Monetary policy effects: new evidence from the Italian flow-of-funds

Abstract

New evidence on the transmission of monetary policy to the economy is provided through an analysis of the effects of a restrictive monetary policy shock on Italian flow of funds over the period 1980 to 2002. Firms reduce issuance of debt and decrease the acquisition of financial assets, providing no support for the existence of strong financial frictions. Following the shock, in the first quarter households increase short-tem liabilities and diminish the acquisition of liquid assets and shares. The public sector increases net borrowing during the first 2 years. Financial corporations decrease their borrowing for three quarters while in the same period the foreign sector increases borrowed funds. We claim that our results shed new light on the role of the financial decisions of the economic sectors in the transmission mechanism of monetary policy.

Acknowledgements

We specially thank Stefano Neri for useful suggestions and comments and Francesco Nucci for a helpful discussion. We thank Massimo Caruso, Larry Christiano, Riccardo De Bonis, Leonardo Gambacorta, Rustam Ibragimov, Andrea Nobili, Luigi Federico Signorini, the anonymous referee and seminar participants at the 38th Annual Conference of the Money, Macro and Finance Research Group, at the 12th Conference of the Society for Computational Economics, at the XV International Tor Vergata Conference on Banking and Finance, at the SaDiBa conference on flow of funds, at Bank of Italy and at University of Rome, Tor Vergata, for helpful comments and discussions. Massimo Coletta helped us with the flow-of-funds dataset. Any remaining errors are our own. The views expressed are those of the authors and do not involve the responsibility of the Bank of Italy. Even if the work is the product of a joint effort, sections 1, 3 and 5 are attributable to Francesco Columba and sections 2 and 3 to Riccardo Bonci.

Notes

¹ Details of the model are provided in Appendices A and B.

² In local currency.

³ The exchange rate since January 1999 is a constant because of the adoption of the single currency.

⁴ From 1980 to 1981: average interest rate on fixed term advances; from 1982 to 1998: auction rate on repurchase agreements between the Bank of Italy and credit institutions; from 1999 onwards: interest rate on main refinancing operations of the European Central Bank (ECB). The latter interest rate does not present a particular break at the beginning of stage 3 of Economic and Monetary Union (EMU) with respect to the Italian repo rate, even if the convergence of interest rates, begun since 1993, accelerated in 1998 (circumstance that we acknowledge with a dummy variable).

⁵ Christiano et al. (1999) demonstrate that in a VAR, the dynamic responses of the variables to a monetary policy shock is invariant to their ordering in the nonpolicy and policy blocks, while the distinction between nonpolicy and policy variables matters.

⁶ We also checked for a treatment of the exchange rate as a policy variable without detecting significant changes in the results (see also note 13 and Neri (2004)).

⁷ Some of the variables in our specification are nonstationary (see graphs in Appendix A), nevertheless we chose not to impose cointegrating relations, in line with the empirical approach to model the effects of unexpected monetary policy shocks of the literature (see, for instance, Bagliano and Favero, 1998), loosing some efficiency but without impairing the consistency of the estimators or arising issues of misspecification. This approach hinges on Sims et al. (1990) who demonstrate that standard asymptotic tests are still valid if the VAR is estimated in levels, even if some variables display unit roots (see also Hamilton, 1994). Moreover we focus, like the comparable literature, on the short-run dynamic responses and not on the long-run dynamics.

⁸ We tried to use an alternative monetary policy indicators like reserve aggregates, in line with Christian et al. (1996). Difficulties in interpretation of these data, particularly at the beginning of the 1980s, put us in the same position of De Arcangelis and Di Giorgio (2001) who considered the monetary policy in those years to be not well described by a market-based approach. Therefore, we were reassured on our choice of interest-rate indicators.

⁹ The three dummies are also related to the three more relevant perturbations of the monetary policy in the period observed. The dummy in the third quarter of 1992 accounts for the contraction of monetary policy during the exchange rate crisis of autumn 1992; the second dummy, in the first quarter of 1995, corresponds to the monetary restriction that contrasted inflationary pressures and the exchange rate depreciation; the dummy in the third quarter of 1998 considers the series of interest rate cuts put in place to achieve convergence of the national interest rates to the common level of the new currency area started in 1999.

¹¹ The responses of the variables to a monetary policy shock were computed with 1000 Monte Carlo simulations over 16 quarters; following Sims and Zha (1999) the confidence bands are 1 SE wide, corresponding to a 68% confidence interval, since ‘[…], for characterizing likelihood shape, bands that correspond to 50 or 68% posterior probability are often more useful than 95 or 99% bands and confidence intervals with such low-coverage probabilities do not generally have posterior probabilities close to their coverage probabilities.’

¹² We do not find what is known in the literature as the ‘price puzzle’, that is, an increase in the price level (measured by the consumer price index), after a monetary restriction, contrary to the theory that predicts instead of a decrease (Kim and Roubini, 2000). The inclusion of the price of imported raw materials between the endogenous variables has the scope of tackling the price puzzle. This is in line with Christiano et al. (1996) who include the price of commodities, along the conjecture of Sims (1992), to take into account inflation indicators in the reaction function of the central bank that may be missing from the VAR model.

¹³ This result allows our results to be exempt from the ‘exchange rate puzzle’ (also excluding from the sample the last 4 years when the exchange rate is constant), that is an impact depreciation of the currency after a monetary contraction (see Sims, 1992; De Arcangelis and Di Giorgio, 2001, and for Italy, Chiades and Gambacorta, 2004), mainly we believe for the different identification scheme adopted and the inclusion of the price of imported raw material, given that also a restriction of the sample to the years examined in the two quoted works on Italy does not change our results. Since we have no evidence of exchange rate puzzle, though with a limited statistical significance, we did not deem necessary to depart from the recursiveness assumption [which we prefer also for preserving comparability with Christiano et al. (1996)] results to allow simultaneous causality between the policy rate and the exchange rate as other authors did to address the puzzle (Smets, 1997; Clarida et al., 1998; Dornbusch et al., 1998, Gaiotti, 1999). Nevertheless for robustness sake, we allowed for simultaneous causality between the two rates adopting an identification scheme as Kim and Roubini (2000) widely considered adequate to deal with the exchange rate puzzle, without detecting any relevant change in the impulse responses (results are available on request).

¹⁴ Consistently with the presence of a liquidity effect we have no evidence of the ‘liquidity puzzle’ previously found in the literature, i.e. when monetary policy shocks are identified as innovations in monetary aggregates and innovations appear to be associated with increases rather than decreases in nominal interest rates. The inclusion as a policy variable of the monetary aggregate M2, following among others the suggestion of Leeper and Roush (2003) has precisely the role of avoiding the insurgence of the liquidity puzzle.

¹⁵ Notably for Italy, Gaiotti (1999) describes in detail the transmission of monetary policy from 1967 to 1997.

¹⁰ In Fig. 1, the shaded areas correspond to the three recessions of the Italian economy as identified by Altissimo et al. (2000), respectively between March 1980 and March 1983, March 1992 and July 1993, November 1995 and November 1996. The monetary policy is relatively tight in the period before each recession and the stance becomes looser during the recession period (with the possible exception of the first period, when the policy rate is highly volatile). Our measure of monetary policy is also consistent with the period of monetary restriction from 1994 to 1996, during which inflationary pressures arising from the exit of the Italian lira from the European Monetary System (EMS) exchange rate mechanism in 1992 and the depreciation shock in 1995 were counteracted (Gaiotti, 1999).

¹⁶ During the period of observation, apart for the major methodological break in 1999 when new monetary aggregates definitions were adopted, M2 witnessed changes in its definition; moreover different definitions of M1 are conceivable. Finally the two monetary aggregates can be considered as evaluated at the end of each period, as (simple or moving) averages, and seasonally adjusted or not.

¹⁷ This fact may be due to the average small size of the estimated policy interest rate shock in the 4 years considered relative to that in the previous part of the sample.

¹⁸ The presence of the foreign sector characterizes Italy as an open economy.

¹⁹ In the present work we consider a genuine ‘consumer’ household sector, while in the Italian flow of funds the household sector comprises also ‘producer’ households (small unincorporated firms and sole proprietorships with less than five employees). We prefer to include the latter among nonfinancial firms, so to include all the producer units in the nonfinancial sector, regardless of firm size or of legal form. The other sectors are consistent with the ESA95 (European System of National Accounts) classification, which is also applied in the Italian flow-of-funds. Financial firms include banks, money market funds, financial auxiliaries and insurance corporations and pension funds (the Bank of Italy is excluded). The general government sector includes central government, local government and social security funds. The foreign sector includes all the nonresident units.

²⁰ In the former dataset, time series showed a discontinuity in 1995 because of differences in the compilation methodology, in classification criteria and in the accounting principles introduced with the adoption of the ESA95.

²¹ For a review of the literature see, among others, Bagliano and Favero (1998), Bernanke and Mihov (1998), Rudebusch (1998), Christiano et al. (1999), Kim (1999), and Walsh (2003).

²² One would expect an increase in government deficit due to higher interest payments on the public outstanding debt.

²³ Quite interestingly for our focus on financing and investment decisions, Dedola and Lippi (2005) found that output responses to monetary policy shocks differ among different industry sectors, and are systematically related to the output durability, financing requirements, borrowing capacity and firm size, both in Italy and in other industrialized countries. Gaiotti and Generale (2002) estimated the effects of monetary policy on the investment behaviour of Italian firms with a panel dataset, finding that financial variables do actually matter.

²⁴ See also Bernanke and Blinder (1992) and Gertler and Gilchrist (1993).

²⁵ In Italy, during the 1970s, financial aggregates were subject to quantitative constraints and market mechanisms operated weakly due to a number of real and nominal rigidities.

²⁶ The responses of the flows of assets and liabilities of households were much stronger in Christiano et al. (1996).

²⁷ Although in the Italian financial accounts there is no distinction between deposits and currency (that sum up to M1), we know from monetary statistics that, on average, currency in Italy in the period examined accounted for only one-sixth of M1.

²⁸ This could reflect some sluggishness in the response of bank deposit rates as found by Gambacorta and Iannotti (2007), especially before the introduction of the Consolidated Law on Banking in 1993.

²⁹ The slight decrease observed in the first quarter after the shock is not statistically significant.

³⁰ This could be the case particularly for the restriction in Italian monetary policy between 1994 and 1996.