In defence of technology shocks

In his attack (pdf) upon macroeconomic theory, Paul Romer is especially critical of the real business cycle view that recessions are caused by negative technology shocks. He calls them “phlogiston shocks” and says:

there is no microeconomic evidence for the negative phlogiston shocks that the model invokes nor any sensible theoretical interpretation of what a negative phlogiston shock would mean.

Simon accuses him of attacking a straw man, saying that “the insistence on productivity shocks as business cycle drivers is pretty dated.” And the standard undergraduate textbook, Carlin and Soskice’s Macroeconomics, says RBC theory “is not the mainstream view.”

Looking through a macro prism, such scepticism is reasonable. How can people forget how to do things? Yes, there can be intellectual regress (which is what Romer alleges of DSGE models!) but surely not at the frequency of business cycles.

However, if we ditch representative agent thinking and think instead of firms as being inherently heterogenous, the notion of a negative technology shock seems more reasonable.  

Xavier Gabaix points out that even in a large economy aggregate fluctuations can arise from the failure of one or two big firms. This is especially possible if those firms are important hubs, whose troubles plunge supplier or customer firms into trouble: as Daron Acemoglu shows, networks are crucial in transmitting (or dampening) firm-level shocks throughout the economy.

It seems to me that this is a plausible description of the financial crisis. Banks became less able to supply credit than we thought; this was a firm- or industry-level negative technology shock. And because banks were key hubs, this shock was transmitted to the wider economy.

You can squeeze this into DSGE-style models, as (for example) Michael Wickens (pdf) and Hashem Pesaran (pdf) have done. Whether you should do so, though, is another matter: I fear that such exercises might be examples of economists seeing something happening in the real world and wondering if it is possible in theory.

However, evidence on the importance of firm-level shocks is not confined to 2008-09. In Coping with Recession Paul Geroski and Paul Gregg estimate that just 10% of firms accounted for 85% of the drop in employment in the 1989-91 recession. It’s not clear whether this was due to an identifiable single technology shock or what. But this is strong evidence that heterogeneity matters. They wrote:

 Describing what happens during recessions using simple macroeconomic aggregates and representative firm models of the economy produces a seriously distorted picture of events. Recessions are about what happens to differences between firms much more than they are about what happens to firms on average.

Now, I’m not saying here that recessions are always and everywhere technological phenomena; recessions differ and maybe there’s no single general theory of what causes them. Nor does this view necessarily have policy implications. One could argue that whatever causes recessions, the response should be expansionary macro policy.  

There might, however, be one implication of this. If recessions are occasionally due to technology shocks, then they might be sometimes be inevitable and unpreventable by macro policy: the consistent failure of economists to forecast recessions is consistent with this (though not proof of it).

What I’m doing here is making a plea. In throwing out the bathwater of representative agents and the very dirty bathwater of equilibrium business cycles, economists should be careful to keep the baby of possible technology shocks.

Comments

[Dave Timoney]

I suppose you could argue that the rise and fall of Nokia, and the impact this had on Finland's economy, is an example of a technology shock.

[Luis Enrique]

I can't find the reference but I have long thought that a technology shock in econ model is actually quite close to a demand shock, which some people would find more palatable. A technology shock is something that makes combining inputs to create output more difficult. But in a Says law supply side model, you just assume whatever is made is sold, problematically. If on the other hand you start thinking about a demand shock as making it harder to make stuff and then *sell* it, well that's not too different from something that makes it harder to combine inputs and make output that is sold. I haven't explained that every well, but I am pretty sure I have seen high status econ profs argue something similar.

All these shocks are shorthand for something else. Something that one hopes is real but we treat as too hard to model explicitly. I agree with you that it is perfectly reasonable to think that things happen out there in the economy which can be thought of as hits to the economy, and perhaps reasonably captured in a model by a stochastic process.

(personally I find the next step - that the representative agent optimally reduces labor supply - much more objectionable)

the problem is that its hard to tell the difference between reasonable use of short hand for realistic things, and fantastical nonsense.

I could write down a model in which Godzilla rampages through the economy at random, destroying some proportion of the capital stock, and then use the data on recessions to identify the occurrence of rampages, because in my model the only way for recessions to occur is when the capital stock is destroyed. That's reductio ad absurdum, but I think the force of Romer's argumentis that mainstream macro models are more complicated versions of the above, and have done a bad job of separating out sensible shocks processes from Godzillas (and perhaps have not even really tried to look for alternative evidence/ ways of motivating the shocks that they use)

[aragon]

What are technological shocks?

Real Business Cycle vs Minsky Moments?

Shocks are a result of the failures of finance (Minsky). Speculation, Ponzi Schemes, and Fraud.
A sudden destruction of capital and withdrawl of credit.

http://www.economist.com/news/economics-brief/21702740-second-article-our-series-seminal-economic-ideas-looks-hyman-minskys

"Central bankers seem to agree. In a speech in 2009, before she became head of the Federal Reserve, Janet Yellen said Minsky’s work had “become required reading”. In a 2013 speech, made while he was governor of the Bank of England, Mervyn King agreed with Minsky’s view that stability in credit markets leads to exuberance and eventually to instability. Mark Carney, Lord King’s successor, has referred to Minsky moments on at least two occasions."

Speaking as a non-economist. No technology required.

Back to the dead (luddite) horse...
Economists are re-arrange the macro deck chairs on the Titanic.

The real Technological Shock, will re-write economics destroying capitalism and finance in the process, through massive technological unemployment. See the reverse luddite fallacy.

Economic Singularity (http://pandoras-brain.com/)

https://calumchace.wordpress.com/the-economic-singularity/

https://calumchace.wordpress.com/2016/08/29/bumps-on-the-road-to-the-economic-singularity/

"it is different this time because this time the machines are coming for our cognitive jobs."

https://calumchace.wordpress.com/2016/09/18/the-reverse-luddite-fallacy/

"Technological unemployment is not impossible. It is complacent and irresponsible to say that it is, based on no evidence beyond what has happened in the past. We should be discussing it and working out how to handle it if it does happen. Failure to do so is the Reverse Luddite Fallacy, and it could be extremely dangerous for all of us."

"Toto, I've a feeling we're not in Kansas anymore." - Dorothy in Wizard of Oz.

[aragon]

Sorry just reading Paul Mason.

Information Technology is the culprit, when Finance holds a smoking gun!

Yes, IT changed things, with the rise of Facebook, Google and Twitter, but not too fundamentally. This time it is different. Real progress has been made in Artificial Intelligence.

Post Real Economics is about to become plain Post Economics.

AI may still be narrow (No general AI equivalent to the human brain). But it is starting to extract understanding from natural language (NLU).

Through the application of Deep Neural Networks and Deep Learning developed in 2006.

This allow AI to replace jobs in Call Centres, Warehouse and Transport just the first wave (2021 onwards)

Millions and Millions unemployed with no prospect of replacement jobs.

Economists examine the growths between the railway sleepers while ignoring the oncoming train.

The economic singularity where economics collapses...

[aragon]

Make what you will of 'Peak Andy'

http://www.bankofengland.co.uk/publications/Pages/speeches/2015/864.aspx

[aragon]

The problem:

"At present, owners of capital have the whip hand – no longer literally, but legally – when it comes to running public companies."

Andy Haldane (See above)

Milton Friedman and the dumbest idea in history, all that matters is shareholder value, resulting in huge rent seeking by finance.

We see this with Philip Green and his billions on his yacht in the Med, while BHS collapses, workers discarded and the pension fund is rescued by the state.

Rent seeking is how the super rich remain super rich, as for Peak Andy, Economists jobs won't be taken by computers or robots, they will simply become irrelevant, in a post-economic society.

p.s.
As someone with a technical background (Technologist?) I have never placed any faith in the mathematical models, DGSE or otherwise, and the rules of the game are about to change radically. How we manage the transition is the question, and so far we are true to form (maximum chaos).

Pitchforks at the ready...

[reason]

REAL Business Cycle theory - and then you use the financial sector as your example (after most business cycle theory thinks that business cycles arise from money imbalances). Come on, you are pulling our collective legs here.

[reason]

" But this is strong evidence that heterogeneity matters."

Couldn't agree more, so why is this support for models with sector wide (representative agents) aggregation. This whole post is just silly.

[Luis Enrique]

reason

I think word 'real' is supposed to distinguish from monetary as in central banking stuff. If monetary policy is held constant but then, say, a bank collapses, that's more like a shock to the real economy, involved in the production and trade of goods and services. OK you can then think about how that feeds back to monetary variables, but I think way econ use word 'real', thinking of banks as part of the system of production, which explode. tech shock not so daft.

I could be wrong

[Vic Twente]

If you're defining the institutional role of the banking system as "technology", then really you're just saying "everything that's not K or L is 'technology'". That sounds very much like Romer's anonymous friend's description of Prescott's offhand definition of technology.

The assertion has close to zero informative value.

[B.L. Zebub]

@Chris,

"In his attack (pdf) upon macroeconomic theory, Paul Romer is especially critical of the real business cycle view that recessions are caused by negative technology shocks. He calls them 'phlogiston shocks"."

This was only one of Romer's criticisms. What about the others?

Concrete examples from Romer's paper:

• A troll who makes random changes to the wages paid to all workers
• A gremlin who makes random changes to the price of output
• Aether, which increases the risk preference of investors
• Caloric, which makes people want less leisure

I also notice that neither you nor Simon Wren-Lewis comment on the charge behind all these things: that they are unmeasurable and unobservable.

[Luis Enrique]

Vic

yes technology, at this level of simplification, is everything that determines how much output given combinations of K and L produce. Yes it is unobservable and only measurable as a residual if you accept a bunch of assumption.
http://personal.lse.ac.uk/casellif/papers/growthaccounting.pdf

note that Romer himself is famous for using technology at a similar level of abstraction

other than breaking production up into smaller chunks with numerous different 'technology' parameters of similar nature, I don't really know how would could ever model technology in way that makes it much more tangible. That problem is not escaped by using words rather than maths.

some of his trolls and gremlin are explicitly modelling 'tricks' that introduce something real and important into the model in a way that people can use - such a sticky prices - without ever being meant to be literal representations of what really happens.

I am not saying everything is fine and dandy with how these devices is used, merely arguing that the case for using them isn't as daft as it's made out to be.

[reason]

Luis Enrique
"yes technology, at this level of simplification, is everything that determines how much output given combinations of K and L produce"

Don't you mean
"... K and L CAN produce" - or are you saying that deficient demand is a technology shock? Come on I thought you were smarter than that.

[Luis Enrique]

well I am evidently not smart enough to write carefully enough for you in quickly written blog comments. Yes you are right, it's potential output.

[polymath]

"Technology shocks" are not only unobservable. They are also exogenous. They are the randomness in a random variable.

Models containing "technology shocks" had nothing to say about banks lending less. "The next recession will be a technological phenomenon" (much less that all are) says nothing about what the next recession will be like.

"Shocks" just a statement about the statistical properties of a set of observations. They're not real things in the theory. This is deadly to the theory, and if the field privileges that theory over other theories that fit the data it is deadly to the field. In that respect Romer is right.