(This is an essay about Tom Sowell’s Knowledge and Decisions. I prepared it for a panel discussion at the Hoover Tom Sowell celebration, which will necessarily be much shorter. This will get edited and updated for Hoover, and I’ll keep an updated pdf version on my webpage. This is too long for email, so come to grumpy-economist.com if you run out of email space.)
The Giant’s Shoulders
Suppose that a frost in Florida kills half the orange crop. How do we adjust? People have to drink less orange juice. Those who like grapefruit juice should substitute more, so those who really love their OJ or take statins can still drink it. Some should use frozen for a while. Oranges from California should be shipped to East Coast destinations. And so on.
How does all this happen? Is there a government marketing campaign “drink less OJ for a while,” as California attempts with water? Does a government orange juice board equitably distribute the scarce supply to the most needy?
No. Prices silently guide people, who ruthlessly act in their selfish interest or to maximize profits of their companies, to make all these adjustments in the most efficient way.
Prices convey information, including what we now call “private information,” like who can more easily get along without orange juice for a while or what plot of land is best suited for growing oranges. Prices are the wires that connect the processors of the economy. Prices are the pheromones that socially organize our ant colony. Prices are signals, wrapped in incentives. Prices are efficient, sufficient statistics for decisions, giving just enough information. You don’t need to know whether there was a freeze in orange country, or a train derailment, or a new fad raising others’ demand. You just look at the price, and your own self-interest, and you act with the socially efficient response.
So wrote Friedrich Hayek (1945) in his celebrated essay “The Use of Knowledge in Society.” I read this essay, as many economists do, in graduate school in the early 1980s. I thought it was wonderful. Hayek undeoes a grand conceit: Introductory economics courses teach the “first welfare theorem:” that the market, if absolutely perfect, can just achieve the same result as a benevolent planner. The classes quickly move on to describe how hypothetical planners can solve myriad market “imperfections.” But planners can never have or process the necessary information. The theorem should be rewritten: Planners can never hope to achieve what markets do.
But I also noticed that despite the then 40 year passage of time, few others seemed to have followed up on this spectacular insight. Hayek seemed to stand alone. Every other paper we read spawned thousands of followers. Where were Hayek’s followers? (They were there, of course. But unlike Hayek, “Austrians” never made it to the academic mainstream that gets assigned in graduate schools.)
Then I found the just-published Knowledge and Decisions, Tom Sowell’s economic masterpiece. (I read several other eye-opening Sowell books at the same time, especially his works on race, and culture. But it is not my task to share that experience today.) This was a world-view-changing experience, a few crucial steps down the yellow-brick road that made me the classical liberal with many adjectives that I am today.
Sowell encyclopedically expands on Hayek. But he also drew on many others. Vaugh (1981) writes that Sowell
Systematizes much of the fragmented work of the last twenty years on such topics as the economics of property rights, law, regulation, bureaucracy, public choice, and the evolution of economic and social institutions.
Others note that Sowell synthesized the “UCLA view” of the economy, combining a bit of Chicago and a bit of Austrian economics.
Sowell studies markets and prices, sure, and all that goes wrong when you silence prices. But the heart of his book studies other social organizations, including political organizations, courts, corporations, regulatory agencies, universities, bureaucracies, cultures, communities and even families. Each of these organizations also produce, procure, validate, and transmit information to “decision-making units,” which then face varying incentives to acquire and use information in private or socially efficient way — or not.
The result is a comprehensive and novel view of social organization, including economics. In traditional economics, as repeated in introductory classes, we regard the economy as a machine for producing and allocating scarce goods and services. No. Information itself is the important scarce good in the economy. When you buy even a pencil you are really buying the skill and knowledge of the lumberjack, sawmill operator, carbon miner, pencil machine designer, lawyer, accountant, store operator, and so on. Much traditional analysis of political and social institutions focuses on morals, power, social justice and so on. No, information and incentives are the key determinant of political structure and outcomes.
Sowell’s style is rich by profusion of example, though each synthesized in a general statement. He richly contrasts the actual operation of poltical instiutions with the folly of common enthusiasms that ignore incentives and information in the quest to assign decisions and power to the state. “Central Planning” is so passé that nobody would advocate it by name, yet its poisonous successors travel the world around us.
Knowledge, Decisions, and Incentives
Sowell gives a detailed vocabulary to talk about information, which I will try to highlight.
Modern society organizes a vast amount of dispersed knowledge. “Primitive savages” knew a lot that few of us do. We wouldn’t last long in their world. Yet each one knew pretty much the same thing. Each must make his own spear. Each of us knows a tiny slice of a vast mosaic of complex knowledge. We can fly a plane, but few have any idea how to design one. The magic of AI is only the latest small addition to this set of communication mechanisms.
Ideas must be produced. Discovery gets the headlines, but the social process of authentication is more important since there are so many bad ideas. People must sift through ideas, boil them down, and figure out which are the few good ones. Authentication is often a social and cultural process, even in science. Science too is social, after all, and scientific method just a particular cultural etiquette. Since Sowell wrote, “science says” politics in climate, gender issues, public health and more only emphasize this fact. I would add that prioritization is an important element of this social consensus building. We are awash in ideas. What’s important, and to the problems we each face?
Cultural, legal, and educational institutions refine and pass along consensually authenticated knowledge, freeing us from rethinking every decision from scratch. Your grandmother really was right about the value of getting a job, marrying a nice girl/boy, having children, and saving. Cultural and institutional knowledge, like prices, is informationally efficient, passing along the lessons of immense experience but without that experience itself. In 1340, farmer Smith and farmer Brown might have had an altercation about Smith’s goats grazing Brown’s pasture. In that and thousands of controversies and experiments, the institutions of property rights developed. (Some are deeper, possibly genetic as well as cultural. Monkeys and dogs understand property rights.) We can benefit from that experience with a simple legal and cultural commitment to private property, without remembering the actual cases.
AI is a good metaphor for social as well as human intelligence, The institutions remember and pass on an appropriate action and behavior, without the specifics of the “training data” behind it.
That informational efficiency leads to some fragility. Purists, from peaceful libertarians to religious reformers to Marxist revolutionaries, are tempted to rethink all law and cultural norms from scratch, since they are passed on only the norms but not the experience underlying it. This sensitivity to embedded knowledge leads Sowell, like Hayek to a conservative distrust of rapid change, especially by force. “Imagine no possessions” (Lennon 1971) forgets thousands of years of experience. Nobody in a market has to understand how markets work — as they evidently do not! The tough nut of democracy is that a majority of voters have to understand a bit of how the system works and why.
But not all laws, cultural norms, and institutions are wise. Experience and cultural consensus often leads to fallacies, from slavery and bleeding the sick to the contemporary luxury beliefs of the woke. Our social and political mechanism for authenticating ideas is rife with lies told in the name of power.
Throughout, Sowell emphasizes feedback, how institutions adapt to new ideas and experience, or fail to do so. Cultural institutions are more susceptible to feedback than laws, and laws more than constitutions.
That varying rate of feedback is sometimes wise, as new consensus beliefs are often false. The founders of our pretty good political structure balanced feedback with checks, balances, and brakes. We don’t amend the constitution lightly.
In fact, though “conservative,” most of Sowell’s book is a devastating attack on the consensual knowledge and institutional structures of his day. Much has gotten worse, exactly as he foresaw. This is the eternal tension of a conservative libertarian. But it does not imply a contradiction. We are part of the feedback that makes institutions better. One can simultaneously value the wisdom of institutions and advocate radical change.
Culture vs. economics is a tension in Sowell’s work, and also between him and other analysts, especially of the fortunes of Black America and other less prosperous groups. How much is the effects of culture, and how much is the effects of incentives, such as the awful incentives offered by the welfare state? Cultures evolve, and respond to incentives.
The main theme of the book is not just that our social and political processes have evolved to incorporate a lot of bad ideas, but more deeply just how dysfunctional our social and political processes for authenticating ideas and incorporating feedback have become. The ideological madness of the last decades has only followed Sowell’s gloomy forecast. Cults and bubbles too have social and cultural information verification systems, that go wrong.
Knowledge underlies decisions. Knowledge must be in the right place and time, available to the right “decision-making unit,” and that unit must have the right incentives to make proper use of the knowledge.
There is nothing Sowell dislikes more than claims that “society” should do this or that. The right question is who and how the decision will be made. And not “who” in the political sense but “through what processes and under what incentives and constraints, and with what feedback mechanisms to correct the decision if it proves to be wrong.”
The point is perhaps most important in political organization.
One of the peculiarities of the American Revolution was that its leaders pinned their hopes on the organization of decision-making units, the structuring of their incentives, and the counterbalancing of the units against one another, rather than on the more usual (and more exciting) principle of substituting “the good guys” for “the bad guys”…
250 years of the latter has not yet brought down the former, but they’re trying hard.
Academic economists almost always ignore this question. We study what the hypothetical perfectly informed benevolent dictator should do. And usually that means “us,” if only someone would put us in charge. Until someone does, and then academics seldom perform better than others put in a system full of informational frictions and incentive distortions.
Catastrophic failures of information and decision are illuminating. At Pearl Harbor, the US military had the information that an attack was coming but it did not get to the right place at the right time. Secrecy and verification plague military information. The Vietnam war comes up often. Sowell might have analyzed Iraq, both sides of the Russian invasion of Ukraine, or Israel on Oct 7 2023 in the same way.
Each “decision-making unit” often does not make a final controlling decision, but decides on a move in a game with others.
Even in the extreme case of slavery,… The practical limitations of sheer subordination were repeatedly demonstrated by the various economic incentives which had to be resorted to …, especially for getting higher quality work performed.
Knowledge can be power.
…the lower level unit can use its knowledge advantages to evade, counteract, or redirect the thrust of orders from its nominal superiors.
That’s now called the “deep state.” Costs of information pervade the analysis. It would be prohibitively costly for a higher unit of a hierarchy to obtain the information — and doing so would just create a new hierarchy. The Provost cannot read all tenure candidate’s papers. The mice will play, cat or no cat, since they know what the cat cannot see.
The top, however, generally has greater larger-scale knowledge and more accountability. The lieutenant knows the terrain facing his platoon; the general knows the overall strategy and where the supplies are; the President knows the political goal. The point of the hierarchical bureaucratic organization, from military to regulatory agency to corporation to university is that this difference of knowledge is inherent. They must coordinate their actions though they can never have each other’s information. Perhaps Sowell’s point is that there should not be hierarchical organizations, and in many cases he may be right. But the optimal size of the firm, family, government, military, and university is not atomic.
Sowell distinguishes general knowledge, which is easy to transmit, from specific knowledge which is not.
…agricultural production involves much highly specific knowledge about the characteristics and contours of particular plots of land,…By contrast, an expert on the manufacture of steel can specify the exact quality of steel that will be produced by given combinations of iron ore and coal at given temperatures. For these reasons, steel production has been successfully centrally planned and controlled in various countries, whereas agricultural production has had such chronic problems and periodic disasters in centrally planned economic systems that even the most centralized communist governments have had to make major exceptions in agriculture, allowing decentralized decision-making of various sorts.
Even steel and rockets weren’t that successful, as variety, innovation, and cost eluded the Soviets, but the point is taken.
The scale of a bureaucratic organization depends on successfully separating informational needs. Sowell contrasts the scale of grocery and department store chains, which sell “standardized items,” with fancy restaurants, in which the owner must pay “constant attention to the demeanor of the staff and the delicacy of the chef.” There are now chains, like McDonalds, Applebees, and Starbucks, that prosper by offering better and more consistent quality than what sole owners can provide. They have standardized management training, accounting, HR, food preparation, and customer service, that is not as good as the most talented owners can provide, but much better than attentive but less skilled mom and pops can do.
A decision-making unit must have information, but must also have the incentive to act correctly on that information. In a market, prices communicate information and individual self interest provides the incentive. Social, political, and bureaucratic systems have constructed incentives, that sometimes work and sometimes do not.
I like to call what we do “incentive economics,” rather than “free market economics,” because, as in the entirety of this book, our core insight is that incentives—facing economic budget constraints and, here, information constraints--rather than moral character explain most decisions. We are more alike than we think, and would do things we find puzzling in others given the same incentives and information.
Many people bemoan captured bureaucracies, that end up offering protected profits to regulated industries, as moral or ethical failings. Captured bureaucracies simply respond “very rationally and ably” to the incentives offered by the regulated firms. As a result,
Reform by ‘throwing the rascals out’ seems less promising than reform by structure of incentives.
That lesson has only become truer as the scope of bureaucracy has increased. Another generation of bemoaning regulatory idiocy may eventually sway voters, but only a change in the structure of regulation is likely to change the outcome. Of course the incentives of who does that changing poses the eternal problem. Beware passive voice and “we,” just as much as “society” as the subject of political sentences.
Since Sowell and the public choice school wrote, regulatory bureaucracy has evolved from capture by industry to a two way capture by politics. Regulators and the administration that commands them can threaten the arbitrary persecution to coerce companies to support political efforts. It would be interesting to hear Sowell’s analysis of this development.
Bureaucratic organizations need to create artificial incentives, in the absence of the utility and profit maximization of market actors. This is always imperfect because of constrained information. Measures such as “attendance records, test scores, output per unit of time, seniority, etc.” are only correlated with the desired behavior, and necessarily rough given as always the constraints of information. But people then reach for the incentives, not the desired outcome. Teachers, paid for test scores, teach to the test not to educate the student. Professors gut their classes and hand out pizza and easy grades to boost teaching ratings. Sowell uses the classic example of body count measures in the Vietnam war. Body count is correlated with winning, but winning without slaughter is better, and safer for ourselves as well. Bodies of useless enemy count as much as the more difficult ones that matter.
Counts in arbitrary units of time lead to their own disincentives, as every academic trying to figure out how to spend down his or her research budget or grant before the end of its term will testify. The Soviets were famous for “storming,” producing large amounts at the end of the month or plan period with little regard to quality or plant maintenance, to hit the target.
Slave overseers in the antebellum South …overworked both men and the soil in the interest of current crops at the expense of reduced production years later—when the overseer would probably be working somewhere else.
Incentives rule the world:
…similar structures of incentives produced similar results, even in socioeconomic systems with widely differing histories, ideologies, and rhetoric.
Sowell categorizes decisions as well as information. Some are binary, yes or no, guilty or innocent, while others are continuously variable or incremental tradeoffs, how much ice cream one eats. Law and politics features many of the former, while economics features many of the latter, often a good reason for allowing economic decisions. Some decisions are once-and-for-all, while others are easily reversible. A literature on “real options” analysis expands on this insight. Some decisions are made individually, others as package deals. “…in choosing between political candidates, one must choose one candidate’s whole package—his fiscal policy, environmental position, foreign policy, civil liberties views, etc.—as against the whole package of his opponent’s positions on the same subjects.”
When decisions move from one kind of decision-making unit to another, that category of decision-making process can change. When decisions are moved from market choices to judicial decisions they become all or nothing rather than continuously variable. Writing in the introduction to the second edition, Sowell expands on this theme:
The distinction between incremental trade-offs and categorical “solutions” has been highlighted by recent trends in laws, policies, and judicial decisions creating such categorical goals as protecting endangered species, eliminating the last “vestiges” of segregation, creating innumerable “rights,” and promoting “safety” of many sorts—all this not explicitly at all costs, but often in practice treating costs as somehow unworthy considerations to be almost always over-ridden.
Categorical decisions lead to ignoring tradeoffs and budget constraints, as well as running roughshod over continuously variable decisions.
One particular kind of safety, for example, may be achieved by creating vastly greater dangers of another kind, as when pesticides are banned to eliminate their residual dangers in the environment, at the cost of a thousand-fold increase in the incidence of deadly, insect-borne diseases, such as malaria.
Nuclear power has been regulated to the point that plants must spend millions of dollars to reduce potential radiation losses far below the danger posed by a banana. The result has been vastly greater carbon emissions. Electric vehicle mandates plan to ban all internal combustion engines, even in the most suitable applications — a fire truck, say, that must operate for days at a time fighting fires in remote areas. Moral crusades brook little compromise. Much policy has taken on that character lately.
When decisions move from informal, negotiated under the discipline of competition, to formal but legislated, to the courts, and especially the Supreme Court, that moves the “locus of decision-making” away from feedback, flexibility, and continuously-variable to all or nothing. But the all-or nothing of the law then faces the gritty reality that thousands of decisions are all different, and law must write rules for each case. For example,
In the case of Griggs v. Duke Power Co… the U. S. Supreme Court in effect transferred the decision as to whether to use mental tests in hiring decisions from employers to the federal courts in general and, ultimately, to itself in particular. The basis for this sweeping transfer of decision-making authority from thousands of highly disparate and complex settings across the country to nine individuals in Washington was the simple plausibility of the idea that the validity of a test depended on its having a “manifest relationship to the employment in question” which could be demonstrated conclusively to third parties. Reasonable as this might seem at first, it presupposes far more knowledge than anyone possesses…. Most decisions in most aspects of life cannot be demonstrated conclusively to third parties.
Sowell notes already in the early 1990s the beginning of a reversal of this trend. Roe v. Wade created a single set of decisions for the whole country on just where the legal boundaries of abortion would be. Dobbs returned that decision to state legislatures, who negotiate a continuous set of incremental restrictions. Though contrary to my policy preferences, one must recognize a “who makes the decision” structure with far more and more timely feedback. The disappearance of the issue from the open wounds of Washington partisan warfare is just as remarkable.
Another reversal:
A landmark in this new trend was the 1995 case of United States v. Lopez…the simple, common sense conclusion that someone carrying a concealed weapon near a school is not engaged in interstate commerce came as a thunderbolt.
One may pray for the day when it occurs to the court that growing wheat in your own backyard to bake your own bread (Wyckard v. Filburn) deserves the same treatment.
The important consequences of the fact that government bureaucracies face categorical mandates rather than incremental (marginal) trade-offs in the presence of diminishing returns, which are ubiquitous:
Categorical mandates such as “eliminate pollution” and “eliminate racism” make it
virtually inevitable that governmental agencies would eventually end up doing things which seem irrational as isolated decisions…If the agency is supposed to fight discrimination against minorities, it must successively expand its concept of what constitutes “discrimination” and what constitutes a “minority.”…Urgent tasks such as securing basic civil rights for blacks ultimately give way to activities designed to get equal numbers of cheerleaders for girls’ high school athletic teams….The ineffective record of “affirmative action” policies is in sharp contrast with the record of “equal opportunity” laws in the years immediately preceding.
One reason for the difference was the different set of incentives presented by the two policies. “Equal opportunity” laws provided penalties for specifically proven discrimination. “Affirmative action” laws penalized numbers that disappointed administrative agencies, and made defenses against “rebuttable presumptions” costly and uncertain.
Decisions may also be instantaneous or sequential. Sowell notes many sequential decisions gone wrong, ending up far from where they started. He offers as examples the Vietnam War,
Many early supporters of the Vietnam war came ultimately to the position that it was not worth the cost, after the full cost had been revealed by time, …
school desegregation
…the progression from The Supreme Court’s Brown decision in 1954 that the state cannot classify students by race for differential treatment to its controversial “busing” position in which that is what it requires states to do. Years of opposition to desegregation of the public schools led to progressively tighter judicial control, designed to overcome the various strategies of opposition, delay, and evasion—ultimately arriving at a point the opposite of the court’s original premise or intention.
and the outbreak of WWII. In understanding similar phenomena, I refer to the nursery rhyme of the old lady and the fly. She eats a spider to catch the fly, a spider to catch the fly, a bird to catch the spider, and so on until she swallows a horse, and dies. Of course.
Contemporary examples abound. Banking regulation started with a simple idea: bail out depositors to stop a run, regulate bank risks to constrain the consequent incentives for bank and depositor risk taking. Banks got around regulations, larger and larger sets of assets were bailed out, and we now sit with hundreds of thousands of pages of regulations, interpretations, and an army of supervisors under the Dodd-Frank act and Basel, unable to see plain-vanilla interest rate risk at Silicon Valley Bank, or to be used at Credit Suisse. Few at the outset saw this coming, and if any did they would have been dismissed as paranoid slippery-slope fanatics.
Health and health insurance regulation started with a tax deduction for employer-provided health care, and free care for wounded soldiers. Each scheme exploded with unintended consequences and behavioral responses, leading to the immense mess we have now. Few at the outset saw this coming, and if they did they would have been dismissed as right-wing paranoids.
The income tax started as a simple device to raise revenue in war from a few very wealthy people. But then you have to tax corporations to keep people from incorporating, and one patch after another leads to today’s obscene mess.
The euro started as a simple common currency without fiscal union, and evolved in one crisis expedient after another to a massive fiscal transfer scheme that horrifies its founders (Cochrane, Garicano, and Masuch 2025).
To Sowell,
Praise or blame is not the point. What is important is to understand (1) when a situation facing us is part of a sequential decision making process, and what that implies, and (2) to understand when our own institutions set up sequential decision-making processes when there is an alternative decision-making process available.
Here though, I found myself questioning. I think that’s just what Sowell would want: question, analyze, refine. Some decision-making processes ought to be deliberately sequential. Sequential decisions are also reversible, in the Sowell taxonomy, offering the chance to avoid a disastrous chain of events. Indeed the usual story of the Vietnam war is not the inevitability of disaster, but regret at the many ignored off-ramps.
The FAA, after analyzing many airplane crashes, notes that most follow a similar chain of events, in which one mistake after another compounds, each one offering an off ramp. The pilot may not feel that well, accepts a long flight in bad weather, proceeds as the weather worsens, ignores a malfunctioning instrument, accepts an unsuitable approach, and ends up crashing. The final misapplication of controls is the direct cause, but pilots are trained to recognize a sequence of events and get off the ramp early.
With this tension in mind, I might add another categorization. Sequential decisions lead to disaster when the short-run cost-minimizing outcome of each small bad decision is to continue to patch the initial bad decision. Reversing course means starting all over, and would be enormously expensive. Regulation fails. Adding more regulation is not stupid. It is the short-run easiest way to patch a leaky boat. Sequential decisions benefit from increased real options if it’s easy to get off the train and start over.
Structuring sequential decisions offers suggestions for reform. On my mind, the Administrative Procedures act specifies cost-benefit analysis (often ignored) and extensive public comment before a regulation is implemented. Afterward, however, these evaluations and decisions are fixed indefinitely time. There is no structure for evaluating how a regulation or program actually works, collecting actual costs and benefits, adjusting the incentives and who-decides, in the light of experience, other than an administration taking on the task to restart the regulatory process. A sunset requiring regular review might well provide the feedback needed to make a properly sequential decision, given the lack of information at the outset.
Relationships and organizations can be informal and voluntary, such as friendship, dating, and club membership, or formal and enforced by institutions that can assess behavior by rules, impose penalties, such as marriage, rental contracts. Of course even voluntary relationships involve investments in which current decisions lead to future costs and benefits. And we voluntarily enter in to enforced relationships, as a form of precommitment. Marriage is the obvious example.
Sowell naturally praises the flexibility, feedback and ability to incorporate private information in informal relationships. “Lindbergh’s flying across the Atlantic alone was less of a feat than if he had flown across the Atlantic with a committee.” Small businesses are financed by family and friends who know the character of the entrepreneur. The theory of finance adds that such businesses can also obtain collateralized debt financing, but not equity financing.
In this vein, Sowell has kind words for word-of-mouth referrals, “old-boy” networks and other methods of hiring good employees that are often excoriated by social justice advocates. They exist not just to perpetuate the advantages of the elite, but also to allow the employer to use “soft” information known by people who also wish to keep their relationship with the old boy intact. “Employers who hire new employees by word-of-mouth referrals from existing employees get around the problem confronting banks” “Decisions made through informal relationships can be more readily individualized than in an organization bound by its own rules.”
This insight is a part of the debate between government social programs vs. charity (the latter, perhaps subsidized in general terms). Social programs must dispense money by rigid rules, which can never really evaluate the core question, whether a person in needs is unlucky or just lazy. Families, churches, and communities are far better able to tease out this private information.
…a worker suffering a minor injury of a sort described in the rules may receive a windfall gain, while another worker psychologically devastated by the ending of a love affair is expected to continue carrying out all duties as if nothing had happened.
Dates are more likely to be made in cultures where this implies little beyond a short-run commitment to be at a certain place at a certain time, than in cultures where overt expressions of interest in an individual of the opposite sex, or subsequent displays of affection toward such individuals imply matrimonial intentions—and where failure to follow through brings social ostracism or even risk to life and limb.
Dating is, in the US, a sequential decision with many off-ramps, but still leading towards marriage. Many countries of Sowell’s latter type feature marriages arranged by family, who can often use “old boy network” soft information to learn about character and suitability that daters struggle with. Which is better? It is a debatable question, but a good one.
Foreign investments are more likely to be made in a country where the proceeds can be withdrawn at will in convertible currency than in a country where legal barriers make this impossible or political barriers make it costly. Similarly, the existence of such instruments of future decision variability and reversibility (i.e., nonprecedence) as brakes and steering wheels is all that makes most people willing to ride in automobiles at highway speeds. Liquidity of assets and the existence of options markets serve similar functions in the economic sphere.
Property rights, which allow investment, are a good example of formal institutions that impose costs and precommitments for good effect, despite the later inflexibility. Non-reversible decisions are also good in some cases.
Someone who is going to work for many years to have his own home wants some fairly rigid assurance … that he cannot be dispossessed by someone who is physically stronger, better armed, or more ruthless, or who is deemed more “worthy” by political authorities.”
Admittedly, I have just touched the surface, both of Sowell’s ideas and the breadth of his applications. The book is not easy to summarize, which is both praise and a limitation. But I must respect the reader’s patience.
The book is not just a dry analysis of institutions, but up front on the conservative / libertarian critique of government that follows. Government institutions miss the incentive and feedback mechanisms that of profit maximization (or losses!) Induce; it has little authentication of assumptions; dispersed knowledge does not flow easily. These problems are
…compounded when the assumptions underlying social policy take on the status of fashionable axioms, which become immune to contrary evidence (as for instance the assumption that school busing promotes racial integration and is desired by black parents)
This tendency has gotten a lot stronger! That “woke” resembles a religious cult as well as fashionable status-marking luxury beliefs has been noticed by many commenters.
The book is full of thoughtful asides from the main topic. One stood out. Some libertarians are pacifists and isolationists. I admire the purity of their view, but not its practicality. Tom likewise.
“Freedom has cost too much blood and agony to be relinquished at the cheap price of rhetoric.”
“If you are not prepared to use force to defend civilization, then be prepared to accept barbarism.”
“Everyone may be called “comrade,” but some comrades have the power of life and death over other comrades.”
Capitalism the only “ism” that is not fundamentally based on the use of force to take A’s property or labor and give it to B, with the state taking a cut along the way. Capitalism is freedom. Thankfully, it is also informationally efficient. If we had to choose between freedom and prosperity under the yoke of socialism, our choice would be a lot harder. But capitalism must be defended, and by force if needed.
The Conversation and the Future
Knowledge and Decisions did not emerge from a vacuum, nor does it speak in a vacuum. But it has not had the influence I think it deserves.
This is not a dead book, to be admired on the shelves for its one time accomplishments, This is a live book, which can and should inspire new thinking today. As Hayek inspired Sowell, and left plenty for Sowell to do, so Sowell should inspire a new generation of economists, political scientists, cultural critics, and public intellectuals to stand on his shoulders, and write the next great book on information in markets and social and political organization.
Contemporary economics largely ignores Sowell’s insights and approaches. It is also, like the economics of Sowell’s time, a target-rich environment. Contemporary information economics is full of criticism of markets and other decentralized mechanisms, leaning unthinkingly to state intervention, just as the economics of Hayek’s time ignored the information problems facing the planner. It ignores look-out-the-window realities, and has not been subject to the kind of validation or refutation that Sowell’s fact-based analysis can offer. Sowell also left quite a few important questions open, especially how to reform our social and political systems, and, as Sowell might put it, how and by whom that decision can be better made.
Mainstream economics ignored Hayek for 45 years until Sowell produced his masterpiece. That it has largely ignored Sowell for another 45 years means that the ground is ripe for another masterpiece.
The 1970s and 1980s were the beginning of a boom in the economics of information. I will mention just a very few.
Fama (1970) is the centerpiece of the “efficient markets” revolution in finance, which investigates the hypothesis that asset markets “efficiently” incorporate all information available to investors, a classic bit of Hayekian information transfer. His instant critics complained that due to various irrationalities and frictions that the market did not do a good job (Shiller 1992).
Akerlof (1970) wrote in the “market for lemons” of a situation in which private information makes markets break down. While competitive prices induce people to consume the right amount, according to their privately-observable demand, Akerlof shows how people with private information about supply can ruin a market. If used car buyers cannot tell good cars from lemons, but used car sellers can do so, only the worst quality car can be sold.
In theory, beautiful. I long for a Sowell response. I might offer two words: Car Max. “Eppur is vende” (and yet sells) a modern Galileo might respond. Like many beautiful theories, this one suffers if one looks out the window, let alone confronts it with the kind of detail that a new Sowell could bring to it.
The lemons parable is the prime rationalization for our massive health insurance regulatory state, offered by leading experts for decades, such as the influential Einav, Finkelstein, and Fisman (2023) recently. People know more than insurers about their health, so healthy people will not buy health insurance. The market collapses. Or so the story goes. But really, now, do people, armed with their aches and pains, really know more than a health insurer, armed with your full medical records and tests, about the present value of their health costs can possibly know? In fact, if anything, asymmetric information goes the other way, but that is not what the theory is about. That can lead to a remediable pre-existing conditions issue, but it does not cause market failure. In reality, health insurers are prohibited from using information that they have. Adverse selection results from government mandate to ignore available information, not consumer’s inherently superior information or anything else wrong with markets (Cochrane 2015). It is waiting for a Sowellian response.
I wish Sowell had picked a little more on the academic information economics as well as the general progressive fallacies in popular currency. This “adverse selection” and “moral hazard” of supposedly distorted incentives due to information problems, arguing against Hayek, were already common currency in 1980. That fruit remains low-hanging.
Milgrom and Stokey’s (1982) “no trade theorem” is a version of the lemons theorem for asset markets. If someone offers to buy your Tesla stock, you should think, “surely he knows something I don’t know,” and refuse to trade. His bid rises until the price reveals his information. With no trade. As the joke goes of the senior market for economists, prices change, with no trading volume.
Related, Grossman and Stiglitz (1980) prove that informationally efficient markets are impossible. If markets are efficient, then there is no reward for collecting the information and trading on it which make markets efficient.
And yet it sells. There is a lot of trading volume. Grossman and Stiglitz also provide an answer, which dominates the theory of market microstructure to this day. Introduce some “liquidity” traders who buy or sell without information. Then the “informed” traders can hide a bit, and arbitrageurs will buy and sell by offering a bid-ask spread. But this is still rudimentary. It is not at all clear that society as a whole invests too much or too little resources into producing information to make stock prices efficient (Cochrane 2013). Is it worth billions of investment in fiber-optic cable for the S&P futures and index to match 20 milliseconds sooner? Understanding how information becomes incorporated into prices through a cacophony of trade, how markets come to a consensus, and the efficiency of this information-production structure is a great task for a new Sowell.
Contemporaneously to Knowledge and Decisions, and subsequently, Fama and Jensen (1983a,b) and Jensen and Meckling (1976) are emblematic of a huge literature examining how information and incentive problems pervade the boundaries of the firm, organizational structures within firms, nonprofits, governments and other organizations, as well as the nature of their financing. The “agency theory” in which a “principal” such as stockholder wishes to set up a contract for an “agent” such as a CEO to work on the principal’s behalf, though the agent has information the principal cannot possess, is an analysis very like Knowledge and Decisions. The vast “mechanism design” literature, including the famous “revelation theorem” (Myerson 1979, 1981) describes ways to set up auctions and other mechanisms so that people do, in fact, “reveal” their private information. “Personnel economics” (Lazear 2009) studies classic Knowledge and Decisions questions, as framed by Lazear and Shaw (2007):
Why should pay vary across workers within firms -- and how “compressed” should pay be within firms? Should firms pay workers for their performance on the job or for their skills or hours of work? How are pay and promotions structured across jobs to induce optimal effort from employees? Why do firms use teams and how are teams used most effectively?
Decision theory incorporates information limitations. A huge “real options” literature studies the value of “reversibility,” in sequential decisions, or keeping one’s options open. Spending a bit more to build a plant that can run on either oil or gas preserves the less costly option later. Robust control theory (Hansen and Sargent 2007) studies formally how to make decisions when you don’t even know what probability to attach to uncertain surroundings.
More recently, economic growth theory starting with Romer (1990) ends up being a theory of the incentives for the production and application of ideas. Growth in the end comes from productivity, and productivity in the end comes from ideas. But ideas are “nonrival.” If you use my car, I can’t use it. If you use my great recipe for spaghetti amatriciana, I can still cook dinner. Ideas are hard to develop, but easier to use when someone else has done so. And, beyond temporary patent protection for a tiny fraction of ideas, there is nothing stopping someone else from using ideas. Thus, growth theory concludes that a free market will generically invest too little in the production of new ideas.
Some growth theorists jump to argue for R&D subsidies. Sowell would object. One must subsidize the right, productivity-enhancing ideas. Subsidies for basic science have had some great successes, but governments have also subsidized terrible or just wasteful ideas, from religious investigation to decolonialization studies. Moreover, the step from the transistor to the iPhone, and the organizational improvements of big box stores, stem from applied and process research where government subsidy has never worked. Ideas for cutting costs and doing things more efficiently are most of productivity. I long for a Sowellian response, and I wish the authors clamoring for R&D subsidies had read Sowell.
Influence and the future
So there is a thriving economics of information. And it’s influential. After writing the above, I notice that just about every citation got a Nobel prize. And that is an unfortunate fact about this wonderful book. People who have read it love it, and report the same a world-view-changing experience I had. But it has not had the impact one would hope for in the academic culture that produces, authenticates, and passes on economic ideas that one would hope for. Knowledge and Decisions has 929 citations, only one of which is in the works I cite above. That’s pretty good, but not Fama and Jensen’s (1983) 36,000. Why not? Can we apply Knowledge and Decisions idea to understanding that outcome?
In part, it’s a book, and it’s words, not equations or regressions. His method is essentially historical. Much of the culture of information transmission in economics works differently.
Cynically, the secret of success is to develop a new methodology, publish it in a “top 5” journal, that answers an interesting question in the first instance, but which a generation of graduate students and followers can use in their quest to do economics, to write and publish papers, get tenure and high-salary offers, and be cited in turn. Maybe that’s not a great social process for developing ideas. Economic economics is pretty far from a market-based activity. But given what it is, that’s a partial explanation.
Less cynically, counterfactual history, theoretical equations, and cause and effect econometrics follow naturally as we solidify the kind of deep historical and conceptual investigation that Tom brings us. In the end, writing down math doesn’t mean it’s right, but if you can’t write down the math it is almost certainly wrong. We aren’t that original, but we are a useful part of the social authentication process. That Sowell chose to take the first step and leave us the latter is not criticism, it’s just comparative advantage. But it leads plenty for us to do. And the historical method is, I think, starting to have a resurgence in economics as we discover the limitations of formal econometric techniques, and if so Sowell’s analysis may have more readers.
But Fama and Jensens’s work is also just words. Here, perhaps, publishing a book and staying aloof from the academic subculture of journals, conferences, and PhD teaching, while enhancing its public impact, lowers its academic impact. Sowell of course doesn’t care. He consciously writes for a much larger general public. But academia is one source of our social consensus information distillation mechanism, and we we might care.
The book also mixes analysis of information problems with broader incentive problems, social and political commentary and philosophy. It is a comprehensive personal summary and world view. That makes it a great read, and a powerful experience for its fans as it was for me. But it is less likely that someone will use Sowell’s analysis in a narrower subsequent investigation.
The book is rich with examples and applications, which I have largely not passed on in the interest of brevity. But these examples are proudly offered as examples, cherry-picked illustration of the basic points. They are explanations, not testable propositions. Broadly, the approach is historical, and interpretive, rather than attempting to be “scientific,” as economics usually does, well or badly. That’s fine. Any investigation of a set of ideas must start by seeing if it offers a plausible explanation of what we see, and then move on to a plausible explanation of a wide variety of experience as Sowell masterfully does. But it makes the examples prone to selection bias. How many times have the same conditions not led to the same answers? Eventually, I hope Sowell’s followers will answer questions like this.
For example, as above, Sowell leads to the impression that all sequential decisions lead to bad outcomes, as the Vietnam war did. But thinking about it, I noted that sequential decisions are often reversible, leading to greater flexibility.
Reform
I wish Sowell were more often the reformer than the curmudgeon, much as I revel in libertarian curmudgeonliness. Bureaucratic organizations can operate better or worse. How are some, both private and public, relatively honest and efficient, and others rampantly corrupt and makework ratholes? Some Nordic bureaucracies seem better than ours, and many third-world bureaucracies much worse. Shouting into the wind for another half century just how idiotic the outcomes are and how bad the decision-making process is may not change the trend, as it has not much changed since 1990. My other Hoover Hero, George Shultz, always emphasized that we must tend to an optimistic, constructive end.
As “who will decide” (and how) is the right question, “who will decide who decides” (and with what set of incentives, information, and decision-making units) is even more important, and Sowell is often not clear about that. Regulatory agencies have poor structures. How will our political system reform them? I’m intrigued by proposals to change the regulatory process: shot clocks, sunsets, retrospective cost-benefit analysis, greater congressional review, and so forth. Most of us advocate moving regulatory activity from central, uniform, federal control to states. States should feel greater “feedback” and flexibility to adapt to local information. But a look at most state and local governments reveals just as much if not more dysfunction and less oversight. What would Sowell do?
Related, there is a real tension between “conservative” appreciation of how legal and cultural institutions encode a good deal of experience, and the classical liberal or libertarian despair at how the legal and cultural institutions of our day have evolved to what seem to us like dysfunction. Just what happened in about 1930 that led steadily improving institutions to get steadily worse? It is a familiar tension to Chicago, Austrian, UCLA, CATO institute, and other “spontaneous order” fans. When do we look at the world and see spontaneous order, and when do we look to see spontaneous dysfunction, remediable if only people would read our books and decry the idiocy of it all? Why is the rise of the left a spontaneous order too?
Summarizing the second half of the book, Sowell writes,
…the locus of decision making has drifted away from the individual, the family, and voluntary associations of various sorts, and toward government. And within government, it has moved away from elected officials subject to voter feedback, and toward more insulated governmental institutions, such as bureaucracies and the appointed judiciary.
But why has it so drifted, and what are we to do about it? Moving out one layer, why has the machinery of our political system moved in a dysfunctional direction?
Sowell’s trends leave one wanting to throw the rascals out and put better people in charge. But one cannot have a more anti-Sowell thought. How can we, as voters, citizens, and intellectuals, participate in this grand though frayed information processing system, to provide the feedback that it’s not working, and change the structure of the system so it will work better? How, for example, could the Administrative Procedures Act be reformed, to rein in the expanding arbitrary and stifling power of the regulatory state? What court cases could we bring? Can we get the Federalist Society, or create a new such society, to finally overturn Wyckard v. Philburn, the foundation of federal economic control, and proclaim that the federal government has no business regulating wheat you grow in your own back yard to bake your own bread? Are law and its interpretation insufficient, and we need to organize constitutional change?
We are not helpless. We live in a grad information processing machine, consisting of the least-bad constitutional structure ever created, surrounded by active interest groups, media, chattering classes, and thinkers like ourselves? We, sitting here today and talking and writing at the Hoover institution are part of that process, just as much as the Democratic Socialists of America debating in another room somewhere. How can we better play our part? What laws can we advocate passage, and organize interest groups to pass?How can we act in that system to bring Sowell’s and his followers’ insights to bear?
None of this is criticism. The book is a masterpiece as it is. Adam Smith wrote in much the same way. The pin factory was an illustration of a point, not a testable proposition about division of labor. “Testable propositions” are just one, rather convoluted and often quite misleading, form of rhetoric that actually accounts for surprisingly little of the adoption of even scientific ideas (Kuhn 1962, McCloskey 1983). Impact on academic literature via the strange rituals of academic culture, rather than a broader readership, is not the measure of a book. It just helps to explain my sorrow that more academics do not pay attention to the important lessons of this book for academic work. But it implies that there is much one can do to build on this pathbreaking book, rather than just remember warmly the experience of reading it and admire it on one’s bookshelf.
In the context of Hayek, Sowell writes
The hallmark of such a thinker is not personal notoriety but the fact that a hundred years after he is dead and forgotten, men who never heard of him will be moving to the measure of his thought.
Keynes had similar thoughts about the lasting influence of dead economists. Both Sowell and Hayek have moved past this stage; men (and women) will remember them. And if societies are not moving to the measure of their thought, they will have taken a dramatic turn away from freedom and prosperity.
Concluding Remarks
I close with a personal note. In these events, we are supposed to tell a few funny stories about our warm interactions with the celebrated person. But I have never met Tom Sowell, not in my 30 years as an economist at the University of Chicago, not in my 10 years at Hoover; not once since that fateful day in about 1981 that I started reading his books.
I am not complaining. Something I quickly learned about Tom is that, like so many great scholars who remain productive in later years, he jealously guards his time. The most legendary stories I have heard about Tom are, in fact, stories of his refusal to go to even the most appealing diversions open to celebrated academics and public intellectuals.
I respect and admire that. Tom shouldn’t waste his time hanging around the coffee room, chatting with me. Tom communicates well enough via the written word, using his eloquent command of the English Language. So, I will have to conclude without the funny stories that habitually accompany these events, but with great thanks to Tom for everything I have learned by reading his work over many decades.
References
Akerlof, George A., 1970. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism.” The Quarterly Journal of Economics 84, 488–500. https://doi.org/10.2307/1879431.
Cochrane, John H. 2013. “Finance: Function Matters, not Size.” Journal of Economic Perspectives 27, 29–50.
Cochrane, John H., 2015. “After the ACA: Freeing the market for health care”
in Anup Malani and Michael H. Schill, eds. The Future of Healthcare Reform in the United States. Chicago: University of Chicago Press, 161-201.
Cochrane, John H., Luis Garicano, and Klaus Masuch, 2025. Crisis Cycle: Challenges, Evolution and Future of the Euro.Princeton: Princeton University Press.
Einav, Liran, Amy Finkelstein and Ray Fisman, 2023. Risky Business: Why Insurance Markets Fail and What To Do About It. New Haven: Yale University Press;
Fama, Eugene F., 1970. “Efficient Capital Markets: A Review of Theory and Empirical Work.” The Journal of Finance, 25, 383–417. https://doi.org/10.2307/2325486
Fama, Eugene F., and Michael C. Jensen, 1983a. “Separation of Ownership and Control.” The Journal of Law & Economics 26, 301–25. http://www.jstor.org/stable/725104.
Fama, Eugene F., and Michael C. Jensen, 1983b. “Agency Problems and Residual Claims,” Journal of Law and Economics 26, 327-349.
Grossman, Sanford J. and Joseph E. Stiglitz, 1980. “On the Impossibility of Informationally
Efficient Markets” American Economic Review 70, 393-408.
Havender, William R., 1982. “Review Essay: About Knowledge and Decisions.” https://www.aei.org/articles/a-review-essay-about-knowledge-and-decisions/
Jensen, Michael C., and William H. Meckling, 1976. “Theory of the firm: Managerial behavior, agency costs and ownership structure.”Journal of Financial Economics. 3, 305-360.
Lazear, Edward P., 2009. Personnel Economics. Cambridge: MIT Press.
Lazear, Edward P., and Kathryn L. Shaw, 2007, “Personnel Economics: The Economist’s View of Human Resources.” Journal of Economic Perspectives 21, 91-114.
Hansen, Lars Peter and Thomas J. Sargent, 2007. Robustness. Princeton: Princeton University Press.
Kuhn, Thomas S., 1962. The Structure of Scientific Revolutions. Chicago: University of Chicago Press.
McCloskey, Deirdre (then Donald N.) 1983. “The Rhetoric of Economics,” Journal of Economic Literature 21 481-517. https://www.jstor.org/stable/2724987
Milgrom, Paul, and Nancy Stokey, 1982, “Information, trade and common knowledge,”
Journal of Economic Theory 26, 17-27. https://doi.org/10.1016/0022-0531(82)90046-1.
Myerson, Roger B., 1981. “Optimal Auction Mechanism Design,” Mathematics of Operations Research 6, 58–73.
Myerson, Roger B., 1979. “Incentive Compatibility and the Bargaining Problem.” Econometrica 47, 61–73.
Romer, Paul M. 1990. “Endogenous Technological Change.” Journal of Political Economy 98 S71–102. http://www.jstor.org/stable/2937632.
Shiller, Robert J. 1992. Market Volatility. Cambridge: MIT Press.
Vaughn, Karen I. 1980. Review of “Knowledge and Decisions” by Thomas Sowell.”Journal of Economic Literature 18, 1559-156. https://www.jstor.org/stable/2724067.
Excellent essay. Thoughtful and fair.
The most approriate quote from Hayek (The Use of Knowledge in Society) is, “we must look
at the price system ... as a mechanism for communicating information if we want to understand its real function”.