Selgin properly points out that the Fed has exercised plenty of discretion and has not slavishly adhered to the Taylor rule by any means. However, I still think Kocherlakota has a point.
In my view, coming up with a simple formula relating a setting for the federal funds rate to past deviations of inflation from target and output gaps is a fools errand. Of course, I favor a level target growth path for nominal GDP rather than a target for inflation and an effort to minimize output gaps. But I don't think coming up with a fixed rule for changing the federal funds rate based upon past or even forecasted deviations of nominal GDP from target is wise. Nor do I favor coming up with a fixed rule for adjusting the quantity of base money according to past or even forecasted deviations of nominal GDP from its target level.
In a market economic system, setting prices and quantities is an entrepreneurial decision. It cannot be distilled into a rule. I do think that an economy needs a nominal anchor. But no one is proposing that the federal funds rate serve as a nominal anchor. And Taylor-type rules don't make the quantity of base money into the nominal anchor either. (The two percent inflation target is the only nominal anchor with a Taylor rule.)
I don't think the Federal Reserve should worry about the federal funds rate at all, but it should adjust the interest rate it pays on the reserve accounts held by banks and the quantity of base money it creates. It should also adjust the interest rate it charges for loans to banks. Its goal should be to keep the quantity of base money equal to the demand to hold it. Changes in the interest rate it pays will influence the demand. Open market operations and lending to banks will determine the quantity. The decision on interest rate and quantity needs to be forward looking, like all entrepreneurship.
The demand for base money also depends on nominal GDP and so these decision do need to be constrained by the need to keep nominal GDP on target. But the rule should relate to the commitment to the nominal anchor, preferably a target growth path for nominal GDP--not the level of base money or the interest rates the Fed pays or charges. The Fed's only commitment should be that it will set interest rates and a quantity of base money consistent with nominal GDP being on target in the near future (for example, one year from now.) The Fed should make no commitment as to what either these interests rate or the quantity of base money should be.
Of course, the Fed has a monopoly on the issue of base money. But even so, I don't think that means that its pricing and quantity decisions should be based upon some mechanical rule. It needs to be understood to be entrepreneurial like all such decisions, even when a producer has a monopoly on the provision of an important product. Just because a firm has a patent on a lifesaving drug doesn't mean that its price and quantity should adjusted by some mechanical rule.
Putting all of the entrepreneurial eggs in a monopolist's basket is not the ideal course. Introducing competition so that interest rates and the quantity of money are determined competitively, arising out of the decisions of many entrepreneurs should be the goal. But can that be combined with the constraint of a decent nominal anchor?
I have long argued that the Fed should float its interest rates--paying less and charging more than market-determined short term interest rates. It should adjust the quantity of base money to try to meet the demand to hold it at market determined interest rates In effect, the determination of the interest rates that Fed sets would be farmed out to a competitive market. Of course, changes in the quantity of base money will influence market interest rates in the short term and so indirectly the interest rates the Fed would charge and pay. The institutional framework would be consistent with the Fed gauging its open market operations to keep short term market interest rates at a level it believes is consistent with a quantity of money adjusting to the demand constrained by the nominal anchor.
I have also advocated the complete privatization of hand-to-hand currency. The quantity of that important portion of what is now base money would then be jointly determined by competitive forces. Each bank would entrepreneurally determine its issue of currency and the total quantity of currency would be the sum total of those decisions.
However, since I believe it is both desirable and highly likely that any such currency would be redeemable with the remaining portion of base money--reserve deposits at the Fed--this would not take away the key Fed monopoly and the Fed's need to act entrepreneurally to determine the appropriate quantity of reserves. That is, meeting the demand by banks to hold them at market determined interest rates and consistent with the nominal anchor.
I believe that it may be possible to design a clearing mechanism that uses index futures convertibility to enforce the nominal anchor without there being any base money at all. The position of each bank on the futures contract would vary with its net clearing balance at the clearinghouse. If any net debit balance would be secured by specified securities (like T-bills) and the interest rate charged on net debit balances is more than that generated by the securities and the interest rate paid on net credit balances is less, then desired balance for each bank would be zero. Under such a system there would be a quantity of reserves. It would be the total of the net credit balances of banks and always matched by the total of the debit balances of other banks. It is just that in equilibrium the quantity would equal the demand to hold them which would be zero. The system is disciplined by the index futures contract so that if nominal GDP is expected to stray from target, the private interests of the banks participating in the system will drive changes in both the market determined quantity of money (largely checkable deposits) and market determined interest rates to return it to target. It is all just a variation on the mechanics of clearing in the Black-Fama-Hall system developed by Greenfield and Yeager years ago.
Well, maybe it won't work. But that brings me back to Kocherlakota. He seemed to be criticizing a mechanical rule relating the federal fund rate to inflation and output gaps. But his substitute is that the Fed should remain strongly committed to its goal. To me, that rings true. I think that the Fed, or our monetary institutions generally, should be strongly tied to the nominal anchor. The Fed should not be able to adjust the nominal anchor on the fly. For me, that is, it should not be able to adjust the target growth path for nominal GDP period by period. But it should not be tied to any mechanical rule for manipulating market interest rates or broad conglomerations of monetary assets or even the interest rates it pays or charges or the quantity of the monetary liabilities it issues. Taylor type rules are a bad idea.
Kidland and Prescott's model has the Fed changing the nominal anchor period by period to exploit a short run phillips curve and obtain modest decreases in unemployment. This would be like having the Fed raise the target growth path for nominal GDP period by period to try to create temporary booms in real output. That is a bad idea.
Kydland and Prescott didn't show that keeping the quantity of money growing at a constant rate despite fluctuations in velocity is better than adjusting the quantity of money to keep nominal GDP growth, inflation, output and employment more stable. Nor did they show that manipulating interest rates according to rule based upon past inflation and estimated past output gaps when shifts in investment demand or saving supply are causing changes in the natural interest rate are better than adjusting interest rates in a way that keeps nominal GDP, inflation, real output and employment more stable.
Selgin points this out as well, noting that older arguments for rules as opposed to discretion emphasized the difficulty in forecasting velocity and the natural interest rate. I grant that, but I think that a fixed rule for manipulating interest rates or base money are really just nonstarters.