As investors continue to dissect the factors behind hedge fund performance, academic studies too are shedding further light on some previously obscure ingredients to a hedge fund’s success that could in fact be quite critical. Herman Melville once wrote, “A thousand fibres connect us with our fellow men; among those fibres, as sympathetic threads, our actions run as causes, and they come back to us as effects,” and this seems particularly pertinent in today’s increasingly interconnected financial services environment.
Marc Gerritzen of Berenberg, Jens Carsten Jackwerth of the University of Constance and Alberto Plazzi of the Swiss Finance Institute and University of Lugano will shortly be publishing a fascinating paper that seeks to shed light on how hedge fund managers influence each other, be this via having worked together at the same institution or indeed remaining in ongoing social contact.
In their soon to be published research paper, entitled Birds of a Feather – Do Hedge Fund Managers Flock Together?, the trio analyse publicly available data provided by the Financial Conduct Authority to analyse the past and present employment trends of UK-regulated hedge fund managers, including social ties and prior employment experience.
The paper comes at a time when investors and regulators are enjoying unprecedented levels of oversight and transparency on an industry that was once considered somewhat obscure. It has paved the way for more cogent research on hedge fund managers’ behavioural patterns and the way this impacts fund performance. Almost two decades ago, I wrote an article in an FT journal that asked how investors and risk officers could elicit new insight on fund managers’ likely risk management behaviour using techniques employed by the US Navy to assess its officer candidates. At the time, the hedge funds industry was still germinating, there was less public money in alternative investment schemes, and the overall levels of transparency were a far cry from what they are currently.
In 1998 we had less data to play with: there were fewer hedge fund databases, and a much smaller sample of managers reporting. Available data on managers and their movements was paltry, and the focus had to be more on trying to assess the personal characteristics of the fund manager, rather than data on past activities. The emphasis was more on the potential that an individual had to wreck a fund, than dissecting the contribution of the individual to alpha.
The ties that bind
This new research paper reveals how fund managers who share a common experience in the industry are likely to have been exposed to similar training and to share many of the same personal connections. They assert that these characteristics tend to be ‘mobile’ and to manifest themselves later, when managers are setting up their firms. The authors contend that the social ties between hedge fund managers explain a significant part of their success and that investors should be taking this into account when evaluating performance. Indeed, they go further to argue that social interaction between managers could play a significant part in their overall success.
“Examining the work trajectories of UK hedge fund managers through our definitions of social ties reveals a highly interconnected world. In fact, the great majority of funds in our dataset share connections of some sort. We utilise this information to ask whether social ties and work experience can explain similarities in hedge fund returns,” the authors say.
Theresearch was restricted to funds which report in USD. The data it draws on consists of monthly information on 21,547 hedge funds (organised in 9,147 management companies) from January 1977 to December 2012, of which 16,374 are dead funds and 5,173 are live funds. Keeping the dead funds addresses potential survivorship bias.
The authors note that the backfill bias is due to the possibility of reporting older (and typically higher) returns in the database at the time of joining. If performance is correlated with prior work experience, this practice would induce a selection bias and lead to over-estimation of the effect of social ties. The paper follows the practice established by prior studies and removes the initial 12 months of each fund’s return data history.
Investors already recognise that managers leaving successful teams on the trading desks at investment banks, or who have cut their teeth in the tough environments within leading hedge funds, bring with them some of the magic fairy dust that is accrued during such an experience. Indeed, senior hedge fund managers themselves will seed former employees, confident in the knowledge that their abilities, honed in their original shop, will also bear fruit in their new firm.
Funds of funds, the authors argue, should be paying more attention to social ties and social connectivity ought to be added to the formal due diligence process.
The paper adopts a different modelling perspective from the standard factor-based analysis of the past. Mandatory filings with the Financial Conduct Authority, which have been carried out since 2002, provide an ideal source of data about the movement of individuals between financial institutions and asset management firms in the UK.
“Using the FSR [UK Financial Services Register] has clear advantages with respect to other available sources. The fact that the FCA requires reporting rather than voluntary disclosure increases the completeness and accuracy of the FSR information, which is comparable to existing databases on US executives such as the widely used Boardex. The resulting sample should be devoid of any selection bias and is survivorship-bias free as the FSR also keeps track of dead companies,” the authors say.
The research assesses whether a fund manager’s professional career, the essential training he might have received within the industry where he originally worked, will affect his approach to money management.
For example, managers coming out of the insurance industry may develop a different approach to risk management from former employees of banks. Managers may have therefore learned portable skills which will have an intrinsic effect on the way they run both their portfolios and their businesses. Much of this may even be unconscious bias, but it is something investors ought to be aware of.
Correlations and performance
The authors of the paper also add to the growing body of theory that assesses social connections between managers, for example managers who may have worked together in the past, and may still exchange ideas. The sharing of views and information may lead to potential correlations in trading behaviour. The consequences of these social ties may indeed show up in funds’ returns and the various other components of performance: exposures to risk factors (i.e. beta), abnormal performance (i.e. alpha), and the unexplained (mean zero) idiosyncratic component.
Investors will probably be most interested in how the research helps them to zero in on managers with the most potential for alpha. It certainly does seem to indicate a strong basis for social ties as part of the overall analytical process:
“We find that industry experience of a manager significantly influences future alpha with managers from pension funds and banks outperforming and managers from investment management underperforming. We also explore the economic gains from loading on connected funds through a bootstrap exercise. There, we randomly group funds into portfolios and then sort these portfolios based on the degree of industry connectedness. We document that the decile of most connected funds outperforms the decile of least connected funds by a significantly positive abnormal return of about 60 basis points per month. Overall, the results lend further support to the claim that managers’ connections have ultimately beneficial effects on performance.”
The economic value of hedge fund managers’ networks was assessed via complementary analyses: one area of focus was regression analysis using industry dummies, where these were defined based on prior managers’ work experience. This allowed the authors to zero in on industry rather than company background using a more distinctly limited set of factors, which enhanced statistical power. Sharpe ratio, alpha and information ratio were used as measures of performance, with performance calculated over the evaluation period and industry dummies defined at the end of the network period.
The research found that hedge fund managers who worked at a pension fund or bank prior to becoming a hedge fund manager delivered superior performance with respect to their peers. For example, alpha was more than doubled if the manager hailed from the pension fund industry. The effect was somewhat similar for former bankers. Interestingly, the only other significant industry effect was former investment managers, where negative effects dominated. This is a phenomenon which has been observed periodically in the hedge funds industry, when larger numbers of portfolio managers and analysts leave larger investment management businesses to set up hedge funds, attracted by the higher levels of fees hedge funds have traditionally charged. Many, however, demonstrate that they are less effective in a more unconstrained environment. This is certainly consistent with anecdotal evidence The Hedge Fund Journal has from allocators.
Crucially, the authors allow for hiring climate in their results: with higher levels of recruitment in the finance industry – i.e. overall net number of new hires – the average skill of managers declines. Lower hiring numbers mean only the most skilled managers can find jobs, but this also affects performance.
The paper notes that the social networks of managers who previously worked in banking have contributed heavily towards performance and skills acquired.
The authors argue that the components of the UK hedge fund industry are densely linked through social ties, which are found to be important determinants of proximities in any two hedge fund pairs. In particular, managers who may have worked in the same industry, and for the same employer in the past, may demonstrate how this contributes to risk exposures and alpha. Social connections measured by an overlap in prior employment experience will only explain the differences in the idiosyncratic component of returns. The paper claims these connections play a much greater role for funds that invest in styles that are particularly sensitive to the exchange of relevant information, such as event driven or merger arbitrage funds. However, this area of research must also touch on the more contentious issue of the degree to which IP within systematic approaches can filter out from a firm as employees move to new funds. Although the research does not focus on this element per se, further work may provide interesting analysis on this topic.
Competing channels
Can we conclude that social connections are ultimately responsible for similarities in hedge funds’ performance? The paper addresses several competing channels. For example, managers may self-select, and find themselves working for the same employer because of similar preferences and risk profiles, or there may be other network-related conduits like access to local information that are responsible for similarities in trades. Adding managers’ personal characteristics or controlling for geography does not, however, dissipate the effect of prior employment connections.
The authors recognise the potentially more challenging task of controlling for managers’ skills. The argument here is that (past) employers may hire individuals with similar levels of skills. The research exploits the fact that a subset of managers in the data have previous experience in the hedge funds industry. They then use the abnormal performance in the previously managed fund as a control variable. If skilled managers tend to outperform their peers over time, they argue, this should go a long way toward absorbing the effect of innate intelligence.
The fact that fund pairs of connected managers are closest in performance poses the question of where the differences occur in the average returns of hedge funds (not pairs) that are grouped based on the extent of their connectedness. Loading on portfolios of connected funds generates a positive spread in terms of risk-adjusted performance compared to unconnected funds. The authors’ conclusion is that the exchange of information through social ties ultimately has a positive effect on performance.
What are the implications of this research for the industry? Certainly they imply that managers’ social ties should be considered when evaluating the performance of a single hedge fund or a portfolio, for example with funds of funds. Social ties should also be an important aspect of investors’ due diligence processes when deciding which funds to invest in.
The paper argues that: “On the other hand, the evidence…that connected funds exhibit more similar return components suggests that loading on these funds may harm portfolio diversification. Moreover, correlated trades among funds of the same network may lead to the funds trading securities based on the same signals. This behaviour could endogenously generate a liquidity spiral when the signal reverts, and depress their performance. Which one of these effects prevails is ultimately an empirical question.”
Conclusion
This study will make an important contribution to the existing body of work on the subject, particularly on the risk factors underpinning the performance of hedge funds. Little work has been done on the topic of social connections, but the results do seem to bear out that further analysis needs to be done on social ties and work histories by institutional investors.
The authors argue that managers with overlapping work experience are more likely to demonstrate idiosyncratic components of their hedge fund returns than non-overlapping managers. As the authors say, the results may be capturing the exchange of information among managers in a network that lead to similar bets, for example, distressed securities, convertible securities or merger arbitrage. These are hedge fund strategies that are more reliant on communications – and hence social networks – than others.
In the digital age, we are confident that social connectivity between managers will play a bigger role in the performance of managers across a range of strategies. In particular, we envisage that more complex, machine-driven analysis will soon be able to shed yet further light on this topic. Given that some hedge funds now report daily or weekly NAVs under European directives (e.g. UCITS), there may be scope for more detailed mapping of performance versus established career and social overlap.
Outside the FCA data resources, for example, other publicly available information resources are being added to on a weekly basis which can provide additional insights on portfolio managers. Further work has yet to be done on this, and it requires a considerable level of data analysis that is currently only within the province of sophisticated algorithms, but will eventually deliver further insights on managers’ behavioural biases which can only contribute further to investors’ insights.