April 2015

Sun Mon Tue Wed Thu Fri Sat
      1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30    
Blog powered by Typepad

« Why We Need the Liberal Arts | Main | Peter Lewin on Uncertainty »


TrackBack URL for this entry:

Listed below are links to weblogs that reference Keynesians stories do not imply Keynesian policies.:


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

Peter zeroes in on the essential weakness of Keynesian arguments for stabilization policy. If markets fail, why would politics succeed. How is that policmakers have information markets do not.

The competing visions of stabilisation policy have been defined by Franco Modiglani and Milton Friedman respectively.

• Modiglani considers the Keynesian vision of macroeconomic policy to be “a market economy is subject to fluctuations which need to be corrected, can be corrected, and therefore should be corrected.”

• Milton Friedman was far more circumspect: “The central problem is not designing a highly sensitive [monetary] instrument that offsets instability introduced by other factors [in the economy], but preventing monetary arrangements becoming a primary source of instability.”

It seems that macroeconomics needs to learn a lot from engineering control theory, where dealing with instabilities in feedback and feedforward control systems is a standard problem. A market system can be mathematically viewed as a control system, where the price signals generated by demand changes feed back to the production. The stability of the system will depend upon the normal time constants of the demand changes (mens clothes have a slower time constants that teenage girls clothes) and the supply response time. If the natural frequency of the demand changes is the same as or faster than the supply response time, the system will be unstable and can create a bubble and crash.

If someone put a delay time in the response of your car to steering wheel changes, you will tend to over-correct and crash. Just visualize an emergency that required a direction change in your car and the car didn't respond for a second or so, you would put in more direction change then the car would over-respond. You would then crash. This is why their is a computer between the control stick on some high performance fighter jets and the control surfaces -- to prevent these instabilities.

When the government adds to the supply response time via permit delays, restrictions, impact reports and hassles, you can shift from a stable system to an unstable system that will form a bubble and crash.

Often government action can actually create instability, like in the housing markets. Notice the areas of the country without zoning laws and lot of time delays didn't have a bubble and crash in housing.

Good piece.

I have written a couple of pieces along similar lines. Comments welcome.

Government spending is far stickier than wages or prices: http://www.insofisma.com/wp2/government-spending-far-stickier-than-wages-and-prices/

The long run is here. Keynes is dead. http://www.insofisma.com/wp2/the-long-run-is-here-keynes-is-dead/

Frank, I look forward to reading the posts at your links.

Deweaver, when I teach Milton Friedman on monetary policy I tell my students to imagine trying to steer a boat with a faulty rudder across a lake. The lag in response is long and variable (cannot be predicted) - the result is to amplify the fluctuations (weaving) and most likely miss the target on the other bank.

So far so good. But the analogy of a dynamic control problem only takes us so far. If we knew the number of markets, could specify all of the products involved, and could estimate the lags, even probabilistically, we could theoretically devise an optimizing algorithm - one that would probabilisitcally minimize the mean squared error or similar. In fact, we are not really dealing with this kind of problem. We are dealing with a much more complex system in which the number of variables is both large, unknown and continually changing. It is much more akin to a biological system than a mechanical one. It is complex and adaptive for some configurations of the enclosing institutions (Hayek). An evolutionary mindset is more accurate than a mechanical one. Either way, we may agree that discretionary macro-policy of the kind envisaged by Keynesians is impossible.

Deweaver -
This, of course, assumes that the macroeconomic problem is similar to the control system problem you describe. There's probably a good reason why engineers model their systems one way and economists model theirs another.

"Why should policy-makers know any more about the future than the agents in the field? And why should we trust them to want to mitigate the cycle, rather than exacerbate it to their advantage? "

The fundamental flaw in your argument is the belief that policy makers need to know *exactly* how consumers or business will react to AD expansion.

It is enough to know that the empirical evidence is overwhelming that businesses increase capital goods investments or build up stock in response to surges in demand for their products, and consumers, especially those at the lower end of the income scale, spend extra income on goods and services.

If I, as a policymaker, cut taxes to increase aggregate income, there is a perfectly good inductive arguments for what will happen and why it will be a successful stimulus for the economy.

If you don't think induction is justified, philosophically speaking, it is easy to reformulate one's prediction in terms of Popper's falsificationism by hypothetico-deduction:


And I have not even touched on the moral case for government support. Why should a high level of suffering - unemployment, poverty and hardship - be tolerated in the community when Austrians/libertarians can hardly be certain either that action by private agents in free markets will lead to a return to prosperity quickly and smoothly?

LK: State agents don't need to know "exactly" in order to do better, just "enough". The Hayekian case is that they not only don't know enough, but can't know enough. And that still ignores Public Choice.

As for morality -- don't go there. Putting peaceful people in prison for not agreeing to theft is the very foundation of the state, without which it does not even exist.

Lord Keynes: (I love your name). What BZ said. The moral case - all government action involves coercion and distortions (taxes have to be levied). This would seem to place a big burden of proof on the proponents to show that its benefits will outweigh the costs. I don't think the evidence supports this - not at all. But then history doesn't speak to us in one voice. No doubt you hear a different academic scribbler.

I don't think I implied that Keynesian macro-policy needs to know "exactly" the affects of AD in order to be successful - but it does need to know where to apply the AD, doesn't it? Unless you want to claim that all AD stimuluses are equal. On what basis?

And, yes, there is the formidable Public Choice problem - Buchanan-Wagner, Democracy in Deficit.

"The moral case - all government action involves coercion and distortions (taxes have to be levied)."

If you're appealing to Rothbard's natural rights ethics, that is one of the most shoddy and unconvincing ethical theories ever formulated:



Even Edward Feser (a libertarian himself) agrees that Rothbard's justification's natural rights are utterly flawed:


"I don't think I implied that Keynesian macro-policy needs to know "exactly" the affects of AD in order to be successful - but it does need to know where to apply the AD, doesn't it?"

Where to expand AD? On middle classes and lower income earners and businesses? - just how it's been done successfully many times in the past 65 years in many nations, for which there is overwhelming empirical evidence on how it works.

Lord Keynes: I was NOT referring to Rothbard or natural rights, simply to the undeniable fact that all government action involves coercion - it implies coercion (try refusing to pay your taxes). In order to justify it you have to justify the coercion. Not rocket science.

Where to expand AD? On middle classes and lower income earners and businesses? - just how it's been done successfully many times in the past 65 years in many nations, for which there is overwhelming empirical evidence on how it works.

You clearly don't understand the problem. The question was rhetorical meant to draw attention to the fact that not all AD spending is going to have the same effect - something that Keynesians clearly deny or ignore. There is a structure of production and consumption, output is not an amorphous, homogeneous glob. You think in terms of income classes. I was talking much more fundamentally.

In any case, even the Keynesians seem to admit that stimulus spending on cronies has a different effect from spending on food stamps.

I suppose what matters is that I believe neither increases employment (quite the opposite) and you do. But you still have to prove it, you have the burden - that was my point. [I don't see the overwhelming evidence you claim].

The comments to this entry are closed.