Facebook Blogging

Edward Hugh has a lively and enjoyable Facebook community where he publishes frequent breaking news economics links and short updates. If you would like to receive these updates on a regular basis and join the debate please invite Edward as a friend by clicking the Facebook link at the top of the right sidebar.

Wednesday, May 16, 2007

The Low Fertility Trap

In a recent post on Demography Matters Claus Vistesen has offered us a nice summary of the latest revision of the low fertility trap hypothesis as advanced by Wolfgang Lutz et al. As summarized below, there are three basic components to this hypothesis:



As Lutz says the key idea is that once fertility falls below a certain level (and even in the event that the hypothesis proved to be well founded this level could only be determined empirically, on the basis of actual experience) a self-reinforcing demographic regime may be established from which it is hard to escape, in the sense of raising fertility back up towards replacement levels. The cut-off point which Lutz et al start from is 1.5 (and in this they take their lead from a proposal by Peter Macdonald in this paper ). This figure does seem to have some coherence in terms of actual experience to date, since with the exception of Denmark - which did briefly fall under 1.5 tfr in the 1990s - no country seems to have gone below it and come back up again.

The explanatory mechanisms we are offered are full of self-reinforcing feedback processes, as can be seen from the diagram below (incidentally for a better look click over the image):




As I said, the whole process is described in terms of three basic components (LFT1,LFT2 and LFT3 respectively). The first of these - the negative momentum component - does not, of course, affect fertility (in terms of the TFRs of those of childbearing age) directly, but it does affect the shape of the population pyramid, and in this way it affects the future shape of the pyramid by influencing reproductive cohort size and also may influence social attitudes to childbearing. There may also be an impact on the third mechanism, the economic one, as I shall try and argue below.

But first an illustration of the negative momentum phenomenon at work. Fortuitously, and as part of the Mothers Day celebrations in Germany, the Federal Statistical Office recently put the following results of a microcensus online:

As reported by the Federal Statistical Office on the occasion of Mothers Day on 13 May, about 1.9 million mothers in Germany aged 15 to 64 years raised young children of under three years in the household in 2005. That are about 154,000 mothers less than in April 1996. This is shown by current results of the microcensus, the largest household survey in Europe. In this context, children include not only natural children but also stepchildren, adopted children and foster children. The microcensus results also show that the number of mothers with young children in the former territory of the Federal Republic (excluding Berlin) was down by 225,000 from 1996 to 1.5 million in 2005, whereas in the new Länder (including Berlin) it rose by 71,000 to about 346,000 in 2005.

The upshot of what they found in the census is that were 150,000 or so less mothers of children under three in 2005 than there were in 1996. That is a cohort width shrinkage of something like 5 - 6% in nine years, which I think is quite a lot. And of course with below replacement fertility on and on it goes. (The reasons for the apparent disparities between East and West would need further examination, although several possible explanations immediately spring to mind, including some rebound from the very dramatic collapse in East German fertility which took place after 1990. There may however also be changing cohort size elements from previous generations at work, as we all know demographic processes tend to cast a long shadow).

Turning now for a moment to the ideational mechanism, and over and above the material which Claus refers to, I recently came across the following interesting box diagram which shows cross generational changes in the proportion of women who regard the answer "none" to the question about the desireable number of children to be appropriate (again, click to enlarge):




Now what is striking here is the situation in the German speaking countries, where the percentage in Austria who are willing to answer "none" is now 12.6%, while in Germany it is 16.6, and in both cases this is a large and significant change over the previous generation. This conforms to the argument Lutz et al themselves put forward when they say:

"The sociological mechanism of the Low Fertility Trap Hypothesis (LFT 2 in Figure 1), is that the assumed is that young people are socialized in a way that they internalize the family size norms which they experience around them in term of actual fertlity. This suggests that ideals will decline about one generation length after the decline in TFRs. This timing fits quite well to the German speaking countries which were the first to enter a steep fertility decline in the 1970s and which in the 2001 Eurobarometer are the first to show significantly lower personal family size ideals for the younger age groups."

The change in attitudes to childlessness reveals yet one more mechanism whereby this decline in average desired family size may be working. What would be interesting would be to know more about why childlessness is becoming more socially acceptable, and why the effect is so strong in German speaking countries.

Moving on the LFT3, the economic component, Lutz et al tend to focus this on a cohort-type effect, based on the changing relative income position of young people vis-a-vis their predecessors 20 odd years ago:

"We study data on earnings by age, gender and education for full time workers over time. As can be seen by Figures 2 and 3, individuals aged 20-34 earn the least. The Figures also show that the young have profited less from the economic growth relative to the other age groups, and that in Canada they actually experienced a decrease in real wages (while other age groups had a rise) since 1981. In the UK women aged 20-34 had a positive, but relatively slow wage growth, where the 35-49 year have a 66% higher wage growth than those aged 20-34. The negative development of income for women in their peak-childbearing years may represent one reason why that fertility timing is postponed and fertility outcome is lowered."

Now this data is interesting, and the situation revealed obviously forms part of the environment in which young people take their childbearing decisions, but perhaps from the point of view of the fertility trap hypothesis something is missing, since these changes do not, in and of themselves, seem to be produced by demographically induced processes and as such they may not be considered self-reinforcing. What might these changes be due to? Well, globalisation processes for one, and technological changes for another. When I say technological I am thinking of the so-called "changing skill bias of work" argument, which would seem to imply that in order to achieve the higher salaries which go with higher value added work young people need to acquire more education and relevant experience. This factor certainly may act as an incentive to delay having children.

However, in line with the exogenous shock idea, the data Lutz et al seem to have studied is cross-country data regardless of Tfr, whilst what would really be needed to validate the economic mechanism Lutz et al advance as part of their trap-feedback hypothesis would be data which correlated earnings for young people (under say 30) with Tfrs, and showed that as the Tfr dropped below 1.5 this income effect became more pronounced (or vice-versa if you prefer).

But in the final analysis the key point about both the globalisation and the technology arguments is that they are better seen as exogenous (ie external to the system) shocks, which have a "level effect" impact in terms of the timing of the first birth decision. For the fertility trap hypothesis to have internal coherence what we need are endogenous processes, ones which are internally driven by the model and repeat and repeat.

Based on the work which Claus and I have been doing on the Life Cycle Model of consumption and savings in the context of rising median ages (see, for example, this post)I think we may well have come up with just some of these.

Essentially the argument, as it is presented here, is that as median ages rise beyond a certain point - 42/43 let's say - the structural characteristics of the economy change. While younger economies - let's say with median ages in the 35 - 39 range - are driven by large scale borrowing (on aggregate), domestic consumption surges, and, of course imports and current account deficits to match the domestic savings weaknesses, the more elderly ones can exhibit higher relative savings levels (Japan, German, Italy, Finland, possibly Switzerland), can no longer rely on domestic consumption to anything like the same extent, and increasingly come to depend on export growth for GDP growth.

Now, of course, this produces a mechanism whereby four things happen:

1) In order to compete for exports these economies have a permanent pressure on their tradeable sectors, whereby outsourcing is continuous and ongoing, wages are continuously compressed, and structural reform is permanent. Since the very export dependence is only further reinforced by the continuing process of change in the population pyramid (ie domestic demand never "recovers" as such) this is all self-reinforcing. That is the more time passes the more there is downward pressure on the wages of young people.

2) Due to the comparatively lacklustre economic growth performance there is a constant shortfall in the tax income necessary to guarantee existing welfare and pension commitments. This shortfall is produced by the low levels of trend growth (think Italy, Germany and Japan) which you can generate exclusively on the basis of export growth. Since the changing pyramid structure (here is another part of the feedback loop) means that an increasing part of the voting population comes to be over 50, the tendency, as we are in fact seeing, is to attempt to maintain welfare commitments by increasing the tax burden, which affects the consumption and earning possibilities of the young directly.

3) Migration factors. The general lack of growth in the economy, and the tendency towards increase retirement ages and higher participation rates at the older ages, all mean that there is a relative lack of well paying jobs at the entry level, a phenomenon which makes outward migration an increasingly attractive proposition for educated young people (again, as we are seeing in Germany and in Italy). This out-migration once more feeds back into the structural evolution of the population pyramid. If the out migration is in part compensated for by in-migration of lower skilled workers, then this tends to retard the process of moving towards higher value work, a feedback which one more time would seem to find reflection in lower wage levels on average in the younger age groups.

4) Impediments on pro-natal policies. The pressure on fiscal resources which result from the previous three factors mean that effectively it becomes increasingly difficult to generate the resources to finance really meaningful pro-natal policies which might attempt to "tease" fertility back up towards a higher level. As time goes by this problem only gets worse.

OK, these are really simply a set of working notes. Comments, as always, welcome.

No comments: