Measuring the impact of austerity

Ben Chu has a good article disposing of some of the nonsense ideas associated with austerity (which refuse to die, because they are useful to politicians, and much of the media is generally clueless). Perhaps the most silly, which I encounter a lot, is that the UK has not really endured austerity because debt has been increasing, or some other irrelevant measure has been rising.

If trying to reduce the deficit - what economists call fiscal consolidation - had no adverse effects on the economy as a whole it would not be called austerity. Austerity is all about the negative aggregate impact on output that a fiscal consolidation can have. As a result, the appropriate measure of austerity is a measure of that impact. So it is not the level of government spending or taxes that matter, but how they change.

An obvious measure to use is the change in the deficit itself, generally adjusted for changes that happen automatically because output is changing. I have used that measure many times, because it is produced by the OBR, IMF and OECD among others. But it is not ideal, because the impact of changes in taxes on demand and therefore output is generally smaller than the impact of a change in government spending, because some of any tax increase comes from reduced saving. (This is also true, but perhaps to a lesser extent, of government transfers.)

There is no simple way of dealing with this measurement problem, because the amount of any tax increase people will find from their savings will depend in part on how long they expect taxes to be higher. As a result, some people prefer to focus just on government spending to measure fiscal impact (although the data you will easily find is government consumption, and as fiscal consolidation normally involves cuts to government investment it is important to add that on). However it is also possible to apply some simple average propensities to consume from tax cuts and transfers to get a fiscal impact measure.

This is what the Hutchins Center fiscal impact measure does for the US.

These are not multipliers (so are different from what the OBR does for the UK, for example [1]), but just the direct impact of government spending and taxes on aggregate demand and hence GDP. The average total impact is something like 0.4%, so this would be fiscal policy that was in this sense neutral.

Compare the mild 2001 recession with the much larger 2008/9 recession. In both cases during the recession fiscal policy was strongly counter-cyclical, helping to reduce the recession’s impact. After the 2001 recession ended, fiscal policy continued to support the recovery for around two years: these were the Bush tax cuts. The recovery in GDP was reasonably strong: growth from 2003 to 2005 of 2.8%, 3.8% and 3.3%.

In 2010, we had a much deeper and longer recession, but the fiscal support was only marginally greater than 2001, despite interest rates being stuck at their lower bound. On this occasion fiscal support was strongly opposed by the Republicans. It continued for another year and a quarter, and then became strongly contractionary from 2011 to 2015. GDP growth was slower than in the previous recovery, despite the deeper recession: from 2010 to 2014 2.5%, 1.6%, 2.2%, 1.7%, 2.4%. This is not surprising, as fiscal policy was reducing GDP by around 1% during 2011,2012 and 2013, rather than adding the normal 0.4%.

The speed and extent to which austerity was applied after the Great Recession was very unusual: the textbook says secure the recovery first, allow interest rates to rise, and then worry about government debt. There was no economic justification for switching to austerity so quickly after 2010: the motivation (as in the UK) was entirely political. It produced the slowest US recovery in output since WWII. (This is a very useful resource in comparing US upswings.) As I showed here using simple calculations, if total government spending from 2011 had remained neutral instead of becoming sharply contractionary, US output could easily have got close to capacity (as measured by the CBO) by 2013.

To subtract 1.5% from GDP would not matter if something (consumption, investment or net exports) filled its place. But that will only happen by chance or because of a monetary policy stimulus, and monetary policy was stuck in a liquidity trap. This is the real crime of austerity. Decreasing demand and output just when the economy is beginning its recovery from the deepest recession since WWII is as foolish as it sounds, but to do this at just the time that monetary policy was unable to effectively fight back is macroeconomic madness. As I will argue in later posts, it looks increasingly likely that this has made us all permanently poorer.

[1] If somebody publishes similar estimates for the UK, please let me know. Personally I think it makes more sense to publish data like this than use a multiplier based analysis, simply because these measures are more direct, and involve fewer ‘whole economy’ assumptions. Crucially, there are no implicit assumptions about monetary policy being made. It would be interesting to know why the OBR decided not to take this approach.

UK Growth has been even worse than we thought

That is one headline on the Office for National Statistics (ONS) latest data revisions. Output in the UK economy is now estimated to be currently almost 4% below its previous peak, compared to previous estimates of 2.5% below. Or alternatively, the headline could be that the UK never had a double dip recession: at the beginning of 2012 growth was flat rather than falling by 0.1% (not annualised), a 0.1% that has been reallocated to the subsequent quarter. The chart below shows the old and new data for GDP growth, quarter on quarter. So GDP went fall, flat, fall, which technically is not a recession. I’ll leave you to decide which the more informative headline is.*

As you can see the big revision is in how much GDP fell in the recession. GDP is now thought to have decreased by a little over 5% in 2009 as a whole, compared to the previous estimate of -4%. At this point I cannot resist telling a small story about this number, but for those who are fed up with my personal anecdotes there is a serious point about inflation to follow. I make a weak attempt to connect the two at the end.

Quarter on quarter changes to UK GDP (not annualised): ONS

At the beginning of 2009, I was asked to attend a breakfast meeting with the then Chancellor, Alistair Darling, along with some non-academic economists. I had never attended one of these before, so I did not know what to expect. I had not met Darling, but all the other economists invited appeared much more comfortable with the format and surroundings, so to be honest I was rather nervous. Academics in particular can appear out of touch because they do not have all the latest data at their fingertips.

Sure enough, one of the first questions Darling asked was just how bad we each thought things could get. I cannot remember what each person said, but the general view was that GDP could fall by as much as 3% in 2009. I was the last to give my opinion. I could have ducked out, but instead I remembered one thing from my earlier days as a forecaster. This was that forecasts typically underestimate the extent of large swings in GDP, particularly if they are globally synchronised. So I said that I thought things could be worse than that, and GDP could fall by 5%.

Impossible! was the immediate retort of one of the other economists: someone who is very well known and very sensible, although I will not say who it was here. This person then used their detailed knowledge of the data to say why it was inconceivable that GDP could fall by so much. One by one everyone else agreed that although things were bad, they could not get that bad, and 5% was an outlandish number. Just as I wished I had kept my mouth shut, or better still just not come, the senior economist from the Treasury who was there came to my defence: a fall that large could happen, and they described how it might happen. I of course take no pleasure in the fact that my forecast has been vindicated, and it was little more than luck, but it is one of those moments I will not forget.

Now for something more consequential. The chart below compares two different measures of UK inflation: the CPI (green) and the GDP deflator (blue). CPI inflation has been significantly above the 2% target since 2010. In contrast over the last year growth in the GDP deflator has been well below 2%. This is the deflator at market prices, so it includes indirect taxes. The dashed line is the GDP deflator at basic prices, which excludes these. The press release only includes numbers going back to 2010 for this series, but you can see that growth has been below 2% for the last three years. The dotted line is growth in the US GDP deflator - this moved in a more immediately understandable way after the recession, but over the last two years the UK and US measures have not been that different.

Alternative measures of inflation

The fact that output price inflation (which is what the GDP deflator measures) has been below CPI inflation is neither surprising, nor unique to the UK. What is less appreciated is that there is no reason from an economic point of view to focus on one series (the CPI) rather than the other (the GDP deflator) when setting monetary policy. At an intuitive level looking at the output price measure makes more sense, because policy has more control over things produced in the same country. At a deeper level, inflation matters because some prices are sticky, and the GDP deflator generally excludes volatile commodity prices. It should be less influenced by volatility in the exchange rate, so it may be better for that reason too.

I cannot help but reflect on how different UK monetary policy might have been if it had focused on output prices rather than consumer prices. In 2011 interest rates were almost raised (3 out of 9 MPC members voted for doing so), despite the lack of a recovery. Would this have happened if the target inflation measure had been below 2%, as growth in GDP at basic prices was? Since then Quantitative Easing has largely stalled, which would have been very hard to justify if the focus had been on output prices.

One of the reasons often given for focusing on the CPI (which has come up again in discussion of nominal GDP targets) is that this data is available quickly and is not revised. [1] Which brings me back to the beginning, because the GDP deflator numbers for the first quarter of 2013 and earlier have been significantly revised (and are smoother as a result). I have never understood this argument. We should start with why inflation is costly, and then think about how best to measure these costs. If measurements change because information gets better, policy should respond to that. If that causes problems, improve the measurement. Perhaps policy needs to obsess a bit less about this bit of data or that, and think more about the fundamentals of what it is trying to do. 

* I changed the text here from the original version to make the nature of the adjustment clearer. As one economic journalist put it, reallocating 0.1% of GDP between quarters makes no difference in terms of the economics, but revising away the double dip recession will play well for George Osborne politically. I think that says a lot about the quality of political debate.

[1] Another argument is that the CPI is easily understood by the non-economist. If this impresses, why not use wages rather than the CPI, as I suggested here. As wages are clearly sticky, there are good theoretical reasons to focus on this as a measure of inflation.