I like Carl E. Walsh's papers best. There is only one passage that I regard as highly disturbing: "Although the exact channels are model dependent, fluctuations in credit spreads can affect both aggregate demand and aggregate supply. On the demand side, they act as an inefficient tax on investment [assuming that efficiency is defined by Pareto-optimality and not by constrained Pareto-efficiency which takes into consideration asymmetric information and which treats the costs of overcoming these real frictions like any other cost; amv], on the supply side they affect firm borrowing costs and therefore marginal costs." (p. 31)
It was Thomas Tooke, a famous mercantilist, who in 1823 suggested that the price level is a positive function of interest rates: interest rates are costs and if, in equilibrium, prices equal costs, higher interest rates imply higher prices and vice versa. This approach lost repudation due to the Classical insight that - other things being equal - higher costs cannot affect the value of money. Rising costs lower equilibrium profits. If enterprises resume to outside finance, higher finance costs reduces their equilibrium return of capital invested (so that the may produce less, which then allow for a higher pricel level - given the aggregate demand for goods). Of course, modern macro rests on Tooke's approach (which may surprise one or the other, who thinks he/she stands in a general equilibrium tradition).
Ricardo and others argued that low interest rates are related to high price levels. Starting from stationarity, the single money rate of interest, if reduced below the equilibrium long-run rate of return on capital, causes aggregate demand on good-markets to exceed aggregate supply such that the price level rises. Earlier in his paper (p. 13, fn. 13) Walsh explicitely points to such kind of inflation as not being a purely monetary phenomenon, since it focuses on what happens on goods markets (although Milton Friedman had exactly this kind of excess-demand driven inflation in mind in making his famous statement on the purely monetary character or root of inflation). Instead, Walsh resumes to the view that explains the price level over time in the RBC core model with flexible prices by the expected present discounted value of the marginal utilitiy of present and future monetary services. Yet,
(1), DSGE models with nominal frictions - outside RBC core - link price level movements to the sluggish adjustments by firms so that of course their is a good-market link, and
(2), much more important analytically, it is impossible to define the marginal utility of monetary services without at the same time knowing the price level and hence money's purchasing power (money in utility function is free disposable purchasing power in utility function). Hence, the marginal utility of current and future monetary services cannot determine the current price level and price level expecations for the same reason that the value of capital goods cannot determine interest rates.
Finally, equilibrium analysis does not distinguish between costs and prices (net inputs demanded have a negative sign in contrast to net outputs, but there is just one all-embracing price vector with equilibrium values determined simultaneoulsy by the primitives of the model). It is simply illegitimate to partition the price vector into two subsets and then to argue that increasing one subset by a scalar increases the other subset by the same scalar. The concept of relative prices does not allow for that. The scalar is rather attached the the entire price vector. Neutrality is a matter of increasing all prices proportionately, not that of scaling up some prices to scale up other prices. Such a dependence among prices is not formulated in Walrasian-type models. The super-auctioneer moves all prices independently during tâtonnement (households and firms recontract). This independence can be assumed by diagonal dominance. Once the equilibrium vector is found by such a routine, people adjust quantities. Without some kind of non-tâtonnement (Hahn processes and the like) it is be no means clear how adjustment of some prices affects other prices. Thus, scaling up costs does not inform about the level of equilibrium output prices. In general equilibrium analysis, there are only prices (not prices and costs) and also for the RBC core only relative prices matter (the famous zero-degree homogeneity postulate implicit in the existence of each point in the sequence). In the flexible-price benchmark of modern monetary analysis, absolut (marginal) costs levels determine nothing, especially not the average level of output prices: Since these kind of sequence economies are inessential (see Frank Hahn), that is, money has no place in the RBC core, the price level remains undefined.
Of course, this problem is solved by simply imposing either MIU (money-in-utility function) or CIA (cash-in-advance). According to the consensus view this is analytically illegitimate without imposing the relevant real frictions to produce such demand for or use of money (these frictions violate much of the DSGE standard results). But even if we allow for MIU or CIA, the price level is still determined by the supply of goods and services (including assets) and their respective demands which both are in turn affected by the money-induced shifts in budget constraints.
