Tuesday, 25 November 2008

Fiscal salesmanship

Greg Mankiw links to John Taylor's article in the Wall Street Journal looking for alliterative proposals (or predictions) for the expected US fiscal rescue package.

Shame the S has already gone, as I wanted to propose self-fulfilling: successful or squandered. Taylor's article takes the Milton Friedman viewpoint that a temporary stimulus does not work, because people base their spending on expectations of permanent income and will just put a temporary rebate in their savings account. Robert Lucas goes even further and says that even a permanent change does not work because people adjust their spending in accordance with their expectations of future government borrowing.

Both of these are respectable (indeed Nobel prize-winning) economists. But consensus generally allows that a Keynesian stimulus did work in the 1930s and does still work, sometimes, or to some extent. Certainly policymakers still indulge in them from time to time. So why is this, one of the most important questions in macroeconomics, still not settled?

This is not really an economic question, but a psychological one. And economic theory is not yet able to deal with that properly.

So, the answer is that it sort of works and sort of doesn't; and it is easy for both sides of the argument to find support for their theory. Perfectly rational agents probably would adjust their spending to allow for future taxes and exactly counterbalance a stimulus; but we aren't perfectly rational, so we don't; and yet, under some circumstances we do anyway.

We don't spend the entirety of our tax rebates - sometimes we don't seem to spend any at all - so the Friedmanites can prove they are right (Taylor has a graph on the WSJ article, based on the one-off rebate from May 2008, which is pretty convincing). But sometimes we do spend it -in the case of the Bush tax rebate in 2001, which was the first payment of a temporary (five year) tax cut - so the Keynesians can prove they are right too. What's the difference between the two scenarios?

The reason I say the stimulus will be self-fulfilling is that if everyone assumes it won't work, then it won't; and if they think it will, it will.

Knowing what we do about behavioural economics, let's think why someone would save their rebate instead of spending it.
  1. They may feel that they are spending just enough to survive, and are saving less than they want to. The marginal value of saving one more dollar is higher than that of spending it. In theory, this would only happen if they are exactly on the cusp of a marginal change in relative utility between saving and spending - otherwise they would have already redistributed some income into saving.
  2. They may have established spending habits which they have not thought about for a while - having risen, Peter-principle-like, to consume the majority of their income - and a discontinuous boost in income gives them a prompt to evaluate afresh what they will do with it. Like a new year's resolution, they may succumb to virtue and save the money.
  3. They may already be borrowing in order to spend at their desired level; and rather than spending the new money, they pay off part of a credit card balance.
  4. They may feel under pressure because of high debt, and are looking for opportunities to pay some of it off.
  5. They may be expecting a future fall in income and wish to protect themselves against it.
Conversely, why would they spend instead of save?
  1. They might feel they are saving the right amount already and the rebate is a discretionary boost. This could happen, say, if savings rates are high. The rebate is like a little lottery win - and few would argue that people don't spend when they win the lottery.
  2. Similarly, they may have confidence that things will remain stable for them, and additional saving is not needed - in which case there is no reason not to spend the extra income.
  3. Their income may have fallen, so the rebate partially replaces lost income - enabling them to spend at previous levels for at least one more month.
  4. They may simply want some goods or services more than they want the security of having higher savings. In particular, there could be some specific item which has been out of reach but can be afforded now.
  5. They may be expecting a future rise in income, in which case the confidence of increased future saving may permit the gratification of buying something now.
The decision is asymmetric in the sense that people always want a bit more consumption, but they don't always want to save more. This supports the spending argument. Against that, the current environment is one where expectations are for lower, not higher, income.

Generalising, there are two themes which dominate:
  • Is there any established or habitual behaviour which is suppressed by recent changes - if so, they may want to return to it.
  • Is there a strong expectation of the future - such as fear or optimism?
This would indicate that those whose income has been stable recently are more likely to spend the boost unless they are fearful for the future; those whose income has fallen are likely to spend if they previously cut back spending, or save if they have cut back saving.

Most of these reactions are, in strict economic terms, irrational. So our goal is to understand people's irrationality, model it, and relate it to concrete changes in their circumstances; and then choose an appropriate structure for any fiscal boost. And because expectations in turn are influenced by the stimulus itself, make sure that the structure is one that the recipients believe will work.

So here are examples of how a fiscal intervention could work, if its goal is purely to increase spending. Some of these examples may not be consistent with equity or natural justice, and I am not necessarily recommending them:
  • Target it at those whose incomes are stable
  • Target it at those who expect their income to increase - for example young people or graduating students
  • Alternatively, provide it for long enough that the stimulus itself becomes an expected long-term income increase
  • Or, make it automatically adjust to increasing income - the classic "automatic stabiliser"
  • Target it at those who have been saving "enough" already or those who show no inclination to save
  • Aim it at those who have cut their spending recently - we have enough evidence from supermarket loyalty cards to get some understanding of this
  • Package it in such a way that it is least likely to disrupt existing saving habits - for example by a VAT reduction, which does not put a readily-saveable lump sum in anyone's pocket but still leaves us with more money to spend
  • Provide it in a way that is consistent with easy spending decisions - for example subsidising products to make them more attractive, or by giving out $5 bills in bars which sell drinks for $5
  • Ensuring borrowing is available if people want it
  • Doing what you can to change the public conversation, perhaps by providing the right kind of message alongside the policy change; increasing expectations is as important as increasing the cash in your pocket
Thus a fiscal intervention may work well, or not at all. The result is down to the economic conditions prevailing, from which follows the choice of who it is targeted at and how it is presented. In business, this is what we call "selling". A bit of a sales pitch along with the stimulus will do a lot of good.

What can I pick as my alliterative label, then? How about Behavioural, Borrowing-linked and in small Bits? Unfortunately the B has also been taken already, so one more try: focused, fine-grained and fully credible.

Update: Tyler Cowen at Marginal Revolution has posted intelligently and concisely about this. And Mark Thoma at Economist's View has commented on Taylor and written in more detail about stimulus through government spending instead of tax cuts.

No comments: