Style analysis and various forms of style-based investment management have, over the past five to 10 years, become very common in the US and many of the larger European markets. These markets have shown themselves to be characterisable in terms of the major style categories (size, measured by market capitalisation, and value/growth, measured mainly by book value or a variety of dividends or earnings yields). Style-based market analysis and portfolio analysis methods have been widely and usefully applied. Consequently, as foreign interest in Japanese equities intensifies, it is natural to consider how style practices might apply within Japan, as well.
It should be a simple question. After all, many of us remember how, during the 1980s and early 1990s, a number of quantitative model builders were able to construct very successful models of the Japanese market and demonstrate significant outperformance from a systematic focus on certain value factors and negative momentum. Certainly, we might conclude, Japan was a style-based market then and, so, despite the changes in the economic/market cycle, we should expect Japan to continue to be a style-based market now – just at a different phase of the business cycle. It’s an appealingly simple conclusion. But, unfortunately, it is far too simplistic and fundamentally wrong.
During the 1980s and early 1990s, Japan displayed most of the characteristics of a genuine style-based market. Focusing on the key value criteria it seemed abundantly clear. All the main value factors were performing broadly in unison, and were offering positive rewards. And virtually all of the key style existence criteria appeared to validate the significance of each factor as a relevant style:
q The ‘identity’ statistics (which range from 0 to 100) recognised the individualistic performance characteristics of the attribute-sorted portfolios, constructed and rebalanced according to each value factor.
q The high ‘attribution’ numbers (an average of T-statistics) reveal the strength of the significance of each value factor in explaining the returns of individual securities.
q ‘Sector adjustment’ (ie, constructing the attribute-sorted portfolios sector by sector to reduce the distorting effects of differential performance between economic sectors) shows that value themes are independent of sector themes.
Initially, it looks like style has been a long-standing feature of the market.
But, on closer examination, it is apparent that there is something curious going on. Japanese book values are notoriously misleading. Dividends, particularly during the late 1980s, were so low that they did not figure in investment decisions. Earnings and the method of consolidation (or not) of the earnings of non-resident subsidiaries were impossible to interpret. And cashflow and sales calculations were similarly particularly opaque. These considerations clearly raise doubts about the true nature of the Japanese historic value rewards.
The first serious contrary evidence comes from the performance of stocks with positive independent growth characteristics. Normally these stocks perform inversely with stocks with strong value scores. This is certainly so in the US, the UK and much of Europe; and it is also the basic reason why in these markets it is understandable to conclude that the value criteria can measure both value (positive tilts) and growth (negative tilts) stock characteristics. Ultimately, this occurs since premium-rated companies (with low book-to-price ratios and low dividend and earnings yields) are priced at these higher ratings because of their demonstrated, or anticipated, growth potential.
In Japan, however, this does not seem to have been the case. During the period of the apparent strength of value, premium-priced securities were underperforming, yet growth companies were basically flat.
This market dislocation indicates that, at the time, the market was not pricing or rewarding growth potential in the recognised ways. This undoubtedly arose since the internationally recognised valuation criteria did not have the usual significance in Japanese equity assessments and, further, the market itself offered little reward to companies offering genuine demonstrated growth or the potential for rapid growth in the future. It was understood that the market was not competitive and, as a consequence, there were better opportunities to pursue than to wait for the market to recognise and reward growth expectations in share values. Given the concerns mentioned above, this should not be surprising.
So where did the value rewards come from, and should we expect them to return in the current environment?
The next significant clue comes from the performance of the momentum factor.
The return graph shows that, during the phase of value outperformance, just as recent outperformers subsequently underperformed, recent underperformers subsequently out-performed. (This is because the graph plots the relative returns of exactly half the market.) And this is the beginning of the explanation of the historic value effect. This market performance characteristic was both a consequence of corporate behaviour and local market practice and a cause of the appearance of the strong value rewards that typified the 1980s and the early 1990s.
During this period, the active market intervention practices of Japanese security brokers and the share purchase programmes implemented by industrial groups in support of weaker group members caused a slight but significant distortion in share price performance. Shares that had under-performed over the recent past became less prone to further weakness than the rest of the market. It is clear how this resulted in the appearance of the negative momentum effect. And it only takes the immediate recognition that weakening shares also become scored as better (higher) value to appreciate how the value theme of the 1980s and early 1990s was really no more than a reflection of a market anomaly brought on by systematic non-competitive market practices.
The reliability of this ‘theme’ crumbled with the economic shakeout of the late 1990s. Weakening financial services companies (basically banks and brokers) and the weakening share purchase efforts among the major industrial group members brought about the practical end of ‘corporate socialism’ and, in the environment that followed, value weakened dramatically. Investors were no longer able to rely on the systematic support previously given to weakening share prices and, almost immediately, as weakened company shares were permitted to submit to freer market forces, the historic value theme broke-up.
As the negative momentum theme and the value reward both reversed during the late 1990s, the growth style also began to stir. Even though the positive value phase of the earlier period had left growth dormant, as soon as value began to weaken and competitive forces overcame systematic market distorting practices, growth erupted as an investment style, with very dramatic outperformance.
Japan is now displaying strong and very consistent style effects. The new economic and market realism has closed the era of style artificiality and brought with it a recognition of the importance of style factors in the performance characteristics if individual equity securities. The recent rebound in growth stocks and the collapse in value resembles the current style rewards patterns in the other major markets and is entirely consistent with the present phase of the Japanese business cycle. In fact it is even more consistent than in other major markets. Low inflation, low long-term interest rates and stable economic growth (without sharp acceleration) are currently providing a positive background for premium-priced securities world-wide. Fundamentally, this economic environment should favour genuine growth stocks over value stocks but, in a caricature of the more normal market cycle/style reward pattern, most markets are simply experiencing a surge in higher priced stocks – irrespective of their recognisable growth potential. In Japan, however, the current style reward patterns are noticeably more consistent. Value is in decline and companies with recognisable growth features and potential are clearly out-performing. This is quite reassuring.
It is, of course, extremely difficult to forecast the economic/market outlook. Consequently it would be very hazardous to offer a forecast for the future returns of the various significant styles that now operate within the market. But now, much more than before, it is clear that styles are significant features of the market’s performance structure and that the performance of these styles can be analysed and forecast on the basis of recognised criteria in a much more competitive, freer, market environment.
Robert Schwob is director of Style Research in London