The definition deliberately uses the vague term “fiscal variable” for reasons to be discussed later.
Note: this is an unedited draft of a section from my upcoming book on recessions. It refers to section numbers in the manuscript, which I am leaving in place. It skips over most of the usual subjects of debate (and research) around fiscal multipliers, rather it looks at the big picture. For a variety of reasons, the big debate about the fiscal multiplier is the size of the multiplier. The complexities of the topic are discussed in this text, and essentially explain why I am not too concerned with much of the debate around the size of the fiscal multiplier.
Model SIMAs an example of such a scenario, I will take the simplest possible “Keynesian” model: model SIM from Chapter 3 of the textbook Monetary Economics by Wynne Godley and Marc Lavoie.[i] I discussed the workings of this model in detail my book An Introduction to SFC Models Using Python. The advantage of using this model is that it is easily understood even if one is averse to mathematics. Although it has obvious shortcomings, they are easy to see and explain what they are. More advanced models have corresponding greater opacity.
We can also see that model GDP does not immediately jump to the final steady state, and so we immediately see a problem with discussing “the” multiplier: the effect of the change in spending properly has an effect that varies over time. We need to distinguish long-term effects from short-term effects, and this affects empirical estimates of multipliers.
What Fiscal Variable is Changing?The fiscal multiplier is the coefficient relating the change in fiscal policy to the change in GDP, which might be viewed as a sensitivity parameter. The previous example of model SIM illustrates the conventional way of viewing it: the multiple of a spending change to the effect on GDP. For the model economy, there is a spending multiplier of 5, so we multiply the change in spending by 5 to see the effect on (steady-state) GDP.
We can view most governmental programme spending as being a dollar amount under the control of the government – e.g., the salary expenses for permanently employed bureaucrats at a government agency, or defense department acquisitions. However, the other modes of fiscal policy are not properly described in dollar amounts (although that is exactly what conventional analysis does).
- Taxes (other than poll taxes, which are nearly non-existent in modern developed economies) are specified as a tax rate structure. Changing tax rates will have a non-obvious effect on revenue. For example, an increase in income tax rates might be modelled as causing an increase in tax revenue – but that requires an assumption about what will happen to taxable income. If the tax hike causes growth to slow, incomes would be lower than projected, and so the actual amount of revenue will be lower than expected. I would argue that the correct way to look at the problem is to look at the sensitivity of projected GDP to changes in tax rates. Unfortunately, that multiplier will not be as easily interpreted as the spending multiplier in model SIM.
- Welfare state spending programme expenditures depend on the state of the economy. The number of recipients of spending is not directly under the control of the government (putting aside the ability to change eligibility requirements). The government can change programme parameters – e.g., increase the length of time unemployment insurance payments may be drawn – but the actual effect on dollar expenditures is dependent upon the number of people in the programme. We need to keep in mind the counter-cyclical nature of the spending. If everything else is equal, welfare state spending will rise if the economy shrinks. This does not mean that the multiplier on welfare state spending is negative.
The conventional way to budgetary analysis is to look at projected budgets, and work with the projected dollar amounts. One can then use a straight multiplicative multiplier to see the projected effect on growth. However, we see that this is misleading except in the case of straight changes to non-welfare state expenditures.
Further ComplicationsThe other set of complexities with respect to multiplier is that not all taxes and expenditures have the same effect on the economy. We can see the following effects in simple macroeconomic models that are adapted to allow for these dynamics.
- Not all private sector entities will adjust their spending in the same way for the same dollar amount of spending. For example, an increase in interest income going to a pension fund will not prompt any immediate spending in response, rather the income flow will be hoarded. Similarly, richer households appear less likely to spend out of tax cuts than poorer ones.
- Similarly, the economic effects of tax increases will depend on who and what is being taxed.
- If the government purchases goods produced externally, there will be almost no effect on gross domestic product. (The increase in the trade deficit will cancel out the increased government expenditure.)
We would need a detailed macroeconomic model to capture low-level dynamics to properly model the effects of changing fiscal policy in the real world. Since such models are not particularly successful, we are stuck with approximations. This creates a great deal of fuzziness around “multiplier” estimation in practice.
My argument is that any relatively simple macro model that disaggregates the private sector will result in different fiscal policy changes having different sensitivities with respect to their effect on growth. It is only possible to generate a single multiplier in a model where we avoid any attempt at disaggregation (such as simple post-Keynesian SFC models, or neoclassical unitary representative household models without business sectors). Therefore, one should expect empirical multiplier estimation to be an extremely difficult task. As a result, I do not see much value pursuing the literature on multipliers within this text, since that will just tell us what we already can guess from first principles.
Concluding RemarksThe difficulty of estimating the fiscal multiplier means that it is normally difficult to gauge the effects of fiscal policy on growth. Although I believe that fiscal policy is more important than monetary policy, much of the effect is via automatic stabiliser effects. In practice, fiscal policymakers have generally shied away from aggressive fiscal policy adjustments since the early 1990s (the euro area periphery being a key exception, as will be discussed in Section 3.4). Correspondingly, the effects of fiscal policy have often been much less visible than was expected by somewhat sensationalistic market analysis.
Nonetheless, the basic multiplier story remains pertinent for the discussion of recessions. Sufficiently tight fiscal policy will always be enough to generate a recession, and one might hope that a well-timed fiscal boost could prevent recessions (as will be returned to Section 3.5).
[i] Monetary Economics: An Integrated Approach to Credit, Money Income, Production and Wealth (Second Edition), Wynne Godley and Marc Lavoie. Palgrave Macmillan, 2012. ISBN: 978-0-230-30184-9.
(c) Brian Romanchuk 2019