Market volatility

By Will Braynen

Nothing in this article should be taken as financial advice;
I am a software engineer, not a financial advisor.

If engineering were like finance, I would have one colleague telling me, “the probability of an app crash is 91%” while another colleague is telling me, “it is impossible to predict an app crash with any probability higher than 40%.”

If the Economist article doesn't show up in regular Chrome, try using Incognito mode to at least see some of it.

And yet the very next day (on January 5th) after the shiny job report (January 4th) that the government issued in the middle of its shutdown, it was The Economist who, blaming it on a JP Morgan model, forecast that “the probability of a recession in America in 2019 is as high as 91%” (link or page 50 of this week’s issue). Thanks The Economist; 91% sounds very precise. According to this Vanguard study, on the other hand, it is not possible to accurately predict where the market will go a year out. Even “P/E ratios have ‘explained’ only about 40% of the time variation in net-of-inflation returns” the Vanguard study writes. (Of course it sounds like the first one is about the economy and the second is about the market, but never mind that for now.)

But let’s assume that, contrary to the Vanguard study, with enough sophistication it is actually possible to build a predictive model. Confusingly then, according to a Bloomberg’s article from December 9th—so before the shiny job report—a JP Morgan model estimates the probability of a recession in 2019 at only 35%. And 35% is nowhere nearly as scary as 91%. But of course media articles never give us a timestamp for when the model was run. (I wish they did, assuming of course that models are always run on fresh data, so we could make sense of the barrage of conflicting numbers; but they don’t.) And so, without a timestamp showing otherwise, it is true that this more optimistic prediction might have been from a later run of the same exact model as the one mentioned in The Economist, even though the Bloomberg article came out earlier. See? Confusing.

In other words, I have no idea when the JP Morgan model spat out its prediction because no references to their models are ever timestamped with anything more than the date the article was published. At best, I can maybe assume that later articles are referring to later runs of the model. Except that in the case above, the order then doesn’t seem to make sense.

To make things even more confusing, the same article in The Economist also writes that a “different model built by J.P. Morgan analysts, this time based on short-term economic indicators such as car sales, building permits and the unemployment rate, put the probability of recession in 2019 much lower, at 26%”, which is just to say that the economy is doing well even if the markets are in strange unsteady moods. So wait, there are two JP Morgan models?! I guess the Bloomberg article must’ve been referring to this second model? So maybe the order of publication does make sense after all? (Can they version their models or name them or something? This is getting really confusing when googling and cross-referencing with other media sources.)

Oddly enough, I couldn't get the same numbers as Fidelity got in their example. As I understood their example, you start out with $10,000 in 1980 and invest in an S&P 500 index fund. Let’s say you are 28 years of age then. By the time you are 65, which would be in 2017, according to Fidelity your initial investment grows to over $600,000 if you buy and hold throughout. But if you were to miss the market’s best five days, you would end up with $200,000 less. (Missing the best fifty days would devalue your portfolio by more than 90%, leaving you with only $54,000 when you could have had $600,000.) That’s Fidelity’s example. The reason I don’t understand how they got the numbers is this: In 1980, the S&P 500 was valued at $110. By 2017, it was valued at $2500, give or take. In other words, it was worth about 22 times what it was in 1980. But if so, then doesn’t that mean that $10,000 invested in an S&P 500 fund in 1980 would have grown to $220,000? If so, then where did the Fidelity $600,000 ($615,363 to be exact) come from?

But whatever I think of predictions, especially given the full spectrum in the bouquet of predictions of late, markets are definitely volatile. That’s not a prediction; that’s a description. Vanguard, practically on its website’s landing page, assures its users, “You’re not imagining things. The markets have been more volatile lately”, giving us three tips for “weathering the market’s bumpy ride” (link) while Fidelity argues that buy and hold is a better strategy than trying to time the market (link). Media scares us while brokerage houses try to calm us down. (As for JP Morgan modelers—I don’t know what those guys are up to!)

The Wall Street Journal even resorted to surveying professional economists for their opinions (here and here). Accordingly to some sources, the sample size in the WSJ survey was sixty. Sixty economists. (Was it a representative sample of all economists?) 59% percent of these sixty felt rather grim about things. 59% of sixty economists is 35.4 economists… But, as I don’t know the magic they are capable of, I digress. 35.4 it is. I am not going to tell them how to be.

This feels like an odd application of the idea of the scientific consensus, especially when there is lack of consensus (e.g. 59% is not consensus) and with an eye to trying to figure out when exactly to pull money out. Unlikely, but, just to help make a conceptual point: if for example, the market grows by significantly more over the next two years than it will lose in the six months that follow those two years, then in hindsight you would have been better off staying in rather than missing out completely. (But of course see here.) But let’s assume ours is a more modest aim: not trying to figure out net gain or net loss of, say, the S&P 500 over the next five years, but merely trying to figure out if there’ll be a recession this year (2019) or the next (2020) or the next (2021) or the next (2022). (That’s a lot of nexts; I thought 2022 was closer.) If 59% of economists, whether or not they specialize in recession questions and related predictive modeling, guess that a recession is coming sometime between 2020 and 2022 (or whatever the numbers), would it be epistemically responsible of us, based on this, to form the belief that there is a 59% chance that the next recession will come between 2020 and 2022? I am not sure it would be, but that seems to be the implicit suggestion. I am not sure it would be epistemically responsible because I think it is odd to replace a predictive statistical model with human barometers. It is odd unless we believe these 35.4 plus the remaining 24.6 economists to be a representative sample of all buyers and sellers of stocks whose behavior we are trying to predict (or something analogous with the economy instead of the market). But why would they be representative when the whole reason we picked them is that they are special—experts in something or other. The whole point was that this was a biased sample; it’s why we picked ‘em!

Given all this, I’ve been struggling with two basic questions: (1) Why would there be a recession if the economy is doing well? (There are speculative answers offered to this question in The Economist article.) And (2) Should I try to time the market? (Or, assuming this is not a false dichotomy, am I better off “buying and holding”?)

Again, recession (first question above) is about the economy while buying and holding (second question above) is about the market, but never mind that. For now, for more context, below is a graph of the S&P 500 today and historically. This could literally reflect someone’s portfolio if they passively invest in an S&P 500 index fund, for example Vanguard’s VFIAX (interestingly, Vanguard does not have a Nasdaq index fund). So, the graph:

S&P 500, historic prices from ancient times to today

Even if we are dealing with a tilted sinusoidal wave—sinusoidal because our commitment to first principles, business cycle and all (except that the above is for markets, not the economy!), and tilted because of economic growth—it would be crazy to extrapolate anything from the shape we see in the graph above. I mean, imagine this were a physics lab report! (If we must use a physics analogy, then the economy is more like a physical environment in which the underlying causal structure keeps changing given globalization and changing laws and climates.) What is obvious from the above graph is that since 2009 we’ve had a really nice unprecedented run upwards—a nice long “bull run”. I suppose that’s why we hear people like Kai Ryssdal say things like: it’s not a question of whether a recession will happen, but a question of when. (Or something to that effect—can’t remember verbatim.) But then again, the whether-when quip is always true if we believe there is always a next recession, even if that recession comes in 2050. So, if a quip—clever, but definitely panic inducing for those who don’t catch the funny (because they are already a bit panicked).

Of course if you look at the same graph with a smaller time window, it looks more like a real business cycle that’s currently getting overheated:

A representation of the S&P 500 time series that makes it feel like, “yeah, the normal business cycle is a low of 650 and a high of 1500”

But that’s just an artifact of the range we picked for the y axis.

1. Why would there be a recession if the economy is doing well?

With the first graph above in hand, back to my first question: why would there be a recession if the economy is doing well? The answer “because we’ve had a bull run for so long and we are due” doesn’t feel very satisfying for two reasons. Firstly, recessions are ultimately supposed to be about how the economy is actually doing, not about how the market feels about things. Sure, market sentiment is investor sentiment and investor sentiment can sour the economy. I get the idea of feedback loops, but still. Secondly, I neither expect the business cycle to look like a clean sine wave, nor does the business cycle guarantee a large amplitude—and my conception of a recession (which is apparently wrong) is that it somehow involves a large amplitude; correction is one thing, recession is quite another. Or so I thought. That said, is there a technically precise definition of recession and is that definition tied to the economy or the market? I ask this because if recession is just the decline in GDP for two consecutive quarters (as defined in Merriam Webster’s), then I do not see how from that definition alone it could follow that the stock market will drop by a percentage I should care about or that it will even drop at all (although I guess commonsensically it will, because human nature).

2. Should I try to time the market?

No. That’s the short answer. (I am not a financial advisor, so this is not advice; it is merely my thoughts to myself about this; by “I” in the heading, I mean me, not you.)

For the longer answer, I’ve asked myself what’s worse if/when a recession comes: buying today and holding through the end of the recession, or getting the timing wrong?

As I understand it, leaving the market and then coming back in is called timing the market. By ‘market’ I mean the stock market. If you are planning to retire soon (e.g. within ten years), you shouldn’t have all your money in the stock market anyway. But if you are still a good thirty years from retirement, then, if you exit, you’ll have to come back into the market at some point, which means that you would be trying to time the market. And timing the market is, I gather, something I should leave to the professionals.

So, for now, because I don’t plan to hire professionals, I figure that buy and hold makes sense. I think I have the long-enough time horizon and I plan to keep adding more money to my retirement fund at an even rate every month. That way, if there is a recession, (a) I have enough time to recover from a downturn, and (b) I will have bought at all kinds of price points and so, even if I will have overpaid, I will have also underpaid later, so in the long run my buying price will average out. Will that average price be a good price given a thirty-year time horizon? And I guess this was a false dichotomy after all; this strategy doesn’t sound like a buy and hold; instead, it sounds like a whole bunch of staggered buy and holds. Either way, this might force me to think about what I think will have value in the long term.

The trick might be to not monitor the situation too closely. Otherwise, tomorrow I might read something else or receive some other mixed signal from the market (or the media!), inducing me to change my mind again. Because volatility. (Or is the other way around?)

Don’t get me wrong; I am not saying any of the above predictions won’t come true. Mixed signals aside, it’s the reasoning that weirds me out. But sure: we could well believe something that happens to be true, but believe it for the wrong reasons (just as we can believe it for no reasons at all). Even invalid arguments can have true conclusions (while perfectly valid arguments can have false conclusions). And in this information climate, as a layman investor of retirement necessity, I don’t know what conclusions to draw or what to believe. Hence the staggered buy and holds strategy above. Because that strategy is a dominant strategy: it should work okay no matter what the market does. Or so I hope. (Now what to buy, even if you plan to hold it whatever it is, is another question.) What I am really after, after all, is peace of mind and some degree of understanding, not profit maximization per se.

As for The Economist article (or is it “the The Economist article”?), it was probably written before the job report came out and so published anyway despite the job report that came out the day before.

1. Why would there be a recession if the economy is doing well?

2. Should I try to time the market?

Further reading to help calm the nerves