Good Enough For Government Work?


cliff winston is a senior fellow at the Brookings Institution. This article is adapted from his book, Gaining Ground: Markets Helping Government (Brookings Institution Press, 2021).

Illustrations by alex williamson

Published October 11, 2021


Do we really know why government policies seem so inefficient, and why they are rarely reformed? For centuries, economists have been cataloguing sources of the systemic failure of markets to achieve efficient outcomes and then offering government policy fixes. By contrast, little has been written with much analytical depth about the sources of persistent government inefficiencies. The well-known political economy models of government stress the importance of organized special interests, including people in government, who engage in rent-seeking and distort policy formation and implementation or (the flip side) resist efficient policy changes. But as we shall see, the evidence of what’s really going on is weak even if one includes both the conventional “demanders” of policy, who try to purchase influence through lobbying and campaign contributions, and the “suppliers” of policy — government employees and elected officials who are influenced by their personal characteristics and the environment in which they work. The important implication is that it is exceedingly difficult to infer how government can overcome its failures and improve its policies. In contrast, markets have the incentive and ability to correct their own failures over time.

But before going further, let’s briefly consider possible explanations for persistent government failure that simply don’t hold weight, at least not in the context of the United States. First, there is widespread agreement that corruption is just not sufficiently pervasive in the U.S. government to serve as a core explanation for government failure. Nor is there evidence that economic policies tend to be systematically more efficient when one political party or the other is in power. The U.S. economy has performed better when the president is a Democrat rather than a Republican, but this edge mainly stems from circumstances beyond a president’s control — for example, oil price shocks.


Back to the more plausible candidates. Lobbying is certainly high on the list. And there is plenty of circumstantial evidence that lobbyists have influenced the content and enforcement of laws and regulations. For example, bank lobbyists actually wrote the provisions that weakened key regulations in the post-crisis Dodd-Frank Wall Street Reform and Consumer Protection Act. Moreover, many special interests certainly think lobbying works: the health care industry alone spent roughly a half-billion dollars in 2016 alone. But it turns out the benefits that industry realizes from its lobbying expenditures are hard to identify.

In general, it seems, firms have not improved their financial performance by lobbying. In fact, their performance may on balance suffer when corporate leaders use company resources to indulge their own ideological preferences in lobbying. In any case, most firms do not lobby policymakers even when it may appear to be in their interest to do so — and, importantly, most members of Congress do not want to hear from lobbyists anyway.

while lobbying is undoubtedly associated with some changes in public policy, it is difficult to pin down those changes, and therefore difficult to conclude that lobbying generally accounts for a large share of the failure of policy to minimize waste and maximize productivity.

Whatever benefits lobbyists achieve slowed down under the Trump administration. The Economist argued that lobbying for corporate America became even more problematic because (a) Trump was viewed as an outsider; (b) his administration was rarely able to translate official policy into tangible action; (c) many businesses worried about being sullied by association with Trump’s entourage or with his views on race, immigration and foreign policy; and (d) power in the administration shifted very quickly, making it hard to target lobbying. In addition, because Trump retaliated against those who criticized him, it was difficult to distinguish between effective lobbying and times when lobbyists’ goals were simply consistent with Trump’s vendettas. It may prove easier

It may prove easier for corporations to lobby the Biden administration. But while lobbying is undoubtedly associated with some changes in public policy, it is difficult to pin down those changes, and therefore difficult to conclude that lobbying generally accounts for a large share of the failure of policy to minimize waste and maximize productivity. Further complicating matters, special interest lobbying may occasionally promote rather than retard efficiency. Cases in point: lobbying that wards off trade protectionism and restrictions on access to cheaper, cost effective foreign labor.

Campaign Contributions

In light of the growth in campaign spending and widespread condemnation of a legal system that does little to curtail contributions or even force public accountability, the evidence that contributions have a decisive impact on the efficiency of government is surprisingly lacking. MIT professors Stephen Ansolabehere, John de Figueiredo and James Snyder Jr. conducted an extensive survey of research on the effects of campaign contributions on legislation, finding that, in three-quarters of the studies, campaign contributions had no statistically significant effects on legislators’ roll call votes. Or, worse, they had the “wrong” sign — implying that more contributions led to less support. The authors concluded that legislators’ votes depended almost entirely on their own beliefs and on the preferences of their constituents and their party.

You might wonder whether the study has passed its sell-by date (it was published in 2003). But nothing new seems to have happened since the Supreme Court’s 2010 decision in Citizens United v. Federal Election Committee, which allowed corporations to spend unlimited amounts on elections. The predicted flood of campaign spending by corporations failed to materialize. Corporations, it seems, were generally satisfied with the status quo — a state that is preferred by other interests as well — and, in any event, they tend to get involved in campaigns only to defeat a change in policy that would harm them.

More recently, Anthony Fowler, Haritz Garro and Jörg Spenkuch of the University of Chicago and Northwestern University assembled data from some 19,000 elections and 3,000 firms spanning more than three decades. They found little evidence that public policy is distorted in favor of firms as a result of corporate campaign contributions. Why, then, do firms give so much in campaign contributions? They don’t, in fact, contribute much. Many large firms do not even have corporate PACs, and many of those that do make only nominal contributions.

A valid reaction to the preceding evidence is to ask why total lobbying expenditures have grown even if individual firms’ expenditures are not large. The late William Frenzel, a member of Congress for 20 years and my long-time colleague at Brookings, explained that lobbying was primarily defensive — as various policy proposals continue to surface, firms increasingly spend money to maintain the status quo.

This failure to explain government policy inefficiencies may be surprising to those, including me, who were schooled to believe that regulatory inefficiencies were the result of regulators who were captured by the interest groups they were supposed to regulate. Although that explanation struck many economists as intuitively consistent with the economic effects of regulation, supporting evidence is (again) scarce — and probably not likely to materialize as economists’ standards for identifying causal relationships become more sophisticated.

Self-Selection to Work In Government

Supply-side explanations for policy inefficiencies have been advanced much less often than interest-group explanations. The ones considered here are, therefore, to some extent ad hoc. But scraps of institutional and empirical evidence can be cobbled together to assess whether policy inefficiencies could be plausibly attributed to policymakers’ characteristics, their workplace environment or both.

People who self-select to work in government are not a random sample of the population in terms of easily measured characteristics such as age, ideology, military service and union membership — or, for that matter, characteristics that are not always easy to measure, such as competence, technical and analytical skills, ambition and willingness to take risks. Some characteristics, though, may suggest how the self-selected work force could affect both policy and performance.

The economist Francesco Decarolis and colleagues built measures of federal agencies’ competence, accounting for skills, incentives and cooperation among employees, to assess contract-execution efficiency. They found considerable variability in competence across government organizations. They estimated that if all bureaus were able to match NASA’s (high) level of competence, which was on par with private management standards, then delays in contract execution and price renegotiations would decline significantly — with annual cost savings approaching $15 billion. The authors also found that differences in individuals’ ability to cooperate with one another were more important than their skills. (The lack of interagency cooperation is a well-known explanation for the U.S. government’s failure to prevent the September 11 terrorist attacks, or respond more effectively to the events of January 6, 2021.)

Skill deficiencies do, however, appear to be important in specific situations. For example, a participant in several industry teams observed that the FAA’s electronic systems engineers were frequently unable to assess new technology offered by bidders and therefore often took bidders’ claims on faith. The problem is exacerbated by private industry’s ongoing efforts to identify the FAA’s promising hires and lure them away.

Government workers who held jobs that required a PhD or another professional degree were paid 18 percent less per hour than private-sector employees of equivalent education.

Congress shares this skills problem. A 2018 study found that very few Congressional staff had any formal STEM training. As a result, Congress is not up to the task of understanding how technology shapes society. During Facebook CEO Mark Zuckerberg’s testimony in April 2018, senior members of Congress did not seem to understand how Facebook made money or how it targeted ads — and some members could not distinguish Facebook from an email platform or an internet service provider.

In a recent book I wrote with colleagues, we documented the intellectual talent gap, as measured by their grades and the ranking of the law school granting their degrees, between lawyers who worked in the private sector and lawyers who worked in government. Importantly, we found that the government was unable to retain its most able lawyers. Similarly, the Washington Post reported three years ago that the National Security Agency was losing its top talent as highly skilled personnel have migrated to higher-paying, more flexible jobs in the private sector.

Pause here for an important distinction. Accounting for benefits like health care and pensions, the average federal worker’s compensation is actually about 75 percent higher than the average of all workers. And given that it is very difficult to get fired from a government job, the combination of security and compensation is sufficient to keep many government workers on the job for decades.

But that doesn’t mean the ones who stay are the ones you want to stay or the ones you want to attract in the first place. According to the Congressional Budget Office, government workers who held jobs that required a PhD or another professional degree were paid 18 percent less per hour than private-sector employees of equivalent education. That helps explain why Uncle Sam faces a chronic shortage of technically skilled workers and must continually cope with a brain drain. Note, too, that long-termers may well tend to undermine policy performance by their tendency to reinforce a status quo bias because they do not want to disturb their “sweet gig” by advancing change. As noted below, many factors play a role in government’s status quo bias.

Biased Preferences

Government policy inefficiencies may also arise when government bureaucrats follow their own preferences. Alexander Hertel-Fernandez and colleagues found that Congressional staff fail to correctly determine constituent preferences. Moreover, policymakers who indulge their own preferences, however formed, rarely look back to learn from past mistakes.

Jonathan Masur and Eric Posner identified regulatory agencies that have failed to conduct a careful cost-benefit assessment before deciding to implement a regulation or that have issued regulations even when those regs failed a cost-benefit test — in one case, by 1,920 to 1! Richard Epstein pointed out that the Office of Management and Budget, which is supposed to provide general oversight over regulatory agencies, relies on the agencies’ own estimates when they are available.

Presidents Obama and Trump both used their executive authority to finalize hundreds of rules and regulations without evidence that they improved social welfare. Many of Trump’s executive actions simply sought to undo Obama’s executive actions, such as rolling back fuel economy standards, and President Biden is now undoing many of Trump’s executive actions.

Trump also indulged his own “America First” preferences by initiating protectionist policies, such as tariffs and tighter limits on immigration, which clearly harmed consumers’ welfare and firms’ productivity, in return for questionable domestic job savings. Some of his sins were sins of omission. Trump failed to press for the repeal of the Jones Act, which forces the use of costly U.S.-built ships and crews in domestic waters, and he failed to spur airline competition by allowing foreign carriers to fly from city to city within the United States.

Colleagues and I found evidence of growing ideological polarization for or against business firms among justices on the Supreme Court, which could have adverse consequences for policy. The former federal appellate court judge Richard Posner explained such a finding by arguing that when policymakers are not committed to a rigorous method for aiding a decision (for example, cost-benefit analysis), those decisions are by definition ideological because they cannot be anything else.

Workplace Environment

The government workplace environment may inhibit constructive change and promote waste because resources are inadequate to do the job, while employees suffer from low morale. The government’s share of total employment — especially the federal government’s share — has been declining and is now at its lowest point since the 1950s. Some of the decline may be attributable to automation. But the scope and scale of government tasks have certainly grown since the 1950s. So it also reflects resource constraints, political sentiments and poor leadership that reduce morale and undermine the quality of work. This was exacerbated by the Trump administration’s efforts to fight the “dark state” and “drain the swamp” without assessing the consequences. Some specific examples:

  • Trump’s political appointees shut down studies and reduced the influence of government scientists over policy decisions. In the first two years of his administration, more than 1,600 scientists left government — 700 from the Environmental Protection Agency alone.
  • Mick Mulvaney, the former acting director of the Consumer Financial Protection Bureau, openly undermined the agency by allowing the staff to shrink and enforcement to dwindle.
  • Researchers at the Department of Agriculture claimed that the secretary of the department pressured them to leave the Washington office — where they had efficient communication with Congress — as punishment for assessments questioning Trump’s policies: At the same time, there are cases in which inefficiency is driven by inertia or legal protection. Sometimes there is little disagreement that a program or entire agency has outlived its usefulness, but it nevertheless cannot be eliminated. Examples include:
  • The National Technical Information Services agency exists to sell government documents, most of which can now be found on the internet for free. Closing the agency would save the government $50 million per year.
  • Airline-service subsidies are costly but provide few benefits to the intended beneficiaries — air travelers in small communities. In some cases, air service subsidies are provided to airports just a two-hour drive from major hub airports.
  • Between 2010 and 2013, some 4,000 federal employees remained on paid leave for three months to a year, at a cost of $700 million, because the employing agencies didn’t have the will or legal means to get rid of them.

Resource constraints also breed inefficiencies. The Occupational Safety and Health Administration, the Mine Safety and Health Administration and the Consumer Product Safety Commission are widely understood to be weak rule enforcers because they do not have enough personnel to inspect the safety of workplaces, mines and consumer products.

Then there’s the Food and Drug Administration. Before generic drugs can enter the market to compete with branded versions (and thereby drive down prices), the agency must decide they have the same strength, purity and stability as branded drugs. However, the FDA is so understaffed that lags between application and approval average four years.

Similarly, the Patent and Trademark Office has been overwhelmed with roughly 650,000 patent applications in recent years. One consequence: since they do not have enough examiners to spend sufficient time to carefully review applications, there’s an increased probability that patents will be challenged in court. In the interim, however, such patents can chill innovation and competition.

In some cases, the contrast between understaffed and overstaffed agencies is startling. The National Highway Traffic Safety Administration has only three employees for every 1,000 auto fatalities to monitor the increasingly complex software in vehicles. In comparison, the FAA has over 100 for each fatality on commercial aircraft.

Those anomalies are linked in part to the aforementioned problem of getting and keeping technically savvy employees in all levels of government — including elected officials. For example, until Eddie Bernice Johnson (who was trained as a nurse) became chair of the House Committee on Science, Space and Technology in 2019, that committee had not had a chair with any science background whatsoever since the 1990s. Christos Makridis of Stanford concludes that poor pay, apparently, isn’t always the reason for lack of tech savvy; it’s the paucity of opportunities for promotion and rigidity in the work environment.

To be sure, lower pay can be a serious drawback to working in government. For example, the government is seeking to build and sustain a cybersecurity work force, but an astonishing 37 percent of available public sector cybersecurity jobs remain unfilled. The huge gap exists because the public sector must compete with the private sector for the limited supply of qualified candidates with a pay scale that tops out at well below the salaries of information security officers at large companies.

The Problem With Law And Lawyers

A slight digression here. While the law and lawyers are essential for formulating and applying public policy, they both have features that sow the seeds of inefficiency. The strength of strict rule by law is that it establishes clear boundaries and governs interactions. Its weakness is that proposed laws promising constructive change can be blocked by small numbers of opponents. The more veto players in a political system — and the United States has an exceptionally high number of them — the more difficult it is to avoid status quo bias and to adopt socially beneficial policy reforms.

Lawyers account for the lion’s share of policymakers, especially in Congress. However, lawyers, like economists, can be slaves to their training and professional culture. For example, administrative lawyers are comfortable — too comfortable — with highly detailed regulations. The Dodd-Frank Act spawned 14,000 pages of rules in the Federal Register on top of the law’s 2,300 pages.

Moreover, lawyers’ cultural mindset can be quite damaging to social welfare. In our recent book, Trouble at the Bar, my colleagues and I argued that the pervasive role of lawyers in government has had an adverse effect, because lawyers’ training and career development occur in an environment shaped by regulations that reduce competition and innovation and foster status quo bias. By the same token, legal training and practice do not encourage policymakers to acknowledge error and correct their policy inefficiencies by subjecting previous decisions to rigorous retrospective cost-benefit analyses. Precedents are precedents, sufficient to stifle introspection or change.

Interactive Effects of Policy Inefficiencies

Wait, there are wheels within the wheels. In too many cases overlooked by policymakers, the initial impact of policy inefficiencies is magnified by secondary losses. Here are a few examples:

  • Tariffs, it is widely understood, are inefficient because the protected domestic industry can’t make the goods as cheaply as the would-be foreign supplier. But tariffs turn out to be the regulators’ “gift” that keeps on giving. Blocking imports also increases market concentration at home and opens the door to antitrust concerns.
  • SEC regulations intended to protect investors increase firms’ regulatory burdens by requiring more corporate-compliance employees and by rewarding companies with experience in coping with regulation. This can cause weaker firms to exit and outsiders to avoid challenging incumbents, increasing prospects for industry concentration and market power.
  • Inadequate investment in highway maintenance increases road roughness — potholes and the like — which reduces fuel economy and increases emissions. Greater road roughness also may reduce highway safety.
  • Zoning restrictions increase home prices, forcing people to live further from jobs and increasing traffic congestion. The failure to reduce urban congestion has led to larger vehicle sizes, which reduces highway safety and air quality.
Possible Explanations for Policy Inefficiencies

Let’s pull together what we know about the forces that contribute to policy inefficiencies and about the policy inefficiencies themselves to see if we can determine a causal link. Government has an enormous macro- and microeconomic impact on the economy. Hence, policy inefficiency exacts large costs. But as discussed above, the evidence just doesn’t support “demand-side” explanations — in particular, lobbying and campaign contributions — as the major source of inefficiency.

Even when studies conclude that policymakers were captured by a particular interest group, alternative supply-side explanations intrude. For example, Haris Tabakovic (Harvard) and Thomas Wollmann (Chicago) argued that patent examiners awarded more patents to firms that subsequently hired them — and that the standards for those patents were subpar. Score one for the notion of “regulatory capture.”

However, it also is possible that government inefficiency could be attributed to the personnel, supervision and resource constraints in the patent office because the agency does not attract and retain the strongest scientists and engineers. Sometimes marginally qualified examiners are overwhelmed with applications and under pressure to move them through quickly. Moreover, the patent examiners have supervisors who are apparently unable to spend the time needed for quality control. The authors focused on the revolving door, yet it was certainly possible that the overwhelmed examiners who remained in the patent office also approved problematic patents to the firms that, perhaps coincidentally, hired away their colleagues.

Similarly, in the wake of the two crashes of the Boeing 737 Max, critics claimed the FAA had been captured by Boeing because the initial certification process was flawed and the agency was subsequently slow to ground the flawed plane. But consider a supply-side explanation: the FAA left the certification analysis to Boeing because it lacked highly skilled personnel to evaluate the safety of the 737 Max.

I believe the evidence tilts toward supplyside explanations, which can have dynamic effects. This view, by the way, is consistent with the best after-the-fact explanations for why deregulation in the 1980s so dramatically reduced transportation prices. It wasn’t the redistribution of carriers’ rents that, before deregulation, had been taken by the protected incumbent air carriers. It was mostly the consequence of dynamic efficiency gains from innovation — for example, the wider use of hub-and-spoke route structures by airlines — once market forces were unleashed.

Ultimately, most government policy inefficiencies appear to reflect status quo bias caused by some combination of policymakers’ entrenched ideology, the legal system, insufficient incentives for efficiency and lack of resources. As noted, the private sector often prefers this situation, too. Status quo bias inhibits policymakers from learning about the effects of their policies and how they could be improved. It also enables inefficiencies in one area to persist and compound inefficiencies in other areas. The self-selection of people who choose to work in government and the inflexibility introduced by a tangle of rules also undoubtedly strengthen status quo bias and make it less likely that government will reform its inefficient policies.

Status quo bias may be an unsatisfying explanation for inefficiency because it is hard to quantify and encompasses so many hard-topin- down factors. Nonetheless, it is broadly consistent with the evidence. Peter Schuck’s 2014 study of why government fails concluded: “The political process is so complex and opaque, and the role of ideological zeitgeist is so pervasive, that it is usually impossible — even after the fact — to know for sure how weighty each of the many politically relevant factors might have been in determining the policy outcome.” We do know that it takes a long time (assuming it ever happens) to reform policies that are reducing social welfare.

Consider the frustrating effort in the United States to adopt efficient road pricing in big cities. For decades, New York City politicians opposed congestion pricing on the grounds that it would place an undue burden on the large share of their constituents who used their cars to commute, and therefore would not have the option to avoid high peak-hour tolls. Yet Assemblyman David Weprin was the most vocal opponent of the New York tolling plan, even though only 4.2 percent of Weprin’s constituents would have to pay the new toll. Go figure.

congestion pricing was flatly rejected in 2018 by the New York State legislature, which controls the bridge authority. Yet, just one year later, the politics took a seismic shift when it was revealed that New York City’s mass transit system was in desperate need of money

Initially, Weprin prevailed, as congestion pricing was flatly rejected in 2018 by the New York State legislature, which controls the bridge authority. Yet, just one year later, the politics took a seismic shift when it was revealed that New York City’s mass transit system was in desperate need of money, and motorists were targeted as a politically acceptable source of revenue. The realignment of interests is hard to track. But it is likely that some form of congestion pricing eventually will be implemented in Manhattan.

Or consider the effort by former Congressman Bill Shuster to spin off air traffic control from the Federal Aviation Administration, and have it run as a private nonprofit corporation with the goal of improving its efficiency and spurring technological advance. President Trump had made strong statements about the FAA’s failure to modernize the ATC system, touting that reform would be part of his administration’s infrastructure bill. However, it was strikingly absent from the administration plan subsequently released. What’s more, the White House informed Shuster that the president would not make any efforts to enable ATC privatization. Shuster subsequently pulled the ATC section from the bill.

One might assume that special interests opposing ATC reform that might raise their air travel costs got to Trump. In fact, the reason for the about-face is a bit of a mystery, since it was especially hard to follow influence trails in Trump’s Washington. Some even suspect that Trump was ultimately convinced by his (private) jet pilot that the change in ATC governance was a bad deal.

• • •

Nobody disagrees that government policy inefficiencies have been widespread, persistent and costly. But the usual suspects are often hard to pin down when it comes to figuring out why inefficiencies occurred and persisted after they were visible. This is not the conventional view, however, and every policy in the headlines brings discussion of which special interests hold the best cards and how to beat them. I believe it is reasonable to speculate about the causes of policy inefficiencies, but misguided to be optimistic that there is ever a practical adversarial-interests roadmap to guide policy reform.

This is an important conclusion, because it suggests that the answer to government failure is not simply to identify the efficient policy reform and convince policymakers to overcome the obstacles to it. That won’t happen often, because we don’t know — and we may never know — the right buttons to push. More constructively, Gaining Ground does suggest there are many ways in which market forces have corrected government failures over time and have the potential to correct additional failures if policymakers will only give markets a chance. That approach may not appear to result in enough progress for some, but realistically it is the best we are going to do. Better yet, it will result in considerable progress.

main topic: Governance