Thread by @stuartbasten

Yesterday, I shared a few thoughts on the *coverage* of the new #GBD future #population estimates.

Now, I want to spend some time thinking more about their numbers and how GBD present them. SPOILER: I think they are being a bit disingenuous.

THREAD (as they say) (1/x)

I think that this 'implication is deeply problematic and a gross over-simplification. It takes a very deterministic view of the population which is not especially helpful (or scientifically astute). Anyway, let's focus more on the method)... (1a/x)

Lets cut to the chase and start with the interpretation. The 2nd and 3rd sentences here are just self-evident. The study adds nothing new to provide these statements. So let's focus more on the first one. While not self-evident, it's certainly an an 'orthodox view' (2/x)

(2 asides:
1. These are not interpreting anything, frankly.
2. It's not just *access* to contraception; but acceptance and education about use (for women AND men); empowerment to use them; and a link through to changing norms in family, relationships etc.
Anyway moving on)
(3/x)

There is a rather large difference, though, between asserting a *link* between X and Y and stating that X is GOING TO HAPPEN.

(4/x)

However they dress it up, the reference scenario is still grounded in a *qualitative assumption* that these 'trends' which we suppose to affect fertility rates will continue into the future. The other scenarios represent slower or faster manifestations of this.
(5/x)

But, it is really important to remember that this *reference* scenario is NOT the same as a 'business as usual scenario'.
(6/x)

Many of the countries with the highest fertility rates have the most limited resources to invest in (quality) education. As such, it is quite possible that recent gains in education (and the associated demog, social, econ things) may slow as more and more kids need school) (7/x)

Such disruptions to the 'relentless progress' of female educations *do* exist, and they have been linked back to stalls in fertility transition. (See this nice piece in @PNASnews) (8/x)

In other words, we might (normatively) *hope* that improvements in X might occur (which would then accelerate changes in Y), but it is far from guaranteed. (9/x)

The assumptions behind the reference scenario, and the fact that this IS the reference scenario, necessarily lead to the conclusion that there will be the so-called 'jaw-dropping' 'crash' in fertility. (10/x)

The reference scenario, then, is just *a* scenario. The medium variant of the UN WPP is also *a* scenario. GBD makes a big fuss about 23 countries which 'more than halve by 2100'. But, under the low variant scenario of the UN projections fully *48* territories will halve. (11/x)

Remember the low/high variant is medium variant TFR +/- 0.5. This makes a BIG difference over the long run, especially in low fertility countries. This scenario gives S Korea a TFR of 0.68 in 2035; and Singapore stays <1.0 until the 22nd century. (12/x)

Of course this low variant is possible; but it ignores not only policy interventions which may occur, but also possible feedback effects (which could intensify, slow or reverse fertility decline). (13/x)

BUT THE POINT IS THAT IT IS JUST AN ASSUMPTION TO MAKE A PROJECTION. The UN approach is clear. Higher fertility countries will broadly follow a path of lower fertility countries and low fertility countries will follow an adapted path. All this is modelled statistically. (14/x)

(Sorry, @Patrick_Gerland @LeontineAlkema @AdrianRaftery1 for the butchering of your method there. I'm just making a point. Full methods here: population.un.org/wpp/Methodolog….

It is a POSSIBLE trajectory based on a REASONABLE assumption. (15/x)

In essence, I feel that the GBD projections (and their media dissemination) could have been MUCH clearer about the uncertainty which surrounds the qualitative assumptions which the reference model is based on. (16/x)

To be fair, knowing that projections are 'active' agents in shaping the future, I think it would be just as irresponsible to make the worse scenario the reference and create a narrative around explosion and chaos. Both approaches risk harming women and reproductive rights. (17/x)

Now, most of the general public would now say "aha, he's only getting so riled about this because he was involved in the @WiCVienna Projections' (because both the projections and I are so very famous). The paper makes a big deal about how they are 'better than us' (18/x)

(I was mainly involved in the low fertility module. AS mentioned above there are some issues in the #GBD paper about how we manage fertility change in 'lower fertility' countries, but perhaps I'll come to that tomorrow...) (19/x)

So there are, apparently, 7 ways which the GBD projections are better than ours/UN. To be sure, we didn't calculate CCF50. (20/x)

We also did not calculate uncertainty intervals (although given that @IIASAVienna @WiCVienna and @UNDESA staff were/are at the forefront of probabilistic projection methods that seems a bit rum. (21/x)

However, we were inclusive and engaged members of all of regional population societies. We asked people on the ground what *they* thought about the future of *their* society (or the society they study). (22/x)

At workshops held around the world, we worked with experts to interpret these data and try to create holistic 'storylines' (based on the @IPCC_CH model). These storylines were then blended with a statistical model and other assumptions to generate scenarios. (23/x)

We spent an enormous amount of time justifying these scenarios; documenting and reflecting on the process and so on. Look, here is a picture of the book which goes through the whole thing in detail (and this is just a small part of the literature we produced). (24/x)

Then colleagues at @WiCVienna created this amazing 'data explorer' allowing anyone in the world to explore these different scenarios AND WHAT THEY WOULD MEAN FOR THE FUTURE OF DIFFERENT TERRITORIES AND THE WORLD AS A WHOLE
dataexplorer.wittgensteincentre.org/wcde-v2/ (25/x)

So to go back to Japan, you could see what these different storylines meant for the future by year. (I think SSp1 is the best :) (26/x)

I think a big advantage of our approach is to say "look, these are some possible future soc/pol/econ trajectories of the world. Given what we know about the links between these things and demographic change, these are plausible outcomes of these trajectories". (27/x)

What this tries to do is empower the reader/user to 'explore' some possible different demographic futures, and allow them to consider what kind of future they/we might like to see. Then we can think through, together, how we might get there. (28/x)

In other words, it is less 'OMG jaw-dropping fertility in the future and the shit is going to hit the fan' but rather 'these are some plausible future directions; how can we work towards more preferable ones and away from less preferable ones'? (29/x)

In that qualitative sense, I don't feel that you really need uncertainty intervals in such scenarios. The storyline is the scenario. And that's it. such uncertainty intervals around the scenario just create a series of overlapping futures which may not really mean anything.(30/x)

So to sum it up (cos I've got other things to do with my life than moan about this all day). (31/x)

The future is uncertain. If we assume the future of demographic change is tied up with the 'future of everything else', then the future of demographic change is equally uncertain. We can present these uncertainties probabilistically; or with scenarios. (32/x)

They just serve different purposes for different people. However, as demographers, we have a real responsibility to say how we are representing the future, and what underlying assumptions we are making to get there. (33/x)

We must recognise that our words and actions have consequences; that projections are active agents in shaping the future. We must take great care in our 'interpretations' - and perhaps assume the worst. (34/x)

In a world where reproductive rights are under threat; where women are harassed and cajoled into having more/fewer children; where access to family planning is being limited... (35/x)

Where ageism is rife...(36/x)

Where xenophobia and racism are part of everyday life...(37/x)

We have to be crystal clear in our methods, our communication, and our interpretation. Otherwise, we just are contributing to the problems, and are not part of the solution. (38/x)

So sorry that this is a sprawling, messy mess. In future, I will write a nice briefing note for @hkustiems and circulate it!

FIN.