Gendering the Financial Crisis

7/31/2019 Gendering the Financial Crisis

1/27

1

Genderingthe2008FinancialCrisis:

AReexaminationofIntrahouseholdBargainingPowerAnalytics

JohnnyHuynh

PomonaCollege

May6,2012


2/27

2

PartI:Introduction

Neoclassicalmicroeconomicthoughtconsidersthehouseholdtobetheprimary

decision-makingagentwithasetofindividualpreferences.However,manyeconomists

haveidentifiedthathouseholdssometimesdemonstrateinternalconflictandinequity

(Phipps,Burton1995).Impactstohouseholdsarealsodistributedunevenlybetween

spouses,children.ThesefindingsunderminetheParetoefficiencyassumptionwithin

households,whichhascompoundingeffectsonmarketandmacroproductivity(Doss

1996).Tofullyunderstandhouseholdsandtheirroleintheeconomyrequiresananalysis

ofintra-householddynamics.

Iexaminehowcyclicalunemploymentandcreditcrunchesfollowingafinancial

crisisaffecthouseholddecisionsandthewelfareofhouseholdmembers.Particularly,Ilook

atthedisparateeffectsofthe2008financialcrisisonmenandwomen,andreexaminepast

frameworksthatexploreddecision-makingandrisk-bearingwithinthehousehold.This

paperutilizesempiricsfromthePanelStudyofIncomeDynamics(PSID)totestpast

modelsofintra-householdrelations.Becauseofthesubstantialrolethatmortgageshadin

the2008crisis,Iemphasizehousingassets.HousingassetsarealsonoteworthyinNew

HouseholdEconomicsbecausetheyaccruereturnsforboththebreadwinnerand

homemaker.

The2008financialcrisishasbeendescribedtohavetwoseparableeffectsonthe

household.Thefirstisareductionofaggregatedemand,whichledtohighunemployment

andadecreaseinincomeandwealth.Thesecondistheshutdownofmanycreditmarkets

thatledtoinaccessibilityofcredit(Walby2009).Ifocusonbothimpactsindividuallybut

acknowledgetheinteractionofincomeandcreditaccessibility1.


3/27

3

Bustsinthebusinesscyclehavebeenshowntohaveunequaleffectsonmenand

womenemployment,aswellasunequaleffectsinthecreditmarket2.Theseoutcomesare

likelytoalterbargainingpowerswithinthehousehold.Exploringchangesinbargaining

powerhelpstoexplainbehavioralchangesinhouseholddecisionsduetoexogenousshocks

inthebusinesscycle.

Reviewingcurrentliterature,partIIassessesfinancialcrisesdisparateeffecton

menandwomen.Emphasizediscomparativeanalysisbetweenthe1997Asiancrisisand

the2008crisis.PartIIIsummarizestheexistinganalyticalframeworkofhouseholdsand

crises,andlooksattheoreticalmodelpredictions.PartIVteststhepredictionsusingPSID

empirics.PartVconcludeswithrevisionstothemodelsandimplications.

PartII:Empiricsaboutthe2008Crisis

Aftertheearly2000srecessionanduntilthe2008financialcrisis,U.S.GDPgrew

consistentlyattwotothreepercentandpeakedat3.9percentin2003.From2002to2006,

householdsincreaseddurableconsumptionandleveragedhousingmortgagestofund

expenditures(Mian,Sufi2009).Realestatespeculationandanationalhousingbubble

boostedperceivedwealth.Afterthepeakin2003,growthslowlydeclinedeachyearuntil

2008,whenGDPgrowthbecamenegative(BLS2012).The2008recessionprimarily

impactedconstruction,manufacturing,retailtrade,financeandinsurance,andrealestate

industrieswherecorporateprofitsdroppedtotheirlowestlevelsinthedecade(BLS2012).

Theeffectsofthe2008financialcrisiscanbedividedintotwomacroeconomic

effects:higherunemploymentandcreditinaccessibility.Likemosteconomiccrises,the

2008financialcrisiswitnessedasharpdeclineinprivateinvestments,suchastechnology


4/27

4

spending,employment,andcapitalexpenditures.Firmswerealsounabletoborrowin

traditionalcreditmarketsandfundinvestmentopportunities(Campello,etal2010).

DespitetheFederalReservetargetinglowfederalfundsrates,banksmaintainedhigh

reserveratiosandwerehesitanttomakeloans.

Effectstothemacroeconomyalsospreadtohouseholds.Highunemployment,lossof

assetwealth,andpooreconomicoutlookscausedhouseholdstoreducespending(Hurd,

Rohwedder2010).Additionally,householdslackedaccesstofunctioningcreditmarkets,

despitecreditbeingespeciallyimportantduringfinancialcrises(Sullivan2005).

Householdstypicallysubstituteincomewithcredit,usesavings,andreduceconsumption

tomakeupthedifferential.AverageU.S.householdsreducedexpendituresbymorethan

fourpercent,despiteaverageincomeonlyfallinglessthantwopercentfrom2008to2010.

Householdreceiptsdeclinedinallareasexcepthealthcareandat-homefoodexpenditures3

(BLS2010).

Oftheseeconomicimpacts,onecanexplorethedisparitiesincostsincurredamong

menandwomen4.Whilehistoricallymosteconomiccrisesdisproportionatelyaffectfemale

employment5,theinitialeffectsof2008financialcrisiswereconcentratedinconstruction,

financialservice,andautomotivesectors,whicharemale-dominated.Rather,women

occupyjobslesssensitivetothebusinesscycle,suchaseducationandhealthservices.After

thecrisis,femaleunemploymentwasonaveragetwopercentagepointsbelowmale

unemploymentintheUnitedStates.Menwerealsomorelikelytodropoutofthelabor

forceinthe2008crisis(Hartman2009).Thesetrendsareillustratedinthe2003to2009

PSIDhouseholddatainTable1.


5/27

5

Theeconomicrecovery,however,disproportionatelybenefitedmaleemployment

relativetofemaleemployment.TheBureauofLaborStatistics(2012)notesthatfemale

unemploymentrateshaveremainedconstantwhereasmaleunemploymentrateshave

substantiallydecreased,almostreachingparitywithfemaleunemploymentratesby2012.

Thereisnotyetadefinitiveexplanationoftherobustnessofthemaleemployment

recovery,althoughsomecitegovernmentstimulusofconstructionandautomobile

industries,aswellasbudgetcutsinstateandlocalgovernments,whicharefemale-

concentrated(Ruggieri2010).

Themajorityofliteratureonfinancialcrisisandhouseholdsdrawonevidencefrom

the1997Asianfinancialcrisis.Thedifferencesbetweenthe1997Asiancrisisandthe2008

crisisarenotable.Priorto1997,EastAsiaexperiencedsubstantialfinancialliberalization,

dissolvinginterestratecontrolsandderegulatingfinancialmarkets.InmanyAsian

countries,financialrepressionoftenprecededliberalizationpolicies.Newcapitalmarkets

spurredforeignownership,andEastAsiaexperiencedcapitalinflows.Somehave

suggestedthattheinflowofcapitalledtoaspeculativebubbleinrealestateandequity,

sincedemandforclaimsoutpacedtheirproductivecapabilities(Dymski1999).New

financialinstitutionsinEastAsiaalsoexpandedcreditaccesstohouseholdsthatotherwise

reliedonpersonalrelationsforloans.

Asecondaryeffectwasthatforeigninvestmentscreatedmoreopportunitiesfor

marketproduction.Employmentprospectsandhighersalariesgeneratedmoreandsmaller

households,sincehouseholdsweremorewillingandabletoincurthecostsofnottaking

advantageofeconomiesofscale.Greaterjobprospectsandhigherrealwagesthat

accompaniedforeigninvestmentsalsoincreasedlaborforceparticipationamongwomen.


6/27

6

Althoughmaleemploymentalsorose,thedisparityinmale-to-femalemarketemployment

proportionsconverged.

PartIII:ExaminingBargainingPowerTheory

Ahusbandandwifeexhibitcooperativeconflict.Althoughtheyvaluetheothers

wellbeing,theyalsostruggletooptimizehouseholdandmarketproductionunderbudget

andtimeconstraints(M).

ai+aj+X=M (1)

Empiricalliteratureidentifiesthathusbandsandwiveshavedifferenttastes

concerninghouseholdexpenditures.Obviously,husbandsandwivesplacemoreworthto

assetsthataccrueprivatereturns(ai)totheirrespectiveselves.Theyalsovalue

consumptionofpublicgoods(X)inthehousehold(Lam1988;LunderbergandPollak

2008).Dependingonthealtruisticbehaviorofthehusbandorwife,theyderivesomeutility

fromtheirspousesconsumption.

Ui=Ui(ai,X)+iUj(aj,X) (2)

Becausetheutilityprofilesofhusbandsandwivesgenerallydiffer,theoptimalbundleof

privateassetsandpublicgoodsforhusbandsandwivesconflict.Eachspousevalues

differentproportionsofhisorherassetsandtheothersassets.

argmaxUi=BiargmaxUj=Bj (3) ai,aj,X ai,aj,X

Theequilibriumbundleofhusbandsassetsandwifesassetsisultimatelydeterminedby

theirrespectivebargainingpowers().Relativelymorebargainingpowerimpliesan

equilibriumbundlemorecloselyrelatedtohisorherpreference.

B=f( i,j)[Bi,Bj] (4)


7/27

7

Bargainingpowermodelsarecontinuingtobereshaped,andthevariablesthat

influencethemarestillbeingdiscussed.Pollak(2005)maintainsthatbargainingpowerisa

resultofhusbandsandwifesrespectivemarketandhouseholdproductivity.Hecitesthat

wagerates,nottotalearnings,determinehouseholddecisions.Otherpapersnotethat

respectivecontributionstopooledhouseholdresourcesaresignificantvariables.The

distributionofbargainingpoweralsovariesatdifferentlevelsofproductivity(Bittman,et

al2001).Moreover,bargainingtheorymaintainsthathavingoutsideoptionsincreases

bargainingpowerbecausetheopportunitycostofnottransactingislowerfortheactor

withoptionsthanfortheactorwithoutoptions(Muthoo2000).Costsarealsolowerforthe

spousewhocansurviveandthriveoutsidethehousehold(Agarwal1997).

Thismodelsupposesthathouseholdproduction(w)canbeeasilysubstitutedby

marketproduction(h),whereastheinverseisnottrue.Thus,inahouseholdwherethe

breadwinningandhomemakerareequallyproductive,thebreadwinnerwillhavemore

bargainingpowerbecausehehasoutsideoptions.

i=f(wi,wj,hi,hj) (5)

!!!

!!! !!!

!!! 0 !!!

!!! !!!

!!! 0 (6)

Intheirseminalpaper,FinancialCrisis,Gender,andPower:AnAnalyticalFramework

(2000),FloroandDymski(whoseanalyticsIwillrefertoasthebenchmark)explorethe

effectsoffinancialliberalizationandcrisisonintrahouseholddecision-making.Mostly

usingempiricalaccountsfromthe1997Asianfinancialcrisis,theirpaperprovidesa

narrativeattemptingtomodelthebehavioralchangesinriskandassetaccumulation

amongmenandwomen.Thebenchmarkmodelpositsthatthederegulationoffinancial


8/27

8

institutionsandoutwardorientationduringtheonsetofthe1990shadtwonotableeffects

onhouseholds.

First,financialliberalizationincreasesaccesstocredit,especiallytowives,spurring

householdleveraging.Householdsgaincredit-financedpurchasingpoweranduseitto

mitigatebudgetconstraints.ThisisillustratedbyanincreaseinMinequation(1),which

resultsinmorehusbandassets,wifeassets,and/orpublicgoods.Higherincomesand

wealthtriggermoreconsumptionofmarketsubstitutesofhouseholdproduction.Thisfrees

timefornon-productiveactivities,suchasleisure,butalsoincreasesmanyhouseholds

financialfragilitythelikelihoodthatahouseholdwillbeunabletorepayitsdebtifthere

isanexogenousreductionintheflowofincomeoremployment.

Second,themodelconjecturesthatfemaleparticipationintheformalsector

underminespatriarchalnorms.Wivescontributemoretohouseholdborrowingabilityand

therepaymentofdebtobligationsduetoincreasedformalsectorincome.Thisis

demonstratedbyasubstitutionofhouseholdproductionwithmarketproductionin

equation(5),assumingthatproductivecapabilitiesremainconstant.Additionaloutside

optionsprovidewomenmorebargainingpowerinhouseholddecisions.

Thebenchmarkoffersagraphicalanalysisoftheinteractionbetweenbargaining

power,creditdeterminants,andleverage6.Theysuggestthatfemalevoice()wives

bargainingpowerandcreditfactors(C)determineloan-marketleverage(H).

H=f(,C) [0,1],C>0 (7) Themodelsupposesthatfemalesaremorerisk-aversethanmalesbecausefemales

areatgreaterriskofincurringthecostsoffinancialdistress7.Themodelalsoindicatesthat

thefunctionalformisconcave.


9/27

9

!!!

!!


10/27

10

amountthathouseholdsleveragedbecausehighermortgagesimplysmallerdown

payments,ceterisparibus.Moreover,householdcouldborrowmoneyontheirrealestateby

takingoutasecondmortgage,increasingfinancialfragility.Usingthismetric,however,

limitsthescopeofthedatatohouseholdsthatcurrentlyhaveorhavehadatleastone

mortgage.

Thedatademonstratethatforeachyearfrom2003to2009,theproportionof

householdsthathaveremainingmortgagesincreased.In2003,theproportionwas73.2

percentandby2009,itincreasedto74.4percent.Themeanvalueofthisvariablehasalso

increasedforeachyearaswell.Thelargestjumpsoccurredbetween2003and2007,when

meanmortgageincreased17,000pertwoyears.Theseresultsindicatethathouseholds

increasedabsoluteleveragingduringthebuilduptothecrisis.Sincetheproportionand

meanvalueofmortgagesdidnotfallafter2008,thedatasetdidnotyetreflectsignificant

mortgagedefaultsthatoccurredpost-crisis.

Thoughweconcludethatabsolutehouseholdleveragingincreasedfrom2003to

2008,Ialsolookathouseholdleveragingrelativetotime-varianthouseholdcharacteristics,

suchasincomeandhousingvalue.IuseanOLSregressionwithremainingprincipal

mortgageasthedependentvariable.Sincetheonsetofthe2008crisisprimarilyimpacted

certainindustries,Ialsousethesamemodeladjustedforemploymentbyindustry.

MORTGi=0i+1iXi+2iYi+3iZi+4i*INCMi+5i*HVALUEi+i (9)

whereXisatime-invariantcharacteristicsvector,Yisatime-variantcharacteristicsvector

(excludingincomeandhousingvalue),andZisahusband/wifedifferentialvector.Ilookat

theintercepttermandthecoefficients4and5tomeasurehouseholdleveraging.

4i=!!"#$%!

!!"#$!

5i=!!"#$%!

!!"#$%&!

(10)


11/27

11

Anincreaseintheintercepttermorthecoefficientsimpliesthathouseholdsaremorerisk

tolerantbecauseremainingprincipalmortgagesincreaseexogenouslyormortgages

increasemorethanincomeorhousingvalueincreases,respectively.

ThedataforEquation(9)areillustratedinTable2,andTable3adjustedfor

industryemployment.Thelatteronlyincludeshouseholdswhoseheadsareemployedin

construction,manufacturing,retailtrade,financeandinsurance,orrealestate.

Table2:Determinantsofhouseholdleveragingintwo-personhouseholdswithmortgage:OLSmodel;

DependentVariable:REMAININGPRINCIPALMORTGAGE

Variable 2003 2005 2007 2009

AGE(Head) -1582.42 -1628.68 -1948.25 -1923.37 (139.87)*** (146.05)*** (179.05)*** (189.87)***

EDU(Head) 1322.36 1979.94 2910.63 4000.78

(977.52) (1141.7)* (1811.7) (2347.6)*

INCM(Head) 0.25088 0.44419 0.24273 0.15119

(0.0614)*** (0.1115)*** (0.1168)** (.13381)

AGEDIFF 937.524 490.363 117.363 119.630

(276.32)*** (333.27) (62.478)* (37.790)***

EMPLDIFF 6534.38 7496.46 6484.94 4774.47

(3988.0) (4176.2)* (5799.9) (5142.4)

EDUDIFF -27.7840 11.9301 -1496.40 -1728.46

(771.00) (905.16) (1212.28) (1475.9)

INCMDIFF -0.20185 -0.03907 -0.24100 -0.14626

(0.0565)*** (0.1089)*** (0.1288)* (0.1358)

HVALUE 0.37471 0.29559 0.29380 0.32447

(0.0339)*** (0.0247)*** (0.0436)*** (0.0543)***

INTRCPT 78666.5 86927.4 103588 95661.0

(13065)*** (15893)*** (21554)*** (27041)***

N 2099 2122 2133 2147

R2 0.6467 0.5811 0.5226 0.4879NoteRobuststandarderrorsareinparentheses *Significantatthe10%level;**Significantatthe5%level;***Significantatthe1%level

Table3:Determinantsofhouseholdleveragingintwo-personhouseholdswithmortgage;adjustedfor

industry:OLSmodel;DependentVariable:REMAININGPRINCIPALMORTGAGE


12/27

12

Variable 2003 2005 2007 2009

AGE(Head) -1476.61 -1757.06 -2205.49 -2095.70

(162.33)*** (205.42)*** (200.05)*** (243.49)***

EDU(Head) 206.859 303.965 2419.93 -77.2690

(974.48) (1391.6) (1409.1)* (1681.8)

INCM(Head) 0.36831 0.43089 0.20432 0.18434 (0.0911)*** (0.1530)*** (0.1202)* (0.1767)

AGEDIFF 608.675 106.786 825.738 68.0674

(425.04) (441.66) (581.13) (15.872)***

EMPLDIFF 7093.37 8643.63 6286.54 1458.11

(5325.4) (6647.9) (7266.6) (7906.7)

EDUDIFF 958.171 747.966 -396.553 918.883

(977.08) (1104.1) (1274.2) (1546.4)

INCMDIFF -0.31392 -0.39634 -0.19916 -0.21584

(0.09059)*** (0.1236)*** (0.1233) (0.1496)

HVALUE 0.40666 0.34411 0.35560 0.44906

(0.0269)*** (0.0316)*** (0.0220)*** (0.0429)***

INTRCPT 81104.4 103450 103493 128878

(13984)*** (20489)*** (20552)*** (24139)***

N 1010 1102 1119 1062


Therearelittledifferencesbetweentheresultsforbothsamples.Inbothregressions,the

intercepttermincreasesbienniallyfrom2003to2007.Itcontinuesincreasingforthe

unadjustedgroupandslightlydecreasesfortheadjustedgroupbetween2007and2009.

Thereisastatisticallysignificantandpositiveoveralltrendoftheinterceptterm.However,

thecoefficientsforhousingvalueandincomearelessconclusive.Thehousingvalue

coefficientissignificantatonepercentsizeforallyears,butitsvaluedoesnotchange.The

incomecoefficientpeaksin2005anddecreasesforcontinuingyears.Itissignificantatone

percentsizein2003and2005,issignificantatfivetotenpercentsizein2007andisnot

significantin2009forbothsamples.ThesetrendsareillustratedinFigure1.


13/27

13

Theriseintheintercepttermsuggeststhathouseholdsexogenouslyleveraged

duringtheonsetofthefinancialcrisis.Thecontrolvariablescouldnotexplainthisincrease

inremainingprincipalmortgages,whichcouldimplythatthemodelomitsanuncorrelated

variablethatinfluenceshouseholdleveraging,suchasinterestratesorrisktolerance.Since

thetrendsinthehousingvalueandincomecoefficientsareindeterminate,thispaper

neitheracceptsnorrejectsthehypothesisthatperceivedhigherincomesorrisinghousing

pricesinfluencedleveragingbehavior.Thisconclusionisconsistentwiththebenchmark

predictionthathouseholdleveragingwouldincreaseduringthestartofafinancialcrisis.

Itestthesecondbenchmarkpredictionthatwivesaremoreriskadversethan

husbandsusingtwomethods.First,Iexaminebargainingpowermetrics,suchas

differentialsinincome,employment,age,andeducationbetweenhusbandandwife,and


14/27

14

theirinfluenceonhouseholdleveraging.Second,Icomparetheriskprofilesofsinglemale-

headedandsinglefemale-headedhouseholdstoseewhetherthereareinherentrisk

aversiondifferentialsbygender.

Thefirstmethodusesthesamemodelinequation(9)butassessestheinfluenceof

thehusband/wifedifferentialvectorZ.Alldifferentialvariablesarecalculatedby

subtractingthefemalecharacteristicvaluefromthemalecharacteristicvalue.Thus,a

positivedifferentialvalueindicatesthatthehusbandhasahigherincome,age,or

education,orisemployedwhereasthewifedoesnotparticipateintheformallabormarket.

Thesevariablesareproxiesforintrahouseholdbargainingpowerbecauserelativelevelsof

productivityandhavingoutsideoptionsdeterminebargainingpowers.Apositive

differentialsuggeststhathusbandshavemorebargainingpower.Thecoefficientsofvector

Zmeasurebargainingpowerseffectsonremainingprincipalmortgage.Thatis,ifwivesare

relativelymoreriskaverse,thenapositiveincreaseinadifferentialvariablewillincrease

householdleveraging,ceterisparibus.

TheresultsaredemonstratedinTable2andTable3,thoughthesamplesprovide

similarconclusions.

Table2:Determinantsofhouseholdleveragingintwo-personhouseholdswithmortgage:OLSmodel;

DependentVariable:REMAININGPRINCIPALMORTGAGE

Variable 2003 2005 2007 2009

AGE(Head) -1582.42 -1628.68 -1948.25 -1923.37

(139.87)*** (146.05)*** (179.05)*** (189.87)***

EDU(Head) 1322.36 1979.94 2910.63 4000.78

(977.52) (1141.7)* (1811.7) (2347.6)*

INCM(Head) 0.25088 0.44419 0.24273 0.15119

(0.0614)*** (0.1115)*** (0.1168)** (.13381)

AGEDIFF 937.524 490.363 117.363 119.630

(276.32)*** (333.27) (62.478)* (37.790)***


15/27

15

EMPLDIFF 6534.38 7496.46 6484.94 4774.47

(3988.0) (4176.2)* (5799.9) (5142.4)

EDUDIFF -27.7840 11.9301 -1496.40 -1728.46

(771.00) (905.16) (1212.28) (1475.9)

INCMDIFF -0.20185 -0.03907 -0.24100 -0.14626

(0.0565)*** (0.1089)*** (0.1288)* (0.1358)

HVALUE 0.37471 0.29559 0.29380 0.32447

(0.0339)*** (0.0247)*** (0.0436)*** (0.0543)***

INTRCPT 78666.5 86927.4 103588 95661.0

(13065)*** (15893)*** (21554)*** (27041)***

N 2099 2122 2133 2147


Table3:Determinantsofhouseholdleveragingintwo-personhouseholdswithmortgage;adjustedfor

industry:OLSmodel;DependentVariable:REMAININGPRINCIPALMORTGAGE

Variable 2003 2005 2007 2009

AGE(Head) -1476.61 -1757.06 -2205.49 -2095.70

(162.33)*** (205.42)*** (200.05)*** (243.49)***

EDU(Head) 206.859 303.965 2419.93 -77.2690

(974.48) (1391.6) (1409.1)* (1681.8)

INCM(Head) 0.36831 0.43089 0.20432 0.18434 (0.0911)*** (0.1530)*** (0.1202)* (0.1767)

AGEDIFF 608.675 106.786 825.738 68.0674

(425.04) (441.66) (581.13) (15.872)***

EMPLDIFF 7093.37 8643.63 6286.54 1458.11

(5325.4) (6647.9) (7266.6) (7906.7)

EDUDIFF 958.171 747.966 -396.553 918.883

(977.08) (1104.1) (1274.2) (1546.4)

INCMDIFF -0.31392 -0.39634 -0.19916 -0.21584 (0.09059)*** (0.1236)*** (0.1233) (0.1496)

HVALUE 0.40666 0.34411 0.35560 0.44906

(0.0269)*** (0.0316)*** (0.0220)*** (0.0429)***

INTRCPT 81104.4 103450 103493 128878

(13984)*** (20489)*** (20552)*** (24139)***

N 1010 1102 1119 1062


16/27

16


Theeducationandemploymentdifferentialvariablesarenotstatisticallysignificantforany

years,sothispaperdisregardsthemasviablemeasuresofbargainingpowerinthis

regression.However,theageandincomedifferentialsaresignificant,butatdifferentyears.

SinceItestifgenderedriskaversionexistsratherthanlookingattrendsinriskaversion

overtime,significanceforallyearsislessessential.Thedatademonstratethattheage

differentialcoefficientispositiveforallyearsandforbothsamples.However,theincome

differentialcoefficientisnegativeforallyearsandforbothsamples.

Thattheagedifferentialcoefficientispositivesupportsthebenchmarkprediction

thatwivesaremoreriskaversethanhusbands.Ifagesuggestsmorebargainingpower,

thenwiveswhoarelessyoungorareolderthantheirhusbandshaverelativelymore

bargainingpower.Therefore,householddecisionsmorestronglyreflectwivestastes.The

positivecoefficientoftheagedifferentialimpliesthathouseholdswithweakerwifesvoice

willleveragemore.However,theincomedifferentialcoefficientrejectsthebenchmark

prediction.Sincetheincomedifferentialcoefficientisnegative,itsuggeststhathouseholds

withstrongerwifesvoicewillleveragemore.Thatis,wiveshavestrongertastesforrisk.

ThesecondmethodoftestingthesecondpredictionusesasimilarOLSregressionas

equation(9)butexcludesthehusband/wifedifferentialvectorZbecausethedataonlyuses

singleperson-headedhouseholds.

MORTGi=0i+1iXi+2iYi+3i*INCMi+4i*HVALUEi+i (11)

Althoughcomparingcohabitingandsingleperson-headedhouseholdsignoreshow

thedistributionofcostsaffectsriskbehaviorthespousewhoincurslesscostcould

exhibitamoralhazarditdoesprovideinsightonhowgenderitselfaffectsrisktolerance.


17/27

17

Liketheprevioustest,Iusetheintercepttermandthecoefficientsforincomeandhousing

valueasleveragingmetrics.Agreaterinterceptorcoefficientimpliesmoreexogenous

leveragingorleveraginggivenasetincomeorhousingvalue,respectively.Theresultsfor

2003and2009aredemonstratedinTable4.

Table4:Householdleveragingamongsinglemenandwomen:OLSmodel;DependentVariable:

REMAININGPRINCIPALMORTGAGE

2003 2009

Variable Male Female Male Female

AGE -1045.72 -711.786 -1438.34 -1789.36

(203.56)*** (157.91)*** (401.71)*** (253.72)***

EDU 2177.17 -249.786 4597.05 471.003 (1239.3)* (1002.4) (2421.8)* (1327.6)

INCM 0.01746 0.29913 -0.00331 0.30600

(0.0986) (0.1168)** (0.1941) (0.1664)*

EMPLMT 2581.32 8282.05 -868.814 -12433.0

(8565.2) (6025.9) (17554) (11462)

HVALUE 0.45602 0.31251 0.39206 0.41516

(0.0484)*** (0.0529)*** (0.0600)*** (0.0740)***

INTRCPT 32241.1 56853.9 45485.5 121290

(17031)* (15481)*** (47134) (23716)***

N 256 461 232 538


Forsinglefemale-headedhouseholds,theintercept,income,andhousingvaluecoefficients

arestatisticallysignificant,whereasonlythehousingvaluecoefficientissignificantfor

singlemale-headedhouseholds.Thefemaleintercepttermisgreaterforbothyears,andin

2009,thefemaleinterceptisgreaterthanthemaleinterceptplustwostandarddeviations.

Likewise,thefemalecoefficientforincomeisconsistentlygreaterthanthemalecoefficient.

In2003,themalehousingvaluecoefficientisgreaterthanthefemalecoefficient,butby

2009,thefemalecoefficientsurpassesthemalecoefficient.


18/27

18

Theinterceptsandincomecoefficientssuggestthatin2003and2009,femaleswere

lessriskaversethanmales.Femalesdemonstratedsignificantlyhigherinterceptsand

incomecoefficients.Femalesleveragedmoreonrealestateexogenouslyandrelativeto

theirincomes.However,thehousingvaluecoefficientsdemonstratethatmalesleveraged

morerelativetohousingvaluein2003,butby2009,themalesdecreasedtheirleverage

andfemalesincreasedtheirleverage.

Theresultsofthetwomethodsdonotsupportfromthebenchmarkpredictionthat

wivesaremoreriskaversethanhusbands.Theonlyvariablethatsupportsthebenchmark

istheagedifferentialcoefficient.Nonetheless,apositiveincomedifferentialcoefficientand

positivefemaleinterceptandcoefficientssuggestthatfemalesareactuallylessriskaverse

thanmales.

Thethirdbenchmarkpredictsthatincreasedfemaleemploymentintheformallabor

marketincreaseswivesbargainingpower.Iusechildcareasaproxyforfemalebargaining

power.Phillips,etal.(1997)arguethattraditionalgendernormsstillconsiderchildcareto

bethewivesdutydespitechildrenbeingapublicgood.Greaterproportionsofwives

incomepayformarket-substitutedchildcare.Thus,householdswherewiveshavemore

bargainingpoweraremorelikelytopurchasepaidchildcare.

P(DAYCARE=1|!!"#$)P(DAYCARE=1|!!!

!"#$)>0 !!"#$>!!!!"#$ (12) Althoughtheabsoluteproportionofhouseholdsthathadpaidchildcaredecreased

duringtheonsetofthecrisis,Iexploretheisolatedeffectsofbargainingpowertrendson

paidchildcare.Itreatthefinancialcrisisasanaturalexperimentonhouseholdexpenditure

behavior.Sincedisparitiesbetweenhusbandandwifesincomeandemployment

convergedin2009,bargainingpowermodelsshouldpredictthatwivesgainedrelative


19/27

19

influenceduringthecrisis.Householdexpendituresshouldhavemorestronglyreflected

wivespreferences,namelypaidchildcare.Itestthispredictionusingalogitregression

withthedummyvariableDAYCARE,whetherthehouseholdhaspaidchildcare,asthe

dependentvariable.

P(DAYCAREi=1|X,Y,Z)=0i+1iXi+2iYi+3iZi+i (13)

Ifthereisnobargainingpowerdifference,thenafallineithermaleorfemaleemployment

ceterisparibusshoulddecreasechildcareequallybecausetheunemployedspousewill

substitutemarketproductionforhouseholdproduction.Positivecoefficientsforthe

husband/wifedifferentialvectorsuggestthatchildcareisinthehusbandsdomain,

whereasnegativecoefficientssuggestthatchildcareisinthewifesdomain.

TheresultsareillustratedinTable5andTable6.Iusesimilaradjustmentsfor

industryemploymentinthelattertable.

Table5:Probabilityoftwo-personandchildhouseholdwithchildinnon-householddaycare:Logitmodel

DependentVariable:DAYCARE

Variable 2003 2005 2007 2009

AGE(Head) -0.0912 -0.0740 -0.0642 -0.0734

(0.0122)*** (0.0127)*** (0.0143)*** (0.0154)***

EDU(Head) 0.1743 0.0873 0.1403 0.1939

(0.0501)** (0.0489)* (0.0530)*** (0.0575)***

DEBTx10-6 1.3645 3.5496 4.6849 3.3038

(4.0063) (3.6327) (3.4109) (2.995)

HEQUTYx10-6 -2.2124 -0.0763 -1.7900 -0.0468

(1.0446)** (0.5597) (0.8636)** (0.6518)

INCMx10-6 5.0147 3.2095 3.3245 0.1608

(2.2859)** (2.1472) (1.6081)** (1.9976)

INCMDIFFx10-6 -8.2887 -4.2782 -3.5281 -3.6092

(2.3644)*** (2.0934)** (1.6081)** (2.2145)*

EMPLDIFF -0.9427 -0.7236 -0.9992 -0.4166

(0.2376)*** (0.2386)*** (0.3011)*** (0.2579)


20/27

20

EDUDIFF -0.1046 -0.0359 -0.1368 -0.1464

(0.0484)** (0.0506) (0.0578)** (0.0603)**

INTRCPT 0.0495 0.3163 -0.8045 -1.6790

(0.7605) (0.7719) (0.8664) (0.9265)

N 990 934 826 759PseudoR2 0.1228 0.0712 0.0886 0.0745NoteStandarderrorsareinparentheses *Significantatthe10%level;**Significantatthe5%level;***Significantatthe1%level

Table6:Probabilityoftwo-personandchildhouseholdwithchildinnon-householddaycare;adjustedfor

industry:Logitmodel

DependentVariable:DAYCARE

Variable 2003 2005 2007 2009

AGE(Head) -0.0641 -0.0470 -0.0526 -0.0498

(0.0122)*** (0.0106)*** (0.0118)*** (0.0110)***

EDU(Head) 0.1306 0.0700 0.1764 0.1704

(0.0613)** (0.0547) (0.0568)*** (0.0551)***

DEBTx10-6 7.4530 14.3315 3.2875 6.7646

(6.5850) (5.2736)** (3.4820) (3.1985)**

HEQUTYx10-6 -2.6112 0.1456 -0.8859 -0.8059

(1.5175)* (0.7145) (0.8733) (0.9309)

INCMx10-6 13.6091 5.5764 13.4790 0.7889

(5.8967)** (5.1235) (4.1473)*** (4.1786)

INCMDIFFx10-6 -15.9895 -5.3082 -14.6487 -5.7108

(5.0292)*** (4.4213) (4.1943)*** (3.7596)

EMPLDIFF -1.5514 -1.7372 -1.4255 -1.0128

(0.3652)*** (0.3817)*** (0.3814)*** (0.2897)***

EDUDIFF -0.1024 -0.0548 -0.1185 -0.1770

(0.0558)* (0.0501) (0.0522)** (0.0502)***

INTRCPT -0.4817 -0.4229 -1.7297 -1.5425

(0.8496) (0.7584) (0.8079)** (0.7767)**

N 782 874 876 842

PseudoR2 0.1172 0.0790 0.0956 0.0818NoteStandarderrorsareinparentheses *Significantatthe10%level;**Significantatthe5%level;***Significantatthe1%level

Theresultsshownegativecoefficientsforalldifferentialvariablesineveryyearfrom2003

to2009:educationdifference,incomedifference,andemploymentdifference,whichaffirm


21/27

21

pastempiricalstudiesthatconsiderchildcaretobethewifesrole.Iinvestigatethetrends

inthemagnitudesofthesevariablestomeasurechangesinbargainingpower.Areduction

inmagnitude(anincreaseinvalue)ofthecoefficientsindicatesthateitherculturalnorms

areerasingdistinctmaleandfemalerolesinchildcare,whichseemsdoubtfulinthispapers

six-yeartimeframe,orthatthehusband/wifedifferentialisbecominglessunlikelytoaffect

ahouseholdspaidchildcare.

Theunadjustedandadjustededucationdifferentialcoefficients,thoughconsistently

negative,donotofferconclusiveinterpretationssincetheirtrendsareirregularand

unsubstantial.Thevariableisalsonotidealbecauseitistime-invariant.Ithusexcludethe

educationaldifferentialvariablesinceitdoesnotappeartobeausefulmetricfor

bargainingpower.Theunadjustedincomedifferentialcoefficient,however,demonstratesa

negativetrend,especiallyfrom2003to2005whenitsmagnitudehalved.Itsmagnitude

continuestodecreaseandstaysrelativelyconstantduringthecrisis.Theadjustedincome

differentialcoefficientisirregular,largelypeakingin2003and2007;itprovideslittle

informationaboutthetrends.Theunadjustedandadjustedemploymentdifferential

coefficientsdemonstratefluctuatingdeclinesintheirmagnitudesforallyears.Between

2007and2009,thecoefficientsdeclinethemost.Fortheadjustedsample,thecoefficient

dropsfromapproximately-1.4to-1.0,andtheunadjustedsampleisonlystatistically

significantattenpercentsize,suggestingthereisonlyweakevidencethatthevalueis

negative.

Theresultsoftheincomeandemploymentdifferentialcoefficientsagreewiththe

benchmarkpredictions.WhiletheEastAsiancrisismodelssupposedthatfemale

bargainingpowerdecreasedinAsia,thedisparateeffectofthe2008crisisonmalesalaries


22/27

22

andemploymentshouldposittheinverse.Theevidencesuggeststhatwivesbargaining

powerdidincrease,thoughithadbeengraduallyincreasingpriortothecrisis,whichcould

beexplainedbythenarrowinginhusband/wifesalariesandemploymentinallyearsfrom

2003to2008.Theemploymentdifferentialcoefficientsofferthemostconclusiveevidence.

Thelossofstatisticalsignificanceintheunadjustedemploymentcoefficientpointsto

robustgainsinfemalebargainingpowerafter2007.Likewise,thesubstantialdeclineinthe

adjustedemploymentcoefficientsmagnitudeimpliesstrongerfemalevoiceinthe

household.

PartV:Conclusion

Whilethe2008financialcrisisgeneratedtremendouseconomiccoststowhich

orthodoxindicesofeconomicwellbeing,suchasconsumption,GDP,andcorporateprofits,

havepointed,theimpactshavealsoundulatedtodifferentpartsoftheeconomyinless

detectableways.Exogenousshocksofthecrisisdecliningaggregatedemandandcredit

inaccessibilityalsoaffectedintrahouseholdbehavior.Sincethecrisisinitiallyimpacted

male-concentratedindustries,householdsexperiencedshiftsinbargainingpowers,budget

decisions,andriskaversion.Thispaperexploresthegenderedeffectsofthe2008crisison

householdsbyexaminingthreehypothesesputforthbymodelsthatlookedatthe1997

EastAsiancrisis.Ituses2003to2009PSIDdatatoempiricallytestthem.

Firstisthathouseholdsignificantlyleveragedduringtheonsetofthecrisis.Data

fromapanelsampleofabout2100UShouseholdsillustratesthathouseholdsdid

exogenouslyleveragerealestatebyincreasingtheamountandnumberofmortgages,even

whencontrolledforincomeandhousingprices.Thisimpliesthathouseholdsleverageddue


23/27

23

touncontrolledfactors,suchaschangingrisktoleranceorinterestrates.Secondisthat

wivesaremoreriskaversethanhusbandsare.Usingtwomethodologiestotestthe

hypothesis,Icomparerisktoleranceamongsinglewomenandsinglemen,andlooksat

bargainingpowervariablesinmortgageregressions.Bothtestsdisagreewiththe

hypothesis,andinterestinglyoffersomeevidencethathusbandswererelativelymorerisk

averse.Thirdisthatconvergingincomesandemploymentofwivestohusbandsduringthe

crisisincreasedfemalebargainingpower.Usingpaidchildcareasameasureoffemale

bargainingpower,Iinvestigatetheinfluenceofincomeandemploymentdifferentials

betweenhusbandsandwivesonprobabilitiesofhavingpaidchildcare.Theresultsindicate

thatwivesbargainingpowerhadbeenincreasingfrom2003to2007,butsignificantly

jumpedfrom2007to2009.

SinceIemploymodelscreatedusingevidencefromthe1997EastAsiancrisis,the

benchmarkpredictionsandtheempiricalresultsmaydisagreebecauseofdrawbacksof

extendingtheEastAsiantheoryassumptionstotheUnitedStates.Theempirical

conclusionsofthispaperendorsearevisitationoftheintersectionofhouseholdbargaining

powerandfinancialcrises.Italsoillustratesthenecessityoffurtherresearchinto

householdbargainingmodelsfordevelopednations.Especiallyforpolicymakersand

economistswhoattempttorealizethefullcostsoffinancialcrisis,anunderstandingthat

economiccrisesalsoechointohouseholds,andthatunemploymentandcreditshocks

disparatelyaffectmenandwomenisessential.


24/27

24

Notes:

1Theirendogeneity,however,islesspronouncedfordevelopedcountries,suchasthe

UnitedStates,sincetheelasticitybetweenincomeandcreditisweakerthanfordeveloping

countries(Knell,Stix2005).

2Thisisespeciallytrueindevelopingcountrieswherewomenreceivemostoftheircredit

frominformalcreditmarkets.

3Iarguethatisbecausehealthcareisrelativelyinelastictowealtheffects.In-homefood,however,isaninferiorgood,sinceout-of-homefoodexpendituredecreased.

4Thispaperconsidersthetermsmenasmarketproducers,andwomenashousehold

producers.Iacknowledgethatmanyhouseholdsetupsexist,suchasfemale-headed,single-

parent,andsame-sexhouseholds.Drago,etal(2005);Blau,etal(2005);Black,etal(2007)exploretheeconomicsofnontraditionalhouseholdstructures,andsomeanalyticsoverlap.

5Recentlyindevelopedcountries,unemploymentratesofmenarebecomingmoreelastic

tothebusinesscycle,whereasfemaleunemploymentratesarelesselastic.Indeveloping

countries,however,femalestypicallybearhighercostsofeconomicbusts.6Althoughtheirpaperdoesnotspecifyafunctionalform,Iproposetwopossibleformsthat

agreewiththeiranalysis.

H=!!!

!orH= !(1 !) [0,1],C>0

IprovidethefunctionalformtotestwhetherempiricsconfirmFloroandDymskis

graphicalanalysis.

7Althoughsignificantliteraturecitethatwomenaremoreriskaversethanmen,thesestudiesfocusondevelopingcountriesandsinglewomen.SeeKaber(2002),and

Jianakoplosetal.(1998).


25/27

25

Table1:SummarystatisticsMeanswithstandarddeviationsinparentheses

Variable 2003 2005 2007 2009

DAYCARE 0.2129 0.1917 0.1727 0.1764

(0.4095) (0.3938) (0.3782) (0.3813)

AGE(Head) 40.351 41.160 42.078 43.082

(8.2242) (8.0741) (7.9522) (7.9021)

EDU(Head) 13.328 13.310 13.261 13.204

(2.6415) (2.6782) (2.6892) (2.7457)

EDU(Wife) 13.442 13.473 13.483 13.488

(2.6329) (2.6173) (2.6126) (2.6685)

EDUDIFF -0.1263 -0.1863 -0.2514 -0.3016

(2.1248) (2.0867) (2.0780) (2.0825)

DEBT 8410.6 9545.1 12016 14251

(19116) (20104) (24369) (35005)

HEQUTY 68556 98932 126810 91473

(113877) (162503) (199726) (149953)

INCM(Head) 24243 23968 29751 26745

(69937) (41273) (110814) (45795)

INCM(Wife) 8979.9 9681.3 11505 12255

(21617) (23601) (27631) (27327)

INCMDIFF 15346 14504 18271 14715 (69890) (45231) (109342) (49890)

EMPL(Head) 0.9727 0.9634 0.9575 0.9491

(0.1629) (0.1877) (0.2017) (0.2198)

EMPL(Wife) 0.7989 0.8019 0.8035 0.8051

(0.4009) (0.3986) (0.3975) (0.3962)

EMPLDIFF 0.1737 0.1614 0.1539 0.1439

(0.4216) (0.4307) (0.433) (0.4389)


26/27

26

References:

Agarwal,B,BargainingandGenderRelations:WithinandBeyondtheHousehold,

FeministEconomics,Vol.3,No.1(1997)

Bittman,England,Folbre,Sayer,andMatheson,BargainingandTimeinHouseholdWork,AmericanJournalofSociology,Vol.109,No.1(July,2003)

Black,D.,Sanders,S.,andTaylorL.,TheEconomicsofLesbianandGayFamilies,Journalof

EconomicPerspectives,Vol.21,No.2,(2007)

BlauF.,Ferber,M.,andWinklerA.,EconomicsofWomen,Men,andWork,5thedition,PrenticeHall(August,2005)

Campello,M.,GrahamJ.,Harvey,C.,Therealeffectsoffinancialconstraints:Evidencefromafinancialcrisis,JournalofFinancialEconomics,Vol.97,no.3(Sept.,2010)

Doss,C.TestingModelsofIntrahouseholdResourceAllocation,WorldDevelopment,Vo.

24,No.10(1996)

Drago,R.,Black,D.,andWooden,M.,Femalebreadwinnerfamilies:Theirexistence,

persistenceandsources,JournalofSociology,Vol.41,No.4(Dec.,2005)

Dymski,G.,Carvalho,O.,Costa,O.,Fern,L.,andDePaula,O.,AssetBubblesandMinsky

CrisesinEastAsia:ASpatializedMinskyApproach,JeromeLevyEconomicsInstitute

Workingpaper,BardCollege(Sept.,1999)

Floro,M.andDymski,G.,FinancialCrisis,Gender,andPower:AnAnalyticalFramework,WorldDevelopment,Vol.28,No.7,(July2000)

"FocusonPricesandSpending|ConsumerExpenditure|Volume2,Number12."U.S.

BureauofLaborStatistics.U.S.BureauofLaborStatistics.Web.10May2012.

.

Hartmann,H.,GenderImplicationsoftheFinancialCrisisintheUnitedStates,Institutefor

WomensPolicyResearch,(2009)

Hurd,M.andRohwedderS.,EffectsofFinancialCrisisandGreatRecessiononAmerican

Households,NBERWorkingpaper16407(Sept.,2010)

Jianakopolos,N.,Bernasek,A.,AreWomenmoreRiskAverse?EconomicInquiry,Vol.36,

No.4(Oct.1998)

Kabeer,N.,Women,WagesandIntra-householdPowerRelationsinUrbanBangladesh,

DevelopmentandChange ,Vol.28,No.2,(April1997)


27/27

27

KnellM.andStixH.,TheIncomeElasticityofMoneyDemand:AMeta-AnalysisofEmpirical

Results,JournalofEconomicSurveys,Vo.19.No.3(July,2005)

Lam,D.,MarriageMarketsandAssortativeMatingwithHouseholdPublicGoods:

TheoreticalResultsandEmpiricalImplications,JournalofHumanResources,Vol.23,No.4(1988)

Mian,A.,Sufi,A.,HouseholdLeverageandtheRecessionof2007to2009,International

MonetaryFund,(Oct.,2009)

Muthoo,A.,ANon-TechnicalIntroductiontoBargainingTheory,WorldEconomics,Vol.1,No.2,(2002)

"NationalEconomicAccounts."BEA.Web.10May2012..

Phipps,S.andBurton,P.,SharingwithinFamilies:ImplicationsfortheMeasurementof

PovertyamongIndividualsinCanada,TheCanadianJournalofEconomics,Vol.28,No.

1,(Feb.,1995)

Phipps,S.,Burton,P.,WhatsMineisYours?TheInfluenceofMaleandFemaleIncomeson

PatternsofHouseholdExpenditureEconomica,Vol.65,No.260(1998)

Pollak,R.Bargainingpowerinmarriage:Earnings,Wagerates,andHousehold

production,NBERWorkingpaper11239(Mar.,2005)

Ruggieri,D.,Genderperspectivesofthefinancialandeconomiccrisis,InternationalJournalofGreenEconomics,Vol.4,No.3(2010)

SullivanJ.,BorrowingDuringUnemployment:UnsecuredDebtasaSafetyNet,Journalof

HumanResources,Vol.43,No.2(2008)

WalbyS.,GenderandtheFinancialCrisis,UNESCO,(April,2009)

Gendering the Financial Crisis

Documents