Hidden sources of alpha

A look at the big trends among Australia’s superannuation and US pension funds markets through the eyes of Professor David Gallagher, ‘accidental academic” and investment management scholar

Sanlam Investments recently hosted a dinner engagement in Johannesburg with a widely published expert of international standing in investment management and capital markets, Professor David R Gallagher.  This followed his Cape Town talk on in-house investment management, where he likened trends in Australia’s massive $AUD2 trillion dollar superannuation industry to that of South Africa’s pension fund equivalent.

In his Johannesburg talk, Gallagher shared additional insights on how fiduciaries and allocators of capital could address real-world problems, such as how institutional investors could generate outperformance and better meet their risk/return objectives.

Research done on quantitative strategies

Gallagher and his research colleagues conducted substantive research that worked well in both international and domestic markets, by scrutinising fund manager performance using granular data. So rather than just looking at total returns, they also looked at daily trading data and monthly periodic holdings data (i.e. more granular than the US market’s quarterly holdings snapshots required under SEC regulation).

What are the benefits of this?

Says Gallagher, its much like financial management, where you have three major financial statements: P&L, cash flow, and balance sheet. They each tell you a consistent story in aggregate, but not each tells you the whole story on its own. We extrapolated this analogy to the funds management industry where the total return equals the P&L (fund performance above or below the benchmark), the balance sheet is the equivalent of the portfolio holdings at a specific point in time, and the trading data is the equivalent of the cash flow statement. This shows how trades were done to generate portfolio holdings and, in turn, the total returns delivered.

“Using this finer, granular level of data we were able to measure and understand the true skill of an external manager,” says Gallagher. We also looked at additional aspects such as trade skill. If information is short-lived and there is a short window to be able to exploit that data, you cannot see whether the fund manager performed better in terms of total return; you have to look at the number of positions they held and the amount of trading time needed to generate alpha for their clients.

1. Ability to generate alpha in global equity fund managers

Gallagher’s research showed that it was possible for global equity managers to generate an impressive outperformance (alpha) of about 1.2% pa on average. Arguably, anyone grounded in efficient markets theory would declare this improbable, as efficient markets postulate that the average active investor can only generate the equivalent of the average index return, and after fees performance would be negative.

Conversely, however, Gallagher’s research showed that global active equity managers who were actively managing portfolios over 10 to 20 year horizons were able to generate significant risk-adjusted performance persistently over that 10 to 20 year period.

So was that real skill? We like to think so, says Gallagher. In coming to this conclusion his research analysed and took into account all the relevant risk dynamics (momentum, book to market equity ratios, size, etc) and still found that there was persistent alpha generated.

Fund managers buy winners, but they tend to sell too prematurely … 

Then Gallagher scrutinised managers’ buys versus their sells. One could arguably expect, says Gallagher, that with long-only managers, purchases would be significantly positive contributors to performance. On the sales side, however, while still profitable, managers typically sold too soon. The research suggests that if they had delayed their sell and held onto the stock over a slightly longer window of say 60 – 90 days, the outperformance generated for that time window would have been greater than the actual long positions they held over the same period (taking liquidity into account).

This suggests the behavioural bias ‘the disposition effect’ coming into play, where investors generally are too quick to sell their winners (and similarly hold onto their losers too long) because selling would be tantamount to admitting a (behavioural) mistake (i.e. regret).

Is true outperformance possible?

“There is absolutely a skills set out there that enables people to generate outpeformance”, says Gallagher. “Looking at this from a risk-adjusted basis (ie adjusted for size, book to market equity ratio, momentum and all well-known risk factors), we believe there is indeed a skills set out that that is known to generate consistent alpha”.

2. Competition and stock returns

According to the ‘creating destruction’ theory, companies that are dominant in their sector typically underinvest in R&D, inadvertently allowing newer, nimble and more dynamic entrepreneurial competitors to come up from behind and take over. US evidence shows that companies who dominate their industry actually have poorer long-run performance, compared to newer entrants that take over their dominant position. Given that the US is such a large equity market with strong anti-trust rules and a scale unmatched by other markets, they find this something of a contradiction, said Gallagher.

How does the US compare to Australia?

Gallagher and his team found that the opposite trend held true for Australian companies with overweight positions who dominated their industry. Conclusive evidence suggests that in the Australian marketplace – while applying all the risk-adjustment research tools – companies that dominate their industry are able to dominate, generating long-term returns that are significantly better on a risk-adjusted basis compared to newer, smaller entrants who do not have same degree of market share and scale.

“We’re talking extremely large alphas –in excess of 8% per year”, says Gallagher, “on a risk-adjusted basis. And even when applying the caveats of leakage, trading costs etc, we discovered there was significant outperformance; and alphas were persistent over an extended time window”.

As an aside, Gallagher noted that the SA industry is probably more similar in structure to Australia than the US market.

3. Cross-market and cross-sector style rotation strategies

Using international data, Gallagher used a regime switching model to look at asset allocation and how international investors could harvest a better risk-return trade-off, leveraging Markowitz’s 1990 Nobel-winning mean/variance framework.

Bull vs bear

According to this regime switching model, the classic mean/variance relationship could be significantly improved by truncating markets into bear and bull markets, although optimising portfolios by segmenting had quite unique challenges. Bear markets typically have higher sector correlations than bull markets and greater volatility. The theory would suggest that it would be possible to split these two different regimes with their relative volatilities and correlations. Naturally there would be different implications for the variance and co-variance matrix. In doing so, Gallagher found that with the regime switching model it was possible to significantly improve the risk return trade-offs (and improve the mean variance optimisation framework), compared to the conventional method of not splitting.

Style factors for US equity markets

In this particular research, factors were broken up into market factors (momentum, small vs mid-cap vs large cap, book to market equity ratio) and stock characteristics (quality of firms, balance sheet, P&Ls). Gallagher’s team tried to develop a forecasting model using econometric prototypes to try to forecast one quarter out for a holding period of 12-months. Could these factors ex ante be written in a successful way? Yes, says Gallagher. They found strong evidence that using aspects such as quality, net margin, cashflow-to-price and quality aspects of a balance sheet for a company, and by ranking using this econometric model, it was possible to build more portfolios that could significantly outperform US risk-adjusted markets.

Gallagher then went further to attempt to incubate a unique ability to beat the market significantly by applying these quality and style factors. His aim was to see if asset consultants could use these factors as an accurate screen to pick which fund managers would be the better, more consistent outperformers in the ensuring 12 months.

This yielded interesting results. The research revealed that there was an asymmetric contribution to alpha; that is, short positions had a disproportionate contribution to the success of the strategy compared to long positions. But in the US, fund managers were typically long only. So using style factors as a reliable predictor of top-performing fund managers failed to work in the US because these fund managers were long only and couldn’t capture short positions.

Gallagher concluded by saying that extensive research on active managers and asset consultants had been conducted in the past, but what differentiated his work was that it was informed by better systems and more granular information, used to thoroughly understand institutional investor behaviour, performance and risk.

The real debate in the industry, finished Gallagher, is how to calculate the fair split between the value that is created for clients versus the value that is kept by the fund manager. And, arguably, the whole process of scrutinising performance is about establishing this fair and equitable split.