Bayesian Applications in Empirical Monetary Policy Analysis

Abstract

Väitöskirjassa tutkitaan Euroopan Keskuspankin yllätyksellisen rahapolitiikkatoimenpiteen dynaamisia vaikutuksia rahaliiton jäsenmaissa. Lisäksi työssä verrataan estimoidun Keynesiläisen dynaamisen stokastisen yleisen tasapainomallin ennustekykyä perinteisiin ennustemalleihin Yhdysvaltojen aineistolla.

Väitöskirjan kolmen ensimmäisen esseen taustalla on olettamus, että rahaliiton jäsenmaiden rahapolitiikan välittymiskanavat ovat erilaiset vaikka jäsenmaat täyttivät EMU-lähentymiskriteerit ajallaan. Tulos jäsenmaiden epäsymmetrisestä reagoinnista Euroopan Keskuspankin harjoittamaan yllätykselliseen rahapolitiikkaan tukisi väittämää, että euroalue ei olisi optimaalinen valuutta-alue. Väitöskirjassa saatujen tulosten valossa voidaan todeta, että yllätyksellinen Euroopan Keskuspankin harjoittama rahapolitiikka aiheuttaa epäsymmetrisiä hintainflaatiosarjojen reagointeja rahaliiton jäsenmaissa.

Dynaamiset stokastiset yleisen tasapainon makromallit tarjoavat mielenkiintoisen työvälineen talouden muuttujien ennustamiseen ja talouspoliittisten toimien analysointiin. Tämän malliluokan mallien empiirisen soveltamisen kompastuskivenä on ollut se, että näiden mallien sovittaminen havaintoaineistoon on osoittautunut hankalaksi. Neljännessä esseessä esitelläänkin helposti sovellettavissa oleva estimointitapa kyseisten mallien estimoimiseksi. Estimoidun tasapainomallin kyky ennustaa Yhdysvaltojen korkoa, inflaatiota ja tuotantoa on hyvä. Saatu tulos on erittäin mielenkiintoinen, sillä alan kirjallisuudessa uskotaan, että vastaavaan ennustekykyyn päädytään kasvattamalla mallikokoa.

This thesis investigates effects of sudden movement in monetary policy stance in the euro area and assesses forecasting performance of estimated structural dynamic equilibrium model for the United States data. In the first three essays the focus is on an inspection of the dynamic effects of sudden changes in the monetary policy conduct of the European Central Bank (ECB) in EMU member countries. We propose that asymmetric monetary policy responses would imply that domestic monetary policy transmission mechanisms have not necessarily integrated even if EMU convergence criteria were met on time, and it would be at odds that the euro area constitutes an optimum currency area. The fourth essay assesses the forecasting performance of a modern macro model for U.S. data. The statistical inference of the thesis is Bayesian.

In the first essay, we describe the dynamics of year-on-year consumer price inflation responses to an unanticipated expansionary monetary policy shock in the euro area with a vector autoregressive model (VAR) model. The variables and statistically testable short run restriction schemes ensuring identification are derivable from a new Keynesian macro model. A rather surprising finding is that traditional Cholesky identification is only weakly supported by the data. Impulse responses of year-on-year consumer price inflation to an expansionary monetary policy shock are calculated for a VAR model identified by the most probable identification scheme. In this identification scheme we let EMU member country information to affect simultaneously monetary policy instrument. Obtained results suggest asymmetric year-on-year price inflation responses to monetary policy conducted by the ECB.

In the second essay, we first survey, with help of a variant of the Taylor rule, six information sets on which the ECB most likely bases its monetary policy decisions. Assessment of information sets is an obvious accretion to the literature on the conduct of monetary policy by the ECB. In the analysis we approximate euro area s monetary conditions with an estimated VAR model for the suggested information set and calculate identified impulse responses of the difference in year-on-year producer price inflations in the euro area and few peripheral EMU member countries. According to results an unexpected variation in the monetary policy instrument conditioned on the information pertaining to the three largest EMU member countries (Germany, France and Italy) will have asymmetric effects in year-on-year producer price inflation across the EMU member countries.

In the third essay, we apply a new Keynesian open economy macro model in setting identifying restrictions for impulse response analysis of consumer price inflation in the euro area. The relevance of the implied simultaneous parameter cross-equation restrictions is assessed by posterior estimation of the hyperparameter that measures prior beliefs on identifying restrictions. The posterior evidence suggests that prior beliefs on simultaneous effects of model variables are of relevance while identifying the VAR model with an open economy new Keynesian macro model for the euro area. Contrary to outcome of the first essay, the drawn impulse responses support the claim that an expansionary monetary policy shock would not cause evident asymmetric price inflation responses in EMU member countries. However, the impulse responses from recursively identified VAR-model are in line with the ones reported in the first essay.

The fourth essay evaluates a closed economy, log-linearized 3-variable new Keynesian model with an easily implementable method for the Bayesian analysis. It becomes evident that a small-scale modern macro model can rival commonly used forecasting tools, such as Bayesian VARs and forecasts based on random walks. According to the posterior evidence, the model manages to capture evolutions of U.S. macroeconomic variables, price inflation, short-term nominal interest rate and measure of output gap, fairly well.