Stock-specific risk dominates equity returns, finds Sudhir Nanda. But stockpickers should be aware that the contribution of systemic factor risks varies significantly across regions, sectors and time

There are a number of potential ways to add value to managed portfolios. Most investors, for example, are aware of the critical importance of asset allocation, which some studies suggest is by far the most powerful explanation of variations in portfolio returns. Returns on global equities can be partially attributed to their correlations with the global market, as well as with their country, sector and industry. Much research has been devoted to the relative importance of country and sector selection in different markets. In theory, increased global trade and market integration should reduce the potential risk and rewards from top-down country selection, while increasing the significance of global sector selection. The real world evidence for this trend, however, is hotly contested.

More important for investors is the fact that non-systemic factors - primarily stock specific effects - are still the most important drivers of equity returns in global markets. We estimate that these factors explained half or more of the return variability in most geographic regions and economic sectors over a period of more than 15 years ended 31 December 2009. This suggests that security selection remains the most important source of alpha for active equity managers.

However, managers must also pay attention to country and sector factors, as these may
create top down alpha opportunities as well as unintended portfolio risk concentrations.
Moreover, the relative importance of different factors - including the balance between systemic and non systemic effects - varies widely across geographic markets and global sectors, and over different time periods. Disruptions such as the technology bubble of the 1990s or the recent credit crisis may temporarily magnify the importance of certain factors while diminishing others. The recent financial shock produced a spike in global market correlations, reducing the observed significance of stock specific factors. However, this trend appears to be reversing as markets normalise.

The significance of stock-specific factors reflected in our data does not, by itself, guarantee success for bottom-up stock picking strategies: many non-systemic effects are inherently unpredictable, or might simply reflect random ‘noise' in the data, thus providing little or no useful information for security selection. Distinguishing between such short-term shocks, which can reverse quickly, and longer-term secular trends isn't always easy.

To shed additional light on these issues, T Rowe Price's quantitative equity research group conducted a study of return variability in a number of equity universes, including the global large and mid-cap market, the developed and emerging markets, and key regional and national markets such as Asia ex-Japan, Europe, the US and Japan. The study also examined relative factor importance in global economic sectors, using the 10 sector global industry classification standard developed by MSCI.

The purpose was twofold. In addition to examining the now familiar question of country versus sector importance, we also sought to identify those regions and sectors where bottom-up stock selection may have the greatest potential impact on active returns.
Our study used two separate methodologies to measure factor significance:

• Returns on the stocks in each equity universe were regressed in multiple stages against three covariates - the global market, their country market, and their global sector. The residual was assumed to represent stock-specific effects, although it also includes industry effects, fundamental factors such as style and size, as well as random noise. The R squared results from these regressions were averaged across the stocks in each universe on an equal weighted basis.

• Correlations of individual stocks with their global sector, local sector and country were
averaged for each universe, also on an equal weighted basis. The results served as a cross check on the regression analysis and yielded similar findings.

The study covered a period beginning 30 September 1994 and ending 31 December 2009. To smooth short-term volatility, R squared and correlation were both measured over 36-month rolling periods. All returns were in US dollars, which means that currency effects were subsumed in the country factor.

Because sector and country factors cannot be fully disentangled from each other, the sum of the regression R squareds did not equal 100%. However, country and sector R squared were proportionally scaled to make all components sum to 100%. For this reason, the return decompositions reported here should only be used to evaluate the relative, not absolute, importance of each factor.

The most consistent finding in our study was that stock-specific effects (the residual term from the regression) have accounted for the largest single component of return variability in most sectors and geographic markets, implying that individual stock selection remains the most plentiful source of potential alpha for active global managers. This finding was true over nearly all time periods - with the marked exception of the past two years, which have seen a dramatic spike in global market correlations and a corresponding compression in the importance of stock specific factors.

Figure 1 shows the decomposition of returns in a global large and mid cap equity universe of almost 3,200 global stocks, similar in regional concentration to the MSCI All Country World index. The bands in the chart measure the percentage of return that can be explained by each factor. As can be seen, the impact of the credit crisis that began in 2007 dwarfs a similar global correlation spike during the 1998 Asian financial crisis. However, we expect that the effects of the recent crisis, like that earlier episode, will reverse over time.

Factor trends across markets
Leaving aside systemic shocks such as the credit crisis, there still are persistent differences in factor importance across both markets and sectors. These variations may have important implications for active portfolio strategies.

In the emerging markets (EMs), for example, country remains a more powerful return component than global sector - reflecting continued market segmentation due to capital restrictions, perceptions of political risk or other factors. This suggests that managers might need to continually assess the top-down elements of their EM investment process as markets continue to evolve, while recognising that non-systematic factors - primarily stock selection effects - have been and are likely to remain the most important drivers of returns. Furthermore, globalisation does appear to be changing these markets. The relative importance of country declined over the study period, while correlations with global sectors increased, and now rival regional sector correlations. This suggests that some managers might need to rethink their top-down EM strategies as markets continue to evolve.

In the developed markets (DMs), country, global sector and regional sector appear to have equal importance, suggesting that portfolio managers need to pay close attention to all three factors.

Some other geographic trends identified in our study:

• Opportunities for bottom-up stock selection have been relatively more important in the developed than the emerging markets and, within developed markets, more important in Japan and the US than in Europe.

• Even before the credit crisis, the global market had become more important as a driver of European returns. Local and global sector correlations also climbed sharply relative to country correlations over the course of the study period. These trends appear to reflect the introduction of the euro and the further economic integration of the EU.

• Stock-specific factors have become steadily more important in Japan and now explain almost 60% of return variability, down from a high of 71% prior to the credit crisis. The global market and global sectors both have relatively less importance, reflecting the decoupling of Japan and other world markets over the past decade.

• The importance of stock specific effects has also increased in Asia ex Japan, although the credit crisis appears to have sharply reversed that trend, at least temporarily. Regional and global sector correlations have converged across the region, in line with its growing integration into the world economy.

Factor trends across sectors
Our work also found considerable variation in factor importance across global economic sectors (figure 6), which could also have significant implications for stock-selection strategies, although sector data need to be interpreted with particular caution. Among the 10 major global sectors, energy, financials and materials showed the lowest stock specific component to returns, while healthcare and consumer staples had the highest. Not coincidently, the latter two sectors, as well as utilities and telecommunications, showed the least sensitivity to the global market. This appears to reflect the traditional defensive role of these sectors, which has produced relatively low market correlations - reducing, in our methodology, the significance of the global factor. Energy stocks, meanwhile, showed high sector correlations, and thus relatively high global sector R squareds - most likely due to common dependence on oil prices. One implication of these results is that investors in the energy sector should have a global view and would benefit from selecting stocks globally.

These complex dynamics should be kept in mind when assessing the relative alpha potential of global sectors. For example, the relatively high stock-specific content of returns in the healthcare and consumer staples sectors could be attributable to the importance of characteristics such as product innovation or brand strength - precisely the factors that fundamental analysis is geared toward identifying. At first glance, this might seem to suggest that bottom-up research resources could more profitably be deployed in those two sectors compared with others where non-systemic effects appear to be relatively less important. However, this may not be the case. Alpha generation potential can be influenced by a number of other characteristics, such as the dispersion of returns within a sector or market. These dynamics may outweigh the relative importance of stock-specific factors. A high dispersion of intra sector returns, for instance, affects the ‘amount' of available alpha, thus increasing the impact of stock selection.

In our study, for example, a number of defensive sectors, including consumer staples, showed high stock specific content but relatively low dispersion of returns. The energy sector, by contrast, had lower stock specific content but much higher dispersion of returns - implying a better environment for bottom up stock picking, at least over some periods.

In reality, an effective research approach is likely to combine both bottom-up and top-down analytical methods. In some sectors or markets, focusing primarily on stock specific factors might prove to be the most rewarding strategy, while in other cases it may be valuable to develop views on both stock specific and macro issues. As with most aspects of portfolio management, assessing the relative significance of various factors in the global opportunity set is a complex exercise. While our analysis highlights the importance of fundamental research and stock selection as sources of value, it also suggests that exploiting these opportunities requires considerable research expertise and the ability to apply that expertise on a global basis across both countries and sectors.

Sudhir Nanda is vice-president and US small cap equity strategy portfolio manager at T Rowe Price