Skip to main content

Crisis at home: mancession-induced change in intrahousehold distribution


The Great Recessions was essentially a “mancession” in countries like Spain, the UK, or the USA, i.e., it hist men harder than women for they were disproportionately represented in heavily affected sectors. We investigate how the mancession, and more generally women’s relative opportunities on the labor market, translates into within-household redistribution. Precisely, we estimate the spouses’ resource shares in a collective model of consumption, using Spanish data over 2006–2011. We exploit the gender-oriented evolution of the economic environment to test two original distribution factors: first the regional-time variation in spouses’ relative unemployment risks, and then the gender-differentiated shock in the construction sector (having a construction sector husband after the outburst of the crisis). Both approaches conclude that the resource share accruing to Spanish wives increased by around 7–9% on average, following the improvement of their relative labor market positions. Among childless couples, we document a 5–11% decline in individual consumption inequality following the crisis, which is essentially due to intrahousehold redistribution.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2


  1. 1.

    Source: OECD. Main Economic Indicators. Labor: Labor market statistics.

  2. 2.

    So far, the existing literature on the mancession essentially focused on measuring the gender gap emerging with the economic crisis and assessed the vulnerability across different demographic groups (Sierminska and Takhtamanova 2010; Hoynes et al. 2012; Cho and Newhouse 2013). While there exists widespread evidence over the redistributive impacts of economic crises between the households, little is known about the changes in the relative welfare of individuals living in these households. Notwithstanding, related studies analyze the likely effect of the mancession on time allocation among spouses (Starr 2014; Gorsuch 2016; Heiland et al. 2017; Morrill and Pabilonia 2015). Probably the closest study to ours is Lacroix and Radtchenko (2011), who analyze the changes in real earnings among Russian couples and how it affected the sharing rule around the 1998 financial crisis. Notably, they focus on labor supply decisions while we focus here on an identification based on consumption data. Other studies have specifically focused on the gender-differentiated effect provided by the collapse of the construction sector: Farré et al. (2015) instrument the unemployment status of individuals with a measure of their exposure to the bursting housing bubble in Spain to document the medium run impact of unemployment on their mental health. Aparicio-Fenoll (2016) exploits the exogenous variation in return to education for males with respect to females induced by the Spanish housing boom to identify their causal impact on enrollment in post-compulsory education and on completion.

  3. 3.

    Mazzocco (2007), Lise and Yamada (2014), and Theloudis (2015) consider dynamic versions of the collective consumption model.

  4. 4.

    The idea of combining data on people living alone and in couples to retrieve the complete resource sharing rule was already applied in the context of labor supply by Couprie (2007) and calibrated models in a project presented in Laisney et al. (2003). Other applications can also be found in Lise and Seitz (2011), Bargain and Donni (2012), and Cherchye et al. (2012). The assumption of stable preferences across marital status is necessary to make “situation comparisons” (i.e., compare the welfare of adults when living alone or with others) in the terminology of Pollak (1991). Recent studies attempt to test this assumption. While Brugler (2016) shows that consumer preferences significantly differ across singles and members of couples, Hubner (2017) constructs a nonparametric test of the stable preference assumption and rejects it on Russian data.

  5. 5.

    The sharing functions \(\eta _{i}(\boldsymbol {p},\boldsymbol {z})\), \(i=m,w\), are differentiable, comprised between zero and one, and sum up to unity.

  6. 6.

    Bargain and Donni (2012) show that identification results still hold, theoretically at least, when sharing functions depend on total expenditure. Also, this assumption can be mitigated in empirical applications by including measures of household wealth other than total expenditure in resource shares. The fact that scaling functions representing scale economies are independent of expenditure—and hence of the utility level—at which they are evaluated is the “independence of the base” assumption made by Lewbel and Pendakur (2008) in order to retrieve these Engel scales. It is similar to the restriction of the same name in the equivalence scale literature (Lewbel 1991), but it concerns individual utility functions rather than aggregated household utility functions.

  7. 7.

    The right-hand side puts some structure on individual budget shares as a result of the “independence of the base” restriction, using the “basic” budget share function \({w_{i}^{k}}\left (\cdot ,\boldsymbol {p},\boldsymbol {z}_{i}\right ) \) defined for single individuals. The consequence of this assumption is that the budget share equations of person i living in a couple differ from when living alone only in that they are translated over by the elasticity \({\lambda _{i}^{k}}(\boldsymbol {p},\boldsymbol {z})\) and depend on her/his individual resources adjusted by the scaling \(s_{i}(\boldsymbol {z})\). This property of “shape invariance,” as defined by Pendakur (1999), implicitly means that single individuals are used as the demographic structure of reference.

  8. 8.

    We impose homogeneity by expressing all prices relative to the price of the remaining goods. In our simple model with an assignable good and a composite good, Slutsky symmetry mechanically holds because the vector of log relative prices boils down to a scalar.

  9. 9.

    In our simple model, this simply comprises the log relative price of male/female clothing with respect to the composite good for each regional time cell.

  10. 10.

    These are obtained from reduced-form estimations of x on all exogenous variables used in the model plus some excluded instruments (a third order polynomial in household disposable income).

  11. 11.

    Time and regional variation in gender-specific unemployment rate is also used for Britain by Anderberg et al. (2015) who look at the incidence of domestic violence.

  12. 12.

    The employment reduction in this sector has varied from 18 to 55% across regions. Ten percentage points of the post-crisis unemployment rate were imputable to the construction sector alone (Pissarides 2013). The raw correlation between the changes in total and construction employment shares across Spanish regions is 0.70.

  13. 13.

    The model being essentially static, we refrain from including expenditure on education or health. This would require adding dynamics and uncertainty in the structural model, which we leave for future research.

  14. 14.

    It is likely that unemployment is more endogenous to the sharing process in a normal situation because of assortative mating and the fact that potential partners internalize each other’s risk of unemployment before marriage. Inversely, in time of crisis, gender-asymmetric shocks on the labor market may lead to more redistribution as the wife would absorb less of the negative income effect. We check this point in alternative estimations whereby we specify the sharing rule with actual unemployment rates interacted with dummies for pre- and post-crisis periods. As hypothesized, we find that the effect of effective unemployment on sharing is indeed very strong during 2009–2011 while statistically inexistent in the pre-crisis period. The coefficient on our variable of interest, the perceived relative unemployment risk, does not change.

  15. 15.

    Additionally, aggregate data on labor participation of Spanish women displays no dramatic break around 2008. Women’s participation growth rate slightly decreased while remaining positive.

  16. 16.

    Complementing Browning et al. (1994) with endogenous participation, Zamora (2011) estimates the sharing rule in Spain on the 1990–1991 EPF data and shows that it switches with the participation regime. The extent to which these results differ may be explained by different modeling choices (our approach allows identify the complete sharing process while it includes participation as a simple shifter in the sharing rule specification) and by data selection. Indeed, Zamora focuses on the period 1990–1991 and women aged up to 66 years of age (we focus on childless women aged 20–44 in a much more recent period), which also explain that her share of female participants is as low as 30.1% (93% in our sample).

  17. 17.

    Only employed individuals are selected, both in our main sample of couples and in samples of single men and women used for identification. The sample size of the couple data is reduced by about 30%.

  18. 18.

    Among men aged 20–44 living in Spain, the proportion of foreign-born decreased from 22% in 2009 to 19% in 2014. Source: Own calculation from Cifras de Poblaciòn, INE.

  19. 19.

    While the model with three goods has the advantage of simplicity and yields robust results, we have also experimented with a richer specification with K = 8 goods. While it was expected to increase the efficiency of the estimates, results were not particularly more precise. While the effects of most explanatory variables, including those of interest (relative unemployment risk in the first specification and post x husband in construction in the second), were similar to those in the model with K = 3 goods, the constants were different so that the resource share accruing to the average Spanish woman was larger (between 50 and 60% depending on the specification). While the reasons for this variability in the sharing rule remain to be explored (see some enlightening in Tommasi and Wolf 2016), we acknowledge the fact that the estimation of a complete demand system relies on stronger assumptions. More precisely, the identifying assumptions must be satisfied for all the goods and not only for exclusive goods—which make it a more demanding condition. The fact that the estimations of the general model were not especially more precise tends to support this explanation. Detailed results are available upon request.

  20. 20.

    Our scale estimates are in line with the seminal work of Lewbel and Pendakur (2008). In their empirical exercise on the 1990 and 1992 Canadian Family Expenditure Surveys, they find that the average woman benefits from 46% of the total household resources. The scale economies (computed for the reference category, i.e., high school graduates over 40) are 0.7 for women and 0.8 for men, but the hypothesis that scales are indeed under 1 cannot be rejected by the data, due to the large standard error associated with the estimate. Other studies (i.e., Bargain and Donni (2012) for France, or Bargain et al. (2014) for Côte d’Ivoire) find slightly larger scale economies, equally imprecisely estimated.

  21. 21.

    Here, we use individual rather than regional variation so that accounting for regional fixed effects is less of a necessity. Keeping them actually leaves the results unchanged.

  22. 22.

    Note that our measure of construction is not strictly pre-crisis. However, sample selection due to shifts across industries should be limited because for the unemployed, we use information on prior occupation. Sample selection would arise in case of massive reemployment of the newly dismissed construction workers, which is very unlikely in the immediate post-crisis context (see Farré et al. 2015). In fact, according to aggregate data from INE, the construction sector represents 12.2% of total employment over 2006–2007, and much less (8.7%) over 2008–2011. By contrast, in our data, the proportion of husbands in the construction sector is remarkably stable (13.3% in 2006–2007, 12.1% afterwards).

  23. 23.

    As a matter of fact, very few studies have attempted to endogenize household formation and dissolution in a collective model of consumption. Recently, Mazzocco et al. (2014) study the relationship between household consumption decisions (on labor supply and savings behavior) and marital choices. Cherchye et al. (2017) add the assumption that marriages are stable. They show that combining it and the standard assumption of Pareto-efficiency consumption decisions generates strong testable implications for household consumption patterns.


  1. Anderberg D, Rainer H, Wadsworth J, Wilson T (2015) Unemployment and domestic violence: theory and evidence. The Economic Journal. forthcoming

  2. Aparicio-Fenoll A (2016) Returns to education and educational outcomes: the case of the spanish housing boom. J Human Capital 10(2):235–265

    Article  Google Scholar 

  3. Banks J, Blundell R, Lewbel A (1997) Quadratic Engel Curves and Consumer Demand. Rev Econ Stat 79(4):527–539

    Article  Google Scholar 

  4. Bargain O, Donni O (2012) The measurement of child costs: a Rothbarth-type method consistent with scale economies and parents bargaining. Eur Econ Rev 56 (2010-30):792–813

    Article  Google Scholar 

  5. Bargain O, Donni O, Kwenda P (2014) Intrahousehold distribution and poverty: evidence from Côte d’Ivoire. J Dev Econ 107:262–276

    Article  Google Scholar 

  6. Bentolila S, Cahuc P, Dolado JJ, Le Barbanchon T (2012) Two-tier labour markets in the great recession: France versus Spain. Econ J 122(562):155–187

    Article  Google Scholar 

  7. Blundell R, Robin JM (1999) Estimation in large and dissaggregated demand systems: an estimator for conditionally linear systems. J Appl Econ 14(3):209–232

    Article  Google Scholar 

  8. Bourguignon F, Browning M, Chiappori P-A (2009) Efficient intra-household allocations and distribution factors: implications and identification. Rev Econ Stud 76(2):503–528

    Article  Google Scholar 

  9. Browning M, Bourguignon F, Chiappori P-A, Lechene V (1994) Income and outcomes: a structural model of intrahousehold allocation. J Polit Econ 102(6):1067–1096

    Article  Google Scholar 

  10. Browning M, Chiappori P-A, Lewbel A (2013) Estimating consumption economies of scale, adult equivalence scales, and household bargaining power. Rev Econ Stud 80(4):1267–1303

    Article  Google Scholar 

  11. Brugler J (2016) Testing preference stability between couples and singles. Econ Lett 142:15–17

    Article  Google Scholar 

  12. Cherchye L, De Rock B, Vermeulen F (2012) Economic well-being and poverty among the elderly: an analysis based on a collective consumption model. Eur Econ Rev 56(6):985–1000

    Article  Google Scholar 

  13. Cherchye L, Demuynck T, De Rock B, Vermeulen F (2017) Household consumption when the marriage is stable. Am Econ Rev 107(6):1507–1534

    Article  Google Scholar 

  14. Chiappori P-A (1988) Rational household labor supply. Econometrica 56(1):63–90

    Article  Google Scholar 

  15. Chiappori P -A, Mazzocco M (2016) Static and intertemporal household decisions. Journal of Economic Literature 55(3):985–1045

    Article  Google Scholar 

  16. Cho Y, Newhouse D (2013) How did the great recession affect different types of workers? Evidence from 17 middle-income countries. World Dev 41:31–50

    Article  Google Scholar 

  17. Congregado E, Golpe AA, van Stel A (2011) Exploring the big jump in the Spanish unemployment rate: evidence on an “Added-Worker” effect. Econ Modell 28(3):1099–1105

    Article  Google Scholar 

  18. Couprie H (2007) Time allocation within the family: welfare implications of life in a couple. Econ J 117(516):287–305

    Article  Google Scholar 

  19. Farré L, Fasani F, Mueller H (2015) Feeling useless: the effect of unemployment on mental health in the great recession. IZA Discussion Papers 9235, Institute for the Study of Labor (IZA)

  20. Gorsuch MM (2016) Decomposing the increase in men’s time on childcare during the great recession. Rev Econ Househ 14(1):53–82

    Article  Google Scholar 

  21. Heiland F, Price J, Wilson R (2017) Maternal employment and time investments in children. Rev Econ Househ 15(1):53–67

    Article  Google Scholar 

  22. Hoynes H, Miller DL, Schaller J (2012) Who suffers during recessions? J Econ Perspect 26(3):27–48

    Article  Google Scholar 

  23. Hubner S (2017) Its complicated: a nonparametric test of preference stability between singles and couples

  24. Lacroix G, Radtchenko N (2011) The changing intra-household resource allocation in Russia. J Popul Econ 24(1):85–106

    Article  Google Scholar 

  25. Laisney F, Beninger D, Beblo M (2003) Welfare analysis of fiscal reforms: does the representation of the family decision process matter? Evidence for Germany. Technical report, ZEW Discussion Papers

  26. Lewbel A (1991) Cost of characteristics indices and household equivalence scales. Eur Econ Rev 35(6):1277–1293

    Article  Google Scholar 

  27. Lewbel A (2003) Calculating compensation in cases of wrongful death. J Econ 113(1):115–128

    Article  Google Scholar 

  28. Lewbel A, Pendakur K (2008) Estimation of collective household models with Engel curves. J Econ 147(2):350–358

    Article  Google Scholar 

  29. Lise J, Seitz S (2011) Consumption inequality and intra-household allocations. Rev Econ Stud 78(1):328–355

    Article  Google Scholar 

  30. Lise J, Yamada K (2014) Household sharing and commitment: evidence from panel data on individual expenditures and time use. IFS Working Papers W14/05, London

    Book  Google Scholar 

  31. Mazzocco M (2007) Household intertemporal behavior: a collective characterization and a test of commitment. Rev Econ Stud 74(3):857–895

    Article  Google Scholar 

  32. Mazzocco M, Ruiz C, Yamaguchi S (2014) Labor supply and household dynamics. Amer Econ Rev 104(5):354–359

    Article  Google Scholar 

  33. Morrill MS, Pabilonia SW (2015) What effects do macroeconomic conditions have on the time couples with children spend together? Rev Econ Househ 13(4):791–814

    Article  Google Scholar 

  34. Pendakur K (1999) Semiparametric estimates and tests of base-independent equivalence scales. J Econ 88(1):1–40

    Article  Google Scholar 

  35. Pissarides CA (2013) Unemployment in the great recession. Economica 80 (319):385–403

    Article  Google Scholar 

  36. Pollak RA (1991) Welfare comparisons and situation comparisons. J Econ 50 (12):31–48

    Article  Google Scholar 

  37. Sierminska E, Takhtamanova Y (2010) Job Flows, Demographics and the Great Recession. CEPS/INSTEAD Working Paper Series 2010-41, CEPS/INSTEAD

  38. Starr MA (2014) Gender, added-worker effects, and the 2007–2009 recession: looking within the household. Rev Econ Househ 12(2):209–235

    Article  Google Scholar 

  39. Theloudis A (2015) Wages and family time allocation. Technical report, mimeo

  40. Tommasi D, Wolf A (2016) Overcoming weak identification in the estimation of household resource Shares. Working Papers ECARES ECARES 2016-12, ULB – Universite Libre de Bruxelles

  41. Vermeulen F (2002) Collective household models: principles and main results. J Econ Surv 16(4):533–564

    Article  Google Scholar 

  42. Zamora B (2011) Does female participation affect the sharing rule? J Popul Econ 24(1):47–83

    Article  Google Scholar 

Download references


We thank the two anonymous referees for their many valuable comments and suggestions. We are also grateful to Olivier Donni, Frederic Vermeulen, and participants to the ADRES conference and to seminars at ISER (Essex), THEMA (Cergy-Pontoise), AMSE (Aix-Marseille), and LEO (Orleans) for their helpful suggestions. All errors remain ours.

Author information



Corresponding author

Correspondence to Olivier Bargain.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Responsible editor: Junsen Zhang


Appendix A: Model Identification

Identification is obtained in two steps. First, the basic budget share functions \({w_{i}^{k}}\left (\cdot ,\boldsymbol {p},\boldsymbol {z}_{i}\right ) \), for any k and \(i=m,w\), stem from the estimation of (4) on samples of single men and women. They identify the basic budget shares of male and female individuals in couple following the assumption of preferences stability across household types. Second, the resource sharing functions and scaling functions for \(n = 2\) can be identified from a sample of couples. Precisely, the household budget share equation for good k in a couple can be rewritten as:

$$ {W_{2}^{k}}(x,\boldsymbol{p},\boldsymbol{z})=D^{k}(\boldsymbol{p}, \boldsymbol{z}) + \sum\limits_{i=m,w}\eta_{i}(\boldsymbol{p}, \boldsymbol{z})\cdot \left[w_{i}^{k_{i}}\left( x+\log I_{i}(\boldsymbol{z}),\boldsymbol{p}, \boldsymbol{z}_{i}\right) \right] , $$

where \(D^{k}(\boldsymbol {p}, \boldsymbol {z})=\sum \limits _{i=m,w}\eta _{i}(\boldsymbol {p}, \boldsymbol {z})\cdot {\lambda _{i}^{k}}(\boldsymbol {p},\boldsymbol {z} )\). We first eliminate the good-specific function \(D^{k}(\boldsymbol { z})\) by computing the first-order derivative of this expression with respect to x:

$$ \nabla_{x}{W_{2}^{k}}(x,\boldsymbol{p},\boldsymbol{z})=\sum\limits_{i=m,w}\eta_{i}(\boldsymbol{p},\boldsymbol{z})\nabla_{x}{w_{i}^{k}}\left( x+\log I_{i}(\boldsymbol{p},\boldsymbol{z}),\boldsymbol{p},\boldsymbol{z}_{i}\right) , $$

where the left-hand side is known from the data. In the right-hand side, there are only three unknowns (since resource shares sum up to one): \(\eta _{w}\), \(I_{w}\), and \(I_{f}\). Generic identification is therefore obtained with at least three different observations for a constant level of x. With the use of a gender-specific good \(k_{i}\) such as clothing, identification is even strengthen and the formal demonstration is straightforward. Indeed, expression (17) can be rewritten as:

$$ \nabla_{x}W_{2}^{k_{i}}(x,\boldsymbol{p},\boldsymbol{z})=\eta_{i}(\boldsymbol{p},\boldsymbol{z})\nabla_{x}w_{i}^{k_{i}}\left( x+\log I_{i}(\boldsymbol{p},\boldsymbol{z}),\boldsymbol{p},\boldsymbol{z}_{i}\right) , $$

for \(i=m,w\). Differentiating this expression again with respect to x gives the second-order derivative:

$$ \nabla_{xx}W_{2}^{k_{i}}(x,\boldsymbol{p},\boldsymbol{z})=\eta_{i}(\boldsymbol{p},\boldsymbol{z})\nabla_{xx}w_{i}^{k_{i}}\left( x+\log I_{i}(\boldsymbol{p},\boldsymbol{z}),\boldsymbol{p},\boldsymbol{z}_{i}\right) , $$

and taking the ratio of (18) and (19), we have:

$$\frac{\nabla_{x}W_{2}^{k_{i}}(x,\boldsymbol{p},\boldsymbol{z})}{\nabla_{xx}W_{2}^{k_{i}}(x,\boldsymbol{p},\boldsymbol{z})}=\frac{\nabla_{x}w_{i}^{k_{i}}\left( x+\log I_{i}(\boldsymbol{p},\boldsymbol{z}), \boldsymbol{p},\boldsymbol{z}_{i}\right)} {\nabla_{xx}w_{i}^{k_{i}}\left( x+\log I_{i}(\boldsymbol{p},\boldsymbol{z}),\boldsymbol{p},\boldsymbol{z} _{i}\right)} ={\Delta}_{i}^{k_{i}}\left( x+\log I_{i}(\boldsymbol{p}, \boldsymbol{z}),\boldsymbol{p},\boldsymbol{z}_{i}\right) $$

where function \({\Delta }_{i}^{k_{i}}\left (\cdot ,\boldsymbol {z}\right ) \) is known from the first step. This condition uniquely identifies the indifference scales \(I_{i}(\boldsymbol {p},\boldsymbol {z})\) for \(i=m,w\), provided the function \({\Delta }_{i}^{k_{i}}\left (\cdot \right ) \) is not periodic in its first argument – a rather natural requirement. Then, for \(i=m,w\), identification of sharing functions ηi(p, z) follows from (17) and identification of translation functions \(\lambda _{i}^{k_{i}}(\boldsymbol {p}, \boldsymbol {z})\) from (16). Finally, the scaling functions \( s_{i}(\boldsymbol {p},\boldsymbol {z})\) can be computed for \(i=m,w\) from the definition of \(I_{i}(\boldsymbol {p},\boldsymbol {z})\).

Appendix B: Empirical Application

B.1 Engel Curves

We suggest a visual check of Engel curves in Fig. 3. Food, “vice” and housing services are necessity goods: holding prices constant, their demand increases with the total expenditure, but less rapidly. Leisure, clothing, personal care and transport are luxury goods, their shares in the total budget increasing with total expenditure. As previously discussed, the generic identification of the model requires the nonlinearity of the budget share equation for assignable goods with respect to the total log expenditure, which seems the case. A formal test is performed in Table 6. We estimate a reduced form of the model on the subsamples of single individuals and couples, with or without inclusion of Wu-Hausman residuals. Results show that the budget shares for assignable goods are indeed nonlinear with respect to the log of total expenditure. Estimates are not markedly affected by the introduction of Wu–Hausman residuals. We also provide a check for the endogeneity of the total expenditure.

Fig. 3

Engel curves, Kernel weighted local polynomial smoothing

Table 6 Nonlinearities in budget shares of assignable goods, by gender and household structure

Appendix C: Results

As discussed in Section 4.2 of the paper, Table 7 compares the estimates obtained using the restricted sample with our previous results using the main sample. The estimated coefficients for the budget share equations of the baseline model with 3 goods are presented in Table 8. Men and women’s demand for clothing have the same determinants in terms of sign and magnitude. In line with the nonlinearity identification condition, it depends positively on the log of total expenditure, and negatively on its square, for both men and women. Besides, women tend to spend more on clothing items if they live in Madrid or Barcelona, and relatively less if they live in a rural area.

Table 7 Comparing main and restricted samples with the unemployment ratio specification
Table 8 Budget share parameters

The parameters of the scaling function are reported in Tables 9 and 10. The theory predicts economies of scale between the resource share (all the consumption is public) and 1 (private consumption). In other words, the closer the estimated scale to 1, the lower is the degree of joint consumption (scale economies) in the household.

Table 9 Scale economies – unemployment risk specification
Table 10 Scale economies – difference-in-difference specification

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Bargain, O., Martinoty, L. Crisis at home: mancession-induced change in intrahousehold distribution. J Popul Econ 32, 277–308 (2019).

Download citation


  • Mancession
  • Intrahousehold allocation
  • Unemployment risk

JEL Classification

  • C3
  • D12
  • D13