Text

theses

In my dissertation, I attempt to shed new light on the impact of central bank behavior. The first chapter proposes indexes of monetary policy design characteristics in line with inflation targeting (IT). Indexes aim to provide a useful classification of central banks and to examine whether IT intensity matters for sacrifice ratio and inflation persistence. Results show that U.S. and Japan are ranked close to early targeters. Armenian central bank is detected to perform remarkably well as opposed to the literature's descriptions. Noteworthy structural changes are noticed in Mexico and Peru. Higher index level significantly reduces OECD countries' inflation persistence through increased transparency. Higher intensity among targeters does not deliver significantly lower sacrifice ratio or inflation persistence. Announcing official targets does not provide additional benefits. Low degree of central bank discretion and limited financing of the government significantly decrease the sacrifice ratio among OECD countries and in the full cross-sectional sample respectively. Institutional design features need to be mature enough in order to benefit from inflation targeting. The second chapter examines the usefulness of monetary aggregates. A number of dynamic stochastic general equilibrium models (DSGE) and econometric time series models are jointly specified; including vector autoregressive (VAR), random walk (RW), and various autoregressive (AR) and AR with exogenous variable type models (ARX). Model performance is evaluated via novel testing methods developed for assessing predictive and simulation accuracy. Money is found to matter for inflation simulation purposes in simple econometric models, such as ARX models. With regard to predictive point and density analysis, when complex models are examined, M2 enters into the forecast-best VAR-type inflation models. Model selection is found to be sensitive to researcher’s objective function, to target variable and forecast horizon. Particularly, when the objective is policy analysis, the examination of multivariate models show that DSGE-type models often outperform at longer horizons and mainly for output forecasts. In contrast, VAR-type models win for simulation purposes and short-horizon inflation forecasts. However, univariate models without money generally dominate theoretical and atheoretical multivariate models both with and without money from both forecasting and simulation standpoint.

ETD_3770

Ph.D.

Includes bibliographical references

Includes vita

by Demet Tunali

http://hdl.rutgers.edu/1782.1/rucore10001600001.ETD.000064184

doi:10.7282/T3959GJ0

ETD doctoral

The author owns the copyright to this work.

1489408

windows xp

ETD

application/pdf

application/x-tar

1495040

5ce7bb9da416ac75364088ec9db493c97bd73cdc