CHILDCARE ACCESS AND EMPLOYMENT: THE CASE OF WOMEN WITH PRESCHOOL-AGED CHILDREN IN TOKYO

Authors


Abstract

This article presents an empirical study on childcare accessibility and the importance of access to childcare in attaining preferred employment among women with preschool-aged children in Tokyo. The age-wise childcare accessibility of this study takes into account spatial variations in the supply and demand of childcare, as well as “spatial competition,” based on spatially micro areas — blocks. The accessibility reveals a considerable geographic mismatch between childcare center supply and demand, particularly for children aged up to two years. Empirical results show that access to childcare is closely associated with a higher probability of attaining preferred employment among women with preschool-aged children. The association is remarkably strong when a woman has a very young child aged up to two years and when the childcare center is one that is desired. Adequate childcare provision, particularly for children under the age of three, helps to augment active female participation in the labor market.

1 Introduction

With the increasing participation of women in the labor market, childcare has become an important policy issue in many countries (Organisation for Economic Co-operation and Development 2008). Families attempting to reconcile their work and family life face numerous challenges and one of these is to find appropriate childcare for their children while they work. If adequate childcare is unavailable, many parents may abandon the employment they prefer. Other parents may have no choice but to use undesirable (e.g. distant and low-quality) childcare out of necessity, which can hamper not only the work-life balance, but also children's well being and development. The lack of adequate childcare is reported to be a problem in various areas around the world (Blau 2001; Apps & Rees 2005; Del Boca & Vuri 2007). This problem is currently serious in Japan, as there are a large number of children on waiting lists for childcare centers.

The lack of access to childcare facilities may arise not only because of a supply shortage (excess demand), but also because of a geographical mismatch between supply and demand for childcare. The location of childcare facilities is important for many families who rely on childcare to reconcile work and family life (Tivers 1988). Indeed, proximity to home is often an important consideration when selecting a childcare facility (Hanson & Pratt 1990; Kawabata 2010a). Even when a place in a childcare center is available, that opening may remain inaccessible unless the center is located within a reasonable distance from where the family lives. However, the geographic mismatch of childcare arising from the consideration of a reasonable distance from home has not been sufficiently examined, perhaps as a result of data limitations. Childcare accessibility, often termed childcare availability, has been examined for relatively large geographic areas, such as counties and regions (Kreyenfeld & Hank 2000; Gordon & Chase-Lansdale 2001).2 Webster and White (1997) use a more detailed geographic area—a circular area with a radius of 800 meters — when measuring accessibility of childminders. However, limited research addresses the accessibility of childcare centers, taking into account not only the supply, but also the demand that spatially competes for the supply (i.e. spatial competition). Moreover, very few studies differentiate between children's ages in measuring childcare accessibility, although childcare accessibility may differ substantially by age.

In most European Union countries, among couples with children under the age of six, more women express a desire to participate in the labor force than actually do (Jaumotte 2003). A national survey in Japan shows that among women with preschool children aged four or five, the majority (72%) wish to work, but less than half of that figure (37.4%) actually work (Gender Equality Bureau Cabinet Office of Japan 2007). The large disparities between desired and actual employment patterns indicate substantial potential for the promotion of female labor force participation. A large body of literature suggests that greater availability of childcare increases female participation in the labor market (Stolzenberg & Waite 1984; Webster & White 1997; Gordon & Chase-Lansdale 2001; van Ham & Büchel 2006; Herbst & Barnow 2008). Thus far, however, the extent to which access to childcare is important in the attainment of preferred employment is an empirical question that has not been fully explored.

The objective of this research is to shed new light on these under-researched aspects of childcare access and work-life balance. Specifically, this study addresses the following two questions. First, does a geographic mismatch exist between the supply of and demand for childcare centers, and does this differ according to the age of children? Second, is access to childcare centers important in attaining preferred employment among women with preschool-aged children?

To answer the first question, this study calculates and visualizes the accessibility of childcare centers that takes into account the geographic mismatch in their supply and demand, children's ages, and spatial competition, which have rarely been simultaneously addressed. The accessibility is calculated at the block level — a micro area — using detailed spatial data and a geographic information system (GIS). To address the second question, probit models are estimated using a unique survey dataset on balancing work and child rearing and accessibility to childcare centers. This study further examines whether the importance of access to childcare differs according to the following three variables: status of preferred employment (full-time vs. part-time), age of youngest child (younger vs. older preschool child), and type of childcare (desired vs. other-than-desired). The study area is Tokyo's 23 wards, where increasing the number of childcare centers is a pressing policy issue.

The remainder of the paper is organized in the following manner. Section 'Rise in Childcare Demand and the Childcare Policy in Tokyo' describes the rise in childcare demand and the childcare policy in Tokyo. Section 'Method' describes the methodology, and Section 'Empirical Results' presents the results. Finally, Section 'Conclusions' discusses the findings of the study and policy implications.

2 Rise in Childcare Demand and the Childcare Policy in Tokyo

Increasing the number and capacity of childcare centers is a pressing policy issue in Japan. For over a decade, the Japanese Government has endeavored to increase childcare services to eliminate waiting lists at childcare centers throughout the country. In 2001, the Cabinet endorsed the “Zero Wait Listed Children Strategy,” and in 2008, the government implemented the “New Zero Wait Listed Children Strategy.” Nonetheless, the rise in demand has far outpaced the increase in supply; during the 2009 fiscal year, childcare waiting lists in Japan grew by 891 children, even though the total supply of childcare slots increased by approximately 26,000 (Ministry of Health, Labour and Welfare of Japan 2010b).

Childcare waiting lists are an urban problem in Japan. As of April 2010, most waiting lists were found in urban areas, with Tokyo alone accounting for 32% of the total number of children on childcare waiting lists nationwide (Ministry of Health, Labour and Welfare of Japan 2010b). Between 2005 and 2010, Tokyo witnessed a dramatic increase in licensed daycare waiting lists (Fig. 1) (data obtained from the Bureau of Social Welfare and Public Health of the Tokyo Metropolitan Government [TMG]). The total number of children on waiting lists rose by 62%; in particular, the period 2008–2009 showed a remarkable increase. The rise in demand among parents of younger children in the age group of 0–2 years is noteworthy. Between 2005 and 2010, the proportion of 0–2 year olds on waiting lists rose from 76% to 91%. The one-year-old group was predominant, accounting for almost half the total number on waiting lists in 2010.

Figure 1.

The number of children on waiting lists for licensed daycare centers in Tokyo, 2005–2010.

Note that the potential demand not reflected by publicly released data (Fig. 1) is likely to be enormous, as suggested by Zhou and Oishi (2005) and Kawabata (2012). In order to resolve the problem of childcare waiting lists, the TMG is aiming to increase childcare capacity by 22,000 spots over three years from 2010 onward (TMG Bureau of Social Welfare and Public Health 2011).

3 Method

The study area comprised Tokyo's 23 wards, which encompass the central and densely populated area of Tokyo Metropolis, a metropolitan prefecture in Japan (Fig. 2). The study area covers 622 square kilometers with a population of 8.5 million people in 2009. In the remainder of this paper, this area will be referred to as the “Tokyo ward area,” while “Tokyo” will be used to refer to the entire Tokyo Metropolis.

Figure 2.

Tokyo metropolis (excluding islands) and the study area.

This study employs a two-part methodology. The first part examines the geographic mismatch between childcare center supply and demand, and the second part analyzes the importance of access to childcare centers in the attainment of preferred employment among women with preschool-aged children. Each part is described below.

3.1 Geographic Mismatch of Childcare Centers: Accessibility and Data

In order to examine the geographic mismatch between the supply and demand for childcare centers, accessibility is calculated for each basic unit block (kihontaniku). The basic unit block is the smallest geographic unit for which census population data is available. There are 23 wards and 115,501 basic unit blocks within the study area.3 Accessibility is measured in various ways depending on its purpose (Handy & Niemeier 1997). In order to meet the goals of this study, I use the accessibility measurement of Kawabata (2010b), which can indicate not only the mismatch between the demand and supply of childcare, but also their geographic mismatch, with the following measurement:

display math(1)

where a is the age of children; Ai is the accessibility for a resident zone i; Sj is the supply (capacity) of a childcare center j; dij, and dkj are the respective distances by road between resident zones i and k, on the one hand, and a childcare center j, on the other; d0 is the threshold distance for commuting to childcare centers; r is the ratio of those requiring childcare centers to the entire population; and Pk is the population in a resident zone k.

An important aspect is that the accessibility measurement takes into account spatial competition, as described in Harris (2001); the proposed measurement incorporates not only the spatially accessible supply, but also the demand that is spatially accessible to the supply (spatial competition). This accessibility measurement is similar to the two-step floating catchment area method employed by Luo and Wang (2003), among others, in that the measurement combines the following two steps: (i) at each location of a childcare center, the ratio of supply (Sj) to its surrounding demand (within a threshold distance to the childcare center), math formula, is calculated; and (ii) at each demand (residential) location, the ratios (derived in the first step, or math formula) surrounding (within the same threshold distance from) the demand location are summed. Therefore, in essence, an accessibility value represents the supply–demand ratio; accordingly, an accessibility value of one represents a supply–demand balance, whereas a value of greater or less than one indicates excess supply or demand. An accessibility measurement that allows such interpretation is intelligible and useful for policy making. The population-weighted average of accessibility for the entire area equals the supply–demand ratio for the entire area. It must be noted that not taking into account the supply and demand outside the study area can distort the supply–demand ratio, particularly around the border of the study area; however, this is unlikely to be a major problem in this study, because most childcare centers prioritize applicants who are living in the same ward or in Tokyo. In this study, accessibility from home, not from the workplace is addressed; this is because in the Tokyo ward area, a majority of working parents commute on congested trains and accessibility from home, rather than from the workplace, tends to be of greater importance in the selection of childcare centers (Kawabata 2010a).

There are various types of childcare services in Tokyo. This analysis focuses on three types of childcare centers: licensed daycare centers (ninka-hoikujo), TMG-certified daycare centers (Tokyo's ninsyo-hoikujo), and authorized childcare centers (nintei-kodomoen). These three types of centers were selected because their quality and affordability are ensured at certain levels by national or TMG standards,4 they are the major providers of childcare services in Tokyo, and relevant data on other non-licensed childcare centers were unavailable. Quality and cost are often the primary concerns in countries where childcare services are provided by the private sector, such as in the US (Blau 2001), whereas availability seems to be more of an issue in countries where most childcare services are publicly subsidized, such as Italy, Germany, Japan, and the Netherlands (Kreyenfeld & Hank 2000; Del Boca & Vuri 2007; Suzuki 2010). Data on the three types of childcare centers categorized by age were obtained from the Bureau of Social Welfare and Public Health of the TMG, ward offices, and individual childcare centers. The Statistical Bureau of Japan provided spatial data on basic unit blocks in the 2005 census. For the supply of childcare centers (Sj), I used the capacity of the three types of childcare centers as of April 2009. Capacity tends to be smaller for younger children. The total capacity for children aged 0 is 10,158 spots, aged one is 18,592 spots, aged two is 21,876 spots, aged three is 22,954 spots, and aged four and above is 45,885 spots. The distance by road between basic unit blocks and childcare centers (dij and dkj) is calculated using the 2009 road network data and GIS. Here, locations of basic unit blocks are the centroids of the blocks, and locations of childcare centers are the spatial points created using the detailed address matching service of the Center for Spatial Information Science at the University of Tokyo.

The ratio of those requiring the three types of childcare centers to the population (r) is set at 20% for infants aged 0 and 35% for children aged one and older.5 Here, the constant ratios for the Tokyo ward area are used, because the constant ratios make interpretation and comparison straightforward. Also, the Tokyo ward area is the central area of Tokyo and has relatively common central-city characteristics. The population by basic unit block (Pk) for each age was estimated using 2009 population data by age for each subdivision (chocho-aza)6 of a city. The subdivision, which is larger than a basic unit block, but smaller than a ward, is the smallest geographic unit for which population data by age was available for the year 2009. The data for subdivisions were disaggregated into the data for basic unit blocks on the basis of proportional distributions of the 2005 census population at the basic-unit-block level, which were provided by the Statistics Bureau of Japan.

For the threshold distance to childcare centers (d0), I used 500, 750, and 1,000 meters, given that over 90% of travel times to childcare facilities in the Tokyo ward area are approximately 15 minutes or less (Kawabata 2010a). Assuming a walking speed of approximately 50 meters per minute with young children in tow (Segawa & Sadahiro 1996), the three time thresholds are approximately 10, 15, and 20 minutes on foot for 500, 750, and 1,000 meters, respectively.7

Thereafter, the calculated accessibility measurements were visualized using GIS to examine the extent to which areas have a geographic mismatch. The visualization is effective for identifying the geographic mismatch, particularly when the study area comprises a large number of locations (blocks).

3.2 Childcare Access and Attaining Preferred Employment: The Model and Data

The importance of access to childcare centers in attaining preferred employment by women with preschool-aged children is examined by estimating probit models. With regard to preferred employment, this study addresses two employment statuses — full-time and part-time employment. It is assumed that preferred employment is attained if a woman who prefers to work full-time is working full-time or if a woman who prefers to work part-time is working part-time.8 The sample is from a unique Internet-based questionnaire survey on balancing work and child rearing and accessibility to child-care centers, which was conducted over the period 20–25 November 2009. The survey data include answers from 311 respondents among 650 women with preschool-aged children living in the Tokyo ward area. The respondents are widely distributed across the 23 wards and their basic demographic statistics are not very different from the 2005 census data.9 The survey data indicate a considerable discrepancy between preferred and current employment status. Of the 311 women, 263 (85%) prefer to work and are the population of interest, while less than half of them (113, 36%) are actually working. Among those who prefer to work, part-time employment is preferred to full-time employment. The models in this paper use 261 women who prefer to work and answered all the relevant question items. Of the 261 women, 71 (27%) indicated that they prefer full-time employment, while the remainder, 190 (73%), indicated that they prefer part-time employment.

The probit models take the following form:

display math(2)

where y is the binary response variable 1 if preferred employment is attained, and 0 otherwise. X is a set of individual and neighborhood characteristics—age, number of children, presence of a very young child, presence of a full-time husband or partner, work experience, and female unemployment rate, which are likely to be relevant to women's employment outcomes based on labor market literature. Acc comprises the variables on access to a childcare center, and D denotes either a dummy indicating preference for working full-time or a dummy indicating the presence of a child in the age group of 0–2 years.

Table 1 presents variable descriptions and descriptive statistics. The four variables on access to childcare centers are: (i) access to a childcare center (acc1); (ii) access to a desired childcare center (acc2); (iii) access to an other-than-desired childcare center while unable to have access to a desired childcare center (acc3); and (iv) the accessibility of childcare centers at the ward level for the youngest child (acc4), which is calculated from the accessibility values in Equation (1). The variables acc2 and acc3 are used to examine whether the importance of childcare centers differs between desired and other-than-desired childcare centers. The accessibility variable, acc4, is the population weighted-average accessibility at the ward level, larger than the basic unit block level, because the residential information of the sample is limited to the ward level. It is expected that the signs of the four variables on access to childcare centers are positive, indicating that access to childcare centers is positively associated with the attainment of preferred employment by women with preschool-aged children. The interaction terms, Acc  ·  D, are introduced in order to examine two further aspects—whether the importance of access to a childcare center differs by the type of preferred employment (full-time and part-time) and the presence of a very young child in the age group of 0–2 years.

Table 1. Variable descriptions and descriptive statistics
VariableDescriptionMeanSD
  1. SD, standard deviation.
Attained preferred employment1 if preferred employment attained; 0 otherwise0.31 
Age 30–34 years1 if between 30 and 34 years old; 0 otherwise0.35 
Age 35–39 years1 if between 35 and 39 years old; 0 otherwise0.38 
Age ≥40 years1 if 40 years old or older; 0 otherwise0.16 
2 or more children1 if living with 2 or more children; 0 otherwise0.55 
Child age 0–2 years1 if child is aged 0–2 years; 0 otherwise (child aged 3–5 years)0.55 
Full-time partner1 if living with a husband or partner who is working full-time as a regular employee; 0 otherwise0.85 
Work experience before childbirth1 if working up to (less than one year before) birth of the youngest child; 0 otherwise0.56 
Full-time work preference1 if prefers to work full-time; 0 otherwise (prefers to work part-time)0.27 
Access to childcare center (acc1)1 if using childcare center for the youngest child; 0 otherwise0.32 
Access to desired childcare center (acc2)1 if using desired childcare center for the youngest child; 0 otherwise0.25 
Access to other-than-desired childcare center (acc3)1 if using other-than-desired childcare center for the youngest child because unable to use desired childcare center; 0 otherwise0.07 
Accessibility of childcare centers, ward-level, 750 m, value × 100 (acc4)Population-weighted average of accessibility of childcare centers using the 750 meter threshold, for the youngest child, at the ward level (value × 100)97.520.9
Female unemployment rate (ward level, %)Female unemployment rate (%), at the ward level4.960.81
Number of observations 261 

4 Empirical Results

First, I present the results of the geographic mismatch between the childcare center supply and demand. Next, the importance of access to childcare centers in attaining preferred employment for women with preschool-aged children is explained.

4.1 Geographic Mismatch of Childcare Centers

Figure 3a depicts the accessibility of childcare centers by basic unit block for the commuting threshold of 750 meters and one-year-old children, for whom the waiting list is the longest. The map reveals a considerable geographic mismatch in the supply and demand for childcare centers. It is striking that many blocks have accessibility values below one, which indicates excess demand or childcare shortage. When Figure 3a and the corresponding maps for the other age groups of preschool children (not shown) are examined, the extent of blocks with accessibility values below one tends to be larger for children in the age group of 0–2 years than for those in the age group of 3–5 years.

Figure 3.

Spatial distributions of (a) accessibility of childcare centers with commuting threshold of 750 meters, (b) one-year-old children, and (c) childcare centers for one-year-old children in the Tokyo ward area, 2009.

Table 2 summarizes the age-wise percentages of blocks with accessibility values below one and the percentages of preschool-aged children residing in these blocks. For younger children in the age group of 0–2 years, the majority of blocks have accessibility below one, and the percentage of blocks, as well as population with accessibility values below one, is the largest for one-year-old children. A comparison with the spatial distributions of preschool-aged children and childcare centers (for one-year-old children, see Fig. 3b and c) uncovers that blocks with low accessibility have the following three patterns in general. First, there are no childcare centers nearby (within the commuting threshold). Second, nearby centers do not provide care for a particular age group. Third, demand exceeds supply for a particular age group, even when nearby centers do provide care for that age group.

Table 2. Percentages of basic unit blocks and population with accessibility values below one
Commuting thresholdPercentage of blocks with accessibility values below onePercentage of population in blocks with accessibility values below one
500 meters750 meters1,000 meters500 meters750 meters1,000 meters
0 year old65%66%69%67%70%72%
1 year old64%68%72%67%71%75%
2 years old64%53%54%67%57%57%
3 years old50%48%46%53%51%49%
4 years and older51%50%47%54%52%50%

The sensitivity to the two alternative commuting thresholds of 500 and 1,000 meters was examined. When the 500 meter threshold was used, the number of blocks with considerably low accessibility values below 0.25 increased substantially, particularly for children aged 0, for whom both the number and capacity of childcare centers are limited; many blocks do not have childcare centers within the 500 meter threshold. On the other hand, when the 1,000 meter threshold was used, the number of blocks with considerably low accessibility decreased because many blocks have childcare centers within 1,000 meters. However, there were still many blocks with accessibility below one. This result indicates that even when parents can travel a certain distance to a childcare center, many areas still experience excess demand or childcare shortage. Thus, although alternative thresholds led to different spatial variations in accessibility, the finding that low accessibility exists in many blocks was consistent.

The considerable geographic mismatch for younger children appears to be related to the large numbers on childcare waiting lists (see Fig. 1). I investigated this relationship by using simple linear regression models relating the accessibility measurement to the number on the waiting list in 2009 at the ward level for each age group (Appendix Appendix). As the smallest spatial unit for which waiting list numbers were available was the ward, the population weighted-average accessibility at the ward level was used for the regression. As expected, poorer accessibility was significantly associated with a longer waiting list for children aged under three, and the most significant association was found for the one-year-old group, for which the waiting list was the longest. On the other hand, the association for children aged three and over was smaller in magnitude and, subsequently, insignificant. The weak association for the children aged three and older may be related to the fact that they can be accommodated in kindergartens, which is likely to shorten childcare waiting lists.

4.2 Childcare Access and Attaining Preferred Employment

Table 3 presents the estimation results of probit models examining the importance of access to a childcare center in attaining preferred employment by women with preschool-aged children. The five models using different sets of variables were estimated to examine whether the importance of access to a childcare center differs between: (i) a desired childcare center and an other-than-desired childcare center; (ii) a woman preferring to work full-time and a woman preferring to work part-time; and (iii) a woman with a very young child in the age group of 0–2 years and a woman whose youngest child is in the age group of 3–5 years.

Table 3. Probit estimates of attaining preferred employment
 Model 1Model 2Model 3Model 4Model 5
  1. Statistical significance is indicated in the following manner: ***P < 0.01; **P < 0.05; and *P < 0.10. Standard errors are given in parentheses. Marginal effects at the mean of the sample are in square brackets; marginal effects for dummy variables are for discrete changes from 0 to 1. Marginal effects for interaction variables are for discrete changes from 0 to 1 in variables related to access to childcare centers when (a) full-time work preference is 1; (b) full-time work preference is 0; (c) child in the age group of 0–2 years is 1; and d) child in the age group of 0–2 years is 0.
Age 30–34 years−0.077[−0.025]−0.076[−0.024]−0.069[−0.022]−0.117[−0.036]−0.057[−0.018]
 (0.349) (0.349) (0.349) (0.358) (0.363) 
Age 35–39 years0.129[0.042]0.130[0.042]0.138[0.045]0.069[0.022]0.124[0.040]
 (0.353) (0.353) (0.353) (0.361) (0.366) 
Age ≥40 years−0.287[−0.086]−0.284[−0.085]−0.298[−0.089]−0.320[−0.092]−0.332[−0.099]
 (0.406) (0.407) (0.408) (0.411) (0.418) 
2 or more children0.512**[0.161]0.514**[0.162]0.505**[0.158]0.560***[0.171]0.471**[0.149]
 (0.208) (0.209) (0.208) (0.213) (0.218) 
Child age 0–2 years−0.430*[−0.140]−0.429*[−0.139]−0.457*[−0.148]−0.874***[−0.277]−0.936***[−0.199]
 (0.235) (0.235) (0.238) (0.304) (0.310) 
Full-time partner−0.779***[−0.282]−0.780***[−0.282]−0.756***[−0.273]−0.835***[−0.299]−0.902***[−0.330]
 (0.243) (0.243) (0.246) (0.248) (0.250) 
Work experience before childbirth0.669***[0.207]0.670***[0.208]0.644***[0.199]0.653***[0.197]0.594***[0.186]
 (0.217) (0.217) (0.219) (0.221) (0.223) 
Full-time work preference    0.158[0.059]    
     (0.303)     
Access to childcare center (acc1)1.123***[0.387]  1.066***[0.373]0.691***[0.230]  
 (0.196)   (0.238) (0.258)   
Access to desired childcare center (acc2)  1.111***[0.395]    0.605**[0.474]
   (0.213)     (0.270) 
Access to other-than-desired childcare center (acc3)  1.160***[0.433]    1.234**[0.453]
   (0.335)     (0.536) 
Full-time work preference × acc1    0.065[0.408]a    
     (0.420)[0.359]b    
Child age 0–2 years × acc1      0.983**[0.514]c  
       (0.392)[0.263]d  
Child age 0–2 years × acc2        1.321***[0.619]c
Child age 0–2 years × acc3        (0.440)[0.235]d
         −0.033[0.419]c
         (0.705)[0.454]d
Accessibility of childcare centers, ward-level, 750 meters, value × 100 (acc4)0.004[0.001]0.004[0.001]0.004[0.001]0.003[0.001]0.001[0.000]
 (0.006) (0.006) (0.006) (0.006) (0.006) 
Female unemployment rate (ward level, %)0.050[0.016]0.047[0.015]0.035[0.011]0.074[0.023]0.126[0.041]
 (0.129) (0.131) (0.131) (0.132) (0.138) 
Constant−1.423* −1.424* −1.361* −1.162 −1.090 
 (0.767) (0.767) (0.777) (0.783) (0.785) 
Number of observations261 261 261 261 261 
Percentage correctly predicted78.9 78.9 79.3 77.4 79.3 
Log-likelihood value−118.9 −118.9 −118.5 −115.7 −114.1 
McFadden's pseudo R-squared0.26 0.26 0.26 0.28 0.29 

Model 1 examines the importance of access to a childcare center in attaining preferred employment. Access to a childcare center is significantly associated with a higher probability of attaining preferred employment. The estimated marginal effect (at the mean) indicates that the probability of attaining preferred employment is 38.7 percentage points higher when an otherwise-average woman has access to a childcare center. Model 2 divides childcare centers into desired and other-than-desired. Both access to a desired childcare center and access to an other-than-desired childcare center have positive and significant associations with the attainment of preferred employment.

In Model 3, an interaction term is included in order to examine whether the association between access to a childcare center and the attainment of preferred employment differs between full-time and part-time as the preferred employment. The results indicate that the association does not differ significantly according to the type of preferred employment. Model 4 includes an interaction term to examine whether the association differs by the presence of a very young child in the age group of 0–2 years. Indeed, the association is greater for a woman with a very young child in the age group of 0–2 years than for a woman whose youngest child is in the age group of 3–5 years. For an otherwise-average woman with a very young child in the age group of 0–2 years, the probability of attaining preferred employment is 51.4 percentage points higher when she has access to a childcare center. On the other hand, when the youngest child is in the age group of 3–5 years, the difference is much smaller—26.3 percentage points. Model 5 divides childcare centers into desired and other-than-desired centers and adds two interaction terms to examine whether the association between access to each of the two types of centers and the attainment of preferred employment differs by the presence of a very young child in the age group of 0–2 years. A significant difference emerges when the childcare center is the desired center. For an otherwise-average woman whose youngest child is in the age group of 0–2 years, the probability of attaining preferred employment is 61.9 percentage points higher when she has access to a desired childcare center, whereas for a woman whose youngest child is in the age group of 3–5 years, the difference is much smaller, 23.5 percentage points. When the childcare center is an other-than-desired center, the association is not significantly different because of the presence of a child in the age group of 0–2 years.

It must be noted that access to childcare centers and employment are conceivably endogenous; in other words, employment may affect access to childcare centers. Therefore, the estimated results cannot prove the causality, but do provide evidence suggesting that access to childcare and the attainment of preferred employment are strongly associated.

In all of the models, the ward-level accessibility of childcare centers using the commuting threshold of 750 meters shows weak and insignificant association with the attainment of preferred employment. The insignificant result holds true when alternative accessibility with the threshold of 500 or 1,000 meters is used. These results may be partially a result of using accessibility measurements at the ward level, rather than at the basic unit block level (see Fig. 3), because of the limited information from residential areas in the sample. The particularly tight childcare market (excess demand) may also be partially responsible for the insignificant result. In a tight market, a woman might put extra effort in overcoming the spatial barrier by finding other childcare help or moving residence.

Other variables that are significantly associated with the probability of attaining preferred employment in all of the models are discussed below. Living with two or more children is positively associated with the attainment of preferred employment, which might suggest that having more than one child causes a greater financial need to work. The presence of a very young child (under the age of three) is negatively associated with the attainment of preferred employment. Living with a husband or partner who is working full time is also negatively associated with the attainment of preferred employment. Work experience before childbirth is related to a higher probability of attaining preferred employment. These results are reasonable given the existing labor market literature.

5 Conclusions

Accessibility at the block level—a micro area—has revealed the existence of a considerable geographic mismatch between the demand and supply of childcare centers, particularly for children aged two or below. For these children, lower accessibility of childcare centers was significantly associated with a longer childcare waiting list. Key findings from the probit models are summarized in the following three points. First, access to a childcare center was significantly associated with a higher probability of attaining preferred employment among women with preschool-aged children. Second, the association did not differ significantly according to the status of preferred employment — full-time versus part-time. Third, access to a childcare center, particularly a desired center, was significantly more important in attaining preferred employment for a woman with a child in the age group of 0–2 years than for a woman whose youngest child is in the age group of 3–5 years.

The findings of this study have important policy implications. First, resolving the geographic mismatch for children in the age group of 0–2 years helps to reduce the number of childcare waiting lists. The location of childcare centers within a city and the extent of the development of these centers are aspects that have not been addressed by any evident policy thus far. The accessibility measurement for spatial micro areas can be a useful indicator in the development of such a policy. For example, establishing childcare centers in low-accessibility areas with high demand would be an effective approach for reducing the number of waiting lists.

Second, providing adequate childcare helps women attain preferred employment while raising children, which encourages active participation of women in the labor market. It would be beneficial in areas experiencing childcare shortage, as well as promoting female labor market participation. There is a great potential here, for example, as indicated by the substantial gap between current and preferred employment status. Indeed, many countries, including Japan, are developing strategies to help more women participate in the labor market and increasing the availability of childcare is an important policy option in this regard (Ministry of Health, Labour and Welfare of Japan 2010a; European Commission 2011).

Third, providing quality and affordable childcare for very young children (under the age of three) is particularly valuable for the attainment of preferred employment by women. A substantial number of women need childcare in order to continue working while raising children. The lack of access to reliable childcare can not only discourage such women from continuing to work, but also deteriorate their future employment prospects, as gaps in work history are a disadvantage for those who attempt to re-enter the labor market. In fact, the probit results show that continuous work experience is an important factor in the attainment of preferred employment. The probit results also suggest that for women with very young children, access to quality and affordable childcare are of particular importance for continuing employment. As is evident, the positive association between access to a desired childcare center and the attainment of preferred employment was significantly greater for a woman with a very young child in the age group of 0–2 years than for a woman whose youngest child is in the age group of 3–5 years. Further, the survey data employed in this study provide answers regarding the most desirable childcare service. Among women with children in the age group of 0–2 years, a strong majority (73%) selected licensed daycare centers that provide relatively high quality and affordable childcare.

The results of this study warrant further research. The first direction is to refine the models to examine causal relationships between access to childcare and employment, which are likely to be endogenous. The second direction is to use increasingly spatial micro-level data on the accessibility of childcare in the models. The results obtained could be more significant if sufficiently spatially detailed accessibility were incorporated into the models. Further research in these directions would enhance our knowledge about childcare access and female participation in the labor market.

Acknowledgment

This work was supported by the Grants-in-Aid for Scientific Research [22510140,25285080] from the Japan Society for the Promotion of Science. The author is grateful to the Bureau of Social Welfare and Public Health of the Tokyo Metropolitan Government, local municipalities, and individual childcare centers for the valuable childcare data they provided for this study. The author would like to thank Yukiko Abe, Yukako Ono, and participants in conferences, seminars, and workshops for their valuable suggestions. The author also thanks the referees for their helpful comments. Any errors in this paper, however, are the author's sole responsibility.

  1. *, **, and *** denote significance at the 10%, 5%, and 1% level, respectively. Standard errors are in parentheses. The number of observations for each model is 23.

Notes

  1. 1

    “Re-use of this article is permitted in accordance with the Terms and Conditions set out at http://wileyonlinelibrary.com/onlineopen#OnlineOpen_Terms.”

  2. 2

    Gordon and Chase-Lansdale (2001) use data at the zip-code level; however, the supply-demand mismatch is presented at the metropolitan level.

  3. 3

    The median areas of wards and basic unit blocks in the Tokyo ward area are 20.34 square kilometers and 3,456 square meters, respectively. The median number of blocks in wards is 4,066.

  4. 4

    The costs of the three types of childcare centers are regulated so that their maximum monthly fees are set at approximately 80,000 yen (approximately $800 at JPY/USD100). Fees for licensed daycare centers are determined depending on family income and the age of the child, from no fee at all (for welfare recipients, for example) to approximately 80,000 yen per month. Suzuki (2010) estimated that the average monthly fee paid for licensed daycare centers is 20,000–30,000 yen (approximately $200–$300).

  5. 5

    These figures are based on the 2009 ratios in the Tokyo ward area that were estimated on the basis of available data, using the following equation:

    display math(2)

    where a denotes the age of children, Sl the number of children enrolled in licensed daycare centers, St the capacity of TMG-certified and authorized childcare centers (as the number of children enrolled in these two types of centers were not obtainable), Ql the number of applicants for licensed daycare centers minus the number of applicants admitted, Et the number of children in Ql but enrolled in TMG-certified or authorized childcare centers, and P the population. The estimated ratios are 17% (for age 0), 34% (one year), 37% (two and three years), and 35% (four and five years).

  6. 6

    In the 2005 census, the Tokyo ward area comprises 3,139 subdivisions. The median area of the subdivisions in the Tokyo ward area is 172,125 square meters.

  7. 7

    In the Tokyo ward area, the most common means of traveling to childcare centers is by bicycle on days when weather permits, but by foot on days with inclement weather (Kawabata 2010a). Pinch (1984) and Webster and White (1997) use a similar threshold — approximately 800 meters (half a mile) — which Pinch considers a reasonable duration given travel challenges on bad weather days.

  8. 8

    The probit models assume that the association between access to childcare and the probability of attaining preferred employment is the same for women who work part-time, but wish to work full-time, and for women who work full-time, but wish to work part-time. One may question this assumption because when childcare accessibility is improved, women who work part-time, but wish to work full-time, may switch to full-time jobs, but women who work full-time, but wish to work part-time, may not switch, owing to reasons such as financial pressures. In areas that experience childcare waiting lists, however, some women might choose to work full-time in order to use licensed daycare centers, as licensed daycare centers prioritize applicants employed in full-time jobs over part-time workers. If childcare access is improved, such women may switch to part-time jobs. Constructing more refined models that take into account these considerations is a topic for future research.

  9. 9

    Basic demographic statistics of the survey and 2005 census data are presented in Table A1 in Kawabata (2012). Details regarding the survey are documented in Kawabata (2010a).

Appendix I

Regression estimates of the number of children on childcare waiting lists.

 Age 0Age 1Age 2Age 3Age 4 and olderTotal
Accessibility
500 meters−68.04**−334.66***−116.46**−24.45−1.80−380.04*
 (32.07)(120.97)(56.39)(16.61)(4.52)(196.02)
750 meters−68.00**−373.28***−123.01**−24.74−2.37−410.80**
 (32.83)(119.07)(58.85)(17.34)(4.65)(196.27)
1,000 meters−71.68**−388.21***−121.80**−22.68−2.07−392.49*
 (35.66)(126.83)(59.26)(17.63)(4.65)(202.14)
Adjusted R-squared
500 meters0.140.230.130.05−0.040.11
750 meters0.130.290.130.04−0.030.13
1,000 meters0.120.280.130.03−0.040.11

Ancillary