Download Avoiding Disappointment in Investment Manager Selection and more Study Guides, Projects, Research Statistics in PDF only on Docsity!
Avoiding Disappointment in Investment Manager
Selection
Roger Urwin
United Kingdom
Pension funds and other institutional funds have increasingly turned to consultants to pmvide llsearch and advice on the best investment managers to appoint to ma& various mandates. A s a result. the selection of investment managers is a datively common activity carried out by actuarial consulting firms. ? h e w o r d ' b e s t ' i n t h i s c o n ~ i s ~ m e a s u r e d b y t h e p a f m d t h e investmentmanager. In particular, @performance relative to benchmarkor actiue retumis the main determinant of success. hemash&, allowanceis made for riskbymea~uringa~retumperunitof~em,wfiichisoStenrefened toasqtmmthmth
managers have been disappohting The selectionof~~liinager~is an activity which kquently produces a poor outcome: that is, the undaperformance ofthe
'Ihis paper gives two linked qlanations for this pmblan:
the hi@ noise to signal ratio in active ~ l l ~ ~ n a g e r d x r n swhich causes a structural bias towards disappointmentin manager selection
the dkmmmies of s a l e of active management which produce structuml impedimentstoCOIlSiStentou~ormance. 'Ihe appropriate methods of dealingwith this problem are twofold:
maldng realistic allowances for these ditlicultiesin the pafnmame monitoring P- ~ s o m e p a r t o f t h e f u n d ~ i n t o o t h e r i n v e s ~ t s t y l e s - particularly speoalist niche products with smaller 5xms and also index tmcking mandates.
8 Numemus studies have demonstratedthat the pehormances of managers over most paiods oftime am very hamsistent Such studies have been used by
miry1 particubrfy acadarniis, to fllggest that them are no m t m e n t managaS
who have abnamal skis Abnormal skill for a manager is gemdly d M to be
the ability to generate consisten!ly positive active returns, eitherbefm or after
dowing for style. An altematiw definition would be a managerwhose BqKcted activeretumwassignifi~tlypositive.
7 Thisisaparadccc 1fabnormallysldlfidmanagasdoexist.whyisthedataso reluctanttodemomtmkit? IfabN)ITnanyskilfulmanagasdonotBdst,howisit possible for so many pdticmms to believe that they do? ?he resolutionofthis paradoxis thatwNeabnamalskiUamongmanagasdoessdstitisnotvay pasistentiJIthe@tamorvaynaeaaonbleh~short-tam ?hisisexplained furtherbelow.
The Ed&ence of Abaolmal Skill
8 ' I h e ~ a c t i v e r e t u m o f m a n a g a s , a s d ~ a b w e , ~ t b e p o s i t i v e acnass the whole investment management commnmity when deahg with investment markets m which institutional players are dominant In a closed gmupofmanagas(0rcldsystem)whmthebmdmarkistheav~rage~ ofthegmup,theeqectedactive~shouldsumtoaero. Wherethe benchmark is an appropriately m h c t e d market index the expeded active return will lie a little behind the index to d e c t bnsactiion costs. However,such equivalencies wiU not m m s d y apply over shorter periods.
9 In such a closed system there must be managas with below average skills. Such rnanagaswouldhavea~tivearpectedactiveretmnInqualitativetams,the investment managanent lhnwhich has a annpetitive advantage needs a counter-party to hamad with to implement his advantage. ?hat counter-partyis a manager with a competitive disadvantage and lower sldll. In the highly
competitive W of investment maqpnmt the investment h with a
competitivedisadvantage may look at least supa&hUy, to be a wry campetwt
but because they have pocnw than avaage investmentpeople and
p-thqrwilltendinthe~-runtoundapafma&ainsttheavaaged thegmup. ItistheBdstenreoflowerskilledinves~thatallowfirmswith abnormalskillto&t
10 ?he other e v i d m for the existence of abnormally skilled investment fhms comesfiumdinxtlyobsavingskilL I n m y ~ i n t o t h e q u a i i t i e s o f
investmentm-@mex&firms,I see clearand merit evidence of-
pxududng exdent paformance fium sldKul decisions and prcmsscs. However, gventhewmpetitivenahmoffundmanaganenfthe~encesoftransadion
Probleme with Noise in Performance Data
The dispersion of results in any sample of manager performance tends to exhibit normal or lognormal characteristtcs. The extent of the dispersion is directly related to the amount of active risk taken by managers in the sample. There is no obvious link between the dispersion and the degree of skill exhibited by managers in the sample.
A typical pattern of dispersion is a s shown in the distribution chart below. In this example, the active risk of the sample is 2 O h per annum and gives rise to an upper quartile result at 1.4%per annum above the benchmark.
Fitted Distribution of Actual Active Returns
I of Manager Universe I
The figures are for a one year period. The crucial point is that the results in this distribution over just one year reflect the influence of chance or luck (this is noise in the statistical sense) more than the influence of investment skill (this is the signal in the statistical sense). It appears that the noise of one year's performance data is of the order of five times the magnitude of the underlying skill in that performance (ie a noise to signal ratio of 5) in most of the mandates I have analysed. This suggests that the dispersion of skill (as opposed to performance) is much more tightly grouped than the performance data suggests. A distribution of skill a s given in the chart below is realistic given this analysis.
,- Fitted Distribution of Expected Active Returns of Manager Universe
The above position is simply a result of regression in observed data. Investment statistics are subject to the same effects a s many other examples in business, sport and nature. However, in investment the effects are more profound because of the high noise to signal ratio. It is reasonable to ask the question why is the investment system subject to this high noise problem when most other systems we come across in our everyday lives are not? The most plausible explanation lies in the means by which investment market prices are set. Market prices at any point of time represent an equilibrium point between marginal buyers and sellers. Such prices of course directly influence all investment manager performance. This causes a quite arbitrary 'out of process' evaluation of most managers' investment processes and portfolios. One way of looking at this is to say that the market places an 'estimate' on the quality of the investment decisions of managers at any point in time. For many managers such a n estimate is quite arbitrary and prone to 'error' as it reflects the marginal investment activity of that point in time not the coUectiue view.
Two major issues arise in dealing with this skill and performance issue. First, expectations for active performance have been clearly set too high. In this respect it is clear that the performance target in paragraph 12 above is set at too high a level to be realistically achievable in a majority of circumstances. Secondly, we cannot expect that the signal in investment skill will be stationary over time and this causes additional difficulties. This point is addressed below.
and generated excellent performance with a certain value of funds will tend to struggle to construct the same quality of portfolios when the value of funds has grown. Fest. transaction costs will increase as lager transactions attract higher spread and market impact costs. Analyses of the economics of brokers show clearly that large transactions tend to be among their most profitable. Secondly, some transactions which would have been implemented under the smaller asset base, will not be tradable at all once assets have increased. Transaction cost measurement firms refer to this as an implementation shortfall.
The problems of maintaining competitive advantage in perlormance as assets grow can also be seen qualitatively. For a manager to pmduce a good active return he must implement his skill advantage by transactmg with a lower skill counterpart. In simplistic terms, a doubling of the asset base. all else being equal, doubles the number of lower skill counterparts he must deal with. In a competitive marketplace for investment managers. this may be very difiicult
This raises an important distinction between skill and information ratio. We generally use the term skill as an assessment of the quality of the people and processes at work Skill may be maintained as assets grow, but
the information ratio can nevertheless be reduced because of the problems
of illiquidity and of finding suitable counterparts in a reasonably &cient market. In fact, it is probably more realistic to think of skill being capable of sustaining a certain level of added wealth in money terms. In other words, the information ratio will reflect the value of assets under management and all else being equal will have an inverse relationship to it.
Drawing together these points in summary:
there are diseconomies of scale in active management which result from client growth and the problems of rnaintainmg&sponsive flexible investment teams as numbers of portfolios these factors tend to affect skill adversely
there are other diseconomies of scale in active management which result 6rom asset growth and the problems of liquidity; these factors tend to dilute the highly skill manager's performance advantage
taken together these factors suggest that the information ratio of managers will often follow a mean regressive stochastic process.
Summary of the Issues
There are two major problems - both instances of regression - in
considering skill in investment management through the measurement of performance:
short-term results exhibit too much influence from chance to be that helpful in the assessment of skill
long-term results reflect the influences of change that have taken place in fund management h s which will generally have altered the levels of skill at work.
These issues of 'double regression' present particular difficulties in the selection of investment managers. Clearly processes that rely too much on past performance are flawed. We have inferred from our analysis above that exceptionally good past performance is associated with lower future performance. This suggests that the only reliable method of selecting managers should be centred on direct research into the people and processes used by fund management firms (so called 'soft research).
However, even the use of the highest quality research in manager selection processes runs the risk of disappointment. This is because those working with fund managers still do not reliably take regression into account in making judgements about the future and the past. People seem not to expect regression in this situation, where it is bound to occur and tend to invent spurious causal explanations for regression when it is observed (you might refer to these as performance myths1 :
In the face of these problems. how should we act? There are two responses:
being realistic in our targets and expectations
being diversifled in manager choice.
Realism: The Sporting Metaphor
Realism in recognising these lower levels of fund manager skill at work is dimcult. This seems to require a very deliberate re-tuning of our mindsets to take on board some of the peculiarities of the investment world. Comparisons with other competitive systems are helpful in that they bring out the differences that apply in investment. In this regard, sporting metaphors are particularly instructive.
Realism in Performance Monitoring 42 The earlier results in this Paper suggest that a realistic outcome for a highly skilled manager over a one year period should fall in the following sort of distribution (again assuming active risk of 2% per annum).
Fitted Distribution of Expected Active Returns
of a High Skill Manager
43 Performance targets should be set at a level to be consistent with these expected outcomes. Increasingly, I favour targets being set at the mean expected level (0.5%above the benchmark in this example). Targets set at the customary level of 1°h above the benchmark can only be justifled as 'aspirations' or 'stretch' figures.
44 Furthermore the above distribution may be used more extensively in the monitoring process as a control device to test the continuing skill of the investment manager. The question to be considered is have the results achieved so far been consistent with this distribution pattern. Naturally, the distribution can be projected over longer periods of time than one year allowing for a lower annualised dispersion.
Diversification Opportunities
45 The realistic levels of outperformance typical in normal mandates a s illustrated above leave substantial scope for disappointment with active management. In the face of this dmculty, there is a strong case for diversifying into other basic investment management styles. of which these two are particularly attractive:
active investment in narrowly focused situations where future growth of funds does not represent a problem
passive investment using low cost index tracking methods. It appears that the size of assets under management is a material constraint to achieving worthwhile information ratios. Investing in niche products managed by smaller firms provides a realistic opportunity to increase the overall performance expectation of a fund. To make such a strategy appropriate for the governance of a fund. it would be desirable to choose an array of several such opportunities which have independent performance characteristics.
It is also of value to a fund to include diversification in index tracking mandates. This does not improve the information ratio relative to the agreed benchmark. However, it does diversify the risk that active management underperforms the market index which does happen over certain periods (but is not plausible long term). Furthermore, the inclusion of passive management allows direct control over the levels of active risk taken in a fund. This represents a crucial goal in effective investment management arrangements.
Conclusion
The influences of chance and organisational change on performance have been insufficiently recognised in investment management. As a result, funds are exposed to the risk of disappointment in manager selection on quite a frequent basis. This is not because the decisions themselves are necessarily poor, but simply because the seemingly rational way people make decisions in this area leads them to expect too much.
Funds need to recognise realistic measures of skill and out- performance. In the face of the reality of lower probability of achieving worthwhile active performance there is a clear case to diversify into other styles of investment.
Funds should concentrate their attention on what they can realistically control - active risk, rather than what they can only hope to control - active return. As a closing remark I am drawn to the useful maxim 'in an uncertain world. the wise decision is the one you can live with however bad the outcome'.
