Money, Income, Prices, and Interest Rates

This 1992 American Economic Review paper by Benjamin Friedman and Kenneth Kuttner asks whether money remains a reliable indicator of future nominal and real activity, and whether the predictive failure of monetary aggregates that shows up in post-1980 U.S. data reflects a genuine structural break rather than a fluke of sample choice. Using quarterly U.S. data and reduced-form VAR-based Granger-causality tests (with a uniform four-quarter lag length and no structural identification) plus cointegration tests (ADF residual tests and Johansen maximum-eigenvalue/trace tests), they examine four financial aggregates – the monetary base, M1, M2, and total domestic nonfinancial credit – across three overlapping samples: 1960:2-1979:3 (pre-Volcker), 1960:2-1990:4 (full sample), and 1970:3-1990:4 (post-1970). In the pre-Volcker sample all four aggregates have F-statistics significant for nominal income at the 0.01 level; extending the sample to 1990:4 (1960:2-1990:4) causes the base and credit to lose significance while M1 (3.75**) and M2 (4.49**) remain significant at the 0.01 level, and starting the sample in 1970:3 (1970:3-1990:4) instead eliminates significance for all but a marginal M1 result (2.27, significant only at the 0.10 level) – a pattern that repeats for real income. Cointegration results track the same deterioration: ADF tests find M2 cointegrated with income only in the pre-Volcker sample; Johansen bivariate tests find the base, M1, and credit cointegrated with income pre-Volcker but no aggregate cointegrated in 1970:3-1990:4; and trivariate Johansen tests (money, income, an interest rate) find all four aggregates cointegrated pre-Volcker and none post-1970. Turning to a candidate replacement indicator, the paper shows the commercial paper-Treasury bill rate spread is significant for real income at the 0.05 level or better in every sample and specification, remains significant even when a monetary aggregate is included (at which point the aggregate itself becomes insignificant), and accounts for roughly 23-32% of real income variance in 1960:2-1979:3 and 21-26% in 1970:3-1990:4 in the systems built around the base, M1, or credit (the M2-based system shows a distinctly smaller spread share, around 12-15% and 14-22% respectively); further tests show this predictive power is not just a restatement of the bill rate’s own level, since the bill rate remains independently significant when the spread is included. The authors do not adjudicate between two candidate explanations for money’s failure – 1980s financial deregulation/innovation or the 1979-1982 Volcker disinflation’s change in Fed operating procedure – and they are explicit that all findings are reduced-form predictability results, not structural or causal claims.

Benjamin M. Friedman | Macro Paper Warehouse

Money, Income, Prices, and Interest Rates