How does death of a household member affect consumption? The Case of Mexico

• Název práce v češtině: Vliv úmrtí člena domácnosti na spotřebu
• Název v anglickém jazyce: How does death of a household member affect consumption? The Case of Mexico
• Akademický rok vypsání: 2017/2018
• Typ práce: diplomová práce
• Jazyk práce: angličtina
• Ústav: Institut ekonomických studií (23-IES)
• Vedoucí / školitel: doc. PhDr. Julie Chytilová, Ph.D.
• Řešitel: skrytý - zadáno vedoucím/školitelem
• Datum přihlášení: 22.05.2018
• Datum zadání: 22.05.2018
• Datum a čas obhajoby: 17.09.2019 09:00
• Místo konání obhajoby: Opletalova - Opletalova 26, O105, Opletalova - místn. č. 105
• Datum odevzdání elektronické podoby: 25.07.2019
• Datum proběhlé obhajoby: 17.09.2019
• Oponenti: PhDr. Pavel Vacek, Ph.D.

Seznam odborné literatury

ARDINGTON, Cally, BÄRNIGHAUSEN, Till, CASE, Anne and MENENDEZ, Alicia, 2014. The economic consequences of AIDS mortality in South Africa. Journal of development economics. 2014. Vol. 111, p. 48–60.


BEEGLE, Kathleen, DE WEERDT, Joachim and DERCON, Stefan, 2008. Adult mortality and consumption growth in the age of HIV/AIDS. Economic Development and Cultural Change. 2008. Vol. 56, no. 2, p. 299–326.


DERCON, Stefan, HODDINOTT, John and WOLDEHANNA, Tassew, 2005. Shocks and consumption in 15 Ethiopian villages, 1999-2004. Journal of African economies. 2005. Vol. 14, no. 4, p. 559.


GRIMM, Michael, 2010. Mortality shocks and survivors’ consumption growth. Oxford Bulletin of Economics and Statistics. 2010. Vol. 72, no. 2, p. 146–171.


KIM, Jungho and PRSKAWETZ, Alexia, 2010. External shocks, household consumption and fertility in Indonesia. Population research and policy review. 2010. Vol. 29, no. 4, p. 503–526.


WAGSTAFF, Adam, 2007. The economic consequences of health shocks: evidence from Vietnam. Journal of health economics. 2007. Vol. 26, no. 1, p. 82–100.

Předběžná náplň práce v anglickém jazyce

Motivation
Deaths can represent a significant shock to household consumption. For example, due to expenditure required by funeral costs they can affect both the structure and volume of consumption. Furthermore, a death can lead to a permanent change of household composition. This can possibly result in a loss of productive labor force and constraint the resources available to the respective household. Considering that the reaction of consumption to a death is likely to depend on a variety of factors, its assessment is not a straightforward exercise. However, given its relevance regarding intra-household consumption optimization as well as its importance for policy purposes it received significant attention.

The existing evidence on the impacts of death on household consumption is mixed. Grimm (2010) analyses the Indonesian Household survey and finds that a death of a child, adult man or an older person causes a statistically significant increase in consumption. Kim and Prskawetz (2010) reach a similar conclusion based on the same dataset. Contrarily Beegle et al. (2008) find that a death leads to a negative change in consumption after controlling for other events based on the Kagera Health and Development Survey which monitors the Kagera region in Tanzania. Wagstaff (2007) also finds a statistically significant negative impact of death of working age household member for Vietnamese households. Furthermore Dercon et al. (2005) find that statistical significance of the influence of death on consumption is limited. Therefore, a new assessment of the impacts of death on household consumption will be of significant value.

Reliable longitudinal data are necessary for a robust analysis. In my thesis I will rely on the freely available Mexican Family Life Survey (MxFLS) data. MxFLS provides longitudinal household and individual level data spanning over ten years. These will provide me with the opportunity to study not only the magnitude of the impact itself but also its evolution over time. To the best of my knowledge similar approach was adopted only by Beegle et al. (2008) and Dercon et al. (2005). While the former study finds that the impact of death on consumption becomes insignificant five years after the death, the latter does not identify a statistically significant impact of death on consumption. The data also provide me with the opportunity to study the evolution of food and non-food consumption after a death occurs in a household. Grimm (2010) and Wagstaff (2007) address this question. Grimm (2010) finds statistically significant increase in non-food expenditure and statistically insignificant decrease in food expenditure. Wagstaff (2007) distinguishes between urban and rural households and finds that the results vary not only based on these criteria but also based on the equivalence scale used. Furthermore, the above-mentioned studies consider the impact of death to be linear meaning that the change in consumption caused by a death is of the same magnitude regardless whether multiple deaths occurred during the period or not. I am going to test whether with more deaths within one household we can observe a decreasing effect of an additional death on consumption. Ardington et al. (2014) study the similar question for the impacts of deaths caused by AIDS on household assets and find no statistically significant relationship.

Hypotheses
1. Hypothesis #1: Shocks in consumption induced by a death of a household member tend to persist over time.
2. Hypothesis #2: The magnitude of the shock depends on previous deaths in the household.
3. Hypothesis #3: Non-food consumption is affected more by a death of a household member than food consumption.

Methodology
I will begin my analysis by providing a detailed description of the considered variables and the methodology by which the MxFLS dataset was constructed. In this step I will also provide descriptive statistics on the relevant indicators. In the second step of my analysis I will follow the approaches of Wagstaff (2007) and Beegle et al. (2008) and construct fixed-effect models. Total consumption, food and non-food consumption will be used as dependent variables. I will introduce dummy variables describing different types of death and will test for their significance by the appropriate methods. I will distinguish in between child, prime-age and old household members. In the case of prime-age household members I will control separately for male and female deaths. To test whether the magnitude of the impact of death depends on previous deaths within the household I am going to test the statistical significance of dummy variables indicating whether a death already occurred in the household. These will be distinguished based on the time in between the current and previous death.

Expected Contribution
The contribution of my thesis will be threefold. In my thesis I will be the first to use the MxFLS dataset to study the influence of a death on household consumption. Second, I will study whether the magnitude of the shock depends on previous deaths in the household. Third, I will assess the persistence of the shocks in consumption caused by a death of a household member.

Outline
1. Introduction
2. Literature review
3. Dataset description
4. Methodology
5. Econometric analyses
6. Conclusion