I have recently had discussions with a PhD student of mine who was interested in exploring the cyclical link between productivity growth and the economic cycle in the context of the intergenerational debate about ageing and the challenge to improve the former. The issue is that sound finance – the mainstream macroeconomics approach – constructs the rising dependency ratio as a problem of government financial resources (not being able to afford health care and pensions) and prescribes fiscal austerity on the pretext that the government needs to save money to pay for these future imposts. Meanwhile, the real challenge of the rising dependency is that the next generation will have to be more productive than the last to maintain real standards of living and if austerity undermines productivity growth then it just exacerbates the ageing problem. My contention has always been the latter. That governments should use their fiscal capacity now to make sure there is a first-class education and training system in a growth environment to prepare us for the future when more people will have passed the usual concept of working age. This question also is hot at the moment in the Brexit debate in Britain and in this blog post I offer some empirical analysis to clear away some of the myths that the Remainers have been spreading.
In August 2019, the Bank of England released a Staff Working Paper No. 818 – The impact of Brexit on UK firm – uses the Bank’s Decision Maker Panel dataset, to “assess the impact of the June 2016 Brexit referendum on labour productivity, the principal causal mechanism being conjectured is that rising uncertainty has “gradually reduced investment by about 11% over the three years following the June 2016 vote”, which then undermines productivity growth.
In that paper, they make several claims, many of which are unable to be statistically verified, even when using their own results.
For example, they seek to “examine the impact of Brexit exposure on firms’s investment and employment” and say that, in relation to employment, that:
… the results are all negative but none are statistically significant. Therefore, whilst there is some tentative evidence that the Brexit process has led to lower employment, the finding is less robust than for investment.
That statement cannot be made.
The fact is that the regression estimates relating the Brexit variables in their model to employment are, in statistical terms, not different from zero – meaning there is no statistically robust negative impact.
Their results establish that there has been zero impact according to their dataset.
So any further attempts “to quantify the magnitudes of the reductions in … employment that have occurred in anticipation of Brexit” are spurious and should be disregarded.
Another example, is that they:
… report the IV estimates for the impact of the Brexit process on firm level productivity. In all cases we see a negative impact, which is significant in the OLS estimates, but not in the IV estimate.
What does that mean?
IV is the estimation technique, instrumental variables, which is deployed instead of OLS (ordinary least squares), when there is the prospect that the explanatory variables are jointly generated by the same processes that generate the variable to be explained.
Which means that OLS is an invalid estimation technique and IV is used to produce what we call ‘consistent’ estimates. In this case, if the IV estimates in question are not ‘statistically significant’, then there is no relationship discovered.
So the rest of their analysis in this regard is highly questionable.
Which means that the Financial Times article (August 31, 2019) – Brexit has cut UK productivity by up to 5%, says BoE – is unjustified in claiming that:
Brexit has caused British companies exposed to Europe to cut investment plans significantly and has cut UK productivity by between 2 and 5 per cent, according to Bank of England research.
At least without drawing attention to the econometric issues that the paper raises.
I was, however, interested in the Bank’s Survey Data that the paper used.
The most recent data is for – June 2019.
The Bank says that:
The panel comprises Chief Financial Officers (CFOs) from small, medium and large UK companies operating in a broad range of industries and is designed to be representative of the population of UK businesses. The panel has grown rapidly since its inception in August 2016, reaching around 8,000 firms by June 2019. The response rate has been around 40% in recent surveys.
The results are interesting:
1. In September 2016, 37 per cent of the sample said that Brexit as a source of uncertainty was just “One of many sources” that proportion is now (between February 2019 and April 2019), 33 per cent.
9 per cent said it was the ‘main source currently’ in September 2016, and now 21 per cent. A year ago it was 8 per cent. So the rise in uncertainty is more about the process it seems.
2. Between November 2016 and January 2017, there was a 73 per cent probability derived from the sample that the ‘Expected impact of Brexit on foreign sales by 2020’ would either increase or have no material impact (40 per cent for the latter).
Between May 2018 and July 2018, the probability of no material impact had risen to 48 per cent (and 62 per cent for the increase or no impact).
The main rise in the reduced sales probability was in the “less than 10% category”.
3. In terms of the ‘Expected impact of Brexit on capital expenditure over the next year’, the average probability in October 2016 was 57 per cent (no impact), 18 per cent (less than 5 per cent reduction), 16 per cent (more than 5 per cent reduction).
The most recent estimate (November 2018 to January 2019) has those probabilities at 59 per cent (no impact), 18 per cent (less than 5 per cent reduction), 15 per cent (more than 5 per cent reduction).
So the negative outcomes are stable and the no impact probability, despite all the chaos and mishandling of the process has actually risen.
That brings into question the Project Fear scenarios.
Either the respondents are lying or the Financial Times is exaggerating.
4. The average probability (%) of an ‘Expected impact of eventual Brexit deal on sales’:
– February 2017 and April 2017: 39 per cent (Make little difference), 13 per cent (Adding less than 10%), 6 per cent (Adding more than 10%).
– February 2019 and April 2019: 51 per cent (Make little difference), 8 per cent (Adding less than 10%), 4 per cent (Adding more than 10%).
A substantial rise in those who don’t think there will be an issue.
5. Most of the other Brexit-related questions have provided no data until now.
One interesting result (for which there is just one observation) was that the between February 2019 and April 2019, 70 per cent of the respondents said that the “Impact of Brexit so far on overall investment” had “No material impact”.
Now clearly, if we are talking about changes in investment behaviour then a stable majority does not exclude negative results emerging if the minority are negative.
6. The Survey also provides measures of uncertainty both across and within companies by employment, sales and price growth.
The fact is there is hardly any movement in any of these measures since the first Survey (November 2016 to January 2017). If uncertainty was on the rise then
Project Fear continues
Project Fear continues, albeit in a slightly more subdued manner, given, I suppose because of the massive predictive failures of the initial ‘the sky is falling’ campaign that the Remainers assailed the British people with.
Now, there is some circumspection disguising the disaster scenarios.
First, note that the latest GDP figures from the Office of National Statistics for the July National Accounts update
Gross domestic product (GDP) grew by 0.3% in July 2019.
The services sector (around 80 per cent of the total GDP output) grew by 0.3 per cent, and manufacturing and construction increased by 0.3 per cent and 0.5 per cent, respectively. The Index of Production rose overall by 0.1 per cent.
A ‘swallow does not a summer make’, meaning in this context, we should never rely on a single month’s observation.
But the Project Fear story that disaster is unfolding due to Brexit seems to keep getting inconveniently derailed by the actual data.
Second, an example of the more cautious version of Project Fear was the UK Guardian article (September 5, 2019) – The promised Tory tax cuts will only mean more austerity in the long run – written by the mainstream economist Simon Wren-Lewis.
I don’t wish to critique the whole article because it runs the ludicrous argument that if the British government uses fiscal stimulus now to offset the damage that Brexit might cause, eventually, and without doubt, further spending cuts will have to be made and austerity will be resumed.
He also claims that any plans to cut taxes would be dangerous because any “fiscal stimulus should be temporary, so that any consequent increase in the deficit comes to an end”.
So underpinning his logic is a belief in a balanced fiscal state, seemingly irrespective of the state of the non-government spending and income balance.
With an ongoing external deficit, if the British government targets a fiscal balance and succeeds in achieving that, then it is forcing the private domestic sector, already heavily indebted, to run deficits equal to the external deficit.
If the private domestic sector resisted that pressure and wanted to, for example, save overall, then the government pursuit of fiscal balance would drive the economy into recession.
But then the mainstream economists seem oblivious to these sorts of dynamics and fail to understand that for a country with an external deficit, the most likely sustainable position for government is to run on-going deficits in order to finance income growth and private domestic overall saving aspirations.
The point I want to focus on that came up in that article, and is a growing narrative among mainstream economists, is his claim that:
… the Bank of England … suspects that current weak growth reflects a deterioration in supply, not a lack of demand. A recent Bank of England study suggested that the collapse of investment since the referendum may have reduced productivity by between 2% and 5% from what it otherwise would have been. A fall in productivity of that size could well result in today’s flat growth.
Here we have the problem of causality.
First, has there been a collapse in British investment? And if so, has this been the result of the Brexit referendum?
The first graph shows the Total British investment ratio (sum of Business and General Government capital formation) as a per cent of GDP from the first-quarter 1997 to the first-quarter 2019.
The dotted read lines are the average ratios before the GFC and after and for the Post-Referendum period (from the September-quarter 2016).
It is clear that a mean shift occurred during the GFC in capital formation in the UK. There is also some evidence of a slight further reduction in the ratio post Referendum.
But this is because the rise in GDP (the denominator) has outstripped the numerator (Capital formation).
The investment ratio is, on average, higher in the Post-Referendum period than for the Post-GFC period overall.
The next graph shows the Business investment ratio as a per cent of GDP from the first-quarter 1997 to the first-quarter 2019. Clearly given the dominance of business investment (more than 3 times general government expenditure), this series drives the dynamics for the previous total series shown above.
The following facts apply when we concentrate on real expenditure (rather than the ratio):
1. Real expenditure on Total Capital formation rose, on average, by 3.4 per cent per annum from 1997 to the March-quarter 2008.
2. In the Post-GFC period it rose by 0.9 per cent per annum on average.
3. In the Post-Brexit period, it rose, on average, by 2 per cent per annum.
4. Since the June-quarter 2016, that is, in the Post-Referendum period, real business investment has slowed but is still, on average, growing. Thus trying to tell UK Guardian readers that it has ‘collapsed’ is simply false.
5. It is true that growth in real Business investment has contracted in the last three quarters but real capital formation in the general government sector has accelerated, ensuring total investment in the British economy has continued to deliver positive growth.
The next graph shows the evolution of real capital formation from the March-quarter 1997 to the March-quarter 2019, where the index is set to 100 in the March-quarter 2008 (so at the turning point of the cycle).
I deleted the observations for the second-quarter 2005, which saw a huge upward spike in private business investment and a corresponding fall in public capital formation as a result “of the transfer of British Nuclear Fuels Ltd (BNFL). In April 2005, nuclear reactors were transferred from BNFL to the Nuclear Decommissioning Authority (NDA). BNFL is classified as a public corporation in National Accounts, while the NDA is a central government body. The business investment series includes investment by public corporations but not government spending, with the positive spike therefore reflecting the £15.6 billion transfer.”
The GFC stimulus and fiscal austerity is clearly shown (red line) as is the sharp decline in the growth in business investment and the slow recovery post trough.
But real Business investment started to stall in 2014 as the fiscal austerity started to really impact on the economy. That is some years before the Referendum and reflects poor conditions in the economy overall, rather than angst over leaving the EU.
Why does this matter?
Claims that productivity slumps are the reason for slow growth almost assuredly put the cart before the horse.
It is clear that capital formation slumped dramatically as the GFC hit and drove the collapse in real GDP growth.
Spending creates income (and output).
For more analysis of this issue, see the blog post – British productivity slump – all down to George Osborne’s austerity obsession (October 18, 2017).
The consequences the investment behaviour are not solely the short-run spending loss. Capital formation adds to the supply-side of the economy and is a major driver of productivity.
So we get a more likely causality – GFC erodes confidences, capital expenditure drops, GDP drops as a consequence, and the slowdown in capacity building impacts negatively on productivity growth.
Solution – stimulate spending.
The first graph shows the annual growth in output per hour worked (labour productivity) from the March-quarter 1971 to the March-quarter 2019.
The red horizontal lines are the average rates of growth Pre- and Post-GFC.
The mean shift in growth occurred as a result of the GFC and the poor policy response that followed.
The next graph shows this data in a different way by tracing the evolution of output per hour worked (labour productivity) from the March-quarter 1970 to the March-quarter 2019.
The dotted line is an extrapolation of the Pre-GFC trend rate of growth.
The productivity growth slump happened as a result of the GFC. Brexit has really had nothing to do with it (although it is plausible it may turn out to be a second negative shock on an already flat series).
The final graph zooms in on the Post-GFC period (from the December-quarter 2007) to get a better feel for the behaviour of productivity in this period.
We see the decline as the GFC erodes output growth, the improvement as the fiscal stimulus starts to work to buttress renewed growth, then the George Osborne austerity push from mid-2010 creating further negative productivity shocks and then the subsequent, slow, drawn out recovery.
Don’t be fooled by the vertical scale. The overall period is best described as being very flat growth. Overall, the index has risen only 1.5 index points since the December-quarter 2007 – in other words hardly at all.
But almost all that increase has come in the Post-Brexit period, which further casts doubt on Project Fear’s disaster scenarios.
The point is that the productivity slump was driven by the GFC.
Structural explanations of this slump are also unlikely to have traction. The massive cyclical contraction pushed British productivity growth of its past trend. Structural factors work more slowly and we would not witness such a sharp fall if they were implicated.
Why would that cause productivity growth to slumplike it did, then fail to recover?
A major driver of productivity is investment spending – both public and private.
While business investment is cost sensitive (so may respond to interest rate changes), mainstream economists usually ignore the fact that expectations of earnings are also important as are asymmetries across the cycle.
Cyclical asymmetries mean (in this context) that investment spending drops quickly when economic activity declines and typically takes a longer period to recover. So fast drop and slow recovery.
The cyclical asymmetries in investment spending arise because investment in new capital stock usually requires firms to make large irreversible capital outlays.
Capital is not a piece of putty (as it is depicted in the mainstream economics textbooks that the students use in universities) that can be remoulded in whatever configuration that might be appropriate (that is, different types of machines and equipment).
Once the firm has made a large-scale investment in a new technology they will be stuck with it for some period.
In an environment of endemic uncertainty, firms become cautious in times of pessimism and employ broad safety margins when deciding how much investment they will spend.
Accordingly, they form expectations of future profitability by considering the current capacity utilisation rate against their normal usage.
They will only invest when capacity utilisation, exceeds its normal level. So investment varies with capacity utilisation within bounds and therefore productive capacity grows at rate which is bounded from below and above.
The asymmetric investment behaviour thus generates asymmetries in capacity growth because productive capacity only grows when there is a shortage of capacity.
This insight has major implications for the way in which economies recover and the necessity for strong fiscal support when a deep recession is encountered.
These dynamics are covered in my 2008 book with Joan Muysken – Full Employment abandoned.
In some of my earlier work (with Joan Muysken) – for example, here is a working paper you can get for free (subsequently published in the literature) – we developed a model based on the notion that investors facing endemic uncertainty make large irreversible capital outlays, which leads them to be cautious in times of pessimism and to use broad safety margins.
The problem of asymmetry can be attenuated if government steps in during a downturn and arrests the spending collapse. Appropriate fiscal stimulus initiatives can thus shorten any non-government spending declines and limit the investment slump.
This not only shortens the decline into the activity trough but also means that potential output growth can be maintained at previous rates.
The opposite is of course also the case.
Imposing pro-cyclical fiscal austerity of the scale that George Osborne initiated when the Tories came took government in May 2010 is the last thing a government should do when non-government spending is in retreat.
Fiscal austerity in these circumstances exacerbates the typical asymmetry associated with investment expenditure and is a major reason why business investment in the UK has been so weak.
We often focus on the short-term negative impacts of fiscal austerity, but in this case, it also has serious long-term impacts on both the rate of business investment and the potential growth rate (which falls as capital formation stalls).
The longer it takes for business investment to recover, the worse will be the long-term impact on potential GDP growth. In turn, this means that the inflation biases are increased because full capacity is reached sooner in a recovery – often before all the idle labour is absorbed.
So, while George Osborne is long gone, the negative impacts of his policy folly will reverberate for a long time to come. His failings will continue on for many years and the flat productivity growth is one manifestation of that failing.
The Brexit issue is clearly a source of uncertainty.
But it looks from the data to be of a second-order of importance relative to the on-going damage that the GFC and subsequent fiscal austerity caused.
Carping on about Brexit as the source of all ills in the British economy is a popular pastime for the urban elite Remainers.
But it is unlikely to produce a very meaningful narrative to help understand what is going on at present.
It also leads to ridiculous articles in the UK Guardian that suggest fiscal policy will not be very helpful in redressing the damage that poor fiscal conduct over several years, largely generated.
That is enough for today!
(c) Copyright 2019 William Mitchell. All Rights Reserved.