Archive for Portfolio Optimization

Professional Money Managers Admit: Risk Management Can and Should Improve

Posted in Manager Selection, Portfolio Optimization, Risk Management with tags , , , , , , , , , , , on March 23, 2009 by evd101

By Erik L. van Dijk

The EDHEC Risk and Asset Management Research Centre  has done a lot of good work during recent years. Especially in the area of hedge fund research, where it tried to bridge the gap between sophisticated academic work on the one hand, and the use of these concepts in practice on the other.  Lately, their work in the area of Asset Allocation has drawn attention as well. In 2008/09 researcher Felix Goltz of the French institute first surveyed a group of almost 300 high level European money managers, and then – based on the outcomes of the survey – contacted them again to discuss the results. The study led to some remarkable conclusions that shed new light on both the quality of players in the industry AND risk management quality during the credit crisis. We can therefore say that the study corroborated findings discussed earlier in this blog.

You can find the full EDHEC paper in the link below (at the end of this entry), but first we will provide you with a summary of the main results.

The main conclusion of the study is a harsh one:

practical applications of portfolio construction/optimization techniques and risk management in the institutional investment world fall short of what has been described in the academic literature as ‘state-of-the art’.

To some extent your author should feel guilty about this, because our Markowitz-Van Dijk framework for risk management and optimization/rebalancing is one of the more advanced systems that are available to practitioners when trying to better manage risk. Actually, MIT professor Mark Kritzman and State Street Scholars Seb Page and Myrgren, wrote a nice paper in 2007 indicating that our methodology was actually superior to other available methodologies. But there are more good systems. What is important is that risk management is not about doing things right whenever we feel like ‘doing another risk assessment’. No, what it is all about is that we apply the technique chosen ALL-THE-TIME and USING THE RIGHT ASSUMPTIONS.

The EDHEC study indicates that there is still a long way to go. Professional money managers do – as a group – apply optimization and risk management tools, but often not the right ones. Or they use the right ones, but use the wrong assumptions, or they do things right, but forget to continuously apply them.

A few of the major mistakes:

How to measure risk? Absolute versus Relative, and What About Outliers?

Optimization of an investment portfolio is about trying to achieve the maximum amount of return for a specific level of risk during a specific investment period. Or, the minimum amount of risk for a specific level of return. With respect to the calculation of returns there isn’t much of a problem. We all know that it is equal to the price return plus percentage dividend (in the case of stocks) or coupon interest (in the case of bonds).  More problematic is the derivation of the right risk definition. It turns out that by far the largest group of professionals uses an absolute risk definition, be it the so-called Value-at-Risk (VAR) or the volatility. Absolute risk factors are good when analyzing a portfolio decision, without comparing it to some kind of benchmark. However, professional investors are by definion always invested in something. They could for instance hire our friend King Kong with the darts to manage portfolios or flip coins. The market average of such a strategy would then provide a good benchmark, with the difference compared to this benchmark being the quality/skill (or lack thereof!) of the professional money manager. Strange therefore that only a minority of professionals is thinking first and foremost in relative terms, notwithstanding the fact that folkore (?) within professional circles always explains that wealth management for private investors is about absolute risk and return indicators, with the real institutional professionals focusing on relative return and risk indicators. It is true that professionals do pay special attention to alpha as a measure of excess return (= Return strategy minus Return benchmark) and tracking error (= standard deviation or volatility of the excess return), but somehow they do not integrate this in the optimization or risk framework. They seem to use excess return for the calculation of variable fee (bonus!!!!) levels and tracking error as a kind of constraint (‘avoid tracking errors in excess of x percent’).

VAR and the Normal Distribution of Returns

Another mistake made is that – when applying absolute analysis – the majority is using a Value-at-Risk (VAR) framework while assuming a normal distribution of returns. However, we all know that – during extreme periods like the one at hand – the likelihood of extreme events happening is far larger than what the normal distribution tells us. Normal distribution theory assumes that events that are 3 standard deviations away from the average have a chance of occurence of 1 percent maximum. Events 2 standard deviations away from the average have a chance of occurence of 5 percent maximum and events 1 standard deviation away from the average  about one-third. Result: dozens of chief risk or investment officers of even big firms (e.g. Bob Littermann at Goldman) had to tell journalists last year that things that initially were supposed to happen just once every 10,000 years happend two times within a year. In other words: due to the normality assumption there was an underestimation of the ‘fat tails’ in return distributions caused by the fact that people either exaggerate when in panic (excess pessimism) or when too enthousiastic (irrational exuberrance). As a result of the fat tails the value-at-risk is much larger than what can be derived based on the mean and variance. Actually: what is the use of applying VAR when assuming that distributions are normal? In that case the whole distribution is already defined by the mean return and the variance of return!

‘Too complicated for the clients’

Some managers even suggest that they do not apply top-level risk management and optimization techniques because clients don’t ask for them, or would actually not understand them. That is an insane line of reasoning when you would for a second see the investment specialist as a kind of medical doctor trying to cure the financial health of his patient, a.k.a. the client. The average patient in a hospital doesn’t understand anything about Medicine or all the complicated machinery and/or pharmaceuticals used by the doctor. But be sure that patients wouldn’t be too happy if their doctor thinks like this! Thank God it was just a minority of specialists putting forth this line of reasoning. But the fact that alleged professionals dare to think like this is outrageous and indicative of the necessity for increased professionalism within the industry.

The Not-Invented-Here Syndrome

Academic risk and optmization techniques do have a large quantitative component at a level considered difficult by most graduate students or professionals that went for an MBA. It is Ph.D kind of stuff. As so often things that are considered difficult are surrounded with a feel of them being ‘magic’ (at best!) or even ‘pure theory’ (very often!). Result: to the extent that senior management in an investment organization is willing to apply these methods it is too often the case that the one’s that have to decide are not themselves specialists. This creates a feeling of not being totally in control as a result of which there is a preference to use these techniques only when the knowledge is available internally. Why? Well, external specialists cannot be that easily controlled and they are also more expensive. ‘Prudent’ thinking implies that you do not spend too much on things that you don’t totally understand, right? But after some time in this inward-looking culture within the machoist financial industry something else happens. People start to actually believe that they are good internally, not hearing the real specialists (who work outside the firm) anymore and being surrounded within the firm by youngsters that won’t dare to be too critical because that might hurt their career. A so-called Not-Invented-Here syndrome can then set in easily.

The Age- or Vintage-related Knowledge Gap

As long as it is true that senior managers are on average by definition older, with knowledge developments in the young investment science going relatively quick, we will notice that the average younger graduate – still working at lower levels within the organization – is more knowledgeable than the older, senior people. Sure, the older ones have more experience and qualitative knowledge. But risk management and optimization are first an foremost about quantitative skills. And that leads to an age- or vintage-year-related knowledge gap with the ones that probably understand less about risk management and optimization being the ones that have to decide. This is dangerous, and we do strongly believe that to some extent the credit crisis is directly related to this problem. Not to the least, because men have a tendency to be overconfident, as we already indicated in an earlier contribution.

What is New! Sign of Changing Times: Mea Culpa

Not much good news, it seems. However, in one area there was a remarkable difference between the reaction of the top-level professionals to the EDHEC researchers when asked about the survey results now, compared to similar reactions before. Before it was almost standard that – whenever something was academically-oriented or mathematical – practitioners would discard it as being ‘theory’, ‘impractical’ and/or ‘irrelevant’. They would never ever confess that it was too complicated and that they actually did not really understand what it was about. Maybe also due to the credit crisis, and the resulting first indications of some modesty, do we now see that there was something of a ‘mea culpa’ with professionals admitting that they were the ones that were to blame most. There was according to them a big need for additional education. Something that we at Lodewijk Meijer would be pleased to be involved with. But as a kind of mea culpa on behalf of all academics or semi-academics with strong professional roots, we also admit that too often university- or business-school-organized courses were indeed to theoretical. The top scholars never had the responsibility over big multi-asset portfolios that had to be managed on behalf of an end-client represented by a board who applied complex goal functions concerning the way in which the portfolio should be taken care of. And that was difficult, because traditional optimization techniques suffer from the existence of too many constraints. Constraints that are the standard reality of life for practitioners. That is why Harry Markowitz and I myself worked on our new ‘near’ optimization technique, later labeled Markowitz-van Dijk by Kritzman c.s. Scientists with at least some knowledge of hands-on asset management should also try to explain things in a clearer way. But the majority of the professionals with self-criticism made it clear that this was only a secundary reason.

What about the US?

The EDHEC study was primarily about asset management within the European institutional community. What about the US? We said at many occasions that the level of asset management in the US was on average higher than in Europe, and we still believe it is. But even there, the situation is not good but at best mediocre. The focus on stock picking and return is too high, which leads to overconfidence. Results can be improved by better incorporation of risk management and portfolio optimization. It is illustrative how Kritzman’s study indicates that something as simple as efficient rebalancing using Markowitz-van Dijk can easily improve net results by 0.5 percent or more per annum. And that is a lot of money for institutional investors who often have assets under management in excess of USD 500 million. A simple statistical tool can in that case already improve net results by USD 2.5 million per annum! It is therefore not surprising that almost none of the participants in the survey listed ‘costs’ as a prime factor for not applying these modern, state-of-the-art techniques.

Conclusion

Professional money management is an industry under construction. Knowledge levels can and should be improved in a few areas, with risk management and portfolio construction certainly being one of them. Based on the findings in the EDHEC survey it is not surprising that something like the credit crisis could happen. When you are not in the driver’s seat in a risky and changing market, how can you expect to be able to avoid major pitfalls and panics?

It is good to see that the prime actors are at least modest in this respect, something that we did not see sufficiently yet when it comes to bonuses. We are convinced that – with the guidance of the regulators and the interest of the main actors in ongoing education in these areas – improvements are possible. Increased competition between providers, better manager selection by specialized parties who help distinguish between the good and the bad, and interesting opportunities for implemented consultants willing to organize courses will collectively help improve things. But in the meantime end-clients have to make sure that they select professionals that do pay attention to portfolio construction and risk management in a sophisticated manner. Something easier said than done, because of the high abstraction and math levels involved in these applications. Whenever you feel that our support could be of help to you in taking the right decisions in this area, do not hesitate to contact us.

FULL EDHEC RESEARCH REPORT:

edhec_publication_portfolio_construction

 

How do people invest? The difference between group and individual decision taking – with a small side-step to the credit crisis

Posted in Behavioral Finance, Financial Markets, Manager Selection with tags , , , , , , , , on March 11, 2009 by evd101

By Erik L. van Dijk

In our previous entry on Fund Manager Selection we noticed what the typical flaws of private investors and their advisors are. Now, before turning to decision taking by the so-called specialists in part 2 of the series on Fund Manager Selection, it is good to take into account resulst from a great paper by Barber, Heath and Odean from the Haas Business School at the University of California Berkeley. In their paper Good Reasons Sell; Barber, Heath and Odean, published in Management Science in 2003 they point out that individual investors take different decisions when being part of an investment club. 

A lot of the research focusing on differences between private and institutional investors incorporates agency-related factors in one form or another. The Agency Theory, going back to the original work of Mike Jensen and Bill Meckling in the mid-70s, indicates that situations in which a principal / decision taker is working in his own self-interest can lead to different outcomes when – for example because of the decision framework or company structure getting too large – he/she outsources part of the decision taking to an agent. Supposedly the agent is required to act totally in the interest of the principal / decision taker, but one can never exclude that the agent will incorporate his/her own agenda with goals/targets in the overall decision framework. Consequence: the end result can easily be less good than what the principal expected.

Before turning to investment decisions by institutions in a future post on Fund Manager Selection (part 2 of the series started yesterday), it is good to re-analyze the Barber et al. paper. In their paper they publish results from tests in which they distinguished between individual investors who took their investment decisions alone, and those who worked together in a group. Now with investors in an investment group still being the sole owner of their share of the overall group portfolio, the agency aspects don’t apply. Any difference in investment style of people in groups versus people on their own is therefore the sole result of behavioral factors.

The results are astonishing. People do decide differently when in groups. And not just that: people in general use decision processes in which ”good reasons” are guiding them more than conscious cost-benefit-analysis-based decision taking! Both the individual investors and the clubs in the Barber et al. study underperformed the benchmark, both on a net and gross basis. In other words: professional investors as a group do outperform. And that implies that the fun, anecdotes about King Kong beating the specialists with stocks picked by throwing darts are demagogic. The tests of monkey versus specialist do always incorporate 5 or less stocks and one ot the key characteristics of a good specialist portfolio is diversification. Taking away the diversification from the specialist, is like taking away bananas from the monkey. If you don’t give him his yellow, curvy food, he won’t be able to pick any stocks, let alone outperform.  

But the differences between the good, bad and ugly in both investor categories can be substantial. And that holds for both the professionals and the amateurs.

One standard mistake made by private investors is that they forget to diversify. Research has indicated that it is best to diversify a portfolio over 15-20 different, if possible not too-correlated holdings. In case your portfolio is too small: stick to funds and avoid individual stocks or other securities. The average investor club or individual diversified his/her portfolio over 5-10 holdings. The excess risk that results from this is one reason for the relatively meagre performance. Diversification was a bit better in the groups, but there diversification did not follow the lines of correlation reduction. It was more a way to ensure that everyone in the group had something funny and interesting to do, with others being able to mingle in the discussions. I.e. we diversify the research, but please make sure that the group discussions in the meetings that decide on portfolio changes are ‘fun’ with all respected members being able to show how knowledgeable they are.

This phenomenon can explain why securitized portfolios that were so instrumental in explaining the credit crisis were not diversified enough. A good group process would be one in which bottom-up specialists prepare things, bring their research into a meeting with eachother and with top-down specialists, with the latter than deciding ultimately about allocations. That is not what is happening in the case of investment clubs I am afraid. Actually, it is not even true for most institutional decision processes either. As top-down specialists, we believe that our cooperation with Noble Prize laureate dr Harry Markowitz (mr Diversification if you like) has helped us to take an objective stand by using a helicopter view in which we give credit to the bottom-up specialists that we categorized as best-of-breed.

No, decision taking in investment clubs follows a different path. Barber et al. show that ‘good reasons’ are the dominating factors explaining buy decisions. ‘Good reasons’ are labels and categorizations providing a seal of approval / quality stamp to individual stocks, albeit that the same could apply to mutual funds. Factors that fall under the ‘good reason’ definition could be; i) incorporation in a most-admired list (e.g. the one by Fortune); ii) strong sales growth of a firm over the last 5 years; or iii) strong stock returns over the last 3 years. With respect to the incorporation in most-admired lists: compare what we wrote in our previous contribution about advertising by mutual funds! And the 3-year return track record is of course one-on-one in line with what we wrote about raw returns in our previous post. ‘Good reasons’ are useful quality stamps, but not necessarily so for achieving above-average investment results: you have to pay for them. ‘Good reasons’ firms are more expensive, so that net results on ‘good reasons’ portfolios can be lower than on a contrarian one. That is how markets work: if something is more desirable, more people want it. And this higher demand translates into a higher price.

The fascinating thing is that people in an investment club are not agents, but principals! So, why do they decide the way they do? In our contribution on Fund Manager Selection by institutional investors we will show that the old adage ‘No one ever got fired for buying IBM’ is an important explanation in an agency setting. But the Barber et al. research indicates that there is more at stake. Group decision taking incorporates innate behavioral problems that can lead to sub-optimal investment decisions with or without agency complications. Investment portfolios should – in the end – lead to diversified mixes of individual holdings or mutual funds that are sufficiently different. Selection of different things involves different qualities. Different qualities require different type of people/specialists. Group processes with different people might lack the ‘social fun’ or a group process in which all like each other, respect each other, have fun working together. And that is where things go wrong. It is most likely that this problem plays a bigger role, the bigger the organization is. Investment clubs are normally limited to just 10-15 people, and already with those small numbers, the problems are tantamount.

Interestingly enough, it is not true that this problem arises solely in groups. Be they institutional (with agency problems included) or private (principal-only). Barber et al. show that the ‘good reasons’ problem is also bothering most individuals on a stand-alone basis when selecting stocks, albeit to a lesser extent. We can extrapolate this behavioral finding into another plea for making sure that boutique, smaller asset management firms in which a charismatic set of key personnel plays a leading role are not forgotten within a diversified portfolio of funds.

Bottom line: taking the right investment decisions – be they the selection of fund managers or the buying or selling of individual stocks, bonds or other securities – is a complicated thing. Too often, do we show weakness to quality-related factors, or non-investment-related agency factors, or do we fall prey to naive extrapolation of historical results. The ‘good reasons’ work by Barber et al. confirms this.

Conclusion: good investing requires a complicated skill set, that is unfortunately less readily availabe than we would like. It is not very helpful that the average marketing or business development executive of big financial institutions tries to make us believe that they are big because they are the best in everything they provide. The same holds for investor relations personnel (or the CFO or even CEO) of stock-market-listed firms trying to convince investors to buy their shares and hold them longer. The difference between ‘the good’, ‘the bad’ and ‘the ugly’ is complicated when it comes to finding the best investment portfolio. Avoiding ‘bads’ and ‘uglies’ alltogether will lead to underpermance. Buying the bads is not an option either. The best strategy should lead to a diversified mix of ‘goods’ (‘quality’) and ‘uglies’ (contrarian strategy components), but as a decision taker you need a strong stomache to defend the latter category, especially in harsh times.

POSTSCRIPTUM

Application example 1; Emerging Markets 2008

The fact that Emerging Markets – even the ones with strong fundamentals – lost more in the credit crisis than struggling but bigger (‘good reasons’) developed markets gives you an idea about what we mean here.

We at Lodewijk Meijer continue to believe that only a robust, cost-benefit-based approach in which all pluses and minuses are made transparant to the final decision taker (‘principal’) can lead to structural outperformance. That is why we combine ‘best-of-breed’ specialist products of a mix of large firms and boutiques in a framework in which we only play the top-down, objective outsider role based on a framework that we developed together with Noble Prize laureate dr Markowitz. Humble when necessary, bold when based on something and boring when it comes to risk: that is the right way to go for investors in this new era. Unfortunately too many decision takers didn’t come to that conclusion yet.

Application example 2; Will our leader be capable of getting us out of the current crisis?

And that does also seem to hold for the ones that are trying to lead us out of this credit crisis. Initially the G7 leaders and ministers of Finance tried to do it themselves. With the smaller developed nations playing copy-cat in their own country. But hey guys, if you let liquidity first flow towards the Chinese and Arabs through the acquisition of products, oil or direct investments, don’t be surprised that – when suffering a problem – liquidity is somewhere else. The receivers of liquidity in the aforementioned nations did not automatically send it back to Wall Street or London. Sure, sooner or later it will come, but at their convenience. Only now, months down the road do we see markets slightly improve and leaders change talk from G7 to G20. Finally! The initial money spent was printed money of future tax payers. If we want it or not, cost-benefit-based thinking would imply direct talks with the Chinese, the Arabs and other sovereign wealth investors. Our leaders have to be bold and humble at the same time when doing so. Difficult, but possible.