| 1. Characteristic of balanced growth path in the standard neoclassical growth model is theoretical foundation in many DSGE model building, which typically assume that labor supply shocks have a stable characteristic, and then hours worked series are also have stationary characteristic. However, this inference is contrary to the real per capita hours worked series representation continuing volatility characteristics. In this article, we assume non-stationary labor supply shock is the result of the work of a continuing shift of time series an important reason, that per capita hours worked in the study by a continuous sequence characteristics, consider a non-stationary labor supply shocks, which dedicated to measure and analyze the endogenous transmission mechanism of adjusted DSGE models. Specifically, firstly, we use quarterly Taiwan and Japan real per capita GDP and hours worked from 1961: Q1 to 2009 Q4, improve the standard stochastic growth model, thus constructed with four different forms of adjusted DSGE model. At the same time, we use Kalman filtering technique to calculate the state space system, and using Markov Chain Monte Carlo simulation method to achieve Bayesian econometric analysis. Secondly, according to benchmark prior distribution, we build four another adjusted different prior distribution, respectively, and based on baseline prior distribution and the adjusted prior distribution to calculate the marginal data density of the stochastic growth model. So we can compare the size of the marginal data density to evaluate the fitting results of overall hours worked series on the random growth model. Thirdly, based on quarterly data sample set of Taiwan and Japan, and the benchmark prior distribution to calculate posterior predictive distribution plot of hours worked series autocorrelation and standard deviation, further evaluate fitting results of the overall hours worked series stochastic growth model. Finally, based on quarterly data sample set of Taiwan and Japan, and the benchmark prior distribution to calculate output and hours worked impulse response function to evaluate the effect of models for the impact of technology shocks and labor supply response.We found that there is no case of labor adjustment costs, encompassing long-term labor supply shocks on the time series model fitting is superior to contain the impact of short-term labor supply model. However, once introduce the labor adjustment cost factor, the short-term impact on labor supply model fitting time series are stronger. Model with adjustment costs than the model without adjustment costs is better fit the actual data. Through the stable labor supply shocks of adjustment cost model, that is stable and continuing labor force with strong fluctuations in the cost of adjustment model, we can better identify the observed sample autocorrelation and the non-stationary features of hours worked series.2. In today's macroeconomic research, that the RBC model is generally a lack of economic transmission mechanism, but how to solve this problem has become the focus of academic research. This study focuses on building, estimation and evaluation of the"learning by doing"models, that is through the introduction of the"learning by doing"mechanism to improve the shortcomings of RBC models. First of all, for the LBD model building, we will improve the dynamic stochastic general equilibrium model, so that it can encompass"learning by doing"mechanism. Second, for the LBD model estimation and evaluation, we use a Bayesian econometric method, through the prior distribution to put the external information into the parameters, and then analyze and judge the effectiveness of LBD mechanism. Specifically, we will compare the total moment and the impulse response function of DSGE model and VAR models. Although the DSGE models and VAR models are generally data indicated a linear moving average, but because of the constraints of VAR models are far less than DSGE models, which is more suitable as the base model. The constructed model can effectively avoid the distinction between internal and external"learning by doing"mechanism, so that we can determine the contribution of LBD to national income; on the another side, compared with the previous many studies, we used a very high standard to compare the impulse response function: we are not only qualitative comparison of the shape of impulse response functions, but also tests the quantitative match of the impulse response function. In this article, we use quarterly Taiwan and Japan real per capita GDP and hours worked from 1961: Q1 to 2009 Q4, introduce the"learning by doing"mechanism into dynamic stochastic general equilibrium models, and use Bayesian techniques to estimate and evaluate the LBD model. Through this research, we get some meaningful conclusions: the introduction of"learning by doing"transmission mechanism can improve the overall fitting effect; compared with the RBC model, LBD model can simulate through the VAR model to capture some of the important dynamic features.3. The role of monetary policy and economic system agents has closely relationship. Sims (1980) had suggested that the traditional VAR model to evaluate the results of policy interventions. Policy interventions are defined as non-expected departure from monetary policy rules, that is policy impact, and the result of intervention can be assessed by the impulse response function. However, the expansion of monetary policy rules non-expected departure will make agents to change their inherent concept of monetary policy behavior, leading VAR model impulse response forecast failure. One way to solve this problem is to estimate the DSGE model full set, which can be re-calculated alternative policy rules, to predict the impact of structural changes of policy regime.In this paper, we use quarterly China real output growth, inflation and nominal interest rates from 1992: Q1 to 2009 Q3, to construct and estimate a simple New Keynesian monetary DSGE model. In this model, monetary policy follows the nominal interest rate rule, that is under the target inflation rate, monetary policy subject to two regimes Markov switching process. In the first model set, the agents can use the Bayesian rules and observed monetary policy, to inference the current state of monetary policy, and through expectation of future output, prices and interest rates to consider the possibility of regime switch. In the second model set, the agents have complete information on the current monetary policy. Different from previous studies, in our constructed regime switching framework, there is no shift in monetary policy expectations, we only assume the existence of"high inflation"and"low inflation"two different districts, which transition probability constant. While firms and household are trying to infer that monetary policy transmission mechanism, but the central bank has no intention of select the optimal form of monetary policy according to the effectiveness of monetary policy. When we try to estimate the DSGE model, we need to involve a very complicated nonlinear filtering problem, which requires the conventional computer take a long time to calculate. Therefore, in this article, we conduct a two simplified. Firstly, we use log-linear DSGE model approximation. Secondly, we assume that policy rule only depends on observed variables, does not include the system indicators and impact of the policy, not dependent on the underlying model variables (such as potential output). We confirmed that based on these assumptions, using modified Kalman filter to calculate the likelihood function of the regime switching model of is relatively simple. In addition, in empirical analysis, we will combine the prior distribution and likelihood function, and doing Bayesian inference, and thus to evaluate the posterior distribution.4. Dynamic stochastic general equilibrium model has been widely used in the field of macroeconomic analysis. In order to make DSGE model to deal with and solve easier, a typical approach is to use linear rational expectations to model and the local approximate. LRE model in the solution process, however, may appear multiple equilibria, we usually call it indeterminacy. In general, the presence of indeterminacy can lead to the emergence of two results: on the one hand, affect the fundamental shocks conduction, such as technology shocks and the impact of monetary policy transmission; on the other hand, the impact of sunspot shocks can affect the balance allocation efficiency and lead to determinacy circumstances do not appear in the business cycle.In this article, we will estimate DSGE models Bayesian statistical inference method extension to the parameter space uncertainty region, and building weight based on parameter space, and these probability values will be used for conduction weighted in parameter estimation and structure impact. Due to collaboration considerate the certainty and uncertainty of economic system, therefore, this research method allows for the measurement uncertainty based on analysis of the impact of conduction as well as the impact of sunspots on the overall economic fluctuations possible. Specifically, we use quarterly China real output growth, inflation and nominal interest rates from 1992: Q1 to 2009 Q3, while using monetary DSGE model to analysis and test the effectiveness of monetary policy. Firstly, we set the prior distribution, followed by calculating the marginal data density of the certainty and uncertainty, and posterior probability values, which determine the economic system in the certainty or the uncertainty region. Secondly, we based on different prior settings, under different sample interval estimation results, using impulse response function to analysis the impact of the sunspot shocks and the three basic impact (monetary policy shocks, demand shocks and supply shocks) on the three endogenous variables (output, inflation and nominal interest rates), and describe and compare the change degree of the impact of transmission under the certainty and the uncertainty. Finally, based on variance decompositions, we analysis the impact of the contribution structure of the output gap, inflation and nominal interest rates three endogenous variables.5. Today, the dynamic stochastic general equilibrium model is a very important analysis method in the field of macroeconomic, while the model as an effective supplement to the original model has been used in a number of developed countries. However, the DSGE model in emerging market countries has not been applied widely and fully, Liu bin (2008) has pointed out that the existing models of central bank can not solve economic significance of issues of expectation and identification constraints well, which were affect the model result some extent. From the use of foreign experience, the DSGE models are able to solve these problems better. Therefore, based on DSGE model to develop central bank monetary policy analysis model is the direction of our future efforts.As we all know, real output growth, inflation and nominal interest rate are three key macroeconomic variables most concerned by monetary policy authorities. In addition, Ingram and Whiteman (1994) point out that the King et al (1988) proposed the priori information of basic stochastic growth model has great benefit in predicting such as production, consumption, investment and hours worked macroeconomic variables. For this reason, in this text, we have expanded and extended the relevant studies in two aspects. Firstly, we use quarterly China real output growth, inflation and nominal interest rates from 1992: Q1 to 2009 Q3, we will put the prior information got from new Keynesian monetary DSGE model into 3 variables VAR models, and thus measure the predictive ability of the constructed DSGE-VAR model, meanwhile, compare the predictions of DSGE-VAR model, unconstrained VAR model and the Minnesota-VAR model. Secondly, we conducted the following two levels of policy analysis: on the one hand, we have based DSGE model and DSGE-VAR model, detailed analysis and compare on the impulse response functions of identified policy shocks on real output, inflation and nominal interest rate. Specifically, in order to obtain a recognized result, through VAR approximation of DSGE model, we build orthogonal matrix, which will change simplify shocks into structural shocks. On the other hand, owing to DSGE-VAR model, we identify and characterize the concrete results of"moderately easy"monetary policy and the"moderately tight"monetary policy changed. Specifically, in order to predict the effect of the policy rules changes, we assume that the existing poor predictor of a DSGE model and a VAR model without policy regime change, in order to accurately estimate the effect of policy intervention, we modify the predictions of DSGE model, and make it effectively applied under the existing policy system, and this amendment is based on the DSGE-VAR model.
|
PDF Full Text Request