Can Monetary Stabilization Policy Be Improved by CPI Futures Targeting?
Garrison, Roger W., White, Lawrence H., Journal of Money, Credit & Banking
In recent years several economists (Hall 1983; Glasner 1989; Sumner 1989, 1995; Hetzel 1990; Dowd 1994, 1995) have made related proposals with the appealing objective of "using market expectations to direct monetary policy."(1) Sumner in particular argues that a central bank can make its stabilization policy more precise (reduce the deviation of the policy goal from its target value) by using feedback from a market for futures contracts written on the policy goal, and can make its policy more credible (overcome the time-inconsistency problem) by itself taking a position in such contracts. If Sumner's proposal were adopted in the United States and the policy goal were a constant consumer price index, for example, the Federal Reserve would trade in CPI futures contracts, and use information from the trading activity of the public to adjust the money stock.(2) Forcing the Fed to trade in CPI futures contracts would supposedly both bind the Fed to a goal of stabilizing the CPI, and translate that goal into appropriate short-run behavior.
We spell out reasons for doubting that Sumner's proposal would achieve greater precision or substantially greater credibility for monetary policy. First, we find that the public would want to defer taking a position in the market Sumner envisions until it is too late for the Fed to respond. If the public were to speculate in timely fashion and thereby reveal information enabling the Fed to improve its precision, the Fed's effective use of that information would paradoxically destroy the incentive to speculate. Second, because the Fed can print money and lacks an effective budget constraint, the Fed's exposure to trading losses may fail to mitigate the time-inconsistency problem.
Given that the private sector has not sustained a market in CPI futures (Horrigan 1987, Dowd 1995, p. 57), the Fed is to make one: "the central bank could create such a market by offering to buy or sell unlimited amounts of CPI futures at the target price" (Sumner 1995, p. 93). In effect, CPI futures contracts allow speculators to "bet" on the price level at a specified future date, with the Fed serving as the "bookie." The Fed is to note the market's betting position and adjust monetary policy appropriately.
The CPI being measured once a month, contracts are to be available on each upcoming month's CPI. For example, the Fed throughout May offers an unlimited number of contracts on the June CPI at $50,000 each ($500 times the CPI target of 100). It offers them continuously at the same (par) price until the close of business on May 31 (the last day before price data for June begin to be collected). Those choosing to "bet" on the June CPI take either the short side or the long side of a contract, with the Fed obliged to take the other side.(3) If the realized June CPI is 98, or 2 percent below the target level of 100, the Fed pays $1,000 (2 percent of $50,000) on each short position held by the public, and receives $1,000 on each of its own short positions (from members of the public who are long). If the realized CPI is above 100, the payments go the opposite way, again in proportion to the discrepancy. Thus the Fed pays and receives nothing only if the realized CPI is exactly 100; otherwise its net payment is zero only if the price-level bulls are exactly balanced against an equal volume of bears. A financial loss is imposed on the Fed if the public in net terms has bet smartly, for example, if the majority of speculators are long and the June CPI comes in above 100.
The lag in reporting the monthly CPI constitutes an "information lag" for any monetary policy geared to the reported CPI. Overcoming the information lag is the essence of Sumner's promised gain in precision. If the public has more timely price information, and contemporaneously reveals it in the CPI futures market, then by responding to that "early warning" of CPI deviations, the Fed can reduce the variance of the realized CPI around 100. …