On the Accuracy of Economic Observations, second edition, completely revised

Summary

Oskar Morgenstern’s On the Accuracy of Economic Observations is a revised methodological monograph on the reliability of economic statistics. Its central thesis is that modern economics, policy, and business decision-making routinely treat economic data as more exact than they can be. Morgenstern does not reject quantitative economics; he insists that it become more scientific by learning to state the probable errors, limits, and uses of its observations.

We must carefully distinguish between what we think we know and what we really do and can know.

The book’s argument begins from a conceptual distinction between numerical appearance and warranted knowledge. Economic figures acquire authority because they look precise, but their precision is often typographical rather than epistemic. For Morgenstern, accuracy is never an abstract property of a number alone. It depends on the purpose for which the number is used, the method by which it was obtained, and the consequences of error.

The very notion of accuracy and the acceptability of a measurement, observation, description, count—whatever the concrete case might be—is inseparably tied to the use to which it is to be put.

This is the book’s key methodological move: it shifts the problem from “having data” to knowing what kind of error surrounds the data. Morgenstern compares economics unfavorably with the natural sciences, where measurement is habitually accompanied by attention to tolerances, probable error, and the limits of instruments. Economics, by contrast, often publishes aggregates, indexes, national accounts, and other statistical series without making clear how uncertain they are. The result is a false sense of exactness in theory, forecasting, administration, and public debate.

The structure of the work is cumulative. The prefaces frame the revised edition as a warning against complacent statistical use. The main argument then moves from general principles of observation and measurement to the institutional production of economic data, including reporting practices, revisions, and the responsibilities of statistical agencies. Morgenstern’s recurring concern is that the makers of economic statistics often present figures as if their uncertainty were either negligible or the user’s problem.

To throw the burden of estimating the errors and the reliability upon the user, though exceedingly convenient for the maker, is a totally inadmissible procedure.

A distinctive part of the argument is Morgenstern’s insistence that economic observation differs from physical observation because the objects observed are human and institutional. Firms, households, governments, and markets may misreport, conceal, strategically alter behavior, or define categories inconsistently. The economist’s data are therefore vulnerable not only to technical measurement error but also to social incentives and deliberate distortion. This is the force of his contrast between nature and economic actors.

Nature may hold back information, is always difficult to understand, but it is believed that she does not lie deliberately.

Morgenstern’s criticism also bears directly on econometrics and mathematical economics. The danger is not mathematics itself, but the application of exact formal operations to inexact observations without carrying the uncertainty through the analysis. If the underlying data are weak, refined calculation may merely amplify the illusion of knowledge. The book thus challenges economists to connect formal theory with a disciplined account of observational error.

It is at best exceedingly unlikely that such accurate measurement is ever possible.

The book’s practical relevance lies in its demand that economic statistics be accompanied by explicit reliability estimates, error ranges, and statements of method. Morgenstern treats statistical revision as evidence that economic quantities are not final discoveries but provisional constructions. Published numbers may improve, but they remain subject to definitional change, new reports, and correction.

It is almost a statistical axiom that no revision is ever final.

In this sense, On the Accuracy of Economic Observations is both a critique and a research program. It calls for a culture of economic measurement in which accuracy is investigated rather than assumed, and in which users are not seduced by neat columns of figures. Its enduring relevance is that it exposes a central weakness of empirical economics: the authority of quantitative results depends not only on models and methods, but on the often fragile observations from which they begin.