Why multipliers matter
“Nothing in economics is more potent than a simple idea whose time has come”, Gavyn Davies, October 21st 2012
The week before last, the Office for Budget Responsibility (OBR) published a new assessment of ‘fiscal risks’ to the UK. Reporting on threats to the public finances, they warned against new spending commitments:
“… new unfunded ‘giveaways’ would take the Government further away from its medium-term fiscal objective and would only add to the longer-term challenges”.
At the same time they made the observation that the “fiscal position … is more vulnerable than that which prevailed on the eve of the crisis 10 years ago”.
The message from the OBR seems to be that the patient must keep on taking the medicine – even though there’s no sign of it working yet. This assessment is based on an assumption that spending cuts are not damaging the economy, and that the weakness in growth since 2010 is nothing to do with government policy.
This rests on a judgement that the size of the ‘multipliers’ that capture the effects of government spending on the economy are very small. This was a controversial position when the OBR first adopted it in June 2010. But even more controversial is their refusal to debate or change this assumption in the light of increasing evidence that they have called this wrong.
In this blog I set out why multipliers matter, the latest evidence on just how big multipliers might be, how the OBR have reacted, and why this matters so much for the health of the economy and public services.
To Recap: the multiplier
The idea of the multiplier emerged in the 1930s great depression, as a way to explain the argument that increased government spending could pay for itself. The intuition is simple. Here’s how Keynes put it in an appeal to the public, published originally in the Times newspaper:
It is often said that in Great Britain it costs £500 capital expenditure on public works to give one man employment for a year … This is based on the amount of labour directly employed on the spot … but it is easy to see that the materials used and the transport also give employment … If the expenditure is additional and not merely in substitution for other expenditure, the increase of employment does not stop there. The additional wages and other incomes paid out are spent on additional purchases which lead in turn to further employment … Nor have we reached the end. The newly employed who supply the increased purchases of the employed on the new capital works will , in their turn, spend more, thus adding to the employment of others; and so on.
Keynes warns that “some enthusiasts, perceiving the fact of these repercussions, have greatly exaggerated the total result”. Acting against this ‘exaggeration’ are what he called “leakages” at each stage of the process into saving, imports and price.
While the idea is straightforward, capturing or illustrating it in theoretical terms is less so. The most common way is to consider the series of ‘repercussions’ with a leakage only to saving. So the original government spending increases employment and wages and so labour income. This increase is either spent (proportion c) or saved (s = 1 – c). The money that is spent is earned as income by the recipients, which is then spent according to the same average proportion (so a share c of c, which is equal to c2). And so on. Working through these repercussions leads to a total amount of income/spending equal to 1 /1-c times the original increase in government spending.
[NB the repercussions can be represented as a series 1 + c + c2 + c3 + …, which sums to 1/1-c by a standard mathematical result.]
Firms will react to this increase in demand by increasing production, with some of the increase will be met by imports. As discussed in an earlier post, taking into account imports the formula is 1/(1 – c + m), where c is known in the jargon as the marginal propensity to consume and m the marginal propensity to import. Both proportions can be estimated from average experience, using national accounts estimates of annual changes in consumption, imports and GDP. With c = 2/3 and m = 1/3 the multiplier is estimated at 1.5, i.e. if the government increases expenditure by £10bn, this will translate to an increase in GDP (i.e. overall income and spending) of £15bn. (And there will be tax gains and reductions in welfare expenditure, as discussed in the previous post.)
Strictly the (expenditure) multiplier is set in cash terms, but there is obvious interest in the share that goes to increased output and jobs (i.e. the ‘real’ effect’) versus the share that simply goes to price. The whole point of Keynes’s theory is that in conditions of deficient demand, a revival in demand will strengthen the economy; price may go up but as a secondary and generally not problematic consideration.
A brief history of the multiplier since the financial crisis
With the financial and economic crisis of 2007-2009 the worst since the 1930s, a revival of interest in multipliers was unsurprising. The most important action was after the election of President Obama. His Council of Economic Advisers (in a report ‘The job impact of the American recovery and reinvestment plan’) advised that the multiplier was 1.5 (though admittedly they were a little short on detail).
At the time, the FT (23/1/09) reported comments from a couple of prominent academic economists, with Ken Rogoff offering vague support:
“Academic economists are far more uncertain about the impact of the fiscal stimulus than Wall Street … The range of estimates is very wide. But given the situation we’re in it is certainly worth trying”
and disagreement from Robert Barro, who argued that the beneficial impact of additional public sector spending would be ‘crowded out’ by reduced private sector spend (I come back to this below), so that:
“… with partial crowding out the multiplier will be a lot less than one”
The implied uncertainties here were largely ignored by policy makers, who, within a year, began to impose public spending cuts – a.k.a. austerity.
The OBR were created by the Coalition to monitor the government’s progress in repairing the UK public finances. In order to make forecasts about the impact of the government’s policies, the OBR had to make their own estimates of the multiplier. They included the following table in their June 2010 Budget forecast.
The implications need to be understood. The largest category of government spending is departmental spending on public services and public sector wages. Here (‘RDEL’) the OBR operate with a multiplier of 0.6. This means for example that an increase of £10bn in public service spending will translate to an increase of only 7bn in overall GDP. This is only possible if other expenditures fall to offset the increase in government spending (namely investment, trade and consumer demand). The idea is that government spending effectively gets in the way of private sector spending and makes the economy less efficient overall – so private spending is ‘crowded out’ by public spending.
In a footnote, the OBR offer background to these estimates, giving prominence to work by the IMF.
A review of estimates for fiscal multipliers for different policy instruments and countries is available in Fiscal Multipliers, Antonio Spilimbergo, Steve Symansky, Martin Schindler (IMF Staff Position Note), May 2009. Further evidence was taken from papers including: Fiscal Policy Action in the Banking Crisis, National Institute Economic Review, January 2009; Fiscal Stabilisation and EMU, HM Treasury, 2003; Public Investment and the Golden Rule: Another (Different) Look, Roberto Perotti (IGIER Working Paper No 277), 2006; and Estimating Tax and Benefit Multipliers in Europe, Ali J Al-Eyd and Ray Barrell, Economic Modelling (Vol 22), 2005.
Yet within only two years even the IMF were alarmed at the scale of the damage to global growth. Their October 2012 World Economic Outlook included analysis of the impact of fiscal consolidations to date – underpinned by recognition that their forecasts of economic growth had gone furthest astray for those countries making the biggest cuts. Effectively they U-turned on previous policy advice and offered new much higher multipliers:
Our results indicate that multipliers have actually been in the 0.9 to 1.7 range since the Great Recession.
At the time I was working in the Treasury, enjoying the reaction to the IMF findings; my predecessor at the TUC Duncan Weldon justly wondered how the OBR would react – given the importance of IMF work to the original OBR assessment:
The question now is will the OBR follow the course set by the IMF, reassess its own use of multipliers and revise its view of the impact of spending cuts and tax rises?
Gavyn Davies’s (of Goldman Sachs and sometimes Labour Party adviser) reaction is at the top of this blog. But the OBR were undeterred. Some five years later Ben Chu at the Independent challenged outgoing member of the Budget Responsibility Committee Sir Stephen Nickell on underestimating multipliers.
“The IMF had four different measures of multipliers, depending on which branch [you talked to]…. The Great Britain team had multipliers very similar to ours. Olivier [Blanchard – the former chief economist of the IMF] wanted to have much bigger multipliers.”
And Nickell stresses that he is still “perfectly happy” with the OBR’s original judgements on this front.
From theory to practice
We now have several years of data about the impact of spending cuts in the UK with which to test these slightly theoretical arguments.
Underlying the OBR position is the idea that private sector activity is ‘crowded out’ by public sector activity (as in the 0.6 example above). Looked at the other way around, cutting back public sector activity should ‘crowd in’ private sector activity, i.e. lead to stronger economic growth. The OBR original economic forecast was for 2010-15, so we can now see how things turned out.
The chart below shows the OBR forecast against growth before the crisis and then against what actually happened after the crisis. It does so by taking the average GDP growth across each period, and examining how different types of spending were affected. (NB While so far the illustrative calculations have been in terms of cash increases, in reality the spending cuts and multiplier have operated on the growth of the economy.)
- Column 1 shows pre-crisis growth
- Column 2 shows the OBR forecast for growth over 2010-2015
- Column 3 shows the difference between the first two columns. They key point is that the OBR expected investment and trade to be ‘crowded in’ by the cuts in government spending – with green and yellow above the line offsetting the red below the line – so that GDP growth was almost unaffected.
- Column 4 fourth column is what actually happened to growth over 2010-15: it was significantly weaker than expected.
- Column 5 shows the difference between the OBR forecast and outcomes in reality. ‘Crowding in’ didn’t happen and consumer demand (pink) was also hit hard. The OBR argue that their forecast was knocked off course by the impact of the Eurozone crisis on the UK economy. Though trade was actually a little less bad after the crisis than before.
Contributions to GDP growth, annual averages in percentage points
Source: TUC calculations on ONS and OBR figures
In fact, comparing pre-crisis and post-crisis outcomes, (nominal) GDP growth slowed by 1.8 percentage points a year (from 5.4 to 3.6 per cent a year). The contribution of government expenditure was down 1.3 percentage points. Applying to this a multiplier of 1.5, suggests a slowdown in economic growth of 1.9 percentage points (= 1.3 x 1.5). So the actual reduction of 1.8 ppts a year was very close to what might have been expected on the basis of normal experience of the relation between government expenditure and economic activity as a whole.
Moreover ‘crowding in’ didn’t happen in any of the 32 OECD countries where cuts were imposed (here). And the evidence across these countries suggests an average multiplier of 3 – with cuts in government expenditure hitting economic growth very hard. It is not plausible to blame the Eurozone for this global disaster (not least because Germany actually did better after the crisis than before). It is hard to see how this evidence effectively of a (very brutal) natural experiment can simply be ignored. The blame is with the policy approach in the UK, EU and most countries of the world.
The OBR’s report was shortly followed by the circulation of a new report published by the US National Bureau of Economic Research. The author Gabriel Chodorow-Reich finds a US multiplier of 1.7 / 1.8 – working bottom up from regions to the country as a whole.
Long standing critics of austerity were quick to draw attention to the findings. Simon Wren-Lewis tweeted how he had previously assumed 1.5, and so it “looks like I could have gone higher. OBR need to rethink their multipliers”. And Catherine Mann, chief economist at the OECD, backed him up “Collective action generates bigger multipliers. Indeed”.
A proper debate on multipliers is now seriously overdue.
As I discussed in the previous post, a higher multiplier opens up the possibility that the correct way to improve the public finances is to increase rather than reduce spending. While the Chancellor plays the politics of division pitting the public against the private sector, it is perfectly possible that expanding public sector activity will lead to stronger private sector activity. In the meantime, erroneously operating with a too low multiplier means that the economy is being badly damaged for no sound economic reason at all. The consequences of getting multipliers wrong are serious.