July 27, 2016 | Backgrounder on Labor
The federal government pays its employees more than they would earn in the private sector. Economic studies consistently find that federal employees enjoy both higher pay and substantially higher benefits than comparable private-sector workers. Federal pay is inflated because market forces do not discipline it. In the private sector, productivity determines workers’ pay. Businesses that pay workers much less than the value they add lose them to competitors who offer more. Businesses that pay workers more than their productivity level tend to go out of business. In general, market forces cause workers in the private sector to earn what their productivity merits. Bringing federal employee compensation and personnel procedures closer in line with private-sector compensation would not only improve government efficiency, it would also generate significant savings. Combined, the reforms in this Heritage Foundation Backgrounder would reduce federal personnel costs by $333 billion between 2017 and 2026.
The federal government pays its employees more than they would earn in the private sector. Economic studies consistently find that federal employees enjoy both higher pay and substantially higher benefits than comparable private-sector workers.
Alan Krueger, the former Chairman of President Barack Obama’s Council of Economic Advisers, documented this pay premium in the 1980s. Academic researchers have repeatedly found similar results. More recently, researchers at the Congressional Budget Office (CBO), The Heritage Foundation, and the American Enterprise Institute (AEI) re-examined this question. These studies examined different data sources and used different econometric models. They all concluded that the federal pay premium remains considerable—particularly when including employee benefits. Table 1 highlights their results.
The Heritage study found that federal employees receive 22 percent higher wages than similar workers in the private sector. Including the value of employee benefits, the total compensation premium increased to between 30 percent and 40 percent. The CBO found a small wage premium (2 percent), but substantially inflated benefits, producing an overall compensation premium of 16 percent. AEI found a 14 percent pay premium and a 61 percent total compensation premium.
This federal compensation premium costs taxpayers heavily. The Office of Management and Budget (OMB) estimates that taxpayers will spend $337 billion in 2017 employing civilian federal employees. The CBO’s estimates of the compensation premium imply that taxpayers would save $47 billion that year if the government paid its employees according to market rates. That represents about one-tenth of the budget deficit that the OMB forecasts for 2017. The Heritage Foundation and AEI studies imply even greater savings.
How and Why Federal Pay Is Inflated. Federal pay is inflated because market forces do not discipline it. In the private sector, productivity determines workers’ pay. Businesses that pay workers much less than the value that they add lose them to competitors who offer more. Businesses that pay workers more than their productivity level tend to go out of business. Broadly speaking and under normal economic conditions, market forces cause workers in the private sector to earn what their productivity merits.
Market forces do not similarly influence federal pay because the government operates outside the marketplace. It does not have to compete for its taxpayer dollars and it enjoys significant advantages in its ability to finance its deficits. Absent market forces, the federal government determines pay and benefits through formulas set by Congress. In theory, the Office of Personnel Management constructs these pay scales to reflect market wages for similar private-sector jobs, but in practice, government wages frequently bear little resemblance to market pay.
The basic federal pay scale is the General Schedule (GS), which covers roughly 70 percent of federal civilian employees. Separate pay systems cover political appointees, senior executives, and blue collar workers.
The GS consists of 15 pay grades and 10 steps within those pay grades. Each grade requires progressively greater skills and experience. GS grades 1 through 7 denote entry-level positions, while grades 8 through 12 mark mid-level positions and grades 13 through 15 are higher-level and management positions. The GS also incorporates locality pay adjustments to account for cost-of-living differences across the country and overseas.
Table 2 displays the 2016 GS for the Washington, DC, metropolitan area, where one-sixth of all civilian federal employees work. It also shows the proportion of all GS federal employees at each grade. (The table excludes Senior Executive Service employees—the federal government’s top managers—who make a minimum of 20 percent more than GS-15 base pay).
On average, GS employees earn more than they would in the private sector. This is a result of how the federal government sets GS pay levels. The government does so by determining the GS grades that hypothetically correspond to various private-sector jobs. The government then averages pay across the positions assigned to each GS level. However, private firms use nothing like the GS to determine pay levels. Occupations involving similar skill levels can pay very differently depending on market conditions. As a result, GS pay rates often differ wildly from similar private-sector jobs.
Moreover the federal government suffers from GS “grade inflation.” Many federal employees fill jobs with much more senior titles than their skills would merit in the private sector. As a result, they enjoy inflated GS grades and higher pay. This grade inflation explains why over three-fifths of federal GS employees belong to GS grade 11 or higher.
In addition to setting artificially high starting wages and inflating workers’ pay grades, the government’s two automatic pay increases further inflate federal pay. First, federal employees get an annual pay adjustment based on average wage growth. This pay increase functions as a cost-of-living adjustment, but is greater than a typical inflation adjustment because employee compensation rises faster than prices over time. It increases pay across the board for every grade and step of the General Schedule. In 2016, for example, federal employees received a roughly 1.3 percent increase (with slight variations based on local living costs).
Second, federal employees advance up the 10 steps of their GS grade through seniority. Each “step increase” (also called a “within grade increase” or WIGI) brings an approximate 3.0 percent increase in pay. Employees in steps 1 through 3 advance a step every year. Employees in steps 4 through 6 increase a step every two years. Employees in steps 7 through 9 advance a step every three years. Federal employees will thus reach step 10 of their grade after 18 years. At that point, they will earn about 30 percent more than they did in step 1.
Except in cases of extreme misconduct, these step increases occur automatically. If a supervisor attempts to withhold an employee’s WIGI, he must prove that the employee performed unacceptably. The employee can appeal that determination through either a union grievance or to the Merit Systems Protection Board, and most do so. Defending that decision can cost an agency tens or hundreds of thousands of dollars. Consequently, supervisors almost always grant step increases. Only about 0.06 percent—one of every 1,700 federal employees—are denied a scheduled step increase. WIGIs function as a second automatic pay increase, layered on top of the annual pay adjustment. These two automatic raises systematically increase federal compensation costs for taxpayers.
Congress can reduce this burden on taxpayers by reducing the magnitude of automatic pay increases, increasing the use of performance-based bonuses, and ending policies that result in managers giving raises to over 99.9 percent of all workers. Congress should:
Estimated Savings: We estimate that the reductions in salary-step increases would reduce federal personnel costs by $26.7 billion as a stand-alone reform from 2016 to 2027, and by $25.8 billion when combined with our other proposed reforms.
Generous retirement benefits account for the largest difference between public-sector and private-sector compensation. Federal employees receive both a substantial defined-benefit (DB) pension, as well as a considerable defined-contribution (DC) plan. Private employers, on the other hand, typically provide only one type of retirement plan, if they offer one at all. According to recent data from the Bureau of Labor Statistics, 76 percent of full-time, private-industry workers have access to retirement benefits at work, but only 31 percent of workers in the lowest decile of earners had access to retirement benefits, compared to 88 percent of workers in the top decile.
On average, private employers who offer retirement plans typically contribute a maximum of between 3 percent and 5 percent of employees’ salaries to their retirement plan. The federal government’s retirement contribution—equaling between 15 percent and 18 percent of employees’ pay, depending on their year of hire—dwarfs that of private employers.
The federal defined-benefit pension system, the Federal Employees Retirement System (FERS) has an estimated cost of between 14.0 percent and 14.2 percent of payroll. Federal employees who were first hired before 2013 pay 0.8 percent of their pay toward the government’s defined-benefit pension system and taxpayers pick up the remaining 13.2 percent of payroll cost. Employees hired in 2014 and thereafter contribute 4.4 percent of their pay to the FERS, and taxpayers pick up 11.1 percent (1.3 percent of employee contributions are used to pay off the unfunded obligations of the Civil Service Retirement System (CSRS)).
In addition, all eligible federal employees are also enrolled in the government’s 401(k)-style, defined-contribution Thrift Savings Plan (TSP). Employees receive an automatic 1 percent contribution, and up to an additional 4 percent in matching contributions. The combination of an 11.1 percent or 13.2 percent FERS contribution and up to a 5.0 percent TSP contribution makes federal employees eligible for retirement benefits that add between 15.1 percent and 18.2 percent of pay to their total compensation.
Defined-Benefit Plans. DB pensions provide a predetermined monthly retirement benefit, dependent on workers’ years of service, salary, and retirement age. By guaranteeing the benefit level, defined-benefit plans shift investment and other risks (such as increased life expectancies) onto employers.
DB plans were established to provide a guaranteed level of income no matter how long one lives, but many public and private DB plans paid out benefits before they had been accrued, did not provide adequate contributions, and have been plagued by short-sighted and self-interested management. As a result, many DB pensions cannot provide anywhere near the benefits they promised, and the DB structure that was supposed to provide greater security for workers has instead created a false sense of security for many.
In part because of the risk that DB plans place on employers, as well as the lack of ownership and autonomy they provide for workers, private-sector employers have moved away from providing DB pensions. In 2015, only 8 percent of private employers, covering 18 percent of private-sector workers, offered DB plans.
Defined-Contribution Plans. Significantly more private employers provide DC retirement plans. In 2015, 47 percent of private employers, covering 61 percent of private-sector workers, offered DC plans. Larger employers are significantly more likely to provide DC plans; 86 percent of firms with more than 100 employees offered DC plans, compared to 45 percent of firms with fewer than 100 employees.
DC plans remove the risk and future liability of offering retirement benefits for employers. Rather than requiring firms to lock in a future payment stream, DC plans allow employers to provide a specific contribution with no guarantee of the level of future benefits those contributions will generate. While DC plans shift the risk of varying investment returns and unexpected longevity to employees, they also remove the risk for workers of an employer going bankrupt or promising more than he can afford to pay. Moreover, DC plans provide employees with tangible, real-time compensation that they own and control.
Pension Changes. Without taking away any benefits that federal employees have already accrued, and without forcing vested employees to leave the current system, retirement benefits for federal employees should shift from the existing hybrid of oversized DB pensions and above-average DC contributions towards an exclusive—although still generous—DC system that more closely resembles private-sector retirement benefits.
This shift would provide workers with a better comparison between federal and private employment options and would give them immediate ownership and control over their retirement compensation. This would also reduce future taxpayer liabilities by requiring the government to recognize its employment costs up front instead of potentially passing them on to future generations.
Retirees and Near-Retirees. Nothing will change for federal employees who are already retired, those who have left federal service, or who have 25 years or more of federal service. Retirees who are receiving benefits will continue to do so as scheduled under current law and workers who have 25 years or more of service will continue to accrue both FERS and TSP benefits as they do under current law.
New Hires and Non-Vested Employees. New hires and non-vested employees (those who have fewer than five years of service and therefore are not yet eligible to receive FERS benefits) will shift to a new system that provides higher Thrift Savings Plan contributions, but no FERS contributions or benefits. Non-vested employees will receive a portion of the FERS contributions that have been made on their behalf in the form of a lump-sum rollover into their TSP.
Employees will receive an automatic 4 percent TSP contribution (instead of the current 1 percent), and will continue to be eligible for up to an additional 4 percent match, increasing the maximum government contribution from 5 percent to 8 percent.
Current Vested Employees. Employees who have already earned a FERS benefit but who do not have at least 25 years of service will not lose any of their accrued benefits. Going forward, however, they will have three options:
Each of these options aims to provide the same level of government contribution to all employees—a maximum of 8 percent of payroll—regardless of their choice of plan(s). This results in a savings of roughly 10.2 percent of payroll for the government.
Option A: Remain in current system, accruing both FERS and TSP benefits, with changes to future FERS accruals. For employees who elect this option, Congress should:
Replace the highest three years of earnings (high-3) with the lifetime average for FERS benefit calculation. The current system bases employees’ FERS benefits on their highest three years of earnings. Instead of employees’ high-3, their average earnings will be used to calculate FERS benefits.
The current system inflates pension benefits and causes inequities among similar earners because benefits do not reflect lifetime contributions. Two workers could have exactly the same lifetime earnings and exactly the same FERS contributions, yet one with a big jump in earnings at the end of his career could receive twice the pension benefit as someone who had steady earnings throughout his career. This is not a fair or accurate way to calculate benefits.
To preserve benefits that workers already earned under the current system, workers would receive the greater of their existing high-3 years of earnings-to-date, or the average of their earnings accrued after enactment of the change. For example, if John Smith worked for the federal government for 15 years prior to the change in retirement benefits and he plans to work for another 20 years after the change, his FERS benefit will be based on either his highest three years of earnings over his first 15 years of service (pre-reform), or the average of his earnings over his last 20 years (post-reform).
Gradually increase the federal retirement age to that of Social Security and index it for life expectancy. Currently, federal workers can collect retirement benefits as early as age 55 to 57, depending on when they were born. Yet, non-federal workers currently in their fifties cannot collect full Social Security benefits until age 67. Life expectancy was less than 55 years in 1920 when the CSRS federal employees’ pension system was established. It is now above 78 years and rising, leaving little wonder why pension systems face funding shortfalls when retirees collect benefits for years longer than initially projected.
The retirement age for federal employee pensions should be gradually increased to that of Social Security—age 67—and then indexed to changes in life expectancy thereafter. If federal employees want to collect their FERS pensions at an earlier age, they should be able to do that with actuarially fair reductions in benefits.
In the example above, John Smith would be able to collect a little less than half his pension (the part earned over his first 15 years of service, three-sevenths his total pension) at his current minimum retirement age of 57, but would have to wait until age 67 to receive the rest (his remaining four-sevenths representing his last 20 of 35 total years). If he wanted to receive the entirety of his benefit at age 57, he could do so with an actuarially fair reduction in his latter-career portion of his pension (the remaining four-sevenths).
Option B: Maintain already accrued but frozen FERS benefits and receive higher TSP contributions. For employees who elect this option, Congress should:
Freeze FERS, increase TSP contributions. Employees can opt to maintain a future claim to their already accrued FERS benefits, with no changes to those benefits, while forgoing future FERS contributions and accruals. Employees who elect this option would instead receive the higher 4 percent automatic TSP contribution and would remain eligible for up to another 4 percent matching contribution, for a total 8 percent maximum contribution. Employees who chose this option could thereby avoid the increased payroll deduction required of employees who remain in the FERS system.
Using the example of John Smith above, who worked for the federal government for 15 years prior to enactment of the change and will work for another 20 after, John would receive a FERS benefit based on his first 15 years of service. He would be eligible to receive that benefit at his current minimum retirement age of 57, and his benefit would be based on his high-3 to date. Going forward, he would receive at least 4 percent in TSP contributions and up to 8 percent that could be invested and withdrawn at his discretion in accordance with existing laws regarding retirement account withdrawals.
Option C: Shift entirely to the TSP. For employees who elect this option, Congress should:
Permit lump-sum FERS rollover into TSP, higher TSP contributions. Employees could also choose to exit the FERS system entirely and receive a lump-sum benefit based on their accrued FERS benefit. Employees who want to forgo their future FERS benefits can take a lump-sum payout equal to 75 percent of the net present value of their future FERS benefits, and roll this amount over into their existing TSP account. These employees would receive the higher 4 percent automatic TSP contribution from the government, and up to another 4 percent in matching TSP contribution, for a total of up to 8 percent in employer contribution.
Although workers who choose this option would receive only 75 percent of the present value of their future pension costs, that amount would likely generate significantly larger benefits than FERS would provide. All FERS contributions are currently deposited into the notional Civil Service Retirement and Disability Fund (CSRDF), which is invested entirely in U.S. Treasuries. Because Treasuries generate significantly lower returns than stocks and other bonds, individuals who choose a lump-sum payout could end up with significantly greater retirement income than their existing FERS benefit would provide.
Based on current projections for a long-run nominal average 10-year Treasury rate of 4.1 percent and a conservative 7.0 percent nominal rate on stock market returns, individuals who opt for a 75 percent lump-sum payout of their already accrued FERS benefit would be able to receive substantially higher retirement income than the full value of their existing accrued FERS benefit would provide. For example, a 46-year-old federal employee with 15 years of service who will begin receiving retirement benefits at age 62 would accumulate an account balance (based on his 75 percent lump-sum payout) sufficient to provide 75 percent more than his accrued FERS benefit ($22,313 per year compared to an accrued $12,750 benefit). If he opted to withdraw only what his FERS benefit would have provided, he would be left with more than $538,000 in his account upon death at age 81. (The remaining balance could also provide continued withdrawals through age 108.)
Estimated Savings: While much of the reduction in personnel costs from the retirement provisions would not have an immediate impact on government costs, they would nevertheless generate real and substantial savings over the long run. Using a cost-accrual method that recognizes the cost that the government would have to pay in the current year to provide the benefits it promises in that year, we estimate that our proposed retirement reforms would reduce federal personnel costs by $207 billion as a stand-alone reform over the 2016–2027 period, and by $202 billion when combined with our other proposed reforms.
Retiree Health Benefit Changes. In addition to generous pensions, federal employees receive generous health benefits in retirement. Even after federal employees retire, they are eligible to receive the same subsidy for Federal Employees Health Benefits (FEHB) as current workers receive. Only about 15 percent of private employers provide retiree health coverage. With a minimum retirement age of only 57, workers can claim these taxpayer-financed benefits while working at jobs that otherwise would provide their health insurance (resulting in windfall benefits for employers who do not have to pay employees’ insurance costs).
The federal government should eliminate the subsidy for retiree health benefits for new hires. This would not generate immediate budgetary savings over the near term because the federal government does not incur any retiree health costs until employees retire. However, the proposal would generate significant future savings.
Estimated Savings: While the elimination of FEHB retiree benefits would not reduce government personnel costs until future decades, a 2002 study by the CBO estimated the accrual cost of retiree health coverage to equal 6.4 percent of pay. Thus, on an accrual basis (essentially what the government would need to contribute each year to pre-fund retirees’ health benefits), eliminating FEHB benefits for new hires would reduce federal personnel costs by $38.9 billion as a stand-alone proposal over the 2017–2026 period, and by $36.6 billion when combined with our other proposed reforms. Savings would grow significantly in later decades as a larger share of the federal workforce would no longer receive FEHB retirement benefits. (Savings would equal $130 billion over the 2027–2036 period.)
Federal employees receive significantly more paid leave than private-sector employees. A federal employee with five years of experience receives 20 vacation days, 13 paid sick days, and all 10 federal holidays. The average private-sector employee at a larger company receives 13 days of vacation and eight paid sick days, for a combined total of 21 days of paid leave (excluding holidays), compared to 33 for federal employees. Additionally, few private-sector employers give their employees paid leave on every federal holiday.
While the federal government’s generous annual sick leave and the ability to roll that leave over from year to year acts as a short-term disability insurance plan, individuals and the government would likely be better off with a private disability insurance plan, particularly if it also covered long-term disability. Not only is private disability insurance relatively inexpensive, it is also more effective at treating and improving outcomes for disabled individuals.
Though sick leave is supposed to be taken only when needed, specifying a certain amount of sick leave leads some employees to feel entitled to use all of that leave, even if they are not sick. This incentive has contributed to a shift among private-sector employers toward providing employees with Paid Time Off (PTO) that can be used for either vacation or sick leave. The typical private PTO plan provides 21 days of PTO (plus holidays) for a worker with five years of experience. PTO plans are favored by many employees who generally use little or no sick leave because it provides them with more total vacation days.
Congress can bring the amount of paid leave it provides to federal employees in line with private-sector paid leave by reducing both vacation and sick leave, or by implementing a new PTO system:
Under both options, Congress should consider auto-enrolling employees into a private disability insurance plan that would provide more timely, comprehensive, and effective coverage and rehabilitation services than Social Security’s Disability Insurance program.
These policies would boost federal employees’ productivity by increasing the number of days they work, and thus could reduce the number of federal employees needed to carry out government functions. While the proposed paid-leave options would reduce available leave by six to nine days (between three and six fewer vacation days and three fewer sick days), we conservatively assume that actual paid leave would decline by only five days on average, as employees do not currently use all of their paid leave (particularly sick leave). This increase in work days would allow the government to reduce total employment by 2.2 percent.
Estimated Savings: We estimate that these proposed reductions in paid leave would reduce total employment by 2.2 percent and save the federal government $73 billion in personnel costs as a stand-alone reform over the 2016–2027 period, and by $68.4 billion when combined with our other proposed reforms.
Federal employees have the option of choosing from a wide variety of health insurance plans through the FEHB network. The structure of the government’s contribution discourages workers from choosing lower-cost plans, however, which drives up both employees’ and employers’ costs.
Currently, the government contributes 72 percent of the weighted average premiums of all the health insurance plans in the FEHB, but employees must pay at least 25 percent, regardless of the cost of the plan they choose. This reduces federal employees’ incentives to choose less-expensive health care plans because 75 percent of the savings go to the federal government.
The popular Blue Cross Blue Shield (BCBS) Basic FEHB plan is a good example. Individual coverage costs a total of $7,122 in 2016—the government pays $5,342 and employees pay $1,780. The National Association of Letter Carriers (NALC) Value Option plan costs significantly less at $4,482, with the government paying $3,362 and employees paying $1,120. Even though the NALC Value Option plan costs $2,640 less than the BCBS plan, the employee’s contributions will go down by only $660 while the government will reap the remaining $1,980 in savings. This setup discourages employees from choosing lower-cost plans, even if those plans are advantageous to them.
Recognizing that employees will not choose less-expensive plans if they do not reap most of the savings, private employers that offer multiple health plans tend to pass the cost difference of those plans on to employees by requiring employees to pay a higher portion of higher-cost plans and a smaller portion of lower-cost plans. This structure encourages employees to choose plans based on a more accurate valuation and allows employers to provide more equal health insurance benefits to all employees.
The federal government should set its FEHB contribution at a flat-dollar amount equal to 72 percent of the average premium cost for all FEHB plans (the current maximum contribution amount). This would allow workers who choose lower-cost plans to keep all the savings and would increase competition among FEHB plans. Over time, greater competition would reduce the average cost of FEHB coverage to the taxpayers.
Estimated Savings: A 2011 report by the CBO estimated that a similar proposal that would provide a flat FEHB contribution with a specified limit on contribution growth would have saved $42 billion over 10 years. We propose a similar flat FEHB contribution, but instead of a specified limit on growth, we allow market forces to reduce cost growth over time. While we assume that this proposal would result in significant savings (similar in magnitude to those projected by the CBO and growing over time), we do not have sufficient information on how federal employees’ FEHB choices would change over time, and thus, how much the government’s costs would decline. Thus, we do not include estimated savings from eliminating the minimum 25 percent federal employee FEHB contribution in this analysis.
Federal law makes it very difficult to separate poorly performing federal employees from their jobs. Managers who need to fire problematic employees, whether because of misconduct or poor performance, must go through draining and time-consuming procedures that take about a year and a half. Consequently, the federal government very rarely fires its employees, even when their performance or conduct justifies it. In fiscal year (FY) 2013 the federal government terminated the employment of just 0.3 percent of its tenured workforce for performance or misconduct. This system shelters bad employees. It also raises costs for taxpayers; agencies must hire additional employees to get the job done. Congress should make dismissing federal employees less difficult by:
Estimated Savings: While improved hiring and firing procedures would likely reduce federal personnel costs through improved efficiencies, we do not include any estimated savings in this estimate.
Bringing federal employee compensation and personnel procedures closer in line with private-sector compensation would not only improve the efficiency of the government, but would also generate significant savings. Taken together, the above proposed reforms would reduce federal personnel costs by $333 billion between 2017 and 2026.
This amount includes the accrued value of certain benefits that would not register as savings until future years. It also includes the interaction between reform proposals. For example, the proposal to reform salary increases would lead to lower base salaries. This would affect other reforms, such as retirement benefits, that are a function of salaries. Consequently, the combined savings are lower than the sum of the stand-alone proposals as reported in each section above. (The sum of all individual proposals is $346 billion, compared to the combined savings of $333 billion.)—Rachel Greszler is Senior Policy Analyst in Economics and Entitlements, and James Sherk is Research Fellow in Labor Economics, in the Center for Data Analysis, of the Institute for Economic Freedom and Opportunity, at The Heritage Foundation.
Base Population of Federal Employees. In 2015, there were 2.053 million civilian, non-postal federal employees (according to fedscope.gov). This is the population of employees subject to the proposed federal compensation reforms. Historical data from the OPM shows that the number of executive branch, civilian, full-time-equivalent employees increased by an average of 1.59 percent between 2007 and 2017 (including projections for 2016 and 2017). We assume that this federal employment group will grow at a rate of 1.59 percent per year.
Reduced Federal Employment. Changes in paid leave are estimated to reduce the number of federal employees necessary to perform government functions by 2.22 percent compared to current and projected levels. This reduction comes from assuming that the existing structure of 13/20/26 days leave is reduced to 10/15/20, and that sick days are reduced from 13 to 10. While the changes would reduce paid leave by six to nine days, depending on employees’ tenure, we assume that employees do not currently use all their sick leave and that on average total paid leave taken will fall by five days a year.
Salary Base. We use total employment (2.053 million) times average salaries for 2015 ($80,463) to establish a salary base. Salaries as a whole are projected to grow by 3.22 percent per year. This is the CBO’s projection for average growth in the Employee Cost Index (ECI), which is used to adjust salary levels in the General Schedule.
We also apply step increases to the salaries of specific cohorts of workers, as broken down by tenure below. These step increases are roughly 3.3 percent and occur as employees accumulate one, two, three, five, seven, nine, 12, 15, and 18 years of service.
Reduced Salaries. The proposal to reduce the gap in pay between steps 1 and 10 of each General Schedule grade from 30 percent to 20 percent is modeled as a reduction in the size of step increases from about 3.3 percent to 2.2 percent. These step increases are applied to each cohort based on their years of service. Because the last step increase for employees occurs in their 18th year of service, this change does not affect grandfathered employees, as they have 25 or more years of service.
Employees by Tenure. Because our reform proposals would affect employees differently, based on their tenure, we break employee population and salaries out by group including:
The number of employees in each group is estimated based on the three-year average (2013–2015) of employees by tenure. For example, employees with between five and nine years of federal service comprised 15.1 percent, 16.0 percent, and 16.4 percent of total federal employment in 2013, 2014, and 2015, respectively. Thus, we assume that the share of federal employees with between five and nine years of tenure will remain constant at the 2013–2015 average of 15.8 percent of federal employment. This allows us to age our separate groups above while accounting for employee attrition and retirement.
Average base salaries were established using December 2015 data on federal employment and salaries by tenure from fedscope.gov. These salaries equaled $63,358 for not-vested employees, $56,137 for new hires, $82,580 for vested employees, and $97,623 for grandfathered employees in 2015. We adjust all salaries for growth in the PCE, projected by the CBO to average 3.22 percent, and also for step increases as scheduled based on employees’ years of service. Those step increases are roughly 3.3 percent in the current system and 2.2 percent in our reformed system. Step increases occur as employees reach one, two, three, five, seven, nine, 12, 15, and 18 years of service.
Lower salaries and reduced employment affect other estimates in this report, such as pension contributions, paid leave, and total compensation. To avoid double-counting some of the savings, we provide a cumulative or interactive estimate that incorporates the combined effects of our proposed reforms, beginning with the lower base employment and salary levels that result from our proposed reforms.
Pension savings equal the difference between what the government pays now and what it will pay after the reforms. The OPM estimates the cost of FERS to be 14.0 percent of pay for employees first hired before 2013, and 14.2 percent for employees first hired in 2013 or later. Employees hired before 2013 pay 0.8 percent of pay toward FERS, those hired in 2013 pay 3.1 percent, and those hired in 2014 and beyond pay 4.4 percent. However, 1.3 percent of the contributions for employees hired in 2014 and beyond goes to the unfunded liabilities of the federal pension systems. Thus, the government effectively pays 13.2 percent of pay for employees hired before 2013, and 11.1 percent for those hired in 2013 and later.
In addition, the federal government contributes an automatic 1 percent of pay to employees’ TSP accounts, and up to an additional 4 percent of pay as a matching contribution. We assume that the average federal match is 3 percent of pay, making the total government TSP contribution equal to 4 percent of pay.
The government’s combined FERS and TSP contributions are 17.2 percent for employees hired before 2013, and 15.1 percent for those hired in 2013 and later.
Under both the reformed and new systems, the government will contribute a maximum of 8 percent of pay to employees’ retirement accounts. We estimate the average contribution will be 7 percent based on an automatic 4 percent TSP contribution and a 3 percent matching TSP contribution. This generates savings of 10.2 percent for vested employees hired before 2013, and 8.1 percent for employees hired in 2013 or later.
There is no change in pension costs for employees who have 25 years or more of federal service when the reforms begin in 2017.
Choice Among Vested Employees. Employees who are part of the “reformed system” have three options. We assume that one-third of these employees choose option A, one-third choose option B, and one-third choose option C.
The annual savings to the government of each of these options is equal—reducing total costs by 10.2 percent of payroll, from 18.2 percent to 8.0 percent. Option C, however, has a significant up-front cost because of the lump-sum payout, but results in greater net savings than options A and B because the value of the lump-sum payout is only 75 percent of its net present value, meaning the government saves an additional 25 percent of the net present value of option C employees’ accumulated FERS benefits.
Lump-Sum Benefit Calculations for Option C. The initial outlay of lump-sum transfers would register as a significant increase in government costs. We calculate that initial cost but record the savings based on the total impact—that is, the present value of liabilities for the government. This means, rather than reporting the lump-sum transfer as a cost of 75 percent of the present value of the future benefit, we report it as a 25 percent gain of that benefit because the government will not have to pay out that 25 percent in the future. This accrual-based accounting more accurately reports the true value of the reform as measured by what the FERS accounts would otherwise have paid in the future versus the 75 percent value that is actually paid.
The value of the lump-sum rollover and 25 percent savings to the government is calculated based on the accrued benefit that would be available to an average federal employee. In 2015, the average age of federal employees was 46. We use this to obtain a life expectancy of age 81.1 based on the 2013 life expectancy of a 45-year-old. We assume the average employee in the reformed system has 15 years of service (the mid-point of the group, which includes those with as few as five and as many as 24 years of service). We assume that the employee’s highest three years of salary average $85,000 (based on average salaries by tenure), which generates an accrued future FERS benefit equal to $12,750 per year (1.0 percent * 15 years of service * $85,000 = $12,750). The present value of that benefit equals the number of years that the individual would receive it times the benefit amount (20.1* $12,750), or $256,245.
Thus, the individual could choose to receive a 75 percent lump-sum payout equal to $192,206 in 2017. To compare the level of retirement income that the lump-sum benefit would provide compared to the existing FERS benefit, we assume that the lump-sum benefit is invested and earns a 7 percent nominal average rate of return until the individual begins making withdrawals at age 62. We then assume the account is shifted to more conservative investments and earns the 10-year Treasury rate of return, estimated by the CBO to be 4.1 percent.
Although initial calculations for the FERS benefit do not include inflation or wage adjustments, we assume that both for the value of the lump-sum benefit as well as the sake of comparison, that an individual’s benefit would have otherwise been adjusted upwards for inflation so that a frozen or lump-sum benefit transfer does not result in a real loss of benefits over time. Thus, we assume that the individual’s $12,750 per year benefit would have grown 2.0 percent per year, based on the CBO’s projections for inflation as measured by the PCE. This would result in an annual benefit of $17,413 at retirement in 2033, growing to more than $25,700 by the time of the employee’s death in 2053.
Comparisons between the level of income that the individual’s lump-sum account would provide versus his current FERS benefit can be made by subtracting various levels of annual benefits (beginning at 100 percent of the FERS benefit and moving upwards) from the lump-sum account and seeing how high the benefits would go before the account runs out of money when the individual dies at age 81.1, by seeing how long various benefit levels could be maintained, or by seeing how much money would be left over in the individual’s account to pass on to heirs.
Employees who have been continuously enrolled in any FEHB plan (including of a family member) for at least five years immediately prior to their retirement have access to continued FEHB benefits during retirement, including the federal subsidy. (If the employee has fewer than five consecutive years prior to retirement, he is still eligible if he has at least five total years and elected coverage in all years of service.) Ending access to this benefit only for new hires would postpone most savings until employees retire. However, we include the accrual value of this benefit in our calculations. A 2002 study by the CBO included accrual-based estimates of the value of FEHB retiree benefits, reporting a 2002 value of $3,475. This represents 6.4 percent of the average federal salary of $54,656 in 2002. Thus, we subtract 6.4 percent of the pay of individuals first hired in 2017 or later. Using the same methodology as described above to calculate the distribution of employees affected by pension changes based on tenure, we estimate that 6.3 percent of federal employees will be new hires in 2017. This population of federal employees hired in 2017 and later will grow to 53 percent by 2026.
The current budget is represented on a cash-flow basis. Cash-flow budgeting does not register costs until they are paid out, or revenues until they are paid in. In that way, the current budget does not account for the true cost of particular items, such as pensions and other federal employee benefits, which are promises for future payments. Such accounting fails to recognize the true value of liabilities incurred and often leads to excessive or unfunded liabilities (as in Social Security and many other underfunded public and private pensions).
A 2015 CBO study comparing cash accounting to accrual accounting found that federal pension costs were $1.6 trillion greater over the 2015–2025 period when using accrual-based accounting than under the government’s traditional cash-based accounting. Although many of the reforms we propose do not result in immediate cash-flow savings, we report the accrual value of estimated savings to show the true magnitude of savings from our proposed reforms.
 Alan B. Krueger, “Are Public Sector Workers Paid More than Their Alternative Wage? Evidence from Longitudinal Data and Job Queues,” in Richard B. Freeman and Casey Ichniowski, eds., When Public Sector Workers Unionize (Cambridge, MA: National Bureau of Economic Research, 1988), http://www.nber.org/chapters/c7910.pdf (accessed June 9, 2016).
 For an overview of this literature, see Robert Gregory and Jeff Borland, “Public Sector Labor Markets,” in Orley C. Ashenfelter and David Card, eds., Handbook of Labor Economics (Amsterdam: Elsevier, 1999), Vol. 3A, Chap. 31.
 James Sherk, “Inflated Federal Pay: How Americans Are Overtaxed to Overpay the Civil Service,” Heritage Foundation Center for Data Analysis Report No. 10-05, July 7, 2010, http://www.heritage.org/research/reports/2010/07/inflated-federal-pay-how-americans-are-overtaxed-to-overpay-the-civil-service.
 Andrew G. Biggs and Jason Richwine, “Comparing Federal and Private Sector Compensation,” American Enterprise Institute Working Paper No. 2011-02, June 1, 2011, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1991405 (accessed June 9, 2016).
 Office of Management and Budget, Budget of the U.S. Government, Fiscal Year 2017: Analytical Perspectives, Table 8-4 (“Personnel Compensation and Benefits”), https://www.whitehouse.gov/sites/default/files/omb/budget/fy2017/assets/spec.pdf (accessed June 9, 2016).
 OMB forecasts a federal budget deficit of $503 billion in FY 2017. Ibid., Table 4-2. “Federal Government Financing and Debt.”
 George Borjas, Labor Economics, 6th edition (Columbus, Ohio: McGraw–Hill, 2013), pp. 89–94.
 For example, the unionized Detroit automakers paid above-market wages but steadily lost auto share to non-union competitors charging lower prices, such as Toyota and Honda.
 Robert Gibbons, Lawrence F. Katz, Thomas Lemieux, and Daniel Parent, “Comparative Advantage, Learning, and Sectoral Wage Determination,” Journal of Labor Economics, Vol. 23, No. 4 (October 2005), pp. 682–724.
 Kay Coles James, “A Fresh Start for Federal Pay: The Case for Modernization,” Office of Personnel Management White Paper, April 2002, https://archive.opm.gov/strategiccomp/whtpaper.pdf (accessed June 10, 2016).
 Melissa Famulari, “What’s in a Name? Title Inflation in the Federal Government,” University of Texas at Austin Working Paper, 2002.
 62.7 percent of federal GS employees (or an equivalent grading system) are in grades 11 or higher. See Table 2.
 This pay adjustment is not technically a cost-of-living adjustment, although it is frequently referred to and described as such by both the media and federal employees. This pay adjustment is tied to overall pay rates as measured by the Employment Cost Index, not to the price level as measured by inflation indices.
 Employees can reach step 10 in fewer than 18 years. Agencies have the authority to give employees performance-based step increases in addition to the seniority-based increases.
 Ralph Smith, “Those ‘Automatic’ Within-Grade Increases and a Performance Based Pay System,” FedSmith, June 10, 2011, http://www.fedsmith.com/2011/06/10/those-automatic-withingrade-increases-performance-based/ (accessed June 14, 2016).
 For most GS grades, the pay gradient is exactly 30 percent. For steps 1 and 2 the pay gradient is 25 percent. In some high-cost areas—such as Washington, DC—the locality pay adjustment raises GS-15 employees in step 9 and 10 above level IV of the Executive Schedule. Those employees are limited to the pay of level IV and the GS-15 pay gradient is less than 30 percent.
 The step increases would be slightly above 2 percentage points in order to create a 20 percent pay gradient. Moving from step 1 to step 10 involves nine step increases—there is no step 0.
 This reform would be phased in so that, going forward, each new step increase would be approximately 2 percentage points. A step 8 employee would have smaller step increases in step 9 and step 10, but would not have pay frozen or reduced to create a 20 percent gradient.
 This would mean the Merit Systems Protection Board would have no jurisdiction over the decision not to grant a step increase or performance bonus, and unions could not file grievances on account of an agency not granting a step increase or performance bonus.
 Data from the Bureau of Labor Statistics shows an average 5 percent ceiling on employer matches: “Automatic Enrollment, Employer Match Rates, and Employee Compensation in 401(k) Plans,” Monthly Labor Review (May 2015), http://www.bls.gov/opub/mlr/2015/article/automatic-enrollment-employer-match-rates-and-employee-compensation-in-401k-plans.htm (accessed April 14, 2016). Data from the 401kHelpCenter.com shows a 2.7 percent average employer contribution to employee retirement plans: 401kHelpCenter.com, “Benchmark Your 401(k) Plan – 2015,” http://www.401khelpcenter.com/benchmarking.html#.Vw_YjE3lvIV (accessed April 14, 2016).
 The 18.2 percent contribution exists for employees first hired before 2013. They receive a 13.2 percent FERS contribution and up to a 5 percent TSP contribution. Employees hired in 2014 and beyond receive an 11.1 percent FERS contribution and up to a 5 percent TSP contribution.
 OPM estimates total FERS costs to equal 14 percent of pay for employees hired before 2013, and 14.2 percent of pay for employees hired in 2013 and after. Katelin P. Isaacs, “Federal Employees Retirement System: Budget and Trust Fund Issues,” Congressional Research Service, August 24, 2015, https://www.fas.org/sgp/crs/misc/RL30023.pdf (accessed April 21, 2016).
 Employees hired in 2013 contribute 3.1 percent of their pay towards FERS, and the federal government contributes 11.1 percent.
 The 18.2 percent of pay contribution from the government assumes that the OMB’s estimated cost of FERS at 14.0 percent to 14.2 percent of pay is accurate. If life expectancies are higher than projected, or if FERS contributions earn lower returns than expected (meaning that Treasury rates are lower than anticipated), the actual cost would be higher than the 14 percent–14.2 percent range.
 While only 8 percent of private employers offered DB plans in 2015, 18 percent of workers had access to DB plans because larger employers are more likely to offer them. Bureau of Labor Statistics, Database, http://data.bls.gov/cgi-bin/dsrv (accessed April 21, 2016).
 Employees will receive approximately 75 percent of the contributions made on their behalf in the form of a lump-sum payout. The justification for receiving less than 100 percent is both that the employees are not yet vested and that a portion of contributions made on their behalf are used to pay unfunded CSRS liabilities.
 Employees who choose this option will receive a 3 percent contribution from the government and will pay the remaining cost of their annuity. The OPM estimates the current cost of the FERS benefit to equal 14 percent of pay. Using a fixed-income annuity calculator from Charles Schwab, we estimate that increasing the minimum retirement age by 10 years on newly accrued benefits, as well as basing benefits on the average of all years of service as opposed to the high-3, will reduce the cost of the FERS benefit by about 30 percent, to roughly 10 percent of pay. The government would contribute 3 percent and employees would pay the remaining 7 percent. Charles Schwab, “Fixed Income Annuity Estimator,” http://www.schwab.com/public/schwab/investing/accounts_products/investment/annuities/income_annuity/fixed_income_annuity_calculator (accessed June 22, 2016).
 In general, federal employees with 20 or more years of service who retire at age 62 or later receive an annual pension benefit equal to 1.1 percent times their number of years of service times the average of their highest three years of earnings. For example, a federal employee with 35 years of service and an average income of $85,000 over their high-3 would receive a FERS benefit of $32,725 per year. For employees with fewer than 20 years, or who retire earlier than age 60, the 1.1 percent benefit per year is reduced to 1.0 percent. Office of Personnel Management, “FERS Information: Computation,” https://www.opm.gov/retirement-services/fers-information/computation (accessed June 16, 2017).
 The current reduction to FERS benefits for individuals who retire before their full retirement age is 5 percent per year. Office of Personnel Management, “FERS Information: Types of Retirement,” https://www.opm.gov/retirement-services/fers-information/types-of-retirement/#url=Early-Retirement (accessed June 22, 2016).
 Lump-sum payouts would be calculated according to workers’ already accrued FERS benefits. That is, the present discounted value of the estimated amount they would receive in FERS benefits over their lifetime, based on the level of benefit they are eligible to receive as of the date of enactment of the change.
 Authors’ calculations based on the following assumptions: age 46, life expectancy of age 81.1, retirement in year 2033 and death in year 2053, 15 years of accrued FERS benefits, a high-3 salary of $85,000, resulting in an accrued pension of $12,750 per year in current dollars (1.0 percent * 15 years * $85k = $12,750 per year); 75 percent of the present value of 20.1 years of $12,750 annual benefit payments equals $256,275. This assumes the alternative FERS benefit would have grown with prices until and during retirement (equal to 2.0 percent as projected by the CBO for the Personal Consumption Expenditures (PCE). The lump-sum transfer is assumed to grow at a nominal rate of 7.0 percent invested in stocks until 2033 (a total of 20.1 years) when the individual retires at age 62 and begins drawing annual payments. (The individual could retire later, with his account continuing to grow, but we assume 62, since it is the individual’s age of retirement eligibility for his FERS benefit.) From that point on, the balance is assumed to be invested more conservatively in 10-year U.S. Treasuries earning a 4.1 percent annual return. If the individual elects to withdraw the exact amount that his existing FERS benefit would provide, he would have over $538,000 remaining in his retirement account upon death at age 81.1 in 2053. That amount would be sufficient to continue providing the same real level of benefit until 2080, when the individual would be 109 years old. The individual could alternatively withdraw up to 175 percent of what his accrued FERS benefit would provide without running out of funds in his account. The size of account balances and possible benefit levels would vary depending on workers’ earnings, years of service, and years before retirement. In general, workers who are further from retirement would receive larger financial benefits from choosing a lump-sum payout than those closer to retirement because they would benefit from more years of higher market returns on their accounts.
 Life expectancy of age 81.1 is assumed based on the most recent data from the Centers for Disease Control, which shows life expectancy at age 45 to be another 36.1 years, or 81.1 years of age. See Jiaquan Xu et al., “Deaths: Final Data for 2013,” National Vital Statistics Reports, Vol. 64, No. 2 (February 16, 2016), http://www.cdc.gov/nchs/data/nvsr/nvsr64/nvsr64_02.pdf (accessed April 25, 2016). The average federal employee was 46 years of age in 2015.
 Office of Personnel Management, “Continuing FEHB Coverage Into Retirement,” https://www.opm.gov/faqs/topic/insure/?cid=880bfba8-8f8b-4e64-9a72-fae98408fd0e (accessed June 27, 2016).
 Authors’ calculations establish the 6.34 percent of pay cost by comparing the average salary of $54,656 in 2002 to the estimated $3,475 accrual cost of FEHB benefits as reported in: Congressional Budget Office, “The President’s Proposal to Accrue Retirement Costs For Federal Employees,” June 2002, https://www.cbo.gov/sites/default/files/107th-congress-2001-2002/reports/accrual.pdf (accessed May 16, 2016).
 This estimate equals 6.4 percent of the salaries of all employees first hired in 2017 or later. See appendix for additional information on the estimated distribution of employees by tenure, salaries by tenure, and both employee and salary growth over time.
 The specified paid leave allowances are provided for employees once they have three years of experience. We use the example of an employee with five years of service because it allows a comparison to data on private-sector employees.
 Bureau of Labor Statistics, National Compensation Survey. Figures are mean vacation and sick leave days provided to private-sector workers in 2015 in companies with 100 or more employees, for employees who do not have consolidated vacation and sick leave plans.
 Rachel Greszler, “Private Disability Insurance Option Could Help Save SSDI and Improve Individual Well-Being,” Heritage Foundation Backgrounder No. 3037, July 20, 2015, http://www.heritage.org/research/reports/2015/07/private-disability-insurance-option-could-help-save-ssdi-and-improve-individual-well-being.
 Ibid., for employees who have a consolidated sick leave and vacation plan.
 The average private disability insurance premium was $245 per year in 2011. This is less than one-third the comparable cost of Social Security’s Disability Insurance program. See Greszler, “Private Disability Insurance Option Could Help Save SSDI and Improve Individual Well-being.”
 This calculation assumes a 2.2 percent reduction in total employment. Federal employment is expected to grow by 1.59 percent per year based on the OMB’s historical and projected growth in full-time-equivalent employees from 2008 to 2017. The employment cost index, which is used for the GS salary scale, is estimated by the CBO to grow at an average rate of 3.22 percent, and benefits are estimated as a percent of salary based on the average 2012–2017 level reported by the OMB.
 Congressional Budget Office, “Reducing the Budget Deficit: Spending and Revenue Options,” March 2011, https://www.cbo.gov/sites/default/files/112th-congress-2011-2012/reports/03-10-reducingthedeficit.pdf (accessed May 23, 2016).
 James Sherk, “IRS Abuses: Ensuring that Targeting Never Happens Again,” testimony before the Oversight and Government Reform Committee, U.S. House of Representatives, August 6, 2014.
 Heritage Foundation calculations using data from the OPM’s “FedScope.” The rate divides FY 2015 terminations for discipline/performance for employees with three or more years of service by December 2015 total federal employment for that same group of employees. U.S. Office of Personnel Management, “FedScope–Federal Human Resources Data,” http://www.fedscope.opm.gov/ (accessed June 22, 2016).
 U.S. Merit Systems Protection Board, “Addressing Poor Performers and the Law,” report to the President and the Congress, September 2009, p. 33, http://www.mspb.gov/netsearch/viewdocs.aspx?docnumber=445841&version=446988& (accessed June 21, 2016).
 Ibid., pp. 33–34.
 The proposed reforms would not change the current 1.3 percent of payroll that employees hired in 2014 and later contribute to the unfunded CSRS retirement liability. Thus, those employees would continue making a 1.3 percentage point contribution despite not being eligible for FERS benefits. Employees hired in 2013 and earlier do not make that 1.3 percent contribution and therefore would not make any FERS contributions if they elect options B or C, which would shift their future accruals exclusively to the TSP system. Employees hired in 2013 do pay a higher FERS contribution equal to 3.3 percent, compared to 0.8 percent for those hired prior to 2013.
 Xu et al., “Deaths: Final Data for 2013.”
 Congressional Budget Office. “The President’s Proposal to Accrue Retirement Costs for Federal Employees,” June 2002.
 Congressional Budget Office,” CBO’s Projections of Federal Receipts and Expenditures in the National Income and Product Accounts,” September 2015, https://www.cbo.gov/sites/default/files/114th-congress-2015-2016/reports/50819-NIPA.pdf (accessed May 16, 2016).