« Who owns her body? | Main | Employment, Economic Comparisons and EU Trade »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Expectations are the only plausible channel that explains the leading properties of stick prices over real activity. The relationship is very strong even in periods (like the late 19th century) when a trivial fraction of households owned publicly traded equities.

The odd thing is, if you look at most macro model s, the only channel by which changes in equity prices affect real GDP is through the “wealth effect” on consumer spending. Even Tobin’s q gets left out of many of them. There is no such thing as “growth expectations” (or AD expectations or NGDP expectations) in most macro models. Although I suppose Bernanke and Woodford had models that included the “expectations-based IS curve”. But still it seems to be a very underutilized concept.

I guess macro models are built by People of the Concrete Steppes.

But in all seriousness Nick, why is the literature on growth (or AD or NGDP) expectations so tiny relative to the literature on inflation expectations?

Gregor: good comment, and very good question. "But in all seriousness Nick, why is the literature on growth (or AD or NGDP) expectations so tiny relative to the literature on inflation expectations?"

My guess: growth expectations pinned down by rational expectations plus the (unwarranted in NK) assumption that future output had to return to the natural rate regardless of monetary policy?

That's not a good answer to your question.

There are RGDP forecasts out there (I think). It shouldn't be too hard to construct implied NGDP forecasts too, from RGDP forecasts plus inflation forecasts, and for an econometrician to bung them into an empirical IS curve to see what happens. I don't know if any have done that.

Gregor: "There is no such thing as “growth expectations” (or AD expectations or NGDP expectations) in most macro models."

What do you mean? The entire DSGE literature (NK and RBC) incorporates growth, etc expectations. It's the static neoclassical models (IS/LM, AS/AD, etc) that fail to incorporate expectations. Where DSGE models fail, is not in failing to incorporate expectations, but rather in incorporating real-world features of the economy.

E.g. RBC fails to incorporate sources of market failure such as monopoly behaviour, sticky prices or unequal access to borrowing, without which there is no such thing as AD. NK models, on the other hand, fail to incorporate interesting growth models. As it stands, the failures of the former are vastly greater than those of the latter, but there is no reason why ideas from both can't be merged.

To me, the main failure in the modeling of expectations (both NK and RBC), is that the real world, rather than populated by agents with perfect rational expectations, is better described as made up of Bayesians with divergent priors in a world of unknown structural parameters. These agents may *rationally* and inevitably have different subjective expectations which would appear very much like Keynes' animal spirits.

Which brings me to Nick's point that "the whole idea behind Keynesian macroeconomics... is that prices... adjust slowly." Not so! That was the Neoclassical synthesis. Keynes talked about sticky prices, but his *main point* was that expectations can change drastically based on real world observations. Keynes' description is often described as basically irrational investor behaviour, but as  Martin Weitzman and others have demonstrated, even high information (Gaussian) priors may result in extremely fat-tailed posteriors, with extremely volatile moments, even in the face of an infinite number of observations. Real world measurements may be far less informative even than the ones in Weitzman's well controlled model.

K,

I think this is an occasion to distinguish between Keynesian economics and the economics of Keynes.

Nick,

Back on topic: Something is keeping the NK economy in general equilibrium at all times. To the extent that prices can't adjust, quantities do, so I basically agree with you. In the real economy, though, there is a question of which is stickier: wages or employment relationships. I don't see why either has to necessarily always "lead" the other. Also, both could be super-sticky, in which case there are other variables that would "give." Profits, for example.

W Peden,

I didn't mean to quibble about who said what. It was more about *what* is wrong, and *what* is right. In some ways it feels like a long, slow haul back to The General Theory. Of course, we've learned a ton along the way, and quantifying and solving properly micro-founded models has huge benefits. But I think it's striking that we are only just starting to glimpse the mathematical foundations of what Keynes correctly intuited some 80 years ago.

Lars Christensen has a good post on this. Basically, expectations will seem to be backward-looking when there is a lot of uncertainty as to what policy will be, and then only as long as the institutional framework doesn't change (which is consistent with Lucas' Critique).

K: I think Gregor is basically right (though I can't claim to be keeping up with the literature to know for sure).

Ever since the 1970's, there seems to have been oodles of *empirical* studies trying to see if the Phillips curve is forward-looking or backward-looking, or how much of each, then trying to get the theory to fit the econometrics. We've got the evidence on prices (both goods prices and asset prices).

I just don't remember any such vast literature *empirically* testing whether quantities are forward-looking or backward-looking. Maybe I missed it. Yet it seems just as do-able and just as important as whether prices are backward-looking or forward-looking.

Who else has an opinion and/or information on this?

K: "I don't see why either has to necessarily always "lead" the other. Also, both could be super-sticky, in which case there are other variables that would "give." Profits, for example."

Maybe. But even if there's "labour hoarding", Y will adjust when demand falls and P is sticky, even if L doesn't.

anon: Yes. It is a good post, as I re-read it. It has taken a long time for Scott's "long and variable leads" to slowly filter right through my brain.

"I think Gregor is basically right"

Right about what? He seemed to be claiming that the agents in most macro models don't form expectations about growth, AD, etc. That's just not true for NK/RBC models. I don't know of *any* DSGE model where agents *don't* form expectations of the future model dynamics. Usually model-consistent, perfect information rational expectations over all possible future states of the model.

Back in a couple of weeks. Have fun, play nice, keep a lid on things.

K,
Correct me if I'm wrong (I probably am) but in most NK DSGE models that central banks use, expectations of AD are essentially pinned down by inflation expectations as that is what most modern central banks target. The Phillips curve relationship takes you from weak AD to lower-than-target inflation to expectations of central bank easing. So you have inflation expectations and expectations of the future path of the policy instrument in the model. But you don't have expectations of AD (NGDP or RGDP) or growth per se. What if households and firms don't have the same Phillips curve relationship in mind that is embedded in the model? What if sometimes they expect a recovery in real output following a recession and sometimes they don't? Does it matter for the pace of the recovery? Can we measure what households are expecting not just for inflation but for real output as well?

Have you ever seen the data on household expectations of future income growth in the Michigan Consumer confidence series? In every other post-WWII recession that we’ve seen, expectations of future nominal income growth dip but then quickly recover – usually even before the real economy starts to recover. But in the current cycle it has stayed at its recession trough of roughly 0% for 3 and a half years. Prior to the recession it was 3-4%. At the same time, inflation expectations are roughly where they were prior to the recession. I'm trying to figure out what that means. I think it's very much related to the sluggishness of the recovery and the extremely low nominal interest rates that we're seeing.

My initial point (as Nic suggested) was that I don't know of any models that take in data like this as an input and use it to tell you something about the economic outlook. Do you?

Hmm, I don't quite agree. Wages are forward looking but looking at wages is backward looking. That's because expectations drive wages but wage statistics are compiled in a lagged fashion and there is price inertia. Still when prices are set it is done by expectations; rationality requires this.

Stock prices are also forward looking but whether looking at stock prices is forward or backward looking depends on the trading volume and liquidity. Unregistered securities are priced too infrequently for instance to be used as forward looking metric.

I'm puzzled. Where is the evidence that wages and prices are backward looking? It seems to me that wages and prices are forward looking, but move very sluggishly.

Take the example of mid-2008 in the US. During the prior 12 months inflation had been quite high, about 5%. On the other hand the Fed had a 2% inflation target. Wage increases negotiated at that time were quite modest, reflecting the Fed's promise to hold inflation down, not the actual experience of inflation over the previous 12 months. What am I missing?

I agree that a spurt in NGDP would be mostly real growth, but that's partly because I think wages are currently above equilibrium.

I would have to say, quantities are all forward looking, with forward curves, costs, and prices that are somewhat rigid in the near term. One asks the question how much to produce at forward prices/costs. It's hard to imagine a case where yesterday's quantities matter unless you have limited storage space (for example, a utility only has limited storage space for coal, so they might slow down deliveries after a mild summer/winter, but their desired inventory level is a function of the weather forecast and future prices). Even if you have a big pile of inventory at historical cost, its a sunk cost.

What is the difference between forward and backward looking quantities if people make forecasts that are heavily biased by extrapolating from the past?

As no one can forecast the future, aren't you just detecting the degree of dependence on lagged values? In that case, the most forward looking quantities would be the ones whose prices are the most flexible and/or whose lookback window is the smallest, so it makes sense that asset prices appear to be more forward looking than other prices, even though they may not be forward looking at all.

Couldn't Farmer's 'forward looking' stock market index just be a proxy for other financial activity, such as lending (which in the real world is a blend of hindsight and guessing about the future)? In which case it's just a question of which financial target(s) makes the most sense for a CB? (Granted, Roger might argue that it's a hugely important distinction, as policy should target wealth, not income: http://www.nber.org/papers/w17479).

More radically (post-Keynesian), perhaps stock markets reflect intra-sector balance sheet shifts that, if sufficient to disturb 'nominal' relationships, impact real activity? Thinking of the framework articulated in Godley and Lavoie's recently released book (though the second part of that is mine, so don't blame those two if it's an inane thing to say!). There's probably room for some Minskyan fragility in there too, no?

Art,

"More radically (post-Keynesian), perhaps stock markets reflect intra-sector balance sheet shifts that, if sufficient to disturb 'nominal' relationships, impact real activity?"

British Monetarists Tim Congdon and Gordon Pepper have been writing about such a hypothesis for decades. Tim Congdon once used an article to express how the impact of monetary policy (QE) on asset markets could boost the quantity of money without an increase in bank loans, because banks giving loans and banks buying corporate bonds both boost the money stock. It also is a monetarist theory of the business cycle where physical cash doesn't matter (though reserves do) and where banks need not have any direct interaction with the central bank at all.

So one has a theory of the business cycle which goes like this: macroeconomic policy + endogenous conditions ---> asset prices ---> the quantity of money ---> real activity ---> asset markets etc.

"There's probably room for some Minskyan fragility in there too, no?"

Indeed there is! Cyclical financial instability is the result of asset markets getting distorted. Tim Congdon's pet example is the Lawson boom, where the UK saw a major housing bubble (30% average price increases in 1988 and then a massive exodus from home-ownership from 1991-1995).

Congdon's article-

http://www.standpointmag.co.uk/the-unnecessary-recession-features-june-09-tim-congdon-gordon-brown-alistair-darling?page=0%2C0%2C0%2C0%2C0%2C0%2C0%2C0%2C0%2C0%2C0

Congdon on monetary causes of economic instability-

http://critical-reaction.co.uk/3010/05-02-2012-the-pathologies-of-capitalism

The debt-deflation hypothesis which Congdon mentions at the end is an early example of a monetarist analysis of the relationship between money, asset prices, nominal variables and real variables; it saddens me that monetarists since then have often forgotten about asset prices and just focused on nominal and real variables. Congdon's training was at Cambridge in the 1960s, which is why he and other British monetarists of that generation seem so much closer to Keynesians than their American colleagues.

(Congdon also addressed Wynne Godley's attribution of the stability of the post-war years to Keynesianism in "Keynes, the Keynesians and Monetarism".)

Charles Goodhart is another economist probably classifiable as a British monetarist (which perhaps shows how far British monetarism is from American monetarism) not least because he thinks that broad money aggregates should still play a significant role (though not like in the 1970s/1980s) in monetary policy, that money is neutral in the long-run and that changes in broad money are causally important. I imagine that MMT types will like this article and I hope that it dispels some myths about what monetarists are committed to-

http://www.boeckler.de/pdf/v_2008_10_31_goodhart.pdf

The comments to this entry are closed.

Search this site

  • Google

    WWW
    worthwhile.typepad.com
Blog powered by Typepad