Classification rules everything around me.

Fixing the civil service must start with replacing its fundamental building blocks: jobs, pay grades, and the values we hold dearest

Executive summary

There’s an old joke in Washington that the first thing anyone asks you is, “What do you do?” as they’re trying to size you up. That tendency works in reverse too, however, as it’s also the first question most people ask when trying to understand an employer or an organization –– what they say they do can tell you a lot about what they value.

Underlying every federal program are federal workers; underlying every federal worker is an obscure taxonomy that defines their job. Every year, when it hires a couple hundred thousand people, the U.S. federal government sends millions of signals about itself to the job market, its own workforce, and the broader country. A cursory glance through its job postings reveals some of the most obvious ones: that federal employment is narrow, bureaucratic, procedural, and intelligible only to insiders. It prioritizes and selects for task specialists rather than outcome specialists. These signals shape who applies, who stays, and ultimately what kind of government Americans get. It is not a coincidence that Americans impute those same signals onto the entire enterprise: slow, stuck in the past, unable to innovate, etc. It’s also why many contemporary efforts to reform individual parts of the civil service — experiments with hiring process tweaks, exotic accountability schemes, for example — rarely produce much in the way of results. The “bones” of the house dictate its shape more than the window dressings.

It doesn’t have to be that way. The federal government’s job classification system — the framework that defines what federal jobs are, what they’re called, how they’re described, and how they’re compensated — was never meant to be the terminal state for the bureaucracy, and yet it has remained fundamentally unchanged since 1949. Designed for a workforce of clerks and typists in an era of assembly lines and typing pools, the federal government’s General Schedule (GS) personnel system now struggles to accommodate data scientists, cybersecurity specialists, and the knowledge workers who compose the modern civil service. The result is a system that obscures jobs behind bureaucratic titles, pays software engineers the same as HR specialists because an internal rubric deemed their roles equivalently complex, and forces agencies to go hat in hand to Congress or the president for special carve-outs just to stay competitive. Small wonder that Americans and federal employees alike feel their government is broken: It is operating under civil service rules that have long passed the end of their useful life and keep civil servants trapped in a system that incentivizes all the wrong behavior.

We have imagined the system that might come next, reflecting modern labor market dynamics just as the GS did when it was created almost 80 years ago during the Truman administration. In this model, the federal government would redefine some of its core principles, such as “equal pay for equal work,” to better fit its value proposition to modern work life, pruning back the ways in which its application has grown to obscure the original intent. It would break up the universal pay table in the GS into several occupational pay categories; move from 15 grades to six career steps; free agencies from the requirement of bureaucratic job titling; reduce low-value complexity in the Federal Locality Pay system; and let agencies define their own merit-based compensation schemes that better fit the stunning diversity of missions they are asked to undertake. Congress should enact this system on bipartisan consensus and fund agencies adequately to stick the landing.

Summary of findings

The system today falls short in many ways:

1. The system is task specialized, not outcome specialized. The federal government’s system of designing and rewarding jobs is based on which tasks employees complete during the day — never mind that most modern jobs involve a wide variety of tasks centered on broad outcomes. Good marketing specialists aren’t good because they’re just experts at a specific bundle of tasks like proofreading, copywriting, and graphic design; they’re good because they generate positive brand attention using whatever means are most effective.
2. The system is blind to labor markets. A GS-12 Human Resources Specialist and a GS-12 Software Engineer are treated as equivalent units of labor value, but the market prices them very differently. The government chronically underpays for high-demand technical skills while arguably overpaying for administrative roles that have been automated or outsourced elsewhere in the economy.
3. The system is easily gamed. Because classification depends on written descriptions of duties rather than outcomes, clever position descriptions can secure higher grades for work that would be classified lower elsewhere. This produces inequity — the opposite of what the system intends — and fuels persistent grade inflation.
4. Standards cannot keep pace with change. OPM’s occupational standards take years to update. Many date from the 1960s, and even then were out of date. The government is classifying 21st-century work against mid-20th-century benchmarks, forcing managers to shoehorn data scientists into categories designed for file clerks.
5. Managers have no incentive tools. A supervisor who wants to retain a high-performing employee or respond to a competing offer has almost no ability to adjust compensation. Pay is tied to the position, not the person. This strips managers of the most basic management tool and turns them into passive observers of their own attrition rates.
6. Growth in place is impossible. The grade has become the sole proxy for professional worth. Employees who want to be paid more must seek higher-graded positions, even if that means leaving work they love and are good at. The system creates a culture of grade chasing rather than mission focus.
7. The system sends the wrong signals. Bureaucratic job titles, arcane classification jargon, rigid position descriptions, and 18-year step progressions all communicate that the federal government is a world apart from modern employment. These signals discourage exactly the talent the government most needs to attract.

And yet, it has some distinct advantages:

8. Demographic pay equity. The lockstep nature of the GS means that once employees are hired into a given role, their progression is largely automatic. This makes it structurally difficult for the system to disadvantage women or minorities in ways that more discretionary systems can. A role is a GS-12 regardless of who sits in the seat.
9. Predictability for employees. The system offers profound stability. An employee entering as a GS-07 knows exactly what is required to become a GS-09 and how long it will take to reach Step 10. Raises happen at known intervals rather than at managerial whim. This predictability fosters a sense of security that was intended to insulate public service from market volatility.
10. Budgetary and managerial predictability. The government can plan its personnel costs years in advance because salary escalation is formulaic. A GS-13 represents a known cost trajectory. This makes workforce planning possible at a scale that would be unmanageable if every salary were individually negotiated.
11. Whole-of-government coherence. The common language of grades and series allows employees to move between agencies, allows Congress to legislate across the workforce, and allows managers to understand roughly what a given grade level means regardless of context. This coherence has real value for an enterprise as large and diverse as the federal government.

Congress has been working on this for 250 years, with many themes recurring:

12. What Congress puts in statute stays in statute. Detailed grade-level definitions, specific salary figures, enumerated occupational categories — once these are written into law, they prove extraordinarily difficult to update. The lag between economic change and statutory revision creates pressure for exceptions and workarounds that eventually fragment the system. Future-proofing requires removing as much operational detail from statute as possible, finding alternative mechanisms for congressional control that don’t require amending the U.S. Code every time conditions change.
13. Simplicity is durability. The two longest-lived federal pay systems — the four-grade structure of 1853 and the General Schedule of 1949 — were both relatively simple compared with their predecessors. The baroque Classification Act of 1923, with its dozens of services and hundreds of classes, collapsed within a generation. Comprehensive precision is the enemy of longevity. Systems that try to account for every contingency tend to buckle under their own weight.
14. Agency-specific exceptions are a trap. Every major reform has been undermined by carve-outs: the Foreign Service here, the VA there, new agencies exempted because they needed to stand up quickly. Each exception creates a constituency for its own perpetuation and spurs other agencies to seek similar treatment. If a system cannot accommodate legitimate variation through its standard mechanisms, the answer is to make the system more flexible, not to punch holes in it.
15. Implementation outlasts political moments. The George W. Bush administration’s attempts to reform classification at the Departments of Homeland Security (DHS) and Defense (DoD) failed in part because they were enacted on party-line votes and implemented through controversial regulations that lacked durable support. When Congress changed hands, the projects were strangled before they could mature. Any new system will take years to implement fully, years that will span multiple congresses and at least one change in administration. Reforms that cannot survive shifting political winds will not survive at all.
16. Equal pay for equal work is not negotiable, but it must be modernized. At the heart of every reform cycle has been a persistent demand that similarly situated employees be treated similarly. When fragmentation produces arbitrary differences in compensation, the political system reliably responds with pressure for standardization. The Bush-era reforms stumbled in part because their pay-for-performance components were perceived as threatening this principle. Any new system must maintain legitimate equity while redefining what “equal work” means in a world where labor markets value different skills differently.

Other large employers have moved on already, learning that:

17. Fewer levels are better than more. When organizations flattened their hierarchies in the later decades of the 20th century, they discovered that maintaining 15 or 20 finely graded distinctions created administrative overhead without corresponding organizational benefit. Most meaningful career transitions can be captured in five to eight levels. Distinctions finer than that tend to become arbitrary and invite gaming.
18. Job families matter more than individual occupations. Labor markets don’t operate at the level of “Management and Program Analyst” or “Computer Assisted Ordering Technician.” They operate at the level of finance, engineering, human resources, information technology — broad functional groupings within which people move, recruiters specialize, and compensation surveys are organized. Classification systems that mirror labor market structure are easier to benchmark and easier to explain.
19. Market pricing beats internal point counting. As internet adoption progressed and compensation data became widely available, employers discovered they could benchmark directly against external markets rather than rely on elaborate internal rubrics as a proxy for fairness. The relevant question shifted from “How does this job score on our complexity factors?” to “What does the market pay for this work?” Internal equity still matters, but it is no longer the sole or even primary driver of pay structure in most places.
20. Flexibility within structure is necessary. Modern job architectures give managers discretion to place employees within bands and to differentiate pay based on performance, market conditions, and retention risk, without requiring reclassification of the position itself. This flexibility demands other safeguards (calibration, transparency, analytics, etc.) but it allows organizations to respond to labor market signals without constant pressure for grade inflation.
21. Skills and competencies matter more than task lists. Position descriptions that enumerate every discrete duty a worker might perform are both overspecified (listing tasks that may never recur) and underspecified (failing to capture the judgment and adaptability that define knowledge work). Modern systems focus on what employees need to be able to do, not an exhaustive catalog of what they might be asked to do on any given day.‌

Summary of recommendations

Six design principles should guide an overhaul of the system that brings government into 2026:

22. Outcomes over tasks. The system should define jobs by the results they exist to produce and the capabilities required to produce them, not exhaustive catalogs of discrete duties. Position descriptions that enumerate every task are simultaneously overspecified and underspecified for knowledge work.
23. Simplicity as a design constraint. The system must be simple enough to administer and simple enough to understand. History demonstrates that baroque complexity — whether the 1923 Classification Act’s several discrete services or today’s 400-plus occupational series — eventually collapses under its own weight.
24. Permeability to outside talent. The system must treat career mobility as normal rather than aberrant, making it easy to bring experienced professionals in at appropriate levels and easy for federal employees to leave and return without penalty.
25. Labor market comparability by occupation. The system must benchmark compensation against external markets for each occupation, not rely solely on internal complexity ratings. A cybersecurity expert and a budget analyst should not earn identical salaries simply because an internal scoring system rated their positions as equivalently complex.
26. Mission flexibility through standard mechanisms. The system must accommodate legitimate agency variation through its normal operations, not through carve-outs that fragment the workforce. If a system cannot handle variation, the answer is to make the system more flexible, not carve it up.
27. Redefined pay equity at the occupational level. Equal pay for equal work should apply within occupational families — in which employees actually compete for jobs and possess interchangeable skills — not across fundamentally different labor markets.

This reimagined system would have five core components:

28. Simplified job architecture with approximately 30 occupational families replacing 400 occupations. The many granularly defined occupational series should collapse into broader families organized around functional similarity and labor market dynamics: Information Technology, Finance and Accounting, Engineering, Legal, Medical and Health, and so forth. OPM would provide high-level guidance while agencies gain authority to define specific positions.
29. Six career levels replacing today’s 15 grades. The General Schedule’s elaborate grade structure should give way to meaningful career stages: Entry Level/Developmental, Journey Level, Specialist, Manager, Senior Individual Contributor, and Senior Manager. Each level would have a pay band rather than discrete steps.
30. Occupational pay tables benchmarked to labor markets. Each occupational family should have its own pay table, with ranges for each career level, independently calibrated to its respective labor market and updated annually. The IT table might range differently than the Administrative Support table because those labor markets operate independently.
31. Simplified locality pay with four categories. The 58 locality areas should collapse into four categories — High Cost of Labor, Medium Cost of Labor, Low Cost of Labor, and Remote — reducing administrative complexity while preserving geographic responsiveness.
32. Mixed longevity and merit progression within bands. Within-band movement should combine a guaranteed longevity component (perhaps 1 percent to 3 percent annually for satisfactory performance) with a discretionary merit component (up to 6 percent to 8 percent annually) based on agency-determined performance criteria, with strict transparency requirements.

Implementing this model requires the branches to work together & compromise to carry the day:

33. Specific statutory changes are required. Congress must remove grade-level definitions from statute and replace them with career-level bands (5 USC Chapter 51); transition from the General Schedule to occupational schedules (5 USC Chapter 53), simplify annual pay-setting procedures; update within-grade increase formulas to split longevity and merit; and clean up agency-specific exceptions.
34. Congress: you must lead with crafting detailed, bipartisan legislation. The system rests on statutes that only Congress can change. Reform enacted on party-line votes will be vulnerable to rollback when political control shifts; implementation spanning five to seven years requires durable consensus.
35. Appropriators: Agencies need investment in workforce planning capacity. Most agencies lack in-house expertise for career pathing, competency modeling, and performance-based pay distribution. Reform requires building this capacity, not just granting new authorities.
36. OPM: You need a dedicated compensation function. If OPM is empowered to recommend occupational pay adjustments annually, it requires labor economists and compensation specialists— perhaps 30 to 40 staff — capable of producing high-quality market analysis for each occupational family to navigate to the right place in each labor market, consistent with the administration’s policy.
37. Leaders everywhere: Employee engagement must begin early. The Bush administration’s failure to adequately consult unions contributed directly to the demise of the DoD’s National Security Personnel System (NSPS) and DHS’s MaxHR reforms. Genuine engagement with employee organizations from the earliest stages is essential to building a durable coalition.

Summary of benefits

‌If the federal government adopts this model, we can expect the first substantial change to the way the civil service works, looks, and feels in decades:

38. Agencies could compete for talent in critical fields without lobbying for special authorities or relying on workarounds. The system would accommodate the reality that cybersecurity experts and HR specialists operate in different labor markets while guarding against inequities among agencies created by special exception cases.
39. The principle of equal pay for equal work would be preserved but defined at a level that makes sense. Equity would mean that federal employees in the same occupation at the same career level receive comparable pay, regardless of agency. It would not mean pretending that all work of similar “complexity” deserves identical compensation but rather ensuring that equity is defined expansively.
40. The system would be simpler to explain and administer. Instead of one general pay plan with hundreds of exceptions, there would be a coherent structure where occupational differentiation is built in from the start. The guardrails would provide enough guidance to ensure cross-government equity while allowing agencies to fine-tune rules to their unique situations and missions.
41. Growth in place would become possible. A procurement specialist who excels technically but has no interest in management could continue advancing within their pay band rather than chase a supervisory position to get a raise. The practice of overgrading to create temporary competitive advantage between agencies would no longer be required just to compete for talent.
42. A new job architecture unlocks further reform in other parts of the civil service system. A better foundation for job design makes other types of reform possible: A modern merit progression within grades makes it possible for new models for performance management to work; a move to occupational job families makes new ways of providing training or structuring accountability by function rather than agency viable; defining jobs in ways that applicants understand helps hiring reform to focus more on procedure than job design, etc.

Change is possible, but it will not happen overnight, nor spontaneously. If we want a government that works better, we need a civil service that works better; if we want a civil service the works better, we need to change the jobs we ask people to do in the first place. We will keep failing to reform other things — hiring, bargaining, performance management, and all the rest — if we don’t start by getting the foundation right.