ROSCAs and Savings Behavior

Transcription

1 ROSCAs and Savings Behavior A Woman's Individual Motives for Grouped Participation in a Kenyan Slum Justin Chin

2 Table of Contents The Critical Review... 2 What is a Rotating Saving and Credit Association?... 3 Origin... 4 Introduction... 5 Theory and Hypotheses... 5 Models, Empirical Predictions and Data Analysis... 6 Models... 6 Empirical Predictions Data Analysis Gauss Markov Assumptions: A1 through A Biasedness and Unbiasedness Inferences Conclusion Works Cited

3 The Critical Review This is a critical review about a micro-financing institution (MFIs), which were established to help empower poverty stricken villages in rural areas. They empower them by instilling them with the skills and techniques needed in order to drastically improve their methods of saving money and consumption so that they can better the lives of each other and their families. The rural villages that these families live in generally lack the benefits of industrialized banks, so MFIs are there to compensate for these shortcomings. An MFI in particular called, Rotating Savings and Credit Associations (ROSCAs), has a very interesting and successful method in doing so, as it promotes it's members to work as a group and to partake in various teamwork related activities which when done with efficiency and timely, it can benefit everybody involved. Everyone being the individual member, their household and the entire village as a whole and in due time, hopefully, the entire country they occupy. One way to think about ROSCAs is that it can be a trust game concerning tangible currency and mental rewards. The critical review in question is based on a study of ROSCAs in a slum village called Kianda, which is located in Nairobi, Kenya. Siwan Anderson and Jean-Marie Baland are the authors that conducted the case study, the title of their scholarly article is, "The Economics of ROSCAs and Intra-household Resource Allocation," which in particular is about male and female couples, with or without children, and their savings behavior among each other. The focus is mainly on the woman or wife and their tendency to be predominant and proactive in being a member of a ROSCA. The critical review is going to consist of four segments pertaining to different topics that will all in some way be connected. First, an introduction as to how a ROSCA operates and a brief description about its origin in order to get a better understanding of where it came from. Second, an introduction and overall review of the theories and data within the literature used to conduct their observations. Third, the theories, empirical predictions and data that can and cannot be tied into the world of econometrics with topics such as the hypothesis and alternative hypothesis, the relation to the Gauss Markov Assumptions - A1 through A6 and how there could be violations, biasedness and unbiasedness of the data, and inferences. Fourth, the summary and conclusion of the critical review and article. 2

4 What is a Rotating Saving and Credit Association? ROSCAs are an informal method of micro-financing and since it requires a group of people, it has to be based on some kind of trust system, wherein there are many factors that can benefit and hurt the outcome of that group. Mainly due to the fact that there are instances where everybody knows each other in the group or they don't. There's also the instance that everybody or grouped individuals within in a group knows each other and their personality or financial behavior and that would be a deciding factor into whether or not those individuals decide to participate in the ROSCA. In these groups, people that would generally live in the same village gather on a set dates throughout the year. During these gathering each member would contribute to a set amount of money called a "pot" that is given as a lump sum to one of the contributors. The person receiving the "pot" would subsequently be obligated to contribute to future pots. The process is continued until everyone in the ROSCA received a "pot." The groups usually have various methods in determining the amount of gatherings, contributions, group size and how the winner is selected. For instance, they can draw lots where a winner is randomly selected or they can use an auction method where the highest bidder would receive the "pot." Refer to Figure 1. below for visualizations to how a ROSCA works. Figure Simone Baldassarri. "Rotating Saving Groups" (2011). 3

5 In Figure 1., John is the leader or the "winner" of the pot, since everyone, including himself contribute $10, he would receive the sum of the $40. Continuing the cycle, Mark becomes the leader and wins the pot, the cycle continues up until Carol (the last person) receives the pot. As mentioned, if they decide to continue the group, how they determine who starts the cycle and how much the contribution should be is up to them. Origin As mentioned earlier, a ROSCA is one institution among many in a network of MFIs. Originally, the ROSCAs method of micro-financing was spun off from something called the Grameen Model, which was established by the Grameen Bank, a bank establishment in India that started from practically nothing. Its founder, Professor Muhammad Yunus created the Grameen Bank in 1976 with just $27 to his name. Today, the Grameen Bank caters to 7 million impoverished families by means of loans, savings, and insurance along with additional services. 2 ROSCAs employ most of their methods from the Grameen Model, which is also based on trust. In the Grameen Model, the members are in a group, along with the inclusion of the bank itself, as opposed to the ROSCA that doesn't use the bank, but a treasurer. First, potential members living in rural villages for the group are screened by the bank, typically in groups of five. Second, after the group has been selected, two potential predetermined members are selected to borrow money from the bank, they are then observed for a month. Third, if the potential members follow the guidelines set by the bank for the loan and pay their debt with interest within 50 weeks, the bank then allows the remaining three members to borrow funds, of course with adherence to their guidelines. 3 However, there is also collateral involved within the Grameen Model. It comes from the "collective responsibility" of the group, which happens from the two potential members of the group that initiate the model. If one of them fails on their responsibilities the bank could disband the entire group. 4 When compared to ROSCAs, is that the initial "pot" comprises of each 2 "Grameen Heritage" Grameen Foundation. Trademarks & Copyright 1998 Grameen Communications., 23 Aug Web. 18 May < 3 "Credit Lending Models."Grameen Bank. Trademarks & Copyright 1998 Grameen Communications., 23 Aug Web. 18 May < 4 "Credit Lending Models."Grameen Bank. Trademarks & Copyright 1998 Grameen Communications., 23 Aug Web. 18 May < 4

6 individual member without the fear of the collateral, due to the fact that if a member defaults on their contribution, the ROSCAs treasurer will make up for their default and consider it debt among the remaining members until they can find a replacement. Introduction The Economics of ROSCAs and Intra-household Resource Allocation by Siwan Anderson and Jean-Marie Baland, as mentioned earlier, is a study that was conducted in the slum village of Kianda, from a rural area in Nairobi, Kenya. The focus of their study was to observe the spending behaviors of the women in ROSCAs and whether or not they were in a civil union with a male counterpart or not. They wanted to investigate into the reasoning as to why the women were predominant in participating in a ROSCA. Their level of consumption and the spending behaviors became their main focus for both men and women. The authors cited a case study that was conducted by Hoddinott and Hadadd (1995, pg 5), where men in African households would use more of their income on goods such as alcohol and cigarettes. Contrary to men, the women would spend a majority of their income to satisfy the needs of their children and household necessities. 5 They also cited a different case study where "gender-based differences are most explicit in Africa, where it is commonly believed that men have a right to personal spending money, which they are perceived to need or deserve, while women's income is used for collective purposes." 6 The reasoning behind these behaviors is that the men s level of consumption is instant and is used recklessly or for personal gain, as opposed to women who used their savings more beneficial to the family, which is the main study conducted by the authors. Theory and Hypotheses In the article they state that their theory they give predicts "Income earning women are more likely to save in ROSCAs if they are married." In other words, the wife in a marriage, earning her own income is less likely to oppose her husband who also earns his own income 5 Hoddinott, J., and L. Haddad. "Does Female Income Share Influence Household Expenditure? Evidence from Cote D'Ivoire." Abstract. Oxford Bulletin of Economics and Statistics 57.1 (1995): Online Journal. 6 Bruce, J. "Homes Divided." Abstract. World Development 17.7 (1989): Online Journal. 5

7 since it helps the family. So based on their theory, one might assume that the hypothesis and alternative hypothesis that they will be basing their test are: H 0 : Income earning women are more likely to save in ROSCAs if they are married 7 H a : Income earning women are less likely to save in ROSCAs if they are married Models, Empirical Predictions and Data Analysis In their article, Anderson and Baland used a series of theoretical models and compiled data from a sample of 520 households, 385 of which used ROSCAs and don't have a system in place for random or bidding orders. 8 From their theories, they were able to make models using their collective data that also tie into econometrics, which will be done throughout the critical review. Models The theoretical models below and hereafter are what the authors used to calculate and verify their hypotheses. These models were designed from the perspective of immediate and prolonged consumption from the men and the women in deciding whether or not to participate in the ROSCA and how they would finalize their decisions based on their preferences and necessities. Below is a series of models they used with their descriptions in order to calculate the men and women s consumption and household values: - Current consumption: c t - Woman's utility at time t:u = u(c t) o Assumed greater preference for an indivisible good than husbands - Woman's utility when purchasing indivisible good in one period, D: U = u(c t) + D - Husband's utility at time t :U h = u(c t) o Assumed, <1, where δ, represents the husbands greater preference for an indivisible good 7 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 18. Online Journal. 8 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 20. Online Journal. 6

8 - Husband's utility when purchasing indivisible good in one period, δd: U h = u(c t) + δd - Optimal Household Saving Rate: s R = s 0 - Max. the weighted sum of individual utilities: - Utility of the household at time t: U = ((1 - γ) δ + γ)d + u(c t) - Period of Purchased of Indivisible Good: U = u(c t) - Discount factor ( ), is smaller than one - Wife's weight in household joint decision making, γ, where 0 < γ < 1 - Income received by husband and wife, respectively, each period: Y h and Y w - Household Budget Constraint: Y w +Y h = c t + s - T H = purchase date - Accumulated Savings covering the cost of the purchased indivisible good: 1, - First Order Condition, if the household purchases the indivisible good: s = = - Savings, when positive, increasing through time: s <s, t< TH 1 - Second Order Condition, savings pattern for all inclusive equations(note: This condition must be satisfied, if not, then the household is better off, s = 0 for all t) 9 : Next, the authors come up with models detailing the married couple s decision to participate in a ROSCA, it's given in three stages, where s R = savings through a ROSCA: o Stage 1: Woman chooses her contribution to the ROSCA in each period t, 0 <s R < 1 and theduration of the ROSCA =T R. o (If she decides not to participate in a ROSCA then, s R = T R = 0) 9 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 6-8. Online Journal. 7

9 o Stage 2: the husband and wife jointly choose household savings in each period t, o s 0. o Stage 3: when t = T R, the women receives the pot and the household decides its use. Stage 1, is the Optimal Household Savings Rate equal to the First Order Conditions and stage 2, is where the Second Order Conditions need to be satisfied. Stage 3, represents the time of the cycle to end so the ROSCA earnings can be given to another member in the group 10. Holding the second order condition constant, they were able to come up with an Optimal Sequence in which, "the woman necessarily has an incentive to save, where she has a higher preference for the durable good," such as food, clothing, and school supplies for the children. Below are the models they used and descriptions: - Woman's incentive to save: {s } - Optimal Sequence: The authors then brings up a moral dilemma, in which the household decides not to "increase present and future consumption" by saving up without a ROSCA and still consuming. However, when the woman decides to jointly enter a ROSCA with the household, the dilemma then arises as to how much their optimal contribution should be when they've already saved up to consume. The model they then use is as follows 11 : - Amount Saved: s R 0 - Add Extra Savings (to the ROSCA): σ 0 - Household Dilemma: o such that 1 = σ 10 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 8. Online Journal. 11 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 9. Online Journal. 8

10 Thus, the authors state, that the woman ultimately solves the Household Dilemma by choosing the Amount Saved (s R ), thus giving the following model: - - Where σ, solves the household dilemma Following the solution to the household dilemma, the authors come up with two "propositions." The first is where the woman alwaysparticipates in a ROSCA, if the Optimal Sequence (mentioned above) is satisfied 12 : - Proposition 1: s 0 <sr < s! " o Where the ROSCA contributions are constant over time, s 0 o Woman's Optimal Saving Pattern: s! " Therefore, a woman will always participate in a ROSCA if the amount saved (s R ) is greater than the expected contribution and less than her optimal saving pattern. They also provided a diagram representing the propositionwith the wife and husband s optimal savings schedule, s W for the wife, and s H for the husband. Optimal Savings is represented by the s R line, which represents the household total savings schedule (s R + σ ), as noted earlier from the solution of the household dilemma. The contribution of the extra household savings (σ ) is upward sloping. 12 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 10. Online Journal. 9

11 Figure 2.: ROSCA Contributions and Optimal Saving Schedule 13 The second proposal where the household decides to save, thus makes her ROSCA contribution constant on the lower values, represented by gamma (γ), of her weight in the household decision making and decreasing for higher values of γ. Recall that the joint household savings is {s }, from that the authors introduce the idea of "stock of savings," S k. Where S k is greater than zero, but less than 1(0 <S k < 1), and household savings are greater than 1, represented as {s }> 1. Thus, creating the following model for proposition 214 : - Proposition 2: o m is part of the "optimal dissaving pattern", therefore o S k + # = 1 and $ = S k The authors state that due to proposition 2, "given sufficient accumulated ROSCA savings, the household will decide to save to buy the durable goods, instead of consuming it through an 13 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 11. Online Journal. 14 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 12. Online Journal. 10

12 optimal dis-saving pattern $. It may even decide to contribute additional extra savings to the amounts thus accumulated." They also give an alternative to equation to proposition 2, in the event that accumulated savings exceeds one, the household could elect to buy the durable goods and use the excess savings for consumption. Therefore, S k > 1 and % 0, where 15 : and % = S k - 1 and $ = S k Empirical Predictions The empirical predictions presented by Anderson and Balandpositions around the idea that women strategically participate in ROSCAs in order to protect their income from immediate consumption by their husbands. Before the authors get into what their predictions are, they give "relevant considerations" as to how participating in a ROSCA would be worthwhile if the benefits outweigh the certain cost, such as, her "weight in household decision making being positively related to her individual income, or more precisely, to her share in household income, α, where α = & ' & ' & (," with households under consideration, "women only have access to their own income," and that by participating in a ROSCA it "implies some fixed cost in terms of meeting attendance and other social obligations towards the other members of the group." From that, they were able to create two "conjectures," in other words inferences, which will be discussed in the later in the segment. The two conjectures are 16 : 1. Conjecture 1.: The relationship between the probability of joining a ROSCA and female relative income share, α, is an inverted-u shape 2. Conjecture 2.: ROSCA contributions are non-increasing in α [income share], and even strictly decreasing for α [income share] high enough. 15 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 13. Online Journal. 16 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 15. Online Journal. 11

13 The conjectures explain the function for forecasting the probability of women's participation in a ROSCA relative to their income share. The diagram shows this is in Figure 2. below: Figure 3: ROSCA Contributions as a function of woman's share in household income 17 Also, the inverted-u shape that they're referring to is the shape of the chart, in other words, women with very low income share won't join a ROSCA due to high cost, this deals with conjecture 1, and those with high levels of income share will join and will decrease at its peak, dealing with conjecture 2. This is most likely due to the woman's weight in household decision making, represented by the gamma symbol, γ. The more income share they have, the more weight they hold. The highest level of income sharing occurs in the short-run at α 0, relative to the amount saved s R, but as mentioned, as it increases closer to it decreases at its highest peak and goes toward α Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 16. Online Journal. 12

14 Data Analysis The data used in the article is from a case study that was collect in 1996 to 1997, which as mentioned earlier, is from a slum of Kibera, in the village of Kianda. The data was compiled from 520 households, which contained 620 groups and 385 of which were ROSCAs. They consisted of the information in the following table 18 : Table 1. Basic Information on ROSCAs 19 Variable Mean Mean Number of members Months existed 27.8 Contribute every day (% of roscas) 0.1 Contribute every week 0.35 Contribute every 2 weeks 0.06 Contribute every month 0.49 Length of cycle (median, in months) 6.07 Number of cycles (median, in lifetime of rosca) 3.21 Group comprises only women (% of roscas) 0.65 Group comprises only men 0.06 Group comprises both men and women 0.3 All members are same ethnicity 0.37 Order is unchanged each cycle 0.69 Started group with friends/relatives/neighbors 0.85 Group has secondary role (investment/insurance) 0.25 Based on the mean sample, one can see that 65% of the ROSCAs were comprised of women, while 6% were men and 30% were both women and men. Showing a strong correlation that women were the dominant sex in initiating ROSCAs in Kianda. In table 2 from the article, the authors use more detailed information withtheir variables, including the entire sample, all the ROSCA members, only the women in ROSCAs and the women not in ROSCAs, as shown below: 18 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 17. Online Journal. 19 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 18. Online Journal. 13

15 Table 2: Characteristics of the population, rosca participants, female rosca participants and nonparticipants 20 Variable All Sample Std. Mean Error ROSCA Members Women in ROSCAS Std. Std. Mean Error Mean Error Women Not In ROSCAS Std. Mean Error Participation in ROSCA Total Monthly ROSCA Contribution Female Age Married Earns Labor Income Has at least primary school Monthly individual income if work Household monthly income Monthly food expenditures Monthly luxury expenditures Monthly children expenditures Household size Number of Children Years in Kibera Native Language: Kikuyu Native Language: Luhya Native Language: Luo Native Language: Kamba Native Language: Kisii Number of Observations In this table, the sample of the 520 households represented 2300 individuals, but in order to be consistent with the study those that were younger than 16 years old were excluded, thus decreasing the overall sample size and number of individuals to Excluding people younger than 16 was done to account for biasedness in the study, as the youngest age for marriage in Kenya is 16, thus excluding couples which are most likely not in a binding marriage 20 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 20. Online Journal. 14

16 younger than that age. In creating the biasedness, what they did is effectively separate the people that are able to get married from those that can't by law and in doing so they increased the efficiency of the study. Furthermore, based on the table they were able to confirm that school fees were the largest expenditure, which accounted for 36%, while rent 22%, clothing 18% and medical costs 12%. Also noted by the authors, was that in ROSCAs, people that are "females, working, and married" have a higher proportion rate in a ROSCA when compared to the "average individual." Thus, married women earning an individual income have the higher savings rate, holding the average number of children and income level constant within the two groups, ceteris paribus. 21 Table 3: Probit Estimation of ROSCA Participation 22 Variable δf/δx S. E. δf/δx S. E. Female Couple Female x Couple Food Expenses 8.95E E E E-05 (Food Expenses) E E E E-10 Female Share of Couple Income (Female Share of Couple Income) Lived in Kibera for at Most 2 Years Number of Children Primary School Degree Age (Age) E E E E-04 Kikuyu -3.40E E Luhya Luo Kamba Female Income Share > 0 & 20% Female Income Share > 20 & 40% Female Income Share > 40 & 80% Female Income Share > 80 & 100% Number of Observations Psuedo R Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 20. Online Journal. 22 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 24. Online Journal. 15

17 As you can see with the Psuedo R 2 in the Probit Estimation, the score is 0.29 and since it has a low value, between 0 and 1, this would indicate that there's a low fit in the regression model. This might be from the biasedness that was conducted from the exclusion of anyone younger than 16, but one might not be too sure as that's not indicated. In addition, it's not a true predictor of the outcome of either estimation, since the interpretation of the ordinary least squares (OLS) for R 2 is different for Pseudo R 2 's that would indicate a different variation of the regression model used. In other words, the model they used wasn't, R 2 = 1 -. */0 ) *+), -. */0) * +) -, but instead used one with different y variables, unfortunately the Pseudo R 2 predictor was not disclosed in the article. Furthermore, based on the first estimation, they created a plot chart depicting the "predicted values of ROSCA participation rates" in comparison to the "female income shares and for the average levels of all other variables." The "average levels of all other variables" could be accounted for as adding to the unbiasedness in the study as well. The chart below shows that at the highest level of "ROSCA participation" is when the "female income share" is at its mean, then dips (likely due to the squared terms adding a decreasing marginal effect) and makes the inverted-u shape. This chart also closely resembles a Standard Normal Distribution and also confirms their theoretical analysis of conjectures 1 and 2 23 : 23 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 25. Online Journal. 16

18 The author s estimation of the ROSCA Contributions make use of the residual sum of squares as shown in table 4, below. The residual sum of squares is the total sum of squares divided by the explained sum of squares. Table 4: Estimation of ROSCA Contributions 24 Variable Coeff. S. E. Coeff. S. E. Female Couple Female x Couple Food Expenses (Food Expenses) E E E E-06 Female Share of Couple Income (Female Share of Couple Income) Number of Children Primary School Degree Age (Age) Inverse Mill's Ratio Female Income Share > 0 & 20% Female Income Share > 20 & 40% Female Income Share > 40 & 80% Female Income Share > 80 & 100% Constant Number of Observations As you can see, the 2 2 shows that the correlation between the observed values and the predicted values are 0.05 and 0.04, so the estimated ROSCA Contributions isn t perfectly predictable. Due to the fact that when R 2 = 1, it means that the model is perfectly predictable, as opposed to R 2 = 0, wherein it s not. In addition, they were able to confirm their thesis "as female share in couple income is a significant and positive determinant of ROSCA distributions" and the second estimate of the study shows the "inverted-u shape" (Standard Normal Distribution), which confirms their 24 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 24. Online Journal. 17

19 theoretical analysis, conjectures 1 and 2. The second estimation is displayed in the diagram below 25 : Gauss Markov Assumptions: A1 through A6 As for the Gauss Markov Assumptions, if one were to find violations in the data, they would use the Gauss Markov Assumptions as their guideline, more importantly a violation in unbiasedness (A4), which would be the most prevalent in its detection. Although, the article doesn t explicitly give them, the assumptions below are implied or perceived as being so from the given information in the article: - A1. Linearity in Parameters: The relationships between the variables are linear due to the fact that each variable is assigned a constant coefficient in the regression model. - A2. Random Sampling:Random Sampling was performed as stated by the authors in the article, households, selected through a random process, were interviewed over the course of 4months during the spring of All household members were first surveyed for information ontheir education, work activity, and income. Pg17 25 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 27. Online Journal. 18

20 - A3. No Perfect Collinearity/Multicollinearity: The first estimation in Table 3.for the correlation coefficients and the standard error contain no perfect collinearity. However, the second estimation contained dummy variables as apparent by the absence of Female Share of Couple Income and (Female Share of Couple Income) 2 from the regression. - A4. Zero Conditional Mean (Unbiasedness), E[u x] = 0: The sample comprised of individuals participating and not participating in ROSCAs, married or not and all individuals in the study were over the age of 16 and can be legally married by law. All individuals in the sample passed all needed requirements, allowing them to be or not be married, with or without children and participating or not participating in a ROSCA. - A5. Homoskedasticity: The sample is homoskedastic as the random variables included unbiased estimators in the model. There is homogeneity of the variance in the sample as there is a consistency in the standard deviations. Although the authors make no mention of the Breusch-Pagan test to verify this, the standard errors are constant. Thus, showing the absence of hetersokedasticity. - A6. Normality of the Error Term: There is no mention of the normality of the error/residual term in the article, but one might assume that since there are minimal variations within the data, aside some very few outliers, that they aren t significant enough to cause a problem, seeing as how there is no loss of unbiasedness. Biasedness and Unbiasedness Stated by the authors, as a result, we have decided to consider all groups with a ROSCA as one of their activities in our sample. The alternative would have been to consider groups, which are only ROSCAs, but this could have led to a serious bias. In particular, in the survey, all possible alternative functions of the groups were carefully mentioned, even when the latter was clearly of secondary concern for the respondent. Although, there isn t any statistical data showing unbiasedness, the sampled data took information from participants in micro-financing institutions there were part of a ROSCA and not part of one, thus eliminating the concern for a biased sample. However, as mentioned earlier, the only type of biasedness in the study was the exclusion of the people younger than 16 years of age. Granted, they wanted to include people that were in civil union or married with a family, that were or were not in a ROSCA, but that's not to say that 19

21 people younger than 16 could be couples, have children and be in a ROSCA or vice versa. Only for the fact that they were following the age restrictions set by the law helps to prove the unbiasedness, otherwise it would be biased. Inferences As mentioned earlier, the conjectures 1 and 2 that the authors gave earlier were the inferences that they based on the evidence from the empirical predictions. To recall the conjectures or inferences are as follows 26 : 1. Conjecture 1.: The relationship between the probability of joining a ROSCA and female relative income share, α, is an inverted-u shape 2. Conjecture 2.: ROSCA contributions are non-increasing in α [income share], and even strictly decreasing for α [income share] high enough. According to the article, the inferences they made were based on the woman's weighted household decision making(γ). This was based on four levels of possibilities, from very small, small, medium and large 27 : - The very small γ, doesn't take reference from anything due to the fact no matter the amount accumulated, the household doesn't want to buy the durable good and they don't want to participate in a ROSCA. - They take the small γ inference, from proposition 2, where accumulated savings exceed one, as in the equation below: This is where the wife needs to already have a large enough savings in order to convince her husband to participate. - The medium γ, draws inference directly from proposition 2 when it corresponds to a stock of savings and the household savings path, given the condition that a high enough stock may influence to save in future periods, recall the equation for S k and : 26 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 15. Online Journal. 27 Anderson, Siwan, and Jean-Marie Baland."THE ECONOMICS OF ROSCAS AND INTRA-HOUSEHOLD RESOURCE ALLOCATION."Center for Economic Research (2000): 14. Online Journal. 20

22 - Large γ, is based on proposition 2 and gets inference from the likelihood of the household willing to buy durable goods, but the ROSCA is utilized by the wife to increase the savings rate for the family, being that condition below is met Conclusion The contributions in economic development that Rotating Savings and Credit Associations apply to families in rural villages can be both beneficial to everyone in a family. The observation that Siwan Anderson and Jean Marie-Baland they did in Kianda concerning the savings behavior of a woman's ability to join a ROSCA and the factors that come into play as to why and how they do so was very well detailed and calculated. Most prominently in the woman's ability to benefit at their husbands expense by convincing them to contribute. The fact they were able to demonstrate a woman's weighted decision in a household relative to her husband's based on their shared income, along with her bargaining influence, and both of their consumption preference levels in the village showed that they correct in their theories and hypothesis were correct, from their models, and data with their demonstration of inverted-u shapes charts (standard normal distribution curves). In doing so, they were able to exhibit not only the importance of ROSCAs as a saving mechanism for rural villages, but also their importance to the woman that stand to benefit most from using them, especially those in marriages. They provide a higher accumulation of money that can be used for consumption at a much faster rate when working in a group, as opposed not working in a group. More rural villages and countries around the world should make use of ROSCAs as it has the potential to drive them out of poverty, increasing the quality of life. On a different note, it would be interesting to see how a ROSCA might work in poor suburban neighborhoods. The results could be just as rewarding. 21

23 Works Cited 1. Anderson, Siwan and Baland, Jean-Marie, The Economics of Roscas and Intra-Household Resource Allocation (July 2000). CentER Working Paper No Available at SSRN: or 2. Simone Baldassarri. "Rotating Saving Groups" (2011) Grameen Heritage" Grameen Foundation. Trademarks & Copyright 1998 Grameen Communications., 23 Aug Web. 18 May < 4. Credit Lending Models."Grameen Bank. Trademarks & Copyright 1998 Grameen Communications., 23 Aug Web. 18 May < 5. Hoddinott, J., and L. Haddad. "Does Female Income Share Influence Household Expenditure? Evidence from Cote D'Ivoire." Abstract. Oxford Bulletin of Economics and Statistics 57.1 (1995): Print. 6. Bruce, J. "Homes Divided." Abstract. World Development 17.7 (1989): Print. 22