| Popis: |
This thesis is composed of three parts, under the main theme of Extreme Value Theory, applied to various areas in financial economics. Tail events are low probability, high impact phenomena. Therefore, understanding distributions of various financial economics dataset enables us to understand the underlying risks. Part I explores a banking network in an advanced emerging economy, in the post-crisis period. Although static in time, it captures the dynamics among factors such as balance sheet size, interconnectedness and systemic importance. Network structure is examined by classic network theory, with reference to extreme value theory. It worths noting that interbank network structure sometimes exhibits (double) heavy-tailedness, which resembles an un-balanced interconnected- ness among nodes (banks). Network structure may have an impact on network (in)stability. This is realized by a simulation, using a defined a sequential default mechanism, taking into account all risks, arising from any single (or multiple) bank(s) default(s). Results show credit risk (often referred as first round effect) does not post threat to the banking sector, as much as fire sales and liquidity shortage (often referred to as secondary effects). Part II studies the household income and consumption inequality of the country Spain. It firstly estimates the lognormal body, as well as both tails of the income and consumption distributions, across pre, during and post-crisis periods, using household finance data collected by Bank of Spain. It then examines the Income Elasticity of Spending (consumption) on various goods, including (a) food, (b) other non-durable goods, (c) vehicle, (d) other durable goods. Method- ology adopted included the random effect and fixed effect models. Results show that income inequality causes heterogeneity in household consumption behaviours on various goods. House- hold belonged to different income percentiles display asymmetric impacts on their consumption on various goods, due to income inequality caused by crisis in 2007-2008. Part III studies a UK occupational pension fund. It shows the pension income displays classic Pareto behaviour, as pointed by previous literature studying income distribution. It then shows the negative relationship between pension income and mortality experiences. Applying Expectation Maximization (EM), the pensioner population can be broken down into several sub-populations, according to pension income size. This pensioner bucketing strategy can be used in pension de-risking solution - longevity swaps. In the emerging market for longevity risk transfers, bespoke solutions [swaps written on the pension fund that provides perfect hedges] have so far played a major role, as they allow pension funds and annuity providers to hedge the risk of mortality perfectly. Indexed instruments, such as swaps written on the mortality experience of a reference population, represent a cost-effective alternative to bespoke solutions, but have so far been less popular. One of the main reasons is that they give rise to basis risk, the risk of mismatch between the mortality experience of the hedger and the one tracked by the index. The pensioner bucketing strategy developed in this paper quantifies basis risk, at the same time, shows that the quality of a hedge can only be assessed by jointly considering the cashflows from the hedging instrument and its mark-to-market/model dynamics. Once such broader perspective is adopted, the true magnitude of basis risk can appear substantially lower than expected. Indexed based longevity solution becomes practical. |
