| dc.description.abstract | In this dissertation three different economic issues have been analyzed. The first
issue is whether monetary policy rules can improve forecasting accuracy of inflation.
The second is whether the preference of a central bank is symmetry or not. The last issue
is whether the behavior of aggregate dividends is asymmetry. Each issue is considered in
Chapter II, III and IV, respectively.
The linkage between monetary policy rules and the prediction of inflation is explored
in Chapter II. Our analysis finds that the prediction performance of the term structure
model hinges on monetary policy rules, which involve the manipulation of the federal
funds rate in response to the change in the price level. As the Fed's reaction to inflation
becomes stronger, the predictive information contained in the term structure becomes
weaker. Using the long-run Taylor rule, a new assessment of the prediction performance
regarding future change in inflation is provided. The empirical results indicate that the
long-run Taylor rule improves forecasting accuracy.
In chapter III, the asymmetric preferences of the central bank of Korea are examined
under New Keynesian sticky prices forward-looking economy framework. To this end, this chapter adopts the central bank's objective functional form as a linear-exponential
function instead of the standard quadratic function. The monetary policy reaction
function is derived and then asymmetric preference parameters are estimated during the
inflation targeting period: 1998:9-2005:12. The empirical evidence supports that while
the objective of output stability is symmetry, but the objective of price stability is not
symmetry. Specifically, it appears that the central bank of Korea aggressively responds
to positive inflation gaps compared to negative inflation gaps.
Chapter IV examines the nonlinear dividend behavior of the aggregate stock market.
We propose a nonlinear dividend model that assumes managers minimize the regime
dependent adjustment costs associated with being away from their target dividend
payout. By using the threshold vector error correction model, we find significant
evidence of a threshold effect in aggregate dividends of S&P 500 Index in quarterly data
when real stock prices are used for the target. We also find that when dividends are
relatively higher than target, the adjustment cost of dividends is much smaller than that
when they are lower. | en |
