Published Classic [Journal of Political Economy] doi:10.1086/261969 Vol. 102, No. 6, pp. 1228-1247

The Dynamic Impacts of Monetary Policy: An Exercise in Tentative Identification

Q: What problem with existing monetary-policy identification schemes motivates this paper?

Standard SVAR identification schemes for monetary policy -- Cholesky orderings, nonborrowed-reserves (NBR) innovations, or federal-funds-rate (FFR) innovations -- equate a statistical innovation in a single variable with "the" monetary policy shock, which implicitly imposes an extreme assumption about interest elasticities (treating money as perfectly elastically or perfectly inelastically supplied/demanded) without ever estimating those elasticities. The paper argues these assumptions produce anomalous dynamics -- a liquidity puzzle (a positive reserves innovation fails to lower the funds rate) or a price puzzle (a negative FFR innovation is followed by rising prices) -- because a single-variable innovation conflates supply and demand movements in the money market rather than isolating the supply-side (policy) component. The paper's question is whether jointly identifying supply and demand in the reserves market and in the M2 market can recover a monetary policy shock whose dynamic effects match traditional monetary theory.

Q: How does the paper identify supply and demand shocks simultaneously?

Two separate structural systems are estimated -- one for the reserves market, one for the M2 market -- each consisting of a demand equation (money as a function of its own interest rate, the price level, and output, with the interest elasticity a_1 restricted negative) and a supply equation (the interest rate as a function of money, the 10-year Treasury yield, and a commodity price index, with elasticity a_4 restricted positive), estimated by maximum likelihood following Bernanke (1986), Blanchard and Watson (1986), and Sims (1986). Identification rests on two timing/information restrictions: demanders of reserves or M2 observe the relevant short rate, prices, and output within the month but not the long-term rate, so R_10 is excluded from the demand equation; the Federal Reserve observes current financial-market variables (the funds rate, the 10-year yield, commodity prices) within the month -- not because it targets them, but because they signal the state of the economy -- but does not observe current goods-market data (price level, output, unemployment) because of a roughly one-month reporting lag, so those are excluded from the supply equation. The goods-market variables (unemployment, output, prices, the long rate, commodity prices) are additionally treated as block-recursive, triangularized in that order, so that money-market disturbances have no contemporaneous effect on them. The resulting system has five overidentifying restrictions, tested with a likelihood-ratio (chi-squared) statistic, and ninetieth-percentile probability bands are constructed from a Bayesian Monte Carlo experiment (100 draws from the posterior of the VAR covariance matrix, 10 draws of the structural mapping per draw).

Q: How do the M2-market estimates compare, and does that model fit better statistically?

Over the same 1982:12-1992:4 sample, M2 demand has an interest elasticity of -0.0082 (se 0.0007, highly significant), with prices positive and significant and output negative and significant; M2 supply has an interest elasticity of +253.84 (se 10.79, highly significant). Unlike the reserves market, the M2-market overidentifying restrictions are not rejected (chi-squared(5)=2.715, p=.74), giving the M2 model cleaner statistical support in this sample even though, as the impulse responses show (Q6), its dynamic responses are less sharply estimated than the reserves market's.

Q: How do the M2-market responses compare, and how are M2 and reserves supply shocks related?

The analogous expansionary M2 supply shock produces a similar qualitative pattern -- the 1-month T-bill rate falls sharply at impact and only rises smoothly after about 1.5 years, prices rise quickly and keep rising, the 10-year rate rises after a few months, and output is significant only in the first few months, becoming "only marginally significant in subsequent periods" -- leading the authors to conclude that "responses to a supply shock in the M2 market are less significant than those to a monetary policy shock in the reserves market." A regression of the estimated M2 supply shock on current and three lags of the reserves supply shock (R-bar-squared=0.19, F=6.63, p<.0001) shows the M2 shock is a combination of reserves-market policy shocks and other shocks; the authors interpret the low R-bar-squared as evidence that "a substantial proportion of the variability of supply shocks in the market for M2 is attributable to nonpolicy shocks associated with the behavior of financial institutions," and conclude inference drawn from M2-based studies "should be drawn cautiously."

Q: What do the variance decompositions say about how much of output and price variability is due to policy shocks?

Reserves and funds-rate supply shocks explain over 30 percent of the forecast-error variance of reserves and the funds rate on impact, but this share "diminishes rapidly," which the authors read as showing "the endogenous response of policy is the dominant source of forecast errors in reserves and the federal funds rate even at impact." By contrast, the share of forecast-error variance attributable to policy shocks grows with the horizon for the price level and output: "nearly 30 percent of the error variance of the price level is due to policy shocks after 3 years, whereas over 30 percent of output variability gets attributed to policy shocks" by the same horizon (Table 4).

Q: What happens when the same models are estimated on 1971:1-1979:9 data, and how do the authors interpret the result?

In the reserves market, the 1970s sample breaks the identification: the estimated short-term interest elasticity of demand is +0.0069 (wrong sign), the supply elasticity is -3.922, and neither is significantly different from zero, leading the authors to conclude "the model's restrictions fail to identify demand and supply relationships for reserves in the 1970s." The M2-market model fares better in the 1970s -- a demand elasticity of -0.0027 (significant) and a supply elasticity of +53.57 (highly significant), with the supply shock variance five times larger and the supply elasticity to the long rate four times stronger than in the 1980s sample -- but its overidentifying restrictions are now strongly rejected (chi-squared(5)=14.958, p=.01). The authors treat the reserves-market breakdown as evidence that the Federal Reserve's operating procedure was genuinely different before the 1980s (consistent with their deliberate choice of the 1982:12-1992:4 sample as a period of indirect funds-rate targeting, stable inflation, and post-deregulation financial markets), rather than as a robustness failure of the identification strategy itself.

David B. Gordon

Eric M. Leeper

📄 Summarized from the full manuscript · Human-reviewed for faithfulness before publication

Canonical DOI

In brief

How can you tell a deliberate policy move from the market's own reaction to it? Gordon and Leeper model supply and demand in the market for bank reserves jointly, rather than assuming one side is fixed. On monthly United States data from late 1982 to 1992, an easing move lowers the federal funds rate for about eight months, raises output and prices within roughly six months, and cuts unemployment with a delay -- the pattern theory predicts but earlier approaches missed. Applied to the 1970s the same restrictions break down, which they attribute to the Fed's different procedures then. It matters because conclusions about policy hinge on such assumptions.

What this paper finds — and why it matters

This 1994 Journal of Political Economy paper by David B. Gordon and Eric M. Leeper proposes a structural vector autoregression (SVAR) that identifies monetary policy shocks by modeling supply and demand simultaneously in the reserves market and, separately, in the M2 market, rather than treating a single monetary aggregate or interest rate as predetermined the way standard Cholesky, nonborrowed-reserves, or federal-funds-rate identification schemes do. In each market a demand equation (the monetary aggregate as a function of its own interest rate, the price level, and output, with the interest elasticity a_1 restricted negative) and a supply equation (the interest rate as a function of the aggregate, the 10-year Treasury yield, and a commodity price index, with elasticity a_4 restricted positive) are estimated jointly by maximum likelihood, identified by two restrictions: private demanders of money observe only the current interest rate, prices, and output within the month (the long-term rate is excluded from demand), while the Federal Reserve observes current financial-market variables (funds rate, 10-year yield, commodity prices) but not current goods-market data (price level, output, unemployment) because of a roughly one-month reporting lag, so those goods-market variables are excluded from the supply/Fed-behavior equation; the goods-market block itself is treated as block-recursive, unaffected contemporaneously by either money market. Using monthly U.S. data over 1982:12-1992:4 with six lags, the estimated interest elasticity of reserves demand is -0.028 (se 0.014) and the reserves supply elasticity is +30.28 (se 10.95); for M2 the demand elasticity is -0.0082 (se 0.0007) and the supply elasticity is +253.84 (se 10.79); the reserves-market model narrowly fails an overidentification test (chi-squared(5)=11.864, p=.04, which the authors describe as “not… a strong statistical rejection”), while the M2-market model comfortably passes (chi-squared(5)=2.715, p=.74). A one-unit expansionary supply shock in the reserves market produces a liquidity effect – the funds rate falls sharply and reserves rise – that lasts about 8 months before the rate climbs back above baseline after roughly two years, with the price level and industrial production rising significantly within about six months and unemployment falling with roughly a three-month lag, responses the authors call “fully consistent with traditional analyses,” in contrast to the liquidity-puzzle and price-puzzle results the paper documents earlier from reduced-form and single-variable-innovation identifications. The analogous M2 supply shock produces qualitatively similar but “less significant” responses, and a regression linking the two shows the M2 supply shock loading on current and lagged reserves supply shocks with R-bar-squared only 0.19, implying that a substantial share of M2 supply-shock variation reflects nonpolicy behavior of financial institutions rather than monetary policy, so the authors caution that inference from M2-based studies “should be drawn cautiously.” Variance decompositions show policy shocks explaining a large share of reserves and funds-rate forecast-error variance on impact but declining quickly, while their share of price-level and output variance rises to roughly 30 percent by three years out. When the same reserves-market model is re-estimated on 1971:1-1979:9 data, the identifying restrictions “fail to identify demand and supply relationships for reserves in the 1970s”: the demand elasticity flips to the wrong sign (+0.0069) and neither elasticity differs significantly from zero, which the authors attribute to a different Federal Reserve operating procedure in that earlier period rather than to a flaw in the identification strategy itself, whereas the M2-market model still recovers sensible elasticities in the 1970s but now strongly rejects overidentification (chi-squared(5)=14.958, p=.01).

Summary of a classic paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.

Q1. What problem with existing monetary-policy identification schemes motivates this paper?

Standard SVAR identification schemes for monetary policy – Cholesky orderings, nonborrowed-reserves (NBR) innovations, or federal-funds-rate (FFR) innovations – equate a statistical innovation in a single variable with “the” monetary policy shock, which implicitly imposes an extreme assumption about interest elasticities (treating money as perfectly elastically or perfectly inelastically supplied/demanded) without ever estimating those elasticities. The paper argues these assumptions produce anomalous dynamics – a liquidity puzzle (a positive reserves innovation fails to lower the funds rate) or a price puzzle (a negative FFR innovation is followed by rising prices) – because a single-variable innovation conflates supply and demand movements in the money market rather than isolating the supply-side (policy) component. The paper’s question is whether jointly identifying supply and demand in the reserves market and in the M2 market can recover a monetary policy shock whose dynamic effects match traditional monetary theory.

Q2. How does the paper identify supply and demand shocks simultaneously?

Two separate structural systems are estimated – one for the reserves market, one for the M2 market – each consisting of a demand equation (money as a function of its own interest rate, the price level, and output, with the interest elasticity a_1 restricted negative) and a supply equation (the interest rate as a function of money, the 10-year Treasury yield, and a commodity price index, with elasticity a_4 restricted positive), estimated by maximum likelihood following Bernanke (1986), Blanchard and Watson (1986), and Sims (1986). Identification rests on two timing/information restrictions: demanders of reserves or M2 observe the relevant short rate, prices, and output within the month but not the long-term rate, so R_10 is excluded from the demand equation; the Federal Reserve observes current financial-market variables (the funds rate, the 10-year yield, commodity prices) within the month – not because it targets them, but because they signal the state of the economy – but does not observe current goods-market data (price level, output, unemployment) because of a roughly one-month reporting lag, so those are excluded from the supply equation. The goods-market variables (unemployment, output, prices, the long rate, commodity prices) are additionally treated as block-recursive, triangularized in that order, so that money-market disturbances have no contemporaneous effect on them. The resulting system has five overidentifying restrictions, tested with a likelihood-ratio (chi-squared) statistic, and ninetieth-percentile probability bands are constructed from a Bayesian Monte Carlo experiment (100 draws from the posterior of the VAR covariance matrix, 10 draws of the structural mapping per draw).

Q3. What do the estimated reserves-market elasticities look like, and how strong is the identification?

Over 1982:12-1992:4, the estimated reserves demand equation gives an interest elasticity of -0.028 (se 0.014), negative and significant, with prices and output jointly significant (F(2,104)=4.604, p<.01); the supply equation gives an interest elasticity of +30.28 (se 10.95), positive and highly significant, with the 10-year yield also significant (0.456, se 0.101) but commodity prices insignificant. The model’s five overidentifying restrictions are formally rejected at conventional levels (chi-squared(5)=11.864, p=.04), though the authors state they “do not view this as a strong statistical rejection of the model.”

Q4. How do the M2-market estimates compare, and does that model fit better statistically?

Over the same 1982:12-1992:4 sample, M2 demand has an interest elasticity of -0.0082 (se 0.0007, highly significant), with prices positive and significant and output negative and significant; M2 supply has an interest elasticity of +253.84 (se 10.79, highly significant). Unlike the reserves market, the M2-market overidentifying restrictions are not rejected (chi-squared(5)=2.715, p=.74), giving the M2 model cleaner statistical support in this sample even though, as the impulse responses show (Q6), its dynamic responses are less sharply estimated than the reserves market’s.

Q5. What do the impulse responses to an expansionary reserves-market policy shock look like, and why do the authors see this as a validation of the approach?

A one-unit expansionary supply shock in the reserves market (e^s = -1) produces a liquidity effect: the funds rate falls sharply on impact and stays depressed for about 8 months before rising back above baseline after roughly two years, while total reserves rise; the price level begins rising after a one-month delay (forced by the block-recursion assumption) and is rising significantly within about six months; industrial production rises significantly within six months; unemployment falls, mirroring output with roughly a three-month lag; the 10-year rate and commodity prices both rise significantly. The authors describe this full set of responses as “fully consistent with traditional analyses,” in explicit contrast to the anomalous liquidity-puzzle and price-puzzle patterns the paper shows arise under reduced-form or single-variable (NBR- or FFR-innovation) identification of the same data.

The analogous expansionary M2 supply shock produces a similar qualitative pattern – the 1-month T-bill rate falls sharply at impact and only rises smoothly after about 1.5 years, prices rise quickly and keep rising, the 10-year rate rises after a few months, and output is significant only in the first few months, becoming “only marginally significant in subsequent periods” – leading the authors to conclude that “responses to a supply shock in the M2 market are less significant than those to a monetary policy shock in the reserves market.” A regression of the estimated M2 supply shock on current and three lags of the reserves supply shock (R-bar-squared=0.19, F=6.63, p<.0001) shows the M2 shock is a combination of reserves-market policy shocks and other shocks; the authors interpret the low R-bar-squared as evidence that “a substantial proportion of the variability of supply shocks in the market for M2 is attributable to nonpolicy shocks associated with the behavior of financial institutions,” and conclude inference drawn from M2-based studies “should be drawn cautiously.”

Q7. What do the variance decompositions say about how much of output and price variability is due to policy shocks?

Reserves and funds-rate supply shocks explain over 30 percent of the forecast-error variance of reserves and the funds rate on impact, but this share “diminishes rapidly,” which the authors read as showing “the endogenous response of policy is the dominant source of forecast errors in reserves and the federal funds rate even at impact.” By contrast, the share of forecast-error variance attributable to policy shocks grows with the horizon for the price level and output: “nearly 30 percent of the error variance of the price level is due to policy shocks after 3 years, whereas over 30 percent of output variability gets attributed to policy shocks” by the same horizon (Table 4).

Q8. What happens when the same models are estimated on 1971:1-1979:9 data, and how do the authors interpret the result?

In the reserves market, the 1970s sample breaks the identification: the estimated short-term interest elasticity of demand is +0.0069 (wrong sign), the supply elasticity is -3.922, and neither is significantly different from zero, leading the authors to conclude “the model’s restrictions fail to identify demand and supply relationships for reserves in the 1970s.” The M2-market model fares better in the 1970s – a demand elasticity of -0.0027 (significant) and a supply elasticity of +53.57 (highly significant), with the supply shock variance five times larger and the supply elasticity to the long rate four times stronger than in the 1980s sample – but its overidentifying restrictions are now strongly rejected (chi-squared(5)=14.958, p=.01). The authors treat the reserves-market breakdown as evidence that the Federal Reserve’s operating procedure was genuinely different before the 1980s (consistent with their deliberate choice of the 1982:12-1992:4 sample as a period of indirect funds-rate targeting, stable inflation, and post-deregulation financial markets), rather than as a robustness failure of the identification strategy itself.

Key terms in this paper

Definitions below follow the paper's own usage.

Simultaneous supply-and-demand identification: this paper's structural alternative to Cholesky/NBR/FFR schemes, in which both the monetary aggregate and its interest rate are treated as jointly determined by a demand equation (aggregate as a function of its own rate, prices, output) and a supply equation (rate as a function of the aggregate, the long rate, and commodity prices), with the interest elasticities estimated rather than imposed via an assumed ordering.
Liquidity effect: in this paper, the fall in the relevant short-term interest rate (the funds rate in the reserves model, the 1-month T-bill rate in the M2 model) that accompanies a rise in the monetary aggregate following an identified expansionary supply shock -- here found to last about 8 months in the reserves market -- as distinct from the "liquidity puzzle" the paper documents under standard reduced-form or single-innovation identification, where a positive reserves innovation fails to lower the funds rate at all.
Block exogeneity / block recursion: the assumption, imposed on the goods/financial-market variables (unemployment, output, prices, the long rate, commodity prices, triangularized in that order), that disturbances originating in the reserves or M2 market do not affect these variables contemporaneously (within the month); the paper describes this as consistent with "conventional partial equilibrium analyses" of the money markets and notes the model is "neutral on the structure and form of expectations effects."
Overidentifying restrictions test: the chi-squared(5) likelihood-ratio test of the five restrictions left over once the structural system is exactly identified; used in this paper to judge how well each market/sample combination fits, ranging from a narrow rejection for reserves in the 1980s (p=.04, called not a "strong" rejection) to a strong rejection for M2 in the 1970s (p=.01).
Information-based vs. timing-based restrictions: the paper's two distinct exclusion rationales -- an information-based restriction excludes the long-term rate from private money demand because demanders do not observe it within the month, while a timing-based restriction excludes current goods-market data (prices, output, unemployment) from the Fed's supply/reaction equation because those series are released with a lag, even though the Fed does observe current financial-market variables that serve as 'informational instruments' for the state of the economy.

How this summary was made. Bibliographic fields are pulled from Crossref and OpenAlex and are not model-generated. The summary was drafted from the open-access manuscript , checked by a claim-grounding and calibration review pass, and approved before publishing. Found an error or a misrepresentation? Flag it here — corrections are welcome, especially from the authors.