 I'm going to follow up on look with many of the same observations and data that we are sharing and we are analyzing jointly. But seriously, I'm going to reach conclusions which are eventually somewhat different. We work together on the World Development Report. In a sense, the puzzle is why is it that agriculture remains so badly neglected when the World Development Report was written in 2008? There were about three countries that were meeting the cadept goal of spending no less than 10% of government budget on agriculture. What we see is it went up to 10. I'm not placing causality after the World Development Report. The World Food Crisis game, I think food riots were more convincing than reading a report. The number of countries that met this goal of 10% climbed to 10. If you look at it today, we are down to two. So this sort of a puzzle of the neglect of agriculture, if you look at what's happening with yield that you see on this graph, you see that South Africa is lagging behind, which suggests in a sense there is a kind of trap of law adoption which needs to be understood. So what I would like to do in the presentation today is to say, well, technology upgrading is necessary. There is population growth. There is declining farm size. The answer has to be technology in order for agriculture to perform its role. At the same time, there have been adoption failures. And the failures cannot be blamed on kind of ignorance or bad policies. Most likely the reason why agriculture has not been invested in more is because it has failed to deliver in terms of attempts at pushing technology, promoting technology adoption. You see it's not happening. Governments turn their attention elsewhere and look for other opportunities. So the blame is not on ignorance and bad policies. The blame is on the fact that the recommendations that were made as to how to do it were probably not the right ones. And as a consequence, what we need to do is go back to the design board and find other options that could be proposed as to how to be more successful with technology adoption. So I want to look at the way, in fact, technology adoption has been sought, and which has largely been following what I would call a supply-driven approach, which we said technologies exist and are available for adoption are very profitable. So let's look at the demand side as to what are the constraints on adoption of this existing technology. I'm going to show data that where technology adoption has happened pretty well reaches a ceiling, which is going to be 30%, 40% at most, quite often much less than that, which tells us in a sense that, well, this supply-side approach may not be sufficient. We need a much more demand-side approach, namely identify what kind of demand there is for technology, what kind of traits and what kind of aspects of technology would farmers like to have and then go to the supply side and correspondingly develop the technologies that are going to meet this demand. So I'm going to proceed in seven steps. The first one is to say, role of agriculture where recognize the WDR or so the work that has been done in terms of poverty, the sustainable development goals which have the objective as number one to eliminate absolute poverty. This poverty is mainly role in agriculture also. There's clearly a very strong emphasis on the agriculture and rural areas. You could say this is a selection issue that people who fail to escape poverty remain in agriculture and the solution to poverty is to take people out of agriculture. What we see and look has been important in that in looking at the data is in fact that poverty reduction has to a very significant extent happen in the rural areas and with agriculture and not by moving people outside of agriculture. So what we see is an increasing yield gap and we see that labor productivity is low. My second step is to say why are rural households poor? And there's something quite interesting here is that we would say well poverty depends on labor productivity but labor productivity can be measured in two ways. Either we can measure labor productivity per year or we can measure labor productivity when people work and if we do the latter what we see is that the gap between urban and rural labor productivity per hour worked is not very different and it is very large when you measure it on a per hour basis which in a sense takes us to this idea that there is something wrong with the labor calendars that here at peak time you can see that the upper curve is the urban hours worked per week. The lower one is the rural areas at peak time which is in this case the planting time. More or less there is a similar level of occupation but as soon as you are not at that peak period there's a large gap in employment opportunities between the two and the labor calendars as a consequence are the key reason why rural households are key is that there's a lack of employment opportunities elsewhere that the few periods of the year and here what you see in a sense is the for example the bottom graph you can see that you can add to which is when there is a household enterprise in the household it adds to the labor demands but it does not create counter cyclicality so there are very few activities in agriculture which are counter cyclical to agriculture which means that in this interesting way and Catherine is going to go to issue a kind of household labor decision making what it means that people are not switching activities they're going to specialize in different activities each in one in agriculture others in for example of farm income activities or transformation activities or livestock but it's very difficult to smooth labor calendars in agriculture in the rural areas through finding activities which are counter cyclical to agriculture. So what I'm concluding here is that there is a key role to play to how we look at the labor calendars and for that what we need is to go beyond the productivity of stable crops as Luke was mentioning the green revolution and to look at what we call here the agriculture and the rural transformation. The ag transformation is basically diversification of farming systems kind of spreading the use of labor across the year with crops that succeed each other and as a consequence create labor opportunities in more month of the year and the rural transformation is to look for off-farm sources of income that can be complements to agricultural activity not as in a counter cyclical fashion as I have shown but as a complement to what agricultural employment offers and as a consequence some members of the household can specialize in those activities. So the sequence that we are proposing here is to say well it's important to have the green revolution it's an important starting point it uses a lot of the land it employs a lot of the labor but we have to move on beyond this to ag transformation and to rural transformation in order to spread the labor calendars and provide complementary sources of income. If we look at the way technology has been made available for adoption basically we see two different strategies one is a sort of supplied river strategy which I was saying before assumes that technology is available seeds fertilizer for adoption and that is a matter of kind of identifying the constraints on demand and addressing each of those constraints on demand in order for technology to be adopted. The other alternative is to say well no that's not necessarily the case that those technologies that seem to be profitable are not in fact profitable for many in part because they need complementary factors to the to the technology which is being offered and as a consequence we need to go back to the drawing board and ask ourselves well what what are the what for example what do the extreme non-adopters the extreme non-adopters are sort of your marginal non-adopters people whom you think should be adopting who are in fact non-adopting and trying to figure out why is it that they are not adopting what you are proposing them to adopt and what is it that they would like to see in what is being offered which is different from what is is currently being being offered and then once the this sort of needs assessment has been done to go back to the drawing board in terms of supply not to start with the supply but to start with demand and then correspondingly deliver the kinds of technology and farming systems which are going to meet the expectations and the demands of farmers. So this technology improvement calling the green revolution is actually quite complex in rainfed agriculture 94 percent of the area in sub-Saharan Africa is rainfed rainfed means with a lot of heterogeneity there's a lot of complementary factors that have to be mobilized in order to make the technology acceptable for example if you look at fertilizer what you see is that the fertilizer as currently offered is only acceptable to be 30 40 percent of the land because it lacks either soil fertility or it lacks kind of management of soil acidity and as a consequence without packaging the fertilizer with those other complementary factors there is not going to be adoption the fertilizer is not profitable without those complementary interventions which means in a sense the need to customize the offer of technology to local conditions which is part of the difficulty so a supplied driven strategy would say well let's start from this heterogeneity so sorry so what my next step here is to say well this supplied driven strategy has been addressed to a very significant extent especially using randomized controlled trials a lot of the work done under attire with support of DFID and the Gates Foundation which has in a sense identified many ways of lifting the constraints on demand either in terms of behavior or in terms of market access or in terms of access to credit in terms of access to insurance and as a consequence being able to relax those demand constraints on the technology which is currently available but what we see is that we reach typically a ceiling which is going to be 34 40 percent at most quite often much less than that which tells us that in the end the technology which was made available for adoption may not be the one that was necessary and needed and desired by a majority of the farmers so that takes us on the demand side namely how do we start with identifying demand and they have been efforts and they are at the work that you did in the old times at SEAT the work of participatory reading the work in terms of CL the kind of participatory research with groups of farmers the use of the farmer field schools all of those were kind of efforts to try to elicit demand in order to identify what kind of technology would be necessary right usually the approach was too narrow was focusing too quickly on the expected solutions to the technological gap instead of starting with the problem the problem in essence should be not low yield but maybe poverty to say well if farmers are poor what is it that they would like to have in terms of technology that would overcome the constraints to improving incomes right there's an interesting approach that we were just discussing was just in Linn the so-called science and technology backyard platforms in China what you do is you take teams of scientists to the field and jointly identify with the farmers what are the presumed constraints on adoption and the presumed constraints on adoption may not be kind of not only technological or credit or insurance they emerge broader such as infrastructure such as behavior such as understanding and information right so I think what we need to do is to go back to the drawing board of technology to in order to identify what is the unmet demand and start investing in technology on the supply side only after we have correspondingly address what is it that research should be addressing so in conclusion we said a great revolution for good quality rainforests of South Africa is necessary but hard to achieve due to heterogeneity of conditions so the the main difficulty with the demand side in a sense is that we are talking about customization because we are talking about heterogeneity on the other hand their research is costly and so they are huge economies of scale in investing in research the reason why it works so easily with the great revolution is that it addressed irrigated farming where you have large areas which have similar conditions that you can invest in a technological package which is going to be applied very extensively in terms of area so you can you can capture economies of scale as soon as you customize what you do is you go to a situation where you recognize heterogeneity so pretty quickly you are going to face issues of economies of scale can you actually cost effectively develop those new technologies given the fact that demand is atomized and hence your technological solutions need to be customized and each customization effort is going to have a cost which is going to meet eventually reduced demand because of the so of the heterogeneity of conditions under which technology is being used right so what i'm saying here is that we need to kind of go back to the drawing board to look at not only how to break the constraints on demand to a supply driven technological effort but how to design a demand driven strategy which is going to be able to address the economies of scale in research which is a big challenge the cgi has importantly invested into this in the past it has met with strong limitations there has been if anything a move away from the participatory research participatory reading from the cl approach from the farmer field school and what we need to do is to find kind of new ways especially using it in order to be able to make more scale effective cost effective for me an economies of scale perspective the fact of customizing the supply of technology to the heterogeneity of demand right so that's my conclusion at the end i think the reason why we see a neglect of agriculture and a lack of technological progress may not be so much because of misunderstanding in terms of the recommendations to invest more in agriculture but the fact that the way we have proposed the investment in agriculture was not cost effective did not meet the demands for technology as they correspond to what farmers have we need to go back to the drawing board for this to engage into more participatory research but to find new institutional designs in order to do this in a cost effective fashion so thank you very much