Bryan Caplan  

Keynesianism vs. Pump Priming

Political Freedom and Economic... Me on KQED FM on Friday...
Keynesianism is widely seen as a "pump priming" rationale for government intervention.  The government sees the economy in the doldrums, gives it a much-needed jolt, and then the private economy gets back on feet - with no need for further assistance.

If you pay close attention to the simple Keynesian model, though, there's no pump priming to be found.  The model begins with the accounting identity Y = C + I + G.  In English, output equals consumption plus investment plus government purchases. 

The model then adds a simple consumption function:

C = a + bY, where a>0 and 0<b<1.  At low levels of consumption, people consume more than they earn; but when they earn an additional dollar, they consume some but not all of it.

Both I and G, finally, are exogenous.  I is whatever the animal spirits make it.  G is whatever the government wants it to be.  Substituting:

Y = (a + bY) + I + G


Y = (a + I + G)/(1-b)

(1-b) is the infamous "multiplier."  Every dollar of G boosts output by $1/(1-b).

Now take a close look at the final equation.  If you need to boost G to get actual output up to potential output, when can you cut G back to normal levels?  The answer, in the simple Keynesian model, is never.  Boosting G changes consumption, but not the consumption function.  And I, as always, is a product of animal spirits - not government policy.

You can naturally fiddle with the simple model to make it work the way you think it should.  The simplest fix: Assume investment is an increasing function of past G.  The problem is that it's hard to see why this assumption would be true.  It works if the extra G leads to the arrest of the Bolsheviks.  Stories relevant to modern economies are harder to tell.

Once your imagination starts running, in fact, it's easy to see why current I might be a negative function of past G.  Maybe higher G signals hostility to business - regime uncertainty, as Higgs puts it.  And if the G actually does something useful, it could depress the animal spirits by pre-empting the grand designs of private investors.

Believers in the simple Keynesian model could welcome my conclusion.  Perhaps they want a permanent increase in G.  Maybe this is what Keynes had in mind when he smiled favorably upon the "socialization of investment."  At the same time, though, the absence of a pump-priming effect limits the appeal of Keynesian remedies.  Keynesians can't honestly say that a temporary relaxation of free-market principles will get the economy firmly back on its feet.  Their solution for a sour economy is the permanent expansion of the state - nothing less.  And if that's all they've got to offer, most people will probably hold out and hope for a better solution to come along.
P.S. According to Investopedia, "pump priming" is another brain child of... Herbert Hoover:
The phrase originated with President Hoover's creation of the Reconstruction Finance Corporation (RFC) in 1932, which was designed to make loans to banks and industry. This was taken one step further by 1933, when President Roosevelt felt that pump-priming would be the only way for the economy to recover from the Great Depression. Through the RFC and other public works organizations, billions of dollars were spent "priming the pump" to encourage economic growth.
I might have known.

Comments and Sharing

COMMENTS (9 to date)
Steve Horwitz writes:

Bryan - Here's the evidence on Hoover and pump priming:

That's from a nice blog post by Jonathan Bean ( ). I have a Policy Brief type thing coming out from Cato that makes the case for Hoover being the father of the New Deal that uses that cartoon among other things.

Norman writes:

In the equation,

Y = (a + I + G)/(1-b),

the reason output would fall below potential is due to investor pessimism (animal spirits cause a drop in I). You could say the same is true of autonomous consumption 'a', but it amounts to the same thing. In this case, government would increase G only as long as investors remain pessimistic. That is, simple Keynesianism advocates an implicit government spending function where G is a function of I (well, technically deviations of I from the level that keeps output at potential).

Whenever investor optimism returns, Keynesianism says I can and should cut back. So while it's not pump priming, it's hardly correct to say it implies permanent increases in G (unless you want to make the case that Keynes thought of investment as a random walk). It's more like mandatory insurance.

All this sets aside the fact that most principles textbooks--Keynesians, all--includes interest rates for a reason.

Karl Smith writes:

Norman is right

A very traditionalist Keynesian might say welfare is maximized with G(I) where government spending is extra high when animal spirits are high AND extra low when animal spirits are high.

However, a more advanced version would suggest I(Y)so that in a strong economy animal spirits tend to rise. That's pump priming.

Sigve Indregard writes:

As one who has gone to a Norwegian university, and therefore have been fed a Keynes-heavy macro course, I find both your assertions of exogenous I and your slightly more advanced I = d + dG_(t-1) very odd. What we were taught, in the very simplest case, was I = c + dY, with the addendum that there's probably some lag.

I'll also add that in what is called a "simple Keynesian model", in the freshman year, we add the interest rates to this story, making I = c - di + eY, but you could be right about that not being part of a truly "simple" Keynesian model. I think, however, Keynes alludes to this.

Now, if you truly believe Keynesians argue only through the accounting equivalence, it makes no sense to discuss changes in any variable. But Keynesians discuss shocks: changes in parameters and variables. New technology, for instance, might shift c up or down. Without shocks, a Keynesian economy is rock stable.

With this addition, Keynesian stimulus and cutback achieves its stated target: to flatten the fluctuations of Y.

If you also make some rule for i, for instance that i is proportional to delta Y, it all comes together very nicely. However, I will agree that "original Keynes" included no such idea.

Sigve Indregard writes:

I'll also add that making Y = C + I + G and having I(Y) makes for very "interesting" mathematics for freshmen ;-)

Michael Lynn writes:

So the that phrase "priming the pump" in this context has to be a good 10-15 years or so earlier than investopedia is claiming. Here's a political cartoon with woodrow wilson priming the pump.

Chris Koresko writes:

This is the first time I've seen this formulation (I took a few econ courses in college, but no macro). I played with the equations a little, and concluded that this doesn't make sense.

For one thing, it seems wrong that consumption C is given as a function of output Y rather than of disposable income Y-G. After all, what a person spends ought to depend on the size of his paycheck, not his gross pay, correct?

If we have C = a + (Y-G)*b and solve the first equation Y = C+G+I for Y, we find, contra Keynes:

Y = (a+I)/(1-b) + G

i.e., the multiplier for government spending is unity, whereas for investment it is 1/(1-b), which is always greater than 1.

Is there any reasonable way to justify preferring Keynes' original formulation over this?

mick writes:

Y= C + I + G



C= -I-G+Y

By the powers of algebra 1 the larger G is the smaller C is. The garbage of this equation is not rescued by making C a complex variable.

John David Galt writes:

I agree with your doubts about the effect of G (and in any case, government "demand" is certainly at least partly waste due to inefficiency, since if the public wanted the services government generates enough to pay for them in the market, they would be done without a law to so require).

But to me, the interesting term in this calculation is (1-b), the marginal rate of saving by poor people. Keynes seems to be saying that if you minimize that savings rate and thus maximize b, you maximize the rate of growth. Indeed, if the poor can be induced to spend all their income, then (1-b) = 0 and Y = infinity. And this would seem to hold true even if the government discourages saving by destructive means (for instance, close all the banks, or increase inflation to 1970s-Latin-America rates so that no one will want to hold money for even an hour).

I wonder what Keynes would say if confronted by this reductio ad absurdum.

Comments for this entry have been closed
Return to top