A perennial discussion in monetary policy is rules versus discretion. Should the central bank be constrained to follow a policy rule or should it have policy discretion?
I find this debate generally muddled both in the means-versus-ends question and what is the proper underlying concern. Is the concern bad policy or unclear policy? With constraining the central bank from acting badly or with effective management of expectations? To bind the central bank to a rule, such as the Taylor Rule, is also to put great confidence that you have the correct rule. I prefer to think of the proper concern for central bank policy as a matter of having an appropriate goal and having a clear goal rather than having policy uncertainty.
As for means versus ends, take expectations seriously and clarity of purpose dominates constraining of action (provided actions are credibly connected to said purpose, the purpose is credible given available actions and the central bank is credibly committed to the purpose). After all, if actions are too constrained then that can generate very unfortunate expectations, particularly in crisis or abnormal conditions. Clear expectations also allow the central bank to achieve its goals much more expeditiously through the Chuck Norris Effect or central banking as Jedi Mind Trick.
If we go for a clear goal (understood as a policy target), then the time-horizon of the goal matters. The narrower the time-horizon, the greater the inflexibility and the more likely perverse outcomes are (such as following economic activity downwards, intensifying the spending crash). The time-horizon has to be constraining enough to give content to the goal but not so constraining as to lead to perverse rigidity. If macro-economic stability is the goal, then the time horizon should be the business cycle so as to minimise the danger that the central bank will intensify, rather than smooth, the business cycle.
We can thus expect that an inflation-targeting regime which has the business cycle as its time-horizon would function better than one with a narrower time-horizon. This is certainly the Australian experience, with the Reserve Bank of Australia’s over-the-business-cycle inflation targeting goal. Australia is now 21 years without a recession, which may well be a world record. (This is in dramatic contrast to certain other inflation-targeting central banks one might mention.)
If clarity of central bank goal is the best way to manage expectations, and if the business cycle is the appropriate time-horizon of said goal, then what should be the goal? A straightforward way to look at that is whether the goal encompasses the key functions of money. Money exists to facilitate transactions. It does this more efficiently if its value is sufficiently stable over time. So, the goal of price stability is appropriate, but not at the cost of serious and sustained impeding of its use in transactions; the basic purpose that price stability is desirable for.
Giving the central bank a goal with a time-horizon across the business cycle naturally ties together price stability and use in transactions, since the level of transactions on output will affect price stability. So would a stable spending-on-output-goal, or NGDP targeting.
Thus, central banks should have a clear goal; either price stability over the business cycle or a clear spending-on-output path. (I.e, if the economy wanders off the path, extra steps are taken to get it back on.)
What central banks shouldn’t have is an unclear policy goal that makes mismanaging of expectations easy or a policy goal whose time-horizon is so narrow it can make the business cycle more intense or a policy goal which seriously subordinates the facilitation of transactions-on-output to price stability. Rather, the goal should incorporate acceptance that some trade-offs will be required from time to time; though much less if expectations are properly managed, if the policy target has an appropriate time-horizon, if the goal targets spending-on-output directly.
The right goal, the correct time-horizon and effective management of expectations can be a beautiful thing. Screw those things up, and it can get very ugly and stay that way for a remarkably long time.
ADDENDA This paper (pdf), by a former member of the Bank of England’s Monetary Policy Committee, refers to a two-year time frame for monetary policy. Why two years? Why not 23 months or 21 months or 57 weeks or whatever? This is a numerological constraint with no inherent connection to economic conditions, unlike the over-the-business-cycle time frame that the Reserve Bank of Australia operates on.