This is a note shared among colleagues which was a response to a graph in The Economist. In an article ‘British Class Politics is Back – With a Green Twist’ it argued that younger cohorts were experiencing poor economic growth prospects.
Specifically :’Recently published an analysis showing that lifetime levels of GDP growth have a significant impact on individuals’ political outlook. Cohorts that experienced periods of high growth are more likely to trust the government and have more positive perceptions of living standards. In Britain such voters are dying out (chart omitted). Today’s 75-year-olds witnessed average annual GDP growth of 2.4% in their lifetimes, compared with 1.4% for a 20-year-old. This might help explain why Green voters tend to be younger than Labour voters, and Reform voters tend to be younger than Conservatives.’
Here follows my contribution to the discussion it generated:
Dear Colleagues,
I am pleased that the Economist graph has triggered a useful discussion. Here follows my contribution. You may want to skip over the econometric paragraphs to the emboldened head: What does all this mean.
Simon Chapple was quite right to suggest that we would better look at per capita GDP. I had already written to The Economist suggesting they explore that. I also suggested they use working age population and hours worked in the denominator – measures of productivity.
I can confirm Shamubeel Equaab’s graph. I independently constructed a similar one with broadly the same profile.
What it means is unclear. You could, I think, argue that there was a downturn of some kind in 2000, perhaps similar to what may be The Economist finding, but with greater volatility arising from NZ being a smaller and more buffeted economy. (Incidentally, I was fiddling about with the series after 2015 and, as Simon warned, the short time and high volatility resulted in a garbled series.)
One issue which I have yet to look at is that these calculations result in recent growth rates which are high compared to my recall of a Treasury paper (and the Treasury long-term fiscal projections).
It took me some time to sort out my data. I am impressed by Shamubeel’s quickie on the phone. Congratulations to him. I suspect that Bill’s work involved my sort of tiresome fiddling around.
Bill Rosenberg’s work sort of supports the declining growth of GDP per cap thesis. ‘Sort of’ because many of his trend coefficients are not statistically significant although they are all negative as the hypothesis being tested posits – an R-square of above .25 over 60 observations is ‘significant’ at the 5% level. (Note that he is using a different measure to The Economist and Shamubeel.) The data shows a lot of volatility. It is, of course, easy to pick out the graph which supports one’s thesis (and to be clear, my preferred graph on conceptual grounds has the highest R-square – .35.)
Bill, looking at that graph are there any economies which differ markedly from the trend? Focussing on outliers is a dangerous game, but at least we ought to look at them.
What does all this mean?
I was struck by the abruptness of the downturn in the British series (beginning in about 1995). You don’t often get that sort of change. For instance, many years ago I observed a quadratic rather than an abrupt change in a series, which I eventually tracked down to the 1966 wool price shock. While it was abrupt, it took time to feed through into the data – hence the ‘quadratic’.
Conceptually, The Economist is using is an odd series because it accumulates future economic growth for each cohort. I’ve been trying to construct a model but thus far have not got one which is tractable – which throws any light on the abrupt change. It might suggest a spluttering winding down of the underlying growth rate.
What The Economist was saying is that recent cohorts have experienced lower GDP growth than earlier ones. (Yes it would have been better had they used GDP per cap.)
Observe the series is not about productivity change (which is what the secular stagnation hypothesis is about). The concern may be more about consumption – that younger generations are not getting the same growth in (public and private) consumption (assuming investment shares remain constant) as did their elders. (That also requires the assumption that generational shares are constant over time – they havnt been in the past – and something about intergenerational transfers which are too complicated to detail here.)
(A distinction which is probably not important to the Brits but critical to NZ is that the GDP measure is about production, which in constant price terms over time is not the same as effective GDP – the expenditure which production makes possible – when there is a shift in the terms of trade (so that relatively higher export prices makes a higher level of consumption possible and the reverse). Because the NZ TOT have had a secular trend favouring NZ food exports – unlike the Prebish thesis; see my Globalisation and the Wealth of Nations for an explanation – there needs to be an allowance for a TOT effect, if you want to use The Economist measure as an indication of consumption possibilities.)
There are two further possibilities not hinted by The Economist. One is that the politics needs economic growth because there is so much volatility in the distribution of output, that zero growth means there are ‘equal’ numbers of losers and winners. A little growth reduces the number of losers and the political discontent; much growth reduces the discontented even more. The second possibility is that GDP change has downsides as well as upsides and that one needs growth to offset the downsides. (For instance, pollution may increase with GDP and so some of the additional output needs to be set aside to deal with the pollution.) In effect GDP does not covert into wellbeing at low growth levels or high levels of affluence.
I conclude that perhaps The Economist measure is not conceptually robust enough to be of much significance. But its abrupt change in the trend requires further investigation because it says something about some important underlying processes.
<><>