Monday, 28 October 2013

The ‘official’ cost of austerity

Well, not quite, but probably as close as we will ever get.[1] In a new paper, Jan in‘t Veld uses the European Commission’s QUEST model to estimate the impact of fiscal consolidation in the Eurozone (EZ) from 2011 to 2013. The numbers in the table below include spillover effects from other EZ country fiscal consolidations, so they are best interpreted as the impact of overall EZ fiscal consolidation over this period. There are at least two important things to note about the exercise. First, they do not attempt to analyse the impact of the particular mix between cuts in spending and increases in taxes applied in each country. Instead the ‘input’ is simply the change in the general government primary structural balance each year, which is assumed to be equally balanced between expenditure and revenue measures. (More on this below.) Second, to a first approximation this fiscal consolidation is assumed to lead to no change in short or long term real interest rates during the 2011-13 period.

GDP losses due to Eurozone fiscal consolidation (including spillovers) 2011-13. Source European Economy Economic Papers 506, Table 5.

Impact on GDP 2013
Cumulative Impact 11-13

Of course many would argue that had countries like Spain or Greece not undertaken this degree of austerity, long term interest rates might have been even higher than they actually were. (Perhaps short rates might have also been higher, if with stronger growth the ECB had raised short rates, but remember that tax increases also helped raise EZ inflation.) However a significant amount of fiscal consolidation took place in Germany, and this had significant spillover effects on other EZ countries. It is difficult to see why that consolidation was required to ease funding pressures.

One slightly surprising aspect of the exercise has already been noted. To quote: “As detailed information about the composition of the actual consolidations is not available, it is assumed the composition is equally balanced between expenditure and revenue measures.” As other institutions like the IMF publish exactly that kind of information, I’m puzzled. What the paper does report is that QUEST shows that consolidation implemented through spending cuts has about twice the short run multiplier as consolidation through higher taxes, but of course this is exactly what theory would suggest. In a forward looking model like this it also matters a great deal how agents perceive the permanence of these policy changes.

Of course QUEST is just one DSGE model, which just happens to be maintained by the Commission. An earlier study (pdf) by Holland and Portes at NIESR had important differences in detail, but the bottom line was similar: EZ GDP was 4% lower in 2013, and the cumulated GDP loss was 8.6%. These numbers are of course large, and so it is quite reasonable to say that the proximate cause of the second EZ recession is simply austerity.

Now many would argue that much of this was forced by the 2010 crisis. There seems to be a mood of fatalism among many in Europe that this was all largely unavoidable. I think that is quite wrong. Some fiscal tightening in Greece was inevitable, but if EZ policy makers had taken a much more realistic view about how much debt had to be written off, we could have avoided the current disaster. What ended the EZ crisis was not austerity but OMT: if that had been rolled out in 2010 rather than 2012, other periphery countries could also have adjusted more gradually. And of course fiscal consolidation in Germany and some other core countries was not required at all. If instead we had seen fiscal expansion there, to counter the problem of hitting the ZLB, then the overall impact of fiscal policy on EZ GDP need not have been negative. (Section 5 of Jan in‘t Veld’s paper looks at the impact of such a stimulus.) That means that over 3 years nearly 10% of Eurozone GDP has been needlessly lost through mistakes in policy. This is not the wild claim of a mad macroeconomist, but what simple analysis backed up by mainstream models tell us.

One final point. The UK equivalent to these ‘official’ numbers are the OBR’s estimates of the impact of fiscal consolidation on the UK. While they are significant in size, they are smaller than these EZ numbers. The OBR estimate that UK GDP in 2013 is about 1.5% lower as a result of fiscal consolidation, and the cumulated GDP loss due to fiscal tightening from 2010 to 2013 is a bit above 5%. There is a good reason and a bad reason for this difference. First, the UK is a more open economy than the EZ as a whole, and we would expect openness to cushion the impact of fiscal consolidation. Second, the OBR’s numbers are more crudely derived, based on multipliers that take no account of the zero lower bound or deep recession. The numbers from the QUEST model allow for both, which of course raises the size of the fiscal impact. 

[1] As the disclaimer says, “The views expressed are the author’s alone and do not necessarily correspond to those of the European Commission.” A good example of the official line is Buti and Carnot, but this line does not tend to be backed up by model simulations, which I think is revealing. 


  1. Great post, but a few queries regarding the OBR’s estimates of the impact of fiscal consolidation on the UK. As you indicate in this and a previous post, the OBR estimates the cumulative impact to be somewhere between 5-6% of annual GDP, approximately £100bn.
    The question is, what multipliers did they use and why?
    Why would they choose not to take into account "the zero lower bound or deep recession" in their multipliers, when those were the prevailing conditions?
    Furthermore, you mention above "consolidation implemented through spending cuts has about twice the short run multiplier as consolidation through higher also matters a great deal how agents perceive the permanence of these policy changes".
    Is it known whether the OBR incorporated these factors into their multiplier estimates, or is there reason not to?
    In light of these, what might the 'true' size of the impact of the Coalition's fiscal consolidation actually be?

    1. These are indeed great questions! I too would welcome answers. And I would add another. My impression is that in going back and assessing their OBR model's predictive failures, Chote & Co. chose to stand pat and recalibrate nothing. If true, under such circumstances, should that be considered standard, professional practice?

  2. I wish that governments would not include government spending as a part of GDP. The picture would become clearer as to which is which for those who do not understand the difference. I hear politicians say that it will cost us x number of dollars if the cut this much government spending. I consider that an abstraction in clear thinking.

    1. Most Government spending ends up moving money to the private sector, so you have to include it in GDP. GDP is "production", and companies produce goods and services to supply to the Government, which is really no different to supplying them to you and me.

      And yes, politicians talk about what this and that program "costs", but Government programs are not much more than taxpayers consuming collectively rather than individually.

  3. "Second, to a first approximation this fiscal consolidation is assumed to lead to no change in short or long term real interest rates during the 2011-13 period."

    Actually the assumption is even stronger and even more unrealistic than that. The policy rate is assumed to be at the zero lower bound (ZLB) throughout 2011-13.

    The MRO rate was 1.00% when 2011 started and was raised to 1.25% in April and to 1.50% in July. It was reduced to 1.25% in November and 1.00% in December. In July 2012 it was reduced to 0.75% and then in May 2013 it was reduced to 0.50%. There are insitutional differences in what constitutes the ZLB, but even now the ECB does not consider itself to be at the ZLB.

    The assumption that long term rates are constant is also contradicted by actual events. Greek long term rates fell from 29.2% in February 2012 to 10.2% in September 2013, Irish long term rates fell from 12.5% in July 2011 to 4.0% in September 2013 and Spanish long term rates have fallen from 5.6% in July 2012 to 2.5% in September 2013.

    Rather than constrained by the ZLB, ECB policy has been remarkably active during this period. In particular it's hard not to draw a straight line between the Euro Area's 6-quarter second dip recession that began in 2011Q4 and the increases in the policy rate the preceding two quarters.

  4. Simon,
    Your claim that "10% of Eurozone GDP has been needlessly lost through mistakes in policy" is of course not the conclusion I reached in my paper. You choose here to cumulate GDP level effects over three years and express it as a % of one year's GDP, but according to the model simulations EA GDP is around 4½ % lower in 2013. But aside from that, the finding that fiscal consolidations in the last three years have had large negative output effects does not imply that fiscal consolidations should have been avoided. Most economists would agree public deficits had to be brought back to lower levels, and the question was never one of principle but one of speed. Should the adjustment be frontloaded, or could it be postponed to later, when hopefully conditions have improved and multipliers may be lower ? When you speak of needless mistakes in policy, you would have to be more explicit about the counterfactual of no-consolidation. Allowing deficits to remain high would mean the required adjustment in the future would be larger, and the GDP costs of that later adjustment would depend on what you assume about the multiplier. I am not sure the multiplier will be much smaller in, say, 2016, as two of the reasons why multipliers are higher now will still hold: monetary policy cannot be accommodative for a small member in a monetary union, and financial constraints are unlikely to return to pre-crisis levels. But even if you assume smaller multipliers in the future, the option of postponing consolidations was simply not there for highly indebted countries in the EA, which faced pressure from financial markets, or in some cases had completely lost access to markets. A slower pace of consolidation could have raised general fears of sovereign default and such expectations of default could lead to worse growth outcomes than consolidations per se. It is of course much more difficult to quantify that, but we made an attempt to do this in Roeger and in 't Veld (2013), following the work by Corsetti et al. (2012).

    You suggest the 'official' line of the Commission is not supported by these model simulations. That is not true. The Commission has always acknowledged that consolidations have negative short term output effects, and that multipliers are larger now than in pre-crisis conditions. The main point of my paper was cross-country spillovers, and to show how consolidations in the core EA have affected the periphery. The Commission has always stressed the need for a differentiated approach, which takes into account both the fiscal space and the current economic situation in member states. Buti and Carnot (2013) also acknowledge that the symmetry of the adjustment is a legitimate concern. They underline that the adjustment in the periphery “should not be hampered, and ideally be fostered by concomitant changes elsewhere”. And the Commission's country-specific recommendations for Germany stressed the need to boost domestic demand. Specifically, it recommends to use the available scope for increased and more efficient growth-enhancing spending on education and research and calls for continuing efforts to accelerate the expansion of the national and cross-border electricity and gas networks.

    1. Jan
      Thanks for these comments, which I do appreciate. Can I respond on three points.

      1) The total cost of fiscal consolidation is the cost in all three years, not just 2013. I could have worked out the cost in Euros in each year, and summed, but that would just be a very large number. It seems natural to make sense of this by dividing by annual GDP.

      2) On how much of this cost could have been avoided, I acknowledged that some fiscal consolidation would have been required in the periphery. My argument was that it could have been much less, if default in Greece had been sooner and more extensive, and if OMT had been introduced at the same time rather than in 2012. For the Eurozone as a whole, that fiscal contradiction could – and should – have been offset by fiscal expansion in the Eurozone core. So, for the EZ as a whole, fiscal tightening could have been avoided, with the consequence according to QUEST that GDP would be 4.5% higher this year. This would have led to core EZ debt staying too high, but this should have been reduced once EZ interest rates were above the ZLB. At that point, again for the EZ as a whole, monetary policy could in principle have fully offset the output effects of fiscal consolidation.

      3) You are understandably defensive of the Commission’s position. You rightly point out that the Commission has suggested that some fiscal expansion in Germany would be helpful, and your own paper is in a similar spirit. In this sense, our views are not that different in qualitative terms. I just believe fiscal expansion in the EZ core, together with (2), could have been sufficiently large that EZ GDP could be now 4.5% higher. But this quantitative difference might partly reflect the fact that the Commission has to work within what is politically possible, whereas I can ignore these political constraints when I write.

  5. On the technical points, concerning the assumed balanced composition of consolidations, this is indeed a simplifying assumption. We only have cyclically-adjusted total deficit, total expenditure and total revenues. Using CA total expenditure and applying this proportionally to all expenditure components would have been equally wrong as applying the CAPB. Note that not all expenditure shocks have a large multiplier, and that for general transfers is similar to that of labour taxes (see Coenen et al., 2012).

    You also comment that to a first approximation there is no change in short and long term real interest rates. Just to be clear, monetary policy is assumed not to react to fiscal consolidations, and there is no change in the nominal policy rate in the simulations. The deflationary effects of consolidations do lead to higher real interest rates (this is in fact one of the reasons why multipliers are higher at the current juncture, and always for a small country in a monetary union).

  6. "First, the UK is a more open economy than the EZ as a whole, and we would expect openness to cushion the impact of fiscal consolidation."

    It also gives governments a lot less control over their economies. For example if austerity policies reduce demand enough to wipe out a home market (as arguably happened to the domestic manufacturing industry under Thatcher), they are more likely to be supplied from outside if there are no barriers to entry. Incentives to train domestic workers are also likely to be reduced once that industry and its support network disappears. This means when the economy recovers, they will be supplied by cheaper but trained foreign labour. The removal of capital controls is also likely to aggravate all of the above. A large protected home domestic market is your best cushion.

  7. "....and then a programme of real growth in areas like health and education."

    But that's the catch, isn't it? Labour are very good at talking "about" health and education - education, education, education, remember? - but they are not so good at managing outcomes.

    Anyone who takes your distinction literally, will really be distinguishing between the parties based on rhetoric, and not expected results.

    And that's the real claim. The claim from people like Brand isn't that that the Parties sound alike, but that if you vote for one or the other, the outcome won't be that much different.

    And by the way, I am not saying this to criticise either party, but to point to their helplessness. In the UK, Education and Health are the last two surviving massive "Stalinist" organizations.

    The only way to improve either of them would be a restructuring so radical that neither Party has the stomach for it. So they settle for what's really ineffective displacement activity. They can't improve the miserable standard of teaching in State Schools, so they give sub-par teachers pay increases and call them "Professionals", as if that helps.

    And where the NHS is concerned, all they can do is day to day management of crises and scandals, hoping that the newspapers will eventually get tired of stories about patients being abused and starved.

    The political Parties are helpless because beginning in 1945, Government got itself into managing enterprises that aren't amenable to Civil Service management. It did so with the best of intentions, because it saw the main problem as lack of access to good education and health care for people on limited incomes.

    And in 1945, with very rudimentary health care and quite basic educational needs, perhaps that was the correct choice. Today, however, we have learned that Government should not be in the business of mining Coal or making cars, but neither party is going to seriously consider whether Government should be running schools and hospitals.

    They'll allow the occasional maverick back-bencher to think aloud on the issue, but there will be no movement because it's too risky. On two core issues that affect every citizen, you can vote for whom you like, but you will have no choice of outcomes.