In an interview for Jack Schwager’s book The New Market Wizards, Stan Druckenmiller famously said that he:
“…focused [his] analysis on seeking to identify the factors that were strongly correlated to a stock’s price movement as opposed to looking at all the fundamentals.”
The factors that are strongly correlated to a stock’s price movements are often related to earnings. Same-store sales, revenue per available seat kilometer, etc. But in some cases, stock prices react well before reported earnings come out. Factors such as inventory levels or commodity prices may be reflected almost instantaneously in stock prices. I call these concurrent indicators.
This blog post aims to identify exactly when investors tend to shift their expectations. I will use a number of case studies to identify the factors that can be used to forecast and track expectation shifts. The case studies are all related to one of two situations:
- Release of a new, highly competitive product
- Changes to industry supply & demand
I will then go through a number of leading and concurrent indicators for particular industries that I have found useful.
Consensus estimates are ignored for the simple reason that investor behavior tends to be systematic. Most of the time, investors simply extrapolate the recent past and ignore any events that happen more than six or twelve months into the future. In addition, expectation shifts are often the result of changes to management guidance rather than single-quarter earnings beats.
New, highly competitive product
Nvidia released its new graphics card GTX 1080 on 27 May 2016. Meanwhile, the company has been benefiting from modest market share gains and a speculative boom in the trading of its shares. What has been driving margins and operating profit up is primarily the sales of its GTX 1080 product. Another positive tailwind has been the GPU power required for machine learning – a growing industry. Judging from Google Trends data, speculative interest in the stock started shortly after the success of GTX 1080 was established and media reports demonstrated customer engagement data that was largely positive. The right time to invest would have been a few months or even a year prior to the release of GTX 1080 and related products.
Tencent released WeChat in January 2011. Hong Kong stocks sold off in 2011, which might have contributed to Tencent’s weak stock price performance through 2012. That WeChat was going to be a success was not obvious at the time. Customer engagement numbers for WeChat were strong and Tencent’s had a previous track record of success. But there was a fear that Tencent’s competitive advantage would weaken in the transition from desktop – where Tencent was very strong – to mobile. Tencent MAU stayed below 50 million up until early 2012. At that time, the desktop QQ application had more than 700 million users. The time when Tencent’s share price really took off was in 2013, when WeChat MAU rose from 150 million (20% of QQ’s) to over 400 million (60% of QQ’s). Also worth noting is that WeChat ARPU went from zero in 2Q13 to almost $7 in 2014, coinciding with the rise in the share price. Hence, the right time to invest would not have been the product introduction in 2011, but rather the time when monetisation took off and the new product started dominating as a % share of total MAU.
Great Wall Motor
Great Wall Motor launched its popular H6 SUV at the Shanghai Auto Show in April 2011. However, the share price stayed flat throughout 2011. China’s overall car sales grew steadily yet the stock price didn’t react until 2013. In my view, what happened was that the H6 was released into a slowing auto market. If it wasn’t for Great Wall’s great sales performance, the stock would have fallen further. The reason why the stock price peaked in 2013 was that the second phase of the Tianjin factory came into operation, pushing down utilisation rates, with margins falling from 17% in 2013 to 12% in 2015. Price-book fell from almost 4.0x in 2013 to 2.2x today. Growth expectations have also fallen considerably since then.
Apple released its iPhone in June 2007. The stock performed well initially thanks to strong customer reception. Its outperformance against QQQ narrowed somewhat during the crisis, and compared to June 2007, it took two years for Apple to start to outperform. By that time – mid-2009 – Apple already sold 20 million units per year. In FY2008, iPhone represented 6% of total Apple’s revenues. By FY2009, that number had gone up to 19% – a growth rate of over 200%. So again, the potential of the product itself was not appreciated until the point when sales really took off – until the point when the iPhone represented a substantial part of Apple’s revenue. It is worth noting that both iPod and Mac sales suffered during the crisis, perhaps explaining the stock’s relative underperformance until 2009. As the iPhone was highly priced, its improving sales number quickly led to an impact on the bottom line.
Before Netflix became an OTT platform, it rented out DVD’s by mail. The streaming service came into service on July 2007, at which point subscribers could start watching video via their desktop PCs. As the streaming service was available to all of Netflix’s subscribers, the profitability of its streaming service was not easily measured through the company’s financial statements. Topline growth was initially modest: from 13% in 2008, and then accelerating to 22% in 2009. Netflix’s stock price performed in line with the index up until 2008 and outperforming greatly from early 2009. User data showed that streaming customers exceeded the number of DVD shipments from mid-2009 onwards. That’s when the growth in revenues and subscriber numbers really started to accelerate, leading to a significant rerating of the stock.
Conclusion: major shifts in investor expectations tend to happen when
- Customer reception of the new product is positive
- The new product segment starts to contribute significantly (20%+) to the company’s total revenue
- The new product is being monetised at a rapid rate
- Higher sales growth may be tempered by lower utilization rates if the new product line requires large one-off investments in new factories
Changes to industry supply & demand
JinkoSolar / Solar 2016+
Solar PV module margins are volatile. Demand for PV modules have risen practically every year, but the product is such a commodity than even small mismatches in supply and demand have impacts on producer margins. The primary determinant of PV module margins is factory utilisation rates. Cycles have historically been two or three years long as lead times are short. In 2016, intelligence provider TrendForce expected global cell capacity additions to reach 13GW to 15GW with demand growing just 4GW. The timing of demand was also skewed towards the first half of 2016, as a Chinese solar subsidy was reduced in the second half of 2016. The government also capped total additions for 2016 at 20GW. The extension of the US investment tax credit past 2016 may have led to somewhat weaker US demand. JinkoSolar’s stock price fell from early 2016, well in advance of the expected global module oversupply of late 2016.
Vestas / Wind Power 2010+
During the great financial crisis, financing for wind farm projects dried up quickly. Following the Chinese RMB 4 trillion stimulus program, Chinese investments in wind turbine multiplied several times over. Chinese competitors such as Goldwind increased their revenue from CNY 3 billion in 2007 to CNY 17 billion in 2010. Manufacturing capacity started exceeding actual installations from 2009 onwards, and total industry production didn’t normalize until 2012. Meanwhile, Vestas sales growth recovered after the GFC but the impact on margins from the industry oversupply lingered for four years until 2012. The primary culprit was cutthroat competition from cheaper Chinese competitors. Inventory/sales for Vestas peaked in 2009 and kept falling until 2014. The right time to sell or to short the stock would have been when inventory levels went through the roof in 2008, or when Chinese production capacity ramped up significantly from 2009 onwards.
Chesapeake Energy / Shale oil E&Ps 2014+
Oil exploration and production companies tend to lock in selling prices through long-term contracts, but lower oil prices hit earnings sooner or later. The correlation between crude oil prices vs Chesapeake’s share price yields an almost perfect match. US oil production accelerated in late 2011. Production hit a peak in early 2015. Production growth in 2014 of 1.2 mbd was the largest in over a century, while demand in 2014 was weaker than the year before. Given that US oil production is just 10% of the global number, the massive supply response did not have an immediate impact on the commodity – crude prices stayed strong for three years after US production went through the roof. Inventory levels however didn’t start rising above the 5-year average until end-of-year 2014, some months after the crude oil price had started to correct. From June 2014 commercial hedger positions also reversed from an extreme. Later in 2014, OPEC indicated that it would not cut production. It’s also worth noting that credit spreads bottomed in June 2014. When borrowing costs go up, over-indebted E&Ps have incentives to drill even faster. The right time to sell or short Chesapeake would have been when commercial hedger shorts hit a peak and oil price started to decline.
SK Hynix / DRAM semiconductors 2014+
DRAM spot prices peaked in mid-2014 and fell continuously throughout 2015 in what is known as the “third DRAM downturn”. The market had been on an upward trend the previous two years thanks to an earthquake in Taiwan and a fire at SK Hynix’s Chinese fab. Gartner numbers show that worldwide semiconductor capital spend rose 12.9% in 2014 and 0.8% in 2015. Total device shipments for PCs, smartphones and tablets rose just 1.4% in 2015 and 3.7% in 2016. The supply increase was mostly driven by Samsung’s capex going from $10 billion in 2013 to over $13 billion in 2014/2015. Spot DRAM ASPs peaked in the first quarter of 2015 and fell continuously until the second quarter 2016. SK Hynix’s share price followed the DRAM price closely. The time to sell or short SK Hynix would have been when DRAM prices started to drop following Samsung’s capacity additions in 2015.
Lennar Corp / US residential housing 2006
Following the relaxation of Fannie Mae and Freddie Mac loan repurchase criteria driven by the Clinton administration in the 1990s, mortgage lending standards deteriorated rapidly. Household formation had been steady at 1.3 million per year throughout the prior decades, but rose to over 2 million in the 2001-2003 period. US housing starts rose from 1.5 million in 2000 to a peak of 2.1 million in 2005. The period between 2003 and 2005 was characterised by higher speculative activity in the housing market, as evidenced by the 20%+ subprime share of the mortgage market. The Case-Schiller National Home Price Index peaked in mid-2005. Inventory levels – as measured by Monthly Supply of New Housing – rose from August 2005 until December 2008. Homebuilder Lennar Corp’s share price peaked with broader home prices and weaker inventory turnover numbers in mid-2005. Long-term excess supply lasted for over 3 years until banks tightened their lending standards. The right time to sell or short Lennar would have been just at the time when housing prices peaked and inventories started to rise.
Conclusion: Negative expectation shifts in commodity industries tend to happen:
- A number of years after new production capacity has been added to the market
- When inventory starts rising rapidly
- When commodity prices are peaking and are starting to fall
There are other situations where earnings might change quickly: regional expansions, regulatory events, changes to a company’s business strategy, currency depreciation, recoveries from a crisis, etc. Stock price reactions tend to be gradual in these situations as well. Only when the event has an actual impact on earnings does the stock price really start to move.