| This thesis deals with macroeconomic dynamics. In chapter 1, I study a one-sector growth model with endogenous discount rate of the sort proposed by Meng [2006]. I extend the model into a heterogeneous agents model with respect to initial wealth, and investigate whether the wealth distribution may converge to a degenerate distribution. I find that if an agent's decision only depends on his or her reference group and if consumption is more important in discounting than income around the steady state, then convergence to a degenerate distribution is a unique solution. Furthermore, if an agent's decision depends on average variables of overall society, I find that there exists a continuum of steady states.;In chapter 2, I introduce three mechanisms into otherwise standard Aiyagari [1994] model to generate a realistic wealth distribution. The three mechanisms include: i) a wealthdependent shock: labor income shock is wealth-dependent; ii) misspecification: people do not take into account the dependence of the labor income process on wealth when they make consumption decisions; iii) status-seeking from some threshold: there is a direct utility gain from being wealthy. The main findings are as follows: i) Wealth-dependent labor income shock with misspecification helps to explain wealth concentration but cannot fully explain the share of the top 1% in wealth distribution. ii) In the full model with status-seeking, the share of top 1% becomes closer to the data.;In chapter 3, I build a simple model (two-dimensional discrete dynamical system) to study the interactive dynamics of short-term nominal interest rates of the U.S. and international risk appetite. Main implications from the research are the followings: First, strong interaction between short-term nominal interest rates of the U.S. and international risk appetite can induce bifurcations of the dynamical system: stable fixed point to limit cycle and then to chaos. Second, a numerical experiment suggests two possible explanations for rising variance ratios: the reduction of random shock and the bifurcation of a dynamical system. This finding hints the potential of complexity measures (such as Lyapunov exponent and permutation entropy) as early warning signals. |
