Thursday, August 04, 2011

Supply-side vs. demand-side recessions

Tyler Cowen links to an interesting post about the deflationary impact of the internet. This is interesting not just because it illustrates how technological change impacts the economy, but because it tells us something about recessions. Namely, it tells us that our current recession - like all of the other ones in recent memory - was caused by deficiencies in demand, not in supply. 

Does output falter because people don't want to buy as much stuff, or because we become unable to make as much stuff as we used to? This is a debate that has gripped the macroeconomics profession for many decades. Which is kind of surprising to me, because it is so obvious that demand shocks are the culprit.

The reason is prices. If recessions are caused by negative supply shocks, then we should see falling output accompanied by rising prices (inflation). If recessions are caused by negative demand shocks, we should see falling output accompanied by falling prices (disinflation or deflation).

Draw a supply-demand curve for the whole economy, and it looks like this:

A negative shock to supply - for example, a resource shortage or an increase in harmful government regulation or a "negative technology shock" - will shift either the long-run aggregate supply curve or the short-run aggregate supply curve (or both) to the left. The new equilibrium will have lower output and higher prices. However, a negative shock to demand - for example, an increase in the demand for money - will shift the aggregate demand curve to the left; the new equilibrium will have lower output and lower prices.

In all of the recent recessions, faltering output has been accompanied by lower, or even negative, inflation. This means that demand shocks must have been the culprit. If "uncertainty about government policy" were really the cause of the recession, as many conservatives claim, then we would have seen prices rise - as companies grew less willing to make the stuff that people wanted, stuff would become more scarce, and people would bid up the prices (dipping into their savings to do so). I.e, we would have seen inflation. But we didn't see inflation.

So it seems that the stories that conservatives tell about the recession - "policy uncertainty," "recalculation," or even a "negative shock to financial technology" - are not true. The stories that everyone else tells about the recession - "a flight to quality," "increased demand for safe assets," etc. - look much more like what basic introductory macroeconomics would predict.

Of course, conservatives don't necessarily believe in the graph I included above. They may dismiss the idea of a downward-sloping AD curve (and an upward-sloping SRAS curve), and imagine that the AD curve is just a flat line (or maybe even an upward-sloping line?). In that world, supply changes, and only supply changes, determine the level of output, and demand changes, and only demand changes, determine the price level.

That's where the blog post that Cowen linked to comes in. In a world in which deflationary recessions (such as our current one) are caused by negative supply shocks, then positive supply shocks should also leave prices unchanged. In particular, an improvement in technology would cause us to produce more stuff at the same price, rather than more stuff at lower prices. However, if the internet is deflationary, it shows that positive supply shocks do, in fact, decrease prices. The AD curve slopes down, as it should.

The basic point here is about the dangers of doing what Larry Summers calls "price-free analysis". If you ignore prices, it is possible to convince yourself that recessions are caused by technology getting worse, or by people taking a spontaneous vacation, or by Barack Obama being a scary socialist. If you pay attention to prices - as all economists should - it becomes harder to believe in these things.

So the question is: Do conservative-leaning economists push these stories because they believe that we live in a world that is vastly more complicated than anything that can be described in Econ 102? Or is it just because they choose to ignore Econ 102 completely?


  1. If the aggregate demand curve takes into account the central bank's reaction function, then its slope will be determined by the central bank's policy goals, except when the bank makes a mistake or faces a constraint in its pursuit of those goals. In the case of the Fed, given its dual mandate, the policy goals might result in a downward-sloping AD curve regardless of the economy's internal dynamics. Thus the observation that technical advances drive down prices might only reflect the Fed's preference for lower prices when output is high.

  2. Anonymous8:54 AM

    Noah, I totally agree with the argument. This is a balance sheet recession that has shifted the AD curve downward. However, the fact that the AD curve may be upward sloping is not conservative thinking. In their paper "Debt, Deleveraging, and the Liquidity Trap", Gauti Eggertson and Paul Krugman actually show that the AD curve could be upward sloping when caught in a liquidity trap and that could cause perverse effects in the economy...

  3. Anonymous11:44 AM

    I agree with Andy Harless that the central bank's policy reaction matters. Indeed, if the CB can target expected inflation or the price level path, the relevant AD curve is flat and only AS matters. So, in a sense, every recession is a supply-side recession.

  4. Alex Tabarrok1:24 PM

    Of course - but it's not so obvious that prices are pro-cylical. Use an HP filter and you can get a different result. There is a literature on this!

  5. I know little to nothing about economics, but re the conservative argument about "policy uncertainty", a couple of weeks ago on NBC's Meet-the-Press, they had an exec from Honeywell. One of the conservative guests asked him, leadingly,"Wouldn't it help you if you had lower taxes and less government regulations?" Of course he agreed, but one of the more liberal guests followed up with words to the effect of "Of course everyone would agree to lower taxes and regulations, but isn't the real problem lack of demand?" and the Honeywell guy agreed that this was the real problem!

  6. @Alex:

    I am perusing that literature now. As I would expect, there seem to be some "regime changes" at work...recessions in the 60s and 70s seem to have been driven more by supply-side factors, while the Great Depression and the recessions since 1990 had strongly procyclical inflation.

    The point is not that supply shocks have never mattered. But the 70s are long gone. No filter on Earth is going to tell us that inflation has gone up since 2008...

  7. It's no uncommon to hear people insist that we are seeing inflation (commodities!), we've just lost it in how we measure it these days.

    Thus as long as you truly believe in inflation (as some seem to do), then there's nothing inconsistent about the supply-side story (except facts).

  8. But can't policy uncertainty crimp consumer spending?

  9. Anonymous3:15 PM

    I guess one question I have is why "uncertainty about government policy" can only be a supply shock and not a demand shock. Sure, if it plays out as businesses producing less, then I see the how it's a supply shock. But what if it plays out as businesses investing less? Isn't that (at least in the short term) a demand shock?

    For instance, if the uncertain policy was whether or not the tax deduction for corporate jets will survive, won't that paralyze potential purchasers of corporate jets? Isn't that a demand shock?

  10. Anonymous3:26 PM

    This is exactly the analysis that my macro professor used to teach the Keynesian/IS-LM model except he used two sets of insecting graphs that all lined up with one another to "build" the aggregate supply and aggregate demand curves. There were about 10 or 12 graphs in total, arranged so you could track any exogenous change in one market straight through to its effect on output. You could see how Keynes' "lack of a single supply curves for labour" or the liquidity trap would change the model and the ultimate outcome. The prof handed out photocopied sheets with labeled axes and blank contents so we could play out any scenario discussed in class on paper as he followed along on an overhead projector. It was a great teaching method.

    My favourite part was that our textbook featured--in two different chapters--1) a chart of unemployment and inflation over about thirty years and 2) a table of productivity growth over the same period. These two pages of a simple macro textbook explained thirty years of debate among economists in about five minutes.

    In the mid-'70s you saw productivity growth drop (a leftward shift in the supply curve) and simulatneously saw both unemployment and inflation rising at the same time. This "stagflation", of course, seemed to contradict the Keynesian Phillips curve and help revive the anti-Keynes right.

    From the late '70s to early '90s you saw productivity growth remain relatively constant coinciding with a twenty year period in which the Phillips curve held--unemployment decreasing when inflation rose and vis versa.

    Then suddenly a period in the mid- to late-'90s when the Phillips curve once again seemed to cease to apply as unemployment decreased at the same time that inflation was trending lower as well. Of course this was just the opposite of what we had seen in the mid-'70s, but with the supply curve shifting rightward as the use of the internet and personal computing became expanded rather than the supply curve moving leftward after the oil price shocks of 1973.

  11. Anonymous3:40 PM

    "For instance, if the uncertain policy was whether or not the tax deduction for corporate jets will survive, won't that paralyze potential purchasers of corporate jets?"

    No. If the debate is about whether those taxes might go up, then the uncertainty would either have no effect or cause people to be more liekly to buy a corporate jet now before prices go up. But that is a good question.

    On the demand/investment issue, it's true that a decrease in expected future profits from investing in capital can lead to a decrease in demand, but the people saying that "uncertainty" is the issue generally suggest that it's a problem because it leads to reduced supply.

  12. Anonymous3:53 PM

    But can't policy uncertainty crimp consumer spending?

    Depending on what you consider plausible. I'd tend to say no - not in any nuanced way.

    Imagine a public knowledge poll about tax brackets, write-offs and the state of debate in Washington. Think about current events polls you've encountered in the past - you know the results aren't going to be pretty. Policy uncertainty can't have much of an effect on a state of ignorance.

    What (probably) really matters for consumption decisions is whether people have jobs, if the chatter at work suggests they'll get to keep that job, and whether the little financial and political news they get by osmosis sounds happy or sad.

  13. Anonymous4:09 PM

    Hmmm, actually I think the loss of the tax deduction for corporate jets in the future would lead to business buying fewer now. The uncertain policy I was alluding to was actually accelerated depreciation for corporate jets. If I was about to buy a jet, thinking I could depreciate it quickly and then found out that they might change that tax rule and require standard depreciation? Well, I might think twice. Ergo, drop in demand for corporate jets.

    Furthermore, policy uncertainty (how much will employee health care cost me next year) can lead to less investing now as businesses feel compelled to budget for the worst case....


  14. Anonymous4:54 PM

    "If I was about to buy a jet, thinking I could depreciate it quickly and then found out that they might change that tax rule and require standard depreciation?"

    I see what you mean. It would depend on whether the new depreciation schedule applies only to those who buy corporate jets after the new law is passed or applies to those who bought them before as well. If it's the latter, that could result in a decrease in demand, but only to that that uncertainty causes people to just hold on to their money RATHER than spend or invest it elsewhere.

  15. Anonymous10:21 PM

    Is all this as true when the recession is the result of central bank actions in the first place (as has been the case for many post war US recessions)?

  16. Jay said...

    " But can't policy uncertainty crimp consumer spending?"

    Yes, but:

    1) How much? Is it an effect lost in the noise, or is it the dominant effect?

    2) Do we have evidence of a massive drop in demand caused by known factors? (yes)

  17. There are two kinds of people: those who believe there has been low inflation, and those that have actually paid bills for the past two years.

    Which are you, Noah?

  18. Anonymous6:33 AM

    thank you for a most excellent post!

  19. "Do conservative-leaning economists push these stories because they believe that we live in a world that is vastly more complicated than anything that can be described in Econ 102"

    I believe they push those stories because they are focusing on the supply-side shocks that came before, and caused , the demand-side chnages.

    Unless one believe that demand for money suddenly increases for no reason then this view makes sense.

    Further - I think we need a focus on the supply-side factors that prevent the economy from recovering quickly from supply-side shocks. Addressing the demand-side by playing with AD only solves the short-term problem.

  20. Geez, no sooner do I post that you don't see AS-AD in the econ blogosphere -- and even that there would be snickering if you did -- you come along, Noah, and wave one right in front of me.

    I've thought about this very intensively, looked at it from every angle, and I see only one possible conclusion: Noahpinion is not part of the econ blogosphere!

    There! Now everything is in order again.