Chasing hypeAugust 08, 2017
Recent concentration in market breadth, extremes in factor bias and signals from our systematic portfolio’s automatic market risk stabilizer all signal an impending change in market trading behavior.
The last 4 months have seen an historically extreme form of investor behavior. While we would not normally tie a paper in this research series to a particular point in time, we believe recent market activity demands special attention. It forces market participants to determine whether “this time it is different” or a form of mean reversion in behavior is imminent. Our bias is towards the latter.
In the year to July 2017 the MSCI Asia ex Japan total return index (NDUECAXJ Index) is up 29.4%. From its low on 21 January last year it is up 56.6%. These are clearly highly attractive numbers for any long-biased investor in Asia through those periods. Indeed, the Firth Asian Systematic Equity Strategy has substantially benefited from this price appreciation.
Importantly, our systematic strategy derives its signals from analysis of stock fundamentals. It consumes large quantities of income statement, cashflow statement and balance sheet data, along with analyst earnings forecasts and pricing information. The core belief behind this approach is that stocks are ultimately priced on their fundamentals. Ultimately, the correct price of a stock is the discounted value of its future expected dividend stream to you, the investor. Of course determining that is easier said than done. Different expectations are what drives the market and trading activity. Nonetheless, while it often feels like markets are ignoring fundamentals, that is generally not the case.
However, there are two main exceptions to this.
- In a market crash; and,
- Near a market peak.
In both of these circumstances relative stock fundamentals typically play little if any role in determining price action. Stocks are driven by animal sentiment – fear and total capitulation in the case of the first scenario, and greed and ignorance in the case of the latter. On the plus side both scenarios tend to be relatively short-lived. There is only a limited window for sentiment to trump fundamentals before mis-pricing reaches such extremes that investor rationality returns to profit from the inefficiencies.
Given we manage a systematic process we consequently have a portfolio which is not subject to the manager’s personal views. The process is always focused on the fundamentals. An interesting consequence of this is the performance of the process acts as a bellwether for market rationality. In fact the automatic macroeconomic stabilization mechanism employed in the strategy provides explicit signals for future market risk, but more on that below. At its simplest, if our portfolio is outperforming we can be confident that the market is correcting pricing inefficiencies. If our portfolio is underperforming we can be confident that the market is ignoring pricing inefficiencies, and/or creating greater inefficiencies.
Now that may at first glance appear to be a rather arrogant comment. We would never suggest that we are always right and the market is always wrong. However, we are employing fundamentals that, over the medium to long term, have historically been the primary drivers of relative stock performance. In addition, individual month performance is an insignificant indicator of investor sentiment/bias (it is too noisy). But when a series of months exhibit strong bias, of any form, that is when it begins to be possible to infer signs of sentiment outweighing fundamentals.
Narrowing Market Breadth
Now we return to our opening statement regarding recent market activity. We noted the strong performance of the market over the last year or so, with particular attention to the last 7 months. Strong performance in and of itself is no indicator of mis-pricing. Indeed, one of the challenges with our systematic strategy over the last year+ has been that it was identifying too many opportunities. It was signaling strong market upside by the mere fact that it could see there were large numbers of cheap stocks with attractive growth fundamentals.
However, we have observed a significant change in the nature of that rally between the first calendar quarter of 2017 and the subsequent 4 months. Firstly, the dominant driver of market performance has been the technology sector. Performance within that sector has in turn been dominated by 4 stocks: Samsung Electronics, Tencent, Alibaba and Taiwan Semiconductor (TSMC). To put this in context, we estimate that roughly half of year-to-date benchmark index upside was generated by the technology sector (compared with a sector weighting closer to 25% a year ago – over 30% today). Within the technology sector, we estimate roughly 75% of performance was driven by the 4 mega caps. Within some local markets the effects have been even more dramatic. A recent paper from the CCB International strategy team notes that the share of the technology sector in the China MSCI index is now over 50%, from a low of 10% in 2008! This is the first time in the history of this index that one sector has broached the 50% level.
Asia is by no means the only region which has experienced such a phenomenon (note for example the FANG stocks in the US).However, such narrowing of market breadth is a concern. Concentrated performance, while not enough necessarily to signal a market peak, is historically correlated with such peaks.
Factor performance focusing on sentiment
A second point we would like to emphasise is that factor performance has become increasingly extreme over recent months. The best performing factors of the past few months have been those focused on historic return on equity (ROE) combined with low payout (sustainable growth = (1 – payout ratio) x ROE, i.e. an historic indicator of internally-funded growth capacity) and price momentum. This is in stark contrast with the first quarter of 2017, when historic growth performance was still important, but was tempered by significant attention paid to the price for that growth. This is not the case for April through July this year. The price paid for growth was irrelevant for stock selection, and in fact was a negative indicator (backing “cheap” stocks was a losing strategy). We have also seen signs of deteriorating quality in the drivers of growth expectations for a number of high profile stocks.
To demonstrate the magnitude of these changes we generated some simple backtest results. We employed a sample set of a little over 1000 Asia ex Japan stocks, requiring them all to have a market capitalization in excess of USD1b and a range of minimum data requirements. China A and B shares were excluded (China is represented by the H shares).
For the period from 31 December 2016 through to 31 July 2017 we constructed simulated portfolios based on factor rankings. For each factor evaluated we ranked stocks on that factor, split the ranking into 5 groups (quintiles, with the same number of stocks in each), formed market cap-weighted portfolios for each group and measured the performance of each portfolio over the following week. This process is repeated on a weekly basis, ultimately building up a time series of portfolio returns for, say, the lowest 20% of stocks ranked on PE multiples versus the highest 20% of stocks ranked on PE. For the sake of presentation simplicity we limit results here to the performance of sustainable growth, PE, PB and 3 month price momentum. Numbers cited refer to the cumulative returns for the quintile 1 portfolio return less the quintile 5 portfolio return (where quintile 1 is highest sustainable growth or highest price momentum or lowest PE or lowest PB, and quintile 5 is the opposite). Hence, returns are for synthetic long/short portfolios, highlighting factor performance differentials.
Figure 1. Performance of Quintile 1 minus Quintile 5 factor portfolios, first quarter 2017 vs second quarter 2017 + July 2017
Each week a sample of approximately 1000 Asia ex Japan stocks (market cap > USD1b) are ranked on a factor and split into 5 groupings based on that ranking. Stocks within each group are marketcap weighted and the performance of the portfolios are measured over the following week. Results presented are cumulative returns for the quintile 1 portfolio less the quintile 5 portfolio (where quintile 1 is high sustainable growth, high momentum, low PE and low PB). Cumulative performance is shown for the first quarter of 2017 (LHS) and the second quarter + July 2017 (RHS). Arrows indicate the increase or decrease in factor contribution.
In the first quarter of 2017 an investor buying the top 20% of sustainable growth stocks and shorting the bottom 20% would have obtained a return of 27.6% (of course this is not an actual implementable strategy – it is for illustration only to demonstrate relative factor returns). The comparable number for PE multiples was 16.3%. So this value factor generated significant positive performance (for the sake of this exercise we will omit details regarding actual statistical significance, but it is safe to say that this is a significant result). Ranking on PB multiplies only generated a return difference of 0.4%, and 3 month price momentum was a loss-making strategy (-6.6%).
Comparing this with Qtr 2 2017 + July 2017, the return differential between top and bottom ranked sustainable growth stocks was a mammoth 66.5%. Price momentum went from negative to a positive 32.5% return differential. Conversely, backing either PE or PB was a losing strategy, with return differentials between quintile 1 and quintile 5 portfolios of -19.1% and – 39.0%, respectively (results presented diagrammatically in Figure 1).
We believe there are a number of important signals from these results. Firstly, while quintile 1 this year was arguably growth-biased, there was evidently still attention paid to the price of that growth. In that sense, there was significant rationality in investor behavior. Secondly, in Qtr 2 it is evident that valuation was irrelevant, and in fact a negative. Performance was driven by historic ROE (combined with low payout) and price momentum. We would view such a market as driven more by hype than fundamentals.
Such extremes of growth/momentum focus, over an extended period of months, are not common. In our multiple decades of backtesting for the strategy underpinning our systematic fund we found such periods had not lasted beyond 3-4 months (the most extreme being the liquidity-fuelled final rally stage in 2007).
Importantly, we are not suggesting these results herald the end of the bull market. We do not believe such analysis can provide sufficiently robust signals regarding market timing. Nonetheless, we strongly believe these results are indicative of an adjustment in market focus over the next few months. Growth factors may still be important drivers of stock performance, but we anticipate a return to assessment of the cost of that growth. Hence, style bias becomes somewhat less extreme.
Shrinking Systematic Portfolio Beta
A final signal derived from our systematic approach is provided by our portfolio beta. Readers familiar with our process will be aware that we employ stock-level data for purposes:
- Select stocks; and,
- Manage macroeconomic risk.
We believe this process is unique, and comes from experience built up over many years to seek market-level signals from stock-level data (helped along by our published academic research on the identification of hidden macroeconomic signals in stock data).
This paper is not one in which to discuss the nature of that approach. Suffice to say the macroeconomic risk signals are made manifest in the beta of our implemented portfolio (purely by the long stock selection – there is no cash management nor shorting). The portfolio beta changes very slowly over a long (multi-year) period. It is not a market timing signal. It is purely for macro risk management (effectively dampening the amplitude of the cycle). Having said that, we have had investors looking at our portfolio beta as a measure of market risk.
What is interesting today is that the portfolio beta has been shrinking over recent months, from around 1.0 to around 0.9. By the standards of this strategy that is a significant move. Again, we cannot emphasize enough that the beta is difficult to use for market timing. Nonetheless, we can say that market risk is building and our strategy’s automatic stabilizer is adjusting the portfolio beta accordingly.
So, we have extreme sector-concentration of performance, extreme factor bias and signals from our live portfolio of building market risk. Does this herald a turning point in market performance? That is extremely difficult to determine. Does this presage a change in factor selection? We believe that is likely. Stocks have historically always been limited in their ability to trade beyond their fundamentals. Investor willingness to chase hype over reality can only last so long. The last 4 months have seen historic extremes in such activity. The next few months will be a telling test of investors’ ability to put sentiment before fundamentals, and a test of their intestinal fortitude.
Written by: Dr. Hamish Macalister