Friday, May 29th, 2020

Re-Configuring the S&P Sectors

If someone were designing the S&P 500 today, they would not produce an index where the largest single sector accounted for almost 27% of market capitalisation and was over 10x the size of the smallest. Nor would they allow a situation where the bottom three sectors (excluding REITs) were only just larger than the smallest of the remaining sectors. Even after the removal of Alphabet and Facebook, Technology is 10x the market capitalisation of Materials and almost twice as large as Healthcare, the second ranked sector. Energy is now just 3.0% and Utilities are 3.2%. Some of these extreme differences may unwind as the economy returns to normal after the pandemic, but this was already a problem before the virus struck. Eventually, there will have to be major changes and we may as well start thinking about them.

First, the Technology sector will have to be split. We think there should be a new Software and IT Services sector and that the rest of the Technology sector be renamed Hardware and Equipment. Second, Materials, Utilities and Energy need to be combined with other sectors. We would put Materials with Industrials and Energy with Utilities. All this can be done without changing any of the industry categories which the S&P currently uses, thereby retaining full backwards compatibility. The third change is not so important, but we would also clean up the Healthcare sector, by taking Care Providers out and putting them into Consumer Staples. The resulting index would have 10 sectors instead of the current 11. Excluding REITs, the weight of the other nine sectors would range from 6.0% (Utilities and Energy) to 14.7% (Software and Services). Apart from these outliers, the weight of the other sectors would range between 10.1% and 12.0% of the index.

It all sounds very technical and abstract – the sort of thing that quants obsess about – but there is a real point to this. Under the current system, it is very difficult to get a sector-based model to outperform, hence the current fashion for factor-based investing. We have no problem with factors, but a good classification system would naturally group together companies with similar characteristics and separate those with different ones. Hardware and semiconductors are fundamentally different businesses from software and IT services. They outperform at different stages of the cycle. Energy and Utilities will increasingly be dominated by the impact of environmental legislation and parts of these two industries will probably merge or overlap over the medium term.

The second reason for rearranging the sectors is that risk control is more effective and simpler when dealing with nine or ten sectors of broadly equal size. Sophisticated investors don’t need this help, but retail investors (and their advisers) do. If they don’t understand how the US market is structured – and the reason why – there is a danger that their portfolios will be poorly diversified. Alternatively, they may be pressured into buying complicated and expensive investment products that they really don’t understand.

The third reason for doing this, is that it works. Over the last five years, the S&P has generated a total return of 155%, compared with 162% generated by our risk-adjusted momentum process, using the current sector definitions. The same process using the new definitions would have produced 174%, equivalent to annual outperformance of 2.5% vs 0.9%. This comes with slightly higher volatility and slightly bigger maximum drawdown, but the Sharp ratio is 0.63 vs 0.53 for the index. Well-designed sectors make portfolio management easier and the potential rewards are worth the short-term dislocation.

Leave a Reply

Full Blog Archive:


Other Recent Posts