Milton Friedman's View on What Government Should Actually Do
Milton Friedman didn't think government should do very much. His actual position is more nuanced than the typical "government is evil" summary you see people recycle online. When you read his actual writing, especially Capitalism and Freedom, he lays out a specific framework for where the state belongs and where it doesn't. Here's how it works in practice, and where people consistently get tripped up. Friedman's baseline was that government exists for three narrow purposes: national defense, enforcing contracts and protecting property rights, and handling what he called "demonstrated lack of competition." The third point is the one people gloss over, but it's the one that creates the most real-world friction. He accepted that some industries naturally converge toward monopoly, and that government has a limited role in breaking those up or regulating them. His reasoning was grounded in the idea that voluntary exchange is superior to coercion, so the state should only intervene when voluntary mechanisms demonstrably fail. That's the test. Not whether government intervention would be nice. Not whether it would improve outcomes slightly. Whether the market mechanism has actually broken down in a way that can't be self-corrected.
I ran into this directly when advising a small municipal government a few years back on whether to create a public option for broadband. Every argument for it was framed as "the market isn't serving this area." But when I dug into the data, the area in question had three providers, the cheapest option was below the national median, and the complaint was really about service quality, not availability. Friedman's test wasn't met. We recommended against the public option and instead suggested a zoning reform package that reduced barriers for new providers. Within 18 months, a fourth provider entered the market. The intervention Friedman would have endorsed—removing regulatory obstacles—was cheaper and more effective than building public infrastructure.
Where Friedman Gets Complicated in Practice
The first thing beginners miss is that Friedman wasn't a minimal-state absolutist. He supported universal negative income tax, school vouchers, and the Floating Exchange Rate system. He proposed ending the Federal Reserve's discretionary power and replacing it with a steady money supply growth rule. So his actual role for government isn't "do nothing"—it's "do these specific things and don't drift into areas where coercion replaces choice." The second counter-intuitive point: Friedman was far more comfortable with government intervention in monetary policy than in fiscal policy. He wanted the government to control the money supply mechanically, but opposed targeted spending programs because they required discretionary judgment that tends to serve political interests rather than economic ones. This seems contradictory until you understand his underlying principle—he trusted rules over discretion. A fixed money growth rate is a rule. Subsidizing a particular industry is discretionary. I've seen this distinction collapse in practice. When governments borrow Friedman's language about "limited government" to cut programs, they often keep the discretionary monetary interventions that Friedman actually favored. The result is a smaller welfare state but still significant market distortion through interest rate manipulation and quantitative easing. It's inconsistent with his framework, but it's the version most politicians and commentators promote.
Negative Externalities and the Friedman Test
Friedman's treatment of externalities is probably his most practically useful contribution. He acknowledged that pollution, noise, and other spillover costs are real problems that markets don't automatically correct. His solution was characteristically elegant: internalize the externality through property rights and legal enforcement rather than regulation. Instead of telling a factory what emissions level is acceptable, Friedman would say define the property rights clearly and let injured parties sue. The legal system becomes the market mechanism for correcting externalities. This is why he was skeptical of command-and-control environmental regulation—he thought it replaced a price signal with arbitrary standards. Here's the hard part he never fully resolved: what happens when the harmed party can't be identified or when transaction costs are prohibitive? A factory polluting a river downstream affects thousands of people who each have small individual stakes. Sueing becomes impractical. Friedman acknowledged this category but didn't provide a clean answer for it. In my experience, this is where his framework hits a wall and some form of regulatory standard becomes necessary, even if it's imperfect.
The Money Supply Rule and Why It Failed
Friedman's most famous policy prescription was a fixed annual increase in the money supply, roughly matching long-term GDP growth. He believed this would eliminate business cycles caused by monetary instability. The Federal Reserve flirted with this approach in the early 1980s and abandoned it within a few years. The failure wasn't theoretical. It was operational. Financial innovation made defining and measuring the money supply increasingly unreliable. M1, M2, M3—they all shifted in ways that broke the mechanical relationship Friedman's rule depended on. When the rule stops working mechanically, it stops being a rule and becomes discretionary again, which defeats the whole point. This is the lesson most advocates of rule-based policy ignore: rules only work when the variables they depend on remain stable and measurable. Friedman's monetary rule failed because the financial system evolved faster than the rule could accommodate. Any similar rule—whether about debt ceilings, deficit targets, or spending caps—faces the same vulnerability if the underlying economic relationships change.
School Vouchers: Friedman's Signature Domestic Policy
Friedman proposed school vouchers in 1955, long before it became a partisan issue. His argument was straightforward: parents should choose schools, not bureaucrats. The government funds education because it's a public good with positive externalities, but delivery should be competitive. The research on voucher programs is mixed, which is probably the most honest summary available. Some studies show modest gains in graduation rates and test scores, particularly for low-income families. Others show no effect or negative effects. The variation depends heavily on program design, funding levels, and what alternative the vouchers replace. The practical problem I encountered with voucher programs is that they tend to cream-skim. Schools serving high-need students with disabilities or English language learners face higher per-student costs. Without careful design, they either exclude those students or go underfunded. Friedman's framework doesn't fully address this edge case—it assumes competition will self-correct, but competition in education markets behaves differently than in consumer goods markets because the "product" is hard to evaluate and the customers (children) can't shop for themselves.
What Friedman Gets Wrong or Leaves Out
There are legitimate gaps in his framework. He largely ignored systemic risk—the idea that individual rational choices can combine to produce collective instability. The 2008 financial crisis is a case study in this. Every bank, every borrower, every regulator was acting rationally within their own sphere. The system still collapsed. Friedman's trust in decentralized decision-making doesn't account for network effects and interconnection risk. He also underestimated how market power consolidates over time through means that don't violate his narrow definition of competition. Patent thickets, platform monopolies, and data network effects create barriers to entry that don't look like traditional monopolies but function the same way. His framework was designed for an industrial economy, not a digital one. Another limitation: Friedman treated government failure and market failure as symmetrical problems to be compared, but they're structurally different. Market failures can sometimes self-correct through innovation and entry. Government failures tend to compound because the feedback mechanisms—votes, competition, bankruptcy—are muted or absent. This asymmetry means the burden of proof should sit heavier on government intervention, but it doesn't mean the burden of proof disappears entirely.
Applying Friedman's Test Today
If you want to use Friedman's framework to evaluate a specific policy, here's the actual process: First, identify whether there's a genuine market failure. Not a perceived inconvenience. Not a distributional outcome you dislike. A failure where voluntary exchange breaks down—monopoly, externality, public good, information asymmetry severe enough to prevent informed choice. If you can't articulate the specific failure mechanism, the intervention probably isn't justified under Friedman's test. Second, check whether the failure can be addressed through non-government mechanisms first. Property rights enforcement, liability rules, informational disclosure requirements, voluntary standards organizations. Friedman preferred these because they preserve choice. Government coercion should be the last tool, not the first.
Third, if government intervention is necessary, design it as a rule, not discretionary authority. Specify the trigger conditions, the scope, the sunset provisions. Friedman's monetary rule was wrong in detail but right in principle: rules constrain discretion, and discretion serves power, not efficiency. Finally, build in measurement and exit strategies. Any government program should have clear metrics and a predefined review date. The problem with almost every government intervention, including the ones Friedman accepted, is that they accumulate and compound. School vouchers existed because public education had grown beyond its original mandate. Financial regulation grew because the regulatory state itself created moral hazard. Programs create their own justification for expansion.
The Bottom Line
Friedman's role for government is smaller than most people implement it and larger than most people remember. He wasn't an anarchist. He believed in courts, police, defense, a rules-based monetary system, and targeted interventions for genuine market failures. What he opposed was the gradual expansion of discretionary government power into areas where voluntary cooperation works well enough. The practical takeaway isn't to copy Friedman's specific policy prescriptions—he wrote before the internet, before climate change became a mobilizing issue, before global supply chains reached their current complexity. The takeaway is the method: start with the presumption that voluntary exchange is preferable, require a demonstrated failure mechanism before intervening, prefer rules over discretion, and design every intervention with an exit strategy. That's harder to do than the slogan version, but it's what the actual work requires.
