Printer Friendly

An investigation into the positive effect of an educated wife on her husband's earnings: the case of Japan in the period between 2000 and 2003.

Abstract We analyze the effect of a wife's human capital on her husband's earnings, using individual-level data for Japan in the period 2000-2003. We find a positive association between a wife's education and her husband's earnings, which can be attributed to the assortative mating effect as well as the positive effect of an educated wife on her husband's productivity. We divide the sample into those couples with nonworking wives and those with working wives, and also employ an estimation strategy proposed by Jepsen (Review of Economics of the Household 3:197-214, 2005), attempting to control for the assortative mating effect. Our regression analysis provides suggestive evidence that educated wives increase their husbands' productivity and earnings only when they are non-workers and have sufficient time to support their husbands.

Keywords Earnings * Human capital * Marriage * The family. Assortative mating Cross-productivity effect within marriage

JEL D13 * J22 * J24 * J31

Introduction

It is widely recognized that human capital is accumulated through costly investment, such as formal education and working experience (Becker 1964). Human capital is also highly influenced by interaction with surrounding people through sophisticated conversations and the like, and thus economic outcomes such as one's earnings are often associated with family and community backgrounds (e.g., Behrman and Wolfe 1984; Boulier and Rosenzweig 1984; Hauser and Sewell 1986; Corcoran et al., 1990, 1992). (1) Specifically, Benham (1974) was the first to argue that an educated wife improves her husband's productivity and thus increases his earnings, a phenomenon called "cross-productivity effect within marriage." (2) Using United States census data from 1960 to 2000, Jepsen (2005) finds that a wife's education is positively associated with her husband's earnings, but the magnitude of the effect declines over time. (3) Jepsen conjectured that the rapid increase in a wife's labor participation reduced her time to improve husband's productivity, but no direct evidence was provided. Loh (1996) and Gray (1997) find that a wife's labor participation is negatively associated with her husband's earnings, but they do not pay direct attention to the wife's educational level. Therefore, little is known about how much a wife's labor participation reduces the positive effect of her education on her husband's productivity and earnings.

This paper uses individual level data from Japan from 2000 to 2003 to examine whether and how much a wife's labor participation influences the effect of her education on her husband's productivity and earnings. We found that an educated wife improves her husband's productivity and earnings only when she is a nonworking wife and has sufficient time to support her husband.

Empirical Strategy

This paper uses Japanese General Social Survey (hereafter, JGSS) data. (4) JGSSs adopt a two-step stratified sampling method and were conducted throughout Japan between 2000 and 2003. The surveys included standard questions about an individual's and his/her family characteristics through face-to-face interviews. These data cover information related to marital and demographic (age and gender) status, annual income, years of schooling, age, and size of residential area. Spouses' demographics (age and gender) status, job categories, and years of schooling were also obtained.

Table 1 presents the definitions of the variables we use and their mean values. All the observations in our sample (n=5,200) were of married couples. The sample was divided into two groups by the wife's labor participation status: working in one group, and not working in the other group. There was no statistical difference in the mean values of any observed characteristics between the two groups. A husband's annual income (INCOMH) in the working-wife group (around 5.6 million yen) was almost the same as that of the non-working-wife group (5). On average, husbands were 50 years old and had 13 years of schooling. Wives were around 47 years old and had 12 years of education.
Table 1 Variable definitions and means

Variables  Definition                       Non-working  Working   All
                                                   wife     wife

INCOMH     Husband's annual income (in ten          565      561   563
           thousand yen)

EDUH       Husband's years of schooling            12.9     12.7  12,8

EDUW       Wife's years of schooling               12.2     12.3  12.3

AGEH       Husband's age                           49.5     49.6  49.5

AGEW       Wife's age                              46.7     47.0  46.9

Obs.                                               2283     2659  5200

Values are simple averages of yearly values over the period 2000 2003.
The total sample of "non-working wife" and "working wife" is 4942,
which is smaller than the hall sample, 5200. Observations without
data about a wife's work status lead to this difference.


From Table 2, we can see that not only EDUH (husband's years of schooling) but also EDUW (wife's years of schooling) is positively correlated with INCOMH, which is consistent with the cross-productivity effect within marriage (Benham 1974). We also find that the correlation between EDUH and EDUW is 0.65, and that between AGEH (husband's age) and AGEW (wife's age) is 0.95, suggesting that people tend to marry partners of a similar age and educational level. This finding is congruent to the assortative mating in education and age (Becker 1975). That is, productive males tend to marry well educated females, leading to a wife's education being positively associated with her husband's earnings.
Table 2 Correlation matrix

Variables     INCOMH       EDUH       EDUW      AGEH  AGEW

INCOMH             1         --         --        --    --

EDUH        0.35 ***          1         --        --    --
              (0.00)

EDUW        0.31 ***   0.65 ***          1              --
              (0.00)     (0.00)

AGEH       -0.05 ***  -0.31 ***  -0.39 ***         1    --
              (0.00)     (0.00)     (0.00)

AGEW       -0.06 ***  -0.31 ***  -0.40 ***  0.95 ***     1
              (0.00)     (0.00)     (0.00)    (0.00)

As the correlation matrix is symmetric, --indicates the
omitted elements to avoid redundancies. Numbers in
parentheses are p-statistics. ***indicates statistical
significance at the 1% level


In line with Benham (1974) and Jcpsen (2005), the regression model takes the following form:

In[(INCOMH).sub.i] [[alpha].sub.0] + [[alpha].sub.1] [EDUH.sub.i] + [[alpha].sub.2][EDUH.sub.i], + [[alpha].sub.3][AGEH.sub.i] + [[alpha].sub.4][AGEW.sub.] + [Z.sub.i]/[beta] + [[mu].sub.i],

where subscript i denotes married couple i, and the logarithm of [INCOMH.sub.i] is the dependent variable. Regression parameter [alpha]'s are to be estimated, and [[mu].sub.i] is the error term. Since the data on years of work experience is not available, the husband's age is incorporated to capture his work experience. In addition, to control for general market conditions and macro-level shocks, large city and medium size city dummies (size of residential area) and year dummies are incorporated in Z, the vector of control variables, with [beta] as the vector of corresponding coefficients.

Our major focus in this paper is to find out whether the cross-productivity effect is at work; that is, whether an educated wife improves her husband's productivity and earnings (see, e.g., Benham, 1974; Scully, 1979; Kenny, 1983; Wong, 1986; Lam and Shoeni 1993; Lefgren and McIntyre 1996; Huang et al., 2009). If an educated non-working wife spends a certain amount of time to support her husband and consequently raises her husband's productivity whereas a working wife does not have enough time to do so, the coefficient on EDUW is expected to take a positive sign only in a sub-sample of couples with non-working wives but not in a sub-sample of couples with working wives. The assortative mating hypothesis, however, also predicts a positive association between a wife's human capital and her husband's earnings, regardless of the wife's labor participation status (Welch, 1974; Liu and Zhang, 1999; Lefgren and McIntyre 1996; Huang et al., 2009). We are concerned that this assortative mating effect could be sufficiently strong, and the cross-productivity effect might be masked and our hypothesis testing may not work.

In order to alleviate this identification problem between the cross-productivity effect within marriage and the assortative mating effect, we will make our best effort to control for the assortative mating effect. Including husband's own education as a covariate in the regression function is considered as a good way to at least partially control for the mating effect (Huang et al., 2009). Furthermore, Jepsen (2005) proposes controlling for the assortative mating effect by using a sub-sample containing only husbands and wives who have an age difference of more than 5 years. She claims that "this sample represents couples who are less likely to have met each other either in high school or college" (Jepsen 2005, p.204). (6) By minimizing the assortative mating effect, this estimation strategy helps to isolate the cross-productivity effect. As a result, it is expected that a sub-sample of couples with non-working wives will exhibit a significantly positive coefficient on EDUW that is not present in a sub-sample of couples with working wives. (7) Such an estimation result would imply that it takes a certain amount of time for an educated wife's human capital to improve her husband's productivity and earnings. By contrast, a working wife does not have sufficient time to support her husband, and this newly-discovered foregone increase in husband's earnings should be considered as an additional component of opportunity cost to a working wife, though it has never been explicitly taken into account in the existing literature.

Estimation Results

Table 3 presents our estimation results. The results in Columns (1)-(3) are based on the original sample of married couples, whereas the results in columns (4)-(6) are of the sample that excludes couples with an age difference of less than 5 years. The results using the sample of non-working wives are in columns (2) and (5), while the results using the sample of working wives are in columns (3) and (6). As shown in the first row, the coefficient on EDUH takes a positive sign with 1 % statistical significance in all estimations, consistent with the standard theory of human capital. In Columns (1) to (3) the estimated coefficient on EDUW is positive and statistically significant; its magnitude indicates that an additional year of a wife's education increases her husband's annual earnings by 4 to 6 percentage points, which is slightly below the effect of a husband's education but economically significant. This estimation result that the coefficient on EDUW is significantly positive irrespective of the wife's labor participation status implies the assortative mating. When this assortative mating effect is controlled for (Columns 4 to 6), the coefficient on EDUW still remains significantly positive in the sub-sample of the non-working wives (Column 5), whereas the coefficient on EDUW becomes insignificant in the sub-sample of the working wives (Column 6). This estimation result suggests that an educated non-working wife supports her husband and raises his productivity, whereas a working wife does not have sufficient time to support her husband as much. In other words, the cross-productivity effect works only when the wife devotes sufficient time to support her husband.
Table 3 Regression results on husband's annual income

Variables    All currently married.              Difference in age
                                                 between  husband and
                                                 wife > 5 years.

             (1) All         (2)  (3) Worker      (4) All           (5)
                      Non-worker        wife                 Non-worker
                            wife                                   wife

EDUH        0.06 ***    0.08 ***    0.04 ***     0.07 ***      0.07 ***
              (13.9)      (12.2)      (7.12)       (6.90)        (4.66)

EDUW        0.05 ***    0.06 ***    0,04 ***     0.05 ***      0.07 ***
              (8.57)      (6.70)      (5.38)       (3.01)        (3.31)

AGEH          -0.002      -0.002      -0.002     -0.008 *      -0.01 **
             (-0.85)     (-0.64)     (-0.79)      (-1.70)       (-2.01)

AGEW          -0.003      -0.003      -0.001       -0.005  0.002 (0.35)
              (1.15)     (-0.91)     (-0.36)       (1.19)

Constant    4.97 ***    4.68 ***    5.19 ***     5.34 ***      5.16 ***
              (50.1)      (34.6)      (34.3)       (20.7)        (14.3)

Obs.            5200        2283        2659          901           389

Adj             0.16        0.24        0.10         0.21          0.28
[R.sup.2]

Variables

           (6) Worker
                 wife

EDUH         0.07 ***
               (4,53)

EDUW             0.02
               (1.11)

AGEH           -0.005
              (-0.85)

AGEW           -0.007
              (-1.16)

Constant     5.56 ***
               (13.8)

Obs.              473

Adj              0.14
[R.sup.2]

The dependent variable is the logarithm of the husband's annual
income. Numbers in parentheses are t-statistics obtained by
robust standard errors. *, **. and *** indicate statistical
significance at the 10, 5, and 1 % levels, respectively.
Although not reported here, large and medium-sized city,
and year dummies are also controlled for


One would expect that age, acting as a proxy for experience, would have a positive effect on income. However, as shown in AGEH in Table 3, it has a negative effect, significantly so in two cases. I interpret this result as suggesting that the relationship between a husband's age and his income is non-linear. For the purpose of examining this, in addition to AGEH and AGEW, I also incorporate their squares, [AGEH.sup.2] and [AGEW.sup.2], as independent variables. The results of the alternative specification are in Table 4. Furthermore, AGEH and AGEH2 take positive and negative signs, respectively, and show statistical significance in all estimations. It follows that a husband's income increases with his age up to a certain level, but then decreases thereafter. Hence, the relationship between a husband's age and his income is considered non-linear. In respect to the main variables in Table 4, the results of EDUH and EDUW do not change in the alternative specification.
Table 4 Regression Results on husband's annual income

Variables                       All currently married.

                           (1)          (2)           (3)           (4)

                           All   Non-worker   Worker wife           All
                                       wife

EDUH           0.05 *** (12.1)     0.06 ***      0.04 ***      0.06 ***
                                     (10.3)        (6.17)        (6.65)

HDUW           0.04 *** (6.08)     0.04 ***      0.04 ***      0.04 ***
                                     (4.42)        (4.31)        (2.59)

AG EH          0.11 *** (10.1)     0.09 ***      0.12 ***      0.12 ***
                                     (5.62)        (8.34)        (6.23)

[AGEH.sup.2]        -0.001 ***   -0.001 ***    -0.001 ***    -0.001 ***
                       (-10.4)      (-5.46)       (-8.56)       (-6.54)

AGEW            0.02 ** (2.05)     0.05 ***  0.008 (0.52)  0.003 (0.21)
                                     (3.14)

[AGEW.sup.2]   -0.0002 (-1.64)  -0.0005 ***      -0.00004      -0.00001
                                    (-2.68)       (-0.27)       (-0.09)

Constant       2.11 *** (15.7)     1.83 ***      2.19 ***      2.03 ***
                                     (10.1)        (9.85)        (4.71)

Obs.                      5200         2283          2659           901

Adj [R.sup.2]             0.29         0.38          0.22          0.29

Variables          Difference in age
                  between husband and
                    wife>5 years

                       (5)         (6)

                Non-worker      Worker
                      wife        wife

EDUH              0.07 ***    0.07 ***
                    (4.44)      (4.29)

HDUW              0.06 ***        0.01
                    (2.86)      (0.81)

AG EH              0.08 **    0.15 ***
                    (2.20)      (5.33)

[AGEH.sup.2]     -0.001 **  -0.001 ***
                    (2.40)     (-5.44)

AGEW           0.03 (1.15)       -0.02
                               (-1.07)

[AGEW.sup.2]       -0.0003       0.000
                   (-0.77)      (1.17)

Constant          2.02 ***    2.05 ***
                    (2.98)      (3.09)

Obs.                   389         473

Adj [R.sup.2]         0.36        0.21

The dependent variable is the logarithm of the husband's annual
income. Number in parentheses are t-statistics obtained by robust
standard errors, *, **, and *** indicate statistical significance
at the 10, 5, and 1! levels, respectively. Although not reported
here, large and medium-sized city, and year dummies are also
controlled for


Conclusion

Jepsen (2005) finds that, using data from 1960 to 2000 in the United States, an educated, non-working wife increases her husband's earnings, but this effect declines over time, and she conjectures that this is likely due to the secular increase in labor participation by married women.

This directly examined whether and how much a wife's labor participation changes the effect of her education on her husband's earnings by using individual-level data from Japan. We found that a wife's human capital has a positive association with her husband's earnings, for both working and non-working wives. After restricting the sample to married couples with an age difference greater than 5 years to partly control for the assortative mating effect, however, the positive effect of a wife's education continues to be observed only in the sub-sample of non-working wives, whereas the effect becomes insignificant in the sub-sample of working wives.

Our statistical analysis, therefore, provides the suggestive evidence for both the assortative mating effect and the cross-productivity effect within marriage. Moreover, the cross-productivity mechanism is time-consuming, as Jepsen (2005) rightly conjectured. To our best knowledge, this has a new and important implication in considering the labor participation of married women, since the existing literature has not explicitly taken into account this cross-productivity effect within marriage as one component of the opportunity cost to working women.

References

Amin, S., & Jepsen, L. (2005). The impact of a wife's education on her husband's earnings in Malaysia. Journal of Economics, 31, 1-18.

Becker, G. S. (1964). Human Capital: A Theoretical and Empirical Analysis. New York: Columbia University Press for the National Bureau of Economic Research.

Becker, G. S. (1975). A Theory of Marriage. In W. Theodore (Ed.), Economics of the Family: Marriage, Children, and Human Capital. Schultz: University of Chicago.

Benham, L. (1974). Benefits of women's education within marriage. Journal of Political Economy, 82(2), S57-S71.

Behrman, J., & Wolfe, B. L. (1984). Thc socioeconomic impact of schooling in a developing country. The Review of Economics and Statistics, 66, 296-303.

Boulier, B. L., & Rosenzweig, M. R. (1984). Schooling, search, and spouse selection: Testing economic theories of marriage and household behavior. Journal of Political Economy, 92(4), 712-732.

Corcoran, M., Gordon, R., Laren, D., & Solon, G. (1990). Effects of family and community background on economic status. American Economic Review, 89, 362-366.

Corcoran, M., Gordon, R., Laren, D., & Solon, G. (1992). The association between men's economic status and their family and community origins. Journal of Human Resources, 27(4), 575-601.

Gray, J. S. (1997). The fall in men's return to marriage: Declining productivity effects or changing selection? Journal of Human Resources, 32(3), 481-504.

Hauser, R. M., & Sewell, W. H. (1986). Family effects in simple models of education, occupational status, and earnings: Findings from the Wisconsin and Kalamazoo studies. Journal of Labour Economics, 4(3), S83-S115.

Heckman, J., & Hotz, V. J. (1986). An investigation of the labour market earnings of Panamanian males: Evaluating the sources of inequality. Journal of Human Resources, 23, 462-487.

Huang, C., Li, H., Liu, P. W., & Zhang, J. (2009). Why does spousal education matter for earnings? Assortative mating and cross-productivity. Journal of Labor Economics, 27(4), 633-652.

Jepson, L. K. (2005). The relationship between wife's education and husband's earnings: Evidence from 1960-2000. Review of Economics of the Household, 3, 197-214.

Kenny, L. (1983). The accumulation of human capital during marriage by males. Economic Inquily, 21(2), 223-231.

Lam, D., & Shoeni, R. F. (1993). Effects of family background on earnings and returns to schoolings: Evidence from Brazil. Journal of Political Economy. 101(4), 710-740.

Lam, D., & Shoeni, R. F. (1994). Family tics and labour markets in the United States and Brazil. Journal 0. Human Resources, 29, 1235-1258.

Lcfgrcn, L., & McIntyre, F. (1996). The relationship between women's education and marriage outcomes. Journal of Labor Economics, 24(4), 787-830.

Liu, P.-W., & Zhang, J. (1999). Assortative mating versus the cross-productivity effect. Applied Economics Letters, 6(8), 523-525.

Loh, E. S. (1996). Productivity differences and the marriage wage premium for white males. Journal of Human Resources, 31(3), 566-589.

Neuman, S., & Ziderman, A. (1992). Benefits of women's education within marriage: results of Israel in dual labor market context. Economic Development and Cultural Change, 40, 413-426.

Scully, G. W. (1979). Mullahs, Muslims and marital sorting. Journal of Political Economy, 87,1139-1143.

Tiefenthaler, J. (1997). The productivity gains of marriage: Effects of spousal education on own productivity across market sectors in Brazil. Economic Development and Cultural Change, 45(3), 633-650.

Welch, F. (1974). Benefits of women's education within marriage: Comment. Journal of Political Economy, 82(2), S72-S75.

Wong, Y.-C. (1986). Entrepreneurship, marriage, and earnings. The Review of Economics and Statistics, 68 (4), 693-699.

Yamamura, E. (2008). Diffusion of home computers and social networks: a study using Japanese panel data. Applied Economics Letters, 15(15), 1231-1235.

(1.) As an example of social learning, Yamamura (2008) reports a case study from Japan in which people learned how to use computers from neighbors that already owned one.

(2.) Their parents' schooling is also found to be positively associated with his earnings (e.g., Heckman and Hotz 1986; Lam and Shoeni, 1993, 1994).

(3.) It is widely observed that a wife's human capital positively influences a husband's earnings; for instance, in Israel (Neuman and Ziderman 1992), Iran (Scully 1979), the Philippines (Boulier and Rosenzweig 1984), Malaysia (Amin and Jepsen 2005), and Brazil (Lam and Shoeni, 1993, 1994; Tiefenthaler, 1997).

(4.) Data for this secondary analysis, "Japanese General Social Surveys (JGSS), Ichiro Tanioka," were provided by the Social Science Japan Data Archive, Information Center for Social Science Research on Japan, Institute of Social Science, The University of Tokyo.

(5.) It is possible that a husband's earnings are more appropriate than income in this context. However, for this paper, a husband's income is not used, as this data was not available.

(6.) Admittedly, this argument is not entirely convincing, as one does not have to meet in school to mate assortativcly.

(7.) Precisely speaking, the decision making process of a wife's labor participation should be considered to control for self-selection. This is, however, beyond the scope of this note and is an issue to be addressed in a future study.

Int. Adv Econ Res (2012) 18:409-416

Published online: 29 June 2012

[c] International Atlantic Economic Society 2012

E. Yamamura ([??])

Department of Economics, Scinan Gakuin University, 6-2-92 Nishijin, Sawara-ku, Fukuoka 815-0075, Japan

e-mail: yamaci@scinan-gu.ac.jp

Y. Mano

Graduate School of Economics, Hitotsubashi University, 2-1 Naka, Kunitachi, Tokyo 186-8601, Japan

e-mail: yukichi.mano@gmail.com

e-mail: ymano@gmail.com

DOI 10.1007/s11294-012-9368-x
COPYRIGHT 2012 Atlantic Economic Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2012 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Yamamura, Eiji; Mano, Yukichi
Publication:International Advances in Economic Research
Article Type:Report
Geographic Code:9JAPA
Date:Nov 1, 2012
Words:3576
Previous Article:Determinants of interest rate pass-through for emerging market economies: the role of financial market structure.
Next Article:A quick method for assessing economic damage caused by natural disasters: an epidemiological approach.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |