| new week, new abstract |
[Jan. 28th, 2007|12:57 am] |
and again i change my focus. i hope for the last time. it's getting down to the wire, i can taste the end of this ridiculously long time spent working on one paper.
Microfinance institutions (MFIs) have become a popular development tool. Research has shown that the banks help participants, but that they have failed to reach the poorest of the poor. This paper expands on this issue by analyzing whether MFIs affect income inequality. Using a dataset from Freedom from Hunger this paper finds there is relatively no change in income inequality in the program communities with the MFI verse similar control communities without the MFI. The regression analysis provides a possible explanation showing participation was correlated with an 11% increase in income, but the rich and poor participated equally in the program. These results suggest that MFIs in practice are not living up to their hype of massive poverty alleviation, but they are still reaching more of those in need than their critics have argued. |
|
|
| my thesis: in 100 (and ten) words: |
[Jan. 17th, 2007|09:23 pm] |
Microfinance institutions (MFIs) have become a popular development tool. Research has shown that banks help participants, but that they have failed to reach the poorest of the poor. My paper focuses on whether the increase in income in the participant group leads to an increase in income inequality in the community. Using a dataset from Ghana I found that although there is about a 10% increase in income for the participant group, there is no change in inequality in the communities with MFIs compared to similar control communities without MFIs. This is consistent with the view that there many be spillover effects which help those outside of the participant group.
( WHAT DOESNT'T MAKE SENSE )
( WHAT TO DO ) |
|
|
| answering some criticism... |
[Dec. 12th, 2006|10:55 am] |
Q:What's the average loan size and loan period? What's the comparison between loan size and income?
As of March 1997, in all the credit w/ education programs operating in ghana (55 credit associations, about 1500 participants, avg loan size is cedis equivalent of $78 dollars for loan period of 4 months. Average for my dataset from Lower Pra Bank: $60.39 (avg savings: $7.38)
Issue w/ exchange rates apparently huge devaluation of cedis between 93 and 96, from 695 to 1690 cedis to the dollar.
Q: Why is the Gini Coefficient so much lower for the group of participants than nonparticipants or control? Is it because of a variation in demographic factors?
According to the FFH Impact Report the groups are the same socioeconomically (asset value) and by literacy and schooling. However, they do differ in the participate groups have slightly older women with more children. Also had lost more children, no significant difference in alive children. Also more likely to be engaged in non-farm work. Percentage literate is higher in participate group, but not significantly different.
Why don't we just ask them? FFH qualitative follow up survey on 121/152 of the future nonparticipants. Why didn't you join?
25% not interested in working capital loan, engaged in farming, had no time, no good loan investment ideas
22% moved from the area, unable to meet weekly meeting requirement
16% fear of repayment
7% felt repayment would be a problem
14% had not heard about or understood the program
So even FFH says the poorest are probably not participating. Then why is there no significant difference in income quartiles or asset values of participants vs. non participants vs. controls in the baseline survey.
Why....hmmmmm? |
|
|
| before the last post... |
[Dec. 6th, 2006|10:45 pm] |
so this is important to catch you up on where i am before that last post...
the regression i ran for the relative income just didn't seem very valid. i mean how can i interpret the coefficient considering i squared the relative income, what's causing what, and does it even represent inequality? i'm not sure, but michael suggested the model doesnt really work well for what im trying to get at which is basically what happens to the income distribution when the program is started.
So what did I do instead of the regression?
1) Kernal Density Plot, these are basically like fancy histograms where you can see how many people made each level of income and it shows the distribution of income well in the sample, so i can compare between 93 and 96 by group. from this it look like in 93 most people were making the same amt for the participant group, but in 96 some of those people made more flattening it out
2) Graph of Income Quartile Averages- So I looked income quartiles and their averages and the change of these over time. Basically for all 3 groups the income quartiles change is highest for the top two or three quartiles.
3) Gini Coefficient- The most popular way of measuring inequality and my most confusing result. The Gini Coefficient showed that although the change in Gini was highest for the participant group the coefficient itself was much lower at about .3 while the non participant and control groups had higher .5 or .6 Ginis. So why did this happen? There must be some set of characteristics that make the participant group similar compared to the non-participant and control groups which are more diverse. The question is what are these characteristics? |
|
|
| why do people do things? |
[Dec. 6th, 2006|10:37 pm] |
So my presentation to the honors class was fine, but CNISS fellows really kind of went crazy with the criticism. So they suggest I:
1) do a bivariate, univariate analysis of the demographic and other characteristics that I have for the individuals across my sample groups...in english-> look at the average or distribution of age in the participant group and compare it to the control and non participant group to see if there are differences
2) from this i can generate a theory for why people choose to participate in these programs...or who ends up participating. is the self-selection related to income, age, kids?
3) come up w/ a theoretical basis for why i used the controls i did. or throw out the controls and only use ones with a theoretical basis and add new ones that they might not have used before. the rerun the tests and look at the results
4) do a density graph of the program group as a whole and then the future participate group (93 dataset only), see if the future participant group comes from the middle of this distribution
4) clean up the paper/presentation by focusing on what's new and important while summarizing what i replicated and redid using the regression, make sure to include the rational behind everything i decided to do esp the demographic controls
5) don't freak out. |
|
|
| OK! |
[Nov. 27th, 2006|02:01 am] |
So a lot has changed since that Michael Sherraden meeting. I have been neglecting this poor journal...
There was this phase when I wanted to test the effects of just the non-credit aspects of microfinance programs since a lot of them offer business/nutrition education as well as loans. Unfortunately I couldn't access the data on the subject to get a control group. I thought about doing some sort of approximate but it was difficult and not convincing even to me. I did find some relationship between people answering that they would consider profits first when starting a business and having higher profits. I think though this has more to do with an understanding of the term profit then anything else. Anyway it was still interesting.
So that phase seemed unfeasible and not going anywhere. I started thinking about income inequality and relative income because of Becca's thesis on baseball player preformance and income inequality. So basically what I have done is retest the effect of participation on income controlling for a variety of demographic factors like age, married/not married, ect. This was signficant for both income and log income The result is significant and the coefficient is .10-> in english participating in the microfinance program made you 10% increase in income. YAY good news for microfinance right? Well, it is good news. I wanted to also get at this issue of income inequalit and dispersion in the community. Basically my hypothesis is that due to entrepreneurial talent, income position, ect. microfinance is able to help some people a lot thus representing the programs as a whole as very sucessful even though they may not be able to help a lot of the participants.
I noticed the change in standard deviation of income was greatest for the participation group. I also tested to see if there was correlation between relative income (which i measured as an individual variance by group) and participation. Relative income= (Income-mean)^2, or in english the distance an individual is away from the mean, squared to make positive/negative not matter. I found that there was a significant relationship for relative income but not for relative log income. The literature though uses log income instead of income to account for some of the natural tendency of income data to be skewed and include outliers. The relationship between relative income and participation was large though, meaning participation caused a $1143 increase in the relative income of participants in 1996. In 1993 there was a significant positive $9 increase as a result of being a "future participant" which means that even before the program there was a slight increase that might be attributed to entrepreneurial talent or whatever. But it was very small in considering what the effect is after the program.
So what does this mean? Well basically my hypothesis that income is increased by the program but only for a certain group of people why the rest dont do as well is true. But there's no way to know why these people do well and other don't- is in innate ability, amt of income, income quartile to start off with, family, ect? Without the data tracking the same people from 1993 to 1996 there's no way to answer these questions. I think it would be really interesting to find out though. So there's where I am. |
|
|
| (no subject) |
[Oct. 18th, 2006|09:04 pm] |
So I had tea with michael sherraden and he suggested that I use some of the variable related to the children of the women to get at development (i.e. school, nutrition ect). Hmm, this is changing my thesis focus, but maybe it does make sense...
On the stats front, I decided to change my data in STATA format from SPSS because that's the Econ software. Now I'm trying to figure out how to use STATA and its a little more manageable than SPSS but hard to start from scratch. Prof Sened set me up with this grad student to help me with stats so that's nice but I still feel like I don't know what I'm doing and will constantly need his help. Elissa and I talked through some things yest and she knows a lot and seems like she will be good help. She thinks that creating dummy variables for control, participating, and 1996 or 1993 survey will work to split up my data correctly as long as I can figure out how to put in the blank data for the control for the participating variable. I need to figure out how to do it STATA though.
Enough of that if I havent already told you the nobel peace prize was awarded to Yunus the founder of the Grameen Bank, the world's first microfinance bank. Although the literature on the effectiveness of the Grameen Bank economically is mixed and rumors of corruption and beauocratic growth are everywhere, it is great for microfinance as an idea! WAhoooo.
THIS WEEK I NEED TO:
-get started writing the actual thesis: intro, lit review, description of data
-code the variables, separate the categories, and run the replication test
-prepare a presentation on the thesis for class thurs
-turn in a rough draft on wednesday night
-maybe start thinking about emailing jonathan morduch
eeek. |
|
|
| another update! |
[Oct. 10th, 2006|02:42 pm] |
So I have a meeting w/ Michael Sherraden for Friday at 4pm!
My "zero" draft paper is due two weeks from Wednesday.
I met with Patrick and we think in the FFH paper they compared averages between the groups- control, participant, non-participant for each of the variables. So I'm going to try and replicate this using a t-test comparing averages finding standard deviations.
confusing points...
why are the standard deviations so large/greater than then the averages and yet the results are still statistically significant?
they tested some demographic information across groups to find significant differences but then did not control for any of these factors if they were found signficant or so it seems because they didn't run any regressions?
how in the world do i use SPSS program software to do anything?
what i need to do...
-figure out what they did, replicate a table from their paper
-meet with michael sherraden to work out the lit review and sort of background info, take suggestions
-start writing the parts of the paper the i can do (i.e. description of data, theoretical framework, rational for topic's importance...)
-find someone to help with stats? |
|
|
| major update! |
[Oct. 7th, 2006|04:35 am] |
So this journal really needed an update. I think I left with the first draft of my abstract. Well things have been going pretty well but I need to really keep going at this pace. Here's where I am:
*I am definitely doing microfinance! I am so excited about it that I must apologize if I start boring you when you ask me a passing question about how I am doing.
*I'm trying to still test whether it leads to business development
*I HAVE DATA!!! Best thing that has happened recently! But yes I found a dataset from Freedom From Hunger, which I think I mentioned earlier with the excerpt from Jonathan Murdoch's interview. Anyway, I have the dataset which is great
Right now I'm thinking of creative ways to get at showing business development that aren't as obvious as just the avg monthly profit data. Why? Well, first the Fanta language (the data is from Ghana) has a term for profit but often its interpreted more as "savings". So basically people are subtracting not only their business costs from their revenue but also their household costs and expenses before coming up with their "profit". This could actually have understated the effect of the program which is exciting (ooo my bleeding heart bias is coming into this research now isnt it? wait...i thought i was a capitalist bitch!).
So what are my creative ways? Well right now I think the most interesting things that were in the survey are "expenditure on nonfarm tools or business supplies" and things about what assets people own like whether people own a radio. Often the first thing people buy when they move out of subsistence farming is a radio so it could be a measure of business success. I think the nonfarm tools/business supplies could be a good approximate as well. The problem is those indicators were not included in the actual dataset and were just used in a variables the researchers created.
But those issues are kind of aside for now. My first step is to replicate what they did. Yes I need to take their data and try and find their results first. And then we'll see what happens. I am also trying to talk to Michael Sherraden from the George Warren Brown School of Social Work before my next meeting with Itai to get on track with the literature on the subject and because I mean he basically led the concept of microfinance with his theories on asset building for the poor and its amazing that we are at the same university and I didnt realize that. Thanks for reading! |
|
|
| data hunt |
[Sep. 16th, 2006|03:44 pm] |
i need to access this article....grrr
Microcredit and microenterprise performance: impact evidence from Peru
Authors: Dunn E.; Arbuckle J.G.
Source: Small Enterprise Development, Volume 12, Number 4, 1 December 2001, pp. 22-33(12)
Publisher: ITDG Publishing
I am finding some good papers that are sort of dealing with these issues of microfinance and business development. The problem is I can't figure out from the bibliographies where exactly the data comes from?
I called Freedom from Hunger today and left a really awkward message. Everyone cites them in papers though so they must have a really good program for research going on. Need to call again on Monday.
Maybe I need to start searching for gentrification data too as a backup... |
|
|
| get me a job...and some data |
[Sep. 15th, 2006|11:33 am] |
here plz www.rand.org!
So I met with Itai and Benham today both are like FIND DA DATA! I have like maybe 10 days. If I don't find this I need to switch topics. The problem is people have literally (and i do mean literally this time!) had conferences about how bad the data is right now for microfinance and how its hard to do research here. I don't know if I have the skills right now to deal with the statistical problems which would be involved even if I could find a good dataset...which right now I can't. The good news is I love this topic and I'm always thinking about it and wanting to do work for it. The bad news is it may be on its way out...
BUT NO! I will find this data. The search continues... |
|
|
| honors class and microfinance abstract |
[Sep. 7th, 2006|12:11 am] |
Soooo school has officially started which means so has the life of my thesis. Basically, I'm still a bit lost about what I'm doing. I'm still bouncing between microfinance, gentrification, and HFCS. The HFCS model that I found in this one paper for sugar subsidies seems so ridiculously complicated that I kind of want to axe it, but at the same time it also is the one idea with a really clear question (i.e. would a reduction in sugar subsidies lead to a decrease in the use of HFCS). I really like my gentrification idea but the annoying thing is I can't really get any of the papers I found on it in full text. That shouldnt be too hard to solve, but I really need to know the models if I'm going to try and figure this one out. Basically though what's interesting about it? I think what happens to the previous displaced population- in the case of Williamsburg it would be the hispanics and hasidic jews. This takes out a lot of the workforce for factories located here, so does it have a negative economic impact or is this counterweighted by an influx of other economic activitiy (ie. shops, restuarants, bars, ect.). HMMMM
okay but for the abstract I chose microfinance. Against my better desires I will post it, but its pretty thrown together. The data set I talked about with CARE doesn't even exist I just said that cause he said imagine your dream data.
( abstract! ) |
|
|
| microfinance! |
[Aug. 22nd, 2006|09:47 am] |
So you may think I have been neglecting my research because of my posting absence, but it's not true. I've actually been doing a lot of work so here goes...
The newest thing I'm pursuing right now is the microfinance idea. I think I talked about this before but when I say microfinance it's literally what you think of- small loans/savings institutions. They are started in developing countries and can help people out a lot. I emailed my teacher Joy Kiefer and she was really great about it and told me to check out her dissertation on the topic which is about the anthropological angle of a microfinance program in Africa started by www.care.org. She lived w/ a tribe of women in Niger, Africa where CARE's microfinance program was.
Along with this idea is two websites. One which anj told me about - wwww.prosper.com, and one which I found researching, www.kiva.org. The way prosper works is anyone can put up a loan that they wanted funded and then anyone else can fund it in the US/Canada. Prosper works with banks so they can make payments go directly from people's bank accounts. People who want to fund loans bid on them by saying how low their interest rate would be. Kiva is the same idea except its with people to developing countries. So kiva takes microfinance programs in place and puts the loans online. So anyone from around the world can fund these loans (no interest) and you can fund part of them so you can contribute as little as 25 bucks. It's pretty cool. Since they started at the beginning of this year they have had 0 defaults.
So now the problem is I just have programs that I like. There is a lot of data about them available. But I don't have a question. So here's an attempt...
Has the CARE/kiva microfinance program led to higher entrepreneurial activity? Has this led to increases in the incomes of people in the group receiving loans from CARE/kiva? Has the program led to increases in the incomes of people in the society that are not in CARE/kiva?
The problem is CARE hasn't contacted me back and prob won't have as much readily available data as kiva BUT kive is so new that it might be hard to do any legitimate testing.
okay, still working on this...
as my thesis advisor so dramatically said in his email back to me : subject line: struggle, first sentence of body: "Priya, struggle is the name of the game"...
if you read this whole thing thanks! |
|
|
| rut rut rut! get me out... |
[Aug. 2nd, 2006|03:00 pm] |
I'm in a rut. I've been busy than usual, but I'm still in one.
More ideas, still no focus:
Is daylight savings really saving money? compare indiana/arizona to rest of states.
Honor systems- how much do they work? ex. How much more money is made switching from honor system to ticketed turnstile system in a city for transportation
Trying to get more specific.
GENTRIFICATION
---------------
What caused the gentrification of Williamsburg, NYC and how is it affecting the rents and livelihood of its previously latino population?
To do this question I would need: housing data, rent prices broken down by race for nyc, evidence to show a trend for why people started to move there and test the trend
MICROFINANCE
--------------
In a specific case study did the microfinance program lead to more financial independence among participants than regular welfare.
To this question I would need: well kept and unbiased data from a microfinance program, well kept and unbiased data froma welfare program in the exact region and hopefully population. Indicators of financial independence: owning a business,paying back loans, making large investments in purchases, owning one's mode of housing
HEALTH ECON
---------------
Did sugar subsidies and tarriffs lead to an inrease in the use of HFCS and is there any correlation between HFCS use and type 2 Diabetes incline
To do this question I need: sugar subsidy and tarrif data, HFCS use data, type 2 diabetes data |
|
|
| focus focus focus |
[Jul. 27th, 2006|02:38 am] |
"Many of you have provided research questions for Honors in Economics 498 but not all. We need to reduce the honors load for faculty and lack of initiative now is a clear signal."
Okay, where can I get some focus?!? Okay, but anyway I think I'm okay for right now but I have to have a question soon. So I need to throw myself into this. Right now this phrase Prof Sened told me is running through my head...
"all equilibria are local"- what does that mean? local solutions for problems...there's not one overarching world scheme that is going to work for everything and every country
So...
Thanks to Shenbaby's comment I'm going to look into my microfinance and gentrification ideas a little more. Yeah, exactly like that dvd/vhs example Shenbaby for status quo topic. I don't know though how easy it would be to untie all the variables that might affect technology adoption.
MICROFINANCE- Still need to talk to Dr. Kiefer about her study there. But what questions can I ask? I read a paper about how it must be locally integrated like in certain African countries it's normal for a man to provide a service for saving and lending where he literally goes door to door with a can taking collections every week. Or there are sort of shared risk pools of money which are interesting. What could I test specifically though that hasnt been done before...?
I really haven't done anything about gentrification so I'll do that tomorrow.
I found some things on sugar subsidies/hfcs. There's definitely and inefficiency going on with subsidizing sugar when its not needed anymore, but the effects of eliminating the sugar would raise the world price of it. So why the HFCS substitution?
hmm |
|
|
| day 1! |
[Jul. 25th, 2006|07:52 pm] |
I've decided to start a journal about my thesis for a few reasons. I think it'll help me focus and actually be thinking about what I've done for the day. Also, I've found that the more I talk about it with people the easier it is to figure out what I really think. And lastly, if I force myself to write regularly I will force myself to do something each day. That's the hope anyway!
So here we go...
I am a writing a senior honors thesis for economics. It has to be an original study so for those of you who aren't econ-y or have been spared from my constantly worrying and complaining thus far that basically means you are doing a sort of experiment. Think science fair but with out the science and with other variables like money, interest rates, sales ect. So basically the idea is you come up with a question like after event x did variable y increase/decrease whatever, and then you do some statistical junk to figure it out. There are obviously variations on this but thats what it boils down to.
I am researching in part for the Center for New Institutional Social Sciences (www.cniss.wustl.edu). I want to do my thesis on something having to do with developing countries and their economies and institutions but it's hard to find data on that sort of thing. Here are some ideas so far...
( IDEAS )
whoever is kind enough to read this please comment and discuss it would help me out a lot. |
|
|
![[info]](http://l-stat.livejournal.com/img/userinfo.gif){kind=small}
priskool's journal