Morgenstern’s paper is a methodological attack on a deceptively secure empirical foundation of international economics. Gold movements seemed unusually measurable: gold is physically definite, internationally traded, and central to gold-standard theory. Yet the paper’s thesis is that even this “least questionable” component of balance-of-payments statistics proves unreliable when subjected to a simple reciprocal test. If gold data fail, the larger statistical apparatus of international trade and payments is placed under suspicion.
It is the purpose of this paper to examine the statistics of international gold movements for their reliability and usefulness.
The introduction frames accuracy not as an abstract property but as something determined by intended use. Morgenstern insists that data adequate for broad historical description may be useless for theories requiring monthly timing, cyclical turning points, or correlations with exchange rates and interest rates. This is one of the paper’s central conceptual moves: the standard of measurement must come from the theory and policy claims the numbers are asked to support.
This standard, by common scientific practice, can only be derived from the uses to which the data are to be put.
The structure follows this logic. After defining the uses of gold-movement statistics, Morgenstern formulates a test: country A’s reported exports of gold to country B should match country B’s reported imports from A, allowing for transport costs, insurance, and lags. Because gold is high-value and often moved quickly, these allowances should be small, especially between neighboring countries.
The two should coincide, allowing for such factors as costs of transportation and time lags.
The empirical core compares official data for the United Kingdom, United States, Germany, France, and Canada in 1900, 1907, 1928, and 1935, using monthly figures aggregated into quarterly and annual totals. The results are not marginal discrepancies but radical contradictions: ratios fail to cluster around unity, countries report movements others do not, and the prewar gold-standard years fare especially badly.
The results shown are extraordinary.
Morgenstern treats the issue as central because the failure concerns precisely the data most often used to support fine-grained monetary theory. The charts show that even large transactions may diverge sharply between sources; the problem cannot be dismissed as clerical noise in small items.
The statistics are simply contradictory.
His deeper methodological argument concerns the absence of a scientific way to choose between conflicting official observers. If British and American figures disagree, there is no independent reason to privilege one merely because the investigator is British or American. Nor can theory rescue the data, since a plausible model may accidentally fit one erroneous series while contradicting another.
There is no reason why, say, an American economist or a man from Mars should prefer one set of these statistics over the others.
The explanatory section identifies causes without pretending they can be corrected mechanically: unstable lags, transit trade, secrecy, misclassification of gold and silver, incomplete coverage, traveler-carried coin, central-bank privilege, and especially earmarking, where ownership of gold changes without physical shipment. These complications make customs statistics and central-bank statements imperfect substitutes for one another.
They often are unwillingly produced by-products of other activities.
The conclusion is not that all uses of gold statistics are impossible. Morgenstern allows that very broad claims—such as the large inflow of gold to the United States after 1933—may survive. But the paper rejects their use for cyclical analysis, precise capital-flow inference, or delicate correlations. The required principle is proportionality between mechanism and measurement.
the data are as fine and as accurate as the mechanisms they are supposed to corroborate.
The relevance of the work lies in its demand that economists treat data validation as prior to theory testing. Morgenstern turns a narrow statistical inquiry into a general critique of empirical economics: if the best-looking observations are this poor, then conclusions built on more fragile trade, payments, and capital-flow data require renewed burden of proof. The paper ends with an admonition that remains its central lesson.
a good theory cannot be built on shaky data.
This work was divided into 10 sections when it entered the library's research corpus—an apparatus for search and citation, not necessarily the author's own table of contents. Each title opens its summary.
Put a question to this work; the Librarian answers from its 10 sections and cites the passage.
Ask the Librarian