Do 'good' Asian companies give better returns?

Conceptually, there is little reason why having the structures and processes of good CG should lead to strong stock performance. Except that not having them can clearly be very detrimental to a stock.

Searching for the impact of CG on stock performance is however devilishly tricky. Research by CLSA quants team, headed by Chris Lobello, separates out the effect of corporate governance from other factors. Over five years, a 10-percentage-point improvement in a company’s CG score was associated with 7.3% of additional performance for its shares over nine months following an update.

Our own updated tests of stocks sorted by CG going back one, three, five- and ten years indicate that on stocks of companies with better governance scores perform better than lower ones. Breaking this down by CG quartiles, the evidence suggests that in particular it is poor CG stocks that drag down performance. This is hardly surprising given that poor CG companies are more likely to see economic value being hijacked rather than shared equitably with investors. We also find evidence over the last three years of improving CG correlated with stock out-performance, whilst falling

We examined whether performance of a basket selected by attractive price-earnings ratio—a standard valuation measure - can be enhanced by considering CG criteria. A high CG screen imposed on this value approach appears to detract from the performance; however removing the lowest CG stocks from the low PE basket enhances the already strong performance of this value screen over the last ten years. A conclusion similar to that from the CLSA quants team is that CG is not yet being efficiently incorporated in stock prices. This provides investors with the opportunity, by avoiding the worst CG companies, which presumably reduces the risk in one’s portfolio, of nevertheless also getting higher returns.

This finding seems to fly against the axiom of finance that returns cannot be boosted without higher risk, except that the market may not yet have fixed on the means to measure and thus price in risk associated with poor CG.

For investors, the continued relevance of our CG rating of companies and markets is that beyond academic pieces, this remains systematically an under-researched area with important implications for returns of portfolios.

Importantly, it was not just a change in CG that was associated with this strong performance, but the level of CG itself. This indicates that better CG has an impact of stocks that is persistent. As Lobello writes in Nice guys finish ahead, which CLSA published in June last year, the implication is that the market does not quickly digest publicly available information on CG, and that has an effect even over a few quarters.

Lobello accepts that the impact of CG on stock performance could be related to other factors, and some might be associated corporate governance issues. One possibility is the growth of socially responsible investment-related funds.

From 2005 to 2007, SRI assets have increased by more than 18% while the broader universe of all funds under professional management only increased by less than 3%. For these funds, CG is likely to be one of the factors driving the investment decision. It is likely that factors such as these might at least partly explain the correlation between CG and stock performance, however there is little reason to expect these factors to suddenly disappear. Hence in all probability, the correlation between CG and stock performance is likely to stay.

Still, the result reminds us that CG in itself is is not the main explanation of stock performance; other factors like size of the stock as well as market and sector direction remain important.

Strikingly, four of the five years where the higher CG stocks underperformed (2003-2005 and 2007) were periods when Asian equity markets were fairly strong. Indeed, in 2003 the regional index had the strongest performance over the last ten years, up 37%, while the higher CG stocks did relatively the worst compared to the lower CG stocks.

Meanwhile, in three of the four years where the index fell by 10% or more, higher CG stocks outperformed, i.e. in 2000, 2001 and 2008. The exception was 2002, when the index declined but higher CG stocks underperformed; however this was after two consecutive years when higher CG stocks had outperformed. Thus probably by 2002, the selling for investors were in the remnants which were likely to be higher CG stocks that had done better in the earlier two years and were still in their holdings.

Although annual performance from a ten-year period is hardly conclusive statistically, it indicates that higher CG stocks tend to perform better when markets are declining, whilst poorer CG stocks perform better when markets are strong. That the performance of high CG stocks is inversely related with the direction of markets was made in our earlier CG reports. Risk appetite appears to be the underlying reason. Rising markets increase the risk appetite of investors and leads to poorer CG stocks performing better, whilst declining risk appetite in falling markets generally works in favour of the performance of higher CG stocks.

Despite the counter-cyclical performance of high CG stocks, over the long-term the evidence is that these stocks tend to outperform. In the region, over the last ten years, the higher CG stocks have provided a median return of 112.5% versus 96.3% for the lower half of CG stocks that we cover. However this performance need not come from the very top CG companies but may well be from those that are slightly better than average on CG.

For our analysis of relative performance, we use median returns of stocks sorted by CG quartiles or halves, rather than the mean performance. One or two stocks that might have gone up many times over the five-to-ten year period we are looking at can significantly distort the mean. The median gives the typical performance of a stock in that CG group.

We compare the returns of these stocks against the overall basket of stocks in our study rather than against an index for the region, as the index is skewed towards performance of larger caps. This often gives the paradoxical result that the simple average performance of each quartile comes out higher than the index. Each basket, not weighted by market cap, is likely to have performance pushed up by mid- and smaller caps. Each of the CG quartiles or halves could thus outperform a market-cap weighted index where the large caps are likely to have dampened the returns.

Our tests are to determine if on average higher CG stocks outperform, and lower CG stocks underperform. To represent the typical performance of these stocks, we use a simple median performance of the stocks in CG quartiles or halves, rather than market cap weighting the returns which would skew the findings towards the larger caps.

Sorting stocks by CG and examining their relative performance, we find the bottom ranked stocks for CG underperformed in 2009, as well as the last three and ten years. The median 10-year return of stocks where we have a CG rating was 104% with the median performance of the bottom-quartile CG stocks at 91% over the period. For the 2009, the median of the bottom CG quartile across the region underperformed that of the overall sample by 11 percentage points.

However, the top ranked stocks on CG are not consistent outperformers. More often than not, the top CG quartile underperforms the overall basket. These stocks outperformed regionally over the last two years, however they have outperformed in only three of the last ten years. In this ten-year period, the median performance of our overall basket was up 104% while the top CG quartile stocks rose only by a median of 72%. High CG in itself does not seem to be generally associated with stock outperformance.

The bottom quartile of CG stocks regionally underperformed in five of the last ten years, which as we indicated above is likely to be a function of market conditions and risk appetite. However, their relatively greater underperformance in these years (in particular 2009), and more marginal outperformance in the other years, meant that overall the returns of these bottom CG quartile was poor over the period.

These findings support the view that high CG in itself may not be sufficient to generate stock outperformance. However, poor CG is often a key factor that drags the performance of a stock.

By markets, we find a good correlation between the lowest quartile CG stocks under-performing against the overall country basket for China, Hong Kong, Asean markets and Japan over the last ten years. This suggests that either investors are more discriminating about corporate governance in these markets, or that there is greater hijacking of economic value for the poor CG stocks in these markets, or both.

However for our Indian, South Korean and Taiwanese sample, the poorest CG quartile did not underperform the respective country’s basket of stocks over the last ten years.

The findings also highlight the importance of choice of sample. For India, our sample excludes the scandal-ridden computer company Satyam which two years ago admitted that accounts had been falsified—that we no longer cover.

Presuming that our sample in most markets is biased towards better governed companies, and often the worst CG stocks have been dropped from coverage, it is likely to be more difficult to get clear results of the relative performance of stocks sorted by CG. The reality is likely to be that the poorest CG stocks—which are not rated at all—will perform badly.

For South Korea, the bottom CG quartile in our sample has under-performed the country basket over the last one and three years, whilst for Taiwan and India they underperformed in the last year. More likely, when there are significant CG related issues that blow up in any of these markets, and if we include these stocks in our sample, there will be a clearer trend of underperformance from the poorer CG stocks.

Interestingly over 2009 when investors came back strongly into equity markets after the earlier two years of turmoil, there was a clear preference for better CG stocks.

The upper half of stocks by our CG ranking outperformed the lower half in most markets, although not in the Asean markets and Hong Kong. This reflects a general preference for better CG stocks when investors are returning to markets but are somewhat uncertain about the overall economic outlook.

For large caps, we find the lower half of stocks sorted by CG generally outperformed the upper half. Over the last ten years, there was 177 ppts of outperformace.

For Japan, of the top 50 large caps, the lower half for CG outperformed the upper half by a negligible 0.3-ppt margin. This indicates that for the larger caps, investors may be taking the overall CG levels as likely to be reasonably good, and there is no clear underperformance for the poorer CG group.

It could, however, also be a result of mid-cap stocks with large re-rating getting into the larger cap category that may not have as high CG as those already in the large cap group. The impact of re-rating could distort the result of CG ranking and stock performance. Interestingly, for the last year, the higher CG stocks of the large caps have outperformed for both Asia ex-Japan (by 15.7 ppts) and for Japan (by 13.3 ppts).

Improvements. There is reasonable evidence that stocks with rising CG outperform strongly while those with declining CG underperform.

In our universe, 24 stocks had changes in scores of more than 10 percentage points since our last CG rating across the region, where the analyst’s view is that this reflected real change of CG in the companies. Exactly half, or 12 of the companies had a positive change in CG score of 10-points or more with our analysts confirming this reflected real improvement in underlying CG of the company over the last three years. The median performance of these stocks over the three year period to end-2009 was 49.6% (with a mean of +111.7%), compared to the median return for the stocks in our overall universe which over the period produced a return of 2.2% (mean of 23.5%).

Certainly not all of the stocks with CG improvement outperformed. Three of these had an absolute decline. But nine of the 12 stocks with improved CG rose above the median return for stocks in our universe.

Among the top performing in the list are Philex Mining and Manila Electric, where the improvement in CG of 33 to 53 points respectively came with a change in control moving to First Pacific which resulted in massive returns. In the region, Chroma of Taiwan with a 14-point CG improvement, L&T and Bank of Baroda from India also more than doubled with an improvement in CG scores of 13 - 22 points.

Meanwhile the median return of the twelve stocks with notable deterioration in CG over the last three years was -3.4% (mean of +1.3%). This appears only somewhat underperforming our overall basket but note that the biggest disasters are for companies we no longer cover, e.g. Satyam, where the impact of their CG debacles on stock prices would have been much greater. Thus the actual impact on performance for stocks with CG lapses is likely to be much worse than the table below indicates.

Within our current universe, some of the stocks that had a decline in CG nevertheless did well, for instance Sime Darby, Yuanta, Li Ning and Everlight. In most of these cases, business conditions and the performace of the sectors would have contributed to the performance. Others with large declines in CG scores have suffered significant stock underperformance, in particular the Japanese banks where increased dilution and reduced ability to maximise value for shareholders has resulted in a sharp drop in their CG scores.

We back tested the entire universe of stocks under CLSA coverage for those with the cheapest trailing PE at the end of each year for the last 10 years. On these we then applied CG overlays. Our finding is that adding a high CG overlay on the lowest PE stocks does not give added performance. However a CG overlay that takes out the lowest CG stocks is found to give stronger performance.

Stocks with the lowest quartile, or cheapest, PE, rebalanced at the end of each year, gave a huge 43.1% compounded annual return over the last 10 years. This is massive outperformance when compared for instance with the 12.3% compounded return for stocks in the most expensive PE quartile. For this period, the MSCI Asia Pacific index dropped at a compounded rate of 1.6%. As mentioned above, the performance of different baskets not weighted by market cap can each outperform a market-cap weighted index because of the distortions of the large caps; hence we compare the performance of one basket against another relevant basket or the total basket of stocks.

On the cheapest PE basket, we applied various CG screens. From this basket, taking only those in the highest CG quartile of our ranking does not lead to better performance. The basket of stocks with the lowest (cheapest) quartile PE and top quartile CG scores gave an annualised return of 40.5% over the last ten years, underperforming the lowest PE basket without any CG filter which produced a 43.1% annual return.

However, the market seems to hold back somewhat on cheap PE stocks with low CG. Of the lowest quartile PE stocks, if one perversely chose stocks with the lowest quartile CG, one would have managed a return of 34.9%, underperforming the straight lowest PE basket by almost 10 ppts annually. But excluding the lowest CG stocks improves the performance of the cheapest PE basket by 1.5 ppts annually, giving an average annual return of 44.6%.

We find that in seven of the last ten years, taking out the worst CG stocks from the lowest PE basket outperforms against the simple low-PE strategy. (The exceptions were 2000, 2005 and 2008.) Although we do not find any simple relationship between CG and beta, high CG is associated with lower operational risk.

If we take a broader notion of risk rather than defining it solely based on the historical correlation of a stock with the market—its beta—then filtering out poor CG names seems to reduce risk while producing slightly higher returns. This would be a travesty of financial theory. As long as CG is under-utilised for portfolios, this appears an opportunity for investors to exploit which corroborates earlier findings by our quants team in the front of this section.

The number of stocks that have seen significant CG change which we think reflects real change in CG is not large, 25 in all about equally distributed among those with score improvements versus declines.

However, the evidence from the last three years is that those with CG improvements see major outperformance which is likely to have been helped by fundamental factors working in their favour. Meanwhile those with CG disappointments have seen stock underperformance although often other fundamental factors may have worked to push the stocks the other way. •

Amar Gill, CFA, is head of thematic research for CLSA in Singapore and Jamie Allen is secretary-general of the Asian Corporate Governance Association, based in Hong Kong.