Discretion vs Rule: The eatery edition

What do the Reserve Bank and eateries have in common? Both implement rule based policies instead of discretionary policies and both suffer criticism from their clientèle for doing so even though it is in the clients ultimate interest.

I noticed this today when I went to get some food for lunch. A man in front of me was trying to get cash out, when there is a sign that says “no cash out” at the counter. The man was irritated by this rule, he wanted cash and there was cash in the till. The service person tried to explain to him that if they let people get cash out, then they ran out of change in the counter, which causes delays later in the day – furthermore, if they give him cash they run the chance that other people may begin expecting that they can get cash out, and employees would feel more obliged to. As there was a cash machine just outside, the cost of getting the cash somewhere else was very low for the man, however the delays later in the day would have been costly for both the firm and the consumers involved. In this case, the rule improved the social outcome – any deviation from this rule may change peoples beliefs and lead to a case where most people are worse off.

The reasoning behind the eatery’s choice to prevent cash being taken out mirror the situation the Reserve Bank follows. If monetary policy is rule based, households and firms know interest rates will increase at a certain inflation rate, and they will set prices accordingly. If we have a discretionary monetary authority then they may be unable to commit to a given path for monetary policy – we may have time inconsistency.  In this case, the monetary authority may know that it must keep rates high now and in the short term future to control inflation, however when it reaches that future it decides that the cost appears too high to keep price increases in this period down.  If this is the case, consumers will not believe that the Bank will be hard on inflation, and so they will set higher prices as a result – creating more inflation than we had in the rule based case.

Central Banks, such as the RBNZ, try to achieve the rule based outcome by setting inflation targets and being single minded with inflation.  Once they begin discussing things such as output, and exchange rates, then price setters start to doubt the Bank’s commitment to keeping inflation low, which leads to higher ‘inflation expectations’ – in turn leading to higher inflation.

This is the common argument for rule based monetary policy, something I think I subscribe to relatively strongly (as strongly as an economist can subscribe to anything 🙂 ).  My question is, in what way may the rules practiced by the eatery and Central Banks be suboptimal?

  • CPW

    But rules don’t necessarily produce any more credibility than a policy target. Time inconsistency doesn’t vanish, and I would argue that time inconsistency seems to come most often through changes in the policy regime/target rather than interpretation of the policy regime (NZ monetary policy a good example of this).

    Perhaps I misunderstand you, but are you suggesting you’d prefer a computer with some variant of a taylor rule over the current framework? (although you seem to be suggesting an even simpler reaction function than this, nothing but the inflation gap).

    The first problem I see is that it just wouldn’t work – even a rules based framework would have to have some discretion (e.g. can we ignore a one-off price change due to government policy, and when do we get get to re-estimate the variables in our model, or change the model’s workings?).

    The second problem is that I don’t see a single-minded inflation target as appropriate in all situations. I think that too weak growth is more prone to becoming a lingering recession than high inflation is of becoming hyper-inflation. Hence I would generally agree with the path the Fed has taken recently, which I would characterize as suggestive of a belief that the growth risks are strong enough to out-weigh temporarily the inflation risks.

    PS eatery is sub-optimal too in my opinion. The decision to give cash out seems to be very discretion based across businesses, as a result the customer never knows until you get to the counter whether you can get cash. Information costs!! Plus we have a case here of irrational actors putting costs on others by arguing, which we should incorporate into our decision about optimal policy. 🙂

  • “But rules don’t necessarily produce any more credibility than a policy target.”

    By rule vs discretion I’m talking about policy target vs no policy target in a sense, that is why I say:

    “Central Banks, such as the RBNZ, try to achieve the rule based outcome by setting inflation targets and being single minded with inflation”

    “The second problem is that I don’t see a single-minded inflation target as appropriate in all situations.”

    True, but all that matters is that welfare ex-ante is greater from a rule than from an alternatives.

    The question then is, is current Reserve Bank policy appropriate and why/why not? You mention costs from ‘weak growth’, but what type of costs do you mean? (I do agree there are costs, I just want to know which ones you want to discuss) After all if we live in a world that is explained by real business cycle theory, then the business cycle is the equilibrium reaction of agents to a supply side shock, which implies that any policy that aims to ‘stabilise output’ will be sub-optimal.

    PS. “The decision to give cash out seems to be very discretion based across businesses, as a result the customer never knows until you get to the counter whether you can get cash”. I agree in general but the place where I grab lunch is in the middle of a bunch of offices, and will mainly service the same people over and over again. In this case people have full information before they go and buy food. That explains why it might be efficient for them to enforce this policy.

  • “The first problem I see is that it just wouldn’t work – even a rules based framework would have to have some discretion”

    All systems have some discretionary element, but all shades of grey are not the same. Systems that provide greater precommitment result in better outcomes because they reduce the likelihood of time inconsistent behaviour. Even imperfect precommitment devices have value for central bankers.

  • CPW

    Sorry Matt I assumed from your original post that you thought the RB had too much discretion. You/rauparaha think it has an optimal amount? I do (well, it’s close enough).

    Do you actually think we live in a world characterized by RBC theory? I don’t, so obviously I think that monetary/fiscal policy can (in theory at least) stabilise output. I think that recessions as a result of a lack of demand are possible and can be minimized.

    I don’t think agents have completely rational expectations, so I think time inconsistency can potentially have payoffs providing you don’t try and get away with it too often. I’m not certain that such situations actually arise in practice: again using the recent Fed moves as an example, one can always argue that inflation wouldn’t persist in a recessionary environment.

    Even assuming that there was some way to enforce an absolute commitment to a rules-based system (bombs implanted in central bankers skulls triggered by cpi deviating too far from target?) I think it is extremely unlikely that it would maximize welfare relative to a system with discretion. The best justification I can think of is that the feasible gains in terms of lesser inflation/output variability compared to now seem pretty small, but the list of unforeseen circumstances that could lead to major policy errors seems potentially quite big.

    I think we might be having a semantic argument about what a policy target means in terms of rules versus discretion. Implementation of most policy targets seem to involve plenty of discretion, and hence it is never precisely clear what the policy response will be to any given set of economic variables. I agree with the consensus view that credibility and consistency is important for monetary policy; I just think situations could arise where you’d be willing to sacrifice some consistency for short-term gains and it would be welfare-maximizing over the longer-run.

    rauparaha: point taken, but I’d argue that the (notional) independence of modern central banks is about as far as you can take pre-commitment. There doesn’t seem to be any way to further reduce the main remaining time-inconsistency problem which is regime change (e.g. the government deciding that inflation isn’t as important as a low exchange rate). Although maybe free financial markets + exchange rate plus mainstreaming of central bank independance does provide a growing incentive for the government not to deviate from current policy.

    Although now that I think about it I guess making the workings of the RB model public and allowing no deviation from the model forecast interest rate track would have even more commitment value. Or some rule that tied interest rates to expected inflation as measured by inflation indexed bonds? Too much commitment methinks.

  • “Sorry Matt I assumed from your original post that you thought the RB had too much discretion.”

    No sorry, it was supposed to be a support of the status quo type post, I am happy with the current level of discretion

    “Do you actually think we live in a world characterized by RBC theory?”

    No I don’t :P. However, I also do not believe that monetary and fiscal policy should always be used to stabilise output.

    “I think we might be having a semantic argument about what a policy target means in terms of rules versus discretion.”

    Yes definitely, when I’m talking about rule policy I’m mearly stating that the Bank’s actions are constrained in some way, so that they can precommit to the optimal path. When I talk about discretion I am talking about their ability to deviate from this path.

    “All systems have some discretionary element, but all shades of grey are not the same.”

    Thats completely right, it is a continuum of potential choices rather than the black and white – rule vs discretion case I have laid out above. That is what makes the optimal choice so difficult to figure out.