Weekly Note - June 13, 2024

Major Stock Market Indexes are Becoming AI Theme Funds

Major stock indexes are becoming more like theme funds for artificial intelligence. Indexes like the S&P 500 have moved higher because they have relatively high weightings in major technology companies that are currently beneficiaries of the AI theme. Our portfolios are not participating in the AI theme as much as is the S&P 500, largely because a) the DJIA, which we use, does not hold NVIDIA stock, which has returned 144% this year, and b) the weighting technique used by the DJIA, which will be discussed later, makes it less sensitive to themes in general.

While these factors have affected performance this year, avoiding themes is a stronger long-term benefit. Our portfolios will not suffer the large declines associated with once-successful themes that turn into bubbles, and then collapse.

Risk Associated with Index Concentration

Because of how many stock indexes, including the S&P 500, are constructed, the indexes will become more concentrated in the AI-related stocks if these stock continue to perform better than the rest of the market. A high concentration of an index in a narrow theme creates risks for investors. As the prices of the stocks move higher, they can go too high and overshoot what might be a fair value for the theme and become more susceptible to rapid price declines. While we cannot easily determine beforehand when the theme will peak, as prices continue to move higher they are more likely to come to the peak.

The most common index construction scheme is to weight the stocks in the index by market capitalization, which is the price of the stock multiplied by the number of shares it has outstanding (available to investors in the marketplace). This is called “market capitalization weighted,” or "cap weighted" for short. 

Assume we have an index of two stocks. If stocks A and B each have 100 shares in the marketplace but stock A trades for $10 per share and stock B trades for $20 per share, the market capitalization of stock A is $1000 and for B it is $2000. The index will have 33.3% of its total weight in stock A and 66.6% in B. If a year later the price for A remains the same but B increases to $30, the market capitalizations are $1000 for A and $3000 for B. As a result, the index will now have 25% of its total weight in A and 75% in B, because of the price increase of B. Thus, cap weighted indexes tend to add weight to stocks that are performing well and take money away from those performing poorly. But this is buying high and selling low, which is not what one wants to do. One should reduce the weight of the stock as it moves to extremely high levels.

In contrast, the DJIA does not cap weight the stocks it holds. It simply weights them according to a rather old-school scheme based on share price. The result is that the DJIA index is not as aggressive about adding weight to stocks as they move higher in price and it tends not to become highly concentrated in hot themes. It is for this very reason that we make heavy use the DJIA.

Figure 1 below shows the weight of the five companies with highest capitalization in the NASDAQ and the S&P 500. They tend to be beneficiaries of the AI theme, especially NVIDIA.

Figure 1

The NASDAQ has over one-third of its total weight in just five stocks, while the S&P 500 has 26%. These are high concentrations in these technology companies. In contrast, the DJIA has just 13% of its weight in these stocks.

Figure 2 below shows the return of these stocks from January 1 through June 7, 2024.

Figure 2

Even if AI is good for the economy over the long term, the prices of the AI stocks may move higher than is justified over a shorter timeframe. A similar condition happened with the dot-com boom in the late 1990s. We all use the internet now and recognize that the dot-com economic trend has been valid, but there was a dot-com stock price boom-and-bust 20+ years ago that wiped out many investors.

One metric used to estimate if the price has moved too high is to evaluate the Price/Sale ratio, which is the current price per share divided by the sales of the company (over the most recent 12 months) per share. This ratio can be calculated for a company or an index. Figure 3 below shows the Price/Sales ratio for the five companies along with the Price/Sales ratios for various indexes as a comparison.

Figure 3

NVIDIA is often the face of the current AI theme because its chips are in high demand for AI electronics. It has a Price/Sales ratio of almost 40, which means that if sales stayed at the current level for the next 40 years, it would take all of those 40 years of sales revenue to equal the current value of the company. Of course, investors in NVIDIA are implicitly hoping that sales will grow and be much greater than they are now and will justify the current high price.

For comparison, the NASDAQ 100 index (reflecting a mix of 100 stocks) has a price that is about 5 times the most recent 12 months of sales revenue. The S&P 500 and DJIA have even lower prices compared to their annual sales.

The NVIDIA stock price will at some point be sufficiently high that investors conclude that continued price increases are not realistic. At that point, prices will decline. If it does decline dramatically, the AI theme will have been a bubble even if AI continues to spur broader economic growth.

Many industry professionals understand the problems of cap weighted indexes and have tried various approaches to address them. A common approach is to create an index with the same stocks but simply equally weight them in the index, and equal weighted versions are offered by the companies that offer the cap weighted indexes. The equal weighted versions are not nearly as popular. 

Using an equal weighted scheme results in no one stock or narrow theme getting significantly larger than the other stocks in the index. Figure 4 below shows the equal-weighted versions of the S&P 500 and the NASDAQ indexes and weights of those same stocks in the cap weighted version.

Figure 4

As you can see, the weighting technique has a big impact on the weights of these stocks and, in turn, the degree to which the index will be influenced by a theme such as AI. This difference affects performance periodically when strong themes boost the returns of a small number of stocks. Figure 5 shows the performance for 2024 through June 7.

Figure 5

The cap weighted versions of the indexes have performed much better than the equal-weighed versions, primarily because of NVIDIA and the AI theme. Figure 5 makes the equal weighted versions look like poor investments this year. However, over the last roughly 24 years, the DJIA and the NASDAQ have had similar performance, as shown in Figure 6 below.

Figure 6

Over the period shown, the DJIA returned 7.6%, annualized, and the NASDAQ returned 7.7%, annualized. However, the DJIA gives us more reliable signals and the NASDAQ has more dramatic drops, which is why we emphasize the DJIA in our portfolios.