Lies, damn lies and statistics - from covid to GDP data

 

In today’s blog I want to talk about the abuse of statistics in the media with a focus on UK GDP. We have got used to the media blasting us with statistics during the covid-19 pandemic, as journalists with little statistical knowledge use the data without interpreting or analysing them properly. The best example relates to the covid data, whether it is cases, hospitalisations or deaths. Much of the media has used them for their own agenda to attack the government to portray it as incompetent and the sole reason why in their eyes the UK has had a “bad pandemic”.

 

The reality which has become clearer over time is that there are a plethora of reasons for the UK’s experience including mistakes made by the UK government but also by the health authorities (NHS and PHE), by scientists and as a result of underspending on health over several decades. However there is also the health of the nation in respect of obesity and related diseases such as diabetes, cardiac disease, cancer, asthma etc with the UK considered one of the worst in Europe. Then there are non-health factors such as population density and global connectivity, where the UK and particularly London rank at the very top. And finally the data is not compatible on a country-by-country basis anyway with countries using different definitions of for example when a person dies of covid (think Spain with its very narrow definition of a covid death when compared with the UK or other European countries) or because a country is either incapable of recording all or the vast majority of deaths (think India, Mexico or Brazil or indeed most African countries) or deliberately under-records such data (think China, Russia, Iran or North Korea). The WHO estimates that up to 8 million deaths can be attributed to covid globally, which would represent a massive under-recording on the official 3.4 million death toll.

 

In the case of the UK, this under-recording is a relatively small number. In any case latest official data shows the UK does not even make the top 15 countries on deaths per capita, with many mostly East European, and some West European and Latin American countries comprising the top 15. And if we focus on excess deaths per capita which can overcome some of the data issues previously mentioned. the UK is not even in the top 25 worst performing countries for which data exists – The US, Russia, Brazil, Mexico, South Africa, Italy, Portugal and a plethora of East European and other Latin American countries are all above us!

 

However my blog today intended to concentrate on GDP and in particular in challenging the media consensus that the UK government’s “bad pandemic” was also reflected on the economy and in particular in GDP growth. One frequently still reads headlines saying that the UK economy was the worst performing country in the EU or the G7.

 

For 2020, official ONS data shows that UK GDP fell by 9.9%, which by the end of Q4 left the economy some 7.8% below its level prior to the pandemic at the end of 2019. This would represent a collapse without modern precedence – even during the Global Financial Crisis, UK GDP fell by a mere 4.3% in 2009. Moreover, Eurostat data shows us that EU GDP fell by an average 6.1% (and Eurozone GDP by 6.6%) with only Spain, down by 10.8%, worse affected than the UK. However the compilation of these statistics is not comparable as I will explain below. Once adjusted for a fairer comparison, GDP in the UK is estimated to have fallen by a more modest 7.5% leaving it in a better position than not only Spain, but a number of other EU countries including Italy, France, Greece and Croatia.

 

The need for a sizeable adjustment has arisen because of the different ways the UK and the EU treat so-called non-market output - approximately the output of the public sector. GDP, which for all its faults remains the single best summary of national economic activity, effectively measures the aggregate output of all sectors of the economy. This is easily measured in the case of the private sector where we can look at the turnover of a business. But in the public sector, particularly in health and education, it is very difficult to measure output. In the UK, the ONS overcomes this by looking at a wide range of detailed data such as the number of pupils attending school, the number of hours teaching in schools, the number of people visiting a GP, and the number of operations taking place. These measures were introduced to better measure not only output but to assess productivity in these sectors and whether the UK was getting good value for money for any increased government spending.

 

During the pandemic, schools were closed for a prolonged period and many operations were postponed. As a result we saw a big fall in the output of public services here in the UK. Indeed some 2.4 percentage points of the 9.9% fall in GDP in 2020 can be attributed to a fall in public sector output. Within public output we saw Education output fall 16.4% and Health & Social Work output by 8.2%. Nevertheless the UK data greatly exaggerated the fall in output so the ONS has been adding new indicators since to take account of pupils working from home as well as test & trace activity and covid vaccinations, although they are yet to be applied to 2020 data – only current monthly data

 

By contrast in many other countries such data is still not available or not used by national statistical agencies (despite the urging of various international institutions) so they estimate the output of such services by measuring the input ie looking at how much money was spent on them even if fewer services were provided. This is the approach taken by nearly all EU countries and meant we did not see the dip in the output of public services that genuinely did happen during 2020 throughout most of the EU.

 

Another useful measure to use that largely overcomes these intra-country comparison issues is nominal GDP growth, although they miss important volume movements seen in real GDP data. Nevertheless, we can see for the G7, the UK GDP fall is in fact less than in Germany, Italy and Canada, albeit higher than in France. In the US and Japan where the lockdowns were far less severe, overall economic activity was affected much less. So the bottom line is that UK GDP was badly affected by the pandemic particularly as our lockdown was notably long and tight, but the fall was not the worst by any means in the EU or G7.

 

In my next blog I will talk about UK GDP in 2021 and the expected sharp bounce-back in growth which is likely to be the fastest in the EU or the G7. 

Quantitative Easing and its dangers in an inflationary environment

Does a large budget deficit and the associated public debt burden really matter that much if the Bank of England finances it through asset purchases?

 

It is generally agreed during this pandemic that the government should spend vast sums of money to maintain income levels and prevent unemployment rising as much as possible so avoiding potential “scarring” of the economy ie a permanent reduction in productive economic output. There are also times dependent on the state of the economic cycle but also when investment in infrastructure and other forms of capital spending including on humans that means it may be appropriate to run a significant deficit.

 

If a budget deficit is financed through the issuance of government bonds which are purchased by investors (pension funds, insurance companies, households and overseas investors) then the degree to which one can do this successfully will depend on the existing policy credibility of the country, how much more is to be borrowed, the potential growth rate and the global background including the level of world long term interest rates. If a country has to borrow at very high interest rates (whether it is from abroad or not) then the debt burden could escalate and the interest payments will account for an ever higher portion of government spending so potentially squeezing out other spending.  Indeed a simple equation used in economics proves the debt burden will likely “snowball” ever higher if the (real) average interest rate of borrowing is greater than the (real) GDP growth rate.

 

In the UK, the government was in a very strong position to borrow coming into the pandemic aided by strong policy credibility thanks in part to its actions to tackle the deficit previously as well as by the very low global long term interest rates. Moreover, it has not had to borrow from abroad, and the borrowing has been very long-term so avoiding any rollover risk, reducing interest rate sensitivities and maximising the use of low borrowing rates now. Average debt maturities are around 13 years albeit on the decline.

 

It was conventional theory (as I learned it) that a country should borrow to fund any fiscal deficit (i.e. debt financing) if it can rather than finance it by printing money (ie monetary financing). Printing money was a sure way to generate inflation and lead to all the adverse consequences associated with inflation. The extremes of such a path which resulted in hyperinflation were of course seen in the Weimar Republic and in more modern times in Zimbabwe and Argentina.

 

However there are now new arguments which are complex and controversial, that says Bank of England purchases of assets, principally government bonds, known as Quantitative Easing (QE) may be a sort of free lunch. It is also referred to as “unconvential” monetary policy. As a result, increasingly, the UK deficit (and elsewhere) is being financed by QE. It still entails the creation of (electronic) money known as Reserve Money which is part of MO base or narrow money. However, it might not translate into M2 broad money (i.e. remaining largely within the financial sector) preventing it from largely entering the real economy and which is used to buy goods and services, and therefore affecting inflation. Such funding is approaching one-third of the gross public debt burden and being in the form of central bank reserves they only pay an interest rate of some 0.1% much lower than gilt yields.

 

The empirical evidence from the first big wave of QE after the global financial crisis (GFC) is that QE was not inflationary with M2 growth not growing excessively. However, the GFC was a deflationary phenomenon. And particularly important was that banks did not create new bank loans so preventing the operation of the classic money multiplier – they preferred to shore up their own balance sheets and regain profitability. 

 

However, there are real dangers that this will not be the case this time after the pandemic. The world is a potentially much more inflationary place after the initial deflationary impact of the pandemic itself. This reflects explicit coordination of expansionary monetary and fiscal policies (led by the US, but also the UK and the EU), the increased bias it seems of  central banks towards inflation including higher inflation targets (underpinned by a reluctance to raise bank rate given the massive amount of QE), the move back to de-globalisation (and no favourable China deflationary effect) and the disruption of international production chains. Indeed evidence is already concerning. Most notable and most importantly is the US, where M2 growth exceeds 20%yoy, input prices and commodity prices have surged, there are supply constraints and shortages (even in the labour market), and retail price inflation is also now picking up markedly. Nevertheless, some prominent economists still think inflation will not persist with classic “wage-price spirals” not taking off. I am not convinced.

 

So to sum up QE is really an experiment. Under the right circumstances it looks like it can play an important positive role, but we are now in different times than after the GFC and there is a real risk that inflation will not only rise to higher levels but stay there without tough action.