A Transactions-Based Model of the Monetary Transmission Mechanism

Questions & answers

Q1. What three features of real-world monetary economies does the paper emphasize as “often mentioned but seldom explicitly articulated,” and why do they matter?

The paper builds its model around three features: (i) “going to the bank is costly so that people will tend to bunch cash withdrawals,” (ii) “people don’t all go to the bank simultaneously,” and, because of these, (iii) “at any instant of time agents hold different amounts of cash.” Formally, money withdrawals are staggered so that a fixed fraction of the population makes a withdrawal each period (the timing is exogenous, but the size of each withdrawal is endogenously determined by the consumer’s planned spending until the next trip), which is what the paper credits with generating a non-degenerate cross-sectional distribution of cash balances – the key ingredient that makes open market operations non-neutral.

Q2. How does an open market purchase actually work in this model, and why must it be absorbed disproportionately by consumers “at the bank”?

When the government buys bonds on the asset market (“the bank”), consumers who are not currently at the bank (i.e., who have not yet exhausted their cash since their last withdrawal) find it suboptimal to sell bonds for cash and immediately convert that cash back into bonds just to hold idle money; so in equilibrium “only those people who are at the bank at time t” – i.e., those currently making a withdrawal – end up holding the newly created money. Because these consumers must therefore absorb “a disproportionately larger share of the stock of money” than their usual share, banks must lower real and nominal interest rates to induce them to actually consume (rather than simply hoard) this extra cash – the paper’s core mechanism for how a monetary injection generates real effects purely through this uneven, transactions-driven absorption, with no assumed price stickiness anywhere in the model.

Q3. Why does the price level rise only gradually after a monetary expansion, given that prices in the model are completely flexible?

Because the consumers who absorb the new money spend it gradually over the interval before their next bank trip (rather than all at once), “the price level rises gradually through time, even though prices are completely flexible” – a result the paper explicitly contrasts with a hypothetical proportional transfer to every agent, which “would raise all nominal prices by the same percentage and thus have no real effects.” The paper formally shows (Section 4.B) that it cannot be an equilibrium for all prices to simply rise by the same percentage as the money supply while leaving interest rates and consumption unchanged, because the cash-in-advance constraints facing consumers who are and are not currently at the bank are not symmetric, and this asymmetry is what forces the nominal nterest rate to decline as part of the adjustment.

Q4. What wealth-redistribution channels does an open market operation create in this model, and how are they different from “money falling from heaven”?

The paper identifies two indirect redistribution channels, distinguishing the model sharply from a scenario where money simply “falls from heaven” proportionally to existing holdings: first, because agents not currently at the bank hold more money than those at the bank (who hold none), the inflation induced by the money creation “redistributes wealth from those not at the bank to those at the bank”; second, because current-period receipts accrue as interest-bearing deposits, agents not at the bank implicitly lend to those making withdrawals, so “a decline in interest rates redistributes wealth from creditors to debtors,” which further “enhances the relative position of those making withdrawals.” The paper is explicit that “there is no direct benefit bestowed on the recipients of the new money” – the real effects operate entirely through these two indirect channels, not through any assumed first-mover windfall.

Q5. What does the paper conclude about the duration of the price-adjustment transition following a monetary injection, and how is this estimate constructed?

Using the empirical fact that average U.S. money holdings are sufficient to finance about one-seventh of a year’s GNP (the inverse of income velocity), and assuming spending occurs at a constant rate between bank trips (so average money holdings equal half of total spending over the interval), the paper infers that a representative consumer visits the bank roughly every 2/7 of a year, or about 15 weeks, implying a model “period” of about 7-8 weeks; since prices first reach (and overshoot) their new steady-state level after two periods in the model, the paper estimates the transition period during which prices rise in response to a monetary injection at “about 15 weeks.” The authors flag this estimate as imprecise in both directions: it likely understates the true transition if firms hold idle balances rather than instantly depositing receipts at the bank, but overstates it if some consumers receive cash payments directly without needing a bank withdrawal.

Q6. How does the model’s prediction about anticipated future money differ from money-in-the-utility-function approaches like Sidrauski’s, and why?

Because the cash-in-advance constraint puts a hard upper bound on current spending from current cash balances, the paper finds that “current prices [are] less sensitive to anticipation of future money than suggested by the analysis of Sidrauski (1976),” which treats money as providing utility services like a durable good; in the extreme case of logarithmic utility, the paper shows the current price level is entirely unaffected by anticipated future monetary injections (citing its own eq. 2.27a), a stark result that follows jointly from the cash-in-advance upper bound on spending and the assumption that the time between bank trips is fixed rather than responsive to anticipated inflation. The paper notes this last assumption is a real limitation: “a sufficiently large anticipated inflation will cause people to go to the bank sooner,” which would make spending more sensitive to anticipated inflation than the baseline fixed-interval model implies.

Q7. What simplifying assumption does this paper make about output, and what does that imply about the scope of its conclusions regarding money and aggregate output?

The paper explicitly assumes “output is exogenous, so that the model can shed only indirect light on the relationship between money and aggregate output” – its results speak directly to nominal and real interest rates and the price level response to open market operations, not to a fully worked-out theory of how money affects real output. The authors are also explicit that making the time between bank trips endogenous (rather than fixed, as assumed) would likely make the transition slower and more gradual still, and would lower the size of the real-interest-rate decline needed to induce spending by recipients of new money, since a decline in nominal rates would also tend to lengthen the optimal interval between bank trips.

Q8. This record summarizes the September 1982 NBER working paper (“Part 1”) rather than the published 1983 American Economic Review article directly – what should a reader know about that relationship?

The published American Economic Review article (1983, vol. 73, no. 5, pp. 871-880) does not carry the “Part 1” designation used by the two 1982 NBER working papers (No. 973, summarized here, and a companion No. 974 that relaxes the logarithmic-utility assumption to a constant-elasticity form); the AER article is a single, considerably shorter piece (about 10 printed pages) than this working paper (roughly 42 pages), so it is very likely a condensed treatment covering some but not necessarily all of the derivations and extensions discussed here. The model setup, the core non-neutrality argument, the wealth-redistribution channels, and the roughly-15-week transition estimate described above are all present in this working-paper draft and are consistent with the abstract and framing the authors themselves give for the model; any additional results specific to the published AER text that are not in this working paper, or vice versa, have not been separately verified for this record.