Skilled or unskilled? To work or not? Female Decisions under Imperfect Information∗ Alessandra Casarico Department of Institutional Analysis and Econpubblica, Università Bocconi, Milan, Italy; CEPR, London, UK; CESifo Munich, Germany. Paola Profeta Department of Institutional Analysis and Econpubblica, Università Bocconi, Milan, Italy; CESifo Munich, Germany; First draft. Very preliminary: October 2009 ∗ Alessandra Casarico, Università Bocconi, Via Roentgen 1, 20136 Milan, Italy; alessan- [email protected]; Paola Profeta, Università Bocconi, Via Roentgen 1, 20136 Milan, Italy, [email protected]. We thank Carlo Devillanova and Vincenzo Galasso for their comments. 1 Abstract We develop a two-period model to jointly analyze the female decisions of investing in education and of participating in the labor market. In the first period women, who are heterogeneous in talent, decide whether to invest in education, or remain unskilled and start working immediately. In the second period they have a kid and they make their participation decision: the skilled decide whether to enter or not the labor market; the unskilled whether to continue or drop out. Kids are costly, and the cost is divided among parents. This cost may be of a high or a low type, and the true type is revealed only after the child’s birth. Female education decisions are thus taken under imperfect information on the cost of the kid. We solve the model backward and we characterize the participation and education decisions. We find that, due to the imperfect information, there are instances where women who have invested in education find it profitable not to work. We also find that imperfect information may be associated with a higher output than what would result in the case of perfect information, since in the former case highability women with a high cost kid may find it convenient to invest in education and work. We then study a tax-transfer scheme targeted on working women as a specific example of government’s intervention to sustain both female decisions of investing in education and work. In the presence of this policy it’s more likely that more women participate to the labor market when compared to the case where the government does not subsidize the child cost. Their expectation to participate may in turn induce them to invest in education. The policy also reduces the amount of unused human capital related to the presence of educated women who decide not to participate. Keywords: child costs, imperfect information, tax-transfer scheme. JEL Classification: J16, J24. 2 1 Introduction Industrialized countries are characterized by large cross-country differences in terms of female participation and employment rates. The same can be said if one focuses on education and looks at the share of women with upper secondary education. While for instance in Italy only 48% of women in the 25-64 cohort had at least upper secondary education in 2004, in Sweden they were 85%. If we focus on younger cohorts this gap is smaller but not yet closed. Not only are there differences in terms of the share of educated women, but also in terms of the percentages of educated women who work, indicating that female human capital is often underutilized. Among women with at least upper secondary education, 61% of Spanish women works, while more than 74% of women works in countries such as Denmark, Sweden, the Netherlands and the UK. Figure 1 and 2 suggest that countries where the number of educated women and the percentages of educated women who work are higher are those where public subsidies to home produced goods (child care for instance) are larger, or where expenditures on family policies account for a greater share of GDP or where measures of flexibility on the labor market are more used (part-time for instance).1 Scandinavian countries are an example.2 Previous contributions have studied the impact of cross-country differences in institutions on the female labour market participation and on the fertility decision (see Del Boca 1 In Figure 1a we plot the share of female with at least upper secondary education in 2004 (OECD, 2006) and the availability of part-time jobs (same year) based on Eurostat data. In Figure 1b we plot the share of female with at least upper secondary education and the final child care score for children of age 0-2 taken from De Henau et al. (2007). In Figure 1c we plot the share of female with at least upper secondary education and family expenditure as a percentage of GDP taken from the OECD Family Database and referred to 2003. In Figure 2a, 2b and 2c on the vertical axis we have the share of educated women who are employed in 2004 (OECD, 2006), instead of the share of educated women. These are simple correlations and do not imply any causal relationship. 2 While the focus here is on the role of institutions, how good the labour market is in terms of rewarding human capital investments can contribute to explain the cross-country differences in female education and employment. This explanation is however silent on the existing cross-country differences in the gaps in employment rates of women with/without children: while the employment rates of women with children are typically lower than the ones of women without children, for instance in France or Scandinavian countries this gap is smaller than in Italy or Germany. 3 et al., 2007 and De Henau et al., 2007 for instance). The relationship between institutions and educational choices is instead less explored. Including educational choices into the analysis of female employment is however essential to shed a light on what can be done to build further human capital and to reduce the unused human capital of educated women who do not work. Why are there women who invest in education but do not participate in the labor market? Is there a policy which can support the education and participation decisions? In this paper we build a two-period model to jointly analyze the female decisions of investing in education and of participating in the labor market. For simplicity, we consider that these decisions are both binary, i.e. a woman has to decide whether to become skilled or remain unskilled and whether to work or not. In the first period of time women, who are heterogeneous in talent, decide whether to invest in education or not. If they invest, they devote the entire first period to education. If they do not invest, they remain unskilled and they start working. At the beginning of the second period, all women have a child. Kids are costly in terms of time that they require. The cost of the kid is divided among the parents. If they both work, each of them has to buy the own share of required child care time on the market. However, the mother can also live on home production, differently from the father, and thus she can decide whether to stay at home and provide personally her due time to the kid, or to work and buy it on the market. We assume that the cost of the kid may be of two types, high or low, and the true cost of the kid is revealed only after her birth. This heterogeneity captures the idea that some kids are more demanding than others, due for instance to her health status or good nature, or due to the fact that grandparents or informal networks help taking care of the kid. Female education decisions are taken under imperfect information on the cost of the kid: women only know the probabilities that the kid is either a high or a low-cost type. At the child birth, women make their participation decision: the skilled decide whether to enter or not the labor market; the unskilled whether to continue or drop out. Their decision will be based on the comparison between their consumption possibilities when participating to the labor market and when staying at home, assuming that in this latter case consumption depends on a given level of home production. 4 We solve the model backward starting from the participation decision: we identify the cut-off level of ability such that skilled women find it convenient to participate and the minimum unskilled wage required for the unskilled to find it convenient to continue working after child birth. As expected, we find that, for a given level of child cost, if an unskilled woman works, so does a skilled one. We also find that, for a given level of talent, if a woman with a high-cost kid works, so does a woman with a low-cost kid. Knowing the probabilities to participate, given the ability and cost types, we examine the education decision and we calculate the cut-off levels of ability such that women find it profitable to invest in education. We find that, due to the imperfect information, there are indeed instances where women who have invested in education find it profitable not to work. We also find that imperfect information may be associated with a higher output than what would result in the case of perfect information, since in the former case high ability women with a high-cost kid may find it convenient to invest in education and work. The government can intervene to sustain both female decisions. We study a taxtransfer scheme targeted on working women as a specific example of government’s intervention to sustain both female decisions of investing in education and work. In particular, the government provides a subsidy which is proportional to the kid’s cost if the mother works, and which is financed by a proportional tax on wages paid by all workers. In the presence of this policy more women participate to the labor market when compared to the case where the government does not subsidize the child cost. Their expectation to participate may in turn induce them to invest in education. By working as an insurance device against the risk of having a high-cost kid, the tax-transfer scheme also reduces the amount of unused human capital related to the presence of non-working educated women. A recent growing literature analyzes how gender gaps in wage and in participation can arise when the distribution of ability between men and women is the same (Francois, 1998; Albanesi and Olivetti, 2008; Bjerk and Hahn, 2007, Lommerud and Vagstad, 2007). These contributions share a common assumption, namely, skills are given and cannot be accumulated. While the focus of this paper is on explaining female education and participation decisions rather than on gender gaps, we complement the previous studies by including 5 the education decision. The role of education is the focus of a very recent literature which analyses the effects of pre-marital schooling decisions on the marriage market: educational attainment may influence the marital surplus share that can be extracted in the match, by increasing the prospects of marrying an educated partner and by affecting the competitive strength in the bargaining within the couple (Iyigun and Walsh, 2007b; Chiappori et al., 2008). We here abstract from gains in the marriage market as a motivation to acquire education: we will however perform some comparative statics exercise to see how a change in the share of the child rearing costs which are borne respectively by the mother and the father influences education and participation decisions. There is also a literature on the effects of child care subsidies or public child care provision on female employment (see for instance Ohlsson and Lundholm, 1998; Rosen, 1997). However the effects that these policies can have on the decision to invest in education is never addressed, as far as we know. Our paper is also to some extent related to the literature that analyzes the role of imperfect information in education investments. Bernhard and Zilcha (2009) find that, on the one side, better information increases the efficiency of aggregate human capital formation, while on the other side, by decreasing the risk sharing opportunities, if agents are very risk averse, welfare may be higher under a less informative system. In our framework imperfect information may have the positive effect of inducing high ability types, who turn out to have a high-cost kid, to invest in education given their expectation of having an average-cost kid. They would have not found it convenient had they known the true cost. If they end up working as skilled individuals, this will represent a gain of output. The paper is organized as follows: the next Section develops the model, Section 3 studies the implications of imperfect information. Section 4 introduces the tax-transfer scheme and it analyses the role of government and Section 5 concludes. 6 2 The model 2.1 General features We develop a two-period model. Women are heterogeneous in talent , which is distributed on the interval [1 ] with a distribution function and a continuous density function (·) and is known to women, firms and government. In the first period of time women decide whether to invest in education or not. If they invest, they devote the entire first period to education and they become skilled. If they do not invest, they remain unskilled and they start working and receive a salary equal to , which is transferred to the second period at an exogenous interest rate . At the beginning of the second period, all women have a child and they decide whether to work or not. The participation decision is again binary: the skilled decide whether to enter or not the labor market; the unskilled whether to continue or to stop. Their decision will be based on the comparison between their consumption possibilities when participating to the labor market and when staying at home. Consumption takes place at the end of the second period of life when agents consume all their lifetime income. Kids are costly, in terms of the time that they require. We assume that the time required by a kid may be of two types, high or low: = . This is to capture the idea that some kids are more demanding than others, due for instance to her health status or good nature, or to the presence of grandparents or informal networks that help taking care of the kid, thus reducing the amount of time to be financed by parents. The cost of the kid is divided among the parents: the mother is responsible for the amount (1 − ) and the father for the amount , with ∈ (0 1) indicating the division of time of child care in the couple. To start with, we assume that is exogenous and we analyze the implications for our results of different values of . Each parent may decide to provide personally the required child care time for the kid or to buy it on the market. Thus, for the mother and the father the child time requirements are, respectively: 7 (1 − ) = + (1) = + where and are the time that the female and male adult decide respectively to spend personally with the kid and and denote the time bought on the market. We assume that a unit of time bought on the market costs . For each adult, the total amount of time of their second period of life, whose length is 1, may be allocated to labor with = , to home production and to time spent with the kid , so that the following second period time constraints are satisfied: 1 = + + (2) We start with the allocation of time of the mother. If she decides to work, = 1 and thus = = 0. Thus, the mother has to buy all the time of child care under her responsibility on the market: (1 − ) = at the cost of (1 − ). Let = ; then the cost of a kid for a working woman is (1 − ). Working on the market, all unskilled women receive the same wage, equal to , while a skilled woman with ability receives a salary equal to . Thus, the consumption possibilities for a skilled working woman and an unskilled working woman whose kid costs are, respectively: = − (1 − ) (3) = − (1 − ) (4) and If, on the contrary, she decides not to work, = 0. The kid’s time requirement borne by the mother is supplied by her directly and she does not buy any child care on the market, that is (1 − ) = . Thus, given the time constraint, she can devote to home production an amount of time equal to = 1 − (1 − ). Indicating with ∈ (0 1) the amount of home production that a woman can deliver if = 1 when a woman decides ¤ £ to stay at home, her consumption possibilities are equal to 1 − (1 − ) , which can be 8 rewritten as − (1−) Assume for simplicity that = 1; the consumption possibilities for non working women are therefore: = − (1 − ) (5) Differently from the mother, we assume that the father does not produce anything at home, i.e. = 0 for him and thus he always works on the market:3 = 1 and thus = = 0 He receives a wage equal to if he is unskilled and equal to if he is skilled and his ability is equal to .4 Thus, the consumption possibilities for a skilled and an unskilled man whose kid cost is are, respectively: = − = − (6) At the time of the child’s birth, the woman decides whether to work or not knowing the true cost of her kid and comparing her consumption possibilities in the two cases. The education decision instead is taken under imperfect information on the cost of the kid: we assume that with probability the cost is high, 1, and with probability (1 − ) the cost is low, 1, where . Women and firms know and Firms operate in a perfectly competitive environment. 2.2 Individual problem We focus on female education and participation decisions. Figure 3 summarizes the 2stage game and pay-offs: in the first step a woman decides whether to become skilled or not, forming expectations on the cost of her kid; in the second step, the kid is born, the true cost is revealed and the woman decides whether to work or not. 3 This is equivalent to assuming that women have a comparative advantage in home production. The assumption is instrumental to obtaining that men always work, which is in line with the evidence that on average - male participation rates are higher than female ones. A similar result would be obtained if we assumed that male wages are always such that their consumption possibilities are greater when they participate rather than when they stay at home. 4 The distribution of market abilities is assumed to be the same across males and females. 9 We solve the model backward starting from the participation decision. A skilled woman with ability and with a kid of cost (with = ) decides to work if her consumption possibilities in the case she works are larger than those in the case she does not work. Comparing (3) and (5) we obtain: − (1 − ) ≥ − (1 − ) (7) which delivers the following threshold level of ability: ≥ + (1 − )(1 − ) (8) Notice that, when the cost of the kid is high, a higher level of ability is required to work, with respect to the case where the cost of the kid is low. Moreover, the higher the time of child care under the responsibility of the mother, represented by 1 − , the higher the level of ability such that a woman find it convenient to work. Similarly, comparing (4) and (5) the unskilled continue working in the second period if the following condition is satisfied: − (1 − ) ≥ − (1 − ) (9) i.e. if the wage of the unskilled is larger than the following threshold: ≥ + (1 − )(1 − ) (10) As expected, this condition is more easily satisfied for women with a low-cost kid than for those with a high-cost one. Again, it is also more easily satisfied when men participate more in taking child-care responsibilities within the couple (higher levels of ). Since ≥ for all and we have three possible cases depending on the level of wage of the unskilled: • Case 1: + (1 − )(1 − ) • Case 2: + (1 − )(1 − ) • Case 3: + (1 − )(1 − ) + (1 − )(1 − ) We now analyze separately the participation and education decisions in these three scenarios. 10 2.2.1 Case 1: All women work If the wage of the unskilled workers is large enough to induce unskilled women to continue working after the child’s birth, i.e. + (1−)(1−) then all skilled women, given their higher salaries, will also find it convenient to work. Having the expectation that they will certainly work, they decide to invest in education by comparing their consumption possibilities in case they work as skilled or as unskilled. The decision is taken under the expectation that, with probability the cost of the kid is equal to and with probability (1 − ) it is equal to . Remembering that unskilled women work for two periods, a woman invests in education if her consumption as skilled is larger than the one as unskilled, i.e: − (1−)−(1−) (1−) ≥ (1+)+ − (1−)−(1−) (1−) (11) which delivers the following threshold level of ability such that all women with an ability type larger than this invest in education: ≥ (2 + ) (12) Since both skilled and unskilled women will work, the investment in education depends only on the wage premium, i.e. the ratio between the unskilled and the skilled wage. This case is similar to what happens to men: since, as we argued before, men always work, their education decision depends only on the wage premium, i.e. men with a level of ability such that ≥ (2 + ) invest in education and the others remain unskilled. Interestingly, if the wage premium for female is larger than the one for males, women have stronger incentives to educate than men. 2.2.2 Case 2: Unskilled women do not work If the wage of the unskilled workers is below the threshold + (1 − )(1 − ), none of them will find it convenient to continue working after the child’s birth. For the skilled, different possibilities arise depending on their level of ability. • Case 2a 11 All skilled women with a level of ability such that + (1 − )(1 − ) (13) find it convenient to work. Given this expectation, they will decide to invest in education if their consumption possibilities in the case they acquire education are larger than if they do not, i.e. if the following condition is satisfied: ¤ £ ¤ £ − (1 − ) + (1 − ) − (1 − ) £ ¤ £ ¤ ≥ (1 + ) + − (1 − ) + (1 − ) − (1 − ) i.e. if their level of ability is above the following threshold: ≥ (1 + ) + + (1 − )(1 − ) (14) where = + (1 − ) . Interestingly, in this case a woman invests in education on the basis of the expected value of the cost of the kid. • Case 2b All skilled women with a level of ability such that + (1 − )(1 − ) + (1 − )(1 − ) ≤ ≤ (15) find it convenient to work if they have a low-cost kid and not to work if they have a high-cost kid . Knowing this, they will decide to invest in education if the following condition is satisfied: ¤ £ ¤ £ − (1 − ) + (1 − ) − (1 − ) £ ¤ £ ¤ ≥ (1 + ) + − (1 − ) + (1 − ) − (1 − ) (16) which delivers the following threshold level of ability: ≥ (1 + ) + (1 − )(1 − ) + (1 − ) 12 (17) This means that all women with ability levels below this threshold remain unskilled. Notice that not all skilled women work: women who have invested in education under the imperfect information about the cost of their kid may find it convenient not to work if this cost turns out to be high. Their human capital is thus wasted or unused, in the sense that it does not translate into output on the market.5 • Case 2c All skilled women with a level of ability such that + (1 − )(1 − ) (18) find it convenient not to work. Under the expectations of not working it is immediate to verify that no woman invests in education. To summarize, in Case 2 we have found that: low ability women (the ones with (1+) + (1−)(1−) + ) (1−) ≤ abilities ( + (1−)(1−) remain unskilled and do not work; for intermediate levels of ≤ + (1−)(1−) ) women invest in education, and a fraction (1 − ) of them, who turns out to have a low-cost kid, works, while a fraction who have a high-cost kid, does not work; the high ability types ( + (1−)(1−) ) invest in education and work. 2.2.3 Case 3: Women with a low-cost kid work If the wage of the unskilled workers is such that + (1 − )(1 − ) + (1 − )(1 − ) the unskilled women with a low-cost kid find it convenient to continue working after the child’s birth, while the unskilled women with a high-cost kid decide to drop out. Since unskilled women with a low-cost kid work, we are sure that also skilled women with 5 We are not explicitly accounting for the benefits that human capital investment provides when it is not employed on the market. This is not to say that human capital investments by non-working mothers do not have positive effects on the economy (for instance through private benefits to children or through externalities to the society as a whole). However, as long as working on the market does not completely crowd out these effects, there are further benefits (or lower losses) to be reaped when educated women work on the market. On the effects of maternal employment on children’s human capital see Bernal (2008). 13 a low-cost kid do indeed work. The skilled with a high-cost kid instead may be of two types, depending on their level of ability. We analyze them separately. • Case 3a All skilled women with a high-cost kid and a level of ability such that + (1 − )(1 − ) (19) find it convenient to work. Given this expectation, they decide to invest in education if: £ ¤ £ ¤ − (1 − ) + (1 − ) − (1 − ) £ ¤ £ ¤ ≥ (1 + ) + − (1 − ) + (1 − ) − (1 − ) (20) which requires a level of ability above the following threshold: ≥ [1 + + (1 − )] + (1 − )(1 − ) + (21) • Case 3b: All skilled women with a high-cost kid and a level of ability such that ≤ + (1 − )(1 − ) (22) find it convenient not to enter the labor market. Their investment in education depends on the following comparison: ¤ £ ¤ £ − (1 − ) + (1 − ) − (1 − ) £ ¤ £ ¤ ≥ (1 + ) + − (1 − ) + (1 − ) − (1 − ) (23) which delivers the following threshold of ability: ≥ [1 + + (1 − )] (1 − ) This last case is particularly interesting: women whose abilities are such that ≤ + (1−)(1−) (24) [1++(1−)] (1−) ≤ have invested in education but do not find it convenient to work. This is again a situation where human capital is unused. 14 Remark 1 The higher the share of child care time under her responsibility (1 − ), the higher the level of ability such that a woman find it convenient to invest in education. Moreover, the higher the value of home production (), the higher the level of ability to make the educational choice a convenient one. Proof: When is smaller, the threshold levels of ability related to the education decision at Eq. 14, Eq. 17 and Eq.21 are smaller. Moreover 1 − (1 − ) 0 for any guarantees that the same threshold levels are increasing in . The remark clarifies the role of the division of child care duties within the couple, captured by the parameter . A more balanced division (a lower ) helps not only female participation to the labor market (see Eq.8 and Eq. 10), but also female investments in education, with respect to the case where a woman is the main responsible of the kid’s care. Thus, the division of child-care duties within the couple is crucial to understand the level of female education and employment. Similarly, a female comparative advantage in home production (higher ) reduces female education and employment. 3 The role of imperfect information Imperfect information plays a crucial role in the results described. It is intuitive to understand that in our set-up, if there is perfect information on the ability type and kid’s cost at the stage of the education decision, all those who invest in education will certainly work and there is no unused human capital. Consider a woman whose kid costs with = If, drawing upon Eq. 10, ≥ + (1−)(1−), then participating is always preferable to staying at home. In this case, all women with abilities such that ≥ (2+) will invest in education and the others will remain unskilled (irrespective of the ). When + (1 − )(1 − ), the unskilled will not find it convenient to work. As to skilled women, they will find it convenient to participate only if their ability level is such that ≥ + (1−)(1−) . If we turn to the education decision, we find that only those whose ability satisfies ≥ (1−)(1−) (1+)+ + will decide to invest in human capital. Given that the education threshold is 15 higher than the participation one, all women whose ability is (1+)+ (1−)(1−) + are unskilled and will stop participating at the child’s birth. A comparison between the perfect and imperfect information framework delivers interesting results. Having the perfect information context as a benchmark, we can study the implications of the presence of imperfect information for the education and participation female decisions and, in the aggregate, for the level of human capital and output of the economy. On the one side, in the perfect information scenario there is no unused human capital as only those whose ability type and kid’s type will be such that they can work as skilled will actually decide to invest in education. On the other side, it is not clear whether the aggregate level of output is larger or smaller than what emerges in a context of imperfect information. In our first scenario (Case 1) imperfect information does not play any role, since all women work in both contexts and the cut-off level of ability such that women find it convenient to invest is the same in the perfect and imperfect information set-up. A similar result is obtained when no women work (Case 2c), and thus there are no incentives to invest in education. In the other scenarios instead, imperfect information plays an important role, as described in the following proposition. Proposition 2 Imperfect information enlarges the set of women’ ability types that find it convenient to invest in education and turn out to have a high-cost kid, while it restricts the one of those who become skilled and turn out to have a low-cost kid. As a consequence, a context of imperfect information may be associated with a different level of output than the perfect information one. The general result stated above has to be proved and studied separately in our scenarios. The implications on aggregate human capital and output are different in each of them. If unskilled women have no incentives to work, while skilled women do (Case 2a), in a context of imperfect information the decision of investing in education is taken under the expectation that the cost of the kid will be on average equal to , rather than the true 16 cost and . Thus, women with ability types such that (1 + ) + + (1 − )(1 − ) (1 + ) + + (1 − )(1 − ) (25) will not find it convenient to invest in education in the presence of imperfect information, while they would find it profitable in the case they knew from the beginning that their kids were of a low-cost type. Similarly, women whose ability types are such that (1 + ) + + (1 − )(1 − ) (1 + ) + + (1 − )(1 − ) (26) find it convenient to invest in education in the presence of imperfect information, while they would not find it convenient it if they knew from the beginning that they have a high-cost kid. Remember that in this case all skilled women work. Thus, the presence of imperfect information is associated with a loss in terms of human capital and output due to women with low-cost kids and a gain produced by individuals with high-cost kids. It is easy to check that the total number of skilled women is the same in case of perfect and imperfect information.6 However, the skill composition is different, since under imperfect information we gain skilled women with higher level of abilities than those that we lose. Imperfect information increases the risk sharing opportunities (see Eckwert and Zilcha, 2009 for a similar implication) and may thus lead to a better allocation of talents: high-talented individuals become skilled and low-talented individuals remain unskilled, irrespectively of the cost of the kid. This implies that imperfect information may increase the level of human capital and output of the economy. If skilled women work only if they have a low-cost kid (Case 2b), the increase of the number of educated women with high-cost kids in a context of imperfect information does not traslate into an increase of output. Thus imperfect information creates unused human capital. Moreover, by reducing the number of skilled women with low-cost kids who would work in a context of perfect information,7 imperfect information reduces the level of output in the economy. 6 Fewer skilled women with low-cost kids are exactly compensated by more skilled women with high-cost (1+)++(1−)(1−) (1+)++ (1−)(1−) − kids, since (1 − ) (1+)++ (1−)(1−) (1+)++(1−)(1−) = − 7 This result derives from the following comparison: in a context of perfect information, women with low- 17 Moving to our third scenario, when the workers are all skilled women and the unskilled with low-cost kids (Case 3a), the incentives to invest in education for women with highcost kids associated with the imperfect information context8 induce a gain of output. The reduction in the share of skilled women with low-cost kids9 implies, instead, a loss of output. To be more precise, the gain is equal to the difference in the threshold of abilities for the educational choice of women who turn out to have high-cost kids corresponding to the case of perfect and imperfect information, i.e. (1 + ) + + (1 − )(1 − ) (1 + ) + (1 − ) + + (1 − )(1 − ) − (27) which can be rewritten as ∙ + (1 − )(1 − ) − (1 − ) ¸ (28) while, similarly, the loss, which is equal to the difference in the threshold of abilities for the educational choice of women who turn out to have low-cost kids corresponding to the case of imperfect and perfect information, can be written as ¸ ∙ + (1 − )(1 − ) − Thus, if 1 2 (29) there is a positive gain of the number of skilled individuals in the case of imperfect information. Remembering also that the larger number of individuals who invest in education thanks to imperfect information are high-ability types, imperfect information increases the level of output in the economy also in this case. (1+)++ (1−)(1−) , while in a context of imperfect + (1−)(1−) . Since the second threshold is larger than the cost kids invest in education if ≥ information this happens if ≥ first one we (1+) (1−) + have less educated women with low-cost kids in a context of imperfect information. . 8 In a context of perfect information women with low-cost kids invest in education if ≥ (1+)++ (1−)(1−) , while (2+−)++ (1−)(1−) It in the context of imperfect information the threshold is: ≥ is easy to check that the second threshold is smaller than the first one, thus for a larger set of abilities women with high-cost kid invest in education in the presence of imperfect information. 9 The threshold of abilities such that women with ability higher than this become educated is larger in case of imperfect information ( (2+) ( ). (2+−)++ (1−)(1−) ) 18 than in the context of perfect information Finally, if women with low-cost kids work, either as skilled or unskilled, while women with high-cost kids never work (case 3b), the increased number of skilled women who turns out to have high-cost kids, due to imperfect information, does not translate into output. Moreover, since the set of women who become skilled and turn out to have a low-cost kid is restricted, although women with a low-cost kid work also as unskilled, they obtain a lower salary. Thus, in this case imperfect information reduces total output in the economy. 4 The role of government: A tax-transfer scheme In this section we study the introduction of a tax-transfer scheme targeted to working women and how it affects the education and participation decision. For a working woman, the total cost of buying child care on the market is reduced from to (1 − ) with ∈ (0 1). The subsidy to working women is financed by a proportional tax on wages paid by all workers, men and women. We assume that the scheme is balanced, that is ⎤ ⎡ R R (∗ ) + ∗ () + (1 − ) ∗ ()+ ⎦ ⎣ (30) R (∗ ) ) + ∗ () + (1 − ) (∗ # " Z Z ∗ ∗ () + (1 − ) () + ( ) + (1 − ) ( ) = ∗ ∗ where ∗ ( = ) is the threshold level of ability which identifies the participation decision for skilled women with kid of cost-type (which takes a different value in each specific scenario, see sections from 2.2.1 to 2.2.3), and ∗ is the threshold level of ability such that men with abilities above this threshold invest in education10 . The budget constraint above is a general formula encompassing all the scenarios identified. Depending on the participation decision arising in each of the different scenarios some of the terms disappear as women who do not work do not pay the tax and do not receive the subsidy. Considering now the participation decision in the presence of the tax-transfer scheme. A skilled woman with ability and with a kid of cost ( = ) works if: (1 − ) − (1 − )(1 − ) ≥ − (1 − ) 10 This threshold is given by ∗ = (2+− ) See 1− below Eq. 35. 19 (31) which delivers the following threshold level of ability: ≥ + (1 − )(1 − − ) (1 − ) (32) Assume that 1 − − 0: as in the case where the tax transfer scheme is absent, it still holds that a high cost kid requires a higher level of ability for a woman to find it convenient to work. Similarly, the unskilled continue working in the second period if the following condition is satisfied: (1 − ) − (1 − )(1 − ) ≥ − (1 − ) (33) i.e. if the wage of the unskilled is larger than the following threshold: ≥ + (1 − )(1 − ) 1− (34) Again, this condition is more easily satisfied for women with a low-cost kid than for those with a high-cost one. As in the case with = = 0, there are three possible scenarios, each of which delivers different thresholds of abilities for the education decision. The payoffs and the threshold level of ability for the education decision in each scenario are summarized in Figure 4. We here comment only some major findings. Similarly to what we have obtained in Section 2, in the first scenario (Case 1) all women work and their education decision depends on the wage premium and on the tax rate, besides the interest rate. More precisely, all women with ability above the following threshold invest in education: ≥ (2 + − ) 1− (35) Comparing ( 35) and (12) we find that when all women work, the presence of the taxtransfer scheme increases the threshold level of ability associated with the education decision, i.e. it reduces the set of ability types that become skilled. Notice also that in this case a marginal increase of the tax rate reduces the share of skilled women. This is because incentives to education goes hand in hand with incentives to work. As in this case all women are already working, the distortionary effects of the tax dominate and a 20 further use of this policy has no efficacy, since there is no need to use a higher tax rate to increase female participation. In the second scenario, the results that we have obtained when = = 0 are confirmed. In particular, in case 2b unused human capital emerges due to the presence of imperfect information: women who have invested in education under the imperfect information about the cost of their kid may find it convenient not to work if this cost turns out to be high. This happens also in the third scenario, case 3b. What is the role of this policy? First, we show the implications of this policy in terms of female employment and education. In particular, what are the effects of this policy on the presence of unused human capital and on the overall level of output? The following propositions summarize the results. Proposition 3 If (1 − ) [ (1 − ) − ] + 0 and (1 − ) − 0, the tax-transfer scheme makes it more likely that a scenario where women always work independently of their ability and kid’s cost type is realized. Proof: See the appendix The result stated above indicates that the economy is more likely to replicate the scenario described in Case 1 when a tax-transfer scheme is present. The proposition confirms that we have identified a policy which increases female participation, i.e. it increases the likelihood that a woman always works. Moreover, it is intuitive and easy to check that, when all women have the expectation that they will always work irrespective of the idiosyncratic ability and cost type, incentives to invest in education are maximised, that is, the threshold level of ability such that women find it convenient to invest in education is the lowest. As a consequence, this policy also delivers the highest number of skilled women, thus having positive repercussions on the level of human capital in the economy. The total amount of output produced in the economy when all women work is also maximum. Propositions 2 suggests that the tax-transfer scheme makes the realization of the first scenario more likely. The following corollary emphasizes one of the consequences of this result, i.e. that the policy reduces the presence of unused human capital (Case 2b and 21 3b). Corollary 4 The tax-transfer scheme may reduce the emergence of unused human capital. Proof: See appendix Focusing on cases 2b and 3b where unused human capital emerges in the presence of imperfect information, the proposition suggests that while imperfect information cannot be completely eliminated, its effect of generating unused human capital can be attenuated by the government intervention. The tax-transfer scheme can be interpreted as an insurance device against the risk of having high-cost kids. In the presence of the government policy, the difference between having a high-cost kid and a low-cost kid is reduced and so is the possibility that women who have invested in education do not work after the child birth. We may also argue that the absence of such a policy can be interpreted as the real responsible for non participation. Notice however that this potential gain in the output of skilled women with a high-cost kid associated to the policy does not necessarily imply that the total amount of output is larger when the government intervenes. This is because the tax-transfer scheme may enlarge the loss of output due to a reduction in the number of skilled working women with a low-cost kid associated with the imperfect information with respect to the perfect information benchmark11 . In other words, while for skilled women with high-cost kid the policy works as an insurance, for skilled women with a lowcost kid it magnifies the impact of imperfect information as a deviation from the perfect information benchmark. The next proposition summarizes the role of the policy in the other cases. 11 In case 2b this loss is given by the difference in the threshold level of ability characterizing the education decision of women with a low-cost kid in the case of imperfect and perfect information: (1+) + (1−)(1−) − (1−) + (1+) (1+) = 1− + + (1−)(1−) , which, in presence of the tax transfer scheme enlarges to (1+) (1− ) 1− The same emerges in case3b, where the difference has to be interpreted as due to an increase of the number of women with a low-cost kid working as unskilled rather than as skilled in case of imperfect information. 22 Proposition 5 When imperfect information does not create unused human capital, if (1 − ) [ (1 − ) − ] + 0 and (1 − ) the tax-transfer scheme reduces the consequences of the imperfect information, by inducing an equilibrium closer to the perfect information benchmark. Proof: See appendix. The proposition shows that in case 2a and 3a the role of government is to bring the imperfect information equilibrium closer to the perfect information one. Again, the consequences on the level output are not unambiguous. The tax-transfer scheme, by attenuating the impact of imperfect information, may limit the benefits of imperfect information, in terms of output, arising for instance when skilled women work and unskilled do not work (case 2a), while it may also limit the loss of output arising when women with a high-cost kid do not work and those with a low-cost kid work (case 3a). Finally, we may ask whether this policy is welfare improving. Given that in our model utility is defined by the consumption possibilities, it is immediate to show that the taxtransfer scheme may increase the utility of the women who work and of the fathers of their kids. The utility of women who do not work is not affected by the presence of the tax scheme, while the utility of men who have a kid with a non working woman decreases. Thus, the scheme may increase the overall welfare if the gains obtained by the working women and the fathers of their kid is large enough to overcome the loss of the men who have a kid with non working women. 5 Concluding Remarks In a context of imperfect information on the time required to provide child care to their kid, women make education decisions. Imperfect information may induce a waste of talent since women who have invested in education but turn out to have high-cost kids may find it convenient not to work. However, imperfect information may also be associated with a higher level of output if, in the presence of uncertainty, high ability working women who turn out to have high-cost kids are induced to acquire education, an investment they would not have performed had they known from start the high cost of their kid. 23 We have studied the introduction of a tax-transfer policy and shown that its presence makes it more likely that women work irrespective of their idiosyncratic ability and cost type, strengthening the incentives to invest in education. In particular, the government intervention, by reducing the costs of buying child care on the market, diminishes the chances of observing a waste of talent of skilled women who turn out to have a high-cost kid and it induces them to work and realize on the job the return of their educational investment. Our model can be extended in several directions. First, in a political economy context, we may ask whether the policy that we have introduced is politically feasible. In an overlapping generation framework, individuals, men and women, live three periods: in the first they are young and they do not take any decision; in the second they make their education investment or work as unskilled; at the beginning of the third period they have a kid and they make their participation decision. Each time two persons are alive, a man and a woman and two kids are generated, so that the population growth rate is zero. The education decision is done under imperfect information and under expectations on the level of taxation which will be in place in the next period. We will look at fixed point equilibrium such that the expected and the actual level of tax are equal. Two candidates for the equilibrium are: a zero tax rate and a positive tax rate. If individuals expect the tax rate to be zero, it is more likely that women will not work, thus decreasing their incentive to invest in education. Non-working women and the father of their kids will obviously vote for a zero level of the tax rate, thus reinforcing their expectation. On the contrary, if individuals expect the tax-transfer scheme to be in place, women are induced to work and to invest in education, and they will vote for a positive tax rate, together with the father of their kids. Second, as suggested by the figures shown in the introduction, cross-country differences in child-care policies may be associated with differences in female educational and employment outcomes. An empirical analysis founded on microdata would be particularly useful to assess the relationship between the presence of institutions and the female education and participation decisions. Finally, in terms of policy implications, our analysis suggests that promoting an institu24 tional environment in favor of women education and employment may generate economic benefits. The Lisbon strategy which identifies target values for European countries on female employment and on the institutional measures which may favor it (child care services, parental leaves etc.) goes in this direction. In particular, we have stressed that incentives to employment may also reinforce - through their effect on expectations - educational investments. Focusing on them may represent a good strategy to increase the human capital and output of any country, a priority especially in countries where human capital is scarse. 6 Appendix 6.1 Proof of proposition 2 We start from case 2c where a woman never works. If she is skilled, she does not work for ≤ + (1−)(1−) or, in presence of the tax scheme, ≤ + (1−)(1−−) . (1− ) If (1 − ) [ (1 − ) − ] + 0, it is (1 − ) [ (1 − ) − ] + 0 which guarantees that + (1−)(1−−) (1− ) + (1−)(1−) , i.e. the tax-transfer scheme makes the realization of this case more difficult. In particular, it increases the likelihood that a skilled woman with a low-cost kid works. This corresponds to case 2b. Now, in case 2b a woman invests in education if ≥ (1+) (1−) + + (1−)(1−) + (1−)(1−) (1+) (1−) (1−)(1−) + + . Since a skilled woman with a low-cost kid always works. This is not necessarily true for a skilled woman with a high-cost kid. In fact, if ¡ ¢ (1+) + + (1−)(1−) + (1−)(1−) , i.e. (1+) − (1−)(1−)(1−), (1−) a skilled woman with a high cost kid does not work. It is easy to find that, when the tax¢ ¡ transfer scheme is introduced, this condition becomes: (1 + ) − (1 − )(1 − − )(1 − ), which is more difficult to be satisfied for a given level of . Thus, the presence of the tax-transfer scheme increases the likelihood that a skilled woman even with a high-cost kid works. This corresponds to case 2a. In case 2a an unskilled woman never works, since + (1 − )(1 − ). In presence of the tax-transfer scheme, this condition becomes + (1−)(1−−) 1− which, for (1 − ) [ (1 − ) − ] + 0 is more difficult to be satisfied than the first one. 25 Thus, the policy increases the likelihood that even an unskilled woman works. When a woman always works we are in case 1. Now we start from case 3b, where a woman works only if she has a low-cost kid. She invests in education if ≥ [1++(1−)] . (1−) If [1 + + (1 − )]−(1−)− (1−)(1− )(1 − ) 0 a skilled woman who turns out to have a high-cost kid does not work. In presence of the tax scheme, this condition becomes [1 + + (1 − )]− (1−) −(1− )− (1−)(1−)(1−)+ (1−)(1−) 0If (1−)− 0 the tax-transfer scheme makes it more difficult the realization of this case, by increasing the likelihood that even a skilled woman with high-cost kid works. This corresponds to case 3a. In case 3a a woman does not work only if she is unskilled with a high-cost kid: + (1−)(1−) In presence of the tax scheme she does not work if (1 − ) [ (1 − ) − ] + 0 it is + (1−)(1−−) 1− + (1−)(1−−) . 1− Since for + (1 − )(1 − ) the policy makes it more likely that even an unskilled woman with a high-cost kid works, i.e. increases the possibility that a woman always works (case 1). 6.2 Proof of corollary 3 Unused human capital emerges in case 2b and 3b. Starting from case 2b, we may have a non working skilled woman with a high-cost kid if the threshold level of ability which defines the education decision is lower than ¡ ¢ the one that defines the participation decision, i.e. that (1 + ) − (1 − )(1 − − )(1 − ). In presence of the tax transfer scheme this condition becomes ¡ ¢ (1 + ) − (1 − )(1 − − )(1 − ), which is more difficult to be satisfied for a given level of . Similarly, in case 3b the same condition reads [1 + + (1 − )] − (1−)− (1−)(1−)(1−) 0, which, in presence of the tax-transfer scheme becomes [1 + + (1 − )]− (1−) −(1−)− (1−)(1−)(1−)+ (1−)(1−) 0, which, again, is more difficult to be satisfied. 6.3 Proof of proposition 4 Case 1 and 2c are not interesting, since imperfect information has no consequences. 26 Let first analyze case 2a. Imperfect information is associated with a loss in terms of human capital and output due to individuals with low-cost kids that do not invest in education and a gain related to individuals with high-cost kids who would indeed invest. If (1 − ), both this loss and the gain are smaller in presence of the tax scheme. In fact, the loss related to women with a low-cost kid is equal to the difference in the threshold levels of abilities that define their education decision in case of imperfect and (1−)(1−)(− ) − (1+)++(1−)(1−) = perfect information: (1+)++(1−)(1−) (1−)(1−−)(− ) which, in presence of the tax scheme, becomes i.e., if (1 − ) (1− ) it is smaller. Similarly, the gain of imperfect information related to skilled women with (1−)(1−)( −) which, in presence of the tax scheme and a high-cost kid depends on (1−)(1−−)( −) for (1 − ) reduces to The loss related to women with a low (1− ) cost kid is equal to the difference in the threshold levels of abilities that define their − education decision in case of imperfect and perfect information: (1+)++(1−)(1−) (1−)(1−)(− ) (1+)++(1−)(1−) = , which, in presence of the tax scheme becomes (1−)(1−−)(− ) i.e., if (1 − ) it is reduced. Similarly, the gain of imperfect (1− ) (1−)(1−)( −) information related to skilled women with a high-cost kid depends on (1−)(1−−)( −) which, in presence of the tax scheme and for (1 − ) reduces to (1− ) Moving to case 3a where unskilled women with a low-cost kid are the only ones who do not work, the presence of imperfect information generates a gain with respect to the perfect information benchmark in terms of more working skilled women with a high-cost kid and a loss in terms of less working women with a low-cost kid. If (1 − ) and (1 − ) [ (1 − ) − ] + 0 both the gain and the loss are reduced by the presence of the tax-transfer scheme. In fact the loss related to women with a low-cost kid is equal to the difference in the threshold levels of abilities that define their education decision − (2+) = in case of imperfect and perfect information: (2+−)++(1−)(1−) [+(1−)(1−) − ] , which, in presence of the tax scheme, if (1 − ) [ (1 − ) − ] + (1− ) [+(1−)(1−−) − ] Similarly, the gain of imperfect information 0 reduces to (1−)[+ (1−)(1−)− ] related to skilled women with a high-cost kid depends on which, (1−)[+ (1−)(1−−)− ] in presence of the tax scheme and for (1−) reduces to Again, (1− ) 27 the tax transfer scheme leads to an equilibrium closer to the perfect information benchmark. References [1] Albanesi S. and Olivetti C. (2006). Home production, market production and the gender wage gap: incentives and expectation. NBER W.P. 12212. [2] Bernal R. (2008). The Effect of Maternal Employment and Child care on Childrens’ Cognitive Development. International Economic Review, forthcoming. [3] Bjerk D. and Han S. (2007). Assortative Marriage and the Effects of Government Homecare Subsidy Programs on Gender Wage and Participation Inequality, Journal of Public Economics, June, 1135-1150. [4] Chiappori P.A., Iyigun M. and Weiss Y., 2008, Investment in Schooling and the Marriage Market, mimeo. [5] Del Boca D. Pasqua S. and Pronzato C. (2007). The Impact of Institutions on Motherhood and Work, WP Child n.06/2007 [6] De Henau J. Meulders D. and O’Dorchai S. (2007). Parents’care and Career. Comparing Parental Leaves Policies, in Social Policies, Labour Markets and Motherhood: A Comparative Analysis of European Countries (Del Boca D. and Wetzels C. eds) Cambridge University Press. [7] Eurostat, www.epp.eurostat.ec [8] Eckwart, B. and Zilcha, I. (2009) Improvement in information and private investment in education, mimeo. [9] Francois P.(1998) Gender Discrimation without Gender Difference: Theory and Policy Responses, Journal of Public Economics Vol.68, 1-32. [10] Iyigun M. and Walsh R., (2007a), Endogenous Gender Power, Household Labor Supply and the Demographic Transition, Journal of Development Economics, 87, 138-155. 28 [11] Iyigun M. and Walsh R., (2007b), Building the Family Nest: Premarital Investments, Marriage Markets, and Spousal Allocations, Review of Economic Studies, 74, 507-535. [12] Lommerud K. and Vagstad E. (2007). Mommy tracks and public policy: On selffulfilling prophecies and gender gaps in promotion, mimeo. [13] OECD (2006). Education at a Glance. [14] Ohlsson H. and Lundholm M. (1998). Wages, taxes and publicly provided day care, Journal of Population Economics, Springer, vol. 11(2), pages 185-204. [15] Rosen S. (1997). "Public Employment, Taxes, and the Welfare State in Sweden," NBER Chapters, in: The Welfare State in Transition: Reforming the Swedish Model, pages 79-108 National Bureau of Economic Research. 29 Figure 3 Case 1 : wuF > η + ρ H (1 – γ) (1 – η ) work ρH π no work work Skilled 1- π ρL no work work Unskilled π work Educated if αi wFs - ρL ( 1 – γ) η – ρL ( 1 – γ) η wFu– ρH ( 1 – γ) η - ρH ( 1 – γ) η wFu – ρL ( 1 – γ) ρL no work 1( ) η - ρH ( 1 – γ) η ρH no work 1- π αi wFs – ρH ( 1 – γ) η – ρL ( 1 – γ) η αi ≥ wFu (2 + r ) wFs 2a ( ) Educated if αi ≥ wFu (1+ r)+ η+(1-γ)(1 - η) ρL wFs Case 2 : wuF < η + ρ L (1 – γ) (1 – η ) Work if work ρH π no work αi wFs – ρH ( 1 – γ) αi ≥ η+(1-γ) (1 - η) ρH wFs η - ρH ( 1 – γ) η 2b ( ) work Skilled 1- π ρL no work work Unskilled π Educated if η – ρL ( 1 – γ) η αi ≥ __wFu (1 + r)____ wFs (1 – π) wFu – ρH ( 1 – γ) ρH + Work if no work work 1- π αi wFs - ρL ( 1 – γ) η - ρH ( 1 – γ) η no work L η – ρ ( 1 – γ) η η + ρL (1- γ) (1 - η) wFs η+(1-γ) (1 - η) ρL < αi < η+(1-γ)(1 - η) ρH wFs wF s wFu – ρL ( 1 – γ) ρL + 2c ( ) Nobody Educated Don’t work if αi < η+(1-γ)(1- η) ρL wFs Case 3 : η + ρ L (1 – γ) (1– η ) < wuF < η + ρ H (1 – γ) (1 – η ) work no work work Skilled 1- π ρ work Unskilled η - ρH ( 1 – γ) η ) αi ≥ wFu (1+ r +( 1 – π)) wFs αi wFs - ρL ( 1 – γ) η – ρL ( 1 – γ) η wFu – ρH ( 1 – γ) Work if αi ≥ ρH no work + π η + π (1-γ) (1 - η) ρH wFs L no work π 3a ( Educated if ρH π αi wFs – ρH ( 1 – γ) η+(1-γ)(1 - η) ρH wFs 3b( ) η - ρH ( 1 – γ) η Educated if work 1- π wFu – ρL ( 1 – γ) ρL no work η – ρL ( 1 – γ) η αi ≥ __wFu (1 + r +( 1 – π)) wFs ( 1 – π) Don’t work if αi < η+(1-γ)(1 - η) ρH wF s Figure 4 Case 1 : wuF > η + ρ H (1 – γ) (1 – φ – η ) (1 – τ) work ρH π no work work Skilled 1- π ρL no work work Unskilled π work Educated if αi wFs (1- τ) - ρL ( 1 – φ) ( 1 – γ) η – ρL ( 1 – γ) η wFu (1- τ) – ρH ( 1 – φ) ( 1 – γ) η - ρH ( 1 – γ) η wFu (1- τ) – ρL ( 1 – φ) ( 1 – γ) ρL no work 1( ) η - ρH ( 1 – γ) η ρH no work 1- π αi wFs (1- τ) – ρH ( 1 – φ) ( 1 – γ) η – ρL ( 1 – γ) η αi ≥ wFu 2 + r - τ wFs 1-τ 2a ( ) Educated if αi ≥ wFu (1+ r)+ η+(1-γ)(1 - φ - η) ρL wFs (1 – τ) Case 2 : wuF < η + ρ L (1 – γ) (1 – φ – η ) (1 – τ) Work if work ρH π no work work Skilled 1- π ρL no work work αi wFs (1- τ) – ρH ( 1 – φ) ( 1 – γ) αi ≥ η+(1-γ)(1 - φ - η) ρH wFs (1 – τ) η - ρH ( 1 – γ) η i F L α w s (1- τ) - ρ ( 1 – φ) ( 1 – γ) η – ρL ( 1 – γ) η wFu (1- τ) – ρH ( 1 – φ) ( 1 – γ) 2b ( ) Educated if αi ≥ __wFu (1 + r)____ wFs (1 – τ)(1 – π) + Work if + η + ρL (1- γ) (1 - φ - η) wFs (1 – τ) H Unskilled π ρ no work work 1- π H η - ρ ( 1 – γ) η wFu (1- τ) – ρL ( 1 – φ) ( 1 – γ) ρL no work η+(1-γ)(1-φ -η) ρL < αi < η+(1-γ)(1-φ- η) ρH wFs (1 – τ) wFs (1 – τ) L η – ρ ( 1 – γ) η 2c ( ) Nobody Educated Don’t work if αi < η+(1-γ)(1-φ -η) ρL wFs (1 – τ) Case 3 : η + ρ L (1 – γ) (1 – φ – η ) < wuF < η + ρ H (1 – γ) (1 – φ – η ) (1 – τ) (1 – τ) work no work work Skilled 1- π ρL no work work Unskilled π 3a ( ) Educated if ρH π αi wFs (1- τ) – ρH ( 1 – φ) ( 1 – γ) η - ρH ( 1 – γ) η αi ≥ wFu (1+ r +( 1 – π) (1 – τ )) wFs (1 – τ ) αi wFs (1- τ) - ρL ( 1 – φ) ( 1 – γ) L η – ρ ( 1 – γ) η wFu (1- τ) – ρH ( 1 – φ) ( 1 – γ) + π η + π (1-γ) (1 - φ - η) ρH Work if αi ≥ ρH no work wFs (1 – τ ) η+(1-γ)(1 - φ - η) ρH wFs (1 – τ) 3b( ) η - ρH ( 1 – γ) η Educated if work 1- π wFu (1- τ) – ρL ( 1 – φ) ( 1 – γ) ρL no work η – ρL ( 1 – γ) η αi ≥ __wFu (1 + r +( 1 – π)(1 – τ)) wFs (1 – τ) ( 1 – π) Don’t work if αi < η+(1-γ)(1 - φ - η) ρH wFs (1 – τ)
© Copyright 2026 Paperzz