Robert Barro claims that quantitative easing should expand aggregate demand, but that it is equivalent to the Treasury refinancing the national debt by selling T-bills and purchasing longer term government bonds. Why should the Fed be involved in what should be a Treasury operation?
Barro asserts that the Fed fears that when it comes time to sell off the government bonds it has purchased, that these sales will choke off economic recovery. That is why, according to Barro, the Fed is proposing to instead raise the interest rate paid on the reserve balances. But Barro argues that raising the interest rate paid on reserves will have the same contractionary effect as selling off bonds and reducing the quantity of reserves.
I think Barro is correct regarding the consequences of quantitative easing, but I think his approach is wrongheaded. Quantitative easing will raise money expenditures. It is certainly possible that having the Treasury refinance the national debt by selling T-bills and purchasing long term bonds could substitute for the Fed purchasing long term bonds. And raising the interest rate that the Fed pays on reserve balances would have much the same effect on money expenditures and output as the Fed selling off bonds.
From the quasi-monetarist perspective, the problem with the "zero bound" on T-bills (or any security) is that rather than a shortage of T-bills causing a higher price and lower yield on the T-bills, there is a spillover to an added demand to hold money. Assuming that the demand for money was intially equal to the existing quantity of money, the result is a shortage of money and reduced money expenditures on goods and services--a general glut of goods.
If the central bank responds to the shortage of money by expanding the quantity of money as it should, but it does so by purchasing T-bills (or whatever security is at the zero nominal bound,) then while the increase in the quantity of money directly corrects the shortage of money, the increase in the demand for T-bills exacerbates the shortage of T-bills, and with no increase in the price or decrease in the yield of T-bills possible, there is an additional spillover to added money demand, completely offsetting the increase in the quantity of money. The implication is clear--the central bank should not seek to expand the quantity of money by purchasing securities that have yields so low that holding money is a better alternative.
Economists like Barro do not like this sort of messy disequilibrium analysis. Instead, their bias towards equilibrium reasoning results in boiling this down to "T-bills and money are perfect substitutes when the yield on T-bills is zero." From that perspective, if the yield on T-bills is zero, then the quantity of currency and reserve balances plus the quantity of T-bills held by the public forms a single homogenous good. If we call this good, "base money," then when the interest rate on T-bills hits zero, suddenly, the quantity of base money rises by the existing quantity of T-bills. Of course, the change in definition implies that the demand for this base money rises by an equal amount. If the Fed purchases more T-bills with money created out of thin air, the total quantity of "base money" is unchanged.
If the Fed instead purchases long term government bonds in the usual way, expanding banks' reserve balances at the Fed, then the total of base money--currency, reserve balances and zero-interest T-bills, expands, which perhaps results in additional money expenditures on goods and services. From Barro's perspective, if the Treasury were to refinance the national debt by selling more T-bills and paying off long term government bonds, then the effect is the same. "Base money," made up of currency held by the public, reserve balances of banks, and zero-interest T-bills would expand the exact same amount.
But when Barro then asks why the Fed should be doing something that is equivalent a Treasury refinancing, he falls into confusion. It isn't that the Fed is involving itself in the Treasury's determination of how it should fund the national debt. It is rather that with T-bills bearing a zero yield, they are, in Borro's approach, equivalent to base money. If the Treasury sells more T-bills, according to Barro, it is expanding the quantity of base money. It is the Treasury that would inevitably be involved in monetary policy if it issues T-bills with a zero nominal yield.
However, his conclusion is sound. Suppose there is a shortage of T-bills at a zero yield. Rather than allowing this shortage to spill over to a shortage of money, suppose the Treasury refinances its debt, issues and sells more T-bills, and uses the funds to pay off longer term government bonds. The Treasury should be able to prevent any excess demand for T-bills and so avoid the spill-over to an excess demand for money. Of course, if the Fed refinances the entire national debt with T-bills, and there is still a shortage of T-bills, then nothing further can be done by the Treasury. Oddly enough, in that situation, there would be no long term government bonds for the Fed to purchase. The Fed would have to look to purchase other types of assets--perhaps private bonds.
If the Fed and the Treasury are consolidated, then monetary policy is a matter of how to fund the national debt. How much should be funded by the issue of hand-to-hand currency? How much by reserve balances to be held by banks? How much by T-bills? And finally, how much by T-notes and T-bonds of various maturities?
If the monetary authority is independent with some kind of nominal restraint, then it is up to the Treasury to fund the national debt and it is up to the monetary authority to control the issue of currency and bank reserves to hit its nominal target. The monetary authority doesn't have to limit its asset portfolio to government bonds. And to the degree it does purchase government bonds, the rule that it should not be buying any that have zero nominal yields makes more sense than treating all T-bills as base money when their yield falls to zero.
What about Borro's point that raising the interest rate the Fed pays on reserves is just as likely to choke off recovery as the Fed rapidly selling off its asset portfolio? From a quasi-monetarist perspected, this is exactly correct. The increase in the interest rate paid on reserves increases the demand for reserves and base money. By rapidly selling off its government bond portfolio, the Fed is decreasing the quantity of reserves and base money. What is important is excess demands for base money, not whether the excess demand occurs through a higher demand or a lower supply.
More importantly, the reason to raise the interest rate on reserve balances or sell off bonds to reduce the quantity of reserves is to prevent an excess supply of reserves from developing as the demand for reserves fall and commercial banks buy bonds or make additional loans as the economy recovers. The increase in interest on reserves offsets the decrease in demand, and the sale of Fed assets reduced the quantity of base money to match the decrease in demand. The whole point is not to cause a contraction but to prevent an excessive expansion.
The reason to raise interest rates on reserves rather than selling off assets is for the Fed to avoid capital losses on the asset portfolio. If the Fed raises interest rates on reserves enough, it will have higher costs, and perhaps its costs will rise above its earnings on its asset portfolio. But, it can suffer losses for a long time before it runs out of assets. Further, if there remains a large demand for zero-interest hand-to-hand currency for some time, that is a source of profits that could offset losses from paying interest on reserves. If the Fed had to rapidly sell off long term bonds (much less mortgage backed securities,) insolvency could arrive rapidly. Sure, there is no real distinction in present value terms, but in political terms, an open and transparent bailout of the Fed would be very different than reduced payments to the Treasury over the next 20 years.