top of page

Government Agency Misconduct Risk: Accountability Across Public Systems

Feb 19
5 min read

Public agencies carry a double burden on misconduct. They are employers whose staff deserve a safe workplace, and they are public institutions whose conduct is held to a standard of trust, transparency and lawful process. When sexual harassment, abuse of authority or retaliation occurs and is mishandled, the damage spreads past the individuals involved to public confidence in the agency itself.

This article, written with Washington, D.C. as a working context for federal, state and local public systems, looks at misconduct risk in government agencies. It uses the best available federal evidence to show where risk concentrates, why accountability tends to break down across layers of oversight, and what leaders can do about it. It is educational and is not legal advice. Rules differ for federal agencies, state agencies and local governments, and agencies should confirm their obligations with counsel.

What federal survey data shows

The U.S. Merit Systems Protection Board (MSPB) has surveyed federal employees about sexual harassment for decades. Its March 2018 research brief, Update on Sexual Harassment in the Federal Workplace, used the 2016 Merit Principles Survey. Several findings matter for any public employer:

  • About 14.3 percent of federal employees reported experiencing at least one of 12 listed sexual harassment behaviors in the previous two years, including 20.9 percent of women and 8.7 percent of men. The survey measures employee perception and does not establish legal violations.

  • The share fell from the 1994 survey. On the 8 behaviors that both surveys covered, 17.7 percent of women reported experiencing one in 2016 compared with 44.3 percent in 1994. MSPB concluded rates have decreased but remain a problem.

  • Hostile-environment behaviors such as sexual teasing, jokes and comments, and invasion of personal space, were far more common than sexual coercion or assault.

  • Rates differed widely by agency. Among women, the share ranged from about 9 percent to about 27 percent across the agencies MSPB reported on. That spread shows risk is shaped by local conditions, not only by the sector.

Who committed the harassment also matters. In MSPB's Table 5, respondents most often named a coworker (45 percent), followed by other agency employees (27 percent) and customers or members of the public (15 percent), with higher-level supervisors at 12 percent and immediate supervisors at 11 percent. Respondents could name more than one role, so totals exceed 100 percent. Harassment by someone in authority can lead employees to believe resisting or complaining would be futile or risky, as MSPB notes. A control program that looks only at supervisors will miss most of the pattern, and one that looks only at coworkers will miss the people with the most power.

Why accountability breaks down across layers

A public agency rarely answers to one authority. Employees may report to a supervisor, an HR office, an equal employment opportunity office, an inspector general, a union and an outside oversight body. Each layer has its own rules and timelines. The result can be a complaint that every office has touched and no office owns.

A 2025 Government Accountability Office report shows the problem at the policy level. In Sexual Harassment: Actions Needed to Ensure Consistent Agency Policies for Research Institutions, GAO found that none of the 17 research agencies it reviewed had policies fully consistent with federal guidelines issued in January 2025, that five of the 17 had no policies at all when the guidelines were published, and that the office responsible for monitoring implementation had not yet hired staff to do so as of July 2025. GAO recommended that the Office of Science and Technology Policy monitor agency policy development and fully address sharing of reports and investigative information. This concerns federally funded research institutions, not agencies generally, but it illustrates a pattern any public system can recognize: guidance exists, ownership of follow-through does not.

The EEOC's chart of risk factors names several conditions common in public systems: significant power disparities, decentralized workplaces where managers are far from headquarters, and workplaces with high value employees, where leaders may be reluctant to jeopardize an employee's value to the organization. It recommends applying workplace rules uniformly regardless of rank or value to the employer.

Where conventional approaches fall short

  • Annual mandatory training is completed on schedule while complaint handling takes months and employees never learn outcomes.

  • Each office measures its own caseload, and no one measures the whole journey from report to corrective action.

  • Corrective actions are recorded but not tracked to completion, so the same problem returns in the same unit.

  • Political, budget or reputational sensitivity shapes how senior or well-connected employees are treated.

  • Reports from contractors, interns, detailees and members of the public have no clear owner.

Six controls that make accountability real

  1. Assign one owner for the whole journey. Name a role responsible for tracking each report from intake to documented corrective action, even when several offices act along the way.

  2. Build a single case view. Give the owner a shared record that shows each office's step, dates and status, so a stalled case is visible.

  3. Standardize across units. Use the same intake categories, timelines and corrective-action tracking in every component, and compare units rather than averaging them.

  4. Separate the decision from the relationship. Where a report involves a senior official, route fact-finding and decisions outside that official's chain, and document the reason.

  5. Protect against retaliation by design. Check schedule, assignment, performance and access changes for the reporter and witnesses at 30 and 90 days, and make the check someone's named duty.

  6. Close the loop upward. Give executives and oversight bodies regular, de-identified reports on volume, timeliness, repeat patterns and open corrective actions, with a named official accountable for each.

Early-warning indicators that stay organizational

  • Cases that move between offices without a documented decision.

  • Rising age of open cases, or a backlog concentrated in one component.

  • Repeat complaints about the same unit, supervisor or location across years.

  • Corrective actions that are closed without evidence of completion.

  • Employees reporting to outside bodies before internal channels.

  • Large differences in survey results or complaint rates between similar units.

None of these proves misconduct. Each is a prompt to examine conditions, and none should rely on protected characteristics.

How to measure whether it works

  • Median days from report to first contact, to decision and to documented corrective action, by component.

  • Share of corrective actions completed on time with evidence.

  • Repeat-complaint rate by unit over a rolling three years.

  • Retaliation check results at 30 and 90 days.

  • Employee survey items on confidence that reports lead to action, compared across units and over time, in the style of MSPB's federal surveys.

  • Share of cases with a named accountable official and a documented executive review.

Expect reporting to rise at first. A rising count with shorter case times is a healthy sign.

Where RMA fits

The Risk Mitigation Authority approaches misconduct as a human behavior problem, an organizational culture issue and a security issue at once. For a public agency, enterprise risk integration brings misconduct into the same risk register, indicator set and executive review as other enterprise risks, with a named owner for each control. A Risk Exposure Audit is a practical first step. Learn more about RMA. No program can promise to eliminate misconduct or liability. A well-built one makes harm harder to commit, easier to surface and faster to correct.

Related reading

Continue with these related RMA articles:

References

bottom of page