Fiscal multipliers are pretty hard to estimate as they tend not to be particularly stable. Which is another way of saying that the impact on output of a given change in fiscal policy – which we can define as the planned path of government expenditures and tax receipts over some appropriate planning horizon - depends on a very large number of factors for which it is nearly impossible to condition. A non-exhaustive list for fiscal policy might include the extent to which the fiscal expansion is expected, credible, temporary or financed by changes in spending or taxes, as well as the debt instruments issued. The estimated response of the economy will depend how firms, households and banks respond to the changes in fiscal policy but also may be complicated by the response of these same agents to the same shock to which fiscal policy is responding. The responses of other policy makers such as central bank responses to any fiscal policy and that of other fiscal authorities, particularly in a monetary union, will further complicate any attempts to estimate a clean partial.
I illustrated some of these issues in a talk to St Catherine’s Political Economy seminar last month. One early piece of evidence was simply a Table of recent estimates of the multiplier for the UK. The estimates from these sources give a wide variety depending also on the way in which the fiscal stance is changed. The outlier in these estimate are those recently derived by the IMF in the October 2012 WEO (Box 1.1), who used the negative residual (where actual output turns out to be below the forecast) of output forecasts from 2010-11 to find some correlation between the timing of fiscal consolidation and the continuing underperformance of the economy. It turns out though that although the IMF paper tries to apply many controls, they do control for the impact of household and bank deleveraging.
Actually, all kinds of other factors may have led to such an undershoot. Let me illustrate one. In this model, “Reserves, liquidity and money:an assessment of balance sheet policies”, in which firms produce monopolistically competitive goods with sticky output prices, optimising households who are credit-constrained by the extent to which banks lend to them in return for collateral in an economy which is driven over the long run by productivity. This economy can be stabilised by monetary policy changing interest rates or operating open market operations but fiscal policy also has a role by altering the supply of bonds held by the private sector, which act as collateral. In this model we can alter the size of the government debt by increasing expenditure relative to taxes and trace the impact on output under three scenarios: (i) a fiscal expansion when policy rates also rise and when banks increase their lending; (ii) a fiscal expansion when policy rates do not rise and the banks increase their lending and (iii) a fiscal expansion when policy rates do not rise and when banks start to delever and contract their level of activity as part of a consolidation.
The set of impulse responses above are illustrative but show (look at the top left impulse response) that under scenario i) we see a multiplier of around 0.2, under (ii) it is around 0.5 and under (iii) it is around 0.1. The “fiscal multiplier” in a monetary economy, where banks set the level of broad money depends not only on the policy response. But also the impact on other interest rates – the external finance premium or bond rates. And most importantly of all, how the banking sector responds to the state of economy that has led to the need for a fiscal expansion. This simple calibration shows that cautious banks, or deleveraging banks, who increase the level of monitoring work they undertake and hence hold back on creating new loans can per se offset the impact of any fiscal expansion. The key therefore, as we have known for some time, is to fix the banking system. The problem is that the fix also requires deleveraging and consolidation.