Human behavior and tipping point of bank runs

• Main Author: Qiao, Huang
• Other Authors: Huang Weihong
• Format: Theses and Dissertations
• Language: English
• Published: 2019
• Subjects: DRNTU::Social sciences::Economic theory
• Online Access: https://hdl.handle.net/10356/106429; http://hdl.handle.net/10220/47959
• Institution: Nanyang Technological University

To figure out the reason why bank-run phenomenon still frequently occur, this thesis provides insights from animal spirits, the term using in economics goes back as far as John Maynard Keynes (1936). Animal spirits, instead of just employing mathematics and strong economic analysis, would lead people to run on banks without shyness and doubt. A bottom-up behavioral model is built to model bank runs in order to consider the bounded rationality of human beings. It is shown with agent-based modeling approach that animal spirits of the depositors together with endogenous optimism and pessimism can cause liquidity shock. In particular, bank run will be trigged if the animal sprit reaches to a tipping point. The model is tested with a set of empirical data, which shows that the effect of animal spirits is significant in reality. To further explore the internal mechanism of how the animal spirits can trigger the bank-run phenomenon, the nonlinear dynamic probabilities of bank runs based on the global games approach is simulated from the perspective of connectionist model of decision making. In this framework, heterogenous agents hold highly correlated but different beliefs about the real payoffs. When real-world bank runs involve diverse elements, it is hard to conclude bank run is only caused by miscoordination (e.g. Diamond and Dybvig 1983) or the deterioration of the quality of the bank’s assets (e.g. Allen and Gale 1998). Besides, the probability of bank runs (PBR) is neglected by the classical models. Therefore, an internal “mental model” that agents have integrated cognitive-affective network is proposed. As the affective processes play significant role in decision making, the pessimistic announcement may cause bank-run even when the fundamental economics is good. Both the rise of late payoffs R and early payoffs r will decrease the effect of affective process. Whether the increased risk sharing, that is, the increase of r will increase PBR is based on the level of uncertainty. The increasing of late payoff is beneficial for preventing bank runs, but it is hard for bank to offer enough payoff when the economy is too bad. After the conclusion that animal-spirit exists in bank runs and an internal mental mechanism to show the reason behind this phenomenon has been proposed, there are still two key points for the estimation of PBR. First is the best setting for affective process, that is, the parameter in affective magnitude section, second is the tipping point with different combination of early and late payoffs. Thus, the parameters and critical values in bank-run estimation is concerned. For the adjustment of affective-process parameters, the granger causality test and impulse response function are adopted. Based on the selected parameters, the results we have concluded are proved to be correct, and the affective-process effect becomes clearer. To figure out the tipping-point effect by threshold autoregressive (TAR) model, the result shows that the increased risk sharing, that is, the increase of early payoff r will cause this kind of effect, but it does not impact on the tipping-point value. On the other hand, the increase of late payoff R will increase the tipping point, which means that it will prevent even eliminate the bank-run phenomenon. Generally, when PBR passing 0.2, it is possible that bank run happens, and when PBR passing 0.25, it becomes very easy to run to bank and it is too late to restrain. This dissertation develops from the psychological foundation of human behavior and gives explanation for bank-run phenomenon with the framework consist of computational economics, behavioral economics, neuroeconomics and econometrics.