Extreme Dependencies and the Idiosyncratic Volatility Puzzle

在上一个项目的背景下，我们可以证明(i)与市场表现出强烈极端依赖的股票（即崩盘敏感股票）提供了明显的正回报溢价，（ii）流动性也可以观察到类似的效应，（iii）回报模式也在国际上成立。基于基于copulas的测量系统风险的开发方法（该方法允许对非线性依赖关系进行灵活的建模），该持续资助的项目将专注于特殊波动性。由于股票的总收益波动率由其系统性和特质性组成，因此，这是初始项目的自然延续和延伸。如果系统性风险的度量和程度发生了变化，那么在给定的总波动率水平上，特殊波动率的数量也必然会发生变化。在新项目的背景下，将分析以下两个具体子项目：(1)基于copula的特殊波动率和特殊波动率难题（IVP）：我们想分析以传统方式测量的特殊波动率（即不通过风险因素的线性暴露来解释的风险）是否与基于copula的更精确的系统（最终是特殊）波动率测量的特殊波动率不同。此外，我们想要分析股票收益与特殊波动率（IVP）之间理论上意外的负相关关系是否是由于先前工作中对特殊波动率的不正确测量造成的，以及一旦我们使用新提出的基于copula的特殊波动率测量是否会降低。(2)每日股票排名和IVP：许多报纸和网站定期公布日收益最高和最低的股票排名，读者很容易看到。我们预计潜在投资者的注意力将转向这些股票。因此，在第二个子项目中，我们想要分析出现在每日收益排名中的股票（无论它们出现在排名前列还是垫底）是否由于引起关注而在短期内被高估，并且随后与未出现在每日排名中的股票相比表现出负异常回报。此外，我们想分析这种潜在的模式是否也有助于解决IVP。这并非不可能，因为出现在每日排名中的股票也可能表现出更高的特殊波动性。建议的IVP分析可以使用传统的或我们新开发的基于copula的特殊波动率测量方法进行，因此这里建议的分析也与第一个子项目密切相关。

In the context of the previous project, we could show (i) that stocks that exhibit strong extreme dependence with the market (i.e. crash-sensitive stocks) deliver a pronounced positive return premium, (ii) that similar effects can be observed for liquidity, and (iii) that the return patterns also hold internationally. Based on the developed methodology to measure systeamtic risk based on copulas---that allows for a flexible modelling of non-linear dependencies---the projects from this continution grant will focus on idiosyncratic volatility. As the total return volatility of a stock is comprised of its systematic and its idiosyncratic component, this is a natural continution and extension, respectively, of the initial project. If the measurement and extent of systematic risk changes, the amount of idiosyncratic volatility for a given level of total volatility necessarily also has to change. In the context of the new project, the following two specific subprojects will be analyzed: (1) Copula-based idiosyncratic volatility and the idiosyncratic volatility puzzle (IVP): We want to analyze whether idiosyncratic volatility measured in the traditional way (i.e. as risk not explained by linear exposures to risk factors) is different from idiosyncratic volatility for the case of a more precise copula-based measurement of systematic (and eventually idiosyncratic) volatility. Furthermore, we then want to analyze whether the theoretically unexpected negative relationship between stock returns and idiosyncratic volatility (the IVP) is due to the incorrect measurement of idiosyncratic volatility in previous work and whether it is reduced once we use our newly suggested copula-based measure of idiosyncratic volatility. (2) Daily stock rankings and the IVP: Many newspapers and websites regularly publish rankings of the stocks with the highest and lowest daily returns and are thus very visible for readers. We expect that the attention of potential investors is driven towards these stocks. In the second sub-project we thus want to analyze whether stocks that appear in daily return rankings (irrespective of whether they appear in top- or bottom-rankings) are overvalued in the short-term due to the attention they raise and subsequently exhibit negative abnormal returns as compared to stocks that do not appear in the daily rankings. Furthermore, we want to analyze whether such a potential pattern could also contribute in sovling the IVP. This is not unlikely as stocks that appear in the daily rankings probably also exhibit higher idiosyncrativ volatility. The suggested analyses of the IVP can be conducted with traditional or our newly developed, copula-based measures for idiosyncratic volatility and the analysis suggested here is thus also closely linked to the first subproject.