Tag Archive: software

Tech fundamentals still favor Hardware over Software

While the Technology sector has been less dominant in terms of returns this year than it was last year, it remains the largest sector in the US market by value and the focus of much investor attention.

Our view within the Technology sector for some time now has been to favor hardware-related industries over software-related or services areas, and the latest update of both bottom-up and top-down indicators continues to support this view.

The table below shows some of the key bottom-up fundamental metrics we track for all sectors and industries, and focuses on the six US Technology sector industries (using the GICS industry classifications) within our broad 2300-stock US universe.

The revisions breadth measures shown in the table are based on the proportion of analysts covering each stock who have raised their earnings estimates over the last three months net of those who have lowered earnings estimates. That is, the number of upward revisions to estimates minus the number of downward revisions relative to the total number of analysts covering a stock. We use a weighted average of estimates for each company’s current fiscal year (2021 for most companies right now) and next year (2022) to calculate a consistent 12-month forward earnings figure.

Source: Mill Street Research, Factset

The first column shows the average breadth reading for all stocks in each industry on an equal-weighted (“EW”) basis, while the second column shows the same data on a cap-weighted (“CW”) basis. Naturally, the cap-weighted figures give much more weight to the largest stocks in each industry, while the equal-weighted figures will be relatively more tilted toward smaller names. All stocks in our US universe require a minimum market capitalization of $200 million and at least three analysts reporting estimates, among other criteria, so there are no micro-caps or stocks with very few analysts included.

A few key points jump out from the table:

  • Revisions remain net positive in most areas of Technology, in line with the broadly positive revisions activity in US stocks overall.
  • The average revisions breadth readings on a cap-weighted basis are higher in every industry than the equal-weighted readings, and often much higher. The sector-wide average of about 51% far exceeds the equal-weighted average of about 19%. This means the largest Tech stocks have much more positive analyst activity than the average stock in the sector, i.e., a big-cap bias.
  • The strongest industries are those which focus on hardware right now, while software and services industries are substantially weaker.

We see that on an equal-weighted basis, the Technology Hardware, Storage & Peripherals industry has the highest proportion of positive revisions, followed closely by Electronic Equipment, Instruments & Components and then Semiconductors & Semiconductor Equipment. All three of these hardware-related industries also have the highest revisions breadth on a cap-weighted basis.

At the bottom of the list we see the large Software industry, which has marginally negative net revisions breadth on an equal-weighted basis. This means the average Software stock has a roughly equal number of analysts raising versus lowering estimates. However, on a cap-weighted basis, the revisions figure is strongly positive. This means that a few mega-cap names in the industry have strong revisions, while the majority of mid- and smaller-cap names have relatively weak revisions.

A similar but less dramatic picture is seen in IT Services, where revisions are much lower than in the hardware related industries on both equal-weighted and cap-weighted metrics. The cap-weighted figure is still higher than the equal-weighted, as it is for all Tech industries, but by a much smaller margin than in Software.

The overall picture from the bottom-up, stock level analyst revisions data is that hardware makers continue to surprise analysts positively across large and smaller companies alike, while providers of software and tech-related services are much more mixed and dominated by the largest names in those areas.

The second chart below provides some macro, top-down context for the bottom-up fundamentals we see. The top section of the chart plots the year-on-year percent changes in aggregate spending on investment in computer hardware (blue line) and in software (red line) over the last 30 years. The data come from the quarterly GDP data produced by the US Bureau of Economic Analysis. As of the end of 2020, the growth rate of investment in computer hardware had jumped to 16%, the highest reading since the previous recessionary rebound in 2009-10. Growth in software investment spending, by contrast, had only grown 5%, which is near the lower end of its recent historical range.

Source: Mill Street Research, Bureau of Economic Analysis

The bottom section of the chart plots the ratio of software to hardware spending over time. We see that the line has tended to rise, indicating higher growth in software spending than hardware spending on average over the long run. But the latest readings show a drop, reflecting the jump back toward hardware spending that occurred last year and likely is still ongoing. In the longer-term, it seems likely that software spending will again resume its higher growth rate versus hardware, but the preference for hardware is still very visible in current analyst earnings forecast behavior and could last a while longer as the world continues to adapt to the post-COVID landscape.

Semis vs Software trade now favors Semis

Within the broad Technology sector, there are often significant divergences among the various industries. A key intra-sector industry relationship that many investors use as a touchstone is the relative performance of Semiconductors versus Software.

These two industries capture different parts of the Technology ecosystem. Due to their widespread use in so many devices and products, the Semiconductors and Semiconductor Equipment industry reflects demand for hardware, both within Technology (servers, PCs, phones) and in other sectors (e.g. autos), and thus tends to be much more cyclical. Software tends to be much more stable, with more recurring revenue, and nowadays is closer to a service-type industry. There is much less chance of major “shortages” or “oversupply” of software of the kind that semiconductor makers must often deal with.

So even though software and semiconductors are complementary products (each requires the other), it is not hard to see that they can often have significantly different fundamentals and relative returns on an intermediate-term basis.

Our indicators currently show a growing shift in favor of Semiconductors over Software, both in fundamental earnings trends and relative returns. This is a reversal of the trend seen in the 2017-2019 period, when Semis lagged Software, and much of 2020 when relative performance was mixed.

The chart below shows the strong relationship between relative earnings estimate revisions activity in the two industries and their relative returns. The data are drawn from our broad US stock universe of about 2300 stocks (roughly, all US stocks with at least $200 million market cap and three analysts reporting estimates), and the constituents of each industry are equal-weighted in both the revisions and return series. The return series uses a five-day moving average to show the trends more clearly. Earnings estimate revisions breadth measures the average net proportion of analysts raising versus lowering estimates for each stock. Readings above zero mean more analysts raising estimates than lowering them on average.

US Semiconductors vs Software Revisions and Returns

We see that Semiconductor revisions breadth began losing its relative strength versus Software back in 2017, and continued through early 2019. At that point, Semiconductor revisions started to recover while Software continued a slow deterioration, allowing the relative revisions spread (middle section) to turn up from very negative (i.e., pro-Software) readings. That spread turned positive (favoring Semiconductors) in early 2020, just before COVID-19 hit, and then weakened again as many industries weakened simultaneously in the spring.  Both industries then had a simultaneous sharp rebound, along with most of the rest of the market, through the fall.

The last few months are where we again see a distinct divergence. Software revisions have clearly been losing momentum since September while those of Semis have held up and actually grown somewhat stronger. The spread is now quite wide and at multi-year highs in favor of Semis.

The bottom section of the chart shows the relative returns of the two industries, and we see the clear tendency for relative returns to follow the relative revisions. The relative return series has recently broken out of the range it inhabited for most of 2020, and looks set to follow the relative revisions higher. This suggests that Semiconductors should continue to outperform Software as long as the relative earnings indicators maintain their recent clear bias toward Semis.