Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.761908
Title: Interactions between mutual fund flows, asset performances and investor behaviours in United States
Author: Zhang, Dong
Awarding Body: University of Glasgow
Current Institution: University of Glasgow
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Mutual fund is a burgeoning business in not only US but the world. There is a growing tendency that participations of individual investors in financial market are migrated to mutual funds, an indirect channel to invest. Thus, the flows to and out of mutual funds, once a neglected topic, are becoming a new field for financial study. The primary instrument and subject of my PhD is mutual fund flows. Mutual fund flows have special merits for academic research. Firstly, it is purely driven by demands but not supplies, as the supply elasticity of mutual fund is nearly infinite. The characteristic reveals investor behaviours and decisions in a mass scale. Traditional instruments for behaviour studies relies on asset prices and volumes, which are less exogeneous as they are driven by both demand and supply. Secondly, mutual funds specify their objectives and asset classes in prospectus. The characteristics help us understand how investors respond to changing market conditions by changing their exposures on asset classes or styles. Thirdly, a majority of individual investors in mutual funds (as suggested by ICI statistics) provides a natural field for behavioural finance. Fortunately, data is available not only at aggregated level, but also individual and account level, which serve as a great supplement to the existing studies using trading data. The second chapter is based on a simple hypothesis: if flows are (rationally) responding to fund performance, what information does the flow-performance sensitivity convey? How flow, a measure of actual fund investor trading decisions, helps us decompose and finely measure the outcome of these investors? The study is based on several established papers on flow performance relationships in mutual fund market. Warther (1995) is a pioneer paper that discovers a significant correlation between flows and performance. Sirri and Tufano (1998) discovers a convex-shaped flow-performance function and attributes the cause to asymmetrical information. Berk and Green (2004) established a model in which investors trade against good performers and against bad performers but funds themselves suffer from diseconomy of scale. As the fund change in size, it deviates from optimal portfolio size and IV result to better or worse performance. Huang, Wei and Yan (2012) argues that flow performance sensitivity is a rational investor learning process. Based on their arguments, I obtain a simple but effective proxy for investor sophistication: the sensitivity of flows to recent (abnormal) performances. To granularly measure their respective performance, I decompose their performance into three aspects: abnormal returns, fees and timings, a scheme proposed in Fama (1972). The abnormal return is alpha on a four-factor model, which is a traditional before fee, relative measure of whether a fund has beat the market. Fee selection takes into account the average fees that jeopardize the performance and timing cost is measured by “performance gap”, a concept used in Nesbitt (1995); Dichev (2007); Friesen and Sapp (2007); Bullard, Friesen and Sapp (2008). The result is that sophisticated investors earn higher risk adjusted returns and avoid high fees. In addition, investors’ timing performance can be greatly improved by trading less, with the most significant improvements seen on most sophisticated investors. The research question in third chapter is: is there a calendar effect for flow-performance relationship? Does the shape of the function change across the months and what drives the change? The study fills the gap by emphasizing several exogeneous factor of flow-return relationship such as portfolio rebalance and tax-loss selling which interact with calendar dates. Previous literature commonly finds a convex function. Chevalier and Ellison (1997) is first to document the convexity and they argue the convexity may incentivize agency problems. Sirri and Tufano (1998) explained using information search cost and Lynch and Musto (2003) explained with survivalship bias of mutual fund strategies. However, all the study examines only average shape of the flow-performance function. None of them attempt to tackle calendar effect. Calendar effect is potentially a strong determinant of flow performance relationship. Factors such as tax-loss selling (Constantinides (1983)), portfolio rebalance, disposition effect (Kaustia 2011)) and seasonal variation in risk appetite (Kamstra et al. 2017) may interact with dates and change the flow-performance relationship. In this study, I conduct a similar flow-performance regression for each month. The regression is piecewise which separates the sensitivity of mutual fund flows to returns into five parts. I also construct a concise measure of whether a group of funds are bought or sold at any time during the year to disentangle several confounding effects. I find that the shape of the function does change throughout the year and they are affected by tax-loss selling and portfolio rebalance. In fourth chapter, I focus on a special group of funds, the leveraged funds, which mainly caters for day traders. The research question is whether their flows reflect market wide sentiment. Leveraged funds are funds that allows investors to bet on daily performance of stock indexes with leverage and direction. As these funds track only daily index returns and investment horizon longer than one day will result to material deviation from index returns, these funds are unlikely used by mid- or long-term optimizers. As common study suggest too much trading can be harmful (Barber and Odean 2000), I notice that the flows for these funds may be sentiment driven. In this study, I obtain daily flows of nearly 100 largest leveraged funds trading in US and extract the first principal component from these funds. In addition, I follow Baker and Wurgler (2006a) to construct a daily sentiment index (the alternative sentiment measure) from several market variables, which are purposely chosen to be unrelated to fund markets. I find that the first component from leveraged funds is associated with investors’ migration between bull and bear funds and it has strong correlation with our alternative daily sentiment measure. In a later test, the two sentiment measures have similar price impact as a hypothetical sentiment measure would have. I have also examined the limits of arbitrage effect proposed in Shleifer and Vishny (1997). The sentiment component predicts similar cross-section of price revision for up to 7 days into future.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.761908  DOI: Not available
Keywords: HG Finance
Share: