How large a GDP loss under “Level 4”?

I’ve mentioned in posts over the last few days The Treasury’s estimate that the current “Level 4” restrictions will have reduced GDP by about 40 per cent relative to normal levels and the Reserve Bank’s estimate of a 35 per cent reduction.    Both estimates seem a lot more realistic than some of the private sector estimates that were still doing the rounds early in the week.  But I still reckon the Reserve Bank estimate, in particular, is almost certainly too low –  especially if we are concerned, as we should be, with the “true” reduction in the volume of economic activity, not with what SNZ may initially report (accurate measurement is going to be a challenge).

What makes me sceptical?  Well, there is the fact that our partial lockdown seems to be more restrictive, particularly (but not exclusively) on economic dimensions, than those in many other countries.  Thus ANZ yesterday ran this chart of some Oxford index of restrictiveness.

oxford response

Some weeks ago, the French national statistical agency published its estimate that the French economy was then running 35 per cent below normal levels.   Overnight, I noticed the Governor of the Bank of England endorsing the UK Office of Budget Responsibility’s view that UK output had fallen 35 per cent since their lockdown began – and not only will activity levels already have been weakening prior to that but, as the chart illustrates, the UK lockdown is less severe than New Zealand’s.

I also listened again to the Reserve Bank testimony to the select committee on Thursday and noticed that Assistant Governor refer to OECD estimates of a 30 per cent reduction as some sort of baseline estimate for New Zealand, but as I noted in a post here on those estimates they were done (a) for direct effects only, and (b) for a constant (across countries) set of assumptions about lockdowns, just distinguishing between countries by their initial economic structures.  In an initial lockdown period, indirect effects may not matter much, but different degrees of lockdown intensity certainly do.  As just one example, many countries have not prohibited construction work, but New Zealand (mostly) did, and construction alone is perhaps 8 per cent of New Zealand’s GDP.

I was also interested to see this footnote in the Treasury scenarios paper the other day

“MBIE estimates that the essential workforce numbers around 640000, with 510000 of those not able to work from home, while the number of non-essential workers that are unable to work from home is around 1.1 million people, or 49% of the workforce”

I’m not quite sure where they get those numbers from – according to the HLFS, for example, there were 2.648 million people “employed in the labour force” (which includes employers and the self-employed in December) –  but presumably there is some good basis for their estimates.  If they happen to be right that 49 per cent of the workforce is both non-essential and can’t work from home, that alone should produce a very substantial reduction in GDP –  even if less than half if we assume (as I do) that the people unable to work from home are in jobs that are, on average, lower productivity.  Even if the 1.1 million people unable to work from home is really over a denominator of 2.648 million, that is still 42 per cent of the workforce.

But even of those who are left –  able to work from home, or working on site as “essential” –   what do we suppose the average productivity level is like?  It is most unlikely to be normal levels.  And some of those essential services are being very lightly utilised at present –  think of hospitals.  I don’t suppose, for example, that immigration call centres are very busy.

In his presentation to the select committee, the Reserve Bank’s Assistant Governor ran through some sector-by-sector numbers to support his 35 per cent estimates.    The numbers he cited (no sources provided, but perhaps MBIE as well?) were:

Food and accommodation:  16 per cent of normal

Construction:                          25 per cent of normal

Manufacturing:                      60 per cent of normal

Primary:                                   76 per cent of normal

Government:                            80 per cent of normal

That seemed to satisfy Simon Bridges, chair of the committee, who responded “oh, so 30 per cent of the economy at 80 per cent?” and Hawkesby offered no dissent.

However, even though in normal times government revenue and expenditure are equal to about 30 per cent of GDP, the share of government activity (employment, production, purchases etc) is much smaller than that (lots of the rests is just transfers: welfare benefits etc).   The last detailed national accounts I could see on the SNZ website were for the year to March 2018 and in that year government as a whole accounted for only 16.8 per cent of GDP –  3.2 percentage points of that local government.

And even if there are significant chunks of government that are at least as busy as ever (Police, Ministry of Health, Treasury) there will be lots that just aren’t working at all (no work to do, no working from home capability, Level 4 prohibitions or whatever).  Just at a local government level, are libraries open, streets being swept, public gardens maintained, pools open? No, of course not.  Of those who are working, most will be working at less than normal productivity/levels of output (whether because working from home generally isn’t as suitable, distractions of children at home, or whatever).

I was also a bit surprised by the Bank’s assumptions around construction –  sure urgent repairs etc can be done, but at the peak of a building boom, can that have been anything like 25 per cent of the sector? –  and manufacturing, even allowing for the importance primary food processing plays in New Zealand manufacturing (building supplies, including cement, is also a large chunk of New Zealand manufacturing).

I went through the detailed (50+ sub-sectors) national accounts data and did some rough estimates myself.   There are clearly some sectors where there is no reduction in output at all (eg the imputed rent on owner-occupied housing, and actual volumes for public and private rents, and –  most probably –  agricultural production).  There are others where activity will have dried up almost entirely.  There is also quite a bit of uncertainty over all the numbers, but my bottom line was something between 40 and 45 per cent less GDP than normal under “Level 4” conditions.  That is a bit less than I had previously thought, and not that out of line Treasury’s estimates, but still rather higher than the Reserve Bank’s numbers.

There is no real certainty about any of it, but these numbers aren’t just of some obscure academic or historical interest.   Recall that a month’s GDP is roughly equal to $26 billion.   If we allow –  as Treasury does –  that even under “Level 1” there is a 7.5 per cent output loss, and that under Level 4 a 42.5 per cent output loss, that is a $9 billion loss (very little of which will ever be made back) for each month.     That overstates a bit, since the realistic choice at present seems to be between something very restrictive indeed, and some like “Level 2”, which will still have –  on both Treasury and Bank estimates –  incremental output losses.   And whatever governments mandate, there would still be activities citizens simply choose not to do.   But whichever way you cut it, these are still very big numbers that need to be factored in when the Cabinet makes decisions.

One day, when the Official Information Act finally forces the papers out of “the most open and transparent government ever”, we may finally see how their official advisers have attempted to do the sort of cost-benefit analysis that should have been being done at each stage of the process.