Abstract

In the light of the literature on the ‘price puzzle’ (see among others, Sims, 1992; Eichenbaum, 1992; Barth and Ramey, 2001;Tillmann, 2008;Castelnuovo and Surico, 2010), this paper shows that a positive effect of a monetary policy tightening on the level of prices is to be considered as a normal phenomenon rather than an ‘anomaly’ (Castelnuovo and Surico, 2010). Since interest on the advanced capital is part of the monetary costs of production, a change in the interest rate produces a direct effect on the level of prices by means of a mechanism which is typically defined as ‘cost channel’ (Barth and Ramey, 2001; Dedola and Lippi, 2005; Gaiotti and Secchi, 2006). This channel may overcome the uncertain demand channel which is the one usually considered when discussing the mechanisms of monetary policy transmission. As it will be shown, this “cost channel” cannot be regarded as a ‘regime specific phenomenon’, namely as dependent on the passive (active) behaviour of the central bank in response to a change in the inflation rate (Castelnuovo and Surico, 2010), so that the price puzzle is supposed to disappear if the central bank reacts strongly to inflation as in the post-1979 period when Paul Volcker was appointed as Chairman of the Federal Reserve Board.

In order to contribute to this debate, the paper estimates the effect of an increase in the interest rate on the level of prices by means of structural vector autoregression (SVAR) models(Kilian and Lütkepohl, 2017). To do this, we carry out an empirical analysis for the US economy for the 1959-2018 period. Specifically, we use aggregate monthly data provided by the Organisation for Economic Co-operation and Development and the Federal Reserve. In order to detect the relationship among interest rates and the level of price, the following variables are used: (i) the Effective Federal Funds Rate (); (ii) the Consumer Price Index (); the industrial production index (); and the level of monetary hourly earnings (). Furthermore, in order to avoid feasible spurious correlation in the VAR generated by the omission of relevant variables which is supposed to capture price expectations, several and alternative measures of price and inflation expectations are taken into consideration. Particularly, we make use of: (i) Future Prices Paid () released from the Federal Reserve Bank of Philadelphia; (ii) Inflation Expectations provided by University of Michigan (); (iii) ‘Greenbook’ Inflation Expectations released by the Federal Reserve staff before each meeting of the FOMC () (Romer and Romer, 2004). Besides, to consider different monetary policy regimes, we estimate the abovementioned models along different timespans which range from: (i) 1959:01 to 1979:09; (ii) 1979:10 to 2018:08; (iii) 1982:10 to 2018:08; (iv) 1987:02 to 2018:08.

In order to detect the relationship between the interest rate and the level of prices, two main models will be estimated.

Model 1:

Model 2:

By imposing a lower-triangular structure, which is typically derived from economic theory, the employed identification strategies for Model 1 and 2 are summarised in (5) and (6):

Model 1:

Model 2:

where ‘’ indicates an unrestricted parameter and a ‘0’ represents a zero restriction. In the spirit of Bernanke and Blinder (1992) and Sims (1992), in Model 1, the federal fund rate () is the most exogenous variables, the level of prices () can only respond contemporaneously to a monetary policy shock and the level of economic activity can be affected both by the interest rate and the level of prices within the monthly observation. In Model 2, a same identification strategy is assumed by adding the level of nominal earnings () as first variable. We are assuming that nominal wages are exogenous within the monthly observation for three main reasons: (i) wages are determined by a bargaining process led by trade unions which is affected by the bargaining power of wage earners which is determined by several institutional factors; (ii) the bargaining process affected by information delays motivated by the fact that data are released with a different delay and therefore trade unions could not react immediately to variables that they cannot observe. (iii) monetary wages tend to be affected by nominal rigidities and the process of wage adjustment occurs slowly and over a period of time greater than the monthly observation. Monetary wages are not strictly related to the business cycle fluctuations as the wage bargaining process occurs periodically rather than ceaselessly (Azariadis and Stiglitz, 1983). The identification strategies summarised in (5) and in (6) are consistent with the idea of a cost channel perspective: nominal wages and interest rates – being costs for firms – contribute to determine the general level of prices. Despite of the identification strategies imposed in (5) and (6), we have also tested alternative identifications based on the following recursive ordering:  for the Model 1 and  for the Model 2. A similar ordering, which assumes a monetary policy reaction function (e.g., a Taylor rule), can be found in more recent papers (see among others, Castelnuovo and Surico, 2010; Giordani, 2004). Finally, when price expectations are used, the forward-looking variables are alternatively introduced both in Model 1  and Model 2  as first ordered variable, namely as the most exogenous variable. To detect and quantify the causal relationships among the selected variables, we will estimate both impulse response functions (IRFs) and the forecast error variance decompositions (FEVDs) (Kilian and Lütkepohl 2017). Standard errors are estimated through a Monte Carlo procedure based on 1000 repetitions and IRFs will be reported with two-standard error bound, namely a 95% confidence interval.Besides, estimates of FEVDwill be provided and it shows how much of the forecast error variance of  (where ) is determined by each structural shock () identified in our models (Kilian and Lütkepohl, 2017).

Our findings (Table 1) are in sharp contrast with the ones obtained by Castelnuovo and Surico (2010) and Hanson (2004), according to which the ‘price puzzle’ would be regarded as a ‘regime specific phenomenon’. Specifically, the estimated impulse response functions show the existence of the ‘price puzzle’ irrespective of the passive (active) behaviour of the central bank and the inclusion of price expectations in the considered models. In order to provide a clear picture of our estimation, we provide some consideration of our findings by classifying our models as: (i) with or without expectations; and (ii) estimated in the pre- or post-1979 period. By focusing on the first classification, the inclusion of price or inflation expectations lowers the effect generated by a monetary policy tightening. Yet, despite these results are confirmed in almost all selected periods, when the sub-sample 1987:02–2018:08 is considered, some of the estimated models show that the inclusion of expectations engendered a slight greater effect than the one produced in models which not include expectations. When the second classification between the pre- and post-1979 periodis analysed, our findings show that the positive relationship that moves from the interest rate to the level of prices is more affected by the change in the FED operating procedures occurred during the 1979-1987 rather than the passive or active behaviours carried out by the FED in the pre- and post-1979 period. Specifically, despite the effect on the prices is higher in the pre-1979 period, our findings show that, when the monetary targeting period (Volcker era) is excluded (1982:10–2018:08 and 1987:02–2018:08), the relationship between interest rates and the level of prices becomes stronger that the one estimated for the 1979:10–2018:08 period. The larger effect estimated in the post-1979 period occurs when the sub-sample 1987:02–2018:08 is considered, namely when also the target over M1 is excluded. When this period is analysed, the effect on the prices – albeit lower – is close to the one estimated in the pre-1979 period.

Therefore, we can affirm that the introduction of price or inflation expectations does not solve the ‘price puzzle’ which is confirmed even when forward-looking variables are introduced in the models. Additionally, even when the pre- and post-1979 periods are taken into considerations, the positive relationship that moves from the rate of interest to the level of prices is confirmed in both sub-samples.

Peak effect on the level of prices

Models

1959:01–2018:08

1959:01–1979:09

1979:10–2018:08

1982:10–2018:08

1987:02–2018:08

Without Expectations

Model 1

1.66% (53)

1.06% (21)

0.39% (12)

0.59% (26)

0.91% (44)

Model 2

0.82% (28)

1.05% (19)

0.31% (12)

0.60% (26)

0.99% (50)

With Expectations

Model 1 with

0.97% (35)

0.95% (16)

0.35% (12)

0.54% (23)

0.80% (39)

Model 1 with

0.85% (37)

0.80% (17)

0.18% (4)

0.50% (41)

0.50% (40)

Model 1 with

-----

-----

0.23% (12)

0.57% (24)

0.96% (56)

Model 2 with

0.82% (28)

0.94% (16)

0.27% (12)

0.55% (25)

1.00% (52)

Model 2 with

0.69% (23)

0.83% (18)

0.18% (5)

0.50% (41)

0.57% (34)

Model 2 with

-----

-----

0.20% (5)

0.58% (23)

1.00% (52)

Table 1.Peak effect on the level of prices. In ( ) the month in which the peak effect occurs.

References:

Barth III, M. J., & Ramey, V. A. (2001). The cost channel of monetary transmission. NBER macroeconomics annual, 16, 199-240.

Bernanke, B. S., & Blinder, A. S. (1992). The Federal Funds Rate and the Channels of Monetary Transmission. American Economic Review, 82(4), 901-921.

Castelnuovo, E., & Surico, P. (2010). Monetary policy, inflation expectations and the price puzzle. The Economic Journal, 120(549), 1262-1283.

Dedola, L., & Lippi, F. (2005). The monetary transmission mechanism: evidence from the industries of five OECD countries. European Economic Review, 49(6), 1543-1569.

Eichenbaum, M. (1992). Comment on Interpreting the macroeconomic time series facts: the effects of monetary policy by C.A. Sims, European Economic Review, vol. 36(5), pp. 1001–11.

Gaiotti, E., & Secchi, A. (2006). Is there a cost channel of monetary policy transmission? An investigation into the pricing behavior of 2,000 firms. Journal of Money, Credit and Banking, 2013-2037.

Giordani, P. (2004). An alternative explanation of the price puzzle. Journal of Monetary Economics, 51(6), 1271-1296.

Kilian, L., &Lütkepohl, H. (2017). Structural vector autoregressive analysis. Cambridge University Press.

Romer, C. D., & Romer, D. H. (2004). A new measure of monetary shocks: Derivation and implications. American Economic Review, 94(4), 1055-1084.

Sims, C. A. (1992). Interpreting the macroeconomic time series facts: The effects of monetary policy. European economic review, 36(5), 975-1000.

Tillmann, P. (2008). Do interest rates drive inflation dynamics? An analysis of the cost channel of monetary transmission. Journal of Economic dynamics and Control, 32(9), 2723-2744.