A hidden cost of antitrust and regulation is that the political economy can be gamed—competitors (as arguably with Microsoft) instigate cases, imposing large transaction costs and management attention, and fostering a culture where managers spend more time gaming the government than producing better products.

Even superior price mechanisms remain vulnerable to politicization: while economists can estimate bounds on social costs (of pollution, value of time for congestion tolls), the actual tax or toll levels get set politically rather than at the economically optimal level, as seen with intercounty connector tolls being set high and likely facing pressure to reduce.

Across the board, government interventions to correct market failures have turned out remarkably disappointing: while real failures like pollution and congestion exist, there is little academic empirical evidence that government interventions—whether justified or not—significantly improved outcomes, even against the modest benchmark of merely producing net benefits rather than optimal policy.

Antitrust policy is an area of economics pervaded by 'religion'—people hold strong positions regardless of evidence and write pro or con without ever reconciling with what the empirical record shows—even more so than macroeconomics, where at least there is a concerted effort to generate evidence.

Any proposed regulation—especially financial regulation—should be evaluated by demanding the specific mechanism: exactly how it will reduce risk or correct the problem without throwing out important incentives and innovation, rather than relying on the abstraction that 'there's a problem, government should do something.'

Price mechanisms (taxes or charges on emissions/noise) would achieve the same pollution-reduction benefits more cheaply than command-and-control technology mandates, because mandates impose an 'engineering mentality' that pursues a technological goal while treating cost as secondary—as shown by Honda being forced to install catalytic converters even though it already met the same pollution standard.

Distribution and efficiency are often presented as conflicts but need not be: when an economy is inside its production frontier, one can both expand the pie and divide it more favorably, though Roberts is skeptical of this in practice.

Rather than asking whether the textbook market failures are real, the more useful question is empirical: when government is brought in to fix them, does it actually do a good job, make things worse, or is the extent of the failure overstated—and the accumulated body of evidence should guide policy rather than starting from square one each time.

Reviewing antitrust collusion and monopoly studies reveals an absence of positive evidence: government prosecutions (e.g., Microsoft, IBM) cannot be shown to have produced significant price decreases or consumer benefits, which should give pause before launching further large cases such as the Intel investigation.

The deterrence defense of antitrust—that without it collusion and monopoly would run rampant—has been tested via international comparisons of antitrust-law strength (US vs Canada, US vs UK), and even those studies do not find that antitrust laws produce a meaningfully more competitive economy, though this remains an open empirical question.

Studies of OSHA workplace-safety regulation generally fail to find a statistically significant effect: workplace accidents are declining, but regression analysis isolating the counterfactual cannot attribute the improvement to OSHA, and regulatory agencies are often understaffed and cursory in their inspections.

Markets provide powerful incentives for workplace and product safety through liability/court systems, loss of market share, and compensating wage differentials for risky jobs—and these forces have grown stronger as information about employers spreads more freely (e.g., job-market blogs), making it hard to hide unsafe conditions.

Markets exhibit robustness by endogenously responding to externalities: congestion costs are partly under individuals' control through residential and job-location choices (people with high value of time choose shorter commutes), and noise externalities are mitigated by sorting (e.g., locating a School of the Deaf near Logan Airport on cheap land), so markets reduce externality costs even without policy.

Pollution regulation has produced real benefits (air and auto pollution have declined), but government's preferred 'command and control' approach pushed too far, so net estimates show the gains from reduced pollution are roughly balanced out by the higher production and product costs imposed—a cautionary lesson for heavy-handed climate policy.

The textbook justification for government intervention is that where a market failure exists, government is supposed to step in with policies (taxes, subsidies, regulation, or public provision) that make some people better off without making anyone worse off, thereby expanding the pie.

In practice, true (actual) Pareto improvements are very rare; most policy analysis relies on 'potential Pareto improvements' where winners could compensate losers and still be better off—meaning the pie gets bigger even though some people may actually be made worse off.

Monopoly creates a deadweight loss because, to maximize profit, the monopolist reduces output below the level where consumers would willingly pay the cost of additional units—forgoing exchanges that would benefit both monopolist and buyer—because lowering price to capture those marginal sales would require lowering price on all units.

Auto congestion is a consumption externality: a driver in peak periods delays other drivers and imposes a social cost without accounting for it, and the market cannot respond because no one owns the roads—government owns them—raising the question of what could make drivers internalize that cost.

Market failure has a precise economic meaning focused on efficiency: a situation where resources could be reallocated to make at least one person better off without making anyone worse off (Pareto optimality), as opposed to the colloquial sense of markets simply not delivering what someone wanted.

Public production projects (e.g., the Dulles Metro extension) proceed despite large cost overruns because the costs are diffused across taxpayers while benefits are localized—so each taxpayer says it costs little, enabling inefficient projects to move forward.