AD curves don’t actually exist

Here’s Nick Rowe:

I got this idea from reading a Matt Rognlie comment on a previous post. (But Matt may or may not agree with what I say here).

A. Suppose the government sells bonds, and finances those bonds by imposing a 10% sales tax on (say) milk.

B. Suppose the government sells transferable milk quotas, and sells just enough milk quotas that milk prices rise 10% above marginal costs.

C. Suppose the government sells transferable local monopoly rights to sell milk, and those local monopoly rights cause milk prices to rise 10% above marginal costs.

A, B, and C are basically [weasel word] all the same.

.  .  .

If you think you understand how fiscal policy works, in a Keynesian or New Keynesian model, you should be able to see the equivalence. Though macroeconomists who believe it is nominal wages rather than nominal prices that are sticky, and think that the AD curve slopes down (e.g. Scott Sumner), will object that increased monopoly power raises P, which moves us the wrong way along the AD curve in a recession.

Because I’m a really shallow person, my brain cells all light up like a police car flashing light whenever I see my name mentioned in another blog.  Hence I need to comment on this.

Nick’s right that I view the AD curve as downward sloping, but I don’t quite like the way Nick phrased this.  It would be more accurate to say I think it’s useful to view AD as a rectangular hyperbola.  But I certainly don’t think Nick is wrong if he prefers something else, say a horizontal line.  I just don’t think it’s as useful.  In my view the AS/AD model is most useful when it tries to separate out nominal shocks (to M*V) and real shocks (which affect the composition of P*Y).

I should add that it’s not quite right to say that I think “nominal wages rather than nominal prices” are sticky.  I do acknowledge that there is substantial price stickiness, but I don’t view it as being very important.

Let me explain my focus on wages with a thought experiment.  Consider the sales tax increase discussed by Nick, and also consider two possible monetary policy targets, NGDP and nominal national income (NNI).  National income is GDP minus depreciation and indirect business taxes.  Thus a sales tax or VAT increase will drive a larger wedge between NGDP and NNI.  In that case, which should the central bank target?  In my view, the level of employment will track NNI/W much more closely than NGDP/W.  If wages are sticky, then a stable growth path for nominal national income will tend to stabilize employment.  If the central bank targets NGDP, then a rise in the VAT will reduce NNI, and with sticky wages this will also reduce employment.  Price stickiness might play a role in this process (by changing the labor share of NNI), but it would be a very minor role.

The preceding suggests that the most useful way of thinking about AD might be as a given level of NNI, not NGDP.  In that case, the vertical axis of the AS/AD diagram is still P, but the horizontal axis is now real national income, not real output. Or call it “real before-tax net output.”

As a practical matter, NGDP and NNI in the US track each other very closely over the business cycle.  Hence there is almost no difference between a monetary policy that targets 12-month forward NGDP growth at 4%, and one that targets 12 month forward expected NNI growth at 4%.  In other countries with national VATs, there may be a significant difference at times.

PS.  Last night I returned from a lot of travel with a bad cold.  I just feel like sleeping.  I do plan to start on Bernanke’s book today, and know that I have lots of catching up to do.  I have a new post at Econlog.

Tags:

This entry was posted on October 14th, 2015 and is filed under Monetary Theory, NGDP targeting. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response or Trackback from your own site.