Wednesday, September 18, 2013

France's Demographic Advantage

That France has a "demographic advantage" over Germany is well-known, but Paul Krugman (via the OFCE) has data and graphs showing just how big a role demographics will play in the years ahead.

Unfortunately for François Holland, the rewards will come too late.


Passerby said...

I don't want to sound like I'm engaging in Hollande-bashing (a popular sport in France lately), but to be fair it's not like he is going to be deprived of any rightful reward.

If I may use this expression, the seeds of demographic growth have been planted long ago, and Hollande's policies have nothing to do with it. Quite the contrary, his successive budget policies have been disincentivizing having more children.

Up-to-now social benefits were designed to increase financial rewards for families with more than 2 children. This is pretty much over after:
- a first reduction of the "quotient familial" this year
- a second reduction of the "quotient familial" next year
- the announced "fiscalisation des majorations de pension pour famille nombreuse" next year.

bernard said...

comparative demographics with Germany have been a French obsession at least since the first world war.

It is no coincidence that modern demographic projections were actually started by a Frenchman, Alfred Sauvy, back in the late 1920s or early 1930s as I recall. He, using techniques very close to those used today (apt feminine calculating ladies have been replaced by computers), predicted a France of 45 million fifty years ahead, ie by 1980. Of course, there was a slight error in the projection as the population turned out to be some 30% higher.

Not to worry, the famous World Bank projections of the early 1960s also turned out to be "slightly" off, and the hilarious projections made for France in the very early 1990s also painted a declining population fifty years hence. I personally prolonged the model projection another 50 years for fun and announced to the French insurance industry that we were returning to our middle age population a century ahead, which would be good for the environment and for those inclined to love a forested if not entirely active France. I remember ending the conference with the suggestion that demographic projections should be taken with a large pinch of salt.

So, when I see the projection of a German population collapsing, the only question I ask myself is what will be the definition of a German in fifty years and whether perhaps that definition will involve very much less ethnic criteria. Because, you see, nature abhors vacuum. The true news here: they are not about to experience real estate mania, but we French might!

Anonymous said...

To Bernard,

sauf erreur de ma part, in 1980 the poupulation of France was 53 millons.

So it would be a 18% error, not 30%

Not so bad, definitely better than what most economists do.

I do not believe the German projection tooted there, but 18% pmargin of error would be impressive.

bernard said...

to anonymous,

the first point is that when calculating a "forecast" error, standard practice is for one use the forecast as the denominator rather than the actual turnout: population is x% higher than expected, rather than population projected is y% lower than actual turnout. But you're right, I did mix up some number from quoting them from memory. Correction below, which does not change the argument I was making.

My mistake. The population was actually 54 millions in 1980.
Second mistake, his 1929 central projection (see ) was for a 38 million population in 1956, thus 25 years ahead, which was just about the 1929 population according to his article. In reality, the 1956 population turned out to be 44 millions for so-called Metropolitan France, ie not including colonies, dependent territories etc. Thus the error over 25 years is about 15%, which is actually huge. The following document gives a very good hint at what happened with the starting graph at the front of the document: population started growing rapidly in 1946 as opposed to the stagnation that Sauvy was expecting. This is due to a combination of factors obviously: mortality decline (antibiotics invention etc., he could not have guessed), renewal of fertility rates (he did not expect this to happen), net immigration flow (he sort of suspected it might happen, but didn't like it much and did not project in his central case).

I should not put words in Sauvy's mouth of course, but is seems clear reading his 1929 article that had he made a 50 year projection rather than a 25 year projection, he would have projected a population of the same order of magnitude as the 38 millions for 1956, if not lower. The tone of his article is very, very pessimistic.

My point of course was not to denigrate or accuse Sauvy of anything - he was a huge social scientist-, it was to point out that long term demographic projections, while incredibly seducing and powerful in public opinion and politics, almost always turn out to be very far from what actually happens. We know this for a fact from past projections "urbi et orbi" if I may say so.

Of course, people will hope that projections made today for the distant future are of a better "quality", but I would suggest that chances are actually pretty low for this to be the case because the technology of population projection has not changed all that much at all on the one hand, and because the same intractable uncertainties apply to critical factors affecting projections. Think for instance about the sudden arrival of HIV in the eighties, the impact of mass medical care in the long run, the ebb and flow of immigration which is particularly hard to predict (who would have predicted before the financial and european crisis that massive flows of young south european workers would rush totally legally to Germany now?).

In my personal view long run demographic projections are in the end both extremely unreliable and a tool for ideological purposes, really.