The Monetary Transmission Mechanism: An Empirical Framework

Taylor asks what the best empirical framework for the monetary transmission mechanism is, and what that framework implies for the choice between fixed and flexible exchange rates and for the design of an optimal monetary policy rule (pp. 11-12). His answer is not a VAR with an identification scheme for structural shocks, but a synthesis built around his own multicountry structural model (Taylor 1993a) – a model with rational expectations and staggered wage and price setting, comprising a policy reaction function, an uncovered-interest-parity exchange rate equation, an expectations model of the term structure, investment and consumption equations tied to the real interest rate, trade equations tied to the real exchange rate, and staggered price-setting equations for nominal rigidity. The paper reports and discusses results from that structural model and from Romer and Romer’s (1994) reduced-form evidence rather than estimating new equations itself.

Taylor argues that the money demand function for M1 and M2 “seems to have shifted substantially in recent years,” and that credit flow measures are “at least as unreliable,” which has led most structural modelers to focus instead on financial-market prices – the short-term interest rate, the exchange rate, and the long-term interest rate (pp. 11-15). He notes in footnote 1 (p. 12) that if a monetary aggregate with stable velocity could be found, targeting it would have real advantages – it would be “explicit about the nominal anchor for the price level,” provide automatic stabilization, and be easy to communicate – but the instability of observed money demand rules this out in practice, which is why the central bank is modeled as adjusting the supply of high-powered money so as to give the funds rate a desired path (p. 15).

The mechanism runs: (1) the short-term interest rate moves the exchange rate through uncovered interest rate parity under high international capital mobility, so a rate increase appreciates the currency and, given sticky goods prices, raises the real exchange rate in the short run; (2) the higher real exchange rate reduces exports and raises imports, lowering real GDP; (3) the short rate also moves the long-term rate through an expectations model of the term structure, with a persistent rate change passing through nearly one-for-one and a transitory one passing through less; (4) the higher real long rate reduces business fixed investment (equipment and structures), residential investment, durable consumption, and inventory investment, lowering GDP; and (5) GDP and inflation feed back into the short rate via the policy reaction function, closing the circle (pp. 14-18). Taylor emphasizes that this last link is what distinguishes his framework from an “incomplete mechanical transmission story” – monetary policy responds endogenously to the macroeconomic conditions the earlier links generate.

Using Figure 1’s quarterly 1974-1995 US data, Taylor states that the positive relationship between the long-term real interest rate differential and the CPI-adjusted dollar exchange rate against the G-10 is well described by uncovered interest rate parity “at lower frequencies,” so that “the larger and longer swings in interest rates and exchange rates seem to be described accurately” (pp. 15-17). He is explicit, however, that “the high frequency movements are not” well described this way, and that “a well-accepted explanation has yet to be found” for these high-frequency deviations (p. 16) – a caveat he flags as a genuine gap in the framework rather than a solved problem.

In the US, all three components of fixed investment – business equipment, business structures, and residential investment – are significantly negatively related to the real interest rate, and across the full G-7 sample in Taylor’s model, fixed investment is negatively related to the real interest rate in every country (pp. 17-18, 22). Consumption and inventory investment are also reported as sensitive to real interest rates in most countries, though the paper does not tabulate country-by-country coefficients here (referring readers to Taylor 1993a for the underlying magnitudes). Taylor separately acknowledges a theoretical objection – “which interest rate matters for investment?” – to which he “plead[s] guilty” (p. 22): using a single long-term real rate is an empirical simplification, since the five-year bond rate, the mortgage rate, and Tobin’s q may all matter independently.

Taylor writes that while researchers “sometimes find bond supply effects – that is, a change in the supply of long bonds affects the term structure – the effects are too small to explain the errors in the term structure equations on a systematic basis” (p. 18). He immediately qualifies this, however, noting that “there are factors other than short-term interest rates or their expectations that are influencing bond prices” and that “other factors are at least as important as changes in short rates” – an internal tension the paper does not resolve. This 1995 claim predates the quantitative-easing literature (2008-2013), which later found quantitatively important bond-supply effects on term premia, directly challenging Taylor’s characterization here.

Comparing model-implied elasticities across the two overlapping samples, Taylor reports that the real GDP response to a 3 percent unanticipated permanent increase in the target price level fell in all three countries shown in Figure 2 – the US, Germany, and Japan – and concludes “the monetary transmission mechanism has changed so as to reduce the impact of a given change in short-term interest rates,” with “the change in the United States… larger than in Germany and Japan” (pp. 20-21). Within the US specifically, the interest-rate elasticity of investment declined while the elasticity of consumption increased over the same comparison, but Taylor states there is “no general pattern” of this kind across the G-7 more broadly.

Taylor’s simulations of exchange-rate arrangements among the United States, Japan, and Europe show that flexible exchange rate policies produce lower variability in both real GDP and inflation than fixed exchange rate policies (pp. 19-20). He notes separately that the inflation-output variability tradeoff across countries is not observed simply in raw cross-country data, but may emerge once central bank independence is controlled for (p. 23) – a qualification on how the broader stabilization results should be read.

Taylor states that his 1993 interest-rate rule – under which the federal funds rate responds with a coefficient of 0.5 to the inflation deviation from target and 0.5 to the output gap – “has turned out to describe recent Federal Reserve policy very accurately” (p. 15), a self-referential evaluation of the rule from Taylor (1993b). Beyond the UIP high-frequency puzzle (Q4) and the single-interest-rate simplification (Q5), Taylor flags as a standing caveat that the short-term rate is “only one of many factors affecting the exchange rate and the long-term interest rate, and the effects of the short-term interest rate on both are uncertain and variable over time” (p. 14) – a warning he says is “easily overlooked” in the schematic circle diagram.