Measuring Monetary Policy Shocks in a Small Open Economy

23 Pages Posted: 31 Jul 2001

See all articles by Giorgio Di Giorgio

Giorgio Di Giorgio

Luiss Guido Carli University - Department of Economics and Finance

Giuseppe De Arcangelis

Sapienza Università di Roma

Multiple version iconThere are 2 versions of this paper

Date Written: September 2000


In this paper, we present different specifications of a structural vector autoregressive model (SVAR) that can be used to identify monetary policy operating regimes and monetary policy shocks in a small open economy. SVAR has the advantage of imposing a minimal set of theoretical restrictions on the model to be tested. A monetary policy shock is identified with the residual of an equation regressing a monetary policy instrument on a set of variables that are considered relevant for the decisions of the central bank.

We focus on the Italian economy in the 90s and try to establish if monetary policy shocks are better identified using exchange rates or foreign exchange reserves as a conditioning variable for the small open economy framework. In the considered sample, we have two periods of quasi-fixed nominal exchange rates (1989.06 - 1992.09; 1996.11 - 1998.04) and one of free floating (1992.10 - 1996.10). Given the limited span of the subperiods and the monthly frequency of the data, we treat the whole sample as one of managed floating of the Lira and propose different model specifications to check whether the identification of the central bank operating regime and of the monetary policy shocks is robust enough.

Our methodology is based on De Arcangelis and Di Giorgio (1998), which in turn extended to a small open economy the research strategy introduced by Strongin (1995) and further developed by Bernanke and Mihov (1998) for the US. More precisely, we give a structural content to the VAR models by linking econometric analysis with the institutional knowledge of how the market for banks reserves (i.e., the market in which monetary policy is actually conducted) works in Italy. In our estimated models, indeed, identification hinges on a detailed description of the operating procedures used by the Bank of Italy. The advantage of this procedure is that it allows for a direct test of different model alternatives that are nested in the same specification, without imposing one identification mechanism a priori. The correct measure of a monetary policy shock is then selected by the data itself.

Our analysis confirms the view that the Bank of Italy has been targeting the rate on overnight interbank loans in the 90s. This result is obtained with either proposed modeling choices. Therefore, we interpret shocks to the overnight rate as purely exogenous monetary policy shocks and study how they impact the economy. In the model with the exchange rate, following a monetary policy restriction, output declines and shows a statistically significant reduction for about one year after 7-8 months. Although we have not included a commodity price index we find no evidence of a price puzzle. The model does not either exhibit any liquidity puzzle. The model with foreign reserves provide similar results with a more pronounced output response. Also the inclusion of a German interest rate (whose innovations could be interpreted as a proxy for foreign monetary policy shocks) does not modify the identification results and the qualitative responses of output and inflation.

JEL Classification: E42, E52, F41, F47

Suggested Citation

Di Giorgio, Giorgio and De Arcangelis, Giuseppe, Measuring Monetary Policy Shocks in a Small Open Economy (September 2000). Available at SSRN: or

Giorgio Di Giorgio (Contact Author)

Luiss Guido Carli University - Department of Economics and Finance ( email )

Viale di Villa Massimo, 57
Rome, 00161

Giuseppe De Arcangelis

Sapienza Università di Roma ( email )

Dipartimento di Scienze Sociali ed Economiche
P.le Aldo Moro 5
Rome, 00185
+390649910489 (Phone)

HOME PAGE: http://

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics