18/3/2012: Fiscal Stimulus Multipliers - US data and some Irish considerations

In a recent paper, “Fiscal Stimulus and Distortionary Taxation”, Thorsten Drautzburg and Harald Uhlig (published Becker Friedman Institute for Research in Economics Working Paper No. 2011-005 ) estimate the fiscal policy multipliers from the federal spending programmes under the American Recovery and Reinvestment Act (ARRA) of 2009.

Fiscal multiplier is the ratio of output changes to the total stimulus policy-driven change in Government spending and transfers. So positive large multiplier means greater response in economic output per unit of spending, negative multiplier means a fall in the economic output for a unit of spending.

In addition to the traditional literature, the authors include a number of coincident effects:

• Fiscal expansion takes place in the environment of recession, which is also coincident with the Federal Reserve carrying out a monetary easing – referenced as the zero lower bound interest rates policy (ZLB). When the fiscal policy stimulus creates positive impact, this translates into incentives for the Fed to exit ZLB earlier, which in turn reduces economic activity rate of growth. The effect is compounded in the case when wages are sticky (for example, if negotiated via collective bargaining) and/or prices are sticky (for example, if set by regulatory authorities).
• In addition, the authors recognize that implementing / deploying fiscal stimulus in practice takes time in practice.
• Third, government expenditures on stimulus are financed, eventually, with distortionary taxes, creating costly disincentive effects in the future.
• Fourth, welfare transfers matter “to the degree to which they are given to credit-constrained households”.
• Fifth, the authors use Bayesian estimation techniques as well as sensitivity analysis to quantify the uncertain nature of the estimated coefficients in the New Keynesian model.

The study distinguishes between short-run and long-run multipliers in a benchmark model, finding:

• “Modestly positive short-run multipliers” that average 0.51, and
• “Modestly negative long-run multipliers” averaging around -0.42.
• The multiplier is particularly sensitive to the fraction of transfers given to [credit constrained] consumers, is sensitive to the anticipated length of the zero lower bound [the expected period of monetary policy easing], is sensitive to the capital share [in overall stimulus – i.e. the share of stimulus spending directed to capital formation as opposed to consumption] and is nonlinear in the degree of price and wage stickiness.
• Crucially, “reasonable specifications are consistent with substantially negative short-run multipliers within a short time frame” meaning that under reasonably realistic assumptions on price/wage stickiness and model parameterization fiscal stimulus can result in a negative effect on economic growth even in a short-run.
• In the US, the stimulus results in negative welfare effects for agents not constrained by debt. The debt-constrained agents gain, if they discount the future substantially.

Now, the above results are quite interesting in the context of the US economy, but they are even more interesting in the context of Ireland, as a small open economy with fixed interest rates. This is so for a number of reasons:

1. Ireland’s domestic demand is closely linked to imports, which means that unlike in the US, Ireland’s short-run multipliers can be expected to be smaller in magnitude due to losses of economic activity to imports. Note – imports enter GDP and GNP determination as a negative component, so any stimulus funds expended on imported consumption (public and private) will have a dual effect on overall economic activity: in the short run, they will reduce economic activity via imports increases (effects not present in the US economy), and they will reduce, in the longer term economic activity via same pathways as those revealed in the paper.
2. Ireland’s public expenditure is heavily leaned in the direction of consumption supports / income transfers. These provide support for both credit ‘unconstrained’ (non-indebted) poor households (who do not have mortgages and are not subject to the adverse effects of debt overhang) and for households of the unemployed who are constrained by debt overhang (hence significant rise in mortgage supplements payments). Those households, constrained by debt (such as mortgage holders still in employment) and the ones unconstrained by debt and not in welfare net (for example older households with no debt overhang) are therefore direct losers in the short run and in the long run. This, per findings above means that increasing social welfare transfers during the current crisis can lead to reduced economic activity even in the short run, while increasing negative effects of the stimulus in the long run.
3. Our monetary policies are determined outside Ireland and hence we can add higher uncertainty and shorter periods of ZLB duration to the Ireland-specific assumptions (remember, the ECB did hike interest rates into the Irish recession and then repeated the same again). This means that any fiscal stimulus in Ireland will be subject to stronger monetary policy headwinds, further reducing the multipliers in the short run and amplifying long-term costs of such a stimulus.
4. Ireland’s tax system became even more distortionary during the crisis and this process is ongoing. Once again, this amplifies the adverse finding from the US data and interacts negatively with point (2) above.

The study also incorporates consideration of the time-varying differentials between the central bank-set interest rates, government borrowing costs (bond rates) and returns to private capital (cost of private sector credit). The paper shows that “these wedges are indeed the key to understanding the recession of 2007 to 2009.” Although the study does not explicitly quantify these drivers effects, one can suspect that in the case of Ireland, dramatic increases in the cost of Government borrowing, alongside the rise in retail-level interest rates due to banks bust would have much more adverse impact on fiscal stimulus effectiveness. Whether or not these effects are enough to swing the stimulus short-run multiplier to the negative territory we do not know. But it is pretty safe to assume that they will make long-term costs of the stimulus more severe.

The use of New Keynesian model specification allows authors to conclude that “the model here is heavily tilted towards a model in which fiscal stimulus is often thought to work well: we therefore believe that the negative long-run effects of fiscal stimulus should give pause to arguments in its favor. Even at the short horizon, the benchmark multiplier is just around 0.5.”

Another interesting result is that consumption taxes, rather than income taxes are a better way to offset the costs of stimulus in the longer term. This is intuitive and consistent with other evidence. However, the paper finds that “adjusting [raising] consumption taxes only yields a slightly higher multiplier than adjusting labor tax rates.”

Please keep in mind that the Irish stimulus theory supporters have ardently argued that fiscal stimulus must be financed via income tax increases, not consumption tax hikes and have opposed even a modest Budget 2012 shift of tax burden on VAT.