Household Labor Demand And Household Labor Supply1

1

Karsten Hank

Household Labor Demand And Household Labor Supply1 An empirical analysis of the employment of domestic help in German private

households and its effect on female labor force participation

1. Introduction

Although there is already a substantial - mainly "gray" - market for household services in

Germany, the role of the private household as place of work has so far been underestimated

by politicians as well as economists.

Only in recent years, the German government has begun to legislate tax policies designed to

encourage the use of domestic servants as a way to stimulate employment in the household

sector. At the same time, already existing jobs in private households shall be transformed into

such, liable to social security contributions.

Just so, economists have not paid much attention yet to the impact of market-procured domes-

tic work on the female labor force participation decision. So far, this has often been conceptu-

alized in terms of the choice between market work and housework only. Domestic duties,

however, need not be performed by the woman herself: the spread of automated household

appliances could be mentioned here as one possible factor behind the growth in the female

labor force2. Similarly, the use of a domestic servant can be regarded as a readily available

market substitute for self-produced domestic work.

In this paper, data from the German Socio-Economic Panel (SOEP) will be used to examine

the hypothesis that the demand for domestic servants and women’s supply of market work are

joint decisions.

1 This paper is based on work done during a stay at the Center for Policy Research at Syracuse Uni-versity (USA) in 1997. The author would like to thank Prof. Thomas A. Dunn, who, among many others, gave lots of helpful comments. Remaining shortcomings are mine. 2 An empirical investigation of the relation between female labor force participation and the house-hold‘s demand for service goods can be found in Wenke, 1991.

2

2. Overview of the Market for Domestic Work in Germany and Recent Tax

Policies

In the past couple of years, a significant increase in the number of workers who are either not

subject to social security contributions or have a marginal second job3, could be observed in

the German labor market. At the same time, the share of employment in the service sector

grew larger (see Schupp, Schwarze & Wagner, 1995; Schwarze, 1997).

Research, done by the Institute for Social Research and Social Policy (Institut für Sozialfor-

schung und Gesellschaftspolitik, ISG) for the German Federal Department of Labor and So-

cial Affairs, shows that particularly domestic workers account for a large fraction of these

marginal jobs (see Table 1 and Table 2)4.

According to the ISG survey, in 1992 more than 25% of all working persons who were not

subject to social security contributions (about 760,000 persons) worked in private households.

An additional 285,000 persons worked there, having a marginal second job, which is 20% of

all marginal second job holders (see ISG, 1993).

The growth of the number of marginally employed domestic helpers in West Germany can be

explained by numerous factors. Among these are the growing labor force participation of

women, deficits in the compatibility of market work and family (see Klenner & Stolz-Willig,

1996), a growing number of single parents as well as senior citizens with a relatively high

income (see Muntz, 1996). These are, however, rather long term factors. As can be seen from

the ISG figures, there has even been a remarkable growth in the market for domestic work

within only five years, between 1987 and 1992 (see Table 1 and Table 2). This may be due to

the migration of ”cheap” labor from Eastern Europe to Germany during the observation pe-

riod. The wage differential between Germans and Eastern European women allows mutually

beneficial division of labor. This means that a German woman may have an absolute advan-

tage in managing her own household, but she may have a comparative disadvantage compared

to a, say, Polish maid5.

Altogether, there are about 1 million marginally employed domestic workers in Germany,

3 Neither the worker nor the employer pays social security contributions, if the job is regarded as marginal, i.e. less than 15 hours per week and with a monthly wage below 620 DM in West Germany or 520 DM in East Germany (see VDR, 1997). 4 The results of a survey conducted by the ISG in 1997 are not completely published yet. Preliminary figures, however, underline the dominant role of domestic work in this labor market sector (see Da-niels, 1998). 5 This is how Suen (1994) explains the development in the market for domestic labor in Hong Kong, where foreign helpers - especially from the Philippines - have been imported systematically. For ex-amples of foreigners working in German households, see Hanke & Rudzio, 1996.

3

compared to only 34,000 employees in private households, who did pay social security con-

tributions6. Comparing these figures with the 117,000 domestic employees (including 38,000

marginally employed) listed in the 1995 Microcensus (see Emmerich, 1997a), shows that re-

liable data concerning the use of domestic helpers and the job potential in private households

are hard to come by. Other estimates range from 1.4 up to 2.4 million domestic helpers in

Germany (see Weinkopf, 1997a).

One of the main reasons for the underreporting of the official figures is that the German labor

statistics do not include the private household in their definition of places of work (see Hat-

zold, 1986) and do not cover marginal employment to its full extent (see Schwarze, 1992). It

can also be assumed that many self-employed or freelance workers who employ domestic

servants, do not register them as domestic employees but as business employees, so their cost

can be deducted as a business expense for tax purposes (see Muntz, 1996).

This leads to the point that many jobs in the domestic sector are illegal and beyond the control

of state regulations. Employees in private households e.g. do not have to carry their social

security card with them and cannot be controlled at their place of work because of the consti-

tutional protection of dwellings. This is one reason, why standard instruments, which are used

to fight illegal employment or misuse of the social security system in the construction indus-

try, for example, cannot be applied here. It cannot be controlled, whether somebody is in-

volved in illicit work or has multiple marginal jobs (see Bungart, 1997, Weinkopf, 1996).

Hatzold (1988) finds7 that, of all households employing domestic servants, 48% illegally

avoid the payment of social security contributions. The fraction of households that does not

pay the related taxes is even as high as 86%.

Until recently this was tolerated, despite losses in tax payments and social security contribu-

tions (see Hanke & Rudzio, 1996). The pressure to create new jobs, put on politicians by a

record number of unemployed, however changed the evaluation of the private household's

role as employer8.

A practical attempt to set incentives for private households to create jobs subject to social

insurance is found in the 1997 federal tax law. Up to 18,000 DM of the costs for a domestic

servant can be deducted from the household income subject to income taxation9, eligible for

6 This figure is reported in the 1995 labor statistics of the German Federal Bureau of Labor (see Em-merich, 1997b). 7 The survey was conducted by the Ifo-Institute for Economic Research for the German Federal De-partment of Economy. Its sample of 848 West-German households is not representative, though. By its design the survey overestimates high-income households. See also Hatzold, 1986. 8 The state parliament of North Rhine-Westphalia for instance held a hearing on "How to make better use of the job potential in private households" in January 1997 (see Weinkopf, 1997a). 9 The trade union responsible for domestic workers (NGG) suggested a deduction not from the in-

4

all households. In contrast to this, the 1989 law allowed a maximum of 12,000 DM tax deduc-

tion only. Furthermore the eligibility was restricted to households with at least two children

under age ten (for single parents at least one child) or with a person needing care (see BMAS,

1997)10. These changes primarily intend to increase the number of jobs, especially for

low-skilled workers (see Muntz, 1996). At the same time, the introduction of so called

"household checks" is designed to integrate already existing domestic help jobs into the Ger-

man social security system11. While social security contributions usually are calculated by the

employer, in the case of domestic servants the employer has to fill out the household check

only and send it to the health insurance company, which is responsible for collecting the so-

cial insurance payments. On the basis of the information given by the employer (including

duration of the job, payment, and whether the servant is employed elsewhere) the insurance

company calculates the amount of the employers' contributions (see VDR, 1997).

Simulations show that the labor market effect of tax deductions will be relatively weak. Al-

though already existing jobs in the household sector which are subject to social security con-

tributions will be stabilized and a modest growth of their number is predicted, a significant net

job effect is not expected to occur (see Emmerich, 1997a, Muntz, 1996). The advantages of

illicit work or marginal employment still seem to outweigh the utility that could be drawn

from regular employment - for the employing households as well as for the domestic helpers

(see Klenner & Stolz-Willig, 1996). The household e.g. might fear additional costs of regu-

larly employed domestic helpers (like continued payment in case of sickness). The servant, on

the other hand, might not be interested in acquiring own social security benefits, as a majority

of these people is already covered in another way, e.g. by the spouse’s insurance (see Schupp,

Schwarze & Wagner, 1995).

An alternative concept was developed by the Social Democratic Party (see Handelsblatt, 1996,

for example). Its model gives priority to social aspects: only households with at least one

child under age 14 or one person aged 80 or older would be supported. No tax deduction is

planned, but direct transfers in form of vouchers, worth 1,200 DM per month (plus 600 DM

for each additional child or person needing care). The use of domestic services, however,

would only be subsidized, if the servant was engaged via a service agency (see Emmerich,

1997b).

The most discussed form of such service agencies are so called "service pools" (much like come subject to taxation, but from the tax due (see Klenner & Stolz-Willig, 1996). There are doubts, however, if tax subsidies in general are compatible with the German taxation system (see Mundorf, 1996). 10 For legal details of employment in private households, see BMAS, 1996.

5

temporary employment agencies or maid agencies in the US)12. Service pools aim to increase

the number of jobs liable to social security in labor market sectors which are characterized by

a high share of marginally employed workers13. Several small engagements of only a couple

of work hours each, are bundled to a full- or part-time job subject to social security contribu-

tions. In case of the domestic sector, it would be the agency and not the household anymore,

which is the employer of a domestic worker. Supporters of the concept of organizing domestic

services in pools claim that households as well as workers would benefit from it. It is assumed

that a higher quality of the work done and more security for both sides would be guaranteed

(see Weinkopf, 1997b).

Whether such changes would distinctively influence the current structure of employment in

German private households - in terms of hours of work or duties - can be doubted.

Although the work hours of German domestic helpers differ widely, most of the jobs are

part-time, i.e. about 5 hours per week with an hourly wage of ca. 15 DM (Emmerich, 1997a;

Weinkopf, 1997a)14. Hatzold (1986) finds in his survey that not more than 6% of all house-

holds employing domestic help have a full-time servant and only 30% of the part-time jobs

are more than 6 hours per week. The hourly cost of a domestic worker with complete social

security coverage, paid vacation, etc. would be between 30 DM and 35 DM (see Emmerich,

1997b). Assuming a full-time job with standard wage, Drohsel (1996) calculates a monthly

gross income for the servant of 2,340 DM.

Disproportionately high is the fraction of women (often housewives) and foreigners working

in private households (see ISG, 1993). Their main duties are according to Hatzold (1986)

cleaning (64%), washing (25%) and child care (14%)15. Similar results are shown in the 1992

ISG study on marginal employment (see ISG, 1993).

11 A similar policy can be found in France. For details, see Weinkopf, 1996. 12 For alternative approaches, see Drohsel, 1996. 13 For more on the discussion about how to limit marginal employment, see Schwarze, 1993. 14 Domestic servants in Hong Kong for instance usually live in their employers household and are full-time employed (see Suen, 1994). This used to be common practice earlier this century in Germany, too, but is not affordable anymore for most households (see Hatzold, 1988). 15 In comparison, the share of households hiring a nanny in Great Britain, for example, is much bigger. This can be explained by the relatively low provision of non-school child care in the UK (see The Economist, 1996).

6

3. The Structure of Households Employing Domestic Help16

3.1 Data Source

The data used in this paper are drawn from the German Socio-Economic Panel (SOEP), a

longitudinal micro-database, covering socio-economic information on households and indi-

viduals in Germany17.

The survey was started 1984 in the western states of Germany, where about 6,000 households

(including an oversample of foreign headed households) and more than 12,000 individuals

were interviewed. In 1990, the SOEP was supplemented by 2000 households and 4500 indi-

viduals from the eastern states (the former GDR). A sample of recent immigrants has been

added to the SOEP in 1994/9518.

Households in the original samples and their splitoff households are interviewed every year.

Each interview contains a household questionnaire, which collects information on the resi-

dence, number of family members and family income, for example. Additionally, all house-

hold members over age 16 report individual information on standard demographic character-

istics, labor market activity, sources of income, time devoted to various activities, and so on19.

Data from the 1991 and 1994 waves of the SOEP will be examined here to explore the struc-

ture of households demanding market-procured domestic labor. The following question about

the household’s hiring of domestic workers was asked:

"Do you regularly, or occasionally, have someone come in to help with the cleaning or the

household?"

In 1991, questions concerning the employment of baby-sitters and nurses were asked addi-

tionally. However, the number of households that reported paying for one of these services is

very small and will therefore not be discussed here.

Neither the information on how many hours a domestic servant works, nor whether more than

one servant (if any) works in the household, is provided. Thus the demand for domestic help

16 Unfortunately, the number of persons who can be identified as working in a private household is - even if several waves are pooled - too small for an analysis of the labor supply side with the SOEP data. 17 For details see Projektgruppe Panel (1995) and Wagner, Burkhauser & Behringer (1993). 18 As data from this sample are not fully available for the 1994 wave of the SOEP, it is excluded from the analysis done here. 19 Information on children under age 16 are collected in the household questionnaire.

7

will be modeled as a discrete binary choice.

3.2 Descriptive Results20 The SOEP shows that in 1994 as much as 7.0% of all German households regularly employed a domestic helper, which equals 2.3 million households (1991: 6.1%; 2.0 million). An addi-tional 4.1% (1.4 million) occasionally had someone come in to help with the housework (1991: 4.5%; 1.5 million). Altogether, the share of households buying domestic labor on the market was 11.1%, i.e. 3.7 million (1991: 10.6%; 3.5 million) (see Table 3 and Table 4)21. In comparison, estimates by Mediamark Research show that in 1996 a fraction of 9% of all US households (9.4 million) employed professional housekeepers22. Looking for differences in the regional distribution of the use of domestic help shows that the western states of the Federal Republic barely differ from each other23. In East Germany how-ever, the fraction of households with domestic help was only 2.7% (1991: 3.7%). Considering this and socio-economic differences between West and East in general, only households with a West German head will be analyzed in the following24. This increases the share of house-holds hiring a servant to 13.9% (1991: 12.7%) (see Table 5 and Table 6). Especially households with a monthly net income of 5000 DM or more account for a large fraction of those households, which employ domestic servants. In 24.3% (1991: 25.3%) of all households belonging to this income group, at least some of the housework is done by a pro-fessional. The demand for waged domestic labor in low income households (up to 2000 DM monthly net income) turns out to be bigger than one might have expected. With 13.1% (1991: 10.7%), the share of employing households here is nearly as high as in the average. This last finding can be explained in part by a large number of retirees, whose pensions are high enough to afford a housekeeper and who cannot maintain their household without help25. 20 For more detailed tables, see Hank, 1998. Some information on the extend and structure of em-ployment in German private households can also be found in Hatzold, 1986. These results have to be handled with care, though (see Footnote 7). 21 Information on both, household and individuals are required here. If households are included, for which no information on the individual members is available, the number of households employing domestic help increases to 4.1 million in 1994 and 3.7 million in 1991 respectively (without changing their relative share, however). 22 Other estimates – using broader definitions - however assume that in 1994 as much as 17% of US householders employed servants. For both see Dortch, 1996. 23 Unfortunately it was not possible to disaggregate the data and get information on possible differ-ences depending on the community size. 24 This means that only sample A of the SOEP will be used. Foreigners (sample B) and East Germans (sample C) are dropped. 25 47.8% (1991: 37.3%) of all households, in which a person needing care lives, hire domestic help.

8

The large fraction of single-person-households is another indicator for this assumption. As 20.0% (1991: 19.2%) of these households hire domestic help, it is worth taking a closer look at this group. Single households are defined here as households inhabited by one person or a single parent with his/her child(ren). Using this definition, 18.7% of all West German single households employed a domestic helper in 1994 (1991: 17.6%) (see Table 5 and Table 6). Most single households have a monthly net income below 3500 DM, but still a share of about 14% of these households buys domestic labor on the market. While only 11.4% (1991: 9.4%) of the working singles employ someone to help in the home, as much as 29.6% (1991: 27.3%) of the retirees26 demand domestic services. The fraction of two-person-households hiring domestic help, is 12.2% (1991: 11.7%). This share is much smaller in households with three or more members. Thus it can be assumed, that in families with children the intra-household division of labor still follows traditional patterns. This means that one adult works in the market, while the other one (usually the fe-male partner) stays home, taking care of the household and the children27. Also, households with young children are typically headed by younger, lower-paid workers, who might not af-ford paying someone to clean the place, although they might need help worse than anyone else (see Dortch, 1996). It seems plausible that especially Double Income No Kids - households (DINKs) should be frequent users of professional domestic services28. They could afford to pay the cost of a housekeeper and would be comparatively better off, if they let somebody else do the neces-sary housework and spend the saved time elsewhere. If focussing on partner households29 only, it turns out that the fraction of those which employ domestic helpers decreases to 10.1% (1991: 9.2%) (see Table 5 and Table 6). Most of the em-ploying households have a monthly net income of more than 3500 DM. Equally high shares of households hiring domestic servants, however, can be found where no household member works (1994: 11.5%; 1991: 12.5%) and where both partners perform market work (1994: These households, however, account for a very small share of the total population only. 26 A person is defined here as retired, if he/she reports not to be employed, not to be registered as unemployed and is over age 60. 27 The share of households that report having children and employing domestic help is 7.9% (1991: 7.4%), i.e. clearly below the average. 28 For a detailed sociological discussion of the influence of waged domestic labor on the domestic division of labor within dual career households, see Gregson & Lowe, 1994. The authors analyze qualitative data from Great Britain. 29 The category ”partner households” excludes multigenerational households, i.e. households with

9

11.1%; 1991: 11.2%). This means that in partner households, too, retirees account for much of the observed demand for professional household help. Restricting the analysis to those partner households, in which the woman is in the prime age for labor force participation (age 25 to 55), causes another drop in the fraction of households employing domestic help (1994: 8.9%; 1991: 7.7%) (see Table 5 and Table 6). Here, it is in-deed the dual career households that, with a share of 11.8% in 1994 (1991: 11.1%), use do-mestic services the most. This indicates a connection between the woman’s allocation of time and the household’s labor demand. A clear pattern of the female’s hours worked at home and on the market and the fraction of households employing a housekeeper supports an assump-tion like this. Only 7.0% (1991: 5.9%) of the households in which the woman spends more than 3 hours per day on housework employ a domestic helper. On the other hand, as much as 17.7% (1991: 19.0%) of the households where the woman’s weekly hours of market work exceed 40 hours use market-procured domestic labor. In a crossectional analysis, the use of domestic help in West German households appears to be quite stable, if different subpopulations, income groups, etc. are compared in 1991 and 1994. A longitudinal analysis however shows that it is not necessarily the same households which use professional housekeepers (see Table 7)30. 25.4% of all households which employed do-mestic help in 1991 did not do so anymore in 1994. On the other hand, as much as 35.5% of those which demanded domestic services in 1994 were new users, i.e. households that did not hire in 1991. A total of 1.9 million households (8.8%) in the western states of Germany em-ployed somebody in both years. In the following, a theoretical framework for an analysis of the connection between the household’s labor demand and the female partner’s labor supply will be developed (Section 4) and tested afterwards (Section 5).

more than one adult couple. Children however are included. 30 Unfortunately the sample of households which change their demand for domestic help is too small for an analysis of their socio-economic characteristics.

10

4. The Theoretical Model

In this paper, the household’s demand for domestic labor shall not be analyzed on basis of the

neoclassical production theory, but based on the neoclassical household theory, which is also

used for the analysis of labor supply.

The core of the model developed here will be in the tradition of ”the theory of the allocation

of time” models of Becker and Gronau (see Becker, 1965, Gronau, 1977), though it is ex-

tended to allow for the hiring of domestic help.

In traditional neoclassical models, labor supply decisions are the result of utility maximization

subject to constraints31.

In the simple labor-leisure-model, an individual’s utility is assumed to be depending on the

amount of market goods and services C (purchased with labor earnings and nonlabor income)

and hours of leisure L that are consumed per period32. If the individual looses successively

equal amounts of L, he would require successively larger amounts of C to maintain the same

level of utility (law of negatively sloped demand curves). In maximizing his utility function

(1) U = U(C,L) ,

the individual faces two constraints33. First, the budget constraint

(2) PC = WM + V ,

where P is the price of a unit of C, W is the fixed wage per hour of work, M is the number of

work hours and V is income derived from sources unrelated to work. Second, the time con-

straint

(3) T = M + L ,

where T is the total amount of time available, which can be allocated to M hours of market

work and L hours of leisure time.

Both constraints can be summarized by a budget line. Its slope equals W / P, the real wage. If

the individual decides to allocate all his time to leisure, the amount of C that can be consumed

equals V / P, the real property income. In the case that all available time is devoted to work,

the full income, i.e. (W / P)T + (V / P), can be spent on the consumption of C.

To maximize his utility, the individual spends his full income on leisure and on consumer

goods, i.e.,

(4) WT + V = WL + PC. 31 For a more detailed discussion of the simple static model of labor supply as it is given here, see Killingsworth, 1983, Chapter 1; Ehrenberg & Smith, 1996, Chapter 6. 32 This formulation implies that work itself does not carry any utility. 33 Constraints beyond the individual’s choice - like discrimination - are not denied, but will not be

11

The left-hand side expresses the maximum income attainable, if all time is allocated to work,

while the right-hand side represents the expenditures on leisure and consumer goods.

The optimal (C,L) combination is the one lying on the highest possible indifference curve34

consistent with the requirement that the individual remains on or below the budget line.

According to the approach of Becker (1965), however, goods and time do not yield utility

themselves, but are better regarded as inputs to the production of various commodities. It is

these commodities that are ultimately consumed and which are the direct source of utility.

They are produced via household production functions

(5) Z= Z(L,C) ,

where Z is a commodity (or activity), L the necessary input of leisure (or nonmarket time) and

C the amount of consumer goods devoted to the production of the activity35.

Thus, in contrast to the simple model, nonmarket time cannot be regarded as leisure only, but

as a different kind of work, performed at home rather than for an employer (three-way alloca-

tion of time). The new time constraint that the individual faces now, can be written as

(6) T = M + L + H ,

where H represents the hours of work in the home36.

As Gronau (1977) points out, goods and services can be purchased on the market or can be

produced at home. They are assumed to be perfect substitutes for each other37, which leaves Z

unaffected by the composition of C38. The value of home goods will be measured in terms of

their market equivalents39.

The production of home goods or services, according to the function

(7) CH = f(H) ,

discussed here. 34 The individual's preferences can be symbolized by indifference curves, which show different com-binations of C and L, each of them giving the individual the same level of utility. At any given (C,L) combination, the slope of the curve equals the negative of the marginal rate of substitution of con-sumer goods for leisure. 35 It takes food and energy combined with preparation time to produce meals, for example. 36 For more on the extension of the labor-leisure-model to a three-way allocation of time to market work, home work and leisure, see Kaufman, 1991, Chapter 3. - Gronau (1997) summarizes recent work on the theory of home production and provides an extensive bibliography. 37 This is of course not true, if the individual attaches intrinsic utility (or disutility) to the time spent on housework, e.g. when it comes to child care (see Pollak & Wachter, 1975). As the focus here, however, lies on domestic work in the sense of cleaning, washing, etc., the assumption of perfect sub-stitution between market and home production seems to be justified. 38 Assumptions that Z is influenced by the composition of C (like Z = Z(CM, CH, L)), turned out to have only very limited predictive power (see Gronau, 1973). 39 The cost of time used for the production of home goods equals the wage rate, if the individual par-ticipates in the labor force. As the value of time for nonworkers exceeds their potential wage rate, it has to be measured by a shadow price, equal to the marginal product of time in the household sector. For more on the value of the housewife’s time, see Gronau, 1973.

12

is, however, characterized by decreasing returns to effort. This means that, if the capital stock

and technology of production in the household are assumed constant, additional hours of

homework increase the amount of home goods produced, but at a diminishing rate (i.e., the

factor input H is subject to declining marginal productivity). This cannot be explained by fa-

tigue or changes in input proportions only. It is also due to a change in the composition of CH:

as H increases, it changes towards activities that have a cheaper market substitute.

Adopting the assumption that there is only one consumption good X40, one can roughly dis-

tinguish the following situations: For low desired values of X, household production alone is

sufficient and the individual does not participate in the paid labor market. At some point,

home production becomes inefficient in the sense that an hour devoted to work at home pro-

duces less X than could be purchased with the income from one hour's work in the market.

Given sufficient demand for X, the individual decides to allocate some time to home produc-

tion and the remainder of work time to the market, where a constant hourly wage is paid41.

While in the neoclassical theory ”household” is regarded as synonymous with ”individual”,

various empirical findings (see Juster & Stafford, 1991, for example) show that the members

of a family take different roles in the production of utility: traditionally husbands specialize in

work in the market, while the wife specializes in the production of home goods42. Even

households in which both spouses are employed, show large disparities between husbands and

wives regarding the time they spend on housework (see Hersch & Stratton, 1994; Beblo,

1998).

Following the male chauvinist model of labor supply of family members43, it is assumed here

that the husband decides on his labor supply without reference to his wife’s labor supply deci-

sions, i.e. solely on the basis of his own wage and the family’s actual property income.

Therefore, the husband’s allocation of time as well as his earnings can be regarded as fixed

parameters of the wife’s labor supply decision44.

At this point two new features shall be added to the model: a minimum amount of housework 40 According to Hicks’ composite commodity theorem, it makes no difference for the analysis of labor supply, how many consumer goods there are, as long as the prices of these goods stay in the same relation to each other (see Killingsworth, 1983). If necessary, the model can, however, easily be extended to the case of two commodities. Then, a welfare function U(Z1, Z2) would be maximized subject to the constraints imposed by the transforma-tion curve between the two commodities (see Gronau, 1977). 41 When it comes to the initial decision to enter the labor force, it should not be forgotten though, that there are some fixed (time and money) costs related to work in the market. 42 For theoretical remarks on the intra-household division of labor, see Becker, 1981, Chapter 2, for example. 43 For details on this and other models of the relation between household membership and labor sup-ply, see Killingsworth, 1983, Chapter 2.1, Ehrenberg & Smith, 1996, Chapter 7. 44 In effect, the husband becomes a kind of income-producing asset.

13

that has to be done to maintain the household (Xmin), and the availability of market-purchased

labor to perform this work.

Xmin depends on household characteristics, like the size of the household, size of the dwelling,

age and number of children, etc. It can be produced by the male or female partner or it can be

bought on the market (by employing domestic help). If the husband’s labor supply for the

market is regarded as fixed, the same can be assumed to be true for his share of work in the

household sector. Therefore, the remaining time needed for the production of Xmin has to be

done either by the wife or by hired domestic help. This makes market-procured domestic la-

bor an explicit component of the female’s time constraint, which is relevant for her labor sup-

ply decision.

Suen (1994) develops a model, in which all housework is done by a domestic servant (if a

servant is hired at all). This model however ignores that home production is the most efficient

way of producing X, as long as the marginal productivity in the household sector exceeds the

productivity of the individual in the market45.

No hiring and no change in the allocation of time to household and market work occur, if the

minimum amount of necessary production is less than the efficient amount. This means that at

any given point in the production of Xmin the marginal productivity at home is higher than the

woman’s wage rate.

If Xmin is greater than the quantity voluntarily produced and if the wage of the hired worker is

greater than the householder's wage, then the householder will devote more of her time to

home production and less time to the market.

In the final case, where Xmin exceeds the efficient amount produced at home and the house-

holder's wage is greater than the wage of a domestic worker, then hiring occurs and the

householder's hours of home work remain unchanged, since her efficiency in this type of

production has not been altered. However, the direction of the change in the number of hours

she devotes to the paid labor market is ambiguous.

Thus, the householder makes several decisions: how many hours to spend in home work, in

market work, and whether to hire a domestic worker. This set of decisions depends on the

market wage opportunities of the householder, the cost of hiring a domestic worker, the

minimum standard of housework, and the efficiency of the householder in home production.

45 Only if the marginal productivity of work at home falls short of the real wage rate at every point of time, there is no home production. In this case we would face the traditional labor-leisure-dichotomy again (see Kaufman, 1991).

14

5. Empirical Results

5.1 Method

The empirical strategy followed here, is to model the household’s hiring decision and the fe-

male partner’s labor supply decision simultaneously46.

In a first step, however, two separate equations will be estimated. As both decisions are re-

garded as a discrete choice with a binary dependent variable, a probit model is used for the

regression47:

(1) ∫′

∞− ⎭⎬⎫

⎩⎨⎧−=′

β

πβ ix

i dttxF2

exp21)(

2

,

where the function F is the distribution function of the standard normal distribution.

As for many nonlinear regression models, the parameters here are not the marginal effects.

For the normal distribution, the density function that corresponds to the cumulative function is

(2) [ ] ββφ )( xxyE ′=

∂∂ ,

where φ(t) is the standard normal density.

To control for a possible interdependence between the hiring and the labor supply decisions, a

bivariate system of equations will be estimated. Along with the simultaneously estimated co-

efficients for both equations, the correlation among their error terms will be calculated. In a

bivariate probit regression, these correlations can indicate the interdependence of the two de-

cisions, holding other factors fixed. The general specification for a two-equation model would

be:

(3) y1* = β1x1 + ε1, y1 = 1 if y1* > 0, 0 otherwise,

y2* = β2x2 + ε2, y2 = 1 if y2* >0 , 0 otherwise,

E[ε1] = E[ε2] = 0,

Var[ε1] = Var[ε2] = 1,

Cov[ε1, ε2] = ρ (see Greene, 1993).

46 In a more sophisticated model, it would be possible to estimate the decisions on how many hours to spend in home work, in market work and whether to hire a domestic worker, altogether. In such a model, the first two equations were tobit equations, which allow for a concentration of observations at zero hours of work. The third equation would be a probit for the dichotomous hiring outcome. This is not done here, however, and must be left to future research. 47 For details and a comparison between the probit model and other regression models for binary out-comes, see Greene, 1993, Chapter 21; Hamerle & Ronning, 1995.

15

For the analysis, the software package STATA is used.

5.2 Variable Description

The dependent variables in the regressions are:

HIRE - If the household reports to employ domestic help either regularly or occasion-

ally, HIRE=1, 0 otherwise.

PARTICIPATE - If the female partner reports to be either full-time or part-time em-

ployed or to be in vocational training or being marginally employed and not to be registered

as unemployed, PARTICIPATE=1, 0 otherwise.

These variables are assumed to depend on the female partner’s market wage opportunities, the

cost of hiring a domestic worker, the minimum standard of housework, the efficiency of the

householder in home production and her non-wage income. These factors are operationalized

as follows:

• Market wage opportunities of the householder

For women who choose not to work, the market wage rate is unobservable. The woman‘s

wage opportunities, however, can be regarded as a function of her education and age. Thus, it

would be possible to use a selectivity-bias corrected regression to construct a predicted wage

(see Heckman, 1979). Findings by Suen (1994), however, suggest that education variables can

be substituted directly into the probit regression without remarkable changes in the outcome

of the other variables’ coefficients. Instead of entering years of education (and their square)

into the regression, a set of dummy variables will be used to measure the householder’s edu-

cation by the highest schooling or vocational degree she ever received. This has two advan-

tages: 1) Repeated years of schooling may even have a negative effect on wages. 2) It is be-

lieved that particularly the German labor market rewards formal qualifications (see

Kreyenfeld, 1996, for example). The reference category for this hierarchy of degrees is having

a low or no schooling / vocational degree. Additionally the female partner’s age and its

square will be used.

UNI highest schooling / vocational degree is a university degree

VOCDEG highest schooling / vocational degree is a vocational degree

HIGH highest schooling / vocational degree is “Abitur” or “Fach-

16

hochschulreife”48

MID highest schooling / vocational degree is “Mittlere Reife”49

AGE woman’s age

AGE-SQ square of woman‘s age

• Cost of hiring a domestic worker

The data do not provide this information. However, if price information in the market for do-

mestic helpers is assumed to be reasonably good, one can treat the price of domestics as con-

stant (see Suen, 1994)50.

• Minimum standard of housework

The minimum standard of housework which is necessary to maintain the household is not

observable, but may be determined in part by the size of the residence, the number of children

in the family and whether there is a person needing care in the household. Therefore, the fol-

lowing variables are entered into the regression:

NROOMS number of rooms bigger than 6 sqm.

NKIDS3 number of children between age 0 and 3




NKIDSBIG number of children older than age 16

CARE1 person needing care in household (0/1),

who is not the female partner (1991 only)

CARE2 person needing care in household (0/1) (1994 only)

The variable CARE1 is constructed in a way that it equals 1 only, if the female partner does

not report to have suffered from a certain illness or disability for at least a year or chronically.

It is assumed that in this case it is not the woman herself, who is the person needing care in

the household. As the information on the woman’s health is not available for 1994, the vari-

able CARE2 is used in the regression for this year (where CARE2=1, if any person in the

household needs care).

48 “Abitur” usually takes 13 years of schooling, “Fachhochschulreife” 12 years of schooling. 49 “Mittlere Reife” usually takes 10 years of schooling. 50 The attempt to measure the cost of domestic helpers by calculating the hourly wage of marginally employed women in West Germany turned out to be not very useful.

17

• Efficiency of the householder in home production

This is not observable and must be dealt with in the empirical model. However, the variable

ILLNESS suffering for at least a year or chronically

from a certain illness or disability

is used, as it can be assumed, that this affects the woman‘s ability to maintain the household.

This variable is available for 1991 only51.

Finally, two variables are entered into the equations, which measure possible household in-

come which is unrelated to the female partner’s labor earnings. These are:

OWNER owner of residence52

NOLABINC woman’s monthly non-labor income

The variable NOLABINC was constructed by subtracting the woman’s net wage from the

monthly household net income. The result was divided by 100. This increases the otherwise

very small coefficient by 100. To interpret the coefficient, it has to be divided by 100 again.

5.3 Regression Results

Only West German partner households53, where the female partner is between 25 and 55

years old, will be considered. The age restriction is due to the particular interest in the relation

between the hiring and the labor supply decision. Table 10 and Table 11 display the results of

the regressions.

In the hiring equation as well as in the labor force participation equation, the education dum-

mies show the expected signs. As this set of variables represents a hierarchy of degrees, it is

not surprising that especially having a university degree has a strong and highly significant

impact on both decisions. The probability of employing a domestic servant is 21 percentage

51 Another question in the SOEP asks, whether a person is officially registered as having a reduced capacity for work or being severely disabled. Since this question is directly related to the woman‘s ability to work, it was decided not to use it for the probit estimation. 52 The variable OWNER could be influenced by former female labor force participation, which would make the woman more likely to work on the paid labor market today. However, as this variable can be regarded as another indicator for the amount of housework that has to be done, it was decided to keep it in the equation. 53 Regressions were run, using the whole SOEP sample. When it was controlled for household type (SINGLE / PARTNER) and subsample (WEST / EAST / FOREIGN), the coefficients turned out to be significant with the expected signs. It was decided, however, to drop single households, East German and foreign headed households to get a rather homogenous sample, which appears to be more appro-priate for the theoretical framework applied here.

18

points higher for households, where the female partner received a university degree, than in

the case of having no or a low vocational / schooling degree (the result is the same for 1991

and 1994). Other degrees, however, turn out to be insignificant for the hiring decision, while

having a vocational degree or “Mittlere Reife” does have a positive effect on the woman’s

labor supply decision. The influence of having a university degree on a woman’s probability

to join the labor force is very high in 1994, where her probability to work in the market is 23

percentage points higher compared to a woman in the reference category. In 1991 however,

the effect of a university degree is not higher than that of having “Mittlere Reife” only (14

percentage points).

Growing older raises the probability for hiring domestic help by about 1.5 percentage points

for each additional year (the marginal effect is about the same in 1991 and 1994). The age

function has a concave shape, peaking at 58 (1991: 45). This means that the maximum prob-

ability to employ a servant would be reached, if the female partner was 58 years old (45 years,

respectively). The peak age for joining the labor force is 32 years (1991: 27). As young work-

ers tend to be lower-paid and middle agers usually enjoy the highest earnings of their life-

time54, the time gap between both peaks could be expected.

Most of the variables used to estimate the minimum standard of housework come out with the

expected signs in both equations and are statistically significant. The number of rooms posi-

tively affects the household’s hiring decision (the marginal effect in 1991 is, however, ex-

tremely small) as well as the existence of young children in the household. In 1994, additional

children up to the age of 16 raise the probability of employing help between 3 and 4 percent-

age points each (here, too, the effects are weaker in 1991). If the children become older,

however, the coefficients get a negative sign: each child in the household, being older than 16

years, lowers the probability to hire by about 3 percentage points (in both years). Turning to

the equation for PARTICIPATE, in 1991 and 1994, a decreasing negative influence of chil-

dren on the female partner’s labor force participation can be observed, as the children age55.

The outcome of the CARE variables56 used in the regressions differs a lot between both years.

While CARE2 (in 1994) is insignificant for the hiring decision, but shows the expected strong

and negative impact on the labor supply decision, the opposite is true for CARE1 (in 1991). A

person needing care in the household, who is not the female partner herself, raises the prob-

ability to hire domestic help by 20 percentage points, but is not significant in the participation 54 For details on life-cycle aspects of labor supply, see Ehrenberg & Smith, 1996, Chapter 7. 55 For an analysis of women’s labor force transitions triggered by child births, see Gustafsson et. al., 1996. 56 Note that multigenerational households are excluded from the analysis. This means that, if the exis-

19

equation. The effect of ILLNESS (as an indicator of the woman’s health and her capability to

maintain the household - in 1991 only) has the same direction: a small, but positively signifi-

cant impact on the household’s hiring decision, but no significance for the woman’s labor

supply.

Being owner of the residence is insignificant for the decision to hire domestic help as well as

for the female partner’s labor force participation decision in 1994. In 1991, however, it turns

out to be at least significant for the hiring decision: compared to households which do not

own their residence, the probability to employ somebody is raised by 4 percentage points.

The woman’s non-labor income shows the expected signs in both equations, although the ef-

fect is very small. For a 1 percentage point increase in the probability to hire, the monthly

non-labor income would have to increase by 1000 DM in 1994, by roughly 700 DM in 1991.

The impact on the female’s labor supply is stronger, but still small: a growth of her non-wage

income of 300 DM (1991: 200 DM) lowers the probability to join the labor force by 1 per-

centage point.

Turning to the bivariate probit regression, we find that the coefficients barely differ from

those discussed above, when the household’s hiring decision and the woman’s labor supply

decision are estimated simultaneously. The variable RHO denotes the correlation coefficient

between the error terms of the bivariate probit regression. The correlation coefficient is esti-

mated to be 0.30 (1991: 0.34) and statistically significant. This indicates a complementary

relation between the household’s labor demand and the female partner’s time allocation deci-

sion.

tence of a person needing care is reported, this person is not one of the couple’s parents.

20

6. Concluding Remarks

The bivariate probit model in this paper estimates that the presence of a domestic helper in the

family and the female partner’s propensity to engage in market work are closely connected57.

It is necessary, however, to keep a proper perspective on the importance of market-procured

housework for female labor supply as a whole (and vice versa).

In the analyzed sample, the fraction of households employing domestic help in 1994, was

only 8.1% (1991: 7.2%). Comparing this with the 61.8% (1991: 60.8%) of working women,

shows that the effect of the household’s labor demand on the overall labor force participation

rate need not be large.

Conversely, the total positive effect of women’s labor force participation on the use of do-

mestic help can be regarded as extremely small. A major share of households employing do-

mestic servants does not participate in market work at all. While in partner households the

share of those employing a servant is equally high, whether none or both of the partners work

(about 11.5% in 1991 and 1994), almost every third retired single hires a housekeeper. This is

a share about three times higher than that of singles participating in the labor market.

Considering this, it has to be questioned, if the legislated tax policies are designed appropri-

ately. As retirees usually do not pay taxes, they do not benefit from the possibility to deduct

the cost of a domestic worker from their income subject to taxation. Those households, on the

other hand, which could benefit the most from today’s regulations, do not seem to have a very

high demand for domestic help. If there was an actual need, sufficient to cause a significant

increase in the number of regular jobs in private households, it can be assumed that they

would already buy domestic work on the informal labor market more often than they do.

57 Suen (1994) gets a similar result using data from Hong Kong.

21

7. Appendix Table 1: Workers not Liable to Social Insurance in Germany 1987 - 1992

1987 (West Germany)

1992 (West Germany)

1992 (East Germany)

Total

2,284,000

2,616,000

363,000

- in private households

570,000 (25%)

732,000 (28%)

29,000 (8%)

Source: ISG, 1993 Table 2: Marginal Second Job Holders in Germany 1987 - 1992

1987 (West Germany)

1992 (West Germany)

1992 (East Germany)

Total

539,000

1,217,000

257,000

- in private households

97,000 (18%)

256,000 (21%)

29,000 (11%)

Source: ISG, 1993 Table 3: Employment of Domestic Help in German Private Households 1991

Fraction of households employing domestic help

Number of households employing domestic help

(million) 58 YES

10.6%

3.5

- regularly

6.1%

2.0

- occasionally

4.5%

1.5

NO

89.4%

29.8

Source: German Socio-Economic Panel, wave 8 (1991), weighted by HHHRF

58 If households are included in the analysis, for which no information on the individual members is available, the number of households employing domestic help increases to 3.7 million (without changing their relative share, however).

22

Table 4: Employment of Domestic Help in German Private Households 1994 Fraction of households

employing domestic help Number of households

employing domestic help (million) 59

Yes

11.1%

3.7

- regularly

7.0%

2.3

- occasionally

4.1%

1.4

No

88.9%

29.4

Source: German Socio-Economic Panel, wave 11 (1994), weighted by KHHRF Table 5: Employment of Domestic Help in Different Subgroups 1991 Fraction of households


Employing domestic help (million)

All households (n = 6348) *

10.6%

3.5

West German households (n = 3392) **

12.7%

3.2

West German single households (n = 1083) ***

17.6%

1.9

West German partner house-holds (n = 2181) ****

9.2%

1.3

West German partner house-holds (woman 25-55 years old) (n = 1496)

7.7%

0.7

Source: German Socio-Economic Panel, wave 8 (1991), weighted by HHHRF * Information on both, household and individuals required. ** Only sample A of the GSOEP is used. Foreigners (sample B) and East Germans (sample C) are dropped. *** Includes single parents and their child(ren). **** This does not include multigenerational households, i.e. households with more than one adult couple.

59 If households are included in the analysis, for which no information on the individual members is available, the number of households employing domestic help increases to 4.1 million (without changing their relative share, however).

23

Table 6: Employment of Domestic Help in Different Subgroups 1994 Fraction of households


Employing domestic help (million)

All households (n = 6055) *

11.1%

3.7

West German households (n = 3226) **

13.9%

3.4

West German single households (n = 1067) ***

18.7%

2.1

West German partner house-holds (n = 2032) ****

10.1%

1.3

West German partner house-holds (woman 25-55 years old) (n = 1386)

8.9%

0.7

Source: German Socio-Economic Panel, wave 11 (1994), weighted by KHHRF * Information on both, household and individuals required. ** Only sample A of the SOEP is used. Foreigners (sample B) and East Germans (sample C) are dropped. *** Includes single parents and their child(ren). **** This does not include multigenerational households, i.e. households with more than one adult couple. Table 7: West German Households - Employment of Domestic Help 1991 - 1994 (longitudi-nal) (n=2700) Did households which employed in 1991 ...

... still do so in 1994?

YES60 NO

74.6%

25.4%

Did households which employed in 1994 ...

... already do so in 1991?

YES NO

64.5%

35.5%

Source: German Socio-Economic Panel, wave 8 (1991) and wave 11 (1994), weighted by HKHHRF (longitudinal weight)

60 The share of West German households employing domestic help in both years is 8.8% (1.9 mil-lion).

24

Table 8: West German Partner Households (Woman 25-55 Years Old) – Descriptive Sample Statistics (unweighted) 1991 Variable61 Mean Standard Deviation62 HIRE

.07

-

PARTICIPATE

.61

-

UNI

.08

-

VOCDEG

.70

-

HIGH

.01

-

MID

.03

-

AGE

39.00

8.81

AGE-SQ

1598.39

703.66

NROOMS

4.18

1.94

NKIDS3

.20

.47

NKIDS6

.18

.44

NKIDS12

.34

.63

NKIDS16

.18

.42

NKIDSBIG

1.29

1.07

CARE1

.01

-

ILLNESS

.24

-

OWNER

.53

-

NOLABINC

34.93

17.27

Source: German Socio-Economic Panel, wave 9 (1991)

61 For a detailed variable description see section 5.2 of this paper. 62 For binary variables standard deviations are not displayed.

25

Table 9: West German Partner Households (Woman 25-55 Years Old) - Descriptive Sample Statistics (unweighted) 1994 Variable63 Mean Standard Deviation64 HIRE

.08

-

PARTICIPATE

.62

-

UNI

.09

-

VOCDEG

.71

-

HIGH

.02

-

MID

.03

-

AGE

37.34

8.92

AGE-SQ

1480.38

702.12

NROOMS

4.07

1.50

NKIDS3

.22

.46

NKIDS6

.20

.45

NKIDS12

.34

.63

NKIDS16

.21

.47

NKIDSBIG

1.24

1.08

CARE2

.01

-

OWNER

.52

-

NOLABINC

38.38

19.92


63 For a detailed variable description see section 5.2 of this paper. 64 For binary variables standard deviations are not displayed.

26

Table 10: West German Partner Households (Woman 25-55 Years Old) - Probit Estimates for HIRE and PARTICIPATE 1991

PROBITS

marginal effect probit coefficient

probability

BIVARIATE PROBIT65 -

probit coefficient probability

Variable HIRE PARTICIPATE HIRE PARTICIPATE UNI

.2079 1.1672 .000

.1420 .4001 .015

1.1860 .000

.3963 .016

VOCDEG

.0203 .2789 .131

.0751 .1953 .036

.2882 .113

.1938 .037

HIGH

*

- .0040 - .0104

.975

- 3.9243 .994

- .0113 .973

MID

.2066 .2659 .519

.1450 .4134 .062

.2614 .525

.4156 .061

AGE

.0159 .1982 .029

.0231 .0606 .203

.1735 .051

.0621 .191

AGE-SQ

- .0002 - .0022

.044

- .0004 - .0011

.058

- .0019 .078

- .0011 .054

NROOMS

.0033 .0411 .050

.0056 .0146 .531

.0419 .043

.0115 .587

NKIDS3

.0195 .2428 .143

- .4384 - 1.1506

.000

.1838 .261

- 1.1493 .000

NKIDS6

.0281 .3499 .031

- .1629 - .4275

.000

.3208 .046

- .4177 .000

* as HIGH = 0 predicts the failure perfectly, STATA dropped the 19 observations of this category

65 Note that STATA does not provide marginal effects for bivariate probit regressions.

27

Table 10: West German Partner Households (Woman 25-55 Years Old) - Probit Estimates for HIRE and PARTICIPATE 1991 (continued)

PROBITS


probability

BIVARIATE PROBIT -


Variable HIRE PARTICIPATE HIRE PARTICIPATE NKIDS12

.0316 .3943 .001

- .0941 - .2470

.001

.3883 .001

- .2476 .001

NKIDS16

.0093 .1162 .470

- .0431 - .1131

.256

.0830 .605

- .1084 .275

NKIDBIG

- .0298 - .3716

.000

.0048 .0125 .832

- .3443 .000

.0130 .825

CARE1

.1973 1.0724 .009

- .1929 - .4893

.183

1.0594 .010

- .4921 .182

ILLNESS

.0193 .2176 .093

- .0101 - .0266

.752

.2270 .077

- .0256 .761

OWNER

.0390 .4882 .000

.0246 .0646 .423

.4697 .001

.0699 .383

NOLABINC

.0015 .0193 .000

- .0054 - .0141

.000

.0184 .000

- .0141 .000

RHO

_

_

.3397 .000

N

1477

1496

1496

P(SAMPLE)

.0724

.6076

_


28

Table 11: West German Partner Households (Woman 25-55 Years Old) - Probit Estimates for HIRE and PARTICIPATE 1994

PROBITS


probability

BIVARIATE PROBIT66 -


Variable HIRE PARTICIPATE HIRE PARTICIPATE UNI

.2143 1.1098 .000

.2285 .7091 .000

1.1089 .000

.7100 .000

VOCDEG

.0117 .1261 .493

.1080 .2826 .007

.1253 .493

.2721 .010

HIGH

*

-.1985 -.5051 .120

-4.1537 .991

-.5168 .111

MID

.0197 .1758 .653

.1308 .3767 .097

.1689 .665

.3782 .096

AGE

.0119 .1216 .143

.0537 .1426 .004

.1122 .172

.1430 .004

AGE-SQ

-.0001 -.0010 .294

-.0008 -.0022 .000

-.0009 .338

-.0022 .000

NROOMS

.0139 .1426 .000

.0251 .0667 .027

.1436 .000

.0683 .023

NKIDS3

.0458 .4699 .005

-.4956 -1.3172

.000

.4342 .000

-1.3120 .000

NKIDS6

.0445 .4560 .003

-.1780 -.4730 .000

.4448 .009

-.4684 .000

* as HIGH = 0 predicts the failure perfectly, STATA dropped the 20 observations of this category

66 Note that STATA does not provide marginal effects for bivariate probit regressions.

29

Table 11: West German Partner Households (Woman 25-55 Years Old) - Probit Estimates for HIRE and PARTICIPATE 1994 (continued)

PROBITS


probability

BIVARIATE PROBIT -


Variable HIRE PARTICIPATE HIRE PARTICIPATE NKIDS12

.0315 .3230 .008

-.1627 -.4324 .000

.3048 .012

-.4337 .000

NKIDS16

.0300 .3078 .027

-.0913 -.2426 .014

.2975 .031

-.2412 .014

NKIDBIG

-.0356 -.3652 .000

.0053 .0141 .832

-.3510 .000

.0115 .862

CARE2

.0247 .2135 .630

-.2950 -.7568 .014

.2079 .632

-.7435 .014

OWNER

.0147 .1525 .257

.0221 .0588 .495

.1572 .241

.0574 .506

NOLABINC

.0010 .0107 .000

-.0035 -.0094 .000

.0100 .000

-.0091 .000

RHO

-

-

.2976 .000

N

1365

1385

1385

P(SAMPLE)

.0820

.6180

-


30

8. References

Beblo, M. (1998). Intrafamily Time Allocation: A Panel-Econometric Analysis of the Divi-sion of Housework Time among German Couples. Paper presented at the Third SOEP User’s Conference, Berlin, July 1-3, 1998. Becker, G.S. (1965). A Theory of the Allocation of Time. Economic Journal, v75 (Sept. 1965), 493-517. Becker, G.S. (1981). A Treatise on the Family. Cambridge. Bundesministerium für Arbeit und Sozialordnung (BMAS) (1996). Arbeitsplatz Haushalt. Bonn. Bundesministerium für Arbeit und Sozialordnung (BMAS) (1997). Hilfe im Haushalt. Bonn. Bungart, J. (1997). Stellungnahme des Bundesinnungsverbandes des Gebäudereini-ger-Hand-werks. Die Krankenverversicherung, Dec. 1997, 335. Daniels, A. (1998). Im Labyrinth. DIE ZEIT, n48, November 19, 1998, 26. Dortch, S. (1996). Maids clean up. American Demographics, v18, n11, 4-8. Drohsel, P. (1996). Hausarbeit als Erwerbsarbeit. Aus Politik und Zeitgeschichte, B3-4/96, 40-46. The Economist (1996). At your service. Dec.14, 1996, 56. Ehrenberg, R.G. & Smith, R.S. (1996). Modern Labor Economics. Reading (and others). Emmerich, K. (1997a). Wann rechnet sich die Haushaltshilfe? IAB Kurzbericht, 4/97, 1-4. Emmerich, K. (1997b). Förderung der Beschäftigung in privaten Haushalten durch Dienst- leistungsagenturen und Dienstleistungsgutscheine. Arbeit und Sozialpolitik, 7-8/97, 52-55. Greene, W.H. (1993). Econometric Analysis. New York. Gregson, N. & Lowe, M. (1994). Waged Domestic Labor and the Renegotiation of the Do-mestic Division of Labor within Dual Career Households. Sociology, v28, n1, 55-78. Gronau, R. (1973). The Intrafamily Allocation of Time. The Value of the Housewives' Time. American Economic Review, v63, n4, 634-651. Gronau, R. (1977). Leisure, Home Production and Work - the Theory of the Allocation of Time Revisited. Journal of Political Economy, v85, n6, 1099-1123. Gronau, R. (1997). The Theory of Home Production: The Past Ten Years. Journal of Labor Economics, v15, n2, 197-205. Gustafsson, S., Wetzels, C., Vlasblom, J.D. & Dex, S. (1996). Women’s Labor Force Transi-

31

tions in Connection with Child Birth. Paper presented at the Second SOEP User’s Conference, Potsdam, July 10-12, 1996. Hamerle, A. & Ronning, G. (1995). Panel Analysis for Qualitative Variables. In: G. Arminger, C.C. Clogg & M.E. Scobel (Eds.), Handbook of Statistical Modeling for the Social and Beha- vioral Sciences (pp.401-452). New York; London. Handelsblatt (1996). SPD will Beschäftigung in den Privathaushalten fördern. May 30, 1996. Hank, K. (1998). Household Labor Demand and Household Labor Supply. University of Bo-chum. Hanke, T. & Rudzio, K. (1996). Was tun mit den illegal beschäftigten Haushaltshilfen? DIE ZEIT, n25, June 14, 1996, 25. Hatzold, O. (1986). Private Haushalte als Arbeitgeber. ifo-Studien zur Arbeitsmarktforschung 4. München. Hatzold, O. (1988). Der Privathaushalt als Arbeitsplatz. ifo-Schnelldienst, 12/88, 20-27. Heckman, J.J. (1979). Sample selection bias as a specification error. Econometrica, v47 (Jan. 1979), 153-162. Hersch, J. & Stratton, L.S. (1994). Housework, Wages, and the Division of Housework Time for Employed Spouses. AEA, v84, n2, 120-125. Institut für Sozialforschung und Gesellschaftspolitik (ISG) (1993). Sozialversicherungsfreie Beschäftigung - Untersuchung im Auftrag des Bundesministers für Arbeit und Sozialordnung. Sozialforschung 181a. Köln. Juster, F.T. & Stafford, F.P. (1991). The Allocation of Time: Empirical Findings, Behavioral Models, and Problems of Measurement. Journal of Economic Literature, v29 (June ‘91), 471-522. Kaufman, B.E. (1991): The Economics of Labor Markets, Chicago. Killingsworth, M.A. (1983). Labor supply. Cambridge. Klenner, C. & Stolz-Willig, B. (1996). Arbeitsplatz Haushalt, In: G. Pohl & C. Schäfer (Eds.), Niedriglöhne (pp.194-213). Hamburg. Kreyenfeld, M. (1996): Assimilation of Ethnic Germans and East Germans into the West Ger-man Labor Market. University of Bochum. Mundorf, H. (1996). Steuersystematik schafft Gerechtigkeit. HANDELSBLATT, May 31, 1996. Muntz, S. (1996). Beschäftigungspotentiale im Bereich privater Haushalte. ifo-Schnelldienst, 17-18/96, 38-45. Pollak, R.A. & Wachter, M.L. (1975). The Relevance of the Household Production Function and Its Implications for the Allocation of Time. Journal of Political Economy, v83, n2,

32

255-277. Projektgruppe Panel (1995). Das Sozio-oekonomische Panel (SOEP) im Jahre 1994. Viertel-jahrshefte zur Wirtschaftsforschung, v64, n1, 5-15. Schupp, J., Schwarze, J. & Wagner, G. (1995). Zur Expansion der versicherungsfreien Er-werbstätigkeit in Deutschland. DIW-Wochenbericht, 50/95, 857-862. Schwarze, J. (1992). Geringfügige Beschäftigung in der Erwerbsstatistik. Mitteilungen aus der Arbeitsmarkt und Berufsforschung, 4/92, 534-543. Schwarze, J. (1993). Marktwirtschaftliche Möglichkeiten zur Begrenzung der geringfügigen Beschäftigung - zwei politikfähige Vorschläge. Sozialer Fortschritt, v42, n2, 42-48. Schwarze, J. (1997). Nebenerwerbstätigkeit in Deutschland nimmt zu. DIW-Wochenbericht, 22/97, 406-412. Suen, W. (1994). Market-procured housework: The demand for domestic servants and female labor supply. Labour Economics, n1, 289-302. Verband deutscher Rentenversicherungsträger (VDR) (1997). Förderung der sozialversi-che-rungspflichtigen Beschäftigung in Privathaushalten. VDR Info, 1/97, 15-17. Wagner, G., Burkhauser, R.V. & Behringer, F. (1993). The English Language Public Use File of The German Socio-Economic Panel. Journal of Human Resources, v28, n2, 429-433. Weinkopf, C. (1996). Ansätze zur Professionalisierung von Dienstleistungen – Vor-tragsma-nuskript zur Anhörung der SPD-Fraktion im niedersächsischen Landtag "Di-enst-leistungen in Privathaushalten: Neue Wege zu mehr sozial abgesicherten Arbeitsplätzen" am 07.November 1996 in Hannover. http://iat-info/iatge.de/am/dynamdok/weinkopf/hannover/startdok/1.html. Weinkopf, C. (1997a). Stellungnahme zur Anhörung "Das Potential von Arbeitsplätzen in privaten Haushalten besser nutzen" am 22.Januar 1997 im Landtag Nordrhein-Westfalen. http://iat-info.iatge.de/am/dynamdok/weinkopf/landtag/startdok/1.html. Weinkopf, C. (1997b). Dienstleistungspools - ein Ansatz zur Professionalisierung von Dienst- leistungen in Privathaushalten? http://iat-info.iatge.de/am/dynamdok/weinkopf/dienpool/ startdok/1.html. Wenke, M. (1991). Erwerbsverhalten der Frauen und Dienstleistungsnachfrage privater Haushalte, RWI-Mitteilungen, v42, 75-98.

Documents

Household Labor Demand And Household Labor Supply1