Liquidity Effects and the Monetary Transmission Mechanism

Questions & answers

Q1. What empirical regularity is the paper trying to explain, and why do earlier models fall short?

The paper opens by noting that “existing quantitative general equilibrium models, which allow capital accumulation, are inconsistent” with the “conventional wisdom” that “an expansionary monetary policy disturbance generates a persistent decrease in interest rates and a persistent increase in output and employment,” and that while a newer class of models (including the authors’ own Christiano and Eichenbaum 1991) can rationalize the contemporaneous response of interest rates, output, and employment to a money shock, “a key shortcoming of these models is that they cannot rationalize persistent liquidity effects.” The paper cites a shift in the empirical literature (Bernanke and Blinder 1990; Christiano and Eichenbaum 1991 and forthcoming; Strongin 1991; Eichenbaum forthcoming; Sims forthcoming) away from identifying monetary policy shocks with innovations in broad monetary aggregates like M1, on the grounds that such innovations largely reflect money-demand rather than money-supply shocks – a shift that, the authors argue, “provides strong empirical evidence in support of the conventional view” that motivates the paper’s theoretical exercise.

Q2. How does the model generate a liquidity effect at all, and why does the basic version fail to make it persistent?

In the model, households allocate a fraction Q of their beginning-of-period money stock to consumption purchases before the current-period money shock is realized, while the rest is lent to financial intermediaries, who in turn lend it to firms needing working capital; because this allocation “is not a function of the period t realization of monetary policy,” a positive money shock is absorbed disproportionately by the financial-intermediary/firm sector rather than by households, which is what forces interest rates down to induce firms to absorb the extra liquidity (“the basic liquidity model”). However, the authors show that in this basic version, “households face zero costs of adjusting sectoral flows of funds over different periods of time,” so “all flows are instantly adjusted in the period after a monetary disturbance,” meaning “the liquidity effects generated by the model are purely transitory” – the central problem the rest of the paper sets out to fix.

Q3. What specific modification generates persistence, and how large does the implied adjustment cost need to be?

The paper introduces a small resource cost of reorganizing financial flows, modeled as the fraction of agents’ time (Ht) spent adjusting Q toward its new desired level, with a convenient functional form chosen “to investigate the potential of [this] mechanism for generating persistence effects” rather than to model the underlying adjustment technology explicitly (“explicitly modeling the costs involved in adjusting Qt is obviously not easy”). The key quantitative finding is that the needed cost is minuscule: “the maximal value of Ht occurs… at 0.0133 percent of steady-state hours worked,” which, taking steady-state hours worked as roughly 320 per quarter (the postwar U.S. sample average), “translates into a loss of [about] three minutes [in] the quarter after the shock” – leading the authors to conclude that “only… small costs of adjusting sectoral flows of funds are needed to generate substantial persistent liquidity effects.”

Q4. What do the basic cash-in-advance (CIA) model and the basic liquidity model predict differently about the response to a positive money shock, and why?

In the basic CIA model, where households’ cash allocation Q can respond immediately to the current money shock, a positive shock is neutral in its real effects: it simply produces “a proportional jump in current and future prices and wages, leaving all other variables unaffected,” and because Q responds positively to the shock, interest rates must actually rise (not fall) to induce financial intermediaries and firms to absorb a disproportionate share of the new cash – driving employment, consumption, and output down, “just the opposite of the conventional view.” In sharp contrast, the basic liquidity model (where Q cannot respond within the period) implies the interest rate Rt and investment It fall on impact while consumption Ct and hours Lt rise in response to a positive money shock, because the fall in Rt shifts labor demand to the right while the rise in Ct shifts labor supply to the left, both increasing hours worked and output – the pattern the paper identifies with the “conventional view” the empirical literature it cites supports.

Q5. What do the model’s impulse responses imply about the dynamics of the system after a persistent liquidity effect is generated, and how does this compare to the basic models?

With the small adjustment cost in place, on impact the system’s response is identical to the basic liquidity model, but because “the percentage of the money stock absorbed by financial intermediaries and firms remains high for many periods after the money shock,” the interest rate reverts to its steady-state value only slowly as households gradually adjust Q, so consumption rises slowly to its steady-state level while investment declines gradually to its own steady-state level, and – because a persistently high interest rate continues to depress labor demand – hours worked and the real wage stay low, and the marginal cost of hiring labor stays high, for an extended period after the shock, rather than reverting within a period or two as in the basic liquidity model.

Q6. What future research directions and predictions does the paper flag as not yet tested?

The central untested prediction the paper highlights is that “a proportionately large share of money injections is absorbed by firms [to] finance variable inputs”; the authors note they are pursuing this test “in joint work with Chari [and] Evans… using flow of funds data” to trace where injected money actually goes in the economy relative to this model’s prediction. On the welfare side, the paper states that “consistent with results in Fuerst (forthcoming), optimal monetary policy [in] our models does not [correspond to] the zero-nominal-interest-rate rule advocated by Milton Friedman (1968),” because in this class of models the monetary authority can redirect cash to the financial sector via open-market operations more quickly than private agents can achieve the same reallocation through adjusting their own nominal saving decisions – a friction advantage for the central bank, relative to private agents, that the authors argue would make it welfare-improving to let the money supply respond to unanticipated changes in real production opportunities (e.g., technology shocks), a pattern they liken to an interest-rate-smoothing rule “of the sort allegedly pursued by the Federal Reserve Board in different subperiods of the postwar era.”