Karl Smith has a recent post "Why is the US Government Still Collecting Taxes?: Should Lambs Lay Down with Lions Edition" arguing that we should be dramatically reducing tax collection because interest rates on government bonds are so low that, after correcting for inflation, every dollar in the national debt is shrinking over time. Karl’s post follows up on Ezra Klein’s post "The world desperately wants to lend us money," and my grad school professor Larry Summers’s post "It’s time for governments to borrow more money." Ezra and Larry want the government to pay ahead on things it is going to buy anyway and make other productive investments.
Larry, coming from his experience as Secretary of the Treasury makes a strong argument that the U.S. government should be borrowing long rather than short, in order to lock in amazingly low interest rates on long-term government bonds. Since the U.S. government’s position in relation to the bonds it has issued depends on what the Fed does as well as what the Treasury does, this argues against the Fed buying long-term government bonds when Larry’s argument says we should be selling long-term government bonds on net. Larry has convinced me by what he writes in "It’s time for governments to borrow more money." The Fed should not be buying long-term government bonds.
But for the sake of stimulating the economy, the Fed needs to be buying large quantities of some type of asset. This is not just my view, implicit as a possibility in my post "Balance Sheet Monetary Policy: A Primer," but the the official view of the Fed itself, since even the mostly stay-the-course policy the Fed announced at its last FOMC monetary policy making meeting involves buying large quantities of assets. (“Large” is less than “massive.”) So it is unfortunate that the legal authority of the Fed to buy assets is limited to a relatively narrow range of assets. See Mike Konczal’s interview with my undergraduate classmate Joseph Gagnon on the legal constraints the Fed faces. Short of buying foreign assets or otherwise going outside the Fed’s comfort zone, that probably means that the Fed should be buying mortgage-backed securities such as those issued by Fannie Mae. In general, when it uses balance-sheet monetary policy (what the press calls “quantitative easing”), the Fed should lean towards buying assets that have a high interest rate.
It is important to note that many of the objections to more vigorous use of balance sheet monetary policy in the U.S. are really objections to buying long-term Treasury bonds, not objections to buying other assets. Anytime you see an argument against balance sheet monetary policy, check whether it is an objection to buying any kind of asset or only to buying long-term Treasury bonds. And sometimes there are objections to buying Treasury bonds and other objections to buying mortgage-backed securities that even combined do not apply to other assets. So I wish the Fed had the kind of authority the Bank of Japan has to buy a wide range of assets, including commercial paper (CP), corporate bonds, exchange traded funds (ETF’s), and real estate investment trusts (REIT’s). I got this list of assets from the Bank of Japan’s own website on the range of assets it is buying. See also my post on how the Bank of Japan should use this authority quantitatively: "Future Heroes of Humanity and Heroes of Japan."
Given the low interest rates the U.S. government is facing, even aside from stimulating the economy it should be spending more—mostly on things it would buy later anyway and other productive investments. On stimulating the economy, my proposal of Federal Lines of Credit is gaining some traction: you can see my posts on what Brad DeLong and Joshua Hausman and Bill Greider have to say recently, as well as what Reihan Salam said early on. In what I write myself on Federal Lines of Credit, including the academic paper “Getting the Biggest Bang for the Buck in Fiscal Policy” that I flag at the sidebar, I emphasize the importance of not ultimately adding too much to the national debt as it will stand, say ten years from now. This is actually consistent with saying that the government should be spending money now on things the government is going to spend money on sooner or later anyway, regardless of which party is in power, such as maintenance of military assets and stores, and the government should be spending money now on things that will add to the productivity of the economy so much that they will generate more than enough tax revenue to pay for themselves.
The existence of government investment that has this property of generating more tax revenue in the future than needed to make the payments on the money the government borrowed to make the investment is the spending counterpart to being on the wrong side of the Laffer curve so that you could cut taxes and get more revenue. I am not sure how many government investments meet this stringent test, but if any do, everyone should be in favor of them, regardless of their political viewpoint. To have that statement be true, I am assuming that the investment is something noncontroversial that everyone would be glad to have for free (not something like a national stem-cell laboratory). The reason everyone should be in favor of such a self-financing investment is that the American taxpayers would be getting a better deal than “free.”
Now, a government investment being self-financing in this sense is quite hard because for this, it is not good enough to have benefits great enough to pay for the direct cost to the government (the naive cost-benefit test). The extra payments to the U.S. Treasury alone (typically a minority of the total benefits) must be enough to pay for the investment. If on net the U.S. Treasury is out money at the end of the day for the sake of other benefits, then the dollars from the U.S. Treasury have to be counted as costing more than dollar for dollar since each dollar of government spending typically has a deadweight loss from tax distortion on top of it. (See for example Diewert, Lawrence and Thompson’s paper on this. There is a good discussion in this paper accessible to anyone even before the first equation.) As long as the U.S. Treasury is out money at the end of the day, the sophisticated cost-benefit test can easily flip from thumbs up to thumbs down depending on how big the tax distortions are—and economists don’t agree on that. I am on the side of believing there are relatively large tax distortions. (See my first post, "What is a Supply-Side Liberal," which is also now at the sidebar.)
Interest rates won’t stay low forever, and the U.S. government, unlike the government of the United Kingdom of Britain and Northern Ireland, does not sell consols that would lock in a low interest rate forever (though maybe it should in the current environment). Moreover, extra debt probably raises the interest rate on existing debt. So the cost of extra debt is higher than the interest rate itself. (This is the monopoly problem from Economics 101: the U.S. government is a monopolist in the original issuance of its own debt, and so has to consider the effect of issuing more debt on the price and interest rates of existing debt.)
The last words in this post are is in response to Karl Smith’s post, which motivated the lamb and lion illustration. (Karl’s “lamb” is the U.S. government, while the “lion” is the bond traders in the financial markets). If, as I think they will, interest rates will some day go above not only inflation, but above the long-run growth rate of the economy as a whole (which matters among other things because it is the growth rate of the tax base), and the government cannot lock in a lower rate forever using consols, then any money the government borrows will cost us sometime in the future. The low interest rates now prevailing will make government borrowing now cost less in the future, but it won’t make the cost zero. Doesn’t that mean that we should do all our taxing later, when the amounts of money will have shrunk rather than now? It would if the cost of getting an extra dollar of tax revenue were constant over time. But a basic result from public finance is that the cost of getting an extra dollar of tax revenue goes up as you try to get more and more tax revenue as a fraction of GDP. That is why I was so confident in my post "Avoiding Fiscal Armageddon" that there is some limit of government spending as a fraction of GDP that we shouldn’t go beyond, even if that limit is higher than the 50% that I used as an example (and as a reasonable representation of my own judgment given my current knowledge.) Given the pressures on the government budget that I talk about in "Avoiding Fiscal Armageddon," I will be shocked if tax rates are not higher in the future than they are now. So if the cost of getting an extra dollar of tax revenue goes up with the rates you already have, then there is an argument for moving toward raising tax rates at times such as now when tax rates are low compared to what they will be later. This “tax smoothing” argument counterbalances the implications of current lower interest rates, which favor lower taxes now. On balance, given the low interest rates the U.S. government is paying, tax rates should certainly be lower now than we expect them to be in the future, but there is a question of how much.
Why is this called a “tax smoothing” argument? The reason is that raising tax rates when you think tax rates will be higher in the future—or following similar logic, lowering tax rates when you think tax rates will be lower in the future—tends to equalize tax rates now relative to expected tax rates in the future. I have not personally done any research on tax smoothing, but I learned about it from proofreading all of the math in Greg Mankiw’s paper "The Optimal Collection of Seigniorage: Theory and Evidence," when I was Greg’s research assistant in graduate school, and I have heard more about it from my colleagues at the University of Michigan.