Intra-household Income Transfers and its Effects on Children’s Nutrition and Health in Peru 

Get Complete Project Material File(s) Now! »

Information Asymmetry and Distance

In this section, I present a theoretical framework to represent the interactions between transfer behavior and information asymmetry, motivated by distance. In particular, I aim to concep-tualize the idea that distance generates information deficiencies that encourage donors and receivers to act strategically. Living far from each other (geographic distance) or having no parentage (social distance), both donors and receivers can easily hide positive income shocks and, therefore, avoid transfer cutbacks, from the receiver’s perspective, or transfer pressure, from the point of view of the donor.
The positive character of the shocks is one of the key elements behind the configuration of strategic behavior due to information asymmetry and distance. Negative shocks, resulting, for instance, from natural disasters (droughts, earthquakes, etc.), are more likely to induce Private Income Transfers, Information Asymmetry and Distance. Theory and Evidence individuals to communicate about them, despite the distance that may exist between transfer partners. The start point is the model advanced by Cox (1987), where transfers are characterized by two fundamental attributes. The first one is that they are motivated by impure altruism. In contrast with the pure altruistic model (Barro, 1974; Becker, 1974, 1981), where transfers materialize the way agents value the well-being of the others, this formulation has the advantage of allowing them to act also motivated by the exchange of services (Bernheim et al., 1985). This second mo-tivation, although more complex, is better suited to model transfers in a context of asymmetric information and strategic behavior. However, altruistic motivations continue to be the key-stone of transfer behavior and the reason why these transactions require a specific modeling, beyond a pure market economy setting.
In this context, services have a very particular nature. They stand for any action of assistance or work done in order to please someone, that generates income (money or in-kind) transfers, in return. Some examples are help with household chores, support in home production, lend a summer house to a neighbor, pay the rent for a student, look after a sick relative or visit an ail-ing friend. Although, at first sight, these exchanges may seem like a typical market transaction, they differ in several aspects.
In some instances, services are only provided to certain agents or under very specific circum-stances, like taking care of a nephew or give inn to a friend during the winter. It is also very likely that they do not have market substitutes, as they usually involve affections like caring, trust, etc. In addition, very frequently, what is being exchanged and its value is not always precisely known and « payment » conditions are very uncertain, as transfers may not necessarily occur immediately, but later, or be deferred, or be indirect, or even never occur.
The second characteristic of the Cox model of transfers is that they depend, mainly, on the actual income of agents, which implicitly entails that the actual income of one agent is perfectly observable by the other.
Although I concur with the relevance of income in the determination of transfers, hereby I relax the perfect information assumption and consider, instead, that at a given distance, agents only observe the pre-shock income of the each other. As in many other economic models dealing with information asymmetry, I assume that information frictions are only problematic in the short-run, while agents find the way to address their own information requirements, and dis-appear in the long-term. Complete information before the shock is compatible with a long-run equilibrium setting were information circulates well and agents know the income of the each other.
Finally, other important assumptions, present in the Cox model and here as well, are the fol-lowing: (i) there is only one period, (ii) the income of agents is exogenous, (iii) agents are credit constraint, (iv) transfers are one-sided9 and (v) there are only two agents, one transfer donor, labeled with subscript d, and one transfer receiver, labeled with subscript r.
Under this setting and considering information asymmetry and strategic behavior, two differ-ent benchmark transfer regimes are particularly relevant. The first consists on transfers going from an impurely altruistic donor to a non-altruistic receiver. The second, on the contrary, en-tails transfers going from a non-altruistic donor to an altruistic receiver. The following lines provide a detailed analysis of these two regimes, in order to derive some testable predictions about the relationship between transfer behavior and income, when the last is not perfectly observable, in each context.

Sample Characteristics and Empirical Strategy

Given the nature of Familias en Acción and the characteristics of the data, this analysis focuses on poor households living in a rural or a small urban municipality of Colombia. Table 1.3 displays some basic descriptive statistics regarding the socioeconomic characteristics of this sample before the program started (i.e. 2002). A little more than half of the households live in urban areas and 37% live in denser populated rural areas. They are, on average, composed by Almost 20% of these households are headed by a single parent, most of whom are women. Household heads are on average 45 years old and one third is illiterate. Only 3% are unem-ployed and most work as paid employees or self-employed (38% in each case). As already mentioned, households in the sample are very poor, 89% are below the poverty line and 53% fall into the range of extreme poverty.42 In addition, 26% live in inadequate housing, 17% have no access to basic services, 35% live in overcrowded dwellings, 6% have at least one child aged 7 to 11 not attending school and 19% live in high economic dependence. The average monthly household income43 is 496,047 COP (around 198 US Dollars) and consumption amounts to 227,780 COP (91 US Dollars). On average, households have savings for 29,995 COP (12 US Dollars) while loans amount up to 57,050 (23 US Dollars).
Private transfers are very important in this context.44 According to Table 1.4, almost half of the households report having received a private transfer in the previous year: 20% in the case of money transfers and 39% in the case of in-kind transfers. Money transfers come mostly from relatives and represent, on average, 17% of household income. By contrast, in-kind transfers come mostly from donors living nearby and their contribution reaches 17% of total consump-tion. As expected, very few households deliver money and in-kind transfers, 11% and 17% respec-tively, which represent, on average, 3% of total income and 4% of total consumption. This is natural, given that most households in the sample are poor and represent Familias en Acción eligible families. Most households privilege delivering transfers to nearby locations, specially in the case of in-kind ones. Despite its relevance, data shows that, transfers are not received and delivered in a very regular basis. In most of the cases, households are involved in transfer transactions only once or three times in the year (Figure 1.2).
Although it is not the most common trend in the data, households can be involved in several transfer transactions at the same time. Just 13% of the households simultaneously received and delivered transfers (811 cases), with only 6% of them receiving and delivering transfers from and to a close partner, and 31% receiving and delivering transfers from and to a friend. It is also rare to receive or give transfers to more than one type of partner. From the total number of households receiving transfers, only 11% simultaneously received them from close and far locations, and 12% from relatives and friends. Similarly, from those delivering transfers, only 3% delivered them to close and far locations, and 9% to relatives and friends, at the same time (Table 1.5).
Finally, Table 1.6 presents the evolution of private transfers between 2002 to 2003. The percent-age of households receiving and delivering transfers registers a general increase. In the case of transfers-in, its incidence raised in 21 percentage points, mostly driven by in-kind transfers. What is more interesting, however, is that these gains are much more important in the case of transfers involving partners living far and friends. When it comes to their values, however, the evolution path is less clear. Although aggregate transfers increased in both money and in-kind types, in some cases, these sums actually decreased (e.g. for all money transfers-in disaggregated cases and in-kind transfers-in from relatives).
Something similar is observed for transfers-out. Although the incidence of the aggregates re-mains stable, all the disaggregated cases registered an increase, with the most important gains observed for transfers delivered to far locations and friends.45 This time, however, the values associated to the different geographic and social distance sub-categories, show all decreasing trends.

Empirical Strategy

The empirical strategy used in this chapter relies on a difference-in-difference method (DID), consisting in comparing changes in private transfer outcomes between Familias en Acción eli-gible and non-eligible households before and after the program. The empirical specification is given by Equation (1.9): Yi,t = ↵t + ⌘i + P i,t + R ⇥ ↵t + Z ⇥ ↵t + « i,t (1.9).
where Yi,t denotes the transfer outcome of interest for household i in year t (for example, in the case of being participating in money transfers-in from a close partner, Yi,t represents a dummy variable equal to 1 if household i received a money transfer from a household from a nearby location in year t), Pi,t is a dummy variable equal to 1 if the household i is located in a treatment municipality and 0 if it is located in a control municipality in year t, R is a set of region dummies, Z is a set of zone type dummies (urban46, populated center47 and rural), ↵t represents time, ⌘i accounts for household fixed-effects and « i is an error term.
To the extent that treatment status is a random event, would yield an unbiased estimate of the average impact of Familias en Acción eligibility on private transfers. Although the program was not randomly assigned, its evaluation design was made in such a way that in the data, treatment and control households should be alike. Tables 1.7 and 1.8 present simple test of differences in means in order to check how different were these households before the program started.
Results suggest that treatment and control households do not differ significantly in terms of income and several measures of wealth. This indicates that selection into the program may not be so strongly linked to initial household socioeconomic differences. However, there are other dimensions about which treatment and control households do not seem to be so comparable. Eligible households are, for instance, less likely to live in urban areas; tend to have less children between 7 and 11 y/o not attending school; have less adults; have more children below 7 y/o; have younger and more literate household heads, less household heads unemployed and less household heads in self-employment; consume less and accumulate less savings (Table 1.7).

Social Well-being

The identification threats outlined above signal that, although the empirical evidence presented in this subsection illustrates well the theoretical predictions outlined in Section 1.3, according to which the responsiveness of transfers to income shocks is partly explained by information asymmetries associated to distance, one can not discard that other devices are at stake. The viability of alternative mechanisms highlights the importance of being cautious when it comes to draw strong and definitive conclusions from these findings. However, it is important to note that, by acknowledging the role that information asymmetry and distance fulfill in the configuration of private transfers, this analysis does not intend to question the legitimacy of other channels.
In addition, I am also aware of other caveats that limit the interpretation of the findings and restrict the assessment of its public policy implications. A shortfall in private transfers received and an increase of transfer pressure on Familias en Acción eligibles, might have very important consequences in terms of social well-being. Besides the particular implications on the expected program effects, little studied up to now, these new transfer transactions could have key re-distributive impacts. Nevertheless, these effects will depend on variables over which there is no information in the data. For instance, the socioeconomic characteristics of the other sides of the transactions, e.g. the donors cutting transfers-in to Familias en Acción eligibles and the households receiving transfers from them, and the new transfer relationships these unobserved households could start afterwards.
By assuming, for instance, that households cutting transfers to Familias en Acción eligibles are, after the subsidy, comparatively poorer, i.e. Ir + ✓ > Id59, and considering they may now start to deliver transfers to third households (subscript t) even more poor, i.e. Ir + ✓ > Id > It, the final social impact of the program can be indeed positive. On the contrary, if these households are still richer and do not engage in any new transfer transactions, this effect might be very negative. Similarly, if, thanks to the subsidy, Familias en Acción eligible household engage in transfer transactions with poorer households, i.e. Id + > Ir > It60, the final impact will be even greater.

The 2008 Global Economic Crisis

The dispersed location of migrants, observed in the Ecuadorian case, may be very convenient in order to explore whether or not the 2008 global economic crisis, observed in some of these Between 2008 and 2009 a global recession struck nearly all advanced economies14, with 29 out of the 34 OECD countries being in recession by the last quarter of 2008.15 Finland, Ireland, New Zealand, Portugal and Sweden experienced the first slowdown during the first quarter of 2008; Austria, Chile, France, Germany, Hungary, Italy, Japan, Luxembourg, the Netherlands, Spain, Turkey and the United Kingdom during the second; Belgium, Denmark, Estonia, Greece, Mexico, Slovenia and the United States during the the third quarter and Canada, the Czech Republic, Iceland, Israel and Switzerland during the last quarter of this year.
The 2008 global recession had profound impacts on employment.17 However, labor markets adjusted in very different ways across countries. Although most economies saw a relatively small decline in labor input (total hours worked), in the United States and Spain, for instance, it fell sharply and even faster than output. These two countries also suffered major labor force participation and employment declines, in contrast with Germany, Japan and other European countries which opted for cuts in working hours. The different adjustment patterns in terms of labor input, working hours and participation, damped (or amplified) the unemployment effects of the recession. While many European countries did not experience any unemployment effect, in Spain and the United States, where the decline in output was below-average, the rise in unemployment was much higher than average. On the contrary, in Germany, output declined much more and the unemployment rate actually fell.

Identification Strategy

To analyze how variations in household remittances imply adjustments in the labor supply of its members, one might estimate the following fixed-effects regression across the whole sample of individuals and each of the groups of analysis: children, adult men, adult women and old adults: Yit = ↵i + t + htRem + htS + ⇢itX + « it (2.1).
where Yiht represents the outcome of interest for individual i in household h at period t. That is, participation and hours/week spent in market and household work. ↵i is the individual fixed-effect. t accounts for the year effect. Remht measures the amount of remittances received, at the household level.28 Sht denotes a set of time varying household characteristics including composition (number of children under 5 years old, number of children between 5 and 19 years old, number of men between 20 and 60 y/o., number of women between 20 and 60 y/o and number of adults over 60) and household head characteristics (age, age squared, years of education and a dummy equal to 1 if the household head is female). Xit is the set of individual characteristics (age, age squared and years of education). Finally, « it represents the error term.
The main identification assumption for an unbiased estimate of ht the amount of remittances, received in the month preceding the survey, is exogenous to the change in last week labor supply. However, this might be too strong as assumption, taking into account the potential time varying unobserved components of labor supply decisions, potentially correlated with changes in remittances, and the reverse effect of these labor supply adjustments on remittances. Concerning the first problem, omitted variable bias, one might think on the variation in any household unobserved characteristic (e.g. money consciousness, risk taking nature or simple differences in household resources, tastes and labor market opportunities, among others) that enables members to work more regardless of any trend in remittances received.

Table of contents :

1 Private Income Transfers, Information Asymmetry and Distance. Theory and Evidence from Colombia 
1.1 Introduction
1.2 Literature Review
1.2.1 Theoretical Background
1.2.2 Empirical Evidence
1.3 Information Asymmetry and Distance
1.4 Familias en Acción
1.4.1 Characteristics of the Program
1.4.2 Data
1.5 Sample Characteristics and Empirical Strategy
1.5.1 Descriptive Statistics
1.5.2 Empirical Strategy
1.6 Results
1.6.1 Main Estimates
1.6.2 Identification Threats
1.6.3 SocialWell-being
1.7 Conclusions
1.8 Figures and Tables
2 Remittances and Labor Supply in the Hearth of Ecuadorian Migrants 
2.1 Introduction
2.2 Data and Descriptive Statistics
2.2.1 Data
2.2.2 Descriptive Statistics
2.3 The 2008 Global Economic Crisis
2.4 Empirical Strategy
2.4.1 Identification Strategy
2.4.2 Quantile Regression
2.5 Results
2.5.1 Remittances and Unemployment Abroad
2.5.2 Individual Labor Supply
2.6 Potential Threats to the Exclusion Restriction
2.6.1 Return Migration and Re-migration
2.6.2 Selection in Migration Patterns
2.6.3 Confounding Macroeconomic Variables
2.6.4 Sample Attrition
2.7 Conclusions
2.8 Figures and Tables
Appendix
3 Intra-household Income Transfers and its Effects on Children’s Nutrition and Health in Peru 
3.1 Introduction
3.2 Background
3.2.1 Peru’s Elderly Population
3.2.2 Pensión 65
3.3 Data and Descriptive statistics
3.3.1 Households with Children under 5 y/o
3.3.2 Children under 5 y/o
3.4 Empirical Strategy
3.5 Results
3.5.1 Household Monetary Spending
3.5.2 Children’s Nutrition and Health
3.5.3 Sensitivity Tests
3.5.4 Validity Analysis
3.5.5 Potential Confounding Factors
3.6 Conclusions
3.7 Tables and Figures
Appendix
Bibliography 

GET THE COMPLETE PROJECT