DeLong on Deficits
In a recent blog post, UC Berkeley economist Brad DeLong made an arithmetic case that, "We need bigger deficits now!" As an Austrian economist I naturally find his conclusion horribly mistaken, but it will still be useful to go through DeLong's argument and identify exactly where he goes astray.
"Normal" Economic Arithmetic
To his credit, DeLong has always been careful to restrict the case for deficit "stimulus" to cases of economic depression. In other words, DeLong agrees with non-Keynesians that under normal circumstances, government budget deficits per se do not boost economic growth and in fact are generally counterproductive unless the specific items being purchased are socially valuable. DeLong writes,
In normal times, a boost to government purchases or a cut in taxes produces a limited increase in production and employment while adding a substantial increase to the national debt. The increase in debt raises interest rates, which crowds out productivity-increasing private investment spending and, dollar for dollar, leaves us poorer after the effect of the stimulus ebbs.
The borrowing must then be financed at a significant interest rate, and thus paid for with higher taxes, which reduce incomes by increasing the wedge between the private rewards and the social benefits of expanded production.
It's nasty business.
If nothing else, then, it's good that opponents of government deficits now have the above quotation in their back pocket. Once the economy gets out of the so-called "liquidity trap" (as the Keynesians define it) at least we will have the authority of Brad DeLong to back up our calls for immediate cuts in government spending.
Unfortunately, even DeLong's explanation for why deficits are bad rests on dubious economics, at least from an Austrian viewpoint. Let's walk through his arithmetical example:
Consider a $100 billion boost to government purchases or cut in taxes financed by borrowing from abroad. In a normal year, the Federal Reserve will worry about inflation, and raise interest rates somewhat to offset the inflationary impact of the fiscal boost. The multiplier will thus be something like 0.4 — we will spend $100 billion on government purchases or tax cuts and gain perhaps $40 billion in extra production and associated employment out of it.
Already DeLong's on shaky ground. Here and in the remainder of his treatment of this example, DeLong doesn't distinguish between a spending increase and a tax cut. In other words, DeLong thinks that in normal economic times, the nation gains "$40 billion in extra production and associated employment" out of an extra $100 billion in government spending or a $100-billion tax cut.
Yet surely from the standpoint of the actual welfare of the people, it makes a huge difference whether taxpayers get to retain $100 billion more of their money, as opposed to the government spending an additional $100 billion. Regardless of how the government's expenditures are financed, they necessarily divert real resources (labor, steel, gasoline, etc.) away from private-sector uses and into politically chosen projects.
Although we can't add up the amount of subjective happiness among different people, there is a definite sense in which 50,000,000 households each spending $2,000 more on projects they themselves choose represents more valuable economic output than if that same $100 billion is spent by bureaucrats.
And notice the difference isn't that government bureaucrats are a lower form of life than private-sector individuals. From a neutral economic analysis, handing the $100 billion to the bureaucrats as their salaries would allow them to buy "$100 billion worth" of goods and services. But in their capacity as bureaucrats, purchasing items on behalf of the US government, the incentives are different. There is no longer a presumption that money is only spent on things that (to the purchaser) are worth more than the money being spent."If people decide to save less, and the interest rate rises, that isn't "bad" but simply announces the new situation. In the same way, if people decide to buy fewer potato chips and more tofu, then market prices need to adjust to communicate this fact to the relevant people."
Let us return to DeLong. After explaining the benefits of a $100-billion increase in the federal government's budget deficit, he begins listing the costs of the action. The first two are straightforward enough (the increase in the national debt and the supply-side disincentives of raising tax rates to pay for the higher interest payments). But I am troubled by DeLong's treatment of the third problem with higher deficits:
Third, there is another effect. The Federal Reserve's fight against inflation, which demands an increase in interest rates, will have reduced investment. Due to the $100 billion in government purchases, perhaps $40 billion of private investment that would have been made won't be made — it will be crowded out. As a result of the lower capital stock, some $4 billion a year of income that would have been earned won't be. Thus the net cost of the $100 billion in government spending will be a reduction in Americans' disposable incomes of $7.4 billion per year.
In the first place, notice how DeLong gets the (standard) result that a bigger government deficit leads to higher interest rates. He doesn't simply say that there is a market for loanable funds, and if the government enters that market trying to borrow more money, then the price of borrowing (i.e. the interest rate) will go up.
No, DeLong goes through a black box mechanism by which the government's extra borrowing will (a) start to push up price inflation, which will then (b) cause the Fed to raise interest rates.
This is part of the problem with Keynesian economics. It's true, in certain circumstances it can spit out an answer that is reasonable — such as this case, where DeLong agrees that a government deficit would not be justified. But if the analysis itself is crude, then we shouldn't be surprised that often the Keynesian framework spits out the wrong answer — which we'll see in the next section.
Specifically, the problem with DeLong's framework is that he treats interest rates as something that the Fed can control at will, and which function mainly as a governor on price inflation. This is so standard now in economics that it sounds uncontroversial.
To see why this is a strange way to proceed, let's tweak the example. Suppose the government spends the $100 billion specifically on oil to add to the Strategic Petroleum Reserve. Then DeLong might say, "Because of the government's massive new purchases, production and employment in the oil industry will expand, which adds to economic growth. However, to counteract the rising price of gasoline for motorists, the Fed would begin a massive purchase of wheat futures in order to drive up the price of bread. The higher price of food would force consumers to scale back their purchases of gasoline, thus lowering the price spike in that sector and partially offsetting the government's stimulus."
Incidentally, don't spend too much time trying to make sense of the above "logic." There isn't any, and that's my point — this is how it should sound to us when someone says, "By having the government borrow more money, this increases total spending but starts to push up prices, so we need the Fed to raise interest rates to reduce total spending and take the pressure off prices."
One final observation on this issue: In and of itself, a government deficit is not (price) inflationary. When the Treasury sells $100 billion in new bonds and spends the money, that means there are $100 billion less in the pockets of the private sector. In the same way, if I borrow $100 from my friend to go buy some groceries, this action doesn't push up "the price level." I have $100 more to spend, but my friend has $100 less.
Now it's true, there is a grain of truth in DeLong's analysis. An Austrian might say something like this: When the Treasury runs a higher deficit, this increases the demand for loanable funds and would tend to increase the interest rate. If the Fed doesn't want interest rates to rise, it can "monetize" the new debt by creating money out of thin air and purchasing outstanding Treasury bonds, thus increasing the supply of "savings" to perfectly offset the government's increased demand for savings. The new money of course will tend to cause price inflation.
So in this (Austrian) framework, it's true that if the Treasury increases its borrowing, and the Fed doesn't want price inflation to result, then it will allow interest rates to rise, by refraining from creating new money. At a certain level, this explanation is similar to DeLong's, in which the government's borrowing will push up price inflation unless the Fed steps in to raise interest rates.
Even so, the actual mechanics are different in the two stories, and it's important to understand what's really going on. In the next section we'll see what can go wrong when using the Keynesian framework.
"Depression" Economic Arithmetic
After conceding that deficits are no panacea under normal circumstances, DeLong claims,
But right now that arithmetic doesn't apply. We have instead depression economic arithmetic. And in depression economic arithmetic things are very different.
First, more government spending does not lead the Federal Reserve to raise interest rates to fight inflation. The Federal Reserve has pushed interest rates to the floor and wishes it could drive them into the basement — to −5 percent per year or so. Thus the multiplier on the government's spending is not 0.4 but more like 1.5. We do not get $40 billion of additional production and employment for $100 billion; we get something closer to $150 billion. And there is no crowding out of private investment; on the contrary, there is likely to be crowding in. If one-third of the extra production flows through to corporate profits, and if one-quarter of those extra corporate profits are then invested in projects paying a pretax net real rate of return of 8 percent per year, then future productivity is boosted to the tune of $1 billion a year.
Here now we see the huge problem in treating interest rates not as a price helping to coordinate consumer preferences and producer decisions in an intertemporal framework, but rather as a lever with which to accelerate or brake "the economy." Again, to see just how weird this is, suppose instead of using interest rates we used fuel prices. After all, wouldn't it make most people feel richer, and wouldn't it stimulate most businesses, if the Fed could push oil down to $1 a barrel right now? (It's true, this might devastate some people associated with the oil industry, but 0 percent interest rates devastate people with savings right now.)"It's pointless to try to wade hip-deep into DeLong's calculations, because they are meaningless."
Yet most economists would (I hope!) realize that it would massively distort the economy if the Fed were to somehow push down oil prices. (The Fed could do this, at least for a while, by buying oil at the spot price and then selling it to American distributors at $1 per barrel. The loss would be no problem, because the Fed can create dollars out of thin air. It's true, this might seem like a reckless policy to "stimulate" the economy, but then again so is this.)
It's pointless to try to wade hip-deep into DeLong's calculations, because they are meaningless. He is talking about dollars and cents without any reference to the actual production of real goods and services. This emphasis isn't due to sloppiness on DeLong's part; it merely flows from his reliance on a Keynesian model.
In an earlier article I did point out specific problems with (a different example of) DeLong's stimulus accounting. Yet in this case, it seems to me such an effort is wasted. Even if DeLong's argument is internally consistent, it still rests on quicksand from an Austrian point of view.
Although he juggles several different factors, the crux of DeLong's argument seems to be this: during normal times, government deficit spending pushes up the interest rate (unless the Fed monetizes it, which would cause price inflation). So any government attempt to stimulate aggregate demand leads to higher interest rates, which reduce aggregate demand and thus mostly offset the benefits of the stimulus.
In contrast, during a depression, government deficit spending does not push up interest rates, because they are going to be close to zero in any event. Thus the current stimulus to aggregate demand is not counterbalanced by rising interest rates, and the extra output today is obtained on the cheap.
To repeat, the problem here is the Keynesian focus on macro aggregates such as "total spending" as opposed to the microeconomics of relative prices — and the deployment of scarce resources toward the production of those goods and services that satisfy consumer preferences.
The interest rate is a price; it really means something. If people decide to save less, and the interest rate rises, that isn't "bad" but simply announces the new situation. In the same way, if people decide to buy fewer potato chips and more tofu, then market prices need to adjust to communicate this fact to the relevant people.
DeLong takes it for granted that right now the interest rate "ought" to be in the neighborhood of −5 percent, but that is a result of his Keynesian model. Indeed, we really don't know what the term structure of interest rates would have been, had Bernanke stood back and let AIG and some major banks collapse in September 2008.
In other words, we don't know for sure that Treasuries without massive Fed intervention would currently have such low yields, which is the entire basis for DeLong's argument. Indeed, the ostensible purpose of last year's "quantitative easing" by the Fed was to lower long-term Treasury yields. So it is a bit odd to point to low long-term yields as proof that trillion-dollar deficits today aren't distorting the free market.
At times the Keynesians do offer a hint of how it is physically possible that government budget deficits can boost the real standard of living of the community: by putting idle resources back to work, more total "stuff" gets produced. In an earlier article I specifically tackled this Keynesian argument.
Brad DeLong's recent arithmetical argument for bigger government deficits is flawed on several counts. He treats government spending as equivalent to household spending, and his focus on aggregates overlooks the coordinating function of interest rates. In the end, he offers little explanation of how letting politicians borrow an extra $100 billion to spend on anything at all could possibly make the whole country richer.
Note: The views expressed on Mises.org are not necessarily those of the Mises Institute.