Are recessions predictable?

Amidst all the talk of a recession, this chart presents something of a puzzle. It shows that there’s a decent correlation (0.57) between annual changes in the All-share index and in UK industrial production*.

The puzzle is that this correlation is contemporaneous. If stock markets anticipated recessions and booms, we’d expect to see the blue line move in advance of the red one. But this isn’t really the case**. Sure, the strongest correlation (of 0.68) comes if we lag industrial production four months. But such a short lag might exist simply because the stock market is a “jump” variable whereas output isn’t. Shares can respond immediately to bad news (such as the collapse of Lehmans in 2008), but because output takes time and must be planned in advance, it can be slower to respond.

For longer lags, the correlation is weak – for example, it’s only 0.23 between price changes and output changes 12 months later.

This lack of a significant lead between share prices and output is especially surprising simply because you’d expect one simply because changes in share prices can cause changes in output – via cost of capital or wealth effects, or because share prices should send signals about future economic conditions.

So, what explains it?

I don’t think the story here is about equities: my second chart shows a very similar relationship between output and sterling’s trade-weighted index.

Instead, all this is consistent with a simple possibility – that stock markets just don’t see recessions coming (or at least don’t anticipate the things that trigger recessions). We know that economists have consistently failed to foresee recessions. Perhaps this isn’t (just) because of their own inadequacies. Maybe it’s because economic fluctuations are inherently unpredictable.

This could be because recessions are the product of complex emergent processes. But there might be a more mundane reason. If some recessions were predictable, policy-makers would loosen policy in advance, thus preventing them***. The only recessions we ever saw would then be the unpredictable ones.  

The message I take from this is that we should be humble about our ability to foresee recessions coming, especially when others aren’t expecting them.

* Given that UK equities are correlated with overseas ones, and UK industrial production with global output, I suspect a similar pattern is true generally.

** The chart is also inconsistent with Samuelson’s quip that the stock market has predicted nine of the last five recessions. If this were the case, we’d expect to see some falls in share prices which aren’t followed by falls in output. However, except in 2003 (and maybe now!) we haven’t.

*** They would then be criticized with hindsight for loosening policy unnecessarily!

Comments

[Peter K.]

As an amateur, I see recessions coming in two flavors: one type are brought on by central banks tightening, the other by popping asset bubbles.

Neither are predictable. Central banks are driven by politics and they tend to be overly afraid of inflation. It's hard to predict political outcomes.

Bubbles are driven by greed and fear. At a certain point the recent housing bubble was unsustainable.

[Magnus]

Wake me up when the blogosphere realises that these things are inherently unpredictable.

It would be blatant trolling to report that said unpredictability is entirely consistent with the Efficient Markets Hypothesis.

[Metatone]

Contra Magnus, and perhaps taking a bit from Peter K. I think it's clear that some recessions are inherently unpredictable. Others are rather predictable and the refusal of economists to make or investigate that distinction is more about them defending their models and preferences than good scientific inquiry.

(Yes, there's a whole science around complex and chaotic systems out there, there's all sorts of criteria for the definition and exploration of predictability, and no, economists largely don't ever reference that literature in their discussions.)

[Ari Andricopoulos]

I find that the return in the S&P500 one year has a 6% coefficient on economic growth across the developed world in the next year. So a 20% fall in the SPX results in 1.2% lower growth, and this is after controlling for a general loss in confidence that would manifest itself by a large rise in debt.

A bit complicated to explain but it does so here:
http://www.notesonthenextbust.com/2015/06/the-government-must-run-deficits-even.html

What I will say is that I think the decline/rise in stock markets are more causal of a recession than predictive of it.

[Matt]

Note that the UK IP series come out about 1.5 months after the data period, so e.g. December's will be out tomorrow. So best fit perhaps 6 months, not 4 months.

[Deviation From The Mean]

Recessions are like snow, you know it is going to happen but can't predict the day and time.

The best you can say is that capitalism goes through cycles and proceed from there. But in order to understand the cyclical nature of capitalism you need more than a graph showing the correlation between annual changes in the All-share index and in UK industrial production.

Trying to ascertain such certainties from such limited variables is the sort of managerialist mindset you often bleat about when you are not bleating about the bubble.

[aragon]

Policy makers cannot find their asses with both hands. So no chance of prophylactic policy changes.

Mainstream Economics doesn't describe the economy properly (unlike MMT).

The following should come as no surprise.
Private debt is unsustainable and when it gets too great it triggers a recession, of course in the UK the Government is still inflating asset prices.

http://fabiusmaximus.com/2016/02/02/nber-research-household-debt-slows-economic-growth-93697/

""The Great Recession was not an extreme outlier," they write, but "followed a pattern we would expect given the tremendous rise in global household debt that preceded it.""

A Minsky Moment?
What is the UK's debt to GDP?

http://touchstoneblog.org.uk/2015/02/uk-household-debt-still-amongst-the-highest-in-the-world/

120% and increasing in 2014 after the deleveraging.

"The financial crisis of 2008-09 is understood to have been driven by the unsustainability of this position; while the crisis has been arrested – up to a point – it is not clear that the underlying problem has been resolved."

http://www.theguardian.com/business/2016/feb/08/whats-holding-back-world-economy-joseph-e-stiglitz

"QE and low interest rates have disproportionately created wealth in the financial sector and inflated asset prices.
It has done little for the real economy. The rules of the market need to be rewritten."

What was that about not been able to predict recessions? What are the prospects for the UK?

[chris]

@ Matt - my chart shows IP in real time - eg November's data (released in Jan is correlated with November's share prices. Stock markets shouldn't react to formally-released data; they should reflect economic conditions as they are happening.
@ Aragon, Peter K - I fear we're at cross-purposes. It's not good enough to say that a bubble is unsustainable and will trigger a recession at some time. For practical purposes, we need to know when. Eg,US house prices were unsustainably high in 04-05 (they actually peaked in 06). But anyone who had bet then that this would cause falling prices of shares or mortgage securities would have lost money for many months - and would probably have been forced to close their positions.
It's not good enough to be right. You have to be right at the right time.

[Philip George]

When I look at your graph of share prices and production index, it seems apparent that the bottom of the share price line precedes the bottom of the production index. (You can see it in 2008-09 as well as in 2012-13)

The reason is that asset market crashes cause recessions. Surely it is not an accident that the Great Depression, Japan's Lost Decade, and the Great Recession all followed huge asset market crashes.

The arrow of causation is simple. Huge asset market crashes cause huge contractions in net worth of a large part of the population. They respond by cutting consumption in a bid to recoup the net worth. In turn companies respond by cutting output.

The larger the crash, the larger the cut in consumption and the more prolonged the cut. In the 2008-09 crash, for instance, the median US family lost 18 years of net worth. It would have to double its saving rate for 18 years in order to regain the lost net worth. That is why recoveries following asset market crashes are so long-drawn-out.

If you see the graph on http://www.philipji.com/item/2015-10-04/us-recessions-and-recoveries-over-55-years you will see an almost vertical drop in real consumption at roughly the start of every recession.

[chris]

@ Philip - There should be something in what you say, but how much?
One problem is that there's some evidence that stock market wealth effects are small:
http://www.econ.yale.edu/~shiller/pubs/p1181.pdf
I think the 00-03 slump in shares - which did not lead to big falls in output - is consistent with this.
But even if there were large wealth effects, why should they come as quickly as my chart implies? Given that share prices are noisy, and consumption formed by habits, you'd expect a longer time lag than there is.

[aragon]

Chris,

That depends on your purpose. Practical purposes - the big short?

As Joseph Stigliz argues, this is no way to run an economy, with the exception of the top 0.01% who benefit most from the status quo.

First you allude to Gordon Brown - "We have eliminated booms and busts! - except for the unpredictable ones. The Mea Culpa's from Greenspan (Fed Chair) and Mr White (OECD) There was a flaw in my understanding!

But your comment suggest you want to play the big short (as the book/film makes clear), well the banks play a rigged game, as does the state, so knowing when the next recession will occur won't help much.

Keynes said "the market can stay irrational longer than you can stay solvent". It is not a particularly useful observation.

When George Osbourne will consummate his suicide pact with the British Economy is unknown, but the writing is on the wall and in the law in the form of his Fiscal Charter.

Or will some other factor beat him to it, and trigger his political death? Economics is going to hell in a hand cart down so many roads.

As for US house prices, just how detached from reality and irrational the financial markets could get, before the house of cards collapsed? In the case of the Big Short it turns out, there was a trigger date. Reality is not normally so obliging.

I was not reading US Mortgage documents. Michael Burry, it would appear, was in a position to do so, and had access to Capital and Wall St.

[spencer]

I have a thesis that the consensus can not forecast a recession.

Recessions occur when the business community makes a mistake and over estimates demand.
as a consequence they end up with excess inventories they need to eliminate.

But business expectations are generally the consensus economic forecast. So for the consensus to forecast a recession the consensus has to forecast that the consensus is wrong.

Does not happen.

[Magnus]

@metatone so "some recessions are rather predictable" are they? Why don't you sell put options then? On the graph above you'd make a killing.

The only thing predictable here is the lazy caricaturing of mainstream economics.

[Matt]

Hi Chris - I was thinking more of the stock market as a forecasting tool, ie ahead of the official data?