What Is State Capacity? Does China Have It? Does America Really Lack It?
A crash course on the nature of state capacity
In 2024 and 2025, the United States built data centers at a pace that dwarfed China’s. Amazon, Microsoft, and Google poured hundreds of billions of dollars into facilities exceeding a million square feet. Some rose in Loudoun County, Virginia: liberal-leaning, densely populated, exactly the kind of place where common wisdom insists America can no longer build, well, anything!
The same America that supposedly can’t build housing or high-speed rail because of restrictive zoning, environmental reviews, and procedural gridlock built data centers at breakneck speed.
These projects faced real opposition. Data Center Watch documented 142 activist groups across 24 states organizing resistance. Local communities raised concerns about water consumption, energy strain, noise, property values. Over the past two years, $18 billion in data center projects were blocked or delayed, with another $46 billion facing regulatory challenges.
Yet as
points out in his substack, , about how most data centers still got built.Compare this to housing, where neighborhood opposition routinely kills projects. Or transit, where decade-long delays are normal. The contrast should make you suspicious.
Here’s what else got built despite nominal obstacles: Sports stadiums. Cities that can’t fix potholes somehow coordinate across multiple agencies to deliver billion-dollar stadiums with public financing. Highway expansions. The same DOTs that take a decade to plan transit lines fast-track highway widening. Tesla factories. Governor Gavin Newsom personally intervened to help Elon Musk navigate California’s regulatory maze to build factories that other manufacturers said were impossible.
The capacity isn’t gone. It’s selective.
For a decade, a “state capacity movement” has warned that America faces a governance crisis. They’re half right. Government struggles when serving diffuse public interests: people who want housing, riders who need transit, Medicaid patients seeking care. But government coordinates effectively when powerful private interests demand it: corporations with lobbying operations at all three levels of government, connected industries, wealthy individuals who can call governors directly.
That’s not a capacity crisis. That’s asymmetric capacity.
This article will show you what state capacity actually means, where it actually exists in America, and whose interests it actually serves.
The State Capacity Movement’s Federal Fixation
What We Talk About When We Talk About “State Capacity”
The term “state capacity” comes from development economics and political science. It means the ability of government to achieve its goals and produce good policy outcomes. Can the government collect taxes? Enforce contracts? Build infrastructure? Provide basic services? Prevent private violence?
Strong states can do these things. Weak states struggle. The difference matters enormously. There’s a reason Somalia and Switzerland have such different outcomes.
American intellectuals borrowed this concept for a more specific concern. The United States government seems less able than it used to be, or than peer nations are today, to accomplish basic tasks. We can’t build high-speed rail. Infrastructure projects cost multiples of what they cost elsewhere. Permitting takes forever. The basic machinery of governance seems broken.
This observation is correct. But understanding state capacity in America requires recognizing that it operates on three distinct dimensions:
Resource capacity: Can the government mobilize money, people, and equipment? Does it have fiscal resources to hire skilled workers, buy modern equipment, sustain programs over time?
Operational capacity: Does the government have expertise to design and implement effective policies? Can it collect accurate data, analyze problems, coordinate across agencies, adapt when circumstances change?
Political capacity: Can the government make and enforce decisions? Does it have legitimacy with citizens? Can it overcome opposition from organized interests? Can it maintain policy continuity across election cycles?
Strong states have all three. Weak states lack one or more.
The research concludes America has severe deficits in all three dimensions. That diagnosis misses something crucial: where these deficits actually exist.
The Federal Focus Problem
Consider the major works in popular books.
Brink Lindsey’s State Capacity: What Is It, How We Lost It, and How to Get It Back identifies problems across levels of government but directs virtually all its reform proposals at Washington. Staffing the federal workforce. Improving federal tax collection. Enhancing federal information technology. Streamlining federal environmental laws. Lindsey acknowledges that dispersing authority across federal, state, and local jurisdictions complicates policymaking, but dismisses constitutional reform as “a waste of time” because the Constitution is “absurdly difficult to amend.”
Rick Pildes’s “The Neglected Value of Effective Government” makes a powerful theoretical argument: Democratic theory overemphasizes values like accountability, transparency, and fair representation while neglecting effectiveness. But his examples focus on federal institutions. Short House terms. Federal campaign finance laws. The Voting Rights Act’s impact on congressional redistricting. Federal transparency laws. He notes that governance challenges “extend down to the local level,” but the analysis centers on national politics.
Steve Teles’s “Kludgeocracy” anticipated many current concerns, criticizing the byzantine complexity of federal programs and federal-state arrangements. His solutions? End the filibuster. Reduce congressional committee referrals. Give federal agencies more discretion. Establish a “norm” that whoever pays should also administer. Only this last proposal substantively addresses state governance, and it’s actually a call to end cooperative federalism as we know it. Politically impossible.
Even newer entrants follow the pattern. The Roosevelt Institute’s framework for “democratic state capacity” focuses entirely on federal issues: congressional funding for federal agencies, presidential initiatives, congressional and judicial oversight of federal regulators. The Niskanen Center’s “The How We Need Now” concentrates on actions for the White House, Congress, and federal civil society (but to be fair the Niskanen folks have a lot on state and local capacity).
The pattern is bit of a cliche. Identify state capacity as the problem. Concentrate analysis and solutions on federal institutions alone.
This consensus crosses ideological lines. Libertarian economists and progressive think tanks agree American government isn’t working. They agree we need more state capacity. They all look to Washington for answers.
They’re looking in the wrong place.
Imagine your house is sinking. Cracks spider across the walls. Doors won’t close. The floors slope noticeably.
So you hire an architect to redesign your roof. They’re excellent - they propose a stunning new roofline with solar panels! Not to mention it somehow provides better ventilation and improved insulation for the house! The plans are detailed and professional.
But your foundation is crumbling. The roof redesign won’t stop your house from sinking.
That’s where the state capacity literature is right now. It has correctly identified that something is wrong with American governance. The diagnoses of federal dysfunction are often accurate. Congress is gridlocked. Federal agencies are overburdened with procedural requirements. The proposed solutions (ending the filibuster, streamlining NEPA, increasing congressional staff) might well improve federal governance.
But most of what Americans experience as “government” isn’t federal. So federal reforms, however well-designed, can’t fix the problems most people care about most.
If you want to understand why your city can’t build housing, studying the filibuster won’t help. If you want to know why infrastructure is so expensive, analyzing federal agency staffing won’t provide answers. If you’re concerned about failing schools, debating the structure of congressional committees misses the point entirely.
You need to look at your city council, your state legislature, your state transportation department, your local school board.
Who Actually Governs America
Public Employment: Where Government Workers Actually Work
Let’s start with the most basic question: Who works for the government?
The federal government employs about 2 million civilians. Add active-duty military and postal workers and you get to 3.8 million. If you include contractors and grant employees, we are talking about 9.1 million. Even this upper bound requires heroic assumptions about which contractor employees are “really” government workers.
State and local governments employ 19.9 million people.
Read that again. State and local government workforces are more than double the size of the federal workforce, even using the most generous federal counting method. And it might be larger than these numbers suggest, because state and local governments also rely heavily on contractors, but those aren’t included in the 19.9 million figure.
Where Do State and Local Government Employees Work?
According to the U.S. Census Bureau’s Annual Survey of Public Employment & Payroll (ASPEP), which has tracked this data annually since 1957:
March 2019 Data (from the 2019 ASPEP Summary Report):
11.2 million in education (8.3 million at local level in elementary/secondary; 2.9 million at state level, mostly higher education)
1.1 million in hospitals (0.7 million local, 0.4 million state)
1.0 million in police protection (0.9 million local, 0.1 million state)
~0.7 million in corrections (based on 2010 Census data showing 731,692)
Hundreds of thousands more in fire protection, highways, public welfare, sanitation, parks, libraries, and other functions
Nearly three-quarters of these government workers are employed by local governments. Not even states, but counties, townships, school boards, cities, special districts.
If personnel is policy, state and local governments are policy in America.
You might object: Sure, states and localities employ lots of teachers and police officers, but those workers don’t “set policy” the way an EPA official does. Federal regulators make the important decisions (whether to crack down on air pollution, approve a new drug, tighten scrutiny of banks).
This objection is wrong in two ways.
First, teachers and police officers do make policy, both explicitly and implicitly. School boards and police departments formulate policies of general application. But more fundamentally, the day-to-day performance of teachers and officers is policy. A school district with excellent teachers implements a very different education policy than a comparable district with poor teachers, regardless of what curriculum guidelines say. A police force that professionally serves the community implements a different public safety policy than one that shirks duties or abuses citizens. Their collective actions don’t just “influence” policy. They are policy in the most meaningful sense.
Second, most federal jobs aren’t actually that different from state and local jobs. Federal law enforcement (Customs and Border Protection, TSA, FBI) employs about 137,000 officers. The Department of Veterans Affairs employs 371,000 people running what is effectively the nation’s largest healthcare system, accounting for nearly 20% of the federal civilian workforce. The IRS employs 76,000 people, while state and local tax agencies collectively employ many thousands more doing essentially similar work.
The biggest difference isn’t that federal workers do fundamentally different jobs. It’s that there are vastly fewer of them.
Infrastructure: They Build It, They Run It, They Pay for It
High-quality infrastructure has symbolized effective governance since the Roman aqueducts. American politicians understand this. “Infrastructure Week” became a running joke in the Trump administration. The Biden administration passed the massive Infrastructure Investment and Jobs Act and proclaimed it would get America building again.
Yet here’s a basic fact that Washington’s infrastructure discourse obscures: The federal government owns almost no public infrastructure.
Outside of post offices, military bases, and some dams, federal infrastructure holdings are minimal. What the federal government does is provide money (grants to states and localities that actually build and operate roads, bridges, transit systems, water treatment plants).
Most of this money flows through formulas based on population and need, not specific project approvals. A smaller portion comes through discretionary grants requiring competitive applications. The federal government also runs loan programs and credit enhancements. But construction and operation remain firmly in state and local hands.
The numbers bear this out. In 2019 (pre-Infrastructure Investment and Jobs Act), according to the Congressional Budget Office and Census Bureau, state and local governments outspent the federal government across every infrastructure category:
Highways: $131 billion (state/local) vs. $46 billion (federal) in 2017—representing nearly 3:1 state/local dominance
Water utilities (drinking water and wastewater): 96% of total investment came from state and local governments($109 billion state/local vs. $4.5 billion federal in 2017)
Mass transit and rail: State and local governments provided the majority of funding (total spending approximately $75 billion)
Aviation, water transportation, and water resources: State and local governments outspent federal across all remaining categories
Overall in 2017, state and local governments contributed $342 billion (78%) of the total $441 billion in public spending for transportation and water infrastructure, while the federal government contributed $98 billion (22%). Before you bring up Build Back Better and the CHIPS act, I am taking a wait and see approach considering at time of writing and the chaos of current leadership.
These figures include both capital spending AND operations/maintenance. For operations and maintenance specifically (which the federal government largely ignores), state and local dominance is even more pronounced: In 2017, state and local governments covered $240 billion, or 90 percent, of the $266 billion total spending for operation and maintenance.
Your state and city don’t just build your infrastructure. They maintain it, operate it, and pay most of the bills.
Federal money comes with federal strings, of course. Environmental regulations, labor requirements, and procurement rules shape how states and localities build. We’ll come back to this. But the fundamental point stands: If you’re asking “who builds America’s infrastructure,” the answer is state and local governments.
The Cost Crisis and What It Reveals
American infrastructure costs are, bluntly, insane.
Transit costs illustrate the problem most starkly. The Marron Institute’s Transit Costs Project found that U.S. transit projects cost dramatically more than comparable projects elsewhere. The Second Avenue Subway in New York City cost, per kilometer, eight to twelve times the international average. Not double. Not triple. Eight to twelve times.
Compare some representative projects:
Paris built 10 miles of automated subway for $3.8 billion
Berlin built 13 miles of subway for $3.3 billion
New York built 3.5 miles of the Second Avenue Subway for $17 billion
The U.S. doesn’t just have the world’s most expensive subway projects. It has the most expensive highway projects too. Zach Liscow and Leah Brooks found that the cost of building a mile of interstate highway has increased dramatically over time in ways that can’t be explained by rising wages or materials costs. These costs far exceed international comparisons and vary wildly across states within the U.S.
Why are U.S. infrastructure costs so high?
The Transit Costs Project’s analysis points to lack of state and local agency capacity. American transit agencies rely too heavily on consultants because they lack in-house planning expertise. They show little interest in how construction is done abroad. They’re internally divided by politics and labor organization, leading to bloated staffing requirements during construction. When Liscow and Brooks studied highway costs, their leading explanation was the increased power of “citizen voice” since the 1970s (the ability of project opponents to exploit legal and political tools to demand expensive mitigation measures and cause delays).
Let that sink in. The problem in this case isn’t that Congress is gridlocked (unstable funding is its own can of worms). It’s not that federal agencies are hidebound by procedural rules (though they are). The problem is that the governments that actually build things (state departments of transportation, municipal transit agencies, local water authorities) can’t build efficiently.
Property, Contracts, and the Rule of Law
Two features define capitalist economies: secure property rights and enforceable contracts. State capacity scholars rightly treat the security of these rights as crucial.
But here’s what often gets missed: The definition and enforcement of property and contract rights in America is overwhelmingly a state responsibility.
The primary bodies of contract and property law are state common law and state statutes. Yes, federal interventions matter. The Federal Arbitration Act affects contracts. The estate tax influences property transfers. Federal constitutional guarantees constrain state power. But the day-to-day content of property and contract law comes from state courts and state legislatures.
This matters more than you might think.
When State Decisions Shape Global Markets
Consider New York. Sophisticated parties drafting major commercial agreements must choose which state’s law will govern their deal. They choose New York law for about 46% of non-merger commercial contracts. About half of all sovereign debt contracts worldwide are governed by New York law. Seven judges on the New York Court of Appeals in Albany shape the rules for international sovereign debt, cross-border commercial transactions, and major domestic business deals.
Yet when Governor Kathy Hochul nominated Hector LaSalle to the Court of Appeals in 2022, the contentious public debate focused entirely on criminal law, social issues, and an anti-labor track record. No one mentioned commercial law. No one discussed the court’s role in making New York a global financial capital. No one asked how the nominee might affect the ability of governments worldwide to borrow money.
That same year, the New York Legislature began considering the “Sovereign Debt Stability Act,” legislation that would fundamentally change how sovereign debt restructurings work for bonds governed by New York law. The proposed law would create a comprehensive mechanism to restructure sovereign debt, override existing contract terms, and cap creditor recoveries under certain conditions.
This is a state legislature proposing to rewrite the rules for roughly half of global sovereign debt.
Critics (particularly in New York’s financial services industry) warn the Act would raise borrowing costs for vulnerable countries, trigger extensive litigation, and prompt a migration of sovereign debt contracts to other jurisdictions where contractual rights are more predictable. England. Texas. Delaware. Markets would route around New York if New York law became too uncertain.
The Act hasn’t passed. But the fact that it’s being seriously considered reveals something fundamental about American governance: A single state legislature can credibly threaten to reshape global financial markets. Not through superior expertise or careful coordination with international institutions. Simply by exercising the authority over contract law that the federal system gives to states.
State officials themselves sometimes forget they’re in charge of contract and property law. Markets don’t forget.
When State Courts Can’t Handle What They Control
Texas provides a different kind of example. The oil and gas sector generates over $4 trillion in profits annually. Texas produces 43% of U.S. oil and 29% of U.S. natural gas. Over 20 energy-related Fortune 500 companies are headquartered there. Oil and gas disputes originating anywhere in the country often look to Texas law as authoritative.
But for decades, much of this litigation was handled in rural Texas state courts where district judges presiding over “cattle call” dockets heard multimillion-dollar oil and gas disputes alongside family law and criminal matters. Appeals went to understaffed appellate courts covering vast geographic areas. Delays mounted. The state courts shaping rules for a trillion-dollar industry lacked the capacity to handle that role effectively.
Texas recognized this as a state capacity problem. In 2023, the legislature created specialized business courts with statewide jurisdiction for disputes exceeding $5-10 million, along with a new appellate court to review their decisions.The explicit goal: provide uniform, efficient resolution of complex commercial disputes that drive the state economy.
These examples expose a pattern. States control crucial economic infrastructure (contract and property law). Markets depend on this infrastructure working well. When it doesn’t work well, the consequences ripple far beyond state borders thanks to interconnected supply chains (and financial derivatives that are affected by shocks of said supply chains). International sovereign debt markets depend on New York courts and legislatures making predictable decisions. Global energy markets depend on Texas courts resolving disputes efficiently.
Yet state court capacity is largely invisible in national policy debates. The federal fixation means we don’t notice when state courts lack resources, expertise, or institutional design to handle what they control. We don’t connect judicial capacity to economic outcomes. We don’t see court reform as state capacity reform.
This is asymmetric capacity in a different form. Sophisticated parties can navigate inadequate state court systems through forum shopping, arbitration, or expensive lawyers. Smaller parties can’t. When state courts lack capacity to handle what they control, the dysfunction falls unevenly.
State courts are state capacity. When we ignore them, we’re ignoring where a lot of critical core economic governance actually happens.
Zoning and Land Use
Nowhere is state and local control of property rights more consequential than in land use regulation.
Local governments, via power delegated by states, control what you can build through zoning, building codes, subdivision requirements, historic preservation laws, and environmental regulations. Some states have clawed back power, but local control remains dominant and the federal government stays on the sidelines.
Restrictions on building housing have made it unaffordable across entire metropolitan areas. The consequences cascade: endemic homelessness, household precarity, reduced economic growth, increased inequality, racial segregation, environmental damage from sprawl.
Failure at Every Level
Land use is a prime example of failures at every level of government and in how those levels interact.
Federal: Shaped housing markets for decades through policies that subsidized sprawl (highway funding, mortgage interest deductions) and enforced racial segregation (redlining). Has constitutional authority to override exclusionary local zoning but refuses to use it.
State: Delegated land use authority to localities without adequate oversight. Created a system where each jurisdiction externalizes costs onto neighbors by blocking housing. Lacks political will to preempt local obstruction even when exclusionary zoning clearly harms regional housing markets.
Local: Responds to organized homeowners who dominate public meetings and blocks new construction even when most residents want affordability. Planning departments lack capacity to process permits efficiently.
Coordination: No level coordinates with the others. Federal policy subsidizes sprawl while most states delegate unchecked power to localities. Most localities block housing while demanding federal and state infrastructure funding. Nobody takes responsibility because responsibility is genuinely diffused across levels and can be easily gamed by people with extra time and a pen.
The bill offers incentives (competitive grants for jurisdictions that increase housing permits), technical assistance (helping cities identify zoning reforms), and modest penalties (reducing Community Development Block Grants for high-cost cities that refuse to build). It streamlines federal environmental reviews, reforms mortgage rules and manufactured housing regulations, and ties federal transit funding to local pro-housing policies.
What it refuses to do is override local zoning directly. The federal government has constitutional authority under the Commerce Clause and Spending Clause to preempt state and local laws, including zoning. It used this authority before with the Fair Housing Act. But the ROAD Act is a (small) step in the right direction, offering carrots and threatening small sticks, hoping localities will voluntarily reform.
This is the pattern: recognize the problem, identify the barriers, propose solutions that avoid confronting the actual source of dysfunction. The federal government won’t override localities. States won’t override localities. Localities won’t override homeowners. Housing doesn’t get built. Prices soar. Nobody can fix it because no level will act decisively.
When housing becomes unaffordable, voters blame “government” generally. They can’t trace the chain from federal sprawl subsidies through state delegation of authority to city council decisions blocking apartments. This is invisible if you’re only looking at Washington.
Public Safety: The OG State Function
A lot people will reference Max Weber as they defined a state as holding a monopoly on the legitimate use of violence within a territory. The ability to maintain that monopoly (to prevent private violence) is a central measure of state capacity in theory.
In the United States, preventing private violence is almost entirely a state and local function.
The numbers:
1.2 million state and local law enforcement officers (including 780,000 sworn officers)
137,000 federal law enforcement officers
1.5 million people in state prisons and local jails
200,000 people in federal prisons and jails
This decentralization has profound consequences.
Law enforcement responsibilities are scattered among more than 17,000 different agencies. About two-thirds of all officers work for local police departments, another 17% for county sheriffs. This radical fragmentation means we don’t even have good data on the number of crimes committed annually in the United States. International comparisons (straightforward in most wealthy countries) become exercises in estimation and approximation.
Prosecution is local too. District attorneys, elected at the county level, exercise enormous discretion over charging decisions. As John Pfaff argues in Locked In, the power of local prosecutors is the primary driver of mass incarceration, more than sentencing laws, more than drug policy, more than private prisons.
The United States has far more interpersonal violence than peer nations. The homicide rate is six times higher than the European Union average. The gun homicide rate is twenty-three times higher. This isn’t driven by higher overall crime rates (property crime rates are comparable to other wealthy countries). It’s about violence specifically, which is largely a product of firearm availability (state-regulated) and policing resources (state and locally funded).
State and local control of criminal justice means policy varies wildly. Some cities have defunded police; others have hired aggressively. Some prosecutors decline to charge certain offenses; others pursue maximum penalties. Some states have eliminated cash bail; others have expanded it. This variation can be valuable for policy experimentation. But it also means that when criminal justice fails (and by international standards, American criminal justice clearly fails), that failure is rooted in state and local governance.
The Safety Net Is State-Administered, (Partially) Federally Funded
State capacity scholars often use government’s ability to provide social welfare services as a measure of capacity. In the United States, the federal government finances a large portion of the safety net. But states and localities administer it.
The biggest program is Medicaid, which provides healthcare to about 95 million people (nearly 30% of the American population). Total Medicaid spending (including the Children’s Health Insurance Program) amounts to roughly 30% of total state expenditures, almost half as much as states spend on education.
The federal government pays for about two-thirds of Medicaid costs (varying by state). But states are entirely responsible for implementation. They set eligibility rules, design enrollment processes, determine benefits, establish cost-sharing requirements, and set payment rates for providers. They do so with broad “experimental” waivers granted by federal authorities. The result: fifty wildly different Medicaid programs.
States consistently struggle with administration: ensuring adequate provider networks, overseeing nursing home quality, preventing fraud. During the COVID-19 pandemic, state implementation failures led to massive coverage disruptions as states “unwound” temporary enrollment expansions. These aren’t federal failures. They’re state administrative failures.
Unemployment insurance tells a similar story. In a typical year, state unemployment agencies distribute about $40 billion in benefits. When COVID-19 hit and claims surged, thinly-staffed state agencies with antiquated computer systems collapsed under the weight. An estimated $163 billion in fraudulent claims were paid out. From 2013 to 2015 in Michigan, broken computer code led to 40,000 people being falsely accused of fraud, with some facing bills as high as $100,000.
States also administer SNAP (food assistance, $115+ billion annually to 22+ million households), TANF (cash assistance, $30+ billion for low-income families), housing vouchers ($30 billion through local public housing authorities), and school lunch programs feeding millions of low-income students daily.
The October 2025 government shutdown starkly illustrated how state administration of federal programs can fail. For the first time in the program’s 60-year history, SNAP benefits for 42 million Americans were at risk of not being disbursed. Federal courts had to intervene, ordering the administration to use contingency funds. Even then, recipients initially received only 50% of their normal benefits, with significant delays as state agencies struggled to recode their antiquated systems. Some states stepped in with emergency funding; others could not. The crisis exposed the fragility of state-administered safety net programs and the chaos that ensues when 50 different state systems must respond to federal funding disruptions.
Local governments run additional crucial services: public hospitals (roughly one-sixth of American hospitals, disproportionately serving low-income and uninsured patients), mental health facilities (about 7% of the nation’s mental health facilities), homeless shelters (emergency housing in most major cities), and public health clinics (basic healthcare in communities nationwide).
The COVID-19 pandemic starkly revealed the importance of state and local public health infrastructure. County health departments, state epidemiologists, and local hospital systems were the front line of response. Their capacity (or lack thereof) determined how communities weathered the crisis.
When Americans complain about government failing to provide basic services, they’re usually complaining about state and local failures.
A Historical Note
This federal fixation would have baffled earlier generations of Americans.
In the nineteenth and early twentieth centuries, the federal government consumed less than 3 percent of GDP. It was a bit player in most Americans’ lives. State and local governments built the infrastructure, regulated the economy, provided what limited social services existed. When scholars studied American governance, they studied state and local government. When reformers wanted to improve government, they targeted city hall and the state capitol.
And American government then, despite its many flaws, was actually pretty impressive by international standards. Consider the 1893 World’s Columbian Exposition in Chicago. The fairgrounds displayed what American city governments had accomplished: an electrical system with 16 generators powering 172,000 incandescent lights, a scale two to three times larger than Chicago’s entire commercial district. This wasn’t federal prowess on display. It was what American cities could do.
As historian Jon Teaford documented in The Unheralded Triumph, American cities of the late nineteenth century were remarkably successful at meeting unprecedented challenges. By 1900, American cities boasted the most abundant water supplies, brightest street lights, grandest parks, largest public libraries, and most efficient systems of transportation in the world. American cities had better infrastructure than their European counterparts.
The federal budget has since expanded dramatically. Federal spending grew from 3 percent of GDP in 1900 to over 20 percent in recent decades. Federal regulatory power has grown. These changes matter. But they haven’t fundamentally altered the fact that most governance (most of what “the state” does) happens at state and local levels.
What’s changed more dramatically is our attention. National media, national politics, and national policy debates dominate. State and local governance has become largely invisible, even though it affects daily life far more than federal governance does.
When Capacity Appears (For Those Who Can Demand It)
Now we can return to the puzzle from the opening: How does America build data centers and stadiums while failing to build housing and transit?
The answer reveals what American state capacity actually is.
America has coordination capacity. The federal government can act decisively when it wants to. State governments can fast-track approvals. Local governments can override opposition. The administrative machinery works when deployed.
The problem is deployment. That capacity activates for organized interests with money and multi-level lobbying operations. It doesn’t activate for diffuse public interests.
This isn’t a capacity deficit. It’s asymmetric capacity.
Let’s trace how data center coordination actually worked.
Tech giants needed to build massive facilities quickly. They faced nominal obstacles at every level of government. Federal environmental review. State permitting. Local zoning. Opposition from residents and activist groups that blocked or delayed over $64 billion in projects.
But tech giants possess something most constituencies don’t: the ability to operate across multiple venues simultaneously. They lobbied federal agencies and the White House. They negotiated with state governments. They pressured county boards and city councils. When one jurisdiction resisted, they credibly threatened to move the project elsewhere, creating competition among localities for their investment.
What distinguishes these cases? Three things. Money to hire experts who understand all three levels. Organization that enables simultaneous lobbying at federal, state, and local levels. Credible threats to move projects elsewhere if one jurisdiction doesn’t cooperate.
Diffuse constituencies have none of these tools.
The result: Trump issued an executive order accelerating federal permitting for data centers. States provided massive subsidies and fast-tracked approvals. Local governments overrode opposition. In Warrenton, Virginia, residents voted out all town council members who supported Amazon’s proposed data center. Amazon adapted its approach and many similar projects proceeded elsewhere.
The coordination happened. Federal environmental review moved quickly. State permitting bent. Local opposition, while real and organized, got overcome more often than not.
Sports stadiums follow the same pattern.
Cities that can’t fix potholes somehow deliver billion-dollar stadiums with public financing. Multiple agencies coordinate. Federal transportation dollars flow. State bonds get issued. Local zoning changes. Tax increment financing districts get created. Eminent domain gets deployed.
The coordination happens because stadium owners (wealthy individuals, sports leagues with sophisticated lobbying operations) can activate it. They operate across all three levels simultaneously. They can credibly threaten to move teams. They coordinate with officials at federal, state, and local levels to secure subsidies, expedite approvals, and override opposition.
Highway expansions follow the same pattern.
State DOTs that take a decade to plan transit lines fast-track highway widening. Why? Highway construction interests (construction companies, engineering firms, auto manufacturers) are organized at all three levels. They lobby federal transportation officials for formula funding. They work with state DOTs on project priorities. They secure local approvals.
Then there’s Tesla. Other manufacturers said California’s regulatory environment made building new factories impossible. Too many agencies. Too many reviews. Too many opportunities for opposition.
Elon Musk called Governor Gavin Newsom. Newsom personally intervened. State agencies coordinated. Permits moved quickly. Factories got repurposed. Electric Cars get subsidies and sold. Lawsuits about mass pollution get suppressed. Even Gavin Newsom admits “We created the market. There is no Elon Musk, there’s no Tesla without California’s regulatory framework, period, full stop. It wouldn’t exist. It was because of the regulations.” Across the Pacific pond, Xi is better at reminding EV manufacturers who butters their bread.
When powerful, connected interests need government to coordinate across federal, state, and local levels, government coordinates. Environmental reviews that normally take years move in months. Permits that normally take eighteen months get processed in weeks. Opposition that normally blocks projects gets navigated, accommodated, or overridden.
The capacity exists. The question is who can activate it.
When Coordination Fails (For Everyone Else)
Picture yourself as a state transportation director trying to build a subway line.
You need federal money. The project is too expensive to fund entirely with state dollars. So you apply for federal transit grants, competing with every other state. The application takes months. You hire consultants because your agency doesn’t have staff who understand federal grant procedures.
You get the grant. Now you need federal environmental approval. NEPA review begins. The federal government requires an environmental impact statement. This takes two years.
But you’re not done. Your state has its own Little NEPA requiring separate environmental review. Different process. Different timeline. Potentially contradictory requirements. Another 18 months.
Meanwhile, localities along the route each have their own review processes. Public hearings. Design review boards. Historic preservation commissions. Neighborhood consultation requirements. Each provides another opportunity for opponents to demand expensive modifications or delay the project.
You need god only knows how many different permits from how many different agencies at three levels of government. You lost count at 50 and 20 respectively
By the time you’re ready to break ground, costs have doubled. Your project now costs eight to twelve times what a comparable subway costs in Paris or Berlin.
This is American transit construction.
Read that again: eight to twelve times the international average.
Why? Not because American workers are less skilled. Not because American technology is inferior. Because the intergovernmental system multiplies points of failure. Each level requires its own environmental analysis. Each level has consultation requirements with overlapping stakeholder groups. Each level can impose design changes that require re-engineering.
Nobody coordinates to eliminate redundancy. Nobody coordinates to ensure the project gets built efficiently. The capacity to coordinate exists. We’ve seen it work for data centers and stadiums. But transit riders can’t activate that coordination the way tech giants can.
Housing follows the same pattern.
Local governments control zoning. State governments delegate that power but increasingly try to override exclusionary practices. Federal policy shapes housing markets through mortgage policy, transportation funding, and Fair Housing Act enforcement.
Each level has some authority. None has full control. No coordination mechanism exists to align efforts across levels.
Federal policy subsidizes homeownership and sprawl. State policy delegates too much power to localities without enough oversight. Local policy responds to homeowners who oppose new housing.
Housing doesn’t get built. Prices soar. Nobody can fix it because no level can coordinate action across all three levels the way stadium developers can.
Social services show the same dynamic. States administer Medicaid, covering 95 million people. The federal government pays for about two-thirds of costs. But states are entirely responsible for implementation.
This cooperative federalism model works when the federal government provides stable funding and reasonable rules. It breaks down when federal funding becomes a political football.
Trump’s “Big Beautiful Bill” cuts federal Medicaid funding. Virginia voters said these federal cuts affected their family finances and voted accordingly. But states face the immediate consequences. They can’t print money. They must balance budgets. So they either cut Medicaid or slash other services.
When Arkansas faced similar federal Medicaid cuts in 2017, the state responded by reducing eligibility, cutting benefits, and lowering provider reimbursement rates. Thousands lost coverage. Doctors stopped accepting Medicaid patients. Rural hospitals closed. The federal decision cascaded through state budgets and local service delivery.
When Medicaid patients can’t find doctors, voters blame “government” generally. They can’t trace the chain from federal budget decision through state fiscal constraints to their local clinic closing.
The pattern is consistent: Diffuse public interests (transit riders, aspiring homeowners, Medicaid patients) can’t activate the coordination capacity that corporate interests (tech giants, stadium owners, highway contractors) routinely deploy.
The capacity exists. Access to it is unequal.
How Other Systems Coordinate (And What That Reveals About Ours)
China provides the starkest contrast.
China has one of the world’s most complex, fragmented governance structures: 34 provincial-level governments, 333 prefecture-level cities, 2,800+ counties. Local governments control most implementation. They compete with each other, sometimes fiercely.
Yet China builds infrastructure at one-quarter to one-eighth of American costs, depending on project type. When China decided to build an EV industry, different levels deployed different tools in a coordinated strategy. Central government provided R&D funding. Provincial governments offered land and infrastructure. Cities provided consumer subsidies. Each level adapted national priorities to local conditions. Information flowed between levels. Successful approaches spread rapidly.
The result: China went from producing almost no EVs to dominating global production in 15 years.
The key difference: China’s coordination advances state-defined goals. Central authorities set priorities. Local governments implement them with substantial autonomy in execution. The Communist Party manages personnel across levels, rotating officials to spread knowledge and setting performance targets for career advancement.
The authoritarian versus democratic distinction matters enormously for normative reasons. We rightly reject systems that suppress dissent and eliminate political choice. But technically, both systems have coordination mechanisms. China’s mechanisms respond to party-defined priorities. America’s mechanisms respond to lobbying power.
Neither is neutral or “natural.” Both are designed, just designed to serve different interests.
Germany and Switzerland show democratic coordination is possible. These federal systems have strong state and local governments. They build infrastructure efficiently. They maintain excellent schools. They deliver high-quality services.
How? They’ve developed explicit coordination mechanisms.
Germany’s fiscal equalization system redistributes resources from wealthy to poor states, ensuring some level resources for capacity everywhere, even with an “austere” leadership from Merkel to Scholz with a “questionable” taste of policy. Which may explain why Germany is doing better than one would expect.
When Hamburg builds infrastructure, federal formulas ensure Bremen can too. Constitutional provisions clarify which level has authority over what functions. Joint federal-state planning bodies coordinate on shared priorities like university funding and regional development. When coordination is needed, institutional structures make it happen systematically, not through lobbying.
Switzerland’s cantonal system preserves remarkable local autonomy while coordinating through clear constitutional divisions of responsibility, revenue-sharing mechanisms, and horizontal agreements between cantons. When multiple cantons need to coordinate (on transportation, for instance), formal inter-cantonal conferences create binding agreements. The system enables both local variation and coordinated action when needed.
These aren’t perfect systems. But they demonstrate that multilevel governance can coordinate democratically when institutional design prioritizes public coordination over private access.
Here’s a puzzle worth considering: Many countries poorer than the United States have better public services. Their transit works better. Their infrastructure costs less. Their cities are cleaner. If the problem were that Americans fundamentally distrust government or prefer small government, we’d expect the opposite pattern. We’d expect America’s relatively limited government to work well. Instead, we see the reverse: Americans have grown increasingly distrustful of government because it performs poorly.
The state capacity crisis isn’t about American culture or American preferences. It’s about American political institutions, specifically about how state and local governments are structured and how they interact with federal policy.
America has cooperative federalism in theory. In practice: conflict, duplication, and selective coordination based on who can demand it.
The Wrong Tools for the Job
The state capacity literature identified a real problem but diagnosed it wrong. They concluded America lost state capacity. Actually, American state capacity deploys asymmetrically. The system works fine for concentrated interests, fails routinely for diffuse ones.
When major works in this literature focus almost entirely on federal institutions, they miss three crucial facts.
First, most governance happens at state and local levels. State and local governments employ ten times more people than the federal government. They build and operate infrastructure. They maintain public safety. They administer social services. They educate children.
Second, federal decisions shape what state and local governments can accomplish. Federal funding determines state resources. Federal mandates create state obligations. Federal regulations constrain state action.
Third, the coordination capacity exists but deploys selectively. Data centers, stadiums, highways, and Tesla factories prove government can coordinate across levels when powerful interests demand it.
This reframe matters for reform. But before discussing what to do, understand what not to do.
The conventional response to government dysfunction (especially from centrist reformers) borrows from private sector management fads. “Government efficiency.” Metrics and rankings. Performance management systems. These sound reasonable. They’re mostly destructive.
This isn’t systems thinking. It’s bad management dressed up as systems thinking.
The difference matters. Real systems thinking (W. Edward Deming’s Total Quality Management, Stafford Beer’s Management Cybernetics) starts from a fundamental principle: Management is responsible for system performance, not workers. When systems fail, look first at how management designed the system, what incentives leaders created, what constraints management imposed, what resources workers had.
Bad management fads and the government messes (Jack Welch’s Six Sigma, the Clinton administration’s National Performance Review, DOGE, most “government efficiency” initiatives) do the opposite. They treat workers as variables to optimize. They rank and rate individuals. They use metrics to identify “underperformers.” They cut staff and call it efficiency. They blame workers for system failures that management created.
When you hear someone propose “government efficiency,” ask what they mean. If they mean cutting staff and budgets, they’ll make capacity worse and only thing that would come from it is making Rahm Emanuel giddy with joy. If they mean redesigning systems so competent people can do their jobs effectively, that might help. But the former is far more common than the latter.
Real capacity-building requires investment: in people, in systems, in institutional knowledge. The cheapest government isn’t the most effective. Effective systems require resources, training, and time to develop.
This matters because intergovernmental coordination requires functional systems at each level. You can’t coordinate what doesn’t work. States with hollowed-out agencies can’t implement federal programs effectively. Localities without planning capacity can’t participate meaningfully in regional coordination. Cutting capacity in the name of efficiency makes coordination harder, not easier.
What We Can Actually Do
The ideal reform would build systematic mechanisms for intergovernmental coordination that serve public interests. Joint federal-state-local planning bodies. Fiscal formulas that reward coordination. Clear divisions of authority. Information-sharing systems. Germany and Switzerland show these mechanisms can work democratically, even with less than ideal national governments.
You’re not getting that. Not this decade, probably not next. Reform happens piecemeal: one state at a time, one policy area at a time, one level of government at a time. (Unless we learn on how to divide and conquer, you can imagine how mind-numbingly slow the reform process will be!)
Does that mean partial reforms are pointless? No. It means thinking strategically about which reforms create value even without full coordination, and which reforms make future coordination more likely.
If you’re working on federal policy: Focus on understanding what are the actual resources being sent to the states, not just mandates. States can’t coordinate effectively when they’re broke. Federal funding that gives states fiscal breathing room enables capacity-building. Reduce duplicative requirements where federal authority is clear, or use it to overwrite state law to assure that state and local players can’t block it if needed. The federal law (and not just in the US) can preempt duplicative state/provincial/prefecture/whatever reviews in areas of clear federal authority. Reward coordination when it happens through grants that prioritize multi-jurisdictional projects.
If you’re working on state/provincial/prefecture/whatever policy: Build capacity even if federal and local constraints remain. Hire more state agency staff. Develop in-house expertise. Modernize IT systems. When opportunities arise, states with capacity can act. States without capacity can’t. Preempt local obstruction where state authority is clear. State zoning reform that limits local blocking power doesn’t require federal coordination. Make it easier for other states to copy success by documenting what worked and sharing implementation guides.
If you’re working on local policy: Don’t wait for state or federal help. Cities can reform their own permitting, streamline their own approvals, and build their own capacity. Houston built affordable housing by simply allowing more building. Minneapolis upzoned without federal help. Build coalitions for state reform. Cities working together have more power than individual cities complaining separately.
This approach won’t produce optimal outcomes. Piecemeal reform is less efficient than systematic coordination reform. Redundant reviews will continue. Fiscal constraints will persist. Powerful interests will still activate coordination more easily than diffuse publics.
But “less than optimal” beats “paralyzed waiting for comprehensive reform.” The purpose of a system is what it does, and right now the system coordinates for concentrated interests. You will be surprised on what you can do, as I can go on the many different ways local governments can achieve remarkable results (especially with boosting birth rates!). Any reform that makes it slightly easier for diffuse interests to get coordination, slightly harder for concentrated interests to block things, or slightly more likely that good examples spread moves in the right direction.
Take the wins you can get. Build capacity where you can build it. Reform what you can reform. Create examples that demonstrate what’s possible. Build coalitions that can demand more.
The asymmetry won’t disappear overnight. But it can erode, piece by piece, reform by reform.
What You Should Take Away
If you care about housing affordability, infrastructure costs, school quality, or government effectiveness generally, understand this:
State capacity exists in America. Groups, especially well organized ones, can coordinate across federal, state, and local levels. We’ve seen it happen for data centers, stadiums, highways, and Tesla factories. The coordination mechanisms work when activated.
That capacity deploys selectively. It activates for interests that can lobby all three levels simultaneously, threaten to move investment elsewhere, and hire experts who understand each jurisdiction’s rules. It doesn’t activate for interests that can’t: transit riders, aspiring homeowners, Medicaid patients.
This isn’t an accident or natural outcome. It’s the result of institutional design (or the lack thereof). Other democracies coordinate across levels more systematically because they’ve built mechanisms that enable public coordination, not just private access.
The next time someone tells you American government lacks capacity, ask them three questions:
Which government are they talking about? Most governance happens at state and local levels, not federal.
How do different levels coordinate? The problems people care about most involve all three levels. Coordination determines whether government works or fails.
Whose interests does that coordination serve? The answer reveals whether you’re looking at a capacity problem or an access problem.
The state capacity literature correctly identified that American government struggles. But by focusing almost entirely on federal institutions, it missed where most governance happens, how different levels interact with each other, and whose interests the system serves when it does manage to coordinate.
That’s not a neutral analytical oversight. It shapes which reforms get proposed and whose problems get solved.
The capacity isn’t gone. It’s selective. And until we understand that selectivity and how it is causing the crisis, until we see clearly that the system’s purpose is what it does, we even more limited on what we can do and what we can build. That’s where reform must start.