Essays on computational economics
thesisposted on 10.03.2014, 11:32 by Piotr Zbigniew Jelonek
This text consists of two parts. In chapters 2-3 the methods are developed that enable the application of tempered stable distributions to measuring and simulating macroeconomic uncertainties. In contrast to the tools used in finance, these results are applicable to low frequency aggregated data, which typically displays tails of moderate gravity. Thus thay are particularly useful in modelling macroeconomic densities. The new methods may be readily employed in Monte Carlo simulations of possibly skewed, moderately heavy-tailed random variates with arbitrary excess kurtosis. In chapter 4 a computational model of endogenous network formation for the inter-bank overnight lending market is proposed. The structure of this market emerges from interactions of heterogeneous agents who are endowed with assets, liabilities and take into account investment risk. As all the banks are large and their trading affects the prices of risky assets, the costs of price slippage breaks the symmetry of portfolio problem, making inter-bank borrowing and lending more desirable. The model takes into account three channels of contagion - bankruptcy cascades, common component of risky asset returns and erosion of liquidity. The network formation algorithm outputs the ensemble of optimal transactions, the outcome of the corresponding link formation process is pairwise stable. This framework is next employed to investigate the stability of the endogenously generated banking systems.