2: 0.671 < CPC Content ≤ 0.740
- Q3: 0.740 < CPC Content ≤ 0.794
- Q4: 0.794 < CPC Content ≤ 0.815
- Q5: 0.815 < CPC Content ≤ 1.00
The committee then estimated the average time to CPC certification using these same quintiles as shown in Table 5-2. (The trend stops at FY 2020 since too many trainees are still in process between FY 2021–2024 to accurately estimate time to certification.) There is no apparent correlation between the share of CPCs at facilities and time to full certification, although facilities with the highest share of CPCs (Q5) have improved from the slowest time to CPC status in FY 2010 to the fastest by 2020. However, this improvement is not much faster than facilities with the lowest percentile of CPCs in FY 2024. These results imply that the CRWG assumption that assessing facility strength by CPCs would expedite time to CPC status may not be fully correct. The values in Table 5-2 are close to, if not within, the training goals FAA sets for DEVs in Terminals (FAA 2023, 2024). The calculations for Table 5-2 include facilities of all types, but variation in facility staffing, including those that are 10 or 15% below their staffing standard, is captured using the CPC percentile estimates. However, as discussed in the next chapter, time to CPC status at Centers has been growing since FY 2010 and deserves further analysis to determine the causes.
TABLE 5-2 Average Time to CPC Status by CPC Quintile in Facilities, FY 2010–2024
| Average Time to CPC Quintile | Hire FY | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | Grand Total | |
| Q1 | 2.4 | 2.3 | 2.6 | 2.8 | 2.3 | 2.3 | 2.6 | 2.5 | 2.6 | 2.6 | 2.6 | 2.5 |
| Q2 | 2.8 | 2.4 | 2.5 | 2.4 | 2.3 | 2.6 | 2.8 | 2.7 | 2.5 | 2.4 | 2.5 | 2.6 |
| Q3 | 2.7 | 2.3 | 2.5 | 2.5 | 2.3 | 2.5 | 2.7 | 2.9 | 3.5 | 3.4 | 2.8 | 2.8 |
| Q4 | 2.7 | 2.4 | 2.6 | 2.6 | 2 | 2.2 | 3 | 2.6 | 3.1 | 3.2 | 2.3 | 2.6 |
| Q5 | 2.9 | 2.7 | 2.5 | 2.1 | 2.3 | 2.2 | 2.8 | 1.7 | 2.4 | 2.3 | 2.2 | 2.4 |
| Grand Total | 2.7 | 2.4 | 2.6 | 2.5 | 2.3 | 2.4 | 2.8 | 2.6 | 2.9 | 2.9 | 2.5 | 2.6 |
Productivity Impact of Transfers with the ATC Workforce
There is one form of temporary loss productivity that involves transfers within the ATC workforce. These transfers are mostly made by CPCs. (The process for such transfers is described in the next chapter.) If approved, such transfers are made after a CPC requests a transfer to another facility. A few transfers are made each year when a CPC-IT or DEV fails at a facility but is deemed promising enough to certify if allowed to try again at a lower-level facility, but these are small in number. Upon transferring to the new facility, the former CPC is classified as a CPC-IT and begins the certification process anew. This process typically requires 1.5 years, if the candidate is successful, as most are. If unsuccessful at certain new facilities, candidates are allowed to return to their original facility and recertify there. Thus, for some period of time, CPCs involved in transfers are not able to provide productive work until they certify on at least one position. The committee estimates an average of roughly 700 transfers in the FY 2010–2024 period, during which time CPC-ITs are not certified or only partially certified. These 700 annual transfers are second only to the number of hires made in the FY 2010–2024 period (about 1,200 on average) and are therefore an important aspect of overall strength management.
When facility strength is assessed in terms of CPCs and CPC-ITs using the AFN assumption, the impact of lost productivity of CPCs during the transfer process is not considered, thereby overstating the productivity CPC-ITs. The actual contribution of CPC-ITs in the system could be calculated based on their Levels of certification and TOP relative to that of CPCs rather than assuming their productivity is equal to that of CPCs.
CALCULATIONS FOR C: FUTURE ATTRITION4
The air traffic control workforce experiences five types of attrition (see Table 5-3). As classified by FAA, these are
- Retirements
- Resignations, Removals, Deaths (RRD)
- Academy Attrition
- Developmental Attrition
- Promotions and Transfers Out of the Controller Workforce
Average losses are shown in Table 5-3 over the FY 2010–2024 period. Retirements have been the largest share of losses over this period, followed
___________________
4 This section was revised after release of the report to reflect the most current attrition data from FAA.
TABLE 5-3 Attrition Categories, FY 2010–2024
| Attrition Category | Average | (%) |
|---|---|---|
| Retirement | 474 | 36.7 |
| RRD | 85 | 6.6 |
| Academy | 298 | 23.0 |
| Developmental | 102 | 7.9 |
| Promotion and Transfer | 334 | 25.8 |
| Total | 1,293 | 100.0 |
by promotions and transfers, DEV training failures, and Academy training failures. Actual annual losses from the workforce vary considerably as FAA goes through cycles of high and low hiring and resulting peaks and valleys in retirements and Academy and DEV attrition (see Figure 5-2). The most recent wave of retirements fell from a peak of 706 in FY 2015 to 185 in FY 2024; thus, retirements will become a less important source of future attrition for a few years. Meanwhile, the Academy and DEV training losses are expected to increase significantly in the future as annual hires ramp up from 1,800 in FY 2023 to 2,400 in FY 2026.5 Promotions and transfers will also grow as the ATC workforce grows, which will result in more CPCs leaving the ATC workforce to assume the growing number of supervisory and specialist positions in ATO.
Although attrition is modeled for all five categories for each of FAA’s 313 facilities, the sum of all these estimates is only used in generating an aggregate annual hiring goal. Attrition estimates have no role in establishing the staffing standard for individual facilities. AFN’s expectation is that ATO will hire enough applicants each year and will do so far enough in advance to allow individuals to be prepared to step in as CPCs leave the workforce. The hiring constraints that occurred during the FY 2012–2023 period (discussed in Chapter 2) not only resulted in immediate reductions in staff at individual facilities as attrition outpaced hiring, but also minimized the size of the incoming pool of trainees who would be needed in subsequent years as senior controllers leave the workforce.
In practice, anticipation and preparation for attrition takes place at individual facilities by their facility managers. Facility managers should have insight on some forms of attrition, such as retirements and requested transfers. They know about requested transfers, which are not executed quickly since transfers have to be approved at higher Levels as described in Chapter 6 and they know of the retirement eligibility profiles of their most senior CPCs. They also have insight into DEV failures since the failure
___________________
5 This sentence was revised after release of the report to correct the number of hires projected for FY 2026.
SOURCE: FAA. 2024. “Air Traffic Controller Workforce Plan: 2024–2033.” U.S. Department of Transportation.
rates of individual facilities are closely tracked. They may not have as much insight on attrition from RRDs. Another important consideration to bear in mind when reviewing the accuracy of AFN’s attrition estimates is that they are reestimated each year for all facilities after AFN reviews the accuracy of its predictions from the previous years and adjusts its estimation methods accordingly. Thus, any over or under forecast in a single year can be accounted for in the estimates of hiring goals made the following year.
Retirement Projections
Controllers can be hired as soon as they graduate from high school if they have enough appropriate work experience and must retire by age 56 unless granted a waiver.6 Most become eligible to retire by age 50, but only 25% retire in the first year of what can be several years of retirement eligibility for controllers hired at the youngest ages (FAA 2024, 31). To forecast retirements, FAA keeps track of the past years of retirement once a controller reaches eligibility and estimates from this the probability of retirement for each year of eligibility in the future. In running its retirement model for each facility, AFN calculates the probability of retirement for each CPC and CPC-IT at individual facilities based on their year of retirement eligibility and the historic probability of retirement in that year. The committee compared projected with actual retirements over the FY 2010–2024 period: actual retirements were 5% less than projected retirements, which amounts to an average overestimate of retirements of 27 individuals expected to retire who did not retire that year.
Resignations, Removals, Deaths (RRDs)
As one of the smallest categories of loss, both in the past and future, RRDs are of minor importance in estimating future attrition. AFN’s estimation approach is a simple one since RRDs are not easily modeled. It averages the percentage of the total ATC workforce lost to RRDs in the previous 3 years and uses that to estimate future losses. This estimate is applied to each facility each year. Over the FY 2010–2024 period, the average annual forecast error was 8% greater than actual, but it represented an annual average of only 10 greater losses than occurred.
___________________
6 This sentence was revised after release of the report to clarify qualifications for high school graduates.
Academy Attrition
Academy attrition over the FY 2010–2024 period was the third highest of all loss categories on average, and the number will increase as hiring classes expand and its share will become larger as total retirements decrease. This will make estimates of Academy attrition more important in the future. Based on the committee’s estimates, Academy attrition due to training failures has ranged between a low of 7% in FY 2010 to a high of 52% in FY 2017 for hires placed in the Centers Track and between a low of 7% in FY 2010 and a high of 49% in FY 2017 for hires placed in the Terminals Track (see Tables B-9 to B-11 in Appendix B). Thus, there is considerable variation in losses from year to year, perhaps due to variations in the size of the hiring class. One possible explanation for high rates of DEV failures is that as the hiring class size increases, ATO is hiring further down the ranking of candidates who achieve a well-qualified score on its candidate screening test (see Chapter 6).
AFN’s approach to estimating future attrition has been relatively simple. According to AFN staff, they have used the failure rates of the preceding 2–5 years to project the next year’s attrition. With the projected high rates of hiring in FY 2025 and following years, AFN is considering developing a model that will account for hiring class size.
The committee compared the attrition estimates reported in each annual CWP report for Academy attrition with the actual attrition experienced in the following year over the FY 2010 to 2024 period. Over this 15-year period, AFN underestimated Academy attrition by 8%, which represents 321 fewer training successes than estimated, or an average of 21.4 fewer survivors each year than anticipated. The attrition reviewed above applies to the existing workforce. Expected attrition that will occur among annual new hires placed at the Academy is estimated in Step D but is discussed as part of this section’s review of the attrition model accuracy.
Developmental Attrition
DEV attrition is one of the smaller forms of attrition, averaging 102 annual losses over the FY 2010–2024 period (see Table 5-2). DEV attrition is estimated for the existing workforce in the calculations for C and estimated again in D for forecast new hires.
The failure rate among DEVs has varied over the 15-year period depending on the Track DEVs are placed in. The committee estimated DEV training failure rates on an annual basis between FY 2010 and 2020 only since many were still in progress after 2020. AGs and hires with previous experience placed in facilities can be of three separate types. Hires with previous experience are the smallest in number of new hires (representing
roughly 13% of each hiring class). Because many hires with previous experience gained that experience in the military, which does not have Centers, they are almost always placed in the Terminal Track. DEV attrition among previously experienced controllers is generally the lowest each year, but annual attrition has ranged the most from year to year from as low as 0% to as high as 16%. AGs placed in Terminals have annual attrition rates that range from a low of 4% to a high of 11%. AGs placed in the Centers Track have annual failure rates that range from as low as 3% to as high as 10%.
Attrition from training failures in facilities may occur more than 1 year after placement. AFN’s projections of DEV training failures rely on the failure rates of the preceding year to forecast the following year; they account for the failure rates of DEVs who failed in that year who may have been hired 1 or more years earlier. Based on the committee’s comparison of projected and actual DEV training failures between FY 2010 and 2024, AFN’s projected DEV training failures overestimated DEV failures by 21%. The overestimates, however, totaled 399 over the 15-year period, or 26 more survivors per year than anticipated.
Promotions and Transfers
This category of attrition refers to promotions and transfers of CPCs to various kinds of supervisory or specialist positions in facilities or other parts of ATO that are outside of the ATC workforce. They were the second largest form of attrition in the FY 2010–2024 period and will likely grow in importance as retirements decrease and the demand for supervisory and specialist positions grows with an increasing workforce. AFN monitors and accounts for the future demand for these positions because they are filled by CPCs, which creates vacancies among CPCs that need to be accounted for in planning for overall staffing and setting the annual hiring goal.
Some explanation is required to explain why this category of promotion and transfer is necessary because of the varied nature of positions that CPCs will be expected to fill. Facility Support Specialists and various types of Managers and Supervisors are of several different types and are sometimes referred to regarding the pay band that applies to them:
- Traffic Management Coordinators are unionized under the National Air Traffic Controllers Association (NATCA) Air Traffic contract and are in the same negotiated pay tables as CPCs; these employees also have a mandatory retirement age of 56.
- Staff Support Specialists (MSS1) are currently unionized under a contract with NATCA referred to as the “NATCA Integrated Agreement”; their negotiated pay rates are referred to as the MSS1
- pay table from which their acronym is drawn. They are not subject to the age 56 mandatory retirement.
- Operations Supervisors and Supervisory Traffic Management Coordinators are first-level supervisors for controllers and MSS1 employees. Their pay is established in the MSS 2/3/4 pay tables. They are subject to the age 56 mandatory retirement.
- Operations Managers and Support Managers (MSS2 and MSS3) are second- and third-level managers in the air traffic operation and their pay is set in the MSS 2/3/4 pay tables. They are not subject to the age 56 mandatory retirement.
- Air Traffic Managers and District or General Managers (MSS4), who are at the highest levels of supervision, are paid per the MSS 2/3/4 pay tables. They are not subject to the age 56 mandatory retirement.
AFN anticipates CPC vacancies created by promotions or transfers to these Other Than Controller Workforce (OTCWF) positions. AFN models losses in each of the staff categories listed above due to retirements and RRDs using the same procedures described above for the Air Traffic Controller workforce. They also account for transfers of OTCWF employees to other positions in FAA that are outside of the ATC and OTCWF workforce categories because these transfers will create open positions in the OTCWF for which CPCs will be needed to fill. Transfers within the OTCWF that do not create demand for CPCs to replace them are not included.
FAA accounts for the future growth in demand for these OTCWF employees for each of the five categories of staffing above based on its forecasted growth in the ATC workforce. These forecasts are supplemented with specific guidance from ATO about changes it expects to implement in these job categories in addition to those that would come from an increasing workforce (this guidance is manually added to the forecasts). Although these modeling exercises are meant to anticipate losses of CPCs that AFN should estimate in estimating the annual hiring goal for hires that will ultimately become CPCs, they also inform ATO’s planning for an adequate number of future managers and specialists. The committee reviewed projected and actual losses in the promotion and transfer category across the FY 2010–2024 period. For this entire period, AFN’s estimates were 2% lower than actually occurred, or a total of 112 employees or 7.5 per year.
Summary
According to the committee’s estimates, the sum of all sources of estimated attrition calculated at each facility compared to actual attrition over the FY 2010–2024 period indicates that AFN overestimated all forms of attrition
by only 1.1%, or about 15 more modeled hires each year out of projected annual hiring goals for FY 2025 and 2026 of 2,000 hires and 2,400 hires, respectively (FAA 2024). Given that AFN renews its forecasts each year and sets new hiring goals based on modeled estimates, this amount of annual error is reasonable and easily corrected for in subsequent years. The attrition models and procedures that AFN uses as input for setting the hiring goal are appropriate and sufficiently accurate. How well facilities are actually staffed is accomplished by the administrative procedures known as “Plan Execution,” which is the subject of Chapter 6.
CALCULATIONS FOR D: ESTABLISHMENT OF ANNUAL STAFFING GAP
The previous steps in the A−B+C=D modeling process estimate how many CPCs and CPC-ITs facilities should have in estimating A, subtracts from that the number of CPCs and CPC-ITs facilities have on board at the end of the preceding FY in estimating B, then adds the expected attrition the facilities will experience in the next and future years in Step C. The result is the gap between the desired level of staffing and the projected staffing for the target FY. This estimate then informs the setting of the annual hiring goal. Closing the staffing gap based on these hires will be achieved when the surviving new hires certify to CPC status, which is a multiyear process. Because the workforce is constantly in flux, hiring each year with the intent of matching the number of staff needed at each facility is always an ongoing process.
Chapter 2 explained the constraints on hiring that FAA experienced in FY 2013–2023 that caused it to fall far short of modeled estimates of hiring. This explanation is not repeated here. As also noted in Chapter 2, practical considerations also affect the setting of the hiring goal, including budgetary considerations and constraints on ATO’s ability to train new hires at the Academy and in the field. Hiring goals to address modeled estimates that are larger than FAA’s training capacity are typically spread over 2–3 years. As also noted in Chapter 2, ATO’s forecast hiring for FY 2025 and beyond is considerably higher than previous years.
FINDINGS, CONCLUSIONS, AND RECOMMENDATIONS
This chapter has reviewed the modeling and calculations FAA carries out to inform ATO’s decisions regarding the number of new hires it should make each year to ultimately close the gap between facility staffing standards and current CPCs and CPC-ITs and forecast attrition. In the committee’s view, the steps taken and models and estimating procedures are reasonable ones but could be improved.
Finding 5-1: The number of facility FTEs in a given year is a better assessment of facility headcount than the number of total staff on hand at the end of the preceding FY.
Finding 5-2: The CPC Equivalent Workforce expressed in FTEs is a more accurate estimate to apply to the assessment of workforce strength than the number of CPCs and CPC-ITs at the end of the preceding FY. Its use was recommended by the committee for the NASEM 2014 report but was not adopted by FAA.
Finding 5-3: The assessment of workforce strength in developing an annual hiring goal using the AFN method assumes that CPC-ITs are fully qualified, when, by definition, they are totally unqualified when initially transferred and only partially qualified as they progress to full certification.
Finding 5-4: Both the AFN and CRWG assumptions about facility staffing adequacy underestimate the capabilities of DEVs (in the AFN case) and non-CPCs (in the CRWG case) to carry out productive work, and the AFN overestimates the capabilities of CPC-ITs to carry out productive work. Data cited above indicate that CPC-ITs and DEVs can meet their ATO-established training goals while also carrying out a certain level of productive work. In addition, regular stints of independent TOP for DEVs and CPC-ITs are part of their formal training beyond that required for them to maintain currency on their existing certifications.
Recommendation 5-1: As recommended by the committee for the NASEM 2014 report, FAA should further refine the CEW concept and use it for assessing facility workforce strength in establishing the annual hiring goal.
Finding 5-5: The attrition models that AFN has developed and refined over the years are appropriate and adequate for their purpose of estimating future facility losses as part of the process of determining the annual staffing gap and hiring goal. These attrition models are unable to estimate attrition at each facility with precision, but they are not used for this purpose or for setting facility staffing standards.
Conclusion 5-1: FAA’s current roughly 14,000 member ATC workforce requires a constantly flowing pipeline of new hires to replace the roughly 9% of annual losses to the workforce each year. FAA’s ability to implement projected hiring estimated in its overall modeling process, and estimate and carry out hiring levels appropriately, are critical to
rebuilding the size of the ATC workforce sufficient to ensure a safe and efficient National Airspace System.
Conclusion 5-2: If FAA is able to set hiring goals that match modeled estimates in future years, it will be able to make progress in reducing the staffing gap between its facility staffing standards and existing headcount. Congress has authorized hiring at the limit of FAA’s training capacity in the 2024 FAA Reauthorization Act, but constraints on spending to achieve the administration and Congress’s goal to reduce $2 trillion from the federal budget over the next 10 years at the time of this writing may reduce its ability to do so.
Recommendation 5-2: FAA should hire according to its modeled estimates (as revised per Recommendations 4-1–3) of the projected hires needed to ultimately bridge the staffing gap, and Congress should provide the agency with the resources required.
REFERENCES
FAA (Federal Aviation Administration). 2023. “Air Traffic Controller Workforce Plan: 2023–2032.” U.S. Department of Transportation
FAA. 2024. “Air Traffic Controller Workforce Plan: 2024–2033.” U.S. Department of Transportation.
MITRE. 2023. “Collaborative Resource Workgroup Certified Professional Controller Targets: Methodology,” November. The MITRE Corporation.
NRC (National Research Council). 2014. The Federal Aviation Administration’s Approach for Determining Future Air Traffic Controller Staffing Needs. The National Academies Press. https://doi.org/10.17226/18824.
