On November 16, Tesla closed at $408 a share. It had not moved much for two months.
53 days later, on January 8, Tesla closed at $880 a share, up 116%, conjuring an increase of almost half a trillion dollars in market value. In less than two months. To “perspectivize” this incredible number – it is about 9% larger than the entire Gross Domestic Product of Israel.
How to explain this “breakneck rally”?
MORE FOR YOU
The immediate trigger at least is clear. After the market closed on November 16, S&P Global announced that that Tesla would soon be added to the S&P 500 Index, the world’s leading equity benchmark.
There has to be more to the story…
The S&P 500 Effect
- “The S&P 500 index addition effect is around 5-7% percent in the month following the addition announcement date and a large fraction of the price effect is permanent.”
But this effect is thus modest, and may be shrinking as portfolio management techniques have improved. Reviewing data from the period following the 2008 financial crisis, a 2017 study found that –
- “There are currently  no tradable abnormal returns between announcement and event dates in the post-crisis sample period, indicating smoother rebalancing mechanisms by bank’s client facing desks and better services for passive end-investors.The results could be attributed to improved execution algorithms used by the banks, and potentially to the new regulatory reforms…”
The Tesla Anomaly
Clearly, there was no “shrinkage” in the Tesla situation.
What caused this amazing surge?
It Was Not The S&P Index Effect
- The increase due to the membership change is far too small – a few percentage points
- The S&P effect is sometimes explained as a function of increased investor awareness – the idea that a company joining the benchmark attracts more attention, more buying interest. But Tesla already enjoyed an extraordinarily high public profile.
- The positive psychological impact of index membership, as a signal of quality, or stability – the prestige factor, we might call it – is also insufficient to explain Tesla’s performance over the past 53 days
It Was Not The Company’s Fundamentals
- With Tesla now carrying a price-earnings ratio of more than 1600, it was clearly not driven by the company’s earnings performance. Tesla’s profit margin was less than 2%.
- Reported sales were “in line with guidance” — no surprises.
- There were in fact no game-changing fundamental announcements related to Tesla’s business in this period.
So what happened? Was there something different about this index change that amplified or distorted the impact of what should have been a fairly routine adjustment, creating the valuation anomaly we are seeing? Do the index managers bear some responsibility for what many are now calling a “bubble”? Maybe.
How the Addition Was Managed
The change to the index was announced on Nov 16, the announcement date (AD), and scheduled to take effect 35 days later, before the start of trading on Monday Dec 21, the effective date (ED). It was clear from the moment of the announcement that passive index-tracking investment funds with trillions of dollars of assets under management would be forced to buy tens of millions of Tesla shares to align with the new index composition. And they would have to do it more or less all at once, in the open, at whatever the market price might be, and at the last minute (technically that was 4:00 pm, Friday Dec 18).
In such a situation, a reciprocal process quickly develops. Hedge funds and other active investors know that the passive funds will be forced to buy shares to meet the new weightings in the index. They know how much they will buy, and when. So the actives can front-run the passives – legally – buying shares at today’s price to sell to the index funds who will be forced to pay tomorrow’s price.
S&P understood this. It was clear that a very great number of Tesla shares would have to trade in a very short period of time (as described in my previous column). They had to worry about that volume, and the volatility that might develop.
The Volume Surge
The amount of Tesla shares that would trade would be set by the index weighting, which in turn is determined by the price of the shares at the ED.
This, however, became a moving target, as active investors began to buy. On Nov 16, S&P estimated the index-tracking funds’ rebalancing requirement at $51 Bn. Two weeks later the surge in Tesla’s price had driven up S&P’s calculation to $72Bn. A week after that it was $84 Bn – about 130 million shares. (I think that the final number was more than $100 Bn, and maybe as much as $150 Bn – after taking account of benchmarking and closet-indexing funds.)
The timing was set by the deadline – the close of business on Dec 18. An S&P analyst said: “The goal of indexers is to buy Tesla at the close on Friday [to minimize tracking error].”
In the event, Tesla traded 222 million shares on Dec 18, including 69 million shares in the “closing print.” The day’s total was equal to almost 30% of Tesla’s total “float” (i.e., shares available to be traded by the public).
The Volatility Spike
Volatility is not well-understood.
- “Much research has been devoted to modeling and forecasting the volatility of financial returns, and yet few theoretical models explain how volatility comes to exist in the first place.”
But it was clear that there was a risk of high price volatility in this situation.
- “‘This is the largest issue we have ever put into the index and it will disrupt the market the most,’ said Howard Silverblatt, senior index analyst at S&P Dow Jones Indices. ‘Trading on Friday, Dec. 18 will be extremely hectic.’”
Indeed, the market exploded in turbulence as the witching hour approached on Dec 18:
- “Ramming Tesla into the S&P 500 was expected to cause dislocations for other companies as index-fund managers shifted holdings to make room. In the final 30 minutes of trading, 87 stocks jumped by at least 1%, while three fell by that much. All told on Nasdaq, taking every stock into account, a record 1.7 billion shares traded in closing auctions Friday, worth about $150.8 billion.”
The index itself shuddered.
- “The benchmark fluctuated widely heading into the close. The S&P 500 climbed almost 1% in less than an hour heading into Friday’s close, as the benchmark almost wiped out a 1% loss in the last few minutes of trading before turning back in a final fit of selling.”
Mitigating the Surge: Designing The Trading Window
The index managers anticipated the potentially disruptive effects of the Tesla addition, and considered several unusual changes to the “trading window” to mitigate the market stress.
Adding Tesla in Multiple Tranches?
One measure that was considered, but not implemented, was the idea of adding Tesla to the index in stages, instead of all at once. It had never been done before – but there had never been an addition like Tesla before.
- “To avoid missteps, S&P polled big investors on whether they would prefer adding Tesla’s weight all at once on Dec. 21 or split over two trading days in December—an unprecedented move for S&P.”
“Missteps” indeed. Investors were worried, and supported the idea.
- “Asset managers and trading desks across Wall Street have held virtual summits to debate the matter. The vote from many appears to be for the two-day option, partly because of Tesla’s size, along with the potential for elevated volatility in the stock market.”
Some thought the proposal didn’t go far enough.
- “Investors who had shared their opinion with S&P have offered another suggestion that appears to have earned broad support: breaking the trades up over two different quarters, according to people familiar with the discussions. A longer break between the trades would help asset managers digest any sharp moves … and help keep funds in line with benchmarks, investors said.”
It would have been interesting to see how this might have played out. But – somewhat surprisingly – S&P decided to go ahead and add Tesla all at once. An interesting choice, and a bold one. (We’ll look at why they may have made this decision in the next column.)
Lengthening the Trading Window
The step that S&P did take was to extend the trading window (the AD-to-ED period) to 35 days. This, too, was unprecedented. Let’s contextualize it.
Prices are most volatile when large orders surprise the market from either side (buy or sell) without immediately finding a matching level of interest on the other side. An index addition by definition creates large incremental increase in demand for the stock being added – from 4 to12 times the normal volume on the ED when the index funds have to rebalance. Lengthening the AD-to-ED trading window is thought to give more time for the market to find a balance between supply and demand.
This was not always the policy. Prior to 1989, there was no window. S&P’s predilection was secretive. Index changes were seen as “potentially market moving and material… Therefore, all Index Committee discussions are confidential.” The announcement and the execution were simultaneous, even for big changes. In 1983, for example, when the Bell System was broken up, 7 stocks were dropped to make room for the 7 new “Baby Bells.”
- “The exact nature of the change in the S&P 500 was kept confidential until the close of the New York Stock Exchange on November 30, 1983. At that time, subscribers to S&P’s notification service were contacted simultaneously and told of the changes in the index. Although some trading on the Pacific Stock Exchange was possible, most of the trading in these stocks occurred on December 1.”
But the market was changing. Indexing was becoming more than just a market metric. It was evolving into a foundation for index-tracking investment funds. That meant that index changes were beginning to move significant amounts of money.
- “[Prior to 1989] changes were announced after the close of market on Wednesdays, typically around 4:30 p.m. EST, and the change in the index became effective the next day upon the market’s opening. Since index funds are concerned with minimizing tracking error, they must buy (sell) the stock at the time of its addition to (deletion from) the index. However, with the growth in indexing, orders from index funds at the opening bell increased order imbalances and volatility. To ease these order imbalances, Standard and Poor’s began pre-announcing index changes from October 1, 1989 onward.”
This “pre-announcement” policy created the AD-to-ED trading window, a new policy instrument. The window soon averaged 4-5 days in length – which still seems to be the norm. In special cases, usually involving large weighting changes, the window could be lengthened to allow for more time for buyers and sellers to adjust. In 2002, for example, 7 large foreign companies were simultaneously deleted from the index and replaced with US-based companies (including firms such as Goldman Sachs, UPS and Prudential). The move affected 2% of the index weight (even larger than Tesla). The AD-ED window was widened to 10 days. A study of the 2002-2013 period found a few trading windows up to nearly a month in length.
Lengthening the window cuts two ways, however.
The positive side of a longer window is that it gives the market more time to find the balance between supply and demand. It helps stabilize the market when large companies are added. When Berkshire Hathaway was added in 2010, the trading window was 17 days. But a longer window also creates an arbitrage opportunity which many find problematic, giving fast-movers an advantage over the “ordinary investor.” It raises the costs for the index-tracking funds, due to the front-running described in the previous column. In general, a longer window “allows active traders to achieve profits at the expense of index trackers.” Which also impacts all the “ordinary investors” who invest in index funds.
Still, the window is a useful policy tool. The trade-off analysis probably looks something like this:
The bias seems to be in favor of shorter windows most of the time. Index managers probably view the index-tracking funds (and their customers) as their more important constituency. Shorter windows keep their costs down. A study of 228 firms added to the S&P 500 between 2002 and 2013 found that all but 10 were effected in 7 days or less.
Setting the Tesla Window
Following this reasoning, it makes sense that S&P announced a 35-day AD-ED schedule for the Tesla addition, the longest ever (?).
- “The index buyers and potential Tesla share sellers have had ample notice of the Tesla addition — a full month instead of the usual five days that S&P Dow Jones usually provides when changing the index…. ‘Since we’ve given a month’s notice, everyone has had time to plan out their strategies,’ [said S&P]”
The long window worked – sort of. Tesla was added to the index, and life went on.
Sort of. It is also apparent that the 5-week run-up of 70% created a weird momentum in Tesla’s stock (and perhaps in the larger market) that has continued after the effective date. It did not follow the normal pattern of the “S&P 500 Effect” — which would have looked for a new equilibrium to be established following the change. The steady rise in the size of the front running position ended up creating a kind of “coiled spring” – and what happened when it uncoiled was perhaps not exactly what S&P had in mind. The trading in Tesla on the final day may have set a record for the highest dollar volume ever traded in a single stock in one day.
The outcome here does not feel like a new equilibrium. It feels like a bubble. Perhaps we should not equivocate. It is a bubble. The huge volume of front-running here – $100-150 Bn perhaps – may have acted as a technical and psychological accelerant, a ramp to launch this skyrocketing valuation. The unusually long trading window may have distorted the normal pricing process in the market by allowing for so much front-running, which became self-reinforcing. Perhaps it would have been a better idea to introduce such a heavy new component in stages, as many (most?) investors preferred. In the next Part of the series, we’ll consider why S&P may have decided to go against the market advice it was getting.