Frank H. Knight's Risk, Uncertainty, and Profit asks why profit should exist in a competitive economy at all. Suppose every buyer of productive services knew the future. Competition would bid each service up to the value of what it helps produce, and an organizer would earn wages and interest with nothing left over. Knight's 1921 answer is that profit and loss arise from imperfect knowledge of the future; change produces them only when it defeats foresight. His book repays slow reading because it turns that answer into an account of enterprise in which control tends to follow responsibility for guaranteeing other people's returns.
Its twelve chapters fall into three parts. Part I surveys earlier theories that traced profit to change or risk. Part II builds perfect competition strictly enough to show that even continuous change leaves no profit when it follows known laws. Its assumptions, such as complete foreknowledge and agents too small to move prices, then mark where real markets depart from the model. Part III develops uncertainty, enterprise and the salaried manager before turning to social consequences. In his 1921 preface Knight separates explanation from defense. He says the net result is no defense of the existing order, and he denies that any single form of organization suits every purpose.
Knight distinguishes three kinds of probability. Two of them can be measured, either by a priori calculation over equally likely alternatives or by statistical frequency drawn from an empirical classification of instances. His third kind, estimates, covers cases where no valid basis exists for classifying the instance. Probabilities of the first two kinds can enter costs. Knight borrows an illustration from von Mangoldt about champagne bottles that burst in a known proportion. In Knight's cost account that breakage counts as an ordinary outlay of production. A building owner too small to steady his own fire losses can join an insurer, and grouping does the rest. Knight treats the boundary with estimates as a question of degree: no class is perfectly homogeneous, and practically no judgment lacks every basis for comparison.
His hypothetical manufacturer shows why the distinction is important. Weighing a large expansion of his works, he figures what he can measure and arrives at an estimate. Comparable expansions are too few and too unlike his own to yield a tabulated probability that the estimate is wrong, though partial comparisons still inform it. He also judges how far to trust his estimate, and Knight treats that confidence as part of the decision. Evidence about the person then carries weight: a forecaster's past record supports a presumption about his judgment. His long excursus on consciousness reflects 1921 psychology and can be skimmed.
In Knight's account an entrepreneur's receipts mix contractual income, such as wages and returns on property he owns, with pure profit, the residual left after paying the market price of every service he guaranteed. Accounting profit serves legitimate purposes of its own. Knight's concept also charges opportunity costs, and within his theory of profit the contextual benchmark for investment the owner could move elsewhere is the competitive return on safe securities. If a hypothetical firm's specialized machinery would resell for a fraction of its cost, that benchmark applies only to the transferable value. Luck enters too: a single profitable venture is weak evidence of ability, and profit carries no guarantee of moral merit.
A salaried manager seems to direct without bearing risk. Knight answers that in large organizations the decisive act of control is choosing the people who decide, so responsibility rests with the judgment that placed the manager. He treats this contrast as relative, since managers and workers exercise real judgment too. Hired entrepreneurs usually own a share or keep their posts only while the business prospers. Employees who accept imperfect security share both the uncertainty and some of the control. Property is the usual guarantee, though earning capacity or a relative's backing can stand in for it. This arrangement depends on judging general ability, a mapping Knight leaves uncertain. In a footnote he concedes that his theory of enterprise is vague and incomplete.
For Knight, a service's product is whatever depends on it in the situation as it stands. That causal relation holds for machines and harmful acts as much as for virtuous ones, and he calls confusing it with desert a serious error. On his account, payment is ethical only insofar as it is needed to get the service performed. Agreements also affect outsiders, which in his view justifies public support for education and communication and raises hard problems of land use. Readers can reject those remedies and keep the distinction.
His closing chapter finds that research, grouping and control all consume resources and that organization curtails the freedom of those organized. Reducing uncertainty can itself be an uncertain venture. Knight distinguishes freedom of contract from the choices people have in practice, which depend on initial ownership, luck and effects on third parties. Workers, he argues, risk their health, their lives and specialized training without matching control or fair compensation. His comparison of private and state incentives is conditional, and he finds bureaucracy more conservative than reckless. He suspects that entrepreneurs as a class earn less than competitive returns on what they supply, an opinion he rests on statistics he calls inconclusive.
Knight's later prefaces revise the argument. In 1933 he sets price analysis beside institutional history and ethics. In 1948 he abandons the production-period view behind the original text and measures profit over a dated accounting period as receipts minus disbursements, adjusted for the change in asset values. That preface also warns that large extensions of state economic control threaten political liberty. It grants as well that free enterprise tends toward inequality and concentrated power, which political action may need to counter. By 1957 he calls profit a residual that can be negative and no reward for risk-taking. He denies any connection between profit and the use of property. As a case containing every essential of entrepreneurship, he offers two workers on a project, one guaranteeing the other a return. These restatements change parts of the 1921 argument, and the editions do not add up to one consistent theory.
Knight's book suits economically literate readers working on firm governance or uncertainty research who need the risk-uncertainty distinction with its gradations intact. It is not a startup handbook and promises no business success. Expect abstract models, argument by definition and thought experiment, dated psychology and a few period remarks on race. Knight wrote outside the Austrian school, and readers trained on Böhm-Bawerk or Mises will find a rival capital theory. Read the risk-and-uncertainty chapter first. Augustus M. Kelley's 1964 reprint, which includes all four of Knight's prefaces, is available from the Mises Institute: Risk, Uncertainty, and Profit, 1964 Kelley reprint (PDF).