THE DAILY EDGE: 22 July 2024

Note: I will be travelling (Pacific time zone) until August 10. Posting will be irregular and possibly limited by time and equipment constraints.

EARNINGS WATCH

From LSEG IBES:

70 companies in the S&P 500 Index have reported earnings for Q2 2024. Of these companies, 82.9% reported earnings above analyst expectations and 11.4% reported earnings below analyst expectations. In a typical quarter (since 1994), 67% of companies beat estimates and 20% miss estimates. Over the past four quarters, 79% of companies beat the estimates and 16% missed estimates.

In aggregate, companies are reporting earnings that are 5.0% above estimates, which compares to a long-term (since 1994) average surprise factor of 4.2% and the average surprise factor over the prior four quarters of 7.3%.

Of these companies, 57.1% reported revenue above analyst expectations and 42.9% reported revenue below analyst expectations. In a typical quarter (since 2002), 62% of companies beat estimates and 38% miss estimates. Over the past four quarters, 62% of companies beat the estimates and 38% missed estimates.
In aggregate, companies are reporting revenues that are 0.5% above estimates, which compares to a long-term (since 2002) average surprise factor of 1.3% and the average surprise factor over the prior four quarters of 1.2%.

The estimated earnings growth rate for the S&P 500 for 24Q2 is 11.1%. If the energy sector is excluded, the growth rate improves to 12.0%.

The estimated revenue growth rate for the S&P 500 for 24Q2 is 4.5%. If the energy sector is excluded, the growth rate declines to 4.4%.

The estimated earnings growth rate for the S&P 500 for 24Q3 is 7.6%. If the energy sector is excluded, the growth rate improves to 8.8%.

The 70 companies having already reported realized a 16.5% earnings growth rate on a 5.8% revenue gain. Margins keep rising fast!

Trailing EPS are now $230.69. Full year 2024: $243.20e. Forward EPS: $260.30e. 2025e: $278.64.

The combination of rising earnings and declining inflation is keeping the Rule of 20 Fair Value on the rise (yellow line above). Expensive equities are supported by solid fundamentals (rising earnings and slowing inflation).

NBF warns that earnings expectations could be too high:

The frenzy on US stock markets is being fuelled by ambitious earnings expectations, with the consensus forecasting a 13% increase in S&P 500 earnings per share (EPS) over the next 12 months.

But how realistic is this given the current very negative reading of the Bloomberg Economic Surprise Index?

As today’s Hot Chart shows, earnings forecasts have been revised downwards by an average of 7.2% in the three months following a reading of less than -0.5 on the Bloomberg Economic Surprise Index, which is currently the case. As shown, the Energy, Materials and Financials sectors are usually the ones suffering from the biggest downwards revisions, while Health Care, Consumer Staples and Utilities are only marginally reassessed.

Bottom-line: we see a significant downside to the consensus estimates for EPS growth in the coming months.

But the real surprise could be that the economy is not so weak as the surprise factor suggests. Not a moot point considering that equities are often quite susceptible to negative economic surprises:

The initial Atlanta Fed GDPNow estimate for Q2 was +3.9% on April 26. It sank to +1.5%, nearly 40% lower, on July 3rd, heavily weighted down by Personal Income and Outlays data and the very weak ISM PMIs. It has since recovered to +2.7%.

My take on some of the recent crucial stats:

• Real consumer expenditures rose 0.26% in May after –0.13%, +0.8% a.r. in the last 2 months, a sharp slowdown from +1.5% a.r. in Q1. But Wages and Salaries jumped 0.7% in May and are up 5.5% a.r. in the last 2 months with inflation at 2.6% YoY. Absent the rise in the Savings Rate (from 3.5% to 3.9%), real expenditures would have increased 3.7% a.r. in April/May, twice the Q1 growth rate. Strong labor income, higher savings and wealth provide solid spending power going forward.
• Recent ISM data came in very weak but the less widely followed, but more accurate, S&P Global’s PMI surveys were very strong with rising new orders.
• Supporting the stronger S&P Global survey, Industrial Production jumped 0.6% in June following +0.9% in May. Factory output rose 0.4%, in a broad advance that included gains in autos, electrical equipment, appliances and nondurable goods, following May’s upwardly revised 1% increase. Output of consumer goods rose 1% in June, the most in nearly a year. Manufacturing production jumped 0.84% QoQ in Q2, +3.4% annualized, bettering the aggregate of the 5 previous quarters.
• June Retail Sales were much stronger than expected and May’s weak +0.1% originally reported growth rate was revised up to +0.3%. Control Sales rose 0.9% MoM in June after +0.4% in May. Last 4 months: +4.3% annualized with negative inflation.

Overall, the American consumer is in good shape: labor income is strong, inflation is slowing (black), even on “essentials”, and wealth gains have benefitted all income levels.

The Unemployment Rate is up but only because supply has increased. Demand remains solid as S&P Global’s PMIs revealed. Indeed Job Postings have stopped declining since the end of May and even perked up in the first 2 weeks of July.

Economists keep marking down their Q2 GDP forecast, now below +1.8%. The GDPNow, at +2.7%, is above the highest Blue Chip estimate. Barring a major black swan event, the odds are high that economic surprises will surprise on the upside in coming months. Goldman Sachs’ own surprise index has turned up recently.

And, for what it’s worth, U.S. leading indicators are also more positive:

If so, earnings growth will likely broaden beyond the Mag 7. Here’s Goldman Sachs’  David Kostin:

AMZN, GOOGL, META, MSFT, and NVDA are collectively expected to grow 2024 profits by 37% compared with 5% for the median S&P 500 stock. But we have previously noted that consensus estimates of sales growth for the five stocks is forecast to slow from 22% year/year in 1Q 2024 to 17% in 2Q, and further decelerate to 16% in 3Q and 14% in 4Q. NVDA sales growth is expected to slow from 262% year/year in 1Q to 111% in 2Q, 73% in 3Q, and 56% in 4Q.

In contrast, sales growth for the median S&P 500 stock will be accelerating, albeit from a slower pace (year/year growth of 2%, 3%, 4% and 5%).

The recent trend of small-cap outperformance will likely persist unless the macro environment changes substantially, or the mega-cap Tech stocks report 2Q results that causes analysts to raise revenue forecasts for the next several quarters.

In our recent report on AI and US Equities, we discussed that investors have become increasingly concerned about the prospect of “overinvestment” in AI, especially among the four hyperscalers (AMZN, META, MSFT, and GOOGL). These firms collectively spent $357 billion on capex and R&D during the past four quarters – fully 23% of the total spending by all S&P 500 companies!

Consensus estimates of 2024 and 2025 capex and R&D spending by the hyperscalers have increased by $65 billion compared with expectations at the start of the year. However, analysts have lifted their sales forecasts for 2025 and 2026 by only $36 billion – a gap of nearly $30 billion. From an earnings perspective, the shortfall is lower but still substantial ($7 billion) given the profit estimates in 2025 and 2026 were raised by $58 billion.

Simply put, these firms during the past six months have dramatically increased their planned spending on AI initiatives but it is not apparent when the return will come – in 2027, 2028, 2029, or perhaps not at all?

In the late 1990s DotCom boom, sales revisions was the key variable to watch because it ultimately signaled when momentum reversal would be sustained. The potential resumption of the AI trade – and by extension a reversal of the recent underperformance of large-caps vs. small-caps – will depend on revenue revisions. S&P 500 outperformance will resume if big Tech beats and raises its forward sales guidance. If not, then small caps will continue to outperform. GOOGL will report on July 23, MSFT on July 30, META on July 31, and AAPL and AMZN on August 1.

  

Playing small caps? Which ones?

Russell 2000 investors should be aware that:

• Trailing 12-months operating margins is 4.7%, 9pp below that of the S&P 500 companies.
• If the Russell were a single company, it would carry a speculative-grade credit rating: Net debt of 4.4 times trailing Ebitda compares to 1.5 times for the S&P 500 (Grant’s)
• Russell 2000 Q2 earnings are seen down 5.6%, the 6th consecutive down quarter.
• On a trailing 12-month basis, 42% of Russell 2000 companies are losing money. Only 6% of the S&P 500 companies lost money last year.

• The Index P/E of 18.4 is only attractive if you don’t know that this P/E calculation excludes losers. In reality, the P/E is 34.0x trailing earnings per LSEG IBES.

The S&P 600 Index P/E is 17.9x trailing and 15.5x forward. No losers are allowed in this index.

Canada: Retail sales stall in Q2

Retail sales data for May were below economists’ expectations, showing a sharp decrease after an expansion in April (+0.6%, revised down from +0.7%). An increase in sales in the motor vehicle and parts industry mitigated the decline, as the contraction excluding that sector came in at 1.3%, its largest decline since July 2022.

The industrial details behind the decline were not encouraging either, with 8 out of 9 industries registering a monthly contraction, the worst diffusion since July 2022.

Adding to the disappointment, the early estimate for June signaled a 0.3% contraction. According to this preliminary estimate, real retail sales contracted 0.7% annualized in the second quarter of the year.

The picture is even gloomier when accounting for strong population growth, as we estimate that real retail sales contracted 3.9% annualized on a per-capita basis. The clear downward trend in real per capita retail sales since the first rate hikes in 2022 clearly demonstrates the negative impact of restrictive monetary policy on the consumer.

We note that the majority of sectors have seen declines since March 2022, with discretionary spending suffering the most, including real estate-related spending (furniture and building materials). Despite the fact that a first rate cut has been announced and more should shortly follow, we think retail sales could remain muted for the coming quarters. Indeed, rate cuts still leave the monetary policy to a restrictive level, meaning that the mortgage renewal payment shock continues. This is likely to prompt caution from the Canadian consumer, especially as the labour market is cooling rapidly.

AI CORNER

While GOOG, MSFT, APPL, META and NVDA were busy acquiring and developing Gen AI and LLM capabilities, AMZN has been rather quiet. It has invested in Anthropic and is reportedly developing its own LLM (“Olympus”). David says the recent hiring of David Luan and his team at Adept is a significant new step for Amazon.

• Amazon Hires Top Executives From AI Startup Adept for AGI Team

Amazon.com Inc. hired top executives and other employees from startup Adept AI Labs Inc., a move by the e-commerce and cloud-computing giant to bolster the development of artificial general intelligence, an advanced version of AI that can think like a human.

David Luan, Adept’s co-founder and former chief executive officer, will join Amazon’s AGI autonomy team, led by Rohit Prasad, according to an internal memo Amazon provided to Bloomberg. Four other co-founders and an unspecified number of other team members will also join Prasad’s group.

Amazon will license Adept’s technology developing agents, or AI tools that can perform tasks autonomously, to help Amazon build products that can automate software work flows, according to the memo. (…)

“David and his team’s expertise in training state-of-the-art multimodal foundational models and building real-world digital agents aligns with our vision to delight consumer and enterprise customers with practical AI solutions.” wrote Prasad in the note to staff. Luan will report to Prasad as the leader of the AGI autonomy team and the automations team, according to the memo. (…)

Unlike some of its big tech peers like Microsoft Corp. and Google, Seattle-based Amazon hasn’t focused as much on developing a broader, more ambitious artificial general intelligence, or AGI, that can perform most intellectual tasks as well or better than humans.

The Adept agreement also comes at a time that technology companies have stepped up hiring from well-funded AI startups that are building costly foundational models. In March, Microsoft hired a large portion of the staff of Inflection AI and arranged a deal to license its technology for $650 million. (…)

David says that “David Luan is an incredibly insightful and visionary individual. His hiring at Amazon is much more significant than the hiring of Suleyman at Microsoft.”

From Perplexity.ai:

• Luan will lead the “AGI Autonomy” team at Amazon. He will oversee the development of digital agents capable of automating software workflows.
• Prior to co-founding Adept, Luan served as the vice president of engineering at OpenAI.
• As part of the deal, Amazon is licensing Adept’s agent technology, multimodal models, and datasets.
• This move is seen as a significant boost to Amazon’s AI capabilities, which have been perceived as lagging behind competitors like Google and Microsoft.
• The FTC’s interest in Amazon’s hiring of Adept executives follows similar inquiries into Microsoft’s recruitment of employees from Inflection AI.
• This hiring move demonstrates Amazon’s commitment to strengthening its position in the competitive AI landscape, particularly in the development of advanced AI agents and automation technologies. However, it also highlights the increasing regulatory attention on such talent acquisitions in the AI industry.

What are David Luan’s view on applications of LLMs?

• Luan is particularly excited about developing AI agents that can assist humans across various professions. At Adept, his company focuses on building multimodal agents designed to work alongside humans in any profession.
• Rather than full automation, Luan believes AI agents should focus on augmenting and enhancing human capabilities. He envisions a future where AI functions almost invisibly below all software we use.
• Luan emphasizes the importance of developing LLMs for real-world, practical use cases. He believes that co-designing AI systems alongside real users and incorporating their feedback is crucial for progress..
• While acknowledging the power of large-scale models, Luan argues that use-case specific fine-tuning is more critical than simply increasing model size for many applications.
• With Adept, Luan is particularly interested in developing reliable and useful agents for enterprise applicationsLuan highlights the significance of training models on diverse, multimodal data, especially data related to knowledge work, rather than just natural images or text.
• He recognizes the current limitations of LLMs, such as their inability to perform real-world actions, and sees opportunities in developing models that can better understand and interact with computer interfaces.
• Luan is aware of the potential risks and biases associated with LLMs and believes these issues need to be carefully addressed as the technology develops.
• Overall, David Luan’s view on LLM applications emphasizes practical, human-centric use cases that augment rather than replace human capabilities, with a focus on developing reliable, multimodal agents for enterprise settings.

The race is on between behemoths to fill all possible application spaces and build as large and solid user bases as possible as rapidly as possible. Once people get used to specific AI apps and to their dedicated virtual butler (LLMs), it will become very difficult to get them to move.

Yes, there will be excess spending for a while but the chips likely won’t be down until 2026.

BTW, Chinese companies are also rushing to develop AI apps that will be made available outside China. Chinese LLMs are all open source and much lower cost than Western peers. Among major Western LLMs, only LLaMA (Meta) and Mistral are open source.