Money and Interest in a Cash-in-Advance Economy

Questions & answers

Q1. How does the model’s cash-in-advance structure work, and what distinguishes “cash goods” from “credit goods”?

Each period is split into a securities-trading subperiod and a goods-trading subperiod; agents choose their cash balances during securities trading, and in the subsequent goods market they can only buy “cash goods” with currency acquired earlier that period – it is “not possible to acquire additional currency once goods trading has begun” – while “credit goods” are not subject to this constraint, so consumers can substitute between the two goods depending on the relative cost of being cash-constrained. This substitution possibility is the key difference from Lucas (1982) and Svensson (1985), where the whole endowment was effectively a single good determined exogenously; here, because consumers can shift consumption toward or away from the cash good, “money shocks induce real distortions, so that equilibrium quantities and prices must be determined simultaneously” – the paper’s stated main technical challenge.

Q2. What is the paper’s central finding about how current versus expected future money growth affects the real allocation?

The model’s central result, stated in the abstract and proved formally in Section 4, is that “current money growth affects the current real allocation only insofar as it affects expectations about future money growth, i.e., only through its value as a signal.” Formally, if two states of the world imply the same conditional probability distribution over all future outcomes (condition 4.8), then the marginal utility of wealth – and hence the equilibrium consumption allocation – is identical in those two states, even if the money growth rates realized in them differ; conversely, if current money growth and current income are identical across two states but their informational content about the future differs, the allocations will in general differ. In short, “the current rate of money growth plays no direct role in determining the current allocation – only expectations about money growth (and income) matter.”

Q3. If current money growth doesn’t directly affect the real allocation, how can monetary policy still move nominal interest rates?

The nominal interest rate equals the ratio of the expected marginal utility of cash during goods trading later in the current period to the discounted expected marginal utility of cash in the next period’s goods trading (eq. 4.2), and this ratio’s variability depends on the informativeness of any signal available at securities-trading time about future shocks – even in cases where the paper proves the real allocation itself does not depend on that signal’s informativeness at all (Sec. 4, illustrated in Example 3). This decoupling is a central point of the paper: interest-rate behavior and real-allocation behavior can respond quite differently to the same change in the information structure, because interest rates are set before some uncertainty about future goods-market conditions is resolved, while real allocations depend only on the state’s full informational content about the future.

Q4. What do the paper’s examples show about monetary policies with i.i.d. shocks and “interest-stabilizing” policies?

Examples 1 and 2 show that if real income is i.i.d., then any i.i.d. monetary policy yields exactly the same interest-rate behavior and resource allocation as a deterministic monetary policy with the same harmonic-mean growth rate, illustrating concretely that only expectations about future money growth carry allocative consequences. Example 4 works backward from the assumption of homothetic, constant-relative-risk-aversion preferences and a constant (policy-maintained) nominal interest rate to characterize the implied equilibrium allocation and the money-growth process that would support it as an “interest-stabilizing” policy – part of a broader technique the paper uses, of positing a target allocation or price path and solving backward for the implementing monetary policy, “without use of Theorems 1-5.”

Q5. What are the paper’s two main existence results for equilibrium, and how do they differ?

The paper reduces solving for equilibrium to finding a fixed point of a functional equation in a variable that can be interpreted as the value of cash balances (Sec. 2), and offers two existence theorems (Sec. 3): one using the Arzelà-Ascoli lemma and the Schauder fixed-point theorem under weaker conditions, and a second, using stronger conditions on preferences, based on the contraction mapping theorem, which additionally establishes uniqueness of equilibrium. The authors note the contraction-mapping approach, while requiring stronger assumptions, “allow[s] a sharper characterization of the equilibria and are very amenable to numerical simulations” – a methodological tradeoff the paper is explicit about rather than treating one approach as simply superior.

Q6. How does this paper relate to Lucas (1982), Svensson (1985), Lucas and Stokey (1983), and Townsend (1984)?

The paper shares its motivation – reproducing and generalizing the Fisherian theory of interest rates to stochastic environments – with Lucas (1982) and Svensson (1985), but departs from both by letting agents substitute against money, so that (unlike those two papers, where resource allocations were pinned down entirely by exogenous endowments) money shocks here have real effects that must be solved jointly with prices; the model is described as “very closely related to” Lucas and Stokey (1983) and a special case of the framework in Lucas (1984). Townsend (1984) is flagged as the most closely related paper in spirit – agents there similarly hold cash for transactions and monetary policy affects the real allocation – but Townsend’s model allows capital accumulation (not pursued here), while this paper instead offers “richer possibilities for the timing of information arrival and trade” and recursive solution methods not yet known (as of the paper’s writing) to extend to models with capital accumulation.

Q7. What scope conditions does the paper place on its own analysis, by its own account?

The authors are explicit that they abstract from the issues that make the normative study of monetary policy “difficult (and interesting)”: business cycles originating in monetary disturbances are not studied, fiscal authorities are assumed to have access to lump-sum non-distorting taxes, and the only distortion in the system is the “inflation tax,” so that optimal monetary policy in this model always amounts to setting that implicit tax to zero – a result the paper calls “a relatively simple by-product of the analysis,” with the paper’s main focus instead on characterizing the allocative consequences of arbitrary (not necessarily optimal) policies. The paper also explicitly cautions that its results do not deliver “definite signs for the correlations between movements in money and other variables” the way a simple supply-and-demand model of a market for a physical good (its own example: apples) would, because “money is not at all like apples” – a monetary change, once it has occurred, is informationally irrelevant except insofar as it reveals something about the future, so only the full joint distribution of the exogenous shocks pins down predictions about the joint behavior of money, interest rates, and other endogenous variables.

Q8. This summary draws on the May 1985 NBER working paper (No. 1618); how does that relate to the 1987 Econometrica publication?

This record’s source text is NBER Working Paper No. 1618, circulated in May 1985 under the identical title, which was published two years later in Econometrica (1987, vol. 55, no. 3, pp. 491-513); the model structure, the six-section organization (model, existence theorems, properties of equilibria, examples, conclusion), and the paper’s headline “signal” result on money growth and real allocations are all present in this working-paper draft exactly as described above. Because the published version has not been separately checked against this draft for this record, any differences in exposition, numbering, or incremental results introduced during the refereeing process are not reflected here, and this should be read as resting on the working-paper text specifically.