However, solving problems through the federal government is not so simple. Spending programs require funding from taxation or borrowing, and both create collateral damage on the economy. As funds flow through the government, resources are wasted on bureaucracy, mismanagement, and faulty planning. As the remaining funds flow out of the government and into the private sector, they induce people to change their working and investing activities, further undermining the economy. Only a fraction of the tax dollars raised for programs ultimately delivers on the outcomes promised by policymakers.

Economist Arthur M. Okun described this problem as a “leaky bucket.” When the government imposes taxes on some people to aid others, it is not a costless transfer of resources. Rather, it is like transferring water using a bucket riddled with holes. In this study, we describe the many types of holes in the government bucket, and we argue that the leakage of resources is large. We also argue that the government bucket gets leakier as federal programs and bureaucracies continue to expand.

With the government running huge budget deficits, policymakers should rethink the many costly programs that it runs. Some programs are essential, and so policymakers should preserve them while reducing the types of waste or leakage that we identify. But many other programs are so leaky that their costs outweigh their benefits. These programs are spilling economic resources all over and should be eliminated from the federal budget.

Policymakers and the public often notice social or economic problems, such as poverty or infrastructure needs, and assume that expanded federal programs can fix them. People seem to think that taxes can be raised easily and funds expertly steered to individuals and areas of the country that need them the most. With generous funding, just about any problem can be solved with a federal spending program.

Such a romantic vision of the government does not pass scrutiny. From the collection of taxes to the distribution of benefits, many factors create a wedge between the cost of the dollars raised and the program benefits created. There are administrative costs, bureaucratic mismanagement, and pork-barrel politics. Also, taxing and spending generate behavioral changes by individuals and businesses that undermine the economy. Only a portion of every tax dollar raised ultimately delivers on policymaker promises.

In his widely cited book in the 1970s, economist Arthur M. Okun described this problem as a “leaky bucket.”¹ He said that when the government taxes some people to provide welfare benefits to others, it is like carrying water using a bucket riddled with holes. Substantial value “leaks” along the way. We use Okun’s idea in this study and expand on it to explore the many types of leaks or inefficiencies in federal spending programs.

Figure 1 illustrates the flow of funding for a government program. Funds are raised from taxpayers, flow through the government, and then are handed out to program recipients. Leaks occur at each step. Higher taxes generate compliance costs and reduce incentives for working, saving, and entrepreneurship, with the result that taxes cost the private sector more than just the value of the funds raised.

More leakage occurs as funds flow through the government. Policymakers allocate resources to low-value or wasteful activities because of the difficulty in central planning and the political realities of passing legislation through Congress. Resources are also consumed by administrative costs and bureaucratic failures in federal agencies.

Finally, further leaks occur when funds flow to the private sector. One problem is that individuals and businesses change their behaviors in unproductive ways. Individuals receiving welfare may stay on the couch rather than work, while businesses receiving subsidies may focus on lobbying rather than product innovation. These inefficiencies reduce overall incomes in the economy, thus undermining any benefits that federal programs may create.

Some of these leaks or inefficiencies overlap, and some are in tension with one another. Reducing improper payments, for example, may require higher costs for administration. However, most federal programs have many leaks of value that accumulate to the point where the costs are larger than the benefits.

Okun focused on social welfare programs. He said that society faces an equality–efficiency tradeoff: We can transfer money from the rich to the poor, but some of the money will leak along the way. His point was that policymakers should consider this constraint before funding programs. Okun’s idea has broader applicability than just welfare programs. Resource leaks occur for almost all federal spending, whether it is buying weapons for national defense, funding infrastructure projects, or paying benefits for health care programs.

Okun was optimistic in assessing the size of the leaks or inefficiencies in the government. In an interview, he said that you “don’t get 100 percent efficiency in making [a government] transfer. But in many cases you can get 80 or 90 percent if you do it right.”² We argue that combined taxing and spending inefficiencies are much larger than that, and we examine numerous types of inefficiencies that Okun did not discuss.³

In this report we present a structured way to see why solving problems with federal programs is not as simple as policymakers seem to assume. In the next section we will discuss why raising taxes costs the private sector more than just the funds collected by the US Treasury. In subsequent sections we tackle the inefficiencies within the government and the distortions generated by government spending on the private sector. And in the final section we discuss how inefficiencies and distortions—leaks—intensify as the federal government grows larger.

When politicians propose new programs, they make it sound simple: Collect taxes, use the money to achieve a popular goal, and the nation is better off. But the process becomes complex as soon as the government tries to raise money.

Every dollar the government collects in revenue is a dollar that families and businesses cannot spend on their own priorities. That is the direct cost of taxes to the private sector. But extracting taxes imposes additional costs because taxes create compliance burdens as well as disincentives for productive activities. To raise $1 for a government program, it costs the private sector more than $1.

Extracting taxes from individuals and businesses causes damage. Taxes reduce incentives for working, saving, investing, and entrepreneurship, which results in shrinking output and incomes. Economists call this the “deadweight loss” of taxation. Taxes alter individual and business behavior in unproductive ways, and society loses out on beneficial exchanges that would have occurred otherwise. This is leakage from the government bucket.

Suppose that the government taxes wine. Drinkers now pay higher prices, directly losing money to Uncle Sam. But there is a secondary hit: People will buy less wine because of the higher prices. Consumers miss out on the enjoyment those bottles would have provided, and wine producers lose sales. Mutually beneficial trades between buyers and sellers vanish and value is lost.

The losses can be illustrated on a supply-and demand-diagram. Figure 2 shows that before a wine tax of $1 per bottle is imposed, people consumed 100 million bottles at $10 per bottle. After the tax, the consumer price rises, the after-tax price to producers falls, and the quantity traded drops to 90 million. The rectangle is the tax revenue on the 90 million bottles—a transfer from buyers and sellers to the government. The triangle is the deadweight loss: the losses to society from the 10 million mutually beneficial trades that are cancelled.⁴ To a greater or lesser degree, every tax causes this sort of damage by hindering market exchanges. Income taxes, for example, reduce the working and investing efforts of millions of individuals and businesses.

As government spending increases, taxes must rise to fund it, either now or in the future. The heavier tax burden hurts in two ways. First, people perform less productive activities, such as work, saving, investing, and starting companies. Second, they perform more unproductive avoidance and evasion activities to reduce their tax burdens, which wastes resources and misallocates capital.

Economists Michael Keen and Joel Slemrod note that the deadweight loss “suffered by the taxpayer is the same whether the response is in terms of real economy activity … or takes the form of evasion and avoidance.”⁵ The economists note that the British government once taxed windows, causing people to board them up.⁶ That response reduced the tax base but it also destroyed access to sunlight. Today’s methods of avoidance and evasion are more sophisticated, but they destroy value as well.

If the government funds new spending with more borrowing, it delays the tax damage until later. But because the government pays interest on its borrowing, that tax damage will be even larger down the road. Borrowing also causes other negative effects, such as unsettling financial markets. Borrowing does not allow policymakers to avoid the deadweight losses caused by taxation.

How large are deadweight losses? It varies depending on the tax rate, the type of tax, and other factors. But for the federal income tax, studies have found, on average, that the deadweight loss of raising taxes by a dollar is about 40 cents.⁷ The Congressional Budget Office (CBO) found that “typical estimates of the economic cost of a dollar of tax revenue range from 20 cents to 60 cents over and above the revenue raised.”⁸ That means for every $1 billion in higher taxes, the harm imposed on the private sector is between $1.2 billion and $1.6 billion. When considering the costs and benefits of a program, policymakers should include these losses.

Suppose that Congress is considering spending $10 billion on an energy program. Putting aside whether the program is ethical or constitutional, does the program make economic sense? The program’s benefits would need to be higher than the total cost to the private sector of $14 billion, which includes the $10 billion direct cost to taxpayers plus another $4 billion in deadweight losses. In collecting $10 billion in taxes, the government’s bucket sprang $4 billion in leaks.

Deadweight losses are not the only leak caused by taxation. Resources are also wasted by tax compliance, which includes the time, the accountant and lawyer fees, and the household and business paperwork needed to deal with complex tax laws.

The compliance costs imposed by federal taxes were estimated at $546 billion in 2024, which was more than 10 percent of the tax revenues raised.⁹ However, when thinking about funding new programs, we are interested in the marginal impact: How much will compliance costs increase to raise funds for new spending?

The answer depends on the type of tax. Federal payroll taxes are simpler to collect from businesses, for example, than complex corporate income taxes. Also, policymakers could generate higher revenues either by raising tax rates or by closing loopholes in complex tax structures. In the latter case, the marginal compliance costs would be negative because revenues would increase as tax paperwork was reduced.

In recent years, policymakers have favored complicated ways of raising revenues, not closing loopholes. The 2025 version of President Trump’s tariffs imposed annual compliance costs of $39 billion or more, by one estimate, and raised about $132 billion.¹⁰ The ratio of compliance costs to revenues was thus 30 percent or more. There are more than a dozen US tariff regimes, and there were more than 2,000 customs rulings in 2025.¹¹ The Harmonized Tariff Schedule now exceeds 4,000 pages in length. Raising funds by tariffs imposes high marginal compliance costs.

Recent corporate tax increases have also been complex. Congress has cut the corporate tax rate but it has also imposed layers of complicated new rules on foreign income. Overall, corporate tax compliance costs US companies about $119 billion a year, which was 26 percent of the $452 billion collected in federal corporate income taxes in 2025.¹²

In sum, average tax compliance costs are about 10 percent of revenues raised, but marginal compliance costs may be higher or lower. These days, they are more likely to be higher because policymakers are favoring complicated ways of raising taxes.

The CBO does not include tax compliance costs or deadweight losses when it analyzes spending programs. Most federal agencies do not consider them either, even though the Office of Management and Budget has said that agencies may include deadweight loss estimates in their program evaluations.¹³ By ignoring these costs, policymakers are more likely to approve programs that are losers for the economy.

To see why, compare a private charitable project with a government project. Suppose that a philanthropist creates a $10 million project to help poor children by raising the money through market transactions and voluntary contributions, which do not generate the deadweight losses inherent in taxation. If the project yields, say, $12 million in benefits, it is worthwhile.

Now the government attempts the same project with tax financing. It raises $10 million in taxes, which in turn imposes, say, $4 million in deadweight losses and $1 million in added compliance costs.¹⁴ The government version of the project is a loser because the total cost of $15 million is higher than the $12 million in benefits. Even if government management is as efficient as private management, it makes sense to leave this program in the private sector.

With taxes raised, funds for spending programs flow through the government, where more resources leak away. Leaks include failures from central planning, misallocations from congressional politics, program design conflicts, and failures by the bureaucracy. The diminished funds eventually flow out to recipients for lower-value and poorly targeted activities.

The federal government is a giant transfer machine. It takes trillions of dollars of taxpayer money and borrowed funds each year and reallocates them to individuals and businesses as benefits, subsidies, and contracts. How does Congress know which activities to spend on? How does Congress ensure that programs generate benefits that are higher than costs?

Policymakers face huge obstacles to allocating resources efficiently, even aside from the political biases that influence Congress. They lack sufficient knowledge to make accurate choices on which programs to fund and at what levels, and they lack good feedback mechanisms to help them change course when conditions change. They mainly rely on guesswork, and the resulting errors are leakage from the federal spending bucket.

Consider how markets allocate resources: Millions of buyers and sellers pursuing their own interests engage in billions of exchanges. The exchanges are voluntary, which indicates that they are mutually beneficial and generate net value. As supplies and demands fluctuate, they create market prices that help to guide individual and business spending decisions. Prices communicate information about supplies, tastes, and technology, and they create incentives for efficient spending.

Government spending decisions do not rely on such coordinating mechanisms. Its decisions do not arise from voluntary trading and do not generate prices reflecting resource availability. There are no effective feedback signals indicating whether programs are generating net value. Policymakers face an information void, and so their spending decisions are ad hoc—essentially guesses about what might make sense.

Compare the purchase of aircraft in the public sector to the private sector. In the latter, an airline company chooses the number of planes to buy based on the demand for air travel, which is aggregated from millions of individual choices for trips expressed in the marketplace. By contrast, when the Pentagon buys aircraft, the number chosen is decided by guesses regarding future defense needs. There is no market that generates information about the amount of threat reduction gained by buying each additional aircraft.

Consider federal spending choices more broadly. Would new fighter jets, farm aid, or food stamps be the best use of marginal public funds? In markets, prices inform such decisions. If the price of air travel goes up, consumers reduce spending on air travel and increase spending on automobile travel. By contrast, the government flies blind when it allocates spending because there are no solid metrics to rely on. Government program recipients benefit from spending, but that does not mean that the benefits outweigh the costs or that programs are the best use of funds.

Once the government starts spending on a program, it rarely changes direction even as conditions change. The federal government, for example, has spent more than $100 billion on rural broadband hookups over the past 25 years.¹⁵ Even if that spending once made sense, it no longer does because the cost of satellite internet service has plunged.¹⁶ Yet the government continues to spend billions of dollars a year subsidizing expensive land-based rural broadband.¹⁷

Private firms in markets make bad decisions and misjudge technologies and markets. But if they continue down the wrong path, their resources get depleted. A business making bad investments will be punished by financial losses and may face bankruptcy. Ultimately, resources will be reallocated to more promising activities over time.

By contrast, when the government makes poor spending decisions that generate losses for society, those activities may persist for years. Policymakers should rigorously investigate programs and regularly reallocate spending, but in practice that rarely happens. Also, because there are no clear success and failure signals such as profit and loss, policymakers rarely agree on which programs should be cut.

The failure of government planning is exemplified by the Navy’s littoral combat ship (LCS) program, which will cost about $100 billion. Investigations have found that the ships were ill-conceived and poorly constructed. The ships have failed at their planned missions, including mine hunting and anti-submarine warfare, and they have weak survivability, limited endurance, and last only 10 years, not the planned 25 years.¹⁸ The ships also doubled in cost from the original estimates.

The Pentagon has conceded that the LCS “does not provide the lethality or survivability needed in a high-end fight.”¹⁹ Yet the program continued despite growing evidence of failure because “once a massive project gains momentum and defense contractors begin hiring, it is politically easier to throw good money after bad.”²⁰ The government does not have good mechanisms to quickly discontinue wasteful programs before billions of dollars are wasted.

Economist Frédéric Bastiat said that a “public service” provided by government often becomes a “public nuisance” because it gets entrenched and outdated.²¹ The LCS and broadband subsidies were probably bad choices to begin with, but they certainly became resource-wasting nuisances over time.

Congress passes spending bills in large packages that include funding for hundreds of programs and projects. Policymakers do not perform cost-benefit analyses on each spending item to see whether it may create net value for the nation. Rather, bills are passed by members horse-trading over the benefits aimed at different states and interest groups. Many spending items that pass do not have majority support on their own. As such, the bundling of items in large bills can create substantial waste.

Farm bills, for example, are large bundles of subsidies sprinkled around to each part of rural America. When he was in Congress, Joe Scarborough of Florida noticed that dairy subsidies gained support from Maine, Pennsylvania, and Vermont; peanut subsidies gained support from Virginia, Alabama, and Georgia; sugar subsidies gained support from Florida; and cotton, wheat, wool, mohair, and other products gained support from other states. Scarborough argued that “standing alone, not one of these corporate welfare measures could survive the bright light of public scrutiny.”²² Yet when they are bundled, a majority clears it for passage.

There is a similar logrolling process for transportation bills. Rural members want funding for highway projects, which can make economic sense, but to gain a majority for passage, Congress must add spending for light rail systems favored by urban members. Those light rail systems are usually economic losers.²³ There is similar waste with passenger rail funding, but the politics are reversed. Passenger rail can make sense in the urbanized Northeast, but for such funding to gain passage, Congress must add spending on money-losing rural train routes favored by members in less-populated areas.²⁴ In sum, to pass sensible spending, Congress must often include unsensible spending.

With national defense, members will support procuring additional weapons manufactured in their states even when the Pentagon does not want so many of them. Some examples have been the M1 Abrams tank, the C130J transport aircraft, and the F‑22 Raptor aircraft.²⁵ In his book The Wastrels of Defense, congressional aide Winslow Wheeler said that parochial gain in defense spending “has become a full-time preoccupation that permeates Congress’s activities and members’ decisionmaking processes.”²⁶ He argued that members on the defense committees prioritize securing contracts for their states over building the most effective military force for the nation.

Parochial politics push up the costs of defense procurement. Contracts for weapons systems are deliberately spread out across dozens of states to gain political support. The F‑35 Joint Strike Fighter, for example, had its procurement spending spread across 45 states, which made the spending “bullet-proof” in Congress, noted one oversight group.²⁷ The incentive to spread contracts widely likely raises taxpayer costs.

The parochial spending disease runs rampant with earmarking. Rather than having experts in federal agencies at least try to steer funds to high-value projects, congressional earmarks require spending on specific projects in specific districts. These are often items of purely local concern such as roads, bridges, bike paths, and museums. While the House mainly banned earmarks from 2011 to 2020, they have since returned big time. The Transportation, Housing, and Urban Development appropriations bill for 2024 included more than 3,400 earmarks, while the Consolidated Appropriations Act of 2023 included more than 7,000.²⁸

Some members of Congress try to focus legislation on true national interests, but they face a prisoner’s dilemma: If they do not secure funding for their districts, they fear the money will be spent by other members, not for deficit reduction. This is a “common pool” problem, akin to a lake that gets overfished. Even members who favor overall restraint are incentivized to try and grab spending for their districts even for low-value projects.

Once federal programs are enacted, they may perform poorly year after year, yet they are given larger budgets rather than being cut. Members do not want to admit that their favored programs have failed because their careers, reputations, and pride are on the line. Even after a disaster or scandal when a program’s failure becomes obvious, members usually blame the federal bureaucracy rather than the ill-advised program itself.

Furthermore, spending programs often deliver visible benefits to organized groups, but the costs are diffused across tens of millions of taxpayers. Policymakers are barraged with messages and Capitol Hill visits from program advocates. Nearly all witnesses to appropriations hearings favor the programs being examined.²⁹ Members receive campaign support from spending groups, and they look forward to post-congressional careers working with them. This makes them less likely to support cuts even to low-value programs.

The pro-spending bias of Congress is strengthened by “fiscal illusion.” Lawmakers hide the full tax costs of programs from the public by using techniques such as borrowing, inflation, and income tax withholding.³⁰ As a result, the public perceives the price of government programs to be artificially low, and so their demand for more programs from Congress is too high.

In sum, the legislative process generates many leaks or inefficiencies in spending. Low-value programs are enacted by a logrolling process and then become entrenched. Congress can seemingly borrow endlessly and is not forced to make trade-offs. Part of the solution is to devolve programs to the state level, because the states have balanced-budget rules that force trade-offs.³¹ But Congress has not shown much interest in program devolution in recent years.

Congress decides which activities to fund and designs programs to meet its goals. But those objectives can be undermined by program features that work at cross purposes. Sometimes policymakers include program rules that artificially raise costs. Other times, programs use blunt tools to solve complex problems that create internal conflicts. They have leakage built right into the legislation—provisions that are like a fifth column working against the goals.

Federal infrastructure programs often include provisions that raise costs and slow projects. Federal highway aid, for example, comes paired with labor rules under the 1931 Davis–Bacon Act, which raise highway construction costs an average of about 20 percent.³² Excessive environmental rules also raise costs and cause highway project delays.³³ The result of such rules is that the public receives fewer miles of highway for each tax dollar spent.

The costs of federal projects are boosted by Buy American provisions, which mandate the use of US inputs even when they are more expensive. The $550 billion Infrastructure Investment and Jobs Act (IIJA) of 2021 expanded Buy American rules to cover all types of federally funded infrastructure.³⁴ The rules essentially mandate that taxpayers receive infrastructure worth less than the $550 billion they were promised.

Federal urban transit spending provides an example. Because of the Buy American rules, US cities are paying one-third more for rail cars than are European or Asian cities.³⁵ As a result, taxpayers receive less transit service for each dollar invested. Buy American rules also delay transit projects because contractors sometimes have problems finding domestic inputs.

Policymakers may believe that Buy American rules nonetheless create jobs and help the economy, but that is not the case. The rules reduce gross domestic product and the overall number of US jobs.³⁶ A 2024 study found that there is “scant evidence of the use of Buy American rules as an effective industrial policy.”³⁷ So the Buy American rules are a failure all around—a big leak in the government’s spending bucket.

Perhaps it is an apocryphal story, but economist Milton Friedman once toured a government infrastructure project in Asia.³⁸ Armies of workers were using shovels to build a canal. Friedman asked why modern machines were not being used, and a government official explained that using shovels created more jobs. Friedman retorted, “Then why not use spoons instead of shovels?”

Unfortunately, many federal programs use spoons. Consider the CHIPS and Science Act, a 2022 law aimed at boosting the US semiconductor industry. The goal was to build more semiconductor plants in America, but the law imposed rules that raised costs and delayed investments.³⁹ Davis–Bacon mandates ratcheted up costs and slowed the award process.⁴⁰ Also, companies receiving subsidies had to follow costly rules on diversity, environment, and even employee childcare.⁴¹

Regarding the CHIPS Act, the CEO of Microchip Technology, Ganesh Moorthy, told Politico, “The journey to receive grants has taken much longer and been more complicated than we expected,” and added that multiple federal agencies are “all driving their own agendas.”⁴² Meanwhile, the Taiwan Semiconductor Manufacturing Company (TSMC) was promised CHIPS subsidies for a new plant in Arizona, but by 2024 the federal government was still “laying down more conditions that can slow down the process,” and by then negotiations between TSMC and the government had “dragged on for more than 18 months since the 2022 Chips and Science Act.”⁴³

Journalist Ezra Klein has bemoaned “everything bagel liberalism,” which is the tendency for liberals to support programs laced with cost-inflating provisions.⁴⁴ This tendency is not just limited to liberal Democrats, as Republicans also favor cost-increasing rules such as the Buy American provisions.

Another way that Congress builds internal conflicts into programs is by using blunt tools to tackle complex problems. A good example is the Earned Income Tax Credit (EITC), which is a spending program providing more than $60 billion a year in wage subsidies to low-income individuals. The idea is that, because it only goes to people who are earning an income, it gets people off the couch and into the workforce.

The EITC fulfills that goal for individuals at some income levels, but it has the opposite effect for individuals at other income levels. In particular, the credit discourages additional work at incomes where benefits are being phased out as earnings rise. About three-quarters of EITC recipients are in this phase-out range.⁴⁵ Thus, some EITC benefits leak on a net basis; that is, pro-work incentives are partly offset by anti-work incentives.

The $100 billion Supplemental Nutrition Assistance Program (SNAP) is another program with internal conflicts. The “N” in SNAP is for nutrition, yet 23 percent of the program’s benefits are spent on junk food—soda, candy bars, and other low-nutrition foods.⁴⁶ SNAP recipients are allowed to use their benefits to purchase just about any food in grocery stores. The large subsidies for junk food come at a time of record levels of obesity in the nation.

The SNAP program has a 23 percent leak—spending that directly contradicts the program’s stated purpose. The official SNAP webpage claims that the program allows people to “afford the nutritious food essential to health and well-being.”⁴⁷ The reality is that recipients have less nutritious diets than other Americans, including low-income individuals not on SNAP.⁴⁸

In sum, internal conflicts can bake spending leakage right into program legislation. Congress usually needs to make many political compromises in passing bills, and that is why federal programs are usually an inefficient way to solve problems in society.

Congress may think that once it has enacted a program with a goal—such as improving education, health care, or military security—then adequate funding will accomplish the task. But that is often not the case because of failures within federal agencies.

Federal agencies have suffered an endless series of major failures: The Federal Emergency Management Agency failed disastrously in its response to Hurricane Katrina in 2005.⁴⁹ The Secret Service allowed an intruder to jump the fence and walk inside the White House in 2014, and it did not prevent a near-assassination of presidential candidate Trump in 2024.⁵⁰ Federal air traffic control failures have become commonplace, including a catastrophic accident at Washington’s Reagan National Airport in 2025.⁵¹ The Pentagon withdrawal from Afghanistan in 2021 was horribly planned, and 13 US service members died in the chaos.⁵² And federal agencies allowed hundreds of billions of dollars to be stolen from COVID–era relief programs.⁵³

Polls indicate that Americans think the federal government does a poor job. One study found that 74 percent of people believe the federal government is “corrupt,” 85 percent think it is “wasteful,” and 66 percent say it is “incompetent.”⁵⁴ Another poll found that Americans think that “the federal government wastes 59¢ of every $1 it spends.”⁵⁵ And yet another found that 56 percent think that the government is “almost always wasteful and inefficient.”⁵⁶

The public is generally right about the federal government, and there are numerous factors that create such poor performance:

• Absence of Profits. Unlike businesses, federal agencies do not have the straightforward and powerful goal of earning profits. As a result, agencies have little reason to restrain costs, improve the quality of services, or innovate.
• Absence of Losses. Poorly performing federal agencies do not lose money or go bankrupt, so there is no built-in mechanism to end low-value activities and no automatic corrective to inefficiencies. Government activities tend to get entrenched, and resources go to obsolete uses.
• Monopoly. Many federal activities are monopolies, which further reduces incentives to restrain costs and improve quality.
• Measurement. Business output can be measured by profits, revenues, and other metrics. But for the government, it is difficult to measure the quantity and quality of output, which makes it difficult for Congress to set goals for agency leaders or hold them accountable for results.
• Monitoring. Businesses produce audited financial statements, and their products are generally in the public realm. Shareholders, creditors, consumers, and competitors monitor companies in the marketplace. By contrast, government agencies can be secretive about their activities, and the public does not learn about agency dysfunctions until major scandals erupt.
• Rigid Compensation. Federal worker compensation is based on scales that are mainly tied to longevity, not performance. The rigid structure does not encourage high performance, and it reduces morale because good workers see poor workers being rewarded equally. The best workers have incentives to leave, while the poor workers stay decade after decade.
• Lack of Firing. Federal workers have strong civil service protections, and many have labor union protections. Until the Trump administration’s reforms in 2025, only about 0.5 percent of federal civilian workers a year were fired for any reason, including poor performance and misconduct. That rate is just one-sixth the private-sector firing rate.⁵⁷
• Red Tape. Federal operations are loaded with rules and regulations, which reduce efficiency. The government has huge power, so layers of rules are needed to safeguard against abuse. Also, because there is no profit goal in government, detailed rules are needed to monitor workers. As such, thick red tape is an unavoidable feature of the government.
• Bureaucratic Layering. American businesses have become leaner in recent decades, with fewer layers of management. By contrast, “layering has become very extreme” in the federal government, and agency workforces are top-heavy.⁵⁸ As a result, decisionmaking is slowed and it is difficult to hold anyone accountable for failures.
• Agency Capture. Federal agencies can get influenced or “captured” by special interests, such as businesses.⁵⁹ Interest groups may gain influence by providing benefits to federal employees, gaining control over information that agencies need, or by using their relationships with legislators who oversee the agencies. Interest group influence also stems from officials gaining lucrative private-sector jobs after leaving government.

Congress should pursue reforms to improve agency performance, such as by increasing removals of poor workers. But such reforms will not solve deep structural problems. The government will always fall short of competitive private markets on efficiency and innovation.

Congress may wish that programs are executed in an effective manner by expert civil servants, but experience shows that agencies often do not work that way. Members of Congress should temper their enthusiasm for launching programs because there will always be bureaucratic failures and inefficiencies.

When Congress appropriates money for programs, some of it leaks away on federal, state, and local administrative costs. To manage programs, government agencies incur costs for worker salaries and pensions, office space, and equipment. They also incur costs for overhead activities such as finance, human resources, and legal departments.

Administration costs vary widely depending on program complexity. Within the Social Security Administration, the relatively straightforward retirement program spends just 0.4 percent of its budget on administration.⁶⁰ By contrast, the agency’s more complex Supplemental Security Income program spends about 12 percent of its budget on administration.⁶¹

Let’s look at some other programs. With the Department of Commerce, the Economic Development Administration spends about $1 billion a year subsidizing businesses, but about 8 percent of the spending is consumed by federal administration.⁶² The Minority Business Development Agency spends about $80 million a year subsidizing businesses, but about 13 percent of the spending is consumed by federal administration.⁶³

The Department of Agriculture’s largest farm subsidy program is crop insurance, valued at about $15 billion a year. It is administered by 12 insurance firms that are extensions of the federal bureaucracy. The CBO projects that farmers will receive payments net of their premiums of $9.6 billion in 2026.⁶⁴ The insurance firms get paid $2.4 billion for administration costs and $2.7 billion for underwriting gains. Thus, the total taxpayer cost of this program aimed at subsidizing farmers will be $14.7 billion, but $5.1 billion (35 percent) will go to the insurance companies. That is a huge overhead cost for delivering a subsidy, and 2026 will be a typical year according to the CBO projections.

How much excess bloat is there in federal administration? We do not know exactly, but over the years there has been a “steady thickening of the federal bureaucracy as Congress and presidents have added layer upon layer of political and career management to the leadership hierarchy,” reported bureaucracy expert Paul Light.⁶⁵ In 2016, he found that “John F. Kennedy oversaw 450 political and career executives who occupied 17 bureaucratic layers at the top of government. Mr. Trump will soon oversee more than 3,000 executives in 63 layers.”⁶⁶

The number of federal employees grew during President Trump’s first term in office and in Joe Biden’s term in office. When Trump returned to office in 2025, he appointed Elon Musk to head a Department of Government Efficiency (DOGE) to try and reduce overstaffing and improve efficiency with new technologies. In less than a year, the efforts reduced the number of federal civilian workers by 9 percent.⁶⁷

Sometimes the DOGE cuts went too far and were reversed, but other times efficiencies were found. One effort was the redesign of the Free Application for Federal Student Aid. Despite employee cuts at the Department of Education, the new college aid tool launched successfully in 2025. The new tool simplified forms and reduced headaches for families.⁶⁸ Thus, it is possible to have both leaner and more-effective federal agencies.

However, much of the cost of administering federal programs is not within federal agencies, but rather within state and local agencies. More than $1.1 trillion in federal spending is carried out through more than 1,300 grant programs for the states.⁶⁹ Federal officials allocate the grants and impose regulations on the states regarding how to spend those funds. State and local administrators must learn the federal rules, draft plans and procedures, file federal reports, request waivers, audit program recipients, and perform other bureaucratic functions.

There are fewer than 3 million civilian federal employees, but there are 17 million state and local employees.⁷⁰ Some portion of those 17 million employees handle the paperwork created by top-down federal programs for education, welfare, transportation, and other activities. These activities could be funded by the states themselves and operated without any federal bureaucracy or top-down rules. The states, for example, could fund their own K–12 public schools without a federal Department of Education. Thus, the costs of the department and related top-down rules are waste or leakage compared to the states funding their own schools.

How large is the administrative waste created by federal grants to the states? Light estimated that there are about 1.5 million employees in state and local governments supported by federal grants and that about 4.6 million state and local government jobs exist to carry out federal mandates—that is, the rules tied to federal grants and other federal regulations.⁷¹ We do not know how many of these jobs exist just to handle federal grants.

Suppose that 1 million of the 17 million state and local workers just handle federal grants and related regulations, and that they earn the average compensation of state and local workers. That would suggest an annual cost of $110 billion, which is 10 percent of the value of $1.1 trillion in annual grants.⁷² That 10 percent is a rough conservative guess for the additional cost of funding programs federally rather than locally.

Let’s look at some specific examples. In Iowa, officials estimated that if they received the state’s $151 million a year of K–12 grants as a block grant with minimal federal rules, they would save $29 million a year, or 19 percent of total costs.⁷³ In Wisconsin, a study found that state-level bureaucracy consumed $54 million of the $878 million in federal K–12 grants it received—thus as federal money flowed down to the schools, 6 percent was consumed by the state-level middlemen.⁷⁴

Consider the Department of Housing and Urban Development (HUD) programs. The department gives states community development grants of $20 billion a year for items such as roads and housing.⁷⁵ The Government Accountability Office (GAO) found that, on average, states spent 17 percent of these grants on administration.⁷⁶ This is waste or leakage compared to local governments funding their own roads and housing.

The HUD grants for housing vouchers generate bureaucracy in the 2,000 local public housing agencies that administer the program. At least 8 percent of the program’s budget is burned on local administrative costs before aid reaches tenants.⁷⁷ HUD itself spent $2.7 billion in 2025 on federal employee compensation and office costs—costs that would not be needed if the states handled their own housing policies.⁷⁸

Let’s look at one more program: SNAP. It delivered $94 billion in benefits in 2024.⁷⁹ The federal government’s administrative costs were about $1 billion, and state administrative costs were $10.8 billion.⁸⁰ Total administrative costs were thus 10 percent of the program’s budget.

A GAO study looked at administration costs across seven major aid-to-state social programs. The average costs were 18 percent of spending, but there were large differences between the programs and how the costs were measured.⁸¹ Similarly, a study by Julia Isaacs of the Brookings Institution examined administration costs across nine major federal welfare programs, including SNAP, Medicaid, school lunches, housing benefits, and five others. The costs as a percentage of program benefits varied substantially but averaged 12.6 percent.⁸²

In sum, administration costs are difficult to measure and vary between programs. But the costs are not zero and are often 10 percent or more. We also know that extra costs are imposed when the federal government funds state and local programs rather than allowing the states to fund their own programs. Government administration costs are leaks of value that go to bureaucracies, not to the intended beneficiaries of programs.

After taxes are extracted and funds have flowed through the government, more value is lost when spending goes to recipients. Benefits and subsidies often distort the behavior of recipients, which creates deadweight losses. Spending can also displace private-sector efforts, create unjustified windfalls, and go to ineligible recipients through error and fraud. On top of these leaks, spending imposes compliance burdens on recipients and prompts wasteful lobbying efforts.

Of course, not all federal spending is harmful or wasteful. Spending is worthwhile if it fixes failures in markets and generates benefits that are higher than the costs. The government, for example, must spend on national defense activities because these would be underprovided in markets.

However, before policymakers spend, they should be sure that planned programs fill true voids that the states and the private sector cannot fill. And they should consider whether individuals and businesses will respond in ways that undermine any benefits that are created.

When the government spends on benefits and subsidies, the recipients may change their behavior in unproductive ways. They may alter their working, saving, investment, consumption, and entrepreneurial activities. Such responses may generate deadweight losses, akin to the deadweight losses caused by tax increases.

Social welfare programs, for example, can induce the unemployed to rely on government benefits rather than to seek active work. Former senator Daniel Patrick Moynihan of New York said, “It cannot too often be stated that the issue of welfare is not what it costs those who provide it, but what it costs those who receive it.”⁸³

However, we think that it is both the taxing and spending sides of programs that impose costs. Higher taxes to fund welfare programs may induce taxpayers to work less, while the welfare payments themselves may induce recipients to work less. Spending may aid low-income individuals, but at the cost of reducing incentives for them to earn their way out of poverty.

Welfare programs are often designed with benefits that decline as income rises, which creates a tax-like effect discouraging additional work. If the recipients of Medicaid, food stamps, the EITC, and other programs earn more income, they receive fewer benefits. When this effect is combined with income and payroll taxes, earners just above the poverty level can face very high effective marginal tax rates—that is, tax rates on additional earnings.

In a 2019 study, Nina Chien and Suzanne Macartney found that for households with children, marginal tax rates averaged above 40 percent for incomes between 100 and 200 percent of the poverty level.⁸⁴ In a 2023 study, David Altig and coauthors found that among the lowest-income quintile of households, one in four face lifetime marginal tax rates above 50 percent.⁸⁵ In a 2012 study, Elaine Maag and coauthors examined marginal tax rates on a hypothetical low-income single parent with two children and found that the parent would typically face a marginal tax rate of about 50 percent moving from poverty level to twice the poverty level.⁸⁶ In states without Medicaid expansion, earnings that push households just above the poverty level can mean the loss of health coverage, which can result in a 100 percent marginal tax rate if extra earnings are outweighed by the lost Medicaid.

Such high marginal tax rates reduce incentives to work more hours, take promotions, or invest in training to increase future earnings. These responses to government programs are deadweight losses, and are thus leaks from the benefits that the programs may create.

Corporate welfare spending also creates deadweight losses. Farm subsidies, for example, prompt farmers to overproduce subsidized crops and underproduce more valuable crops. These planting changes cause losses to the economy because the overall value of farm production is reduced.

Figure 3 illustrates the losses created by a corn subsidy for farmers. In an unsubsidized market, people buy 100 million ears of corn at 50 cents each. Since markets are voluntary, we know that customers value those ears at 50 cents or more and that the cost of growing the ears is 50 cents or less. Now suppose the government subsidizes farmers 10 cents per ear. Farmers would grow more corn and reduce their investments in other crops. In the figure, the additional ears would cost more to produce than 50 cents, but they would be valued by consumers at less than 50 cents. The subsidy has destroyed value by generating production that costs more than it is worth. This deadweight loss is the black triangle in the figure.

We could make similar diagrams for other spending programs. Congress intends to help individuals and businesses, but it enacts programs that undermine the economy by changing behaviors in markets. Other examples include:

• Unemployment benefits encourage people to reduce job searching.
• Disability benefits encourage people who could work to drop out of the labor force.
• Social Security and Medicare discourage personal saving for retirement.
• Medicare’s payment structure induces the provision of unnecessary services and promotes the substitution of service quantity for service quality.
• Irrigation subsidies cause the overconsumption of water in the arid Western states.

Spending programs can create collateral damage that ripples outward from the initial recipients. Let’s look at farm subsidies again. The direct effect is to increase the output of subsidized crops. But a secondary effect is to push up the demand for cropland, which causes less-fertile lands to be brought into production. In turn, that prompts the greater use of fertilizers and pesticides, which can have negative environmental effects.

With federal health care spending, the subsidies reduce out-of-pocket costs, which encourages the overuse of low-value care that may deliver little health improvement and generate benefits that are less than the costs. Low-value care consumes real resources that cannot be used elsewhere in the health care system and thus generates deadweight losses.

Some federal programs encourage excessively risky actions that people would not take without the expectation of a bailout if things go wrong. When the government shields people from the costs of risky decisions, they take too much risk, which is an inefficient outcome called “moral hazard.”

A good example is federal flood insurance subsidies.⁸⁷ These subsidies have encouraged development in low-lying areas along the East Coast, Gulf Coast, and flood-prone rivers. Private insurance companies would not insure people for homes built in excessively dangerous locations, but the federal program has notoriously paid for homeowners to rebuild their homes in places that have been repeatedly hit by storms.

A related problem is that the federal government often comes in after floods, wildfires, and other natural disasters to aid people living in dangerous places who do not have private insurance. The urge to help people in distress is understandable, but when people expect bailouts from Washington, it generates imprudent decisions that impose costs on society down the road.

Student loan programs create moral hazard. When borrowers expect government forgiveness or income-based repayment, they may borrow excessively, pursue less-useful college degrees, and choose pricier institutions. The subsidies lead to inflated tuition, higher public costs, and lower returns on college degrees. These wasted resources are deadweight losses.

As government programs extend into more activities in society, it creates an overhanging moral hazard whereby businesses and individuals take more risks, believing that the government will bail them out. The huge subsidies handed out during the Great Recession and COVID-19 pandemic, for example, set high expectations for federal relief during the next crisis. Those expectations may induce fewer people to plan ahead and save, leaving them more in need of government aid down the road.

When the government expands spending programs, it imposes a less visible and often forgotten cost called “crowding out.” This is the displacement of private markets and charitable activities by the growing government. Let’s look at crowding out in three policy areas: social welfare, infrastructure, and arts funding.

When the federal government created the unemployment insurance program in 1935, it displaced unemployment programs that were operated by labor unions and other private organizations.⁸⁸ The introduction of new aid programs in the 1930s crowded out charitable giving by churches. One study found that church giving fell by 30 percent in response to the New Deal.⁸⁹

When the government created Medicare in 1965, it displaced private health care plans that covered about half of elderly Americans at the time.⁹⁰ And as the government has expanded Medicaid over the years, it has displaced roughly one person with private health care insurance for every two people added to the program.⁹¹ The provision of Medicaid long-term care coverage has induced families who would have paid for care to take the government benefits, often by manipulating their finances to qualify.⁹²

Economist Milton Friedman noted, “One of the major costs of the extension of governmental welfare activities has been the corresponding decline in private charitable activities.”⁹³ Before the modern welfare state was developed, America hosted a vast range of church-based organizations, mutual-aid groups, fraternal societies, and other charitable groups that focused on social needs.⁹⁴ One study found that when government grants to nonprofit housing and human services groups increase, for example, private donations to the groups fall by about half the amount.⁹⁵

Another casualty of social welfare expansion is a decline in individual responsibility. “State help kills self-help”—when you reduce individual responsibility, people are less responsible.⁹⁶ Without government unemployment insurance, for example, working Americans would have a strong incentive to save for a possible job loss. Without the enactment of Social Security, Americans would have developed a stronger culture of personal saving to build nest eggs to cover retirement.⁹⁷ And without Social Security’s onerous 12.4 percent wage tax, they would have more income to save for retirement.

The rise of government infrastructure spending has crowded out private infrastructure spending. A century ago, most transportation infrastructure was privately funded, but today much is government funded. Government spending on highways, airports, urban transit systems, passenger rail, and other facilities has displaced private investment.

For instance, in the first few decades of commercial aviation, many major US airports were privately owned and operated, including the main airports in Los Angeles, Miami, Philadelphia, and Washington, DC. These were for-profit facilities that were successful and innovative. However, they lost ground over time from unfair government competition. Government airports received subsidies, could issue tax-free bonds, and did not have to pay taxes. Today, all major US airports are owned by governments and are subsidized, but experience abroad shows that private airports can thrive.⁹⁸

Likewise with urban transit: A century ago, nearly all city bus and rail systems were privately owned and operated. But in the 1960s, the federal government began subsidizing government-owned transit systems, and the private share of systems fell rapidly.⁹⁹ In the 100 largest US cities, the private share of urban rail systems fell from 90 percent in 1960 to just 20 percent by the late 1970s.¹⁰⁰

Highway bridges faced a similar fate: In 1932, nearly two-thirds of the 322 toll bridges in the United States were privately owned.¹⁰¹ Then federal and state governments began providing subsidies to government-owned bridges, many of which were “free” to motorists, and private toll bridges began to disappear.

What is the cost of crowding out in infrastructure? We end up with facilities that are more costly, less efficient, and less innovative. Businesses in open markets have strong incentives to continually improve products and reduce costs. By contrast, poorly performing federal infrastructure is rarely restructured. Federal infrastructure businesses, such as Amtrak and the US Postal Service, limp along year after year, needing subsidies and lagging on innovation. Poorly performing employees are rarely fired and there is no pressure for government entities to generate net value for the economy.

Also, when the government provides infrastructure, it squelches competition. This is a problem because major innovations usually come from start-up businesses, not established businesses. FedEx pioneered overnight letter delivery, for example, not the Postal Service. Economic progress depends on open entry—on the ability of entrepreneurs to challenge existing providers. That is hard to do when the existing provider is the government.

Government crowding out is also a problem in the arts. One study found that the creation of the National Endowment for the Arts (NEA) in the 1960s coincided with a decrease in private arts donations.¹⁰² Another study found that when the NEA’s budget was cut in the mid 1990s, private arts contributions increased by 50 cents or more for every dollar of NEA cuts.¹⁰³ Analyst Jack Salmon discusses evidence that “increased government funding not only reduces private donations from donors, it also discourages nonprofits from raising donations from non-government sources.”¹⁰⁴

In sum, the effects of government spending programs are more complicated than simply providing static benefits to the economy. Spending creates negative ripple effects as businesses, charities, and entrepreneurs are displaced by the growing government. A $100 spending program will add less value on net to society than that $100, because private activities and innovations are crowded out.

A share of spending from many programs goes to recipients who claim benefits legally but were not those that Congress was targeting. In some cases, programs confer windfalls on people who are already taking the actions that policymakers wanted to incentivize. In other cases, programs for poor households end up benefiting higher-income households. Congress is imprecise in program targeting, which produces leaks from the federal spending bucket.

Let’s look at windfalls. Until 2025, Congress sought to encourage purchases of electric vehicles (EV) with generous tax credits. However, research found that 70 percent or more of the credit recipients would have purchased an EV without the credits.¹⁰⁵ Congress had wanted the credits to go to people at the margin of buying, but most of the benefits went to inframarginal buyers, and thus the subsidies were mainly a waste. There is a similar problem with rooftop solar panel subsidies.¹⁰⁶

The 2009 Cash for Clunkers program provided subsidies for purchasing fuel-efficient vehicles when people traded in older, less-efficient ones. But a portion of the subsidies simply induced people to advance their already planned purchases. One study found that 45 percent of the subsidies went to consumers who would have purchased a new car anyway.¹⁰⁷ That portion of the spending was a leak from the aimed-at goal.

Federal business subsidies—even if they are justified—face the same dilemma. The government spends billions of dollars a year on subsidies to increase US investment in broadband, energy facilities, semiconductor plants, and other industrial infrastructure.¹⁰⁸ But a portion of the spending goes to companies that would have invested even without the subsidies. Indeed, as soon as the government announces subsidies for an industry, companies have a strong incentive to delay planned investments and lobby for handouts.

In theory, Congress could devise better ways to target the margin to boost incremental investment, but that is difficult to achieve because policies are blunt and the economy is complex. The result is that many subsidy dollars do not move the needle on the stated goals of Congress.

Federal anti-poverty programs illustrate what can be thought of as either windfall spending or poor targeting. Many programs enacted to aid the poor end up benefiting those with higher incomes, either because of poor program design or because of political realities in Congress.

The low-income housing tax credit (LIHTC) is a $14 billion federal program that provides tax breaks for building multifamily housing. Congress was trying to help low-income renters, but studies find that a large share of the benefits go to housing developers. A 2024 study by Evan Soltas, using data from LIHTC projects across 40 states, found that “households benefit from modest rent discounts on subsidized units, but developers capture around half of the subsidy in profits.”¹⁰⁹ So half of the benefits of this anti-poverty program are a leak from the federal bucket.

The $35 billion federal school lunch program was enacted to help poor children, but over time it has grown to subsidize a large share of all children. Originally, lower-income children would receive free or reduced-price meals, while higher-income children would pay full price. But the share of all school lunches provided for free or at a reduced price has increased from 15 percent in 1969 to 72 percent by 2024.¹¹⁰ Policymakers have expanded coverage over time and made it easier to claim benefits. Arguably, school lunch subsidies going to nonpoor children are leaks from the government’s spending bucket.

Similarly, the $20 billion Community Development Block Grant (CDBG) was launched in the 1970s to help the poorest urban areas with funding for basic services. However, every member of Congress has wanted a share of CDBG spending, and so the funds are spread widely to many nonpoor areas. The number of “entitlement communities” that receive funding has doubled since 1975 as affluent areas—such as Falls Church, Virginia; Martha’s Vineyard, Massachusetts; and Stamford, Connecticut—have been added to the program.¹¹¹

The GAO found that the CDBG suffered from poor performance in targeting aid to needy communities.¹¹² A Pew Charitable Trust study found that CDBG and other programs “designed to direct economic activity to struggling areas frequently end up benefiting wealthier communities.”¹¹³ In a 2024 study, federal experts concluded that there has been a “long tradition of research that examines the CDBG formula and finds that it is poorly targeted to community development needs.”¹¹⁴

Yet another program that was created to help poor areas but now funds wealthy areas is the Department of Education’s Title I. This $21 billion program was supposed to provide grants to states for educating the disadvantaged. But Congress broadened the scope over time, and today Title I spending goes to more than 60 percent of all public schools, including schools in very wealthy jurisdictions, such as Arlington and Alexandria, Virginia.¹¹⁵

Finally, consider the $100 billion SNAP program. It was originally targeted at households with incomes near and below the poverty level. It also excluded people who had low incomes but who also had substantial assets. Over time, many states raised the income limits and eliminated the asset test. By 2024, about half the states provided benefits to households with incomes up to 200 percent of the poverty level, and the benefits could even go to households with millionaire levels of wealth.¹¹⁶

If policymakers expand anti-poverty programs to cover the nonpoor, is that spending a wasteful leak? We think so, but it is debatable. Certainly, there is a bait-and-switch game going on. Supporters in Congress and lobby groups frame programs such as school lunches and SNAP as helping the hungry, needy, and destitute. As such, the public may not be aware that part of the spending goes to well-off households who do not need it.

The purpose of identifying leaks in federal spending is to guide policymakers toward making the government leaner and more efficient. Given that the government faces large budget deficits, we think that welfare spending going to middle- and higher-income households is a large leak that should be closed.

Federal spending on benefits, subsidies, and procurement suffers from errors, fraud, and other types of improper payments. The problems stem from complex program rules, people and businesses falsifying paperwork, and criminal organizations targeting vulnerable programs.

All federal programs suffer from improper payments to some degree. Improper payments are spending leaks because the funds do not go to the intended recipients. The GAO has found that social programs—including food stamps, school lunches, disability benefits, health benefits, and the EITC—suffer from improper payments ranging from a few percent of benefits paid to 25 percent or more.¹¹⁷ The federal leaky bucket spills many billions of dollars from improper payments, despite policymakers often promising to crack down on them.

The GAO estimates that improper payments totaled $162 billion in 2024 across 68 federal programs.¹¹⁸ That is a lot of waste, but that figure only covers a portion of federal spending and does “not include certain programs that are susceptible to significant improper payments.”¹¹⁹ The estimate, for example, excludes the huge Departments of Defense and Homeland Security. The GAO estimated that for the entire federal government, the fraud portion of improper payments may be as high as $521 billion a year.¹²⁰

The government could reduce improper payments by imposing more paperwork and more auditing on recipients. That would be a good reform, although it would increase administrative burdens. More paperwork and auditing also conflict with the desire of policymakers to make programs easy for constituents to access. With the school lunch program, for example, many ineligible households receive benefits because administrators check very few applications for accuracy.¹²¹

The largest improper payments flow from Medicare and Medicaid, which together lose about $85 billion a year. Medicare pays more than 1.1 billion claims a year to more than 1.5 million doctors and hospitals, generally without any human eyes checking those payments for accuracy.¹²² Those payments are a target for dishonest health providers, who bill for services not rendered, misrepresent the services provided, bill for unnecessary services, duplicate their billing, and falsify cost reports.

Farm subsidy programs, which cost taxpayers more than $30 billion a year, are subject to substantial errors and fraud. The improper payment rates vary by program: 3 percent for crop insurance, 13 percent for commodity payments, 13 percent for the Livestock Forage Disaster Program, 12 percent for the Noninsured Crop Disaster Assistance Program, and 45 percent for the Emergency Conservation Program.¹²³ It can be difficult for officials to verify the amount of crops that are subject to loss claims, so farmers exaggerate to boost their benefit payments.¹²⁴

Refundable tax credits, which are spending programs, have chronically high improper payment rates. The rate for the EITC, which costs more than $60 billion a year, has been more than 20 percent for decades and was 27 percent in 2024.¹²⁵ The improper payment rate for the refundable American Opportunity Tax Credit is also high, at 28 percent.¹²⁶

There are large errors and fraud in the Affordable Care Act premium tax credits, which cost more than $100 billion a year. The GAO puts the improper payment rate for this program at 28 percent.¹²⁷ In 2025, 2.3 million people falsely misstated their incomes on application forms to boost their premium credits.¹²⁸

The government suffered huge fraud losses on COVID-19 relief programs. The Small Business Administration handed out $1.2 trillion in Economic Injury Disaster Loans (EIDLs) and Paycheck Protection Program (PPP) loans. The agency’s inspector general found that about $200 billion appears to have been stolen and that “at least 17 percent of all COVID-19 EIDL and PPP funds were disbursed to potentially fraudulent actors.”¹²⁹

The pandemic-era Employee Retention Tax Credit was also hit by losses. The IRS found that between 10 and 20 percent of $86 billion in claims were in the “highest-risk group, which show clear signs of being erroneous claims.”¹³⁰ The GAO estimates that $43 billion—or 35 percent—of pandemic unemployment insurance outlays were improperly paid out.¹³¹

Federal aid-to-state programs suffer from substantial fraud because state administrators have little incentive to spend money frugally that comes “free” from Washington. Recent scandals in Minnesota drive the point home. In the Feeding Our Future scam, more than 50 people were convicted of crimes related to stealing $250 million in federal aid for feeding poor children.¹³² Similar scams have come to light in Minnesota’s health care and subsidized daycare programs. These programs combined federal funding with lax state administration.

A growing problem is that criminals systematically attack programs by submitting false electronic claims that government computers pay automatically. In one scam during the pandemic, a California prison inmate and friends submitted hundreds of false claims for the Employee Retention Tax Credit and received $550 million in payouts from the IRS before being caught.¹³³

Similarly, thieves are stealing from the SNAP program electronically.¹³⁴ They are installing card skimmers on terminals at grocery stores and stealing benefits from dozens of recipients before stores even notice a problem.¹³⁵ In one recent scam, a federal employee worked with outside accomplices to steal $36 million from the SNAP program.¹³⁶

Federal procurement spending suffers from fraud and contractor overcharges.¹³⁷ The government buys more than $750 billion of goods and services a year, with the Pentagon accounting for 60 percent of the total.¹³⁸ In one fraud case in 2024, Raytheon was found to have overcharged the government more than $100 million on various weapons systems.¹³⁹ In a case in 2019, the Pentagon paid contractor TransDigm “$1,443 for a three-inch ring called a ‘non-vehicular clutch disk’ which is used in the C‑135 transport aircraft, though it cost the company just $32 to produce.”¹⁴⁰

In a tawdry case of contractor abuse a few years ago, Leonard Glenn Francis cozied up to US Navy leaders to win hundreds of millions of dollars in contracts to resupply ships. He wined and dined naval officers and provided them with cash, gifts, and prostitutes. The scandal exposed “a staggering degree of corruption within the Navy,” noted the Washington Post.¹⁴¹

A six-month CBS News investigation in 2023 looked at defense contracting. The findings were hard-hitting: “Military contractors overcharge the Pentagon on almost everything the Department of Defense buys each year, experts told 60 Minutes.”¹⁴² One of the contributing structural causes is that there are very few suppliers—sometimes only one—of many major weapons systems, and thus there is little competition for contracts.

Procurement programs often cost taxpayers more than promised because of cost overruns.¹⁴³ A 2024 report found that costs for the Sentinel intercontinental ballistic missile system have jumped 81 percent since 2020, to $141 billion.¹⁴⁴ A 2025 report found that 14 of 24 major recent information technology projects at the Pentagon had cost overruns ranging from $6 million to $815 million, with a median increase of $173 million.¹⁴⁵ In past years, the International Space Station famously quadrupled in cost and Boston’s federally funded Big Dig highway project quintupled in cost.¹⁴⁶ Federally funded projects for Veteran’s Administration hospitals, energy projects, and light-rail projects often go far over budget.

A group of scholars recently created a database of nearly 100 million federal contracts awarded over the past 45 years.¹⁴⁷ They noted that federal systems that are “so large and lucrative are vulnerable to grift and corruption. As many as 44% of contracts receive only a single bid, and bidders or potential bidders may collude to maximize their profits. Politicians also have incentives to corrupt the procurement process: securing grants for constituent firms or copartisan entrepreneurs may be rewarded with votes or campaign contributions.”¹⁴⁸

The government proceeds with large projects based on estimated costs, but after projects get underway contractors often revise costs upward. The true final costs of projects determine whether they would pass a cost-benefit test and thus benefit society on net. If policymakers enact projects based on lowballed costs, then they will be making bad spending decisions in many cases.

In sum, federal program benefits are lost through error, fraud, and contractor overcharges. The problems persist year after year, despite lawmakers often promising to fix them. Economist Milton Friedman said, “Nobody spends somebody else’s money as carefully as he spends his own,” and that is certainly true when members of Congress are spending our money.¹⁴⁹

Recipients of benefits, subsidies, and contracts gain from federal programs, but the programs also cost them time and money on compliance. They need to learn program rules, fill out applications, submit grant proposals and invoices, and interact with bureaucracies. Individuals, nonprofit groups, and businesses spend billions of hours every year dealing with government paperwork. These costs are not recorded in agency budgets, but they are costs to society.

The White House Office of Information and Regulatory Affairs estimates that the annual paperwork burden imposed by federal agencies on the public is 11.4 billion hours, but excluding tax paperwork the total is 4.4 billion hours.¹⁵⁰ Given that US average hourly compensation in 2024 was $52 per hour, the cost of those 4.4 billion hours wasted on paperwork can be estimated as $229 billion.¹⁵¹

The Office of Management and Budget data include the time spent reading instructions, filling out forms, and maintaining records, but it does not include some of the broader costs of dealing with the government, such as legal actions, planning, and overhead. Hospitals, businesses, charities, colleges, and other groups that receive federal aid must dedicate large numbers of staff to managing federal grants and contracts. Compliance burdens can be relatively heavier for small organizations.

Consider the large paperwork burden imposed by federal scientific research grants. The National Institutes of Health awarded $33 billion in grants in 2024, and $9.3 billion of the funding was consumed by indirect costs, such as administration and overhead, which was 28 percent of the total.¹⁵² For the National Science Foundation, nearly 20 percent of its funding for research goes toward administration and overhead. Many scientists have complained about how the intense bureaucracy of federal grants takes away from research time.¹⁵³

Medicare and Medicaid impose large paperwork burdens on hospitals and doctors, but private health care funding mechanisms are also complex. A 2019 study found that “about one-quarter of total health care spending in the U.S. is waste, with a price tag ranging from $760 billion to $935 billion.”¹⁵⁴ The “waste” in this study was a very broad category, but the authors estimated that $265 billion of the total was “administrative complexity.”¹⁵⁵

The EITC costs more than $60 billion a year and has more than 20 million recipients.¹⁵⁶ The credit’s complexity has induced half or more of the recipients to use outside tax preparers for filing.¹⁵⁷ One study found that “those eligible for the EITC, who are typically low-income workers with children, would spend between 13 and 22 percent of their refund this year at local tax preparation outlets.”¹⁵⁸ Without the complex EITC, low earners would be more likely to file taxes on their own without having to pay for help.

Some federal compliance costs are unavoidable and there are tradeoffs to consider. For example, some programs impose detailed compliance rules on recipients (which increases costs) to reduce program fraud (which reduces costs). However, Congress often makes programs unnecessarily complex as it tries to micromanage recipient behavior and fine-tune benefits, which is the problem with the EITC.

In sum, compliance expenses are a cost not reflected in the federal budget. Recipients may be better off by taking government payments, but the related paperwork burdens reduce the net benefits and undermine productivity in the economy.

When the federal government enacts benefits, subsidies, and procurements, it generates a gold rush of lobbying by businesses and advocacy groups. Lobbying consumes wealth in the economy that could otherwise be used for creating useful goods and services. Ongoing lobbying on many issues also adds uncertainty to the business environment.

It is true that lobbying can provide useful information to policymakers and is a protected First Amendment right. It is also true that a share of Washington lobbying is defensive in nature: individuals and businesses protecting themselves against unwarranted government actions. Nonetheless, the federal government spends on many activities that it should not be spending on, and the related lobbying costs are a source of waste.

Lobbying has made the economy of the Washington region boom. The region includes three of the top five highest-income counties in the nation.¹⁵⁹ Part of that regional wealth represents leakage from the federal spending bucket. If a defense firm spends millions of dollars lobbying for weapons systems, that business cost will be ultimately paid by taxpayers as the firm wins contracts.

There are more than 13,000 registered lobbyists in Washington, whose activities cost about $5 billion a year.¹⁶⁰ Lobbying tends to track major legislation; it spiked around 2009 with the passage of the stimulus bill and Affordable Care Act, and it spiked in 2025 as President Trump’s major tax bill went through Congress.

Lobbying rose in early 2020 as the government raised spending in response to the COVID-19 pandemic.¹⁶¹ The major airlines, for example, lobbied and won $54 billion in federal bailout payments.¹⁶² OpenSecrets reported a spike in new DC lobby clients when the pandemic hit in order to “influence the government’s response to the virus—and to access its massive pool of stimulus money.”¹⁶³

In recent years, the adoption of industrial policies and protectionist tariffs has boosted business lobbying. Lobbying by steel producers, for example, was up 69 percent in 2024 and another 15 percent in 2025.¹⁶⁴ Nippon Steel lobbying soared in 2024 and 2025 as Presidents Joe Biden and Trump considered blocking its U.S. Steel acquisition.

When businesses increase lobbying in Washington, they divert their focus from innovation in the marketplace. Lobbying activities replace the hard work of making better products. Lobbying can steer Congress away from making good choices for the nation, while simultaneously steering CEOs away from making good choices for their companies.

Intel Corporation spends about $15 million a year on lobbying and won $8 billion in subsidies from the 2022 CHIPS and Science Act.¹⁶⁵ But the subsidies have not prevented the company from performing poorly in the marketplace.¹⁶⁶ Similarly, US auto producers hired 700 lobbyists to win billions of dollars of green subsidies under President Biden, including for electric vehicles.¹⁶⁷ But now the same producers are losing billions of dollars for ignoring their customers and investing too heavily in EVs.¹⁶⁸

The registered lobbyist industry, at $5 billion a year, is only a part of the broader DC advocacy industry. There are more than 2,300 trade associations in Washington and thousands of labor unions, think tanks, law firms, public relations firms, and other organizations that are all trying to influence federal policies.¹⁶⁹

The National Education Association advocates for higher federal spending on K–12 education. The DC-based association has 17 registered lobbyists, but that represents just a fraction of its overall staff of about 700.¹⁷⁰ The NEA’s annual budget is more than $430 million.¹⁷¹ If you think that K–12 education should be fully handled by the states and not the federal government, then you may think that the NEA’s DC spending is a wasteful cost on society.

The NEA is just one of hundreds of organizations that push Congress to increase federal spending on aid to the states, which currently totals more than $1 trillion a year. The National WIC Association—with a $7 million budget—pushes to increase spending on the Special Supplemental Nutrition Program for Women, Infants, and Children program. The Food Research & Action Center—with an $18 million budget—is one of many groups that push to increase spending on the SNAP program.

Federal highway spending is pushed by a large number of trade associations in DC, including the National Stone, Sand and Gravel Association; National Asphalt Pavement Association; Association of Equipment Manufacturers; and American Road and Transportation Builders Association.¹⁷² The last group has a budget of $40 million.

Like the NEA, these highway groups are in DC because what were previously state spending activities have been duplicated at the federal level. Such groups lobby in DC as well as the 50 state capitals. As the federal intrusion into state activities has grown, it has generated additional wasteful federal lobbying.

In sum, DC advocacy activities burn real resources that could be used to produce useful goods and services. There are many highly skilled professionals in Washington who advocate and lobby to boost federal spending on activities that distort the economy and duplicate state spending. Depending on one’s view of federalism and the proper role of the federal government, some share of these DC advocacy and lobbying activities are a waste to society.

The federal government has always had a leaky bucket. Federal taxes have disrupted the economy since at least the 1790s, when new taxes on alcohol were so damaging that they prompted the Whiskey Rebellion. Federal mismanagement goes back at least to the government’s inefficient Indian trading posts of the early 1800s.¹⁷³ And wasteful pork barrel spending goes back at least to the river and harbor bills of the mid 1800s.¹⁷⁴

That said, the federal spending bucket has become leakier over time as the government has grown larger. Federal spending has grown from 4 percent of the gross domestic product in 1930 to 23 percent today. That growth has produced higher taxes, more mismanagement, and more programs with low benefits compared to the costs. Larger government invariably creates larger leaks for numerous reasons.

We discussed how taxes cause deadweight losses. We noted that each added dollar of federal income taxes creates roughly 40 cents in deadweight losses. That estimate is for a government of about today’s size.

However, deadweight losses rise rapidly—more than proportionately—as the government grows larger and raises tax rates. Economist Greg Mankiw explains: “It is a standard proposition in economics that the deadweight loss of a tax rises approximately with the square of the tax rate.… If we double the size of a tax, the deadweight loss increases four-fold; if we triple the size of the tax, the deadweight loss increases nine-fold.”¹⁷⁵ Thus, a 40 percent tax rate is four times more damaging than a 20 percent tax rate.

Since combined federal-state income tax rates are already high, new spending faces a large hurdle for it to make economic sense because of the rising cost of funding it. It is true that a lot of federal spending is currently financed by debt, but that debt accumulates interest charges and will impose even larger burdens on taxpayers in the future.

Interestingly, the rising marginal cost of taxes means that new programs that might have made sense when the government was smaller no longer make sense now that the government is larger. As the government grows, it is more likely that the benefits of new spending programs will fall short of the tax costs.

As the government grows larger, each new spending item is likely to create less value. If the Air Force adds a fighter jet, the marginal benefit to national security will be less than the benefits of the jets it already has. If education spending grows, each added dollar produces less benefit than the last. If food stamps are expanded from 42 to 43 million people, those added recipients are less likely to be in need than the first ones.

Legislators often do not seem to appreciate this idea of declining marginal value. They say things like “education funding helps students” and “defense spending protects the nation.” They confuse the average value of all the current spending with the marginal value of the last dollar spent. The marginal value is lower because we already spend a lot on these activities.

Declining marginal value also occurs as the scope of the government expands. Each new spending activity is more distant from the federal government’s core functions, and thus it likely generates fewer benefits beyond what markets could generate. Historically, the federal government focused on constitutional functions, such as national defense, but over time newer federal activities have been more likely to be duplicative of state, local, and private activities.

Why do policymakers support continued federal expansion despite the declining marginal value of its activities? Perhaps because of the “halo effect” of government. People regard the government’s core functions as so crucial that it creates a positive halo over government in general. The government is powerful, so people assume that it can solve many problems in society. But we have discussed how unlikely it would be for the federal government to find new spending activities where the benefits—net of all the leaks—would outweigh the tax costs.

The huge size of the federal government is overwhelming the ability of lawmakers to allocate resources efficiently. Consider that the federal budget of more than $7 trillion is 100 times larger than the average state government budget of about $70 billion. The federal government has many more employees, programs, contractors, and subsidy recipients to keep track of than any state government. Even if federal legislators spend their time diligently overseeing programs, the job is simply too large for them to do effectively.

The federal government is not just large in size, it is also sprawling in scope. In addition to handling core functions such as national defense, the government runs more than 2,400 subsidy and benefit programs, which is double the number as recently as the 1980s.¹⁷⁶ The government has spread its tentacles into many formerly state, local, and private activities, such as education, energy, welfare, housing, and urban transit.

Congress has neither the time nor expertise to allocate resources efficiently in all these areas. Members spend their time fundraising, securing earmarks for their districts, and giving speeches, but little time ensuring that programs are working well. Members are spread thin, which is evident from the fact that many of them routinely miss all or parts of hearings in their committees.

Members of Congress usually blame government failures on the executive branch, but they are the ones who fail to properly scrutinize programs. They grab power over nonfederal activities, but then they do not have time to monitor how their interventions are working. Congress does little oversight and rarely eliminates failed programs, and so the government becomes more wasteful over time.

The federal government is like a conglomerate corporation that is involved in too many activities. Markets force bloated corporations to shed their low-value activities, but there is no similar mechanism in the government. Milton Friedman said, “The tragedy is that because government is doing so many things it ought not to be doing, it performs the functions it ought to be performing badly.”¹⁷⁷

Leaders of federal agencies are also overwhelmed by the size of bureaucracies. Paul Light has argued that the “ever-thickening hierarchy” in federal agencies is causing more failures.¹⁷⁸ He says that “communication continues to be a major source of failure,” as “information has to flow up through multiple layers to reach the top of an agency.”¹⁷⁹

Another problem is that the more programs the government has, the more these work at cross purposes. Some programs keep food prices high, while others subsidize food costs. Some programs induce people to live in flood-prone areas, while others try to reduce flood risks. Some programs raise housing costs, while others subsidize housing costs. The government promotes breastfeeding but it also subsidizes baby formula. Federal programs subsidize health care and infrastructure, but federal regulations raise the costs of those activities.

In sum, as the government grows, rising tax rates suppress economic activity. As programs expand, the marginal value of spending declines. Policymakers become overwhelmed by all the programs and have little time to prune the ineffective ones. The more programs there are, the harder it is to allocate resources efficiently, and the more likely that programs will conflict. Larger government is leakier government.

Trying to solve society’s problems through the federal government is more costly and less efficient than many people realize. Spending programs require funding from taxes, which damage the economy. A portion of the funds raised is wasted through political misallocation, faulty program design, and bloated bureaucracies. When the remaining funds flow out to the private sector, they distort working and investing, and they generate fraud and other types of waste.

Only a fraction of the tax dollars raised for programs ends up delivering the outcomes that policymakers promise. How small a fraction? Former Council of Economic Advisers chair Michael Boskin described the leaky bucket problem and suggested that the benefits generated by inefficient programs may be just one-third of the costs:

Many federal programs are likely in this “inefficient” category. We discussed how some programs have improper payment rates of 10 percent or more. We noted that anti-poverty spending is often poorly targeted. We described how some programs consume 10 percent or more of funding on administration. And we described how value is lost on central planning errors, program design conflicts, and other failures. Some of these inefficiencies overlap, but they generally accumulate to reduce the benefits of each $1 in spending to less than $1.

Let’s say a $1 program has benefits of $0.40, as Boskin guesses for an inefficient program. With the program costing the private sector $1.40 in tax damage, the benefit-cost ratio would be just 29 percent. Or flip over the ratio and the cost-benefit is about three to one. Edgar Browning, emeritus economics professor at Texas A&M University, came to a similar conclusion in his book on fiscal policy, Stealing from Each Other. He estimated that “it costs taxpayers $3 to provide a benefit worth $1 to recipients.”¹⁸¹

A subsidy project in New York State illustrates the ratio. The state spent $959 million to build a solar panel manufacturing plant for Tesla in Buffalo, but the company and the state misjudged markets and the plant has sat partly idle. A state audit “found just 54 cents of economic benefit for every subsidy dollar spent on the factory.”¹⁸² Those low benefits, combined with the tax damage from funding the plant, creates a cost-benefit ratio of about three to one.

Policymakers try to avoid duds like the Tesla plant, but our study warns that programs have more inefficiencies than they may realize. Policymakers should rigorously vet programs and only approve those promising very high returns because leaks are inevitable. That is true for spending on everything from weapons systems to welfare programs.

Reviving federalism is the other way to tackle the leaks we’ve identified in this study. State governments are required to balance their budgets and thus have stronger incentives to minimize inefficiencies. They can also better target their programs to the needs of local communities. The states suffer program failures—such as the Buffalo Tesla factory—but failures in a single state are not imposed on the whole nation. Indeed, states can learn from failures elsewhere and pursue different strategies.

With the federal government running large deficits, policymakers should reexamine the costs and benefits of every program in the budget. Some programs are essential, and policymakers should focus on reducing the types of leaks we identified. But other programs are so leaky that the costs far outweigh the benefits. These programs are spilling resources all over and should be eliminated.

Edwards, Chris, and Ryan Bourne. “Federal Government Spending Is a Leaky Bucket,” Policy Analysis no. 1020, Cato Institute, Washington, DC, June 23, 2026.