 This is very preliminary work actually and we're looking at the very immediate effects of public works intervention in Rwanda and in recent years public works interventions have actually become quite fashionable I would say again with policymakers particularly in low-income countries because The main why they are so attractive is they have a kind of double dividend So they provide employment on the one hand and on the other hand they do construct physical infrastructure Both sort of important aspects to actually promote and enhance growth if you're looking theoretically however These public works programs which are intended to well provide a transfer to poor households and then increase their consumption And in the long run or in the medium term even their investment capacity safeguarding them against shocks or enhancing their capacity to actually Overcome shocks more easily in the in the longer medium term They might not actually work out that well because there's crowding out effects that could happen So actually you could sort of disinvest into agriculture and in the end up with sort of lower real income effects or Because you provide the certain level of certainty What might actually happen is that households do not start investing in accumulating assets as you intended to do Because they no longer have this precautionary savings motives because they can start and rely on the transfer So even so it does have this natural Quite appealing idea there might actually be some confounding factors And if we're looking in the short run on well here, I've put evidence from The PSNP in Ethiopia, which is pretty similar to the program in Rwanda actually if we're looking at their short run effects And what they find is actually they don't find much changes in consumption They don't really find asset accumulation in the short run So households that receive transfers over 12 months, they don't actually improve that much and we'll actually now look at what's happening in Rwanda and The main sort of approach is that we have a two-round household panel And we look at three indicators. So we look at food consumption asset accumulation and here we're looking mostly at investments in livestock because that's turned out to be what most households do tend to invest in and we look at crop investments and this crop investment is basically just an Indicator to maybe see if they are potential sort of positive medium-term effects So if they actually start investing more into agriculture as sort of a strategy and Well, what do we find? We find that the public works participants are actually households that are relatively better off to those that are targeted by the intervention surprise But on the other hand we do see that households that receive the transfer do increase their consumption in food and they do start to invest already in the short run However, if we talk to those beneficiaries, they all indicate that those improvements are very much short-lived So once off the program, they actually go back to their old state So the vision 2020 Moringa program is the flagship anti-poverty program that is run in Rwanda at the moment It was launched in 2008 and it has three components. So the public works component is the largest component And then it has a direct support component and a financial services component so the direct support component basically is like a Safety net providing a cash transfer to households that are unable to participate in public work So that don't have members that are able to work and the financial services is a complementary component Which basically provides credit to poor households The implementation of the program is faced So it started in the poorest sectors in each of the 30 districts in 2008 and then it has been expanding nationally with an intended Well with the intention to reach national coverage by 2016 at the moment the program is covering about half a million households So compared to the program in Ethiopia It's rather small. So they have about 1.7 million But looking at the size of the country, it's quite quite remarkable still The beneficiaries of the VUP are actually selected by the communities themselves. So they have no Means tested approach, but a community approach to select their beneficiaries and the selection criteria is a two-fold households have to be categorized in the bottom two Ubudae categories and Ubudae categories are Well traditional wealth categories and the bottom two are basically households that live in abject poverty So that don't really have a proper shelter that have to beg to survive that don't have any means to participate in the normal labor markets so those are really the ultra poor and Those households do have do have to have at least one an adult member that is able to provide work So that can actually physically work to participate in public works Initially if the households do fulfill those criteria, they are eligible for public works participation for 12 months And then there's retargeting so how so the the communities go back and they re-ranked the households in the community And then they see if you're now better off than you were before obviously you'll be ranked off two more important aspects of the public works organization wages are set locally to not interfere with the local casual labor markets and the VUP actually Directly transfers the money to beneficiary bank accounts. So beneficiaries have to open a bank account with a local institution Positive aspect is that well it reduces leakage and the risk of corruption on the other hand and it increases sort of Financial literacy among the beneficiaries on the other hand opening the bank account is not for free So they have to pay And in addition households that participate do get some sort of training So they have these regular meetings where they are told about how to best use that transfers They are sensitized to actually save parts of their transfers. They get education on health and Sanity measures and things like that Here a few sort of impressions of what these public worksites look at so the biggest one is So the biggest one you see on from you, it's the right these are the typical terracing works and this is actually where most of the Activities were happening in the early stages and Here to give you a bird's view of what has been generated in terms of public infrastructure through the public works program so far and as you see sort of the largest Projects that were run were actually very much the typical land rehabilitation projects And in Rwanda given that it's a very small country with a very high pressure on land This is actually a quite important aspect in the later two years now There's an increasing sort of shift from this land rehabilitation works to actually physical infrastructure of schools health centers more roads Water networks, yeah, and here the general character Statistics on participation We see that households that are eligible is the first row and the households that have actually benefited We see we have more eligible than actual beneficiaries. So there's some sort of rationing going on One point that is maybe interesting is the average days that is worked per household is around 40 days Initially when they started the program it was intended to actually provide a full year of Work for the households. So that's also not happening. So not only is there rationing, but there might also be some sort of rotation on the projects And Sort of while most of these public works programs and assessments at the moment still don't really are not really able to assess the Community level impact one thing that is maybe interesting is that in the early projects, which were very much focused on this land Rehabilitation so if you look on the last row There was a higher share of the actual project costs that were transferred as wages Whereas later on when we're now generating this physical infrastructure The the the transfer that directly goes to the beneficiaries is like much smaller in percentage terms of the total cost of the project Well to tell you a bit about the data that I'm using for the analysis and it's data that is coming from the normal VUP M&E activities which started up in 2009 and it's some household survey data that was conducted well collected in three cohorts Cohort one is the cohort where the program was launched in 2008 in cohort two the program was launched in 2009 and in cohort three the program was not yet launched at the time of the survey But will be launched in 2010 so for cohort three the 30 sectors We basically have a proper baseline data and then they did a follow-up survey covering all sort of households in the first survey Again in 2011 which basically means that for cohort three We have follow-up information of the households for which we have baseline 12 months after which is quite nice I'm here just graphically to show you where all those sectors are and the sectors that will focus for the empirical analysis are the ones that are in green We'll also do some matching in the paper, but I'm not going to go into detail on that, but yeah, you can read the paper For the approach Some descriptive statistic of who are actually these public works household so if we're looking at baseline characteristics of those public works participants and we compare them to Eligible households so that would fulfill the targeting criteria, but that are not participating. We actually see they are not that much different. So However, there are few sort of distinct factors the households in general They all have around four members between one and two members that are able to work Whereas the non eligible or the eligible non participants have maybe a higher dependency of elderly within the household, so maybe they have Need to provide some capacity to take care of somebody which reduces their ability to actually go and work outside They are pretty poor. So they live in single structures with earth floors. Most of them have land holdings below 0.25 hectares quite small and what we see in terms of targeting the actual Participants while they should be only in Ubud a category one and would be a category two about one third is actually in Ubud a category three. So those are actually the relatively less poor So here are some reasons on why households Don't participate why they can't and why they do participate when talking to two actual households that are eligible There's a lot of information. We got that some actually even so they were offered the position They didn't take it up and when we asked them why didn't you take it up? They gave us well five main reasons So one of the aspects was that the distance to the public works site is too far So most of those sites are pretty much located close to sector administrations because it's easier to kind of organize works around there And for those that live further away from the administration then obviously for them It's harder to get to so we had this example that while you would need to get up at 4 in the morning in order to arrive at your Public Works place at 7 to be there and by that time. I mean you're already tired The work is really hard labor intensive The wages are set too low For the kind of works a household do really perceive them as well I could have better opportunities and maybe go to the capital or the next bigger city and work there Households are unable to pre-finance two weeks expenses and setting up the bank account And I think that's an important issue that has to be targeted So households only receive the transfer after two weeks of actually working and for households that are really poor They have to survive on a day-to-day basis So for them actually to be able to pre-finance these two weeks to participate is the biggest inhibition I would say and Another aspect that has been also mentioned is that households do have certain responsibilities taking care of children Taking care of elderly and households that only have a single member actually said well I don't want to go and participate in public works. I have to stay at home somebody needs to take care of my property So there's obviously selection not only because Households can't participate because they're not enough jobs, but there's also this issue that households Voluntarily choose not to participate But then there were some cases where the project was really big and how they actually did offer Work to households that were a bit richer I'll skip this because I said I'm not going to talk about the matching so what we do it can Econometrically in order to assess the effects is we do double difference estimate we compare before and after situations Using household fixed effects model and we control for household characteristics This maybe just this comes from the matching actually But what is interesting is actually when we look at the determinants for public works participation Within the data it actually pretty much kind of confirms of what households have been telling us So what we do find is well if we're looking at the household members if you have an elderly member You're less likely to participate the distance to the administration seems to matter So the further away you are the less likely you are to participate if you have participated in the community much Selection process and actually you are more likely to participate maybe because you do have the information And we also see that households that are a bit better off are actually equally likely to participate as other very very very poor So households in category one and three are having about the same sort of Magnitudes so the typical public works households are these would a category two households But not the really really poorest This I can skip and here are the results So look we're concentrating on the last two columns Which basically shows the difference in different Estimates and what we do find is that households do increase that per capita food consumption by about 150 random francs per person Which is about 20% an increase in food consumption, which is quite substantial And if we think of while having a four-member household and receiving a transfer of about a thousand random francs Per day then actually it means that about 60% of your transfer is spent on food Well, that's back on the envelope calculations, but just to illustrate the effects another thing and if this effect is quite strong actually we see that households do indeed invest in livestock and While they increase livestock holdings by about point three, which is about one point well one one and a half goats on average if you would sort of translate it into something to associate with and We find some Positive indications that household actually increase investment in crops So we see a higher use of fertilizer when talking to them We see that they actually use better or more seeds and different varieties But those effects are well a bit weaker I would say Then two more points I would like to raise is well basically we're asking them Okay, what drives your investment decision? Why do you invest and what do you invest and why do you spend the money the way you spent it? An important aspect that kind of came up on why households did buy these livestock things is basically well They got the idea from these training and mobilization activities so households were basically told it's a good idea to buy livestock and They seem to have followed the advice and then given that most of the Literature on these public works programs still does only look at the media well short to medium-term impacts we actually tried to well at least Qualitatively look on what's happening in the bit longer run while we still don't have the data and What households do indicate that well while they have been in the V up? situations have gotten better and They are less poor than they used to be but as soon as they are off They're actually going back to their own levels So they have to start selling assets again They have to go back and start work like they used to work casually before which obviously doesn't give them the same level of income than what they used to have and on those that actually managed to invest in and kind of are driving to diversify in terms of their livelihood strategies what they demand is they they want more support in terms of training actually starting up a business or something so there is motivation there to kind of move on And I'm done. So this is the last slide just to summarize One of the drawbacks of the study as it is is obviously that we have a very small sample of actual beneficiary households So we cannot really explore any heterogeneity in the effects when we talk to people what is kind of coming out is that there's Definitely different investment strategies amongst men and women Whereas men tend to invest in to cows bulls so the larger livestock women tend to invest into small ruminants chicken goats rabbits and more into Education actually We find that it's there might be an issue with targeting because it's actually better of households that do participate And but we do find evidence that this public works does improve food consumption and agricultural investment Which is in line with the program of Jeff objectives in the short run and as I said qualitatively these effects might only be temporary because Households do feel that they fall back to their initial levels And obviously I think what should be done at the policy level then and here's the policy implications In terms of program design definitely these obstacles to participation should be addressed somehow either by subsidizing Setting up bank accounts Or changing the payment schedules in either way. Thank you you