Does family policy affect fertility?

by Edward Hugh


Well according to Anders Björklund, of Stockholm University, and based on the Swedish experience, it does. The linked paper is in fact his presidential address to the European Society for Population Economics meeting in Athens in 2001. Björklund's arguments are interesting, especially for those who have been following the discussion we have been having on DM about the birth postponement phenomenon: in fact much of the fluctuation in registered tfrs seen in Sweden can be related to this postponement process (average age at first birth in Sweden is now around the 29 mark), and the conclusion he comes to is a surprising one, that pro-natalist policy has been particularly effective in Sweden (even when compared with other Scandinavian countries) because the package has inadvertently (that is it was not the original explicit objective) incorporated a 'speed bonus' which encouraged families to move rapidly from the first child to second and subsequent ones and this has had an impact on fertility.

Now this impact is important due to the fact that higher order 'spacing' (ie the time between the first and subsequent children) is a key factor in producing a situation that women who want to have more children are, in fact, unable to do so, since as they become older infertility becomes an increasing problem. By encouraging women to have a second child rapidly after the first one some of this 'unwanted childlessness' is reduced. So it is this component in the package that seems to be important, and there would certainly seem to be lessons for others here.

In comparative terms Björklund look at fertility outcomes in Sweden, Denmark, Finland, France, Belgium and the Netherlands. The question he tries to address is the following one:

"Can public policies boost fertility in a modern society by providing financial and in-kind support for families with children? "

He suggests it can, and he advances three arguments to support his claim:

First, family policies were expanded quickly in Sweden. From the early 1960s to around 1980 (i.e., in less than 20 years), the generosity increased markedly – for paid maternity leave, subsidised child care, paid leave to take care of sick children, and universal child allowances. In retrospect, the speed with which these policies were extended is striking. Second, as already alluded to, several policies had a specific design, namely, that they facilitated labour force participation for mothers. So the policies potentially could affect the structure of fertility, i.e., timing, spacing, and socio-economic differentials, as well as the overall level of fertility. Third, pronatalistic arguments did not motivate the policies. In particular, they were not the outcome of a low-fertility period. This fact does not naturally guarantee that the expansion of family policy can be considered exogenous with respect to fertility. But it would have been more problematic to interpret the Swedish experience if, e.g., a sharp drop in fertility rates had preceded the introduction of the new policies.

As he says:

Rapid family policy expansion from the early 1960s to around 1980 implies that those generations of women who planned their childbearing before the expansion, i.e., women born during the 1920s and most of the 1930s, faced markedly different incentives than women born from the 1950s and onward. So I study changes in fertility patterns between these generations of women in Sweden. Because fertility patterns could change for reasons other than policy changes, I compare the evolution of fertility patterns in Sweden with the evolution in other countries. I primarily compare Sweden with its Nordic neighbours (Denmark, Finland and Norway). In many respects, these countries pursued the same family policy, but they did not go as far as Sweden. I also compare the evolution in Sweden with continental European countries (plus the UK) that pursued quite different family policies compared to Sweden.

Björklund provides a convenient table-summary of the child related policies implemented in Sweden (I won't try and summarize these here, but would rather recommend reading the full paper). One interesting detail is worth highlighting:

The Swedish rules that determine maternity-leave benefits also contain elements that create a kind of speed premium on further childbearing. Because maternity-leave benefits are earnings related, a period of no work or only part-time work after a birth would reduce the benefit level after a subsequent birth. But during the late 1970s and early 1980s the rules became successively more generous in allowing the parent to retain the right to the benefit level that she once had obtained by working full-time before the first, or any previous, birth. From 1974 to 1979, the parent could abstain from work-related earnings for 12-18 months (depending on local practice of the rules) and yet retain the right to a previous benefit level for subsequent births. From 1980-85 the interval became 24 months, and in 1986 it was prolonged to 30 months; from 1980 onward, there was no local variation in the rules.

As we will see in coming posts this element may be considered important since, given that the average Swedish woman now has the first child at a rather high age any subsequent delay carries with it an ever increased risk of encountering infertility problems.

Also the provision of child care seems to be important:

Local-government-subsidised child care in care centres is another important result of policy. In the early 1960s only some part-time options were offered, but from the late 1960s onward a major extension occurred. Because local governments have been the major providers of child care, there has always been some variation in coverage and rules; the general ambition has been to provide care that allows both parents to work full time. Further, the fees have only covered about 10-20 percent of all costs; often, the fees have been even lower for the second child and subsequent children. Only families with two working (or studying) parents have been eligible for this subsidised care.

Besides care for pre-school children, local governments provide subsidised morning and afternoon care for school children up to about age 10 (the school-starting age is 7 in Sweden). This care also rapidly increased over the same time period.

There are also further pro-employment elements:

Sweden also used labour relations legislation to facilitate the combination of childbearing and mothers participating in the labour market. In 1979 the parliament took an important decision that gave young children’s parents the right to cut back their working hours from full time to 75 percent. This right makes it easier for parents to take part-time leave. Parents can exercise this right until children reach age 8.

There is also, of course, a cash payment child allowance system:

The major form of cash support for families with children is the universal child allowance that was introduced in 1948. A non-taxable fixed amount per child is paid each month to the mother......the real value of this benefit increased during the period. And in 1982 a higher amount for the third child and subsequent children was introduced. Besides this universal allowance, families with children can become eligible for means-tested housing allowances and social assistance benefits.

Now as he suggests:

The main reason why family policy could affect the level of completed fertility is that such policies cut the cost of having a child.

Also:

The policies could also affect timing of childbearing, i.e., the age at which women decide to have children. To be eligible for the bulk of benefits that Swedish family policy offers, it is important to have become reasonably well established on the labour market. The parental-leave system is earnings related. So women, who have not worked full time or women who have had a low monthly salary the year before childbearing, receive lower benefits for the entire parental-leave period. Much anecdotal evidence suggests that a typical strategy for women in Sweden who plan to have children is:

1. Complete your education.
2. Get a job and work full time for at least half a year; ensure that you get a permanent position so that you can return to your employer after parental leave.
3. Get the first child
4. Work part time for a while, and use subsidised day-care until you get the next child.

Because of the legislation, I expect that in Sweden, the average age for first births became higher than it would have been without implementation of the legislation.


Since however average age at first child has in fact been rising across much of the EU in recent years it is difficult to assess the extent to which this effect operates, and indeed Sweden may simply have had the good luck to produce a policy which in fact was (unintentionally) geared to facilitating decisions which many would have been trying to take even without the presence of the policy.

Also not without interest:

Instead, promotion of gender equality in the labour market was the major goal for these policies.

So in fact the fertility boost was something of an unanticipated spin-off from an attempt to achieve another objective.

Bjorklund is primarily interested in the impact of policy on the level of completed fertility. He looks at the evolution of cohort fertility rates for women born between 1925 and 1958 for a number of countries and he finds that the Swedish rates are remarkably stable around 2.0 over the entire period. Also a before-after perspective suggests that there were no effects from the distinct extension of family policies. By adopting a difference-in-differences approach, however, he was also able to draw another conclusion:

In all other countries, cohort fertility rates declined, and in some countries, they declined considerably. We start by comparing Sweden with its Nordic neighbours. An application of the difference-in-differences approach means that the stable cohort fertility rate in Sweden is compared with declines from the early 1930s to the 1950s by about 0.4 in Finland and Norway and 0.5 in Denmark. So these are the magnitudes of the estimated effects on the level of fertility of the more generous family policy in Sweden compared to the other countries......it is somewhat striking that Sweden is the only country with stable cohort fertility over this period.

As he notes:

The conclusion in the previous section might sound strange for those who have followed the Swedish experience and discussion since the early 1990s. After a peak at 2.14 in 1990, the period total fertility rate declined sharply to 1.50 in 1999 and showed only a minor recovery to 1.55 in 2000.10 (See Figure 7.) Is this development really consistent with the conclusion that Swedish family policy has permanently raised the completed fertility rate?


In order to try and shed some more light on the apparent inconsistency here, he examined cohort fertility rates of women born in Sweden up to 1965 and who had children up until 2000. He found that women born as late as 1965, whose childbearing up to the age of 35 is captured in the data, had already reached the quite high level of 1.70 children on average, and he therefore suggests that it seems reasonable to think that the prospects of reaching a completed fertility rate close to 2.0 are quite good for women born up to 1965. However he cautions:

"low fertility rates during the 1990s suggest that cohorts born after 1965 must space their children very closely to reach the fertility level of the preceding generations of women."

Spacing matters, in other words. One worry he had was whether in fact the policy applied in Sweden might be actually encouraging women to postpone childbearing, since they might be lead to wait till they had relatively higher income levels so as to gain the maximum benefit from the compensatory payments system. However he discounts this conclusion since he finds no striking difference between the evolution of postponement in Sweden and in the other countries for which he was able to acquire data.

"There is a U-shaped pattern in data for all six countries with lows in the mid-1940s. Since then, the age at first birth has increased. In particular, the increase in the Netherlands is marked despite a family policy that is very different from Sweden’s."

The conclusion to his cross country analysis are very interesting indeed. What he concluded was:

First, fertility patterns based on educational level are very similar in all countries for which I was able to find data. Evolution over time, according to educational level, is also stable; the decline in completed fertility rates in countries other than Sweden can be seen in all education groups. Second, postponement of the first birth can be found in all countries for which I have data. Third, spacing patterns for women who have given birth to two, three, or four children are not very different in Sweden, France, Norway, and the Netherlands. Of these three similarities, I am most struck by the last one. Although there is a pattern that is consistent with an effect of the speed premium in the parental-leave system, the similarities among the countries are more marked than the differences...........One explanation could be that the apparent effects on spacing in the micro data are primarily not pure spacing effects but rather effects on the level of fertility. If some of those women, who gave birth to two or more children in short succession because of the parental-leave rules, with other rules would have waited a longer time before trying to get additional children, they might have failed to get these children. In that case, the effect of the parental-leave rules is rather on the fertility level than on spacing conditional upon a final number of children.


That is to say, for those women who actually had second, third children etc, the spacing patterns are very similar across the countries studied, however in Sweden slightly more women had more than one child, and this leads him to think that this result was produced by the fact that those women, had they taken more time to try and have children might never have been able to have them. Fascinating.