Family Planning and Teenage Pregnancy: an economic approach, David Paton
Family Planning and Teenage Pregnancy: an economic approach
David Paton
presented at:
 50 ANOS DE EDUCAÇÃO SEXUAL:
balanços e perspectivas 
Lisbon 12th November 2004
 
1. BACKGROUND 
Improving sexual health amongst young people is an important policy objective in many countries.  High rates of teenage pregnancy are generally perceived by policymakers to be a problem on the grounds that early motherhood is associated with adverse socio-economic outcomes, e.g. related to education, unemployment, earning capability and social exclusion.  Although pregnancy is not necessarily an adverse outcome for all teenagers, there is a presumption amongst policy makers that achieving low rates of teenage pregnancy, particularly amongst younger age groups teenagers will be socially beneficial.

A related issue is the increase in sexually transmitted infections (STIs) currently being experienced by many countries.  The growth of chlamydia infections amongst young people is of particular concern given that this infection is largely asymptomatic, yet is associated with a range of adverse health outcomes, including infertility.  A third element of policy on sexual health is the question of early sexual activity in itself.  There are adverse physical and psychological health outcomes associated with early sexual activity and many countries operate some form of age of consent, implying that sexual activity before this age is undesirably.  However, there is not a consensus amongst policy makers as to whether delayed sexual activity should be a policy objective in itself. 

There exist many possible strategies for achieving policy objectives such as those outlined above.  Some of these attempt to influence sexual behaviour directly (e.g. sex education, abstinence education, provision of family planning and abortion services) whilst others attempt to influence sexual behaviour indirectly by focusing on broader socio-economic influences (e.g. poverty, education, self-confidence, family structures etc).  Many of the direct strategies are controversial and subject to intense debate, particularly those relating to family planning, abortion and sex/abstinence education.  Inevitably moral concerns are often to the fore of such debates.  In this paper, I attempt to circumvent the moral debate (without denying its importance) and to focus on very practical question: what policies actually work in reducing teenage pregnancy rates? 

I shall argue that the discipline of economics provides an excellent framework for analysing these policies.  I shall also provide an overview of the existing evidence base. 

2. An Economic Approach

For a young person, deciding whether or not to have sex involves costs and benefits. 

•         Perceived costs may include fear of pregnancy, STIs, religious &/or parental disapproval

•         Perceived benefits may include desire to have sex or pregnancy, peer approval

A typical approach by economists would be to evaluate not just the value of the costs and benefits, but also the probability of these costs and benefits occurring. 

For example, consider here a simple two 14-year olds, A and B, deciding whether or not to start having sex.  Both teenagers think they would be happier if they had sex, but are worried about getting pregnant.  Economists like to assign numbers to signify levels of happiness (we usually use the term ‘utility’) and that is what I will do here.  Let us assume that both teenagers think that, if they have sex, their happiness will increase by 40 units (the actual value we use does not matter).  A is hopeful of passing her exams and going on to university and is very worried about getting pregnant.  She thinks that her happiness would go down by 100 units if she were to do so.  B, however, has already decided to leave school when she is 16.  She does not actively want to get pregnant, but neither would it be too great a disaster.  If she does, she thinks that her happiness would go down, but only by 60.  Neither of them knows for sure how either sex or pregnancy will affect them, but it is safe to assume that their decisions will be based on their expectations. 

In Figure 1 I use a decision tree to summarise the decision and its consequences.  If A decides to have sex and gets pregnant, she thinks her happiness will go down by 60 units (the increase of 40 units from having sex less the decrease of 100 units from getting pregnant).  Similarly, if B gets pregnant, she thinks her happiness will go down by 20 units (40 less 60). 


Note: a positive value indicates an increase in happiness, a negative value indicates a decrease. 

FIGURE 1: DECISION TREE FOR TEENAGERS A AND B 
The decision each teenager takes will depend not only on the values of happiness they expect from the possible outcomes, but also on the likelihood of those outcomes.  Economists would argue that each will decide to have sex if they think (or expect) that, on average, their happiness will increase.  For example, let’s assume that the probability of getting pregnant is 50% and that both our teenagers know this.  This is how each teenager expects their happiness to change on average if they have sex: 

A: 50% chance of 40 + 50% chance of -100      
→ average or ‘expected’ happiness = -10
B: 50% chance of 40 + 50% chance of -20        
→ average or ‘expected’ happiness = +10
In other words, A expects (on average) her happiness to go down by 10 units if she decides to have sex.  Consequently, she will decide to abstain.  B, however, expects her happiness to go up by 10 units if she has sex.  She will decide to have sex. 

There are lots of ways we could make this model more realistic, but that would also make it more complicated and the simplifications allow us to get over the basic principal very easily.  Let’s use the model to predict what happens if the probability of having sex was to be reduced (e.g. through the availability of family planning) to only 20%.  The expected happiness for A and B can be worked out as follows: 

A: 80% chance of 40 + 20% chance of -100
→ expected happiness = 0.8 x 40 + 0.2 x (-60) = 32 - 12 = +20.
B: 80% chance of 40 + 20% chance of -20
→ expected happiness = 0.8 x 40 + 0.2 x (-20) = +32 - 4 = +28.
Due to the lower risks of pregnancy, the scales have become weighted in favour of having sex.  A expects that her happiness will (on average) be increased by 20 units and changes her decision: she will now decide to have sex.  B expects her happiness to increase by 28 units and still decides to have sex.  In terms of impact on teenage conceptions, B is now less likely than before to get pregnant, but A is more likely to get pregnant.[1]

The exact numbers in our example are not all that important.  In practice, there will be a range of different views amongst teenagers and the perceived benefits or costs of having sex and of getting pregnant will vary according their circumstances.  In general, we can say that if you reduce the risk of pregnancy, more teenagers will start to have sex.  Some of these will get pregnant.  Of course, the ones who were going to have sex anyway will be less likely to get pregnant, but the overall impact on pregnancy rates is impossible to predict. 

When costs are high & benefits low, a youngster is less likely to have sex.  However, for given costs and benefits, if the probability of a bad outcome occurring increases, a youngster will be more likely to have sex.  Government policies that affect costs and benefits or the probabilities of these occurring may have the unanticipated effects of actually changing the sexual behaviour of at least some young people.
 
3. Some Examples

Example 1: what if you reduce poverty/ unemployment/family breakdown?
 
•          Cost of pregnancy increases
•          Fewer youngsters have sex/greater use of protected sex.
•          Pregnancy rates likely to go down.
 
Lots of evidence that economic welfare, family stability, religion etc have impact on teenage pregnancy rates, e.g. Evans et al (1992), Chong-Burn et al (1993), Paton (2002).  Little evidence on STIs.
 
Example 2: what if you increase access to family planning?
 
•          Probability of pregnancy for teenagers having sex is reduced...
•          ...but some youngsters ‘at the margin’ who were abstaining, now decide to have sex...
•          ... and ‘riskiness’ of sexual behaviour may increase
•          Pregnancy rates & STI rates may go up, down or not change
 
Strong evidence that access to family planning has little or no impact on teenage pregnancy rates:, e.g. Churchill et al (2000), Paton (2002), Kirby (2001), DiCenso et al (2002)
No evidence on STI rates.
 
Example 3: what if you emphasise emergency birth control?
 
•          some youngsters substitute EBC for condoms
•          measured pregnancy rate may go down
•          STI rate likely to increase
 
Virtually no evidence on impact of EBC!
 
Example 4: what if you increase access to abortion?
 
•          For some youngsters, costs of sexual activity reduced.
•          Some youngsters who would have given birth now have abortions...
•          ...but more youngsters get pregnant.
 
Strong evidence that easier access to abortion increases teenage pregnancy
 
4. Existing Evidence 

4.1 TYPES OF EVIDENCE 
Existing evidence relating to teenage pregnancy can be divided into two categories, Project Evaluation and Policy Evaluation. 

·        Project Evaluation examines the impact of particular projects on specified outcomes, e.g. the introduction of a family planning clinic at a single school.  The best evidence uses Randomised Controlled Trials (RCTs) in which an ‘intervention’ group is compared to a similar ‘control’ group for which there is no intervention.  Evidence based on ‘observational studies’, in which no control group is present, has been shown to be biased in favour of interventions (Guyatt et al, 2000).  

·        Policy Evaluation examines the impact of particular policies using data aggregated across a region, state or country.  The best evidence in this category will use a multi-variate statistical approach to control for the impact of other factors. 

Both types of evidence should be used in formulating and evaluating policy on teenage pregnancies.

4.2 SEX EDUCATION AND TEENAGE PREGNANCY
4.2.1 PROJECT EVALUATION EVIDENCE
There exist several systematic reviews of the evidence from RCTs.  Most of the reviews have found little evidence that particular sex education programmes have increased rates of sexual activity.  However, there is also little evidence of significant reductions in pregnancy rates. 

For example, the most recently published review of a range of interventions (including sex education programmes) aimed at reducing adolescent pregnancies, based solely on good quality “experimental” evidence, concludes that, to date, primary prevention strategies have had insignificant effects on pregnancy rates, rates of sexual activity and contraceptive use (DiCenso et al, 2002).  Similarly, an influential review of over 250 experimentally-based evaluations of programme for school-age children in the U.S.A. finds few sex education programmes are associated, on their own, with lower pregnancy rates. 

The RCT evidence on abstinence education, although growing, is still underdeveloped.  One such study is included in the DiCenso meta-analysis and we await the publication of a large scale evaluation from the USA by Mathematica Inc. 

4.2.2 POLICY EVALUATION EVIDENCE 
One of the problems with the project evaluation evidence on sex education is that it generally compares groups receiving a particular programme with another group receiving standard sex education classes.  Thus, the evidence cannot generally isolate the impact of sex education in general.  There is a very limited set of evidence in the policy evaluation literature which does this.  Notably, Oetinger (1999) uses data from the US in the 1970s and concludes that sex education programs in the US in the 1970s had some causal impact on both teenage sexual behavior and pregnancy rates.  Specifically such programs were associated with significantly earlier sexual activity for females.  However, the estimated magnitude of the effect was relatively small and no effect was observed for males.  Sex education also lead to slightly earlier pregnancies for some groups of teenagers, but these effects were not always significant.  Oetinger attributes these effects to the role sex education plays in helping teenagers to reduce the risks of sexual activity (p.606).  On the other hand, Evans, Oates and Schwab (1992) find that teenagers who had experienced a sex education program had a lower probability of becoming pregnant.  Although this effect was significant, again the absolute impact was small, relative to other factors. 

Several papers survey the experience in different countries (for example, Alan Guttmacher Institute, 2001; Cheesbrough et al, 2002; Kane and Wellings, 1999).  These examine key features of countries with low and high conception rates.  A conclusion common to this research is that countries with lower conception rates (e.g. the Netherlands) tend to have an open attitude to adolescent sexuality. The problem of data inconsistencies across countries means there has been little or no use of multivariate analysis in these studies to attempt to identify the causal effects of variations in access to adolescent family planning services across countries.  This makes it very difficult to draw policy conclusions from this evidence.  Further, more detailed research into the Dutch experience by sociologist Joost Van Loon (2002) finds that school-based sex education in that country varies dramatically from liberal to extremely conservative and, overall, school-based sex education does not differ greatly to that in the UK.  He attributes lower pregnancy and abortion rates in the Netherlands to factors such the existence of much stronger family ties and better communication between parents and children. 

4.3 ACCESS TO FAMILY PLANNING SERVICES AND TEENAGE PREGNANCY 
4.3.1 PROJECT EVALUATION EVIDENCE
There exist several systematic reviews of the evidence from RCTs.  The key conclusions of this literature are that projects to provide access to contraceptives for young people have not led to significant increases in sexual activity, but neither is there evidence that they have reduced adolescent pregnancy rates. 

Programme encouraging greater access to family planning are included in the meta-analysis by DiCenso et al (2003) which as we have seen found no significant impacts from any type of programme.  The Kirby (2001) review concludes that the provision of contraceptives through schools “does not increase sexual activity”, but also that such provision tended not “to markedly increase the ... use of contraceptives” (Kirby, 2001).[2]  Further, Kirby found little evidence that school or community based family planning clinics reduced pregnancy rates. 

4.3.2 POLICY EVALUATION EVIDENCE 
Although no systematic review of evidence in this category currently exists, a number of researchers have examined the impact of access to family planning on pregnancy rates.  As with sex education, much of the most influential research in this category is qualitative in nature and attempts to pick out key themes or trends in experiences of teenage pregnancy across regions or countries.  This type of approach makes it difficult to identify the causal effect of particular policies or to isolate the individual effects of different policies or characteristics. 

Research focusing on single countries tends to suffer less from data deficiencies and offers the potential for more robust evidence. Rimpela, Rimpela and Kosunen (1992) suggest that increasing use of the contraceptive pill by adolescents in Finland has been associated with a decreasing trend in adolescent conception rates.  However, they find no such association for under 16s.  Further, the statistical methodology in this paper is very unsophisticated and, in particular, the lack of a multivariate approach makes it impossible to draw conclusions about cause and effect. 

In a UK based study, Ingham, Clements and Gillibrand (1999) examine 40 local authorities between 1990 and 1997, 20 of which experienced increasing underage conception rates and 20 experienced decreasing rates.  The authors find that all of the areas with decreasing rates had introduced new adolescent family planning services, whilst only 11 of the areas with increasing rates had done so (Ingham, Clements and Gillibrand, 1999).  Although this may be indicative of a beneficial impact, the lack of multivariate data analysis makes it impossible to know for sure whether or not the introduction of family planning services had a significant causal effect. 

Several recent papers provide statistical evidence using sophisticated multivariate techniques.  Most recently, in a paper published in the Journal of Health Economics, I examine regions within the UK from 1984 to 1997 and find no evidence that increased access to family planning services for under 16s leads to lower conception or abortion rates (Paton, 2002).  A particularly interesting feature of this study is the effect of the so-called “Gillick Ruling”.  In late 1984, Victoria Gillick successfully gained a court ruling that family planning could not be supplied to girls aged under 16 (the age of consent) in England and Wales.  There were widespread predictions of that underage pregnancy rates would rise as a result of the ruling.  The ruling was overturned towards the end of 1985.  In fact, the evidence from Paton (2002) is clear that the Gillick Ruling did not lead to any increase in underage conceptions.  Although Paton (2002) demonstrates this using formal econometric techniques, the lack of impact of the Gillick ruling is clear from an inspection of Figure 2.  There is a clear dip in attendance at family planning clinics for in that year for under 16s (but, but little or no change to conception or abortion rates. 

Clements et al (1998) examine the impact of access to family planning on teenage pregnancy in the Wessex region of the UK in the early 1990s.  To measure access, they use the distance of the teenager from three possible family planning outlets: GP surgeries, family planning clinics and youth oriented family planning clinics.   

For youth oriented family planning clinics, the authors find that teenagers who live between 3 and 7 km from the nearest facility have higher pregnancy rates than those who live closer.  The authors go on to suggest that this finding is evidence that providing youth family planning services works in reducing teenage pregnancies.  If you look closer at the paper, you will discover that they also find that teenagers who live 3-7 km from a youth clinic also have higher pregnancy rates than those who live further away.  Further, they find that the distance from either the nearest GP or the nearest family planning clinic has no impact on the teenage pregnancy rate.  Finally, the authors find no relationship with any family planning measure for pregnancies to those aged under 16.

The limited statistical evidence that has been undertaken in the United States on family planning services suggests ambiguous conclusions.  One national study of black and white adolescents found that more restrictive laws on the sale and advertising of contraceptives was associated with higher pregnancy rates for white adolescents, but no effect was found for black adolescents (Lundberg and Plotnick, 1995).  On the other hand, Evans, Oates and Schwab (1992) use similar data to Lundberg and Plotnick and find that the number of family planning clinics per teenager in the locality has no impact on overall pregnancy rates.  

The lack of solid evidence that access to family planning lowers teenage pregnancy rates is not surprising given the good evidence that the majority of adolescents, at least in the UK who become pregnant do so whilst using contraception.  For example, a recent study of pregnant teenagers in the Midlands in the UK finds that over 71% had discussed contraception with a health professional in the year before they got pregnant and 50% had been prescribed oral contraception (Churchill et al, 2000).  Similarly, Pearson et al (1995) finds that the majority of pregnant teenagers claimed to have used some form of contraception at the time of getting pregnant, whilst well over 90% of the teenagers knew how to get access to family planning. 

4.4 ACCESS TO ABORTION AND TEENAGE PREGNANCY 
The evidence relating to adolescent abortion access almost exclusively involves policy evaluation in the U.S.A.  Data in the U.S.A. are particularly rich in that state level variations in policy provide ‘natural’ experiments in which policy changes can be tested. 

A number of papers examine the impact of ‘parental notification laws’ in which at least one parent must be notified before a woman aged under 16 can obtain an abortion.  Other papers examine a range of other abortion restrictions that have been applied to adolescents in particular states (see, for example, Kane and Staiger, 1996; Ohsfeldt, Robert and Stephan Gohmann,1994; Levine, 2000). 

Without exception these papers find that restrictions to abortion access for adolescents leads to a reduction in overall rates of teenage pregnancy.  Some of these papers find that abortion rates decrease and birth rates increase (but by a smaller amount).  However, Kane and Staiger (1996) even find some evidence that small restrictions to abortion access lead to both abortion rates and birth rates decreasing (Kane and Staiger, 1996). 

The evidence from the U.S.A. to date is strongly suggestive that easier access to abortion for adolescents are associated with higher teenage pregnancy (but not necessarily birth) rates. 

4.5 Socio-economic Variables and Teenage Pregnancy 

Many researchers have examined the impact of socio-economic factors, such as education and family cohesion, on pregnancy and birth rates.  Results relating to these factors tend to be much less ambiguous than those relating directly to reproductive decisions.  In general, teenage fertility rates have been found to be correlated with low educational achievements, unemployment rates, unstable family background, race and religiosity (Evans, Oates and Schwab, 1992; Plotnick, 1992; Chong-Burn, Haverman and Wolfe, 1993; Haas-Wilson (1996); Paton, 2002). 

4.6 Gaps in Current Research 

To date, there has been very little research conducted on the impact of greater access to the morning after pill on conception rates.  Similarly, there is little or no published evidence on the impact of the promotion of family planning on underage rates of sexually transmitted diseases.  I will cover both these issues in more detail in my second paper.  

5. Conclusions 

The key lesson of this paper is that ‘obvious’ policies to tackle issues such as teenage pregnancy may have unexpected (and unwanted!) side effects.  A basic understanding of decision-making behaviour can significantly improve our understanding of how such policies may work.  The evidence so far on measures aimed at influencing sexual behaviour directly, e.g. access to family planning, sex education etc. is not particularly encouraging.  There is little or no convincing evidence that such interventions have worked consistently to achieve lower rates of teenage pregnancy.  One explanation is that the statistical tests used to test for beneficial objectives are just not powerful enough.  Alternatively, the evidence is consistent with the thesis that policy interventions such as greater access to family planning can encourage at least some young people to engage in riskier sexual behaviour.  Perhaps more than anything, it is vital that politicians think through the implications of policy measures in an objective manner. 

References
Alan Guttmacher Institute (2001), Teenage Sexual and Reproductive Behaviour in Developed Countries: can more progress be made?, New York: Alan Guttmacher Institute.
Cheesbrough, Sara, Roger Ingham and Doreen Massey (2002), A Review of the International Evidence on Preventing and Reducing Teenage Conceptions: the United States, Canada, Australia and New Zealand, London: Health Development Agency.
Chong-Burn, An, Haverman, Robert and Wolfe, Barbara (1993), ‘Teen Out-of-Wedlock Births and Welfare Receipt: the role of childhood events and economic circumstances’, Review of Economics and Statistics, 75 (2, May): 195-208.
Churchill D, Allen J, Pringle M, Hippisley-Cox J, Ebdon D, Macpherson M and Bradley S. (2000), ‘Consultation patterns and provision of contraception in general practice before teenage pregnancy: case-control study’, British Medical Journal, 321: 468-89.
Clements, Stephen, Nicole Stone, Ian Diamond and Roger Ingham (1998), ‘Modelling the Spatial Distribution of Teenage Conception Rates within Wessex’, The British Journal of Family Planning, 24, 61-71.
DiCenso, Alba, Gordon Guyatt, Andrew Willan and Lauren Griffith (2002), ‘Interventions to Reduce Unintended Pregnancies among Adolescents: systematic review of randomised controlled trials’, British Medical Journal, 324, 15 (June).
Evans, William N, Oates, Wallace E. and Schwab, Robert M (1992), ‘Measuring Peer Group Effects: a study of teenage behavior’, Journal of Political Economy, Oct., 100 (5): 966-91.
Guyatt, Gordon H., Alba DiCenso, Vern Farewell, Andrew Willan and Lauren Griffith (2000), ‘Randomized Trials versus Observational Studies in Adolescent Pregnancy Prevention’, Journal of Clinical Epidemiology, 53, 167-74.
Haas-Wilson, Deborah (1996), ‘The Impact of State Abortion Restrictions on Minors’ Demand for Abortions’, Journal of Human Resources, 31(10): 140-58.
Ingham, Roger and Steve Clements and Rachel Gillibrand (1999), Factors Affecting Changes in Rates of Teenage Conceptions: 1991 to 1997, Summary Bulletin, London: Teenage Pregnancy Unit.
Kane, Roslyn and Kaye Wellings (1999), An International Review of the Evidence: data from Europe, London: Health Education Authority.
Kane, Thomas, and Staiger, Douglas (1996), ‘Teen Motherhood and Abortion Access’, Quarterly Journal of Economics, May, 111 (2): 467-506.
Kirby, Douglas (2001), Emerging Answers: research findings on programs to reduce teen pregnancy, Washington DC: National Campaign to Prevent Teen Pregnancy.
Levine, Phillip B. (2000), ‘The Sexual Activity and Birth Control Use of American Teenagers’, NBER Working Paper 7601, March, Cambridge, MA: NBER.
Lundberg, Shelly and Plotnick, Robert D (1995), ‘Adolescent Premarital Childbearing: do economic incentives matter?’, Journal of Labor Economics, 13 (2): 177-200.
Ohsfeldt, Robert and Stephan Gohmann (1994), ‘Do Parental Involvement laws reduce Adolescent Abortion Rates?’, Contemporary Economic Policy, 12: 65-76.
Paton, David (2002), ‘The Economics of Family Planning and Underage Conceptions’, Journal of Health Economics, 21, 2 (March),27-45.
Pearson, V.A.H., Owen, M.T., Phillips, D.R., Pereira Gray, D.J, and Marshall, M.N. (1995), ‘Family Planning Services in Devon UK: awareness, experience and attitudes of pregnant teenagers’, British Journal of Family Planning, 21, 45-9
Plotnick, Robert D (1992), ‘The Effects of Attitudes on Teenage Premarital Pregnancy and its Resolution’, American Sociological Review, 57 (6, Dec): 800-11.
Rimpela Arja H., Matti K. Rimpela and Elise A-L. Kosunen (1992), ‘Use of Oral Contraceptives among Teenagers and its consequences in Finland 1981-1991’, British Medical Journal, 305 (October), 1053-1057.
Van Loon, Joost (2002), Deconstructing the Dutch Utopia: sex education and teenage pregnancy in the Netherlands, London: Family Education Trust.
 
 

[1] It might be argued that teenagers will not always act in this rational manner, particularly, for example, when drunk at a party.  Of course, if A gets drunk she may be more likely to decide to have sex than if she stayed sober.  It is still reasonable to suggest that A would be still more likely to have sex if she were drunk and thought that the chances of pregnancy were relatively low.  Similarly, for the model to be useful in explaining the impact of policy, there is no need for every teenager to behave in this rational manner, only that teenagers do so on average.

[2] One possible explanation given for the seemingly contradictory findings is that youngsters simply replace their existing source for contraceptives for the new school-based source.  In this case, the provision of contraceptives has had no real effect at all.  Further, it is difficult to generalise these findings to national or regional policies that may affect all sources of supply.  See also, Cheesbrough, Ingham and Massey (2002).
 

ASSOCIAÇÃO FAMÍLIA E SOCIEDADE
Horário: 9h30 - 13h30
Rua do Lumiar 78 / 1750-164 Lisboa
Tel/Fax: 21 314 95 85 - familiasociedade@sapo.pt