Even before Invesco Perpetual High Income fund manager, Neil Woodford, dropped the bombshell he was leaving to start his own shop, his fund’s performance was prompting a bit of fund-picker’s angst. By which I mean, those who pick active funds put themselves under constant pressure to reassess their decisions. A ‘star’ manager jumping ship is one of the worst ‘new’ decisions to have to wrestle with but it is far from the only one. This is the point about active-manager selection: the game is hard to win but it doesn’t end with the selection – that is only the beginning.
Take the problem of living with passages of poor or mediocre performance. If you bought a fund on the back of an excellent track record and it then proves disappointing, how do you work out whether this is just a passing phase for a skilled manager making big conviction bets that cannot possibly work out well all of the time; whether the skill is specific to a particular market environment that will not repeat; whether he or she was once a skilled manager but has lost his touch; or whether the past record was not actually the product of skill but rather of a few big bets that worked out by chance alone? Why, you might ask, does it have to be so difficult?
The Woodford record
In spite of Woodford’s exceptional reputation, the path of returns for the High Income Fund, when measured against the FTSE All Share Index, presented these difficulties at various points to anyone paying proper attention.
In the chart we show the whole history of the High Income Fund (the earliest series being the income units launched in 1988) compared with the FTSE All Share Index as a measure of the broad market average returns.
These are total returns (income is notionally reinvested), indexed. The data source is Lipper. They are in logarithmic scale to avoid the impression that achieved returns for both the fund and market have risen exponentially: all changes are proportional.
It is easy to see at a glance why Woodford is lauded by investors (whether they enjoyed these returns or joined in after part of the record was established). But this is not how the fund would have been assessed at each point in time, without the benefit of 25 years of hindsight.
Interpreting the data
The income team around Woodford is a ‘high-conviction’ investor, holding a portfolio that (even by income-fund standards) does not look a lot like the main index (the correlation between the fund’s monthly returns and those of the index has been around 70%).
This can make it trickier to address the first decision of the fund picker: to differentiate between luck and skill. However, this is a a problem that eases with more time and track record. In our active fund assessment application, the ‘true’ whole-history alpha (risk-adjusted relative return tested at the 95% confidence level) is around 1% per annum, albeit helped by an even lower beta (fund return variance relative to the market return variance) than might be expected to apply to income funds generally.
Without further analysis, that should persuade anyone that Woodford’s track record could not have arisen by luck alone. But fund pickers don’t typically play this as a numbers game: they want to dig deeper to understand the sources of alpha, perhaps to differentiate yet further between fairly passive style biases and genuinely active bets. For this fund, that digging is going to make the decision quite awkward.
There have in fact only been three phases in which statistically-significant positive alpha has been earned and these are receding deep into the background.
The first, popularly attributed to a call in deeply-unloved tobacco stocks, relates to a brief interval in 1993; but after then tracking the broad market it was entirely given back in 1999. This is readily apparent in the chart. Investors buying after the period of relative gain would have suffered only relative loss.
The typical version of the Woodford narrative is that he outperformed in two bear markets – a factor (were it true) that really ought to find a ready following, as investors love managers who can protect them in the bad times without simply giving it back in better markets. It is not exactly true. There are two more successful phases but they are not exactly in bear markets.
In the chart below we show the returns for the fund’s accumulation units since 1999 and for the All Share Index (the version excluding investment trusts). They are ‘total returns’ (being accumulation units, income is actually not notionally reinvested). To test the popular narrative using data since the fund has outperformed, we’ve separated this history into five distinct phases based on the direction of the fund’s absolute returns (cumulative, not annualised, within the period).
The first phase is from April 1999 to May 2002, during which the index was gradually losing momentum but was hardly a bear market. In this phase, as the story goes, the income team avoided technology stocks completely. Even after a dreadful start to this phase that eroded all the past outperformance, the absence of technology largely explains why the fund returned 26% compared with a total return of -13% for the index.
This was a sector difference effectively ‘imposed’ on UK income funds at that time (so hardly a ‘call’ or bet) and the performance impact did not mark this fund out from its income peers. The end of the phase is marked by the end of these positive fund returns.
This sector difference was nonetheless of no consequence in the next, ‘real’ phase of the bear market. This is the nature of most bear markets: there tends to be no hiding place within the equity market itself. In this next phase, up to the market low of March 2003, the fund lost 29% (in total return terms) and the index lost 28%. So the first significant alpha generation is now over a decade ago.
The bull phase after the market turned is in fact when the Woodford team next pulled ahead. Up to the point of the bull market peak (which coincided for both manager and index) in October 2007, the fund returned on a cumulative basis 187% against 130% for the index. After another shaky start, the relative gains were then spread fairly evenly through the period with strings of monthly outperformance punctuated by much briefer episodes of underperformance. This would normally be the sort of prima facie evidence investors look for of successful changes in strategy, necessary to string together good performance. However, in this period it was (unusually) not necessary to make changes. It was enough to maintain high exposure to utilities and consumer staples and keep avoiding low-yielding sectors – exactly as income-fund managers might be expected as a group to do.
The popular narrative has Woodford’s second call as being to avoid banks and financials in the credit crunch. Indeed he did. But whereas this was the exposure that differentiated between income-fund managers, it was not a source of much protection in the bear market when compared with the index. This is shown in the bar chart as the penultimate phase and is based on the dates of the peak and trough of the index. A total return of -31% versus -39% for the index is good but not much protection. It should not be the source of Woodford’s following.
He and his team were not able to repeat the success of the last bull market in the recent one, starting from the index low point in March 2009. The final phase, which is now nearly as long as the 2003-7 run, shows a cumulative return of 97% versus 104% for the index.
Doubting the reputation
Well before the bombshell, questions were being asked about whether Woodford had lost his touch, even though the recent performance had only been mediocre. Such is the problem of following a star whose reputation, though long-lived, was based on alpha generation between 6 and 14 years before (or even the original episode 20 years ago).
The fact that so much of the relative return history resulted from a small number of persistently-held bets also raised some doubts about whether he or the team had the skills necessary to perform well when ‘normal service’ was restored, in other words when sector relative performance again showed mean reversion and changes in bets were needed in order to keep earning positive alpha.
After all, the three sector bets we have identified, technology, then utilities/consumer staples and then banks, have some characteristics that ought to have made people question their sustainability. The point about the technology bust was that the stocks affected did not recover or were such a small sector of the index as to make not owning them of no consequence to performance in the bull phase: one decision was enough (or enough for a long period of time). Banks were also an unusual one-decision sector, as the scope for recovery in per-share book values, earnings and dividends was always going to be heavily diluted by forced recapitalisation.
The utility and consumer staples sectors, on the other hand, ought to fit the ‘normal’ sector performance rules: you can win big by a single bet in or out but to avoid giving it back you will need to reverse the position. Two decisions will be needed. In fact, a number of reasons have made utilities, in particular, look like a one-decision sector for an unusually long time. This may well turn out to have as simple an explanation as the long bull market in bonds – as far as possible from a one-decision bet as could be imagined. Many reputations have been based on this factor (think of Ruffer and Capital Gearing, for instance) but few have yet closed the bet out, if only because they are finding it very hard (impossible?) to know what to do for an encore.
What should fund holders do now?
Coming back to the trigger for this post, Neil Woodford’s decision to quit, it might be concluded from this analysis that it really makes less difference than is generally assumed. It was always going to be very hard to avoid giving returns back, whether it was him or the team deciding on the encore. Unfortunately, fund pickers are unlikely to be this detached.
If we take a more generous view that it was (until the bombshell) perfectly rational to bet on this fund doing what was necessary to avoid giving back much or any of its positive alpha, we do now need to form a view about the man versus the team. This is not quite as straightforward as it may seem. The key contribution of a star may be as a team leader, in which case the team of stock pickers functions better (such as by avoiding the mediocrity of a committee) with that strong leader in place. I do not know the man or the set-up so I have no idea; but as a fund picker I would probably feel I needed to.
This is not something the data can tell you but I am not convinced that analysis of soft factors, such as teamwork and firm culture, works either. Soft analysis has always ranked very high in the manager-assessment processes used by institutional investment consultants advising pension funds yet recent analysis by Cass Business School of 90% of the US consultant market revealed their selections underperformed by 1.1% pa between 1999 and 2011 (against a simple average, so unbiased by size, of all managers).
In the case of the High Income fund, there are two specific factors that fund pickers will probably focus on.
Woodford may be signalling that his encore lies beyond the remit of an income fund, and he does not want to have to eschew strategies because of the yield rules. If so, this would appear to set him apart from other firms who have ridden the bond coat tails. This possibility ought to unsettle yield-seeking investors generally.
His followers will know that he (and the rest of the Invesco Perpetual group) hold very large and potentially illiquid positions in a lot of small stocks and so investors redeeming units could trigger forced sales against themselves.
If they believe he is as skilled a manager as they could hope to find (as the numbers suggest) and think that either of these specific factors applies, they should follow the star and let him place his new bets elsewhere on their behalf.
The opt outs
The point about this article is that playing the active manager game generates a flow of new and tricky decisions to wrestle with. It is not necessary, provided you have exposure to the systematic returns of equity markets, as you can opt for passive or index-tracking vehicles. Being in with a chance of beating the market is a nice option to have but it is expensive in costs and hassle.
In this analysis, I have reduced the active-manager selection problem to a single manager. But this is only how it presents itself to individuals retaining a single firm with a single strategy as their discretionary manager, as many have chosen Ruffer, for instance. Investors in active funds who diversify active-manager risk do have the choice of opting out of the angst-creating analysis outlined in this article in favour of playing another version of a numbers game. By making a simplifying assumption that past alpha will, on average, persist, they can rely on statistical tests and quantitative rules to make and review selections; and on diversification of active funds within a market, not qualitative research and judgement, to control the errors.
The two opt outs are not mutually exclusive. The alpha option is more valuable when expected beta returns to markets are low; by contrast they offer little value when expected market returns are high enough to secure your objectives. So a sensible strategy is to combine them, solving jointly for the mix of active and passive, based on market conditions, and the selection of active candidates. Passive will tend to dominate and in some markets there may be no funds that pass the tests.