Improving the Recruitment, Retention, and Utilization of Federal Scientists and Engineers (1993)

The federal civil service system, with its strong emphasis on internal equity, has long hampered the government's abilities to compete for scarce talent in the labor market and to reward exceptional individual performance. A number of these instances have involved scientists, engineers, and medical personnel. Accordingly, over the years, a number of special authorities have been employed to enable the government to be more successful in recruiting and retaining technically trained employees, which are described in this chapter, both to underline the need for flexibility in recruiting highly qualified personnel and to indicate the nature of mechanisms that can be usefully employed. More recently, authority to conduct personnel management demonstrations contained in the Civil Service Reform Act of 1978 has been used to address the problems of recruiting, retaining, and motivating scientists and engineers. These demonstrations were the basis for many of the pay-related flexibilities contained in the Federal Employees Pay Comparability Act of 1990 (FEPCA). They are described here, because they provide valuable lessons for the successful implementation of FEPCA, which is analyzed in the next chapter.

The Classification Act of 1949 established the General Schedule, a single, nationwide pay structure for federal white-collar employees that today consists of 15 grades, each with 10 pay steps. The grade level of each position in the more than 400 occupations employed by the federal government is classified on the basis of that position's complexity and

degree of responsibility. Thus each grade includes a very heterogeneous set of jobs—scientific, engineering, legal, accounting, medical, and other professions along with administrators, technicians, and clerical positions—that are deemed to be of equal rank, and they are paid within the same relatively narrow range regardless of geographic location.

The 1949 act, and the 1923 classification act that preceded it, were passed in response to interagency competition for skilled employees that occurred during the world wars; the new civilian war agencies that were created outside the civil service to give them more flexibility proceeded to steal skilled and experienced employees from the agencies under civil service by paying them higher salaries. This is the origin of the emphasis of the current system on internal equity among similar federal jobs without regard for external or market competitiveness (OPM, 1989b). Thus, for example, federal electrical engineers throughout the country who are in positions at a certain level of responsibility are paid basically the same. Moreover, they are paid the same as, say, chemists, mathematicians, and lawyers in the same grade, despite wide geographic and occupational variations in market-determined pay for those jobs across local labor markets.

In 1962 the Federal Salary Reform Act required that General Schedule (GS) pay rates for federal white-collar employees be comparable to pay rates in the private sector for similar levels of work. When federal pay levels nevertheless lagged, the Federal Pay Comparability Act of 1970 established regular procedures for making annual comparability adjustments in GS pay levels. The Bureau of Labor Statistics conducted sample surveys of salaries for 29 "benchmark" occupations across the country, which were used to derive a single average salary for each pay level. This system worked reasonably well for a time, but beginning in 1978 presidents proposed and Congress acceded each year to federal pay raises smaller than those required to achieve comparability with the private sector.¹ By 1990, the average pay gap reached 25 percent.

Table 3-1 displays the increases needed to achieve comparability and those actually granted from 1978 through 1990.

The problem with the comparability system, even when it was working, was that it posited an artificial monolithic labor market with a single national pay rate for the heterogeneous set of occupations classified at each grade level, each with dissimilar skill requirements, local labor market conditions, training and educational background, and geographic and career mobility patterns. Even when "comparability" is achieved under this system, a national average in conjunction with local and occupational diversity means that only some federal employees will be paid near the actual going rate in their field or where they live; the rest will be underpaid or overpaid. The system also looked only at salaries, not at fringe benefits such as health insurance, leave, and retirement, or at job security, even though these factors were often more favorable to federal than private employees.

As the GS pay levels grew more slowly than private-sector salaries generally since 1978, more and more pay disparities developed in more occupations and more localities. Eventually, better methods of comparing federal and private pay began to confirm that the disparities between federal pay levels and national average private-sector pay levels were real gaps in many local labor markets. The Office of Personnel Management (OPM), for example, commissioned a study comparing private and federal pay for certain federal jobs on a local basis, using 1987–1988 survey data. The study, conducted by the Wyatt Company, found that government salaries had fallen so far that it was becoming "difficult, if not impossible, to recruit and retain adequately qualified workers in some occupations and in some locations" (Wyatt Company, 1989:11–12). This problem affected government at all levels and probably in all departments and agencies, the report concluded.

More recently, the General Accounting Office (GAO) has conducted several studies comparing federal with private-sector pay by job and locality, which confirmed that private-sector salaries for similar jobs varied widely from area to area (GAO, 1990b, 1991a). The reports also showed that the general level of federal salaries had lagged so much behind the private sector's that the private sector was paying more for comparable jobs, regardless of locality. One GAO study of 10 occupations at different job levels in 63 metropolitan statistical areas (MSAs) in

1988 found that the private sector paid more in 90 percent of the MSA/job-level comparisons, an average of 21.9 percent more (in the other 10 percent of the MSA/job-level comparisons, the federal government was paying an average of 5.5 percent more) (GAO, 1990b:3).² The second study found that the size of private/federal-sector pay differences for 22 representative occupations (including chemists and engineers) varied substantially across MSAs and, within MSAs, across pay grades and occupations (GAO, 1991a:2). For example, the average private-sector advantage by MSA varied from 6.0 percent in San Antonio to 39.1 percent in San Francisco. However, the private sector paid more overall in all 22 MSAs studied, and the difference was more than 15 percent in 19 of the 22 MSAs.³

The second GAO report (1990a) presented the data for each MSA by occupation and GS level. The data indicate that private salaries for chemists, for example, are more uniform nationally than for most occupations, but the public-private pay gap for chemists is large—20 to 30 percent even in MSAs with a relatively small overall pay gap.⁴

This evidence of widespread salary disparities, small in some localities but large in others, and larger still for certain occupations in high-wage localities, provided support for the passage of FEPCA in Congress. The impetus for the administration's original proposal for pay

reform was based on earlier studies sponsored by OPM of the interplay between federal recruitment and retention trends, public-private compensation comparability, and local labor market variations in pay (e.g., Wyatt Company, 1989; OPM, 1989d). On the basis of these studies, OPM concluded that federal salaries were trailing the private sector ''in most occupations and most areas by widely varying amounts,'' the government was "experiencing significant recruitment and retention problems in some occupations and some geographic areas which cannot be addressed adequately under the existing system," and a shrinking supply of skilled labor in the future would "require more flexible, competitive pay practices to deal with growing recruitment and retention problems" (OPM, 1989e).

OPM cited other trends as indicating problems with the civil service system: the rapid increase in the use of the special rates program in the 1980s (from 11,500 in 1979 to 208,000 in 1992); the growing pressure to adopt more and more personnel demonstration projects as a way for agencies, especially those with science and engineering workforces, to escape the constraints of the GS system; and the growth of legislative initiatives to exempt agencies from the regular civil service provisions in Title 5, United States Code, when OPM would not sponsor demonstrations (e.g., the National Institute of Standards and Technology) (OPM, 1989a).⁵

The same year, the pay and compensation task force of the influential National Commission on the Public Service (Volcker Commission) concluded:

As the government's functions have multiplied since World War II, and the role of science and technology and other forms of expertise has expanded steadily in carrying those functions out, various mechanisms were developed to avoid the rigidities and pay limitations of the GS system. Perhaps the main device in the science and technology area has been the use of grants and contracts to tap the expertise of the academic, industrial, and nonprofit sectors. The government also relies heavily on advisory committees of outside experts. Within government itself, a number of mechanisms have been used to enable agencies to recruit and retain the scientific and technical expertise it needs to carry out critical functions, including various ways to exceed or bypass GS pay levels. These are described in the next section and summarized in Table 3-2.

As long ago as 1949, the Commission on Organization of the Executive Branch (First Hoover Commission) called for locality based pay for technical and clerical workers to enable the government to compete "on an equal footing" with the private sector. Although this advice was ignored when the GS system was created that year, "special," meaning higher, pay rates were authorized in 1954 for shortage occupations, defined as those for which private sector pay was seriously

hindering civil service recruitment and retention. Special rates range from 3 percent to nearly 30 percent above the GS schedule. These special rates were enacted primarily as a "relief valve" to address recruiting needs for scientific, medical, and technical positions (OPM, 1989a:11).

As the regular GS pay rates lagged after 1978, the use of special rates exploded, from 11,700 in 1978 (constituting 2 percent of all GS positions) to 47,000 in 1985 (3 percent), 170,000 in 1989 (10 percent), and 208,000 in 1992 (12 percent). They apply to 47 percent of the white-collar federal employees in Boston, 35 percent in New York, and 30 percent in Los Angeles and San Francisco. Many of the special rates are for technician and clerical jobs in very high wage areas, but nationwide just more than half are for science, engineering, and medical positions.

GAO (1990d:22–26) found that special rates were helpful but not sufficient to make federal pay competitive for certain occupations. One department gave most engineers the maximum allowable special rate, which is 30 percent more than regular starting salaries, but this still left new GS-7 level engineers with $5,200 (21 percent) less than the average offer nationally. The situation varied by location, of course, with one of seven sites employing engineers finding special rates to have a large effect, five a moderate effect, and one no effect.

Meanwhile, the Veterans Administration (VA), faced with staffing an expanded VA hospital system after World War II, was authorized in 1946 to set up an alternative pay system under Title 38 of the United States Code that allows the agency to set special rates for 12 medical occupations at any of its medical centers where higher pay is needed to

be competitive with other health care providers in the community. Approximately 75,000 employees are covered currently. This authority was extended to the National Institutes of Health (NIH) in 1986 when it was having problems staffing its clinical center with nurses and other clinical personnel (Institute of Medicine, 1988). By 1989 approximately 1,100 NIH employees were covered by Title 38 provisions (OPM, 1989a:66). FEPCA permits OPM to extend the pay and other features of Title 38 to employees involved in health care services in any agency.

Another legislative provision dating from the post-war period authorized agencies to establish "P.L.-313" positions, specifically to pay higher salaries to recruit and retain a small number of critically needed scientists and engineers. This authority was used, for example, by the National Aeronautics and Space Administration (NASA) to bring in experts from industry needed for specific advances in the lunar mission effort of the 1960s (Levine, 1982). Today these are known as ST positions and enable agencies to promote their top research and development (R&D) experts to higher paid positions (e.g., GS-16,-17, and-18 pay levels) without requiring them to go into management positions in the Senior Executive Service (SES). In 1991, there were about 250 ST scientists and professionals working in R&D.⁷ FEPCA includes them in the new senior-level positions that agencies can pay anywhere between 120 percent of step 10 of GS-15 and level IV of the Executive Schedule.⁸

In 1964 the Government Employee Salary Reform Act authorized agencies to hire above the step 1 rate for GS-11 or higher grades of the General Schedule, in order to accommodate a candidate's salary history or unusually high or unique qualifications, or to meet the government's special need for the candidate's services. This is another device whose use expanded rapidly, especially after it was delegated to the departments and agencies in September 1988.⁹ A GAO study (1991c) of a sample of 100 advance in-hire appointments at 10 civilian installations found that starting salaries averaged about $7,160 higher than step 1 of the position grade. The majority were used for engineering positions at NASA installations. Agencies also reported faster action in processing these personnel actions—taking only days rather than the months it had previously taken for OPM to approve advance in-hire appointments. FEPCA extends the authority to pay above the minimum to all grade levels.

Finally, a number of civilian white-collar pay systems have developed outside the regular civil service provisions of Title 5, primarily to enable agencies to recruit and retain highly qualified experts,

including scientists and engineers.¹⁰ These include the intelligence agencies (Central Intelligence Agency, Defense Intelligence Agency, National Security Agency); banking and finance agencies (Federal Reserve, Federal Deposit Insurance Corporation (FDIC), Resolution Trust Corporation (RTC), Federal National Mortgage Association); agencies run like private corporations (Tennessee Valley Authority, U.S. Postal Service); and congressional support agencies (General Accounting Office, Congressional Budget Office, Office of Technology Assessment). Typically, these agencies have much more flexible personnel systems in terms of hiring and advancement as well as higher salary structures (up to $150,000 for attorneys at FDIC and RTC). The VA system has already been described, which allows that department to compete more effectively for highly paid medical personnel. The National Science Foundation (NSF) has statutory authority to hire outside Title 5 which it uses for approximately 280 of its 1,180 positions. This excepted service enables NSF to recruit scientists, engineers, and other professionals of distinction and pay them in five broad pay ranges, the top two of which extend beyond the GS-15 level.

Previous groups examining the federal civil service system have recommended the creation of special occupational schedules for groups that do not fit comfortably in the GS pay and job classification structure. A 1972 task force, for example, recommended establishment of four regular pay/job structures, including one for administrators and professionals, and it suggested that special pay schedules be considered for certain groups, including scientists and engineers in R&D as well as health care workers, attorneys, law enforcement officers, and teachers (CSC, 1972). FEPCA includes authority to establish special pay systems for occupations, or groups of occupations, and it creates a Senior Biomedical Research Service in the Public Health Service that is exempt from Title 5 provisions in order to recruit outstanding biomedical researchers and pay them from the minimum rate for GS-15s up to level I of the Executive Schedule ($143,800 in 1992).

The civil service pay and job structure, the General Schedule, has often not been flexible enough for adequate recruitment and retention of highly qualified individuals in high-wage areas or occupations in high demand by private employers. Historically, this has affected federal employment of certain types of scientists and engineers in certain locations, and a variety of devices and mechanisms have been used over the years to deal with these problems, some of which are embodied in FEPCA. Many of them had also been systematically tried out in a series of personnel management demonstrations before they were included in FEPCA.

The Civil Service Reform Act of 1978 included authority to conduct and evaluate personnel demonstration projects requiring waivers from various Title 5 requirements. Currently, there are six projects involving about 21,000 employees. Three of the projects, described below, are efforts to improve the recruitment, retention, and performance of scientists and engineers. (The projects are profiled in Appendix C.)

The longest-running personnel management demonstration project is the Navy's, popularly known as "China Lake" after one of its two sites, the Naval Air Warfare Center in China Lake, California. The other site is the Naval Command, Control and Oceans Surveillance Center, in San Diego, California. The Navy project, which began in 1980, was designed to address the difficulties the two weapons labs were experiencing in recruiting high-quality scientists and engineers and keeping the best of them by making position classification and pay administration more flexible. Navy lab managers were also interested in delegating more managerial control to supervisors by increasing their control over classification, pay, and other personnel matters. Pay increases and promotions are tied more closely to performance ratings.

The 8,700 civilian white-collar positions at the two labs are grouped into five "career paths," each associated with a salary range. There are career paths for (1) scientists, engineers, and other professionals; (2) technicians; (3) technical specialists; (4) administrative specialists; and (5) clerical personnel. The salary range for each career path is divided into four to six broad bands that each span at least two GS grades, which makes it easier to offer desirable recruits a higher starting pay level. It also dovetails with the project's performance-based approach to pay by making it possible to move high performers to relatively higher pay levels than adequate or poor performers because the pay band is much broader than a GS grade.¹¹ This makes it possible to increase pay substantially without having to promote the employee. Finally, since 1987, the labs have been authorized to offer recruitment bonuses of up to 15 percent of starting salary.

Under the traditional GS system, pay levels are adjusted every January, and all employees (except for supervisors in grades GS-13 through-15 who are in a merit review program called the Performance Management and Recognition System) receive the annual raise automatically. Employees also receive within-grade increases of one step (about 3 percent) every one, two, or three years, depending on length of service and a performance rating of fully successful or better. Regular GS employees may be given quality step increases at any time for outstanding performance, which advances the employee to the next-higher step earlier than the normal waiting period, but only a very small fraction receive them. Since more than 90 percent of employees are rated fully successful, and only a few percent receive quality step increases, pay advancement in the GS system is determined almost completely by length of service rather than performance.

In the Navy project, employees receive the annual governmentwide pay adjustment if they are rated at least fully successful. The funding for within-grade increases, sustained superior performance awards, and in-

level promotions, however, is put in an incentive pool and given out in proportion to the points earned in annual performance appraisals. The labs also pay bonuses, mostly to reward performance on temporary assignments or for one-time accomplishments. The incentive and bonus pools together amount to 3.2 percent of salaries at the China Lake lab and 3.3 percent at the San Diego lab.

OPM is conducting the evaluation, which compares the project to two similar labs, the Naval Surface Warfare Center (White Oak, Maryland, and Dahlgren, Virginia) and the Naval Air Warfare Center (Warminster, Pennsylvania). To date, 14 reports have been issued by OPM. Among the key findings, according to OPM, are that, compared to control labs, (1) starting salaries for scientists and engineers have increased substantially (but subsequent pay progression during the first five years is slower under banding than under the GS system of grades); (2) larger pay increases are given to highly rated employees, which has greatly increased the link between performance and pay; (3) turnover among scientists, engineers, and other professionals with high performance ratings has fallen and has consistently been lower at demonstration labs than at control labs (although the turnover at all the labs is affected by trends in the economy, falling when unemployment rises and rising when unemployment falls) (OPM, 1988b, 1991a).¹²

OPM recently completed its analysis of the effect of flexible starting salaries, recruitment bonuses, and pay for performance on recruitment success at the demonstration labs (OPM, 1991b). According to OPM's evaluation, the demonstration labs have been more successful in fulfilling their staffing requirements since the demonstration began. The turnover rate among employees with high performance appraisals has consistently been about 50 percent lower at the demonstration labs compared to the control labs (but turnover has been declining in all labs since 1987, and more steeply at the control labs, even before unemployment rates began to increase). Finally, certain quality indicators (grade-

point averages and job acceptance rates) increased at the Naval Air Warfare Center at China Lake, which had more recruitment problems than the San Diego center.

In a longitudinal attitude survey, the percentage of lab managers agreeing that their center was able to attract high-quality candidates increased at the demonstration labs from 47 percent in 1981 to 60 percent in 1987 before falling to 47 percent in 1989. Meanwhile, the percentage of control lab managers who said they were able to attract high-quality candidates hovered around 50 percent until 1984, before dropping to 34 percent in 1987 and 30 percent in 1989. OPM also found higher and increasing job satisfaction among scientists and engineers at the demonstration labs relative to the control labs—76 percent in 1979 and 81 percent in 1989 at the demonstration labs, compared with 73 percent at the control labs in both years (CBO, 1991:25, citing OPM, 1988c, and unpublished OPM data).¹³ OPM concluded that the demonstration labs were generally better able to recruit and retain scientists and engineers than control labs, and, on average, the quality of recruits was higher. Higher salaries, along with pay for performance, appeared to be the main reason; demonstration labs were able to pay 20-30 percent higher starting salaries than control labs for comparable positions.

In 1986, Congress mandated a personnel management demonstration project to be carried out at NIST. It was a response to the failure to adopt legislation implementing the personnel-related recommendations

of the Packard Panel, which called for the creation of ''a scientific/technical personnel system independent of current Civil Service personnel systems'' at government-operated labs (OSTP, 1983). The project, which began in January 1988, is an expanded version of the China Lake project and has similar aims: to improve recruitment and retention of high-quality scientific, technical, and other employees and to simplify the personnel system in order to increase the control of managers over personnel decisions. It covers 3,000 white-collar employees at research labs in Maryland and Colorado.

Like China Lake, the NIST project features simplified position classification through the consolidation of narrow GS grades into broader pay bands and a pay-for-performance system. As originally conceived, the project added the concept of total compensation (pay and benefit comparability with private sector pay and benefits, which was to be achieved by granting raises larger than the annual governmentwide pay adjustment for the General Schedule. By paying larger comparability increases, NIST could make its salary levels approximate the private sector's and thus improve recruitment and retention of scientists and engineers. Instead, NIST found that pay disparities varied significantly by occupation and site. Larger general comparability increases would have NIST overpaying some employees in some occupations while still underpaying others in one or both sites (the same problem encountered by the GS pay comparability system established in 1970). Instead, the NIST director has opted to pay the governmentwide increase in 1989, 1990, and 1991, and rely on higher starting salaries, performance pay increases, promotions, and bonuses to redress individual pay disparities and reward deserving employees (OPM, 1990:18).

All white-collar occupations are grouped in four career paths: (1) scientific and engineering professionals; (2) scientific and engineering technicians; (3) administrative staff; and (4) support staff. Each career path has a different pay range, divided into five levels reflecting the career stages in each path—entry and trainee, developmental, full performance, senior and supervisory, and managerial. Performance-based pay is allocated to employees through two pools, based on annual performance reviews. The first, for performance-based salary increases, consists of funds formerly used for within-grade step increases, quality step increases, merit pay increases, and promotions from one GS grade

to another now within a single pay band. The second pool, for cash bonuses, includes funds formerly used for Performance Management and Recognition System awards and other awards.

The project included expanded direct-hire authority and agency-based examining authority, which was used to fill 95 percent of the science and engineering vacancies during the first three years of the project. As a result, NIST has been able to reduce the average time to fill vacancies by several weeks (Rosenthal et al., 1991:13).

The project also included authority to give recruitment and retention allowances. This authority, however, has been used very selectively. In the first two years, recruitment allowances averaging about $5,700 were given to 17 new hires and one retention allowance of $5,000 over two years was awarded, all to scientists and engineers mostly in the top rather than entry levels (Rosenthal et al., 1991:10). However, as in the Navy demonstration, regular starting salaries were higher at NIST than before and higher than in a comparison lab in Colorado, which may account for the low use of recruitment bonuses at lower levels.

Finally, pay increases at NIST are strongly related to performance ratings, which they are not in comparison organizations in the Department of Commerce. NIST employees at high rating levels are receiving larger pay increases, and employees with low ratings are receiving smaller pay increases, than their counterparts in the comparison organizations. Also, NIST employees with higher ratings are more likely to receive performance bonuses (OPM, 1990:ii).

According to the third-year evaluation report, NIST has been able to pay new hires more than other agencies for comparable jobs. Overall, the salaries of NIST scientists and engineers increased 5 percent more than those for scientists and engineers in comparable federal agencies between 1987 and 1989. The comparability of science and engineering salaries with the private sector did not change, but they would have been less comparable in the absence of the demonstration project (Rosenthal et al., 1991:Ch. 4).

There have been no significant changes in NIST's workforce quality indicators—undergraduate grade-point averages, quality of graduate schools, and performance appraisal ratings—which traditionally have been high. Turnover rates, already very low, have not changed,

although surveys of leavers indicate that salary has become a less important reason for leaving.

The committee also heard presentations by officials from the U.S. Department of Agriculture (USDA) and its Agricultural Research Service (ARS) about a personnel demonstration project begun in mid-1990 in 140 experimental and 70 comparison sites of the Forest Service and the ARS. The project is largely a testing of a comprehensive simplification and decentralization of the hiring system, but it does include recruitment bonuses and relocation expenses and use of an extended probationary period for scientists in research positions.¹⁴ The committee heard that the recruitment incentives have been rarely used, although they were important in attracting a microbiologist and a plant physiologist. It is too early to evaluate the results of the longer probationary period on the retention and productivity of research scientists.

While they are only quasi-experiments, and they do not measure the effects of the interventions on organizational effectiveness, other effects of the various interventions have been measured, the Navy and NIST demonstration projects are consistent with the proposition that a more flexible pay and position structure improves the ability of federal agencies to recruit more qualified scientists and engineers and to reward and motivate good performers and thus retain them. They also show that the direct cost of such efforts is modest, in part because the agencies can (and do, because of budget constraints) tailor the compensation package to each case rather than increase salaries across the board. In addition, the differences among the demonstrations designed by each agency to

meet its needs show that the various mechanisms can and should be adapted to the particular conditions facing each agency. Thus the agencies faced with implementing FEPCA should consider it an opportunity to design their own recruitment and retention programs. Unfortunately, FEPCA does not include all the devices and flexibilities being used by the demonstration projects. Additional steps needed beyond FEPCA to improve the federal government's capacity to recruit well-qualified scientific and technological personnel are recommended in the last chapter. First, the next chapter discusses the flexibilities offered under FEPCA to agencies faced with attracting and keeping well-qualified scientists and engineers and identifies potential obstacles to effective implementation of FEPCA.