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.