A.  Introduction

The “fiscal multiplier” (often referred to as just the “multiplier”) is simply the ratio of how much aggregate GDP will increase for a unit increase of fiscal spending.  Hence if fiscal spending increases by say $100 and aggregate GDP increases by $200 in response, the multiplier is equal to 2.  The concept is also often applied similarly to tax cuts of some dollar amount.

Under conditions where there is significant unemployment in an economy, an increase in government spending can be expected to have a multiple impact on GDP.  There will be a direct contribution to GDP from the increased production to provide for the demand from government, but also an indirect contribution as those being paid for the initial goods (whether newly employed workers or suppliers of inputs to the production of the good) will in turn spend at least some portion of their higher incomes on other goods or services in turn.  And this process will continue in further rounds.

While the concept is simple, the multiplier in practice is difficult to measure.  It is not a constant, but rather a definitional concept whose value will vary depending on the specific economic circumstances of the time and place.  It has also been controversial, as some economists both historically and even currently do not believe it is possible for an economy to be functioning at less than full employment.  For such economists, higher production from an increase in demand is not possible since the economy is already at full employment, and the multiplier must then always and everywhere be uniquely equal to zero.

But most economists recognize that it is possible for the economy to be at less than full employment.  This is especially clear today in most of the developed world, including in the US, Europe, and Japan, with unemployment high in each of these countries or regions.

The real debate, then, is about the size of the multiplier in a particular situation – whether it is low or high.  If low, then fiscal stimulus will not have much of an effect on increasing GDP, while fiscal austerity will not lead to a big reduction in GDP.  If high in contrast, fiscal stimulus will be quite effective in raising GDP, while fiscal austerity will lead to big reductions in GDP and consequent large increases in unemployment.

Recent work at the IMF, a conservative institution, on the size of the multiplier has brought this debate into the general news.  In particular, in June the IMF published a self-evaluation of the IMF supported (as well as EU and ECB supported) economic program in Greece.  It noted there (on page 21) that the fiscal multipliers assumed in that program turned out, based on actual experience, to have been too low.  This self-criticism was picked up in the general press, and many have questioned how the IMF (and the others) could have gotten this so wrong.

But judging the size of the multiplier in a particular place and in particular circumstances is not easy.  This Econ 101 blog post will discuss why the multiplier will vary in different countries and in different country circumstances.  And while it might be understandable how the multiplier might be misjudged ex ante in some concrete case, what is outrageous is not that initial misjudgment.  What is outrageous is that the policies that had been taken based on that earlier misjudgment were not then revised or reversed to reflect what had been learned.

B.  Why the Multiplier Will Vary

As noted above, the multiplier is not a constant, equal to the some particular value in all countries and under all circumstances.  Rather, it is a concept, expressing a relationship (between changes in GDP and changes in government spending) which will in general vary across different economies and across different circumstances in any particular economy.  Hence even if one had a good estimate of what it might be in one particular country under particular circumstances, one should not assume it would have that some value in another country or even in the same country under different circumstances.

Specifically, one should expect:

1)  The multiplier will vary across countries, depending on the size and structure of those countries:  In a large country such as the US, an increase in spending (both direct and indirect) will be met primarily by supplies originating in the US.  The multiplier will then be relatively large.  In contrast, higher spending in a small and open economy, such as Monaco to take an extreme example, will be met primarily by supplies originating elsewhere.  The multiplier will then be relatively small.   Most economies are in between these two in size, and one would expect the multiplier then also to be in between these two in size.

Note that this will depend not only on the size of the economy, but also its economic structure (the type of goods produced within that economy, as opposed to imported) and the nature of its trade regime.  Some economies are more open than others.

2)  The multiplier will vary depending on the current state of the economy – how far or close the economy is to full employment:  If unemployment is significant, an increase in demand can be met with an increased supply of goods, and an increase in employment of workers to produce those goods.  The multiplier will be relatively high.  In contrast, if the economy is at a time of close to full employment, an increase in demand for certain goods can only be met by reduced production of something else (with a shift in jobs from the latter to the former), so overall output might not rise by much.  In such circumstances the multiplier will be relatively low.

Hence if one had a good estimate of the multiplier in some particular economy at a point in time when the economy was close to full employment, one would greatly underestimate what the multiplier would be in that same economy at a different time when unemployment was high.

3)  The multiplier will vary depending on the form of the fiscal stimulus:  Fiscal stimulus programs can take the form of spending on newly produced goods (such as infrastructure), or on transfer programs to households (such as higher or extended unemployment benefits), or on tax cuts or tax rebates.  But while each might have a similar direct dollar impact on the fiscal deficit, the impact on GDP could vary widely.

Direct government expenditures on newly produced goods, such as new roads or school buildings, will likely have the largest impact on GDP.  The newly produced goods will, with certainty, be produced, and such product is a direct component of GDP (GDP stands for Gross Domestic Product).  And those newly employed to produce such goods (e.g. construction workers) will also then spend most or even all of their new earnings on goods they need.  The multiplier will be high.

The multiplier will also likely be relatively high on transfer programs that go to the unemployed and others who are relatively disadvantaged, as they will spend what they receive on goods that they and their family very much need. The multiplier will be less on transfer programs that benefit those who are better off (such as certain farm subsidies, for example, when they mostly benefit large and relatively well-off corporate farms), as such individuals or firms will likely save a higher share of such receipts.

And the multiplier might be quite small for tax cuts or tax rebates that go to upper income households, as they will likely save much of what they receive.

Hence the size of the multiplier will depend on the nature of the fiscal stimulus program.  Programs focussed on the direct production of goods, especially labor-intensive goods (such as the building and maintenance of much of infrastructure), or on transfers to the relatively less well off, can be expected to have a relatively high multiplier effect.  Programs focussed on transfers or subsidies going to the relatively well off, or tax cuts that accrue primarily to the relatively well off, can be expected to have a relatively low multiplier effect.

4)  The multiplier will vary depending on whether the stimulus (or austerity) programs are temporary or expected to be sustained:   Temporary tax cuts or tax rebates are a common component of stimulus programs, in part because they can be implemented quickly and easily.  However, households receiving a temporary tax cut or a one-time rebate will normally simply save a high share of what is distributed to them (or use the funds to pay down outstanding debt they might have).  The multiplier will then be relatively low or even negligible, as there would be little increased demand for goods to be produced.

5)  The multiplier will vary depending on the direction of change:  Many make the simplistic assumption that if the multiplier has some value for an increase in spending or for a tax cut, one will see the same value for the multiplier for a decrease in spending or a tax increase.  But there is no reason to assume this will be the case.  People will in general respond differently if facing an increase in income (such as from a tax cut) or a decrease (such as from an equal tax increase).  With a tax cut, the households might simply save most of what they receive, resulting in a low multiplier.  But with a tax increase (which one might see as part of an austerity program, for example), the households might be forced to scale back their consumption to pay the higher taxes, resulting in a relatively high multiplier when going in this downward direction.

Similarly, the multiplier impact when a worker is newly hired as a result of a stimulus program will likely be different than the multiplier impact when a worker is laid off as a result of an austerity program.  The multiplier impact is likely to be substantially greater (in the negative direction) when workers are laid off as such workers will likely be forced to scale back their consumption substantially.

6)  The multiplier will vary depending on the policy response of others:   While the government might launch a stimulus program, other economic actors might respond with policy changes of their own.  For example, a Central Bank might raise interest rates when the government launches a stimulus program, due perhaps to a concern on inflation (possibly a mis-guided concern, but nonetheless what they are acting on).  Raising interest rates would lead to a cut in investment, and hence the impact of the stimulus program on GDP might be constrained.  The multiplier would then be low.

Importantly, the ability of the Central Bank to respond by lowering interest rates to a cut-back in government spending, to offset what would otherwise be the contractionary effects of such a cut-back, is important to recognize and take into account.  In times like the present in the US, Europe, and Japan, when the interest rates set by the Central Bank are essentially at zero and cannot go lower, a cut-back in government spending cannot be offset by a cut in interest rates (interest rates are already as low as they can go), and the multiplier will be relatively high.  The fiscal contraction will lead to a large reduction in GDP.  In contrast, if the fiscal contraction is delayed until the economy is closer to full employment, with interest rates then positive and significant, the impact on GDP of a cut-back in government spending can be offset at that point by the Central Bank lowering interest rates, and output will not then fall.  The multiplier will at that point be close to zero.

This has extremely important implications for the design of fiscal adjustment programs.  There may well be a need eventually to reduce public debt to GDP ratios, by cutting back on government spending or increasing taxes.  But if this is done when there is significant unemployment and the Central Bank controlled interest rates are at or close to the zero lower bound, then the fiscal austerity programs will reduce demand and lead to a large fall in GDP (and consequent further rise in unemployment).  One should instead maintain fiscal demand until the economy has recovered sufficiently that one is close to full employment and interest rates are no longer at or close to zero.  At that point, a cut-back in government spending (or an increase in taxes) can be offset by the Central Bank through its management of interest rates, and GDP need not then fall.

Unfortunately, the US, Europe, and until recently Japan, have been doing the opposite since 2010.

The financial markets are another economic actor which can have an impact.  For example, in economies where the foreign exchange rate floats, the foreign exchange markets might respond to a stimulus program with a devaluation of the foreign exchange rate.  This devaluation would make exports more competitive (thus spurring production of exports), and imports more expensive (thus encouraging production of domestic substitutes for what had been imported), which would be expansionary.  The multiplier in such circumstances would then be relatively high.

7)  The multiplier will vary depending on the time frame:  So far we have not made any note of the time dimension, and have implicitly treated all the responses as taking place simultaneously.  But the time dimension does matter, as it takes time to implement programs, and then time for the multiple round responses to work themselves out.  Hence one should be clear on whether one is referring to the multiplier as the response in, say, the current quarter of a year, or over the next year, or over the next several years, or what.  The multiplier will be relatively low if measured as the impact on GDP in the current quarter, fairly large over the next year, and then begin to diminish thereafter.  And one then needs to be clear if one is referring to the multiplier in terms of the impact on GDP only within a certain period, or the cumulative impacts over a multi-year range.

C.  Conclusion

The multiplier is important, and a good deal of work has been done over the years to try to measure what it might have been in a particular time and place.  But factors such as those listed above have not always been taken into account when economists (including at the IMF) and analysts have sought to apply those results.

It has unfortunately been the case, for example, that estimates of the multiplier found when the economy was close to full employment, were then assumed to be similar when the economies at some later time were in a downturn and far from full employment.  Or cross-country differences have been ignored when the multiplier found for some small economy, say, was assumed to apply equally to a large economy.  Or the multiplier that might apply in an expansion resulting from a stimulus program was then assumed to apply similarly in a contraction resulting from an austerity program.  Or no attention was paid to how the multiplier will differ in a stimulus program depending on whether one is looking at new infrastructure work, or transfer programs, or tax cuts.  Or the multiplier for tax cut programs was treated as the same whether the tax cuts were going to the relatively poor or the relatively rich.

This has not always been the case.  Some economists and analysts have been careful.  But there has also been a lack of attention to these issues.  This does not mean one should ignore the multiplier, but rather that one needs to work with care.