Mothers' human capital and childcare in Britain.
In the late-1980s the provision of care facilities for pre-school children moved a few places up the agenda of public discussion in Britain. A few employers began to provide nurseries or childcare advisers; private and voluntary operators opened a few day care centres; pressure groups increased their agitation for tax breaks for childcare expenses and protested at reductions in public daycare provision. The government has held the line against becoming involved, maintaining that childcare is a private matter, best left to the market, though concerning employers as well as parents. As the labour market tightened and the supply of school leavers fell, increasing attention was paid to how childcare might facilitate labour supply. Interest may have subsided from this peak during the recession of the early-1990s, but the long-run concern with the adequacy of the stock of human capital remains, particularly in view of projected shifts in the age structure of the population.
Recent literature (e.g. Mankiw et al 1992) has emphasised the role of human capital not just on the level of potential output in the economy, but also on the growth rate. In the case of the UK economy, skill shortages have been shown to have an effect on productivity (Haskell and Martin, 1993).(1) The stock of human capital in the economy depends not only on initial education and training, but also on work experience via on-the-job training and learning by doing. In the presence of high labour turnover, neither firms nor workers may have much incentive to invest in firm-specific training (see, for example, Jarvis and Prais, 1989). Information asymmetries may reduce incentives for workers to invest in general training, but give firms incentives to co-invest with workers (Katz and Ziderman 1990). High rates of labour turnover will reduce this incentive. Thus the expectation that women will depart from the labour force (especially the full-time labour-force) for a prolonged period on childbirth is likely to limit their acquisition of human capital from training even before maternity.(2) Actual absence from the labour force will lead to depreciation of whatever human capital they have.
Conditional on any level of human capital, sex discrimination in wages and opportunities may result in women's contribution to market work being undervalued relative to men's, thus distorting the privately-optimal domestic division of labour in favour of domestic rather than market production by women. The woman's time spent at home is not valueless: she produces valuable childcare and other domestic services. For these reasons, the value of a woman's time spent at home might well be greater than her market wage net of tax and childcare costs, but less than her social marginal product in the labour market, or at least what that would be were she to be more continuously employed.
Surveys consistently report a large unmet demand for childcare from mothers (Bone 1977, Scott 1989, Simpson 1986, Petrie and Logan 1986). In the PSI study of maternities in 1988, inability to find or afford childcare accounted for almost one third of the women who had intended to return to employment within nine months of childbirth but did not do so (McRae 1993). Most recently the 1990 British Social Attitudes Survey (Witherspoon and Prior 1992) has reported that two-thirds of women with a child under twelve who are not currently employed would like to be (mostly part-time) if suitable childcare were available.
Better childcare could enable the economy to utilise and conserve the stock of human capital embodied in women who become parents. In this article we offer an estimate of the effect of better childcare provisions on mothers' lifetime earnings. We compare these earnings gains and the tax revenue they generate with the costs of providing the care. We believe that this exercise will contribute to the debate by quantifying some of the effects of enhanced childcare provision. Although we illustrate some of the costs and benefits of altering public policy by means of calculations based on a policy of 100 per cent subsidy, we stop short of a complete cost benefit analysis of such a policy, and of any analysis of an optimal public policy.
Childcare may have other effects as well, on both parents and children. It may contribute to the physical and emotional health of mothers (including those without paid employment); by reducing the private costs of reproduction, it may have a positive effect on fertility (Ermisch, 1989). Children presumably benefit from the additional family income permitted by increased parental employment. The direct benefits to children from childcare can include the child's enjoyment of company, as well as learning and developing social and other skills through the educational content of nursery schools, nurseries and playschemes. Such benefits have rarely been quantified. The classic Ypsilanti experiment in Michigan traced long-term labour market returns to a targeted pre-school programme, finding an IRR of 3.7 per cent to two years of pre-school education (Weber et al 1978). The benefits also depend on the type and quality of the provision, as concluded, for example by a broader review by the National Academy of Sciences (1983). The research cited by Cohen and Fraser (1991) and Holtermann (1992) found good quality childcare making a positive contribution to development. Whatever their precise magnitude, there are likely to be intrinsic benefits to children in the child-care provisions envisaged here. Legislation already exists to protect children against poor standards of care. It should go as read that the facilities hypothesised in this paper are of a standard to benefit children directly. Unless they were, we doubt that they would have much of a sustainable impact on labour supply.
There are two further respects in which our exercise stops short of a full cost-benefit analysis. We have not performed the more complicated calculations for the minority of mothers who are not married and in respect of whom saving could be expected on benefits. They are the subject of a separate study by Holtermann (1993). We have also stopped short of calculating the 'deadweight loss' from offering subsidised childcare to those who are already in the labour market, for reasons explained below.
The body of the paper is set out as follows: Firstly we present some background; then discuss the effects of childcare on mothers' earnings; subsequently we compare these with the resource costs of providing the care. We conclude that childcare has an impact on the stock of human capital, as well as on easing the inefficiency and inequity of women's economic disadvantages.
Motivation and method
The story so far
This exercise is motivated in part by comparison with the experience of other European countries. Variations in the degree and pattern of mothers' employment disruption in contemporary Europe are associated with different levels and patterns of subsidised childcare provisions, and other institutions such as the length of the school day (Joshi and Davies, 1992b). For example, in 1988, 74 per cent of the mothers of children under three were in employment in Denmark, compared with 50 per cent in France, 33 per cent in the United Kingdom and 27 per cent in Ireland. Corresponding figures for the proportion of under-threes covered by subsidised day care were: 85 per cent, 20 per cent, 2 per cent and 2 per cent respectively. A detailed reconstruction of mothers' lifetime earnings in Britain, West Germany, Sweden and France showed successively smaller reductions in lifetime earnings compared to a childless base case.(3) This ranking of forgone earnings was attributed to greater availability of childcare facilities, and the long school day in France. In this paper we construct counter-factual simulations of the earnings that might be permitted if British mothers adopted employment histories similar to those of mothers in (the former) West Germany, Sweden and France. For them to do so, we assume that some form of out-of-home childcare would be required, based on experience in these countries. Econometric evidence from the United States indicates that female labour force participation decisions are sensitive to childcare costs (Ribar, 1992).
The method simulates lifetime employment histories and earnings for hypothetical illustrative women, as first suggested by Joshi (1990). The equations used can be found in the Appendix, and the assumptions employed are discussed in detail in Joshi and Davies (1992a). Participation is modelled by a multinomial logit equation which allows for both full-time and part-time employment, and is specified to have effects for the ages and numbers of children. The wage equation allows both for the accumulation of human capital through employment experience (at different rates for full- and part-time employment) and its depreciation by interruptions in employment ('home time'). It also allows for different earnings by those currently in full and part-time jobs. We assume that the illustrative individual is employed whenever the probability of being so exceeds 50 per cent. In this article we confine ourselves to women at the middle range of occupational levels, having GCSE-standard education.
About 53 per cent of British women in their twenties now have educational qualifications at this level, and another 29 per cent have more advanced qualifications (GHS, 1991, Table 10.2). A woman with no qualifications would forgo proportionally about the same, and a graduate (of whom there are far fewer) would forgo less (Joshi and Davies, 1991).
Effect of children and childcare on lifetime earnings
The simulations cover a childless woman, and an otherwise identical woman with two children, born when she is 25 and 28. The earnings costs of motherhood are then calculated as the discounted sum of the earnings of the childless woman from 25-59 less the corresponding figure for the mother. In the base case, the mother stays out of paid work while her children are under school age and returns initially to part-time employment at gradually increasing hours. As compared to her childless counterpart, the mother of two in the base case would spend 15.9 fewer full-time equivalent years in employment. She would be employed full-time for 22 fewer years, but would spend an additional 14 years in part-time employment. The mother would forgo 57 per cent of her childless counterpart's gross earnings between the ages of 25 and 59.(4) We now look at the effect of extensions of childcare on these estimates of a mother's earnings.(5)
We consider only the case of the 'typical' British mother with two children. The standard case (described above) uses no childcare facilities other than those provided as a by-product of the school system. The three extensions to childcare with which we experimented were: one that permitted two extra years of part-time employment while the youngest child is 3 and 4; one where only the years of births are years out of employment, and with part-time employment upgraded to 28 hours per week at full-time rates of pay in all years with any child aged 1-15; and finally another with only two years off for the two births but permitting full-time employment while any child was aged 1-15.
The first extension resembles German-style provision of kindergartens from age three, and could be provided by playgroups or nursery schools. The next variant is inspired by Swedish practice. It would involve care for pre-school children for longer hours and at younger ages than the previous variant: this could be provided by childminders or day nurseries. We have assumed that continuing childcare support when the children are older, complemented by employment protection, would enable women to be employed for long part-time hours, and be paid for them at full-time rates. We have set the hours assumed at 28 per week, roughly the British equivalent of the 80 per cent of full-time hours allowed by Swedish legislation. This applies only when there are dependent children, and the model does not already predict full-time employment.
The final variant assumes provisions for children, similar to those in France, are available which enable our 'typical' mother to be employed full-time. This would involve full day care (for example by childminders or in day nurseries) for pre-school children. Care for children of school age could also be provided by childminders, who in countries like Denmark and Sweden are publicly employed and supported. A radical reform of the school system which could help achieve the French style variant would be a 'continental' school day of the French(6) rather than the German variety. This extension would not involve teachers or formal learning, but supervision of homework, play and feeding on premises preferably in or near the child's school. In this variant our British mother would be employed almost as much as the employment-oriented French mother of the simulations reported in Joshi and Davies (1992b), but she is assumed to take maternity breaks of twelve months rather than three. After her two breaks for maternity leave she is assumed to be in full-time employment at least until she no longer has dependent children.(7) The earnings loss associated with this variant could be cut further by reducing the months of maternity leave, but it must be approaching a limit.(8) The 'Swedish' variant is predicated on larger changes in British employment patterns than the other two variants. It is included to add further choice to either low paid part-time or full-time employment.
The earnings profiles resulting from these experiments are shown in Chart 1 and summarised in Table 1.(9) The earnings gains shown in Table 1 would only apply to those following a 'British' style employment profile. Subject to this proviso, gains would be higher or lower if the woman had a different earning capacity.(10)
The forgone gross earnings for the base case amount to an undiscounted total of |pounds~249,000 over a lifetime, at 1991 prices, setting full-time annual earnings at age 24 to |pounds~10,000. The earnings loss attributable to having two children is 57 percent (undiscounted) of post-25 earnings of a childless woman. The net earnings loss is slightly lower (54 per cent), but if the earnings streams are discounted, the earnings loss increases to 71 per cent of the gross (68 per cent of net) at a discount rate of 5 per cent.
Extending care to part-time cover for the two years when the younger child is three and four (the 'German' variant) does not make much difference: the extra labour supply is only 5 hours per week, on the assumption that the hours of employment while the child is three and four are the same as simulated for age five.(11) Undiscounted lifetime earnings only increase by around 1 per cent of potential earnings after 25, |pounds~6,000. |pounds~2,000 of this is earned in the extra years of part-time employment and the rest in slightly higher rates of pay over the following 27 years.
The second, 'Swedish' style variant, produces undiscounted forgone earnings of |pounds~105,000, half the standard cost, but still a substantial fraction (24 per cent undiscounted, 31 per cent discounted at 5 per cent) of potential gross earnings after motherhood.
In the last ('French') variant there is no part-time employment of mothers of the under 16s, but there is a full year's leave for each maternity. Any pay or allowance at these times is not counted here as earnings. The conversion of part-time to full-time employment reduces total undiscounted gross forgone earnings to |pounds~36,000 (8 per cent of potential). The earnings forgone in this case consist of |pounds~22,000 during the two maternity years and |pounds~14,000 in subsequent pay losses. The undiscounted addition to gross lifetime earnings is |pounds~213,000, adding back 49 per cent of potential earnings after 25. If the earnings streams are discounted to age 25, the relative gain over the standard case rises (from 49 per cent to 57 per cent of potential post-25 gross earnings if discounted at 5 per cent) because the cash stream gained is front-loaded.
Sources of gain
The earnings loss due to having children can be decomposed into four components. Firstly there are the earnings lost in years out of the labour force ('lost years'). Secondly there are losses because a mother may be employed for fewer hours than a childless woman ('lost hours'). These two factors, by lowering employment experience, reduce future earning power: the earnings reduction attributable to this 'lost experience' represents the depreciation of human capital. There is also a penalty in terms of lower hourly rates of pay for part-time employment.(12) The last two together comprise 'lower pay'. The decomposition of undiscounted total costs into these four components is shown in Chart 2. This uses the same procedure for allocating the costs as Joshi (1990), evaluating lost years at base case wage rates. For the base case this method attributes approximately one third each to lost years, lost hours and lower pay. Lost experience accounts for most of the reduced pay, 29 percentage points of the total cost, but 5 points are attributable to the pay penalty for part-time employment. Greater employment experience accounts for about 22 per cent of the earnings gained in moving from the 'British' to the 'Swedish' variant, and about 28 per cent of the earnings gained in moving from the 'British' to the 'French' variant. The human capital component is thus a sizeable fraction of the earnings gain.(13)
TABULAR DATA OMITTED
Sensitivity to policy package and other assumptions
These calculations show that support during both pre-school and school age phases appears necessary to make much of an impact on British mothers' lifetime earnings. Without some support for the employment of mothers during children's school years, any protection of their earning power that might be affected by care for pre-school children would tend to be wasted; without such protection, wage rates after a long break would make a resumption of full-time employment less worth while.
All else equal, the amount of gain will depend on the individual's earning power. As far as the 'typical' employment profile is concerned, the effect of childcare on earnings would be altered by changing the assumption about the number and spacing of births (Joshi 1990).(14) In general, the more woman-years of responsibility for dependent children in a woman's life-history, the more does she stand to gain from childcare services.
Existing investments in childcare
The base-case British mother, whose employment profile we have described above, uses no childcare, but she may be more typical of the early-1980s, when the data for the participation function was collected, than of the 1990s. Over the 1980s, the employment of mothers of pre-school children nearly doubled (from 27 per cent in 1980 to 48 per cent in 1991). The use of both formal and informal childcare increased alongside the employment increase. In 1991 about 23 per cent of all employed mothers used paid childcare, as compared with about 8 per cent in 1980 (Marsh and McKay, 1993). Among mothers with a child under five, full-time employment doubled (from 6 per cent to 12 per cent), and the proportion of these purchasing childcare increased from just under half (47 per cent) to two-thirds (67 per cent) (Martin and Roberts 1984, Marsh and McKay 1993). Many employed mothers who did not make any use of purchased childcare took employment only while the children were at school, while the others used a variety of informal sources (most commonly spouses, parents and parents-in-law). Thus the use of purchased childcare along the lines envisaged in our 'Swedish' and 'French' variants is still quite rare in Britain.
Costs of childcare alternatives
In this section, we place the estimates of the earnings gained from childcare alongside estimates of the costs of supplying the various types of childcare. The policy experiments in the previous section are based on countries where childcare is heavily subsidised, albeit on occasion with income-related contributions from parents. In this section we first compare, for an individual family, total costs with gross earnings benefits. Secondly we examine estimates of the net private gains from using childcare assuming no subsidy, and discuss why more British families do not make the investment themselves. Lastly we examine another extreme assumption--that childcare costs are wholly paid by the state--and calculate the extent to which flowback revenue would cover the expenditure.
Costing childcare provision
The social cost of each childcare variant has been set at two levels of provision, chosen to reflect standards ranging from the average already on offer and a desirable higher quality service. The estimates were compiled by Holtermann (1992). They are expressed as annual amounts at 1991 prices, and all assume that care is purchased for 49 weeks in the year. In the case of 'German' style childcare, care is purchased for the younger child only for two years. The 'low cost' variant assumes that care is provided in an average cost playgroup (|pounds~1421), the high cost variant puts the child in a State Nursery School (|pounds~2300). For the 'French' low cost option, pre-school children (except in years of maternity) are looked after by a childminder at the same annual cost whether one or two children are involved. This sum, |pounds~2050 pa, represents the average actually paid to childminders in several surveys, rather than the rate recommended by the National Child Minders Association. Primary school children participate in after-school and holiday programmes costed on the basis of the few existing clubs at |pounds~1060 per child. Secondary school children fend for themselves. On the high cost 'French' variant, pre-school children (except in years of maternity) attend a day nursery (with a high standard of educational input) at a cost of |pounds~5400 per child. Children of statutory school age right up to 15 participate in after-school and holiday programmes costed at the |pounds~1720 per child needed to reach good standards. We assume that the annual costs of 'Swedish' style childcare are 80 per cent of the 'French' costs.
Table 2 takes one of our hypothetical families and compares the total resource cost of childcare with the resources gained from the programmes. In general, it shows that the resources devoted to out of home childcare would be a lot less than those generated by permitting extra paid work, except for the case of the short 'German' intervention, particularly the high cost version at higher discount rates. This exercise does not, however, quantify the value of childcare provided in the home if the mother does the job herself and forgoes the extra earnings. Presumably the family weighs this benefit (not measured by us) against the full range of costs and benefits of the mother taking paid work.
Net cash gains from childcare
Table 3 shows (in one set of rows) the net gains for the individual family from combining employment and childcare, if the childcare packages were not subsidised at all, but were reliable and satisfactory. These are calculated as earnings gained (after tax and National Insurance) less full childcare costs. From this point of view, the more intensive packages yield substantial gains. The lower cost 'French' variant has the highest net present value, followed by the lower cost 'Swedish' variant.
Why is the British market for childcare so small?
Our calculations on 'German'-style childcare help to explain why the market for this type of childcare may be rather limited. The positive present values of the 'Swedish' and 'French' models, however, lead us to enquire why more British families do not purchase these (or very similar) packages of childcare.
Although the wage rates observed among currently employed women may be higher than those which would be obtained by others, we do not believe that our estimates are badly contaminated by sample selection bias. TABULAR DATA OMITTED The estimation of the wage equation used here (drawn from Ermisch and Wright, 1993) controlled for sample selection bias. The sample selection term, however, was not used in the simulations, and so the simulated wages are those appropriate to the population, not just to the sub-sample of workers.
The British labour market does not often provide the sorts of employment opportunities assumed in the 'Swedish' variant, but this still leaves the 'French' alternative, and in any event only pushes the question from the childcare market to the labour market. It seems unlikely that borrowing constraints form an obstacle to employing the 'Swedish' or 'French' types of care. The annual cash flows are all positive, except for those in the year TABULAR DATA OMITTED when the woman resumes employment after the second birth, using high cost variants.(15) Although there may be unforeseen interruptions to the lifetime employment/childcare plan, childcare decisions can be made year-by-year, and so it seems unlikely that purely financial risks would prove a major deterrent to purchasing childcare.(16) Considerations of imperfect information and hence uncertainty about quality, reliability and continuity of childcare are likely to be very important and to dominate purely financial uncertainty. Any substantial increase in the size of the market for childcare is likely to go hand-in-hand with better information about the quality and reliability of the childcare.
Households, however, maximise utility rather than profits, and it seems most likely that the unsubsidised price of these types of childcare is simply too high for the mother's extra earnings to compensate most British families for the reduction in her time in the home. The woman's time at home has value. Though the gains over a lifetime from the 'Swedish' and 'French' styles of childcare are large, the hourly wage rate, net of taxes and childcare costs is not very high in the early years. For example, at age 32 the net hourly wage of a woman employing the 'Swedish' or 'French' styles of childcare is simulated at about |pounds~3 on the low cost variants, and under |pounds~1 on the high cost variants.
Both theory and some survey evidence suggest that time spent on domestic labour is not reduced hour for hour by time spent in paid labour. Full-time female employees have 20.5 fewer hours of 'free time' per week than housewives (Social Trends, 1993, Table 10.3). Therefore the family might be considered as having to pay her at overtime rates for (at least some of the) time spent in the market, whereas the market would only reward her at straight time (and 'female' straight time net of childcare costs at that).
British mothers commonly cite problems about childcare as an important reason for not taking employment (Martin and Roberts 1984, Witherspoon and Prior 1992, McRae 1993). Women in Sweden, and many in France, do follow these employment patterns, and for them childcare is cheaper.
Impact of childcare subsidy
To illustrate some of the effects of a change in public policy towards subsidising childcare costs, we contrast the zero subsidy case with the other extreme of 100 per cent subsidy. This is more attractive distributionally than a tax relief policy, though less attractive than a subsidy TABULAR DATA OMITTED inversely related to income, as recommended by Holtermann (1992). Table 3 also shows the gains to an individual family (net of tax and National Insurance) which would result in this case. The relative pattern of gains is similar to that without subsidy, except that the low and high cost variants now yield the same cash gain to the family. Table 4 shows, for an individual family, the revenue to the public purse generated by the extra earnings, the 'flowback' into income tax and National Insurance contributions. This is compared to the cost of providing the facilities with 100 per cent subsidy to this particular family. Where the 'flowback' is greater than 100 per cent, as in the case of the low cost 'Swedish' and 'French' variants, the fiscal revenue from the woman's extra earnings more than offsets the costs of providing the childcare. The high cost 'French' variant generates more revenue than it costs at the lower discount rates, but the high cost 'Swedish' variant never quite pays for itself. At subsidy levels below 100 per cent, however, these high cost variants could cover their costs.(17) The 'flowback' covers a much smaller fraction of costs of the 'German' variants, down to as little as 17 per cent in the high cost variant discounted at 5 per cent.
With 100 per cent subsidy about half the (undiscounted) cumulative gains to the individual family of either of the 'Swedish' or 'French' variants are attained after ten years, and 80 per cent after 20 years. The gains to the public sector accrue more slowly. In the case of the low cost 'Swedish' variant, 19 per cent of the cumulative gains have accrued after 10 years and 62 per cent after 20. The gains for the low cost 'French' variant accrue slightly faster: 25 per cent after 10 years and 71 per cent after 20. For these two variants the public sector cash flow is positive in each year. The high cost 'French' variant, though ultimately paying for itself, shows a loss almost as large as its ultimate gain after 10 years, but the loss is only 7 per cent of the final gain after 20 years. The high cost 'Swedish' variant overshoots its final loss massively: after 10 years the cumulative loss is 20 times its final level, and even after 30 years it still shows a cumulative loss about 2 3/4 of its final level.
Towards cost benefit analysis
Our calculations stop short of a full cost-benefit analysis (as attempted by Cohen and Fraser 1991), or even a public-expenditure costing of a policy of subsidising childcare (as in Cohen and Fraser 1991, Holtermann 1992, Holtermann and Clarke 1992). Tables 2 and 4, however, together provide one element in such calculations in greater detail than attempted elsewhere, particularly in accounting for the long-term effects of childcare on the conservation of human capital. Calculations of the type presented above could be extended to cover a distribution of skill levels and family sizes. Some of the studies referred to above have also taken account of wider public expenditure aspects.(18)
Though our results confirm that sizeable resource gains and flowback to the public finances might be expected from greater use of childcare, these calculations of earnings and fiscal revenue gained apply only to the case of a woman whose labour market participation decision changes as a result of using childcare. There will be a deadweight loss from take-up of subsidised childcare by mothers who would have been employed, and for as many hours in any case, presumably making private or informal childcare arrangements. We have no evidence to guide the necessary assumptions about how many of those with existing childcare arrangements, most of which are informal, would switch to the hypothetical subsidised programmes, nor how much paid time would be covered by switches into the programmes. Most of the existing employment of mothers of pre-school children is part-time. To the extent that the subsidies permitted longer hours of employment, there would be benefits, not estimated in detail here to set against the deadweight loss. Where formal care replaces informal, however, there may also be some welfare gains if the former is of higher quality than the latter. A more thorough appraisal of a subsidy policy would require much more detailed formulation of an actual policy than we have attempted. Details of the method of provision (e.g. vouchers versus subsidies to providers), eligibility and monitoring arrangements can be expected to have large effects on both the take-up of any subsidy offered, and the extent of deadweight loss incurred. Bearing all this in mind, the estimates of revenue gains as percentages of costs given in Table 4 can be transformed to give estimates of the 'policy hit rate' required for the public purse to break even. For example, with a revenue gain of 201 per cent of cost (low cost 'Swedish' care discounted at 5 per cent) half of the woman hours getting a 100 per cent subsidy for childcare would need to be 'new' hours for the policy to break even.
We have not attempted to model lone mothers here: this involves complex questions of the interactions of benefit, taxes and wages as well as childcare. The possibilities of positive labour supply responses from this source should be put on the table alongside the considerations of deadweight loss from existing child-care users gaining from subsidies. Holtermann (1993) estimates that a package of policies including better access to childcare could bring 200,000 lone mothers into employment and off benefit. The number corresponds roughly to the shortfall in lone mothers' participation rate compared to married mothers'. Though we have not estimated deadweight loss, neither have we counted in such gains to the Treasury.
Here we have been primarily concerned with the long-run effects of increased provision of childcare on the potential earnings of individual women, but we have alluded to some of the general equilibrium effects in the introduction. As regards short-run adjustment, other considerations also arise in aggregating the individual gains. Some mothers may gain employment at the expense of other workers, and/or the short-run equilibrium wage rate may be lowered. Quantification of these effects depends on a view of short-run labour market dynamics. If the distribution of effective human capital between men and women became more equal, the employment and earnings consequences of short-term fluctuations around the long-run equilibrium rate would be borne more equally by men and women.
Co-investment in childcare
British employers provide only a tiny amount of help with daycare,(19) though they might experience gains in the form of reduced training and recruitment costs from greater provision of out-of-home childcare. The difficulty that individual employers might experience in appropriating such gains has been discussed above. In addition, employees are generally taxed on the value of childcare, so it has no special attractiveness in compensation packages.(20) Perhaps also the recession of the early-1990s is masking a trend towards much higher levels of employer financing. Government policy, however, now encourages partnership between employers, parents and other agencies in the provision of out-of-school hours care by means of pump-priming subsidies. As yet, however, only a very limited scheme is planned, with a target of 50,000 places over the next three years for an outlay of |pounds~45mn (Department of Employment, 1993). The model of joint investment by parents, employers and the state has been advocated by on a wider scale by Holtermann (1992), among others.
1. The benefits of establishing subsidised day care in Britain would include, besides the direct educational benefits to the children, the increased earnings permitted to mothers, not just while their children are receiving day care, but until they retire.
2. Experimental calculations for the 'typical' British mother of two suggest that provisions, such as nursery schools, which permit her to take part-time employment when her younger child is three and four make very little difference to lifetime earnings. Facilities enabling her to take employment full-time or for long, well-paid part-time hours until her children leave school would make the biggest difference.
3. Not every mother is expected to choose full-time employment and childcare over options where she spends more time at home, even if she loses cash thereby: the cash gains forgone give one indication of the value that is implicitly placed on home child rearing. Neither is every mother expected to stay at home. The gains demonstrated as likely to follow from using childcare over the child rearing years may help to explain why the employment of mothers of pre-school children nearly doubled during the 1980s.
4. Our calculations of the public sector revenue flowback suggest that, even in narrow financial terms, childcare subsidies could prove a worthwhile investment for HM Treasury, at least in the case of those mothers who were not previously employed. Extending our calculations to provide aggregate public expenditure costings of a subsidy policy would require: detailed formulation of policy options; estimates of savings on benefits to lone mothers; estimates of deadweight loss resulting from take-up of child care subsidy by those already using other forms of childcare, and estimates of any labour market displacement effects.
5. Childcare is an investment in human capital. Here we have been concerned with the human capital embodied in mothers rather than children. Particularly if daycare continues while children are at primary school, the earnings gained represent conservation of this human capital. For the woman of middle-level skills, whom we have considered, about 20 per cent-30 per cent of the earnings gained arose from wage conservation. For the increasing number of women with higher skill levels, there is more to conserve.
6. Arrangements for the care of children affect, among other things, economic inequality between the sexes. Inequality in the distribution of human capital between the sexes is not just a matter of education and training: employment experience plays a major role in conserving human capital. It seems unlikely (particularly in the light of experience in other European countries) that the part of the nation's human capital embodied in its mothers will be conserved or utilised as fully as it could be, or as these women would wish, unless (among other things) the price of childcare to families is reduced. If this goal is sought, parents, the state and employers may all have a role to play in the financing of childcare.
7. The failure of the market to offer equal opportunities to men and women is a failure of efficiency as well as equity. We have argued that public policy to facilitate childcare has a strategic role to play in reversing this process.
TABULAR DATA OMITTED
TABULAR DATA OMITTED
(1) Easing of skill bottlenecks might also be expected to contribute to short-run price stability in the face of aggregate demand shocks. A more experienced labour force may also reduce the equilibrium unemployment rate (see e.g. Phelps, 1972).
(2) Green (1993), using data from the 1987 GHS, shows that women were significantly less likely to get training than men. Having a dependent child, or being a part-time worker, had a significant, negative effect on the probability of a women (but not a man) receiving training.
(3) For illustrative cases of two child families, and on a common assumption of initial labour force attachment. At a lower level of attachment the ranks of France and Sweden reverse.
(4) These estimates also appear in Joshi and Davies (1992b). They can be compared to the figure of 'almost exactly one half' originally calculated by Joshi (1990).
(5) The participation equation includes a term which lowers the probability of full-time work per child over 16 (ever-born rather than those still present in the household). This term works its way through into the estimated opportunity cost, accounting for 6 per cent of the undiscounted total in the base case. It most plausibly reflects interruptions in employment rather than motherhood per se, and so including it overstates the earnings lost in variants which substantially reduce these interruptions. We have therefore suppressed this term in the simulations of the second and third variants reported below.
(6) More French than the French in fact because it would not include a Wednesday break.
(7) If the woman was permitted by maternity leave provisions to return to her original employer, there could be some benefits to her future earning power for which we have not specifically allowed. The wage equation used here does not allow for the premium on continuity of employment within the same firm, though other analyses of the same data suggest this could exist. The formulation which omits job-tenure from the earnings function was taken as being more generally applicable for this type of exercise, and certainly more convenient in the absence of a model to predict when women change employers.
(8) It is of course possible to imagine foregone earnings approaching zero by imagining employment and care arrangements which permitted mothers to earn as much as they could have without children--day and night care facilities from birth to majority for example--but there is likely to be neither demand for nor supply of such facilities. They not only seem socially absurd but would also be absurdly expensive, and probably only cost effective for their highest paid mothers. The result could conceivably be achieved by complete role reversal between parents, with childbirth taking place during annual paid holiday, and the father staying at home thereafter, but then we should have to be concerned with the opportunity cost of his earnings, very often likely to be higher than hers.
(9) These simulations tend to produce a movement into part-time work by women in their fifties, whether or not they have had children. Thus, for example, the childless woman goes part-time at the age of 53.
(10) In simulations of women with the earning power of graduates, for example, mothers are likely to return to employment anyway by the time their child is three or younger (Joshi and Davies, 1991).
(11) If the hypothetical provisions enabled part-time employment in those two years at longer hours, up to 30 per week, earnings losses could be cut by a further |pounds~11,000.
(12) Job downgrading on return to work is not directly modelled here, but is reflected in the penalty attaching to part-time wage rates. See Joshi (1990, p.43, footnote 10).
(13) This is using an earnings function estimate across all occupations. If the coefficients of the earnings function were allowed to vary across occupations, in some the human capital gains would be greater, and in some smaller.
(14) The effectiveness of provision for the 3-4 year old age bracket, for example, would be somewhat greater per woman if the two births were more widely spaced, or if there were more children.
(15) For both the 'German' variants, on the other hand, the cost of childcare exceeds the earnings gained in both the years when it is employed.
(16) Consider a high cost case: a woman considering return to work after the birth of her second child, and contemplating the use of the high variety of 'French' childcare. She will make a loss if she is employed for under three years. With three years' employment, the internal rate of return on childcare is 7 per cent, rising to 34 per cent after four years, and 58 per cent after five years.
(17) With a state subsidy of two thirds, for example, the high cost 'Swedish' and 'French' variants will yield a public sector surplus at discount rates up to 10 per cent and 25 per cent respectively.
(18) Cohen and Fraser have also estimated the revenue from consumption taxes resulting from the extra earnings of mothers, and the tax and national insurance of the workers in daycare. They have also taken account of savings arising from lone mothers moving from dependence on social security benefits into paid work, as has Holterman (1993).
(19) About 1 per cent of women employees whose youngest child was of primary school age used 'workplace nurseries' (Holterman and Clarke, 1993, Table 1.3). There are, of course, other types of help which employers can give.
(20) There are narrowly drawn cases where in-kind childcare benefits are tax-exempt. The principal conditions are that the employer, perhaps in conjunction with others, makes premises available for a nursery or playscheme and is wholly or partly responsible for the finance and management of the scheme. Additionally, where an employee is not a director and earns less than |pounds~8,500 per year, the costs of childcare directly engaged and paid for by an employer are not taxable. (Inland Revenue, 1992).
Bone, M., (1977), Pre-School Children and the Need for Daycare, London: HMSO.
Cohen, B. and N. Fraser, (1991), Child Care in a Modern Welfare System: Towards a New National Policy, London: Institute for Public Policy Research.
Department of Employment (1993), 'Childcare help for working parents', Employment Gazette, January, 647.
Ermisch, J.F., (1989), 'Purchased child care, optimal family size and mothers' employment: theory and econometric analysis', Journal of Population Economics, no. 2, pp. 79-102.
Ermisch, J.F. and R.E. Wright, (1993), 'Wage offers and full-time and part-time employment by British women', Journal of Human Resources, 28, Winter, pp. 111-133.
GHS, (1991), General Household Survey, no. 22, London: HMSO.
Green, F., (1993), 'The determinants of training of male and female employees in Britain', Oxford Bulletin of Economics and Statistics, no. 55, pp. 103-122.
Haskell, J. and Martin C., (1993), 'Do skill shortages reduce productivity? theory and evidence from the United Kingdom', Economic Journal, no. 103, pp. 386-394.
Holtermann, S., (1992), Investing in Children: Costing and Funding an Educational and Day Care Service for Young Children, London: National Children's Bureau.
Holtermann, S., (1993), Becoming a Breadwinner: Policies to Assist Lone Parents with Childcare, London: the Daycare Trust.
Holtermann, S. and K. Clarke, (1992), 'Parents, employment rights and childcare', Research Discussion Series, No. 4, Manchester: Equal Opportunities Commission.
Inland Revenue, (1992), Tax and Childcare, Personal Taxpayer Series, IR115.
Jarvis, V. and S.J. Prais, (1989), 'Two nations of shopkeepers: training for retailing in France and Britain', National Institute Economic Review, no. 128, May, 58-74.
Joshi, H., (1990), 'The cash opportunity cost of childbearing: an approach to estimation using British data', Population Studies, no. 44, pp. 41-60.
Joshi, H. and H. Davies, (1991), 'Pension splitting and divorce', Fiscal Studies, no. 12, November, pp. 69-91.
Joshi, H. and H. Davies, (1992a), 'Child care and mothers' lifetime earnings: some European contrasts', CEPR Discussion Paper no. 600, London: Centre for Economic Policy Research.
Joshi, H. and H. Davies, (1992b), 'Daycare in Europe and mothers' forgone earnings', International Labour Review, no. 132, pp. 561-579.
Katz, E. and A. Ziderman (1990), 'Investment in general training: the role of information and labour mobility', Economic Journal, no. 100, pp. 1147-1158.
McRae, S., (1993), 'Returning to work after childbearing: opportunities and inequalities', European Sociological Review, vol. 9, no. 2, Sept 1993, pp. 125-138.
Mankiw, N.G., D. Romer and D.N. Weil, (1992), 'A contribution to the empirics of economic growth', Quarterly Journal of Economics, no. 107, pp. 407-437.
Marsh, A. and S. McKay, (1993), 'Families, work and the use of childcare', Employment Gazette, August, pp. 361-370.
Martin, J. and C. Roberts, (1984), Women and Employment a Lifetime Perspective, London: HMSO.
National Academy of Sciences, (1983), Children of Working Parents, Washington DC: National Academy Press.
Petrie, P. and P. Logan, (1986), After School and in the Holidays: the Responsibility for Looking After Schoolchildren, London: Thomas Coram Research Unit.
Phelps, E.S., (1972), Inflation Policy and Unemployment Theory, New York: Norton.
Ribar, D.C., (1992), 'Child care and the labor supply of married women: reduced form evidence', Journal of Human Resources, no. 27, pp. 134-165.
Scott, G., (1989), Families and Under Fives in Strathclyde, Glasgow: Strathclyde Regional Council.
Simpson, R., (1986), The Cost of Childcare Services, in Children and Equal Opportunities, Manchester: Equal Opportunities Commission.
Social Trends, (1993), No 23, London: HMSO.
Weber C.U., P.W. Foster and D.P. Weikart, (1978), An Economic Analysis of the Ypsilanti Perry Preschool Project, Ypsilanti, Michigan: High/Scope Educational Research Foundation.
Witherspoon, S. and G. Prior, (1992), Working Mothers: Free to Choose?, in Jowell, R., L. Brook and B. Taylor British Social Attitudes: the 8th Report, Aldershot: Dartmouth, pp. 131-154.