Migration Decisions Within Dual-Earner Partnerships: A Test of Bargaining Theory

MARTIN ABRAHAM University of Erlangen-Nurnberg

KATRIN AUSPURG University of Konstanz*

THOMAS HINZ University of Konstanz**

Migration Decisions Within Dual-Earner

Partnerships: A Test of Bargaining Theory

This article focuses on the problems facedby dual-earner partnerships arising from theregional coordination of the partners’ careers.It is still unknown whether the fact that couplesare less mobile than singles is caused byhomogeneous preferences within couples or bya process of balancing conflicting interests.Consequently, we analyze the potential conflictsprovoked by work-related migration incentives.We test hypotheses derived from bargainingtheory using quasi-experimental data froma factorial survey of nearly 280 Europeancouples. Our results support the bargainingapproach and confirm that asymmetrical shifts inbargaining power drive the potential for conflict.Women’s willingness to move is generally lessthan men’s, but the impact of employmentprospects differs only slightly by gender.

School of Business and Economics, University ofErlangen-Nurnberg, Findelgasse 7/9, 90402 Nuremberg,Germany.

*Department of Sociology, University of Konstanz,Universitatsstr. 10, 78457, Konstanz, Germany.

**Department of Sociology, University of Konstanz,Universitatsstr. 10, 78457, Konstanz, Germany.

This article was edited by Cheryl Buehler.

Key Words: bargaining theory, conflicting interests, dual-earner partnerships, factorial survey design, householdmove.

It is well known that in modern societies,changing employment patterns affect couples invarious ways. In particular, the increasing labormarket participation of women leads to new chal-lenges for couples living together. One of thesechallenges arises from the necessity to coordi-nate two careers in time and space. For example,the probability of migration drops significantlywhen people form a couple (McHugh, Gober,& Reid, 1990; Quigley & Weinberg, 1977).The migration literature offers a wide range ofempirical findings as well as theoretical expla-nations regarding the determinants of householdmigration (Bartel, 1982; Bielby & Bielby, 1992;Shauman & Noonan, 2007). These findings,however, focus only on actual migration, noton the underlying considerations for or againsta household move. As a consequence, there isinadequate knowledge regarding bargaining andconflicts within couples. Moreover, migrationresearch based on ‘‘classic’’ survey data is usu-ally biased toward the traditional case of men asthe main breadwinners. Women who receive ajob offer with a considerably higher salary thantheir male spouses are still rare (for Germany andthe United States, see Cooke, 2006). Therefore,it is difficult to disentangle gender and structuraleffects on migration behavior.

Looking at household migration from a theo-retical point of view, most researchers stress thenecessity of considering the conflicting interestsof the two partners. Especially in dual-earner

876 Journal of Marriage and Family 72 (August 2010): 876 – 892DOI:10.1111/j.1741-3737.2010.00736.x

Migration Decisions Within Dual-Earner Partnerships 877

couples, new job opportunities for one partnermay correspond with decreasing opportunitiesfor his or her mate. In line with this argu-ment, one of the dominant theoretical modelsfor analyzing dual-earner households is bar-gaining theory. In that framework, the partnershave to bargain over (a) the decision for oragainst migration and (b) the arrangements thatmay follow after the household move. Althoughthis theoretical model seems appropriate for theresearch problem at hand, there are astonish-ingly few empirical tests of it. One reason is thelack of paired data for couples that are necessaryto observe the diverging interests of partners.Another limitation arises from focusing only onthe results of the bargaining process. That is, the-orists have tried to derive hypotheses regardingthe shift of resources allocated to each partnerand the consequences of that shift (Bernasco& Giesen, 2000; Klasen, 1998; Lundberg &Pollack, 1996). One of the main problems ofthis approach is the difficulty of empiricallymeasuring the complete allocation of resourcesincluding immaterial resources such as love,emotional care, and recreational time.

In this article, we aim to fill these gaps byapplying factorial survey analysis, a researchdesign for combining survey research with anexperiment. We requested that couples in Ger-many and Switzerland evaluate hypotheticaldescriptions of better employment opportuni-ties in a distant location for one partner. Thisdesign gave us the opportunity to analyze towhat extent moving preferences conflict withina partnership. By varying the hypothetical situ-ations in an experimental manner, we were ableto identify factors that influence the extent ofpotential conflicts arising from differing migra-tion incentives. The aim was to complement theexisting empirical knowledge on willingness tomigrate within couples and to provide a directtest of bargaining models in family research.

THEORETICAL FRAMEWORK

Bargaining Models

In family research, bargaining models are oftenused to analyze allocation problems amonghousehold members and especially amongspouses or cohabitation partners (Bitman, Eng-land, Folbre, Sayer, & Matheson, 2003; Cooke,2006; Lundberg & Pollak, 1996). At the coreof this theoretical framework is the situation

in which two actors have to decide on theallocation of resources. These can refer to mate-rial resources, like household income, and toimmaterial resources, such as the right to makedecisions, enjoy leisure time, and avoid unpleas-ant household tasks. The result of this allocationdecision depends on the relative bargainingpower of each actor in the partnership.

This conceptual core can be found in dif-ferent theoretical frameworks such as classicalexchange theory (Blood & Wolfe, 1960), socialpsychology (Thibaut & Kelley, 1959), and ratio-nal choice theory (Manser & Brown, 1980). Thecrucial question within all these approaches isthe definition of relative power. Although thereare some differences, most researchers agreethat, to a large extent, the available alterna-tives to the current partnership determine eachactor’s bargaining power (England & Farkas,1986). The partner with better outside optionshas more power in the relationship because he orshe can credibly threaten to leave the relation-ship. These outside options are also called threatpoints. Note that the bargaining process itselfcan stay implicit, that is, without any discussionbetween the partners, but it is assumed that bothpartners are aware of the power imbalance.

Although available alternatives may includethe presence of other attractive partners(Bergstrom, 1996), in modern societies, a per-son’s economic standing is probably the mostdecisive determinant of structural bargainingpower in relationships. Labor market participa-tion and an independent source of income makeit possible for one partner to dissolve an unsatis-factory partnership. The next best alternative isoften the move not into another relationship butinto a single household. Moreover, the higher thepersons’ income, the more that easily that personshould be able to find an attractive alternativepartner. Thus, most research focused on labormarket opportunities for measuring bargainingpower in partnerships (Blau, Ferber, & Winkler,2001; Blood & Wolfe, 1960; Ott, 1992).

The Decision to Move the Household

Within this theoretical framework, householdmoves can be seen as events that are likely tobe affected by the relative bargaining power ofexchange partners, and vice versa (Lundberg &Pollak, 2003). As Mincer (1978) explained, it isunlikely that both partners will receive optimaljob offers at the same time and within the same

878 Journal of Marriage and Family

region. Consequently, a couple has to decidewhether the migration incentive for one partner(the mover) is sufficient for accepting, at leasttemporarily, a worse employment situation forthe other.

Mincer (1978) coined the term tied moverfor a partner who experiences individualdisadvantages by a move that is generallybeneficial for the household. In contrast, a tiedstayer forgoes the better job because his orher individual gains will not be sufficient tocompensate for the overall household loss. InMincer’s classical model the actors will acceptany individual loss as long as the householdutility is maximized. Thus, this model fails torecognize the conflicts that result from the clashof individual interests between the partners.

From a dynamic bargaining point of view,even a higher household utility may not besufficient for a potential tied mover to acceptthe migration. This is because, in the bargainingtheoretical framework, it is expected that theactors rationally anticipate future allocations.More concretely, the actors are assumed toconsider how a new situation affects theirrelative bargaining power and thereby theirindividual benefit of a move. Under theassumption that the bargaining power is mainlyinfluenced by the employment and incomepossibilities of the actors, by definition, a tiedmover will experience a loss of relative power.If this loss is too high, the tied mover may geteven a lower level of household resources at thenew destination than he or she had before themove, although the overall household resourcesare increased by the move (Lundberg & Pollak,2003). Thus, potential tied movers may rejectthe household move, whereas tied stayers mayuse their job offer to renegotiate the allocation ofresources at the present location. The job offercan be seen as a new outside option because,theoretically, the tied stayer has an attractivealternative to the relationship consisting of thepossibility of moving without the partner.

Bargaining models were most often used topredict the allocation of resources in a relation-ship (for a review, see Lundberg & Pollak,1996). In our opinion, however, measuringonly the results of the bargaining process (e.g.,shares of family income used for ‘‘private’’consumptions, the distribution of leisure timesand cumbersome household tasks) is not themost promising strategy to test this kind of the-ory. First, additional factors that are difficult to

control for may influence the actual allocationof resources and duties (e.g., individual pref-erences). Second, and more important, actorstry to solve these kinds of cooperation prob-lems using various mechanisms. These can bejoint investments in the relationship such aschildren, formal marriage contracts, and furthercommitments that settle mutual rights and obli-gations (Bernasco & Giesen, 2000; Ott, 1992).Relation-specific investments enhance the inter-dependence of partners by increasing the sunkcosts of the current relationship and therebypromoting further commitment that stabilizesthe (trust in the) stability of the partnership(Brines & Joyner 1999). Nonbinding agree-ments, in contrast, are considered insufficient. Inintimate relationships, partners may (and oftendo) promise that, because of love, even if thereis a shift in bargaining power, no renegotiationof resources will take place. But it is doubtfulwhether this promise is credible; it is well-knownthat love is no guarantee against conflict in mar-riage or cohabitation. These mechanisms are notcaptured when only the results of bargaining areconsidered.

The Influence of Gender

Bargaining models are, in principal, genderneutral, and this is the point that mostly arousescriticism (for the literature on migrations, seeBielby & Bielby, 1992; Jurges, 2005). Differentincome options influence the bargaining process,and of course, income opportunities differ bygender. But after controlling for those structuralfactors and for the outside options in general, thewillingness to move is expected to be no longergender dependent. A gendered perspective, incontrast, suggests that the bargaining processalso contributes to the production of gender;in other words, it leads individuals to affirmtheir categorical status of male or female (theprocess of doing gender) (see Fenstermaker,2002; West & Zimmerman, 1987). Even ifdoing gender clearly means more than ‘‘[living]up to normative conceptions of femininityor masculinity’’ (West & Zimmerman, 1987,p. 136), gender-appropriate behavior shouldtend to, at least in general, be in line withculturally approved standards. Therefore, thedecision-making process is expected to differby gender especially if the actor’s structuralpositions are incongruent with the gendernorms (Hook & Chalsani, 2008). Because these


institutionalized norms still assign the mainresponsibility for the household income to themale partner (Huppe & Cyr, 1997; Stickney &Konrad, 2007; Szinovacz & Harpster, 1993), themale’s employment options are more likely totake priority over the female’s ones in decisions.

Empirically, the relative impact of thetwo alternative explanations—bargaining versussubscription to gender norms—is still an unre-solved issue (Blackburn, 2006; Lichter, 1983;Maxwell, 1988; Mincer, 1978; Spitze, 1984).By reason of data limitations, it was difficultto decide whether gender disparities in house-hold decision making resulted from employ-ment variables that were not controlled for andtherefore corresponded with a gender-neutralbargaining process or whether they reflectednormative expectations. The classifications ofoccupations typically used were too broad tocapture the specific job and labor market charac-teristics that were unevenly distributed by gender(e.g., geographic ubiquity, length of career lad-ders) (see Shauman & Noonan, 2007). Thus,most results were in line with both theoreticalapproaches. Our experimental design providedmen and women with exactly the same employ-ment opportunities, which we explain in moredetail below. This may help provide insight intothe gender influence on migration decisions.

Hypotheses

The bargaining model allows us to deriveseveral hypotheses regarding how actors makemigration decisions in dual-earner relationships.When evaluating a migration opportunity, thepartners consider the expected changes inrelative bargaining power. If the actors anticipateshifts in bargaining power and if there is anyinterest in preserving the relationship, they willbe less willing to move if one of them will beworse off at the new destination. This leads toour first hypothesis:

H1a: Individual willingness to move will dependnot only on the changes in one’s own situation butalso on the changes in the partner’s situation.

Results in favor of the first hypothesisconstitute, of course, only weak evidence forthe bargaining model because altruism couldalso explain this phenomenon. Altruistic actorscare about the future situation of their partners;hence, they will take into account potential

disadvantages for their partners. The bargainingapproach, however, additionally suggests thatevery actor will consider his or her ownsituation more important than his or her partner’ssituation. Any consideration of one’s partner’sinterests will be in vain if the partnership failsin the future. The risk of a future separationand the interest in receiving a favorable positionin the resource allocation lead to a primaryinterest in preserving one’s own bargainingpower. Therefore, we can hypothesize that nomatter who has the better option, the changein one’s own bargaining power is a strongerpredictor of the willingness to move than is therespective change in the partner’s power.

H1b: Actors’ willingness to move is influencedmore by their own options than those of theirpartners.

The general hypothesis regarding a couple’sconflict potential is straightforward. All unex-pected options to improve the relative bargainingpower of one partner will question the negoti-ated arrangements and evoke clashing interests.The partner with the potential improvement isinterested to move, whereas the other will rejectthe move to preserve the actual power balance.We term this clashing interest conflict potentialbecause the partners may be able to actuallyresolve the conflict (see also Miller, Perlman, &Brehm, 2007).

H2: An actor’s potential increase in relativebargaining power due to a possible migration willlead to a higher conflict potential in the partnership.

Note that, to our knowledge, this is a newhypothesis derived from bargaining theory thathas yet to be tested.

As outlined above, there are many reasonscouples may differ in how they resolve the con-flict, which may result from, for example, certainpsychological characteristics of the partners. Associologists, we are more interested in structuralfactors. One type of solution arises from thestrength of each actor’s commitment to the part-nership (Nock, 1995). A credible commitmentin this sense increases the threshold for leav-ing the partnership and therefore decreases anactor’s bargaining power by reducing the utilityof outside options. In families and relationships,there are three main types of mutual commit-ment: marriage, children, and common property


(especially home ownership). We hypothesizethat all three types of commitment lead to alower level of conflict potential.

H3a: The conflict potential will be lower forcouples who (a) are married, (b) have children,or (c) hold real estate property than for coupleswithout these characteristics.

Couples also differ with respect to the timethey have had to learn about each other andthe relationship. Learning about each other’scharacteristics and preferences allows partnersto assess each other’s future behavior. Conse-quently, the joint history represents a proxy formutual commitment that is not easily captured,such as learning to participate in and appreciateeach other’s hobbies. In addition, the durationof partnership might reflect a positive selectionprocess: Partnerships in which mutual expec-tations were more often fulfilled and in whichtherefore greater trust has been established willhave a lower risk of breakdown. This grownknowledge of each other and amount of com-mitment might cause a belief that, all else beingequal, no renegotiation of resources after themove will take place.

H3b: The conflict potential declines with theduration of the partnership.

Finally, we consider the effect of gender onour research question. In the bargaining frame-work, gender should matter only to the extentthat one gender tends to have less bargainingpower because of structurally worse outsideoptions. A gendered perspective in contrastsuggests that, even when female partners havestrong incentives to move, they may favor theirmale partners’ careers to conform to normativegender ideologies.

H4: Female partners are less willing than theirmale partners to move in response to their ownlabor market options.

That is, we suppose an interaction effect ofgender and labor market options on the willing-ness to move.

METHOD

The insufficient empirical research on the bar-gaining model in the field of families and couplesis mainly caused by a lack of adequate data.

Existing survey data provided little or no infor-mation on the long-term career prospects andlabor market situations at potential destinations(Shauman & Noonan, 2007). Because of thelow numbers of moving households, it wasdifficult to observe the shifts in resources allo-cations caused by household moves. Anotherproblem is the high selectivity of migration:Persons with job-related migrations are gen-erally more career oriented and professionallysuccessful than immobile persons, so they proba-bly would have realized similar gains in incomeand relative bargaining power even without amove (Antel, 1980; DaVanzo & Hosek, 1981).Moreover, in general, only realized migrationand employment opportunities were recorded;therefore, the changes in threat points or thelosses of household gains due to forgone oppor-tunities are unknown. Conflicts and reasons forseparation are difficult to assess with ordinarysurvey data and require paired information fromboth partners that is often not provided. Anadditional problem arises from the high corre-lation between employment characteristics andgender. Observed asymmetric decisions and out-comes may have arisen only because the labormarket conditions created better returns for themale partner and not because his career concernswere fostered to a greater extent (Jacobson &Levin, 2000).

For these reasons, we adopted a researchapproach that uses a factorial survey design(also known as vignette analysis). The keyidea in this design is that the respondentsreact to hypothetical descriptions of situationsor objects (i.e., vignettes) instead of answeringsingle-item questions (see Jasso, 2006; Rossi &Anderson, 1982). By independently varying arestricted number of dimensions of the vignettesin an experimental design, the exact impactof each dimension on respondents’ judgmentsor decisions can be estimated. Therefore, themethod is able to isolate the weight of singlefactors that are often confounded in reality. Itis possible to separate the impact of genderand employment prospects by offering the samemobility incentives to male and female partners.The random allocation of vignettes leads to anoverall independence of migration incentivesand thereby overcomes the selection biases thatlimit the results of usual migration research.

In our study, we used vignettes to describevarious incentives for a household move. As aninnovation to the vignette design we interviewed


FIGURE 1. EXAMPLE OF A VIGNETTE (MALE PERSPECTIVE, OWN JOB OFFER).

Assume, … You at the new location. In the long run, the new job will provide you with no advancement opportunities. If you don’t move, commuting to the new job will take 1 ½ hours each way. It is possible to go there only by car.

You r partner’s and her income prospects are higher there compared to the local labor market.

How much would you like to take the job and move to the new location?

Not at all

1 2 3 4 5 6 7 8 9 10 11 Very much

Note: The inverse situation for the female partner in this example would read: Assume, your partner is offered a net salaryof EUR 2,400 at the new location. . . . Your own chances of finding. . .. How much would you like to move to the new location?

both partners in dual-earner couples with inversesituations. Both partners were presented with thesame set of situations, but one (ego) was giventhe role of a mover, and the other (alter) wasput in the position of a tied mover. The con-stant stimulus consisted of a job offer for egoin another geographical region that providedthe couple similar living conditions. The vary-ing dimensions simulated differently attractiveemployment prospects in this new location forthe two partners. This design allowed us to studythe reactions to changing threat points in the rela-tionship by varying the respective employmentoptions. More concretely, we systematicallyvaried (a) the increase in income and careerprospects for the mover and (b) the employ-ment and income prospects for the tied mover.By conducting computer assisted personal inter-views (CAPI), we were able to simulate theincome prospects as a percentage gain of theactual salary, which makes the situations asplausible as possible. To test our hypotheses,we assumed that ego’s increase in income andcareer prospects improves his or her bargain-ing power in the relationship, whereas alter’semployment prospects in the new destinationaffects the anticipated change in the bargainingpower of ego’s partner.

Concerning this assumption, two points arenoteworthy. First, we manipulated the threatpoints of each partner independently, so it waspossible that both actors’ outside options wouldincrease because of the move and that the relativebargaining in the partnership would not change.Thereby, our design had the advantage that altermight have profited from the move and was notrestricted to the role of the tied mover. Second,

we assumed that the employment situationsof both actors determined bargaining power.Especially if one actor is engaged in highlyspecialized paid work, whereas the other ismostly engaged in housework, this assumptionmay be problematic. To address this problem,we restricted our sample to dual-earner couples.Figure 1 gives an example of the vignettes andshows the answer scale employed in this article.The vignettes also described the commuting timeand the commuting mode (train or car) to the newdestination to make the situations as realistic aspossible. We used these variables as controls.Within a couple, both partners confronted thesame hypothetical situations. The vignettes wereidentical, except one of the partners (ego)responded as the partner who received a joboffer, whereas the other (alter) responded as thepartner who did not receive the offer.

Our dependent variables were each partner’swillingness to move and the conflict potentialin a partnership that results from the migra-tion incentive. The willingness to move scalescored from 1 = not at all to 11 = very much.The mirror-inverted design of our study allowedus to directly compare the alter’s and ego’swillingness to move for each vignette and tomeasure the conflict potential by taking the dif-ference between ego’s and alter’s willingnessto move. The resulting variable ranged from−10 = alter would very much like to move butego does not want to move at all to 10 = egowould very much like to move but alter doesnot want to move at all. A score of 0 indicatedthat both partners shared the same tendency fora move regardless of the strength of that pref-erence. We treated all dependent variables as


FIGURE 2A. WILLINGNESS TO MOVE OF EGO AND ALTER.

0

5

10

15

20

25

30

35

Not at all 2 23 4 5 6 7 8 9 10 Very much

Willingness to move: Ego

Per

cen

t

0

5

10

15

20

25

30

35

Not at all 3 4 5 6 7 8 9 10 Very much

Willingness to move: Alter

Per

cen

tFIGURE 2B. CONFLICT POTENTIAL (WILLINGNESS TO MOVE OF EGO MINUS WILLINGNESS TO MOVE OF ALTER).

0

5

10

15

20

25

Altermorewilling

to move

-8 -6 -4 -2 No 2 4 6 8 Egomorewilling

to move

Per

cent

Note: In the regression analysis, we employed the absolute values of the conflict potential.

continuous. Several alternative statistical modelsfor noncontinuous variables (e.g., ordinal regres-sions, Poisson regressions) did not yield differentresults; therefore, we used the more straightfor-ward interpretation of an ordinary least squares(OLS) model. Because each person evaluatedup to 10 different vignettes, the models had tobe corrected for correlated observations, as isthe case with any repeated measures in within-subject designs (Hox, Kreft, & Hermkens, 1991).We applied random intercept models (Rauden-bush & Bryk, 2001; Snijders & Bosker, 1999)that account for dependent observations byestimating one joint random intercept for allobservations from one single respondent or cou-ple. The characteristics of the vignettes variedon Level 1, and either individual or couple char-acteristics were modeled as Level 2 variables.

The used sample of 200 vignettes wasa fraction of all possible combinations ofthe six variable dimensions (maximization of

the D-efficiency with orthogonalization of allmain effects and first-order interactions in aso-called Resolution III, according to Kuhfeld,2005; Kuhfeld, Randall, & Garratt, 1994). Tenvignettes were presented to each person; that is,we used 20 different decks of vignettes and 80different questionnaires (20 decks × 2 versionsfor mover vs. tied mover × 2 versions for menvs. women).

All interviews were conducted with couplesliving in Germany and Switzerland. Both coun-tries represent conservative welfare regimes withgendered labor market participation rates, highoccupational gender segregations, and genderpay gaps. For instance, in 2007 in Germany,only 64% of women between 15 and 64 years ofage were employed compared to 75% of men;the overall gender pay gap was 23% (EuropeanCommission, 2009). In both countries, tradi-tional gender role beliefs are persistent (Stickney& Konrad, 2007, p. 807).


RESULTS

Descriptive Results and Sample Description

First, we provide a brief overview of somedescriptive results for our nearly 280 interviewedcouples. Only heterosexual couples wereincluded in the sample to address the hypothesesabout gender and to reach a higher standardiza-tion of the sample. Two universities (Konstanzand Bern) collaboratively conducted the researchproject using the CAPI method. The samplewas selected by the network contacts of our 60interviewers. For the research target at hand,a convenient sample seemed adequate becausethe central requirement was a random allocationof the experimental stimuli to the respondents.This procedure guaranteed the independence ofthe vignettes’ and respondents’ characteristicsthat was crucial for a strict test of our hypothe-ses. For example, it ensured the independenceof migration incentives from the respondents’attributes and therefore prevented the selec-tivity bias of nonexperimental survey data onmigrations. To restrict interviewer effects, therecruitment was restricted to an average of sevencouples per interviewer. All interviews wereconducted between June 2007 and February2008. Independently, both partners in a cou-ple responded to the vignettes and providedsome basic information about themselves andthe partnership. This was guaranteed throughoutby the presence of an interviewer who preventedmutual consultations. We oversampled coupleswithout children living in their household (suchcouples represented nearly 90% of our sample)to gain adequate variance in the willingnessto move; children are often seen as a strictobstacle to migration. Similarly, the sample wasrestricted to salaried employees living in onejoint household on a daily basis (no commutingor long-distance relationships) because of thewell-known fact that self-employed persons andolder people are especially immobile (on Ger-many, see Jurges, 2005). Moreover, we wantedto simulate a real decision-making dilemma byeliminating the desire to establish a joint house-hold as an incentive to move.

An interesting observation is that 32% of allegos reported zero willingness to move, althoughthe minimum gain was a 30% higher income (seeFigure 2). Obviously, income prospects were notthe only determinant of migration decisions. Thisfinding is in line with the general low tendencyto move, as reported in the migration literature

(Taylor, 2007). A further result worth noting isthat in about 39% of the vignettes, alter wasmore willing to move than ego (the partner whoactually got the job offer).

Table 1 presents descriptive data for thedependent variable and the sample. Becausethe vignettes’ characteristics were all a directresult of our experimental plan, they alwaysoccurred with about the same frequency. Thevast majority of our respondents were between20 and 45 years old. All respondents wereemployed for at least 19 hours per week. The vastmajority was employed full-time, and only 13%were employed fewer than 35 hours per week.

Test of the Hypotheses

Our first two hypotheses (1a and 1b) weretested using the analyses of ego’s and alter’sindividual willingness to move. Table 2 showsthe results of a random intercept model thatcontrols for the individual variables. Negativecoefficients indicate a lower willingness tomigrate as the independent variable increases.The effects found in Models 1 and 2 implicatethat both ego and alter showed a higher tendencyto move as the employment conditions atthe destination improved. Moreover, the datasupport our first hypothesis: Actors consideredtheir partner’s incentives in addition to theirown incentives. Ego was more willing to moveas alter’s employment and income prospectsimproved at the new destination, and vice versa,alter showed a higher willingness to move asego’s migration gain increased. Because of theexperimental, standardized design, we couldcompare the coefficients of alter and ego directly.An incentive to move for either partner was morerelevant for the person who would profit fromit than was the same incentive for his or herpartner. For example, many career prospectsfor ego (compared to none) increased ego’swillingness to move by an amount of 0.91 butalter’s willingness to move only by an amountof 0.76. In contrast, alter’s willingness to movewas influenced more strongly by his or her ownbetter income prospects (1.60) than by ego’swillingness to move (0.87). Both ego and altertook their partners’ situation into account butgave their own hypothetical gains more weightthan their partners’ throughout.

An interesting finding is that female egosdisplayed a lower willingness to move ascompared to male egos (for similar results,


Table 1. Sample Description

N Range Mean SD Median

Dependent variablesWillingness to move

Ego 2,772 1 – 11 3.90 3.04 3Alter 2,770 1 – 11 3.91 2.93 3

Potential for conflicts (absolute value) 2,762 0 – 10 2.55 2.38 2Vignettes’ characteristics

Gain of income for ego (%) 2,780 30 – 70 49.82 14.07 50Career prospects for ego (ref.: none)

Some 2,780 0 – 1 .34 .47Many 2,780 0 – 1 .32 .47

Commuting time (hours) 2,780 0.75 – 3.0 1.62 0.79 1.5Only reachable by car 2,780 0 – 1 .48 .50Employment prospects for alter (ref.: little)

Moderate 2,780 0 – 1 .32 .47Good 2,780 0 – 1 .33 .47

Income prospects for alter at destination (ref.: smaller)Equal 2,780 0 – 1 .33 .47Better 2,780 0 – 1 .33 .47

Respondents’ characteristicsFemale respondent 556 0 – 1 .50Age 547 20 – 50 30.93 4.97 30Duration of employment (months) 551 0 – 360 61.96 58.03 45Duration of residence (years) 552 0 – 45 20.33 11.70 24Net income (euros)a 555 360 – 5,014 1, 935.61 865.32 1, 750Propensity to lose the actual jobb 555 1 – 11 3.15 2.49 2University graduate 556 0 – 1 .44 .50

Households’ characteristicsLiving in Switzerland 278 0 – 1 .23 .42Real estate property 278 0 – 1 .25 .43Married couple 278 0 – 1 .35 .48Duration of partnership (months) 278 2 – 325. 88.03 56.79 78Children living in household 277 0 – 1 .12 .33

Married, long duration of partnership and 277 0 – 1 .08 .27children in household

Educationc

Male partner has higher education 278 0 – 1 .32 .47Female partner has higher education 278 0 – 1 .29 .45

Incomed

Male partner has higher income 277 0 – 1 .77 .42Female partner has higher income 277 0 – 1 .17 .38

Agee

Male partner is older 269 0 – 1 .44 .50Female partner is older 269 0 – 1 .06 .23

aNet income: adjusted for the different purchasing power in Switzerland and Germany. bPropensity to lose the actual job:scored from 1 = very unlikely to 11 = very likely. cEducation: 0 = spouses have the same education, 1 = male or femalespouse has higher education (at least 1 year more schooling). dIncome: 0 = the monthly net income of the two spouses differby less than 100 euros, 1 = male or female partner has at least 100 euros more income. eAge: 0 = the age difference is lessthan three years, 1 = male or female partner is at least 3 years older than his or her spouse.


Table 2. Summary of Linear Regression Analysis for Variables Predicting the Willingness to Move (Random InterceptModels)

Model 1: OwnJob Offer (Ego)

Model 2: Partner GetsJob Offer (Alter)

B SE B B SE B

Vignettes’ characteristicsGain of income for ego (10%) 0.30∗∗∗ 0.03 0.23∗∗∗ 0.03Career prospects for ego (ref. none)

Some 0.74∗∗∗ 0.10 0.45∗∗∗ 0.10Many 0.91∗∗∗ 0.10 0.76∗∗∗ 0.10

Commuting time (hours) −0.03 0.05 −0.17∗∗ 0.05Only reachable by car (ref.: also by train) 0.00 0.08 −0.07 0.08Employment prospects for alter atdestination (ref.: little)

Moderate 0.45∗∗∗ 0.10 0.84∗∗∗ 0.10Good 1.03∗∗∗ 0.10 1.98∗∗∗ 0.10

Income prospects for alter at destination(ref.: smaller in comparison with theactual destination)

Equal 0.52∗∗∗ 0.10 0.67∗∗∗ 0.10Better 0.87∗∗∗ 0.10 1.60 ∗∗∗ 0.10

Respondents’ characteristicsInterview conducted in Switzerland −0.01 0.38 0.02 0.32Female respondent −0.80∗∗ 0.29 0.11 0.25Age −0.06 0.03 −0.04 0.03Duration of residence −0.02 0.01 −0.03∗ 0.01Real estate property −1.13∗∗ 0.33 −0.64∗ 0.29University graduate 0.41 0.31 0.23 0.28Duration of employment (10 years) 0.02 0.03 −0.01 0.02Net income (1,000 euros)a 0.28 0.20 −0.10 0.18Propensity to lose the actual job 0.14∗∗ 0.05 0.11∗ 0.05

Constant 2.56∗ 1.01 2.72∗∗ 0.89Observations 2,704 2,671

(271 interviewees) (268 interviewees)Variance

σν 2.02∗∗∗ 0.10 1.69∗∗∗ 0.08σε 1.95 0.03 1.97 0.03

Note: Gray shading marks vignette variables directly concerning the employment prospects of the respondent (instead ofthose of his or her partner).

aAdjusted for the different purchasing power in Switzerland and Germany.∗p < .05, ∗∗p < .01, ∗∗∗p < .001.

see Baldridge, Eddleston, & Veiga, 2006;Bielby & Bielby, 1992). Without controllingfor the absolute income level of male andfemale respondents, this difference was evengreater (results not displayed). A reason forthis observation may be that women prefer ashorter distance from their home to their placeof employment (McLafferty & Preston, 1997)and rely on social support from relatives and

other networks when they have children orwhen they anticipate their family obligationsin the future. As a consequence, they may bemore reluctant to leave a carefully constructedwork arrangement (Kalleberg & Rosenfield,1990). Employment opportunities for better paymay be less beneficial for women because itmay be difficult to achieve the same kindof balance between household responsibilities


and employment responsibilities in a new job.We return to this when we present modelsdifferentiated by gender.

As for the other individual determinants,our results are mostly in line with previousfindings from survey research on migration.People tended to be less mobile when they ownedreal estate property (Deane, 1990) or as the timelived in one place increased (Baldridge et al.,2006; Jurges, 2005). The respondents’ actualemployment characteristics, education level,and current network embeddedness, however,did not have a significant influence on thewillingness to move. This may be the resultof effects working in opposite directions: Ahigher education level may, for example, lead tobetter earning prospects at the actual destinationand increase the potential gains of migration.Because these variables served only as controls,we refrained from deeper analyses.

Taken together, the outcomes from theindividual models are mainly consistent withprevious results based on migration research andprovide strong evidence for our first hypothesis.

We next examined our hypothesis regardingthe conflict potential. It was measured as dif-ference between ego’s and alter’s willingness tomove and hence represents a variable at the cou-ple level. Table 3 presents the results of a randomintercept model that includes partnership-levelvariables at Level 2. For ease of interpretation,we used the absolute values of the conflict poten-tial as dependent variable; greater values alwaysindicate a greater disagreement in interests, who-ever is more willing to move. Therefore, a pos-itive (negative) regression coefficient indicatesan increase (decrease) of the conflict potentialas the respective variable increases.

Starting with ego’s mobility incentives, aninteresting observation is that, as income andemployment prospects improved, the potentialfor conflicts also increased. This result is con-sistent with our second hypothesis: A one-sidedincrease in bargaining power will raise the dif-ference in interests and, as a result, the conflictpotential. Because of the characteristics of a mul-tivariate regression analysis, the effect in ques-tion is indeed one sided because we controlledfor alter’s bargaining power in the multivariateanalysis. It is also exactly that aspect that makessense of alter’s results that appear to be simi-larly counterintuitive. Alter’s improving incomeand employment prospects increased the conflictpotential. Because the indicators of the partner’s

bargaining power were always kept constantby the multiple regression models, we couldinterpret any actor’s changes in employmentprospects as one sided. Furthermore, bargainingtheory is very clear on this point. If alter’s (ego’s)incentive to move increases and ego’s (alter’s)incentive stays constant, we should observe ahigher conflict, and the findings sustain this pat-tern. Consequently, the data support our corehypothesis, H2.

For the third hypothesis, we have to lookat partnership characteristics. We assumed thatknowledge about each other and the level ofcommitment might be variables that counteractthe conflict potential. Models 1 – 4 in Table 3present the individual effects of the four relevantvariables: marriage, children, and estate propertyas indicators of commitment and duration ofthe partnership as an indicator of learning andcommitment. The results mainly support ourhypotheses. The conflict potential decreased if acouple was married, had children, or ownedreal estate and decreased as the length ofthe relationship increased. The existence of amarriage and children showed the expectednegative influence on the conflict potential,but both effects failed to reach statisticalsignificance. In the full Model 5, the effectsof the variables were all still negative but nolonger significant. A possible explanation for thismay be the relatively homogenous sample withhigh correlations of these partnership attributes(e.g., marriage and partnership were correlatedat .45). Even with a random sample and morecases, it would be difficult to disentangle theeffects because of an underlying mechanism:Marriage and children make a partnership morestable, and couples with a longer commonhistory are more likely to marry and to becomeparents. To avoid potential multicollinearity, wecompared an extreme group (couples that weremarried, had a long partnership duration, andhad children) with the rest of our sample. Longduration was defined by more than 6.5 years andapplied to about half of our sample. The resultspresented in Model 6 support our hypothesis thatthese determinants, especially in combinationwith one another, are associated with less conflictpotential in a partnership.

The variables indicating homogeneity of edu-cation, income, and age in Models 5 and 6served as controls for different compositions ofrelationships. In the bargaining literature, theactors who possess more of these resources

Migration Decisions Within Dual-Earner Partnerships 887T

able

3.Su

mm

ary

ofLi

near

Reg

ress

ion

Ana

lysi

sfo

rV

aria

bles

Pre

dict

ing

the

Pot

entia

lofC

onfli

cts

(Ran

dom

Inte

rcep

tMod

els)

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Mod

el5

Mod

el6

BSE

BB

SEB

BSE

BB

SEB

BSE

BB

SEB

Vig

nette

s’ch

arac

teri

stic

sG

ain

ofin

com

efo

reg

o(1

0%)

0.14

∗∗∗

0.03

0.14

∗∗∗

0.03

0.14

∗∗∗

0.03

0.14

∗∗∗

0.03

0.15

∗∗∗

0.03

0.15

∗∗∗

0.03

Car

eer

pros

pect

sfo

reg

o(r

ef.:

none

)So

me

0.11

0.10

0.12

0.10

0.11

0.10

0.11

0.10

0.12

0.10

0.12

0.10

Man

y0.

32∗∗

0.10

0.31

∗∗0.

100.

31∗∗

0.10

0.31

∗∗0.

100.

34∗∗

0.10

0.34

∗∗0.

10C

omm

utin

gtim

e(h

ours

)−0

.06

0.05

−0.0

60.

05−0

.07

0.05

−0.0

70.

05−0

.07

0.05

−0.0

70.

05O

nly

reac

habl

eby

car

(ref

.:al

soby

trai

n)0.

050.

080.

050.

080.

060.

080.

060.

080.

050.

080.

050.

08E

mpl

oym

entp

rosp

ects

for

alte

rat

dest

inat

ion

(ref

.:lit

tle)

Mod

erat

e0.

100.

100.

110.

100.

100.

100.

100.

100.

120.

100.

130.

10G

ood

0.51

∗∗∗

0.10

0.52

∗∗∗

0.10

0.51

∗∗∗

0.10

0.51

∗∗∗

0.10

0.52

∗∗∗

0.10

0.51

∗∗∗

0.10

Inco

me

pros

pect

sfo

ral

ter

(ref

.:sm

alle

rin

com

p.w

ithac

tual

loca

tion)

Equ

al0.

23∗

0.10

0.24

∗0.

100.

23∗

0.10

0.23

∗0.

100.

24∗

0.10

0.24

∗0.

10B

ette

r0.

64∗∗

∗0.

100.

65∗∗

∗0.

100.

64∗∗

∗0.

100.

64∗∗

∗0.

100.

65∗∗

∗0.

100.

66∗∗

∗0.

10C

oupl

es’

char

acte

rist

ics

Mar

ried

coup

le−0

.31

0.17

0.03

0.21

−0.0

00.

20C

hild

ren

livin

gin

hous

ehol

d−0

.48

0.25

−0.4

90.

28R

eale

stat

epr

oper

ty−0

.51∗

∗0.

19−0

.4∗

0.21

−0.4

9∗0.

20D

urat

ion

ofpa

rtne

rshi

p(1

0m

onth

s)−0

.04∗

∗0.

01−0

.02

0.02

Mar

ried

,lon

gdu

ratio

nof

part

ners

hip

&ch

ildre

nin

hous

ehol

da−0

.76∗

0.33

Edu

catio

n(r

ef.:

sam

eed

ucat

ion)

Mal

epa

rtne

rha

shi

gher

educ

atio

n−0

.19

0.20

−0.1

40.

20Fe

mal

epa

rtne

rha

shi

gher

educ

atio

n−0

.25

0.20

−0.2

20.

20In

com

e(r

ef.:

sam

ein

com

e)b

Mal

epa

rtne

rha

shi

gher

inco

me

0.31

0.37

0.30

0.37

Fem

ale

part

ner

has

high

erin

com

e0.

490.

410.

470.

41A

ge(r

ef.:

sam

eag

e)c

888 Journal of Marriage and FamilyT

able

3.C

ontin

ued

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Mod

el5

Mod

el6

BSE

BB

SEB

BSE

BB

SEB

BSE

BB

SEB

Mal

epa

rtne

ris

olde

r0.

340.

180.

35∗

0.17

Fem

ale

part

ner

isol

der

0.31

0.38

0.31

0.37

Inte

rvie

wco

nduc

ted

inSw

itzer

land

−0.2

40.

21−0

.25

0.21

Con

stan

t1.

35∗∗

∗0.

251.

28∗∗

∗0.

241.

37∗∗

∗0.

251.

58∗∗

∗0.

271.

45∗∗

0.53

1.32

∗0.

52O

bser

vatio

ns(c

oupl

es)

2,76

2(2

78)

2,75

2(2

77)

2,76

2(2

78)

2,76

2(2

78)

2,65

8(2

67)

2,65

8(2

67)

Var

ianc

eσ

ν1.

21∗∗

∗0.

071.

21∗∗

∗0.

071.

20∗∗

∗0.

071.

20∗∗

∗0.

071.

17∗∗

∗0.

071.

17∗∗

∗0.

07σ

ε2.

000.

032.

000.

032.

000.

032.

000.

032.

010.

032.

010.

03

a ‘‘L

ong

dura

tion

ofpa

rtne

rshi

p’’

isde

fined

bym

ore

than

6.5

year

s.bC

ases

are

coun

ted

as‘‘

sam

ein

com

e’’

ifth

em

onth

lyne

tin

com

eof

the

two

part

ners

diff

ers

byfe

wer

than

100

euro

s.c C

ases

are

coun

ted

as‘‘

sam

eag

e’’

ifth

eag

edi

ffer

ence

isfe

wer

than

3ye

ars.

∗ p<

.05,

∗∗p

<.0

1,∗∗

∗ p<

.001

.

(education, income, and age) as compared totheir partners are often considered to be morepowerful. For example, a higher age indicates, ingeneral, a higher advancement of occupationalcareers (Hallerod, 2005). Job-related migrationsare, however, assumed to be more financiallyrewarding for younger people because of thelonger time span left for labor market pay-offs (Becker, 1993). Hence, any increase in theage difference between partners should increasethe conflict potential regarding migrations. Wefound a significantly higher conflict potentialonly in Model 6 and in those partnerships wherethe male partner was at least 3 years older thanhis female partner. That we did not observe cor-responding results for the partnerships in whichthe female partner was older may be the conse-quence of the small number of those couples inour sample; only 6% fell into this category (seeagain Table 1). Similarly, the high homogeneityof our sample and the high tendency of positiveassortative mating lead to small variances in theeducation and income differences between thepartners and may explain why the variables didnot have strong effects.

In addition, we tested several interactioneffects of the career and employment prospectsof the two partners to find out whether theconflict potential was dampened if both partnershad good opportunities at the destination. Theregression estimates, which are not displayedbut are available on request, revealed that theinteraction terms mainly showed the expectednegative effects on the conflict potential butfailed to reach significance.

Finally, we focused on possible gendereffects. To test H4, which states that maleemployment characteristics have a strongereffect on the willingness to move than femaleemployment characteristics, we estimated sep-arate models for male and female respondents.The results shown in Table 4 do not providevery strong support for gendered evaluations ofthe vignettes. Comparing the results for menwith the results for women in Model 1 (ego)and Model 2 (alter), the gender gap was, ingeneral, in line with the hypothesis but notstatistically significant. We analyzed overall dif-ferences between men and women by usingChow tests (Wooldridge, 2003) and the differ-ence of single variables by interaction terms withgender, estimating in each case a joint model formale and female respondents. There was onlyone significant difference for alters (Model 2):


The coefficient for egos’ career prospects wassignificantly higher in the case of female altersthan in the case of male alters. This means thatwomen in the position of alter valued the careeroptions related to their partners’ job optionsmore than men did. Further analyses (not pre-sented here) demonstrated that the overall resultswere similar when we controlled for differencesin age, education, and income in partnerships.The evaluation, therefore, seemed to not be influ-enced by actors’ actual relative power within thepartnership. Thus, the evidence regarding H4was mixed overall: A female ego’s willingnessto move in general was lower (see Table 2), butthe impact of incentives did not vary much bygender. Only alter’s willingness to move wasmore influenced by ego’s employment opportu-nities when alter was the female instead of malepartner.

DISCUSSION

The goal of this article was twofold. Weattempted to shed light on the influence ofboth partners’ job opportunities in migrationdecisions within dual-earner households, and weproposed a new way to test bargaining theory infamily research. On the basis of a factorial surveyof about 280 European couples, we demonstratedthat there is a substantial potential for conflictsin a partnership when work-related incentivesfor a move arise. As predicted, the conflictpotential in a couple varied according to eachpartner’s change in relative bargaining power.Moreover, we found clear evidence that coupleswere able to minimize this conflict throughmutual commitment and knowledge of eachother (measured by the duration of partnership).Hence, our data support the bargaining model.In addition, we found little evidence for therelevance of gender regarding the influence ofbargaining power on the individual willingnessto move. This result has to be treatedcautiously because couples without children areoversampled in our data, and couples withchildren tend to develop a more gendereddivision of labor (Baxter, Hewitt, & Haynes,2008). Even though we found only slightevidence for gendered decisions, these mightin some constellations already be sufficient totrigger a continuously widening of the careergap between men and women because—froma bargaining point of view—subsequent careerand migration decisions will especially promote

the career of the more successful and thereforepowerful spouse.

An important limitation of our study is that weobserved only hypothetical decisions and no realactions. It is still unclear to what extent factorialsurveys produce valid results (Eifler, 2007).Fortunately, much research evaluating similarmethods (conjoint analysis and stated choiceexperiments) exists. This research showed, allin all, an astonishingly high correspondence ofhypothetical and experimental decisions withreal observations (Blamey & Bennett, 2001;Louviere, Hensher, & Swait, 2000). Moreover,there was evidence that the willingness tomove is a feasible predictor of actual relocationbehavior (Brett & Reilly, 1988). In addition,further analyses of migration behavior inGermany found that variables such as real estateproperty and occupational characteristics havea similar impact on the willingness to movereported in the analyses on hand and on realmigration propensity measured with the GermanSocio-Economic Panel (Nisic & Auspurg, 2009).Thus, we do not have any major reason to doubtthe overall external validity of our results.

Another critical aspect is the occurrence ofa social desirability bias. The sensitivity of fac-torial surveys for those effects in general iscontroversial. In our case, the risk seems minorbecause we did not ask very sensitive ques-tions. Social desirability should, by definition,intensify the measured relevance of norms. Thatwe observed only weak evidence for genderednorms is, therefore, a hint for the robustnessof our results, even if social desirability cannotbe completely excluded. Regarding the conve-nience sample, we encourage replications of ourdesign with other sampling techniques.

Another direction for future research wouldbe a test of the supposition that the actors antic-ipate future allocations. This could be done byadditional answer scales that question the antici-pated share of resources or even obligations (e.g.,housework tasks) in the case where a householdmove has taken place. Gender dynamics under-lying the decision-making processes not visiblethrough the quantitative data on hand mightbe addressed by combining vignette analyseswith qualitative interviews (Ganong & Cole-man, 2006).

In general, the results of our study emphasizethe relevance of a couple- or household-orientedapproach when labor market processes are stud-ied. Individual career mobility is often directly


Table 4. Summary of Linear Regression Analysis for Variables Predicting the Willingness to Move of Men and Women(Random Intercept Models)

Model 1: Own Job Offer (Ego) Model 2: Partner Gets Job Offer (Alter)

Men Women Men Women

Variable B SE B B SE B B SE B B SE B

Gain of income for ego (10%) 0.35∗∗∗ 0.04 0.25∗∗∗ 0.04 0.22∗∗∗ 0.04 0.27∗∗∗ 0.04Career prospects for ego (ref. none)Some 0.85∗∗∗ 0.14 0.57∗∗∗ 0.14 0.20 0.14 0.66∗∗∗ 0.15Many 0.90∗∗∗ 0.14 0.84∗∗∗ 0.14 0.67∗∗∗ 0.14 0.85∗∗∗ 0.14Commuting time (hours) 0.08 0.07 −0.14∗ 0.07 −0.21∗∗ 0.07 −0.15∗ 0.07Only reachable by car (ref.: also by

train)0.08 0.11 −0.12 0.11 −0.02 0.11 −0.09 0.12

Employment prospects for alter atdestination (ref.: little)

Moderate 0.51∗∗∗ 0.14 0.33∗ 0.14 0.82∗∗∗ 0.13 0.78∗∗∗ 0.14Good 1.09∗∗∗ 0.14 0.86∗∗∗ 0.14 1.83∗∗∗ 0.13 2.06∗∗∗ 0.14Income prospects for alter (ref.:

smaller in comp. with actuallocation)

Equal 0.53∗∗∗ 0.13 0.55∗∗∗ 0.13 0.62∗∗∗ 0.13 0.62∗∗∗ 0.14Better 0.92∗∗∗ 0.13 0.90∗∗∗ 0.14 1.61∗∗∗ 0.13 1.51∗∗∗ 0.14Constant 0.68 0.37 1.35∗∗∗ 0.37 1.17∗∗ 0.36 0.92∗ 0.36Observations 1,438 1,334 1,336 1,434(Interviewees) (144) (134) (134) (144)Varianceσν 2.36∗∗∗ 0.15 1.99∗∗∗ 0.13 1.82∗∗∗ 0.12 1.84∗∗∗ 0.12σε 1.99 0.04 1.92 0.04 1.90 0.04 2.05 0.04

Note: In this regression analysis, only vignettes’ characteristics were included. Gray shading marks variables concerningthe male partner. Coefficients in bold indicate significant differences between men and women.

∗p < .05, ∗∗p < .01, ∗∗∗p < .001.

linked to regional mobility, and we have shownthat new individual career options for one part-ner create potential conflicts in the relationship.Cooperative solutions require balanced powershifts or, at the least, credible commitment to thepartnership. The scope of this finding is proba-bly not restricted to migration decisions becausethe same theoretical mechanism can be assumedfor other issues such as fertility decisions. With-out solutions to the conflicts connected withchoices that asymmetrically change the employ-ment prospects of the partners and therefore shifttheir balance of bargaining power, suboptimallevels of family welfare will arise (Jurges, 2005;Ott, 1992).

At least in our sample of European dual-earner couples, the responses to the incentivesfor regional moves differed only slightly bygender. Apart from the fact that women tendedto be more resistant to moving than their

male partners, we found gender roles to beless obviously important than has often beenstated. The high prevalence of tied womenreported in the literature seems to result morefrom their inferior labor market positions thanfrom traditional gender roles in the household.We found that both male and female partnerstake their respective partner’s perspective intoaccount, and they did it mostly in the same way.In a more general view, this means that theefforts to achieve greater gender equality in thelabor market and in families go hand in hand.Better career prospects for women and morebalanced (migration) decisions presuppose eachother.

REFERENCES

Antel, J. J. (1980). Returns to migration: Literaturereview and critique. Santa Monica, CA: RAND.


Baldridge, D. A., Eddleston, K. A., & Veiga, J. F.(2006). Saying ‘‘no’’ to being uprooted: Theimpact of family and gender on willingness torelocate. Journal of Occupational and Organiza-tional Psychology, 79, 131 – 149.

Bartel, A. P. (1982). Wages, nonwage job character-istics, and labor mobility. Industrial and LaborRelations Review, 35, 578 – 589.

Baxter, J., Hewitt, B., & Haynes, M. (2008). Lifecourse transitions and housework: Marriage,parenthood, and time on housework. Journal ofMarriage and Family, 70, 259 – 272.

Becker, G. S. (1993). Human capital: A theoreticaland empirical analysis with special reference toeducation. Chicago: University of Chicago Press.

Bergstrom, T. C. (1996). Economics in a family way.Journal of Economic Literature, 34, 1903 – 1934.

Bernasco, W., & Giesen, D. (2000). A bargainingapproach to specialization in couples. In J. Weesie& W. Raub (Eds.), The management of durablerelations: Theoretical models and empirical stud-ies of households and organizations (pp. 42 – 43).Amsterdam: Thela Thesis.

Bielby, W. T., & Bielby, D. D. (1992). I will followhim: Family ties, gender-role beliefs, and reluc-tance to relocate for a better job. American Journalof Sociology, 97, 1241 – 1267.

Bitman, M., England, P., Folbre, N., Sayer, L., &Matheson, G. (2003). When does gender trumpmoney? Bargaining and time in household work.American Journal of Sociology, 109, 186 – 214.

Blackburn, Mc. K. L. (2006). The impact of internalmigration on married couples’ earnings in Britain,with a comparison to the United States (ISERWorking Paper 2006-24). Essex, UK: Institute forSocial and Economic Research.

Blamey, R., & Bennett, J. (2001). Yea-saying andvalidation of a choice model of green productchoice. In J. Bennett & R. Blamey (Eds.), Thechoice modeling approach to environmentalvaluation (pp. 179 – 201). Cheltenham, UK: Elgar.

Blau, F., Ferber, M. A., & Winkler, A. E. (2001). Theeconomics of women, men, and work. EnglewoodCliffs, NJ: Prentice Hall.

Blood, R. O., & Wolfe, D. M. (1960). Husbands andwives: The dynamics of married living. Glencoe,IL: Free Press.

Brett, J. M., & Reilly, A. H. (1988). On the roadagain: Predicting the job transfer decision. Journalof Applied Psychology, 73, 614 – 620.

Brines, J., & Joyner, K. (1999). The ties that bind:Principles of cohesion in cohabitation and marri-age. American Sociological Review, 64, 333 – 355.

Cooke, L. P. (2006). ‘‘Doing’’ gender in context:Household bargaining and risk of divorce inGermany and the United States. American Journalof Sociology, 112, 442 – 472.

DaVanzo, J., & Hosek, J. R. (1981). Does migrationincrease wage rates? An analysis of alternative

techniques for measuring wage gains to migration.Santa Monica, CA: RAND.

Deane, G. D. (1990). Mobility and adjustment: Pathsto the resolution of residential stress. Demography,27, 65 – 79.

Eifler, S. (2007). Evaluating the validity of self-reported deviant behavior using vignette analyses.Quality and Quantity, 41, 303 – 318.

England, P., & Farkas, G. (1986). Households,employment, and gender: A social, economic, anddemographic view. New York: Aldine.

European Commission. (2009). Report on equalitybetween women and men 2009. Luxembourg:Office for Official Publications of the EuropeanCommunities.

Fenstermaker, S. (2002). Work and gender. InS. Fenstermaker & C. West (Eds.), Doing gen-der, doing difference: Inequality, power, andinstitutional change (pp. 105 – 114). New York:Routledge.

Ganong, L. H., & Coleman, M. (2006). Multiplesegment factorial vignette design. Journal ofMarriage and Family, 68, 455 – 468.

Hallerod, B. (2005). Sharing of housework andmoney among Swedish couples: Do they behaverationally? European Sociological Review, 21,273 – 288.

Hook, J., & Chalsani, S. (2008). Gendered expec-tations? Reconsidering single fathers’ child-caretime. Journal of Marriage and Family, 71,978 – 990.

Hox, J. J., Kreft, I., & Hermkens, P. (1991). Theanalysis of factorial surveys. Sociological Methodsand Research, 19, 493 – 510.

Huppe, M., & Cyr, M. (1997). Division of householdlabor and marital satisfaction of dual incomecouples according to family life cycle. CanadianJournal of Counseling, 31, 145 – 162.

Jacobson, J. P., & Levin, L. M. (2000). The effects ofinternal migration on the relative economic statusof women and men. Journal of Socio-Economics,29, 291 – 304.

Jasso, G. (2006). Factorial survey methods forstudying beliefs and judgments. SociologicalMethods and Research, 34, 334 – 423.

Jurges, H. (2005). Gender ideology, division ofhousework, and the geographic mobility offamilies. Review of Economics of the Household,4, 299 – 323.

Kalleberg, A. L., & Rosenfield, R. A. (1990). Workin the family and in the labor market: A cross-national, reciprocal analysis. Journal of Marriageand the Family, 52, 331 – 346.

Klasen, S. (1998). Marriage, bargaining, and intra-household resource allocation: Excess femalemortality among adults during early Germandevelopment, 1740 – 1860. Journal of EconomicHistory, 58, 432 – 467.


Kuhfeld, W. F. (2005). Marketing research methodsin SAS: Experimental design, choice, conjoint andgraphical techniques. Cary, NC: SAS Institute.

Kuhfeld, W. F., Randall, T. D., & Garratt, M. (1994).Efficient experimental design with marketingresearch applications. Journal of MarketingResearch, 31, 545 – 557.

Lichter, D. (1983): Socioeconomic returns to migra-tion among married women. Social Forces, 62,487 – 503.

Louviere, J. L., Hensher, D. A., & Swait, J. D. (2000).Stated choice methods: Analysis and application.Cambridge, UK: Cambridge University Press.

Lundberg, S., & Pollak, R. A. (1996). Bargaining anddistribution in marriage. Journal of EconomicPerspectives, 10, 139 – 158.

Lundberg, S., & Pollack, R. A. (2003). Efficiency inmarriage. Review of Economics of the Household,1, 153 – 167.

Manser, M., & Brown, M. (1980). Marriage andhousehold decision-making: A bargaining anal-ysis. International Economic Review, 21, 31 – 44.

Maxwell, N. (1988). Economic returns to migration:Marital status and gender differences. SocialScience Quarterly, 69, 108 – 121.

McHugh, K. E., Gober, P., & Reid, N. (1990). Deter-minants of short- and long-term mobility: Expec-tations for home owners and renters. Demography,27, 81 – 95.

McLafferty, S., & Preston, V. (1997). Gender, race,and the determinants of commuting: New York in1990. Urban Geography, 18, 192 – 212.

Miller, R. S., Perlman, D., & Brehm, S. S. (2007).Intimate relationships. Boston: McGraw-Hill.

Mincer, J. (1978). Family migration decisions. Jour-nal of Political Economy, 86, 749 – 773.

Nisic, N., & Auspurg, K. (2009). Faktorieller Sur-vey und klassische Bevolkerungsumfrage imVergleich – Validitat, Grenzen und Moglichkeitenbeider Ansatze. In P. Kriwy & C. Gross (Eds.),Klein aber Fein! Quantiative empirische Sozial-forschung mit kleinen Fallzahlen (pp. 211 – 246).Wiesbaden: VS-Verlag fur Sozialwissenschaften.

Nock, S. L. (1995). Commitment and dependency inmarriage. Journal of Marriage and the Family, 57,503 – 514.

Ott, N. (1992). Intrafamily bargaining and householddecisions. Berlin: Springer.

Quigley, J. M., & Weinberg, D. H. (1977). Intra-urban residential mobility: A review and synthesis.International Regional Science Review, 2, 41 – 66.

Raudenbush, S. W., & Bryk, A. S. (2001). Hierarchi-cal linear models: Applications and data analysismethods. Thousand Oaks, CA: Sage.

Rossi, P. H., & Anderson, A. B. (1982). The factorialsurvey approach: An introduction. In P. H. Rossi& S. L. Nock (Eds.), Measuring social judgments:The factorial survey approach (pp. 15 – 67).Beverly Hills, CA: Sage.

Shauman, K. A., & Noonan, M. C. (2007). Familymigration and labor force outcomes: Sex differ-ences in occupational context. Social Forces, 85,1735 – 1764.

Snijders, T. A., & Bosker, R. J. (1999). Multilevelanalysis: An introduction to basic and advancedmodeling. London: Sage.

Spitze, G. (1984). The effect of family migration onwives’ employment: How long does it last? SocialScience Quarterly, 65, 21 – 36.

Stickney, L. T., & Konrad, A. M. (2007). Gender-roleattitudes and earnings: A multinational study ofmarried women and men. Sex Roles, 57, 801 – 811.

Szinovacz, M., & Harpster, P. (1993). Employmentstatus, gender role attitudes, and marital depen-dence in later life. Journal of Marriage and theFamily, 55, 927 – 940.

Taylor, M. P. (2007). Tied migration and subsequentemployment: Evidence from couples in Britain.Oxford Bulletin of Economics and Statistics, 69,795 – 818.

Thibaut, J. W., & Kelley, H. H. (1959). The socialpsychology of groups. New York: Wiley.

West, C., & Zimmerman, D. H. (1987). Doing gender.Gender and Society, 1, 125 – 151.

Wooldridge, J. M. (2003). Introductory economet-rics: A modern approach. Mason, OH: Thomson.

Migration Decisions Within Dual-Earner Partnerships: A Test of Bargaining Theory

Documents

Transcript of Migration Decisions Within Dual-Earner Partnerships: A Test of Bargaining Theory