Monday, March 17, 2008

Are we already experiencing slower trend-growth?

What if the expansion of the long-run aggregate supply curve through productivity growth has slowed down already? Not right now, but a few years back, say around 2002. And say that the expansion of expenditure managed to keep on going thanks to a wealth bubble, deficits, and cheap credit; and that inflation managed to stay down thanks to cheap imports of both goods and inputs?

What are the policy implications?

Forget about whether you believe or not in an upward-sloping short-run aggregate supply, whether you're a New Keynesian, an RBC-er, or a microeconomist playing the macro game (like yours truly); if the expansion of the long-run AS curve has slowed down, then we should bite the bullet and accept that those GDP-growth numbers will be slower for a while and be grateful if we avoid inflation catching up and staying with us.

And we should definitely be wary of policy measures, whether monetary or fiscal, attempting to return us to the growth rates of the 1990s: these would instead quickly morph into inflation boosters with little to show for real income (and with the usual redistribution effects).

Oh! But isn't that what we've been seeing since January?

My "evidence" for the trend-productivity slowdown since 2002: A short while ago, I blogged around this graph by Robert Gordon, showing trend productivity dropping since perhaps 2002; EconWeekly has a similar one, together with a lot more analysis on the sources of growth and an explanation on the methodology.

Again, some basic dynamic (and expectations-augmented) ADAS will help here. This is a graphical representation of the expansion of the (static) aggregate demand and aggregate supply curves (represented by their "dotted" counterparts), with inflation in the vertical axis and GDP growth in the horizontal (and all the hand-waving this requires).

In the graph below, we start from point A, the long-run equilibrium. We shock the system so that the dynamic long run aggregate supply growth curve (LRAS) contracts (to LRAS'): point A is no longer a long run equilibrium and long-term growth will have to slow down to that given by LRAS'.

I am arguing we were in a point like A until the very end of 2007, with inflationary and "slow-down" pressures being kept at bay by the factors mentioned above (bubbles, deficits, cheap credit, cheap imports) and perhaps also some of the usual rigidities/inertias. But we cannot stay out of long-run equilibrium in the, well, long-run.

How will the system return to equilibrium if policy cannot magically increase long-term growth to what it was before? Suppose we let the market economy work its impersonal magic:

  1. This overheated economy puts pressure on input prices. When this pressure finally gets passed onto input prices, the dotted AS curve contracts and slides along the AD curve, causing inflation to pick up just as the growth rate decreases. This seems like what we've been seeing over the last few months.
  2. At the same time, expenditure could contract not just because of price increases for consumption and investment goods (which cause a movement along the AD curve), but because of the decrease in wealth produced by the bursting of the housing-bubble and by the collapse of credit markets.

This is shown in the graph below and the move from point A to point B:

Of course, the curves don't need to contract by the same amount. We might end up with higher or lower inflation in B than in A depending on whether aggregate expenditure/demand or aggregate production/supply contract faster. Keep in mind that growth will be given, in the end, by the position of the dynamic LRAS curve and is the same whether the adjustment happens with any mix of SRAS or AD contraction; the only difference is how high inflation will be.

For example, say that the adjustment happens with NO contraction of the AD curve, but only SRAS slowdown. In that case, the AS curve must contract more and the equilibrium will "slide" up the AD curve until they meet over the LRAS curve; the end result will be the same growth rate with higher inflation and likely a longer adjustment time.

As an aside: should we care about inflation being higher or lower? In this model, not really; but if the LRAS curve has a negative slope because of all the inefficiencies inflation brings on the price system (plus its distributional issues), then higher inflation leads to lower the long-term growth. I will talk a little more about this later.

What would expansionary policies achieve in this case? In the tamest case, expansionary policies like those given by the Fed and the White House/Congress since January will achieve what was described above as "keeping the AD curve in place", forcing the adjustment to take place solely through the SRAS.

This could be the case if policy makers have accepted that the economy has slowed down so that the discussion hinges only around the speed of adjustment and what tolerable inflation is, not about the growth (and employment) level.

I'm not sure what the advantage of that would be beyond showing voters that the government is "doing something."

What happens if we have not accepted this slowdown and instead try to keep returning to the previous level of long-term growth through expansionary policies? That is when things get scary.

We start again from point A. Rising input prices contract the SRAS curve; however, as it begins to contract, expansionary policies expand the AD curve so that we move to point B: same short-term growth, but with higher inflation.

Point B becomes the new point A for the "next" period! In this "next" period, policy either sends another expansionary boost to the system (to counter the continuing contraction of the SRAS as input prices keep rising), thus moving the economy from B to a point above it; or the contraction finally begins then, the same thing but delayed.

But is it really the same thing? No, it's even worse: the contraction would begin from a higher inflation level. When we finally make it to the long-term equilibrium, we will have the same growth rate as in the previous (pure market adjustment) scenario but with higher inflation.

No, no: it really is even worse! To keep the short-run equilibrium above the long-run equilibrium, policy must be unexpected. The economy is already expecting last period's inflation and that is precisely what will contract the SRAS curve: real prices returning to equilibrium. Policy must surprise contracts, catch them unaware, with an even higher inflation to keep the real price of production inputs (including real wages) below equilibrium levels. That's what makes producing above the long-run equilibrium possible. In other words: inflation must spiral upwards at an accelerating pace to keep us producing above the long-run equilibrium.

(Of course, at this point, we need to tip our hats in Milton Friedman's direction: inflation is a monetary phenomenon. Fiscal policy can result in price shocks, but a process of sustained price increases must feed from the expansion of the money aggregates. Appropriately for this argument, it's the Fed that's doing most of the heavy expansionary lifting these days.)

One might ask: but can't the government simply keep inflating the economy to keep us moving vertically from A to B (to C to D) each period? Of course not: economics would not be called "the dismal science" if that was the case!

What is not captured in this simple model is how the underlying mechanics of the system begin to break down as inflation accelerates (the negative slope of the LRAS referred to above). Eventually, the real side of the economy, production and income, enters free-fall: negotiating contracts (whether for goods, labor, or credit) becomes the art of divining future inflation and how to compensate for it, since the higher the inflation rate, the greater the penalty for failing to adequately account for it. In other words, transaction costs go up and we all end up spending more time trying to figure out how to work our way around inflation than in being productive. And the poor get harder hit as more of their wealth is kept in assets closer to cash: inflation is a tax on money holders.

So eventually, keeping us above long-term growth is unsustainable and we return to the long-run growth at a higher inflation level. Which, when we add detail to the model, actually means lower long-term growth than otherwise.

And once inflationary expectations settle in, they become little nasty self-fulfilling prophecies. To get rid of them, policy makers need to create recessions until they are beaten down like whack-a-moles through slower growth. A government-induced slowdown like in the early 80s: not nice at all.

The morale of the story: If productivity growth is slower, so is long-term growth; accepting this means letting short-term growth slow down and prices adjust; not doing so means entering a losing fight which would only result in higher inflation and, very possibly, even slower future growth.

UPDATE: Gabriel points out that total factor productivity (TFP) is endogenous and thus not policy-independent. Although I completely agree with him (since I dare call myself an institutional economist, I can't do otherwise!), I don't think this affects my argument unless the expansionary policies implemented to fight the slowdown also increase TFP. While I don't have any particular endogenous-growth model in mind, I don't think that what the Fed or the White House/Congress are doing these days would be among the arguments in one. So with that caveat, I'll still stand behind my post... for now.


Gabriel said...

Productivity (TFP) is a black box. You can't really take it as exogenous, since various stuff can forecast the Solow residual to a certain extent (e.g. military expenditure, Hall has a paper).

What you need is a theory of TFP, which might highlight some policy causes for the productivity slowdown, it might point to a slowdown in research, etc.

P.S. People still call themselves "New Keynesians" and "RBC-ers"?

ram said...


Excellent point! You are totally right: TFP is endogenous and treating the Solow residual as exogenous is so ... so last century. As the institutional economist that I claim to be, shame-on-me for not pointing it out!

However, I don't think this affects the argument as long as the expansionary policies implemented to fight the slowdown do not influence the factors affecting TFP.

And as always: thanks for the comments.

p.s. I think they still call themselves that... But at the end of the day, the macros I've met, whatever they call themselves, seem to base their analysis in one of those two schools to the exclusion of the other. (Well, there are some monetarists left around, true.) The constraints imposed by paper publishing, methinks!

Gabriel said...

Right, right, I was not paying attention before.

It also dawns on me that, technically, you don't need a sloped SRAS curve, because with a "transactional need"-based money demand, a productivity drop with lead to a drop in money demand and from there, with constant or expansionary money supply, you get inflation.

Anyways, with economies with the natural rate property, the eventual outcome is the same, in terms of output, and inflation is the only thing monetary authorities control, but this doesn't mean that they shouldn't try to make *some* part of the adjustment be in inflation rather than output.