 Good morning. Good morning to all. So we start our day two yesterday after the opening sessions, we had mainly two, two substantial sessions. One was by Jacob introduction to SDG 231 and 232. So we went through all the steps from the goal goal to the target to the indicators and the link with the indicators and the methodology and the plan of action and possible sources for collecting the data. And after that we had the important so the step one for this process is the identification of the smallholders. And Ida Khalil made first of all a presentation on the methodology going through all the steps, all the discussions and why we selected the approach that is being proposed by FAO, meaning that we have a relative threshold instead of absolute threshold. And then once this was decided how to make it work. So she made a very detailed presentation on that and at the end we had a numerical presentation with the data. So she showed the step by step how to compete the threshold, how to combine the three thresholds and then to identify the smallholders. And we had very lively sessions with a lot of questions and very, very relevant questions that we have taken note of. And after that we decided that the time was not enough so people should take the data and try to look at it and see if they can try to compute themselves, the threshold and to identify the indicators to reproduce the example that was shown. So I think that was my brief summary of yesterday. I don't know if some of you have been able to go through the exercise and to get some results. If that's the case we can give the floor to one or two people who went through and if they have additional questions they can ask the questions before we start the sessions of today. The floor is up to the participants. Anybody has tried to reproduce the exercise? Please kindly raise your hand and we will give you a floor. Yeah, we have two people. I'm Tahidul Islam. I have solved the problem. The data that you have that you have sent to my mail, it is okay and I did it easily but in our data, in our data, his data, I feel some problem from the data set. In that data set, we have no number of labor days utilized by the small scale of procedure producers. We cannot find this data but your data is okay and we can solve it easily in the exercise. Okay, okay. So we did it first for getting the below 40 percent small scale procedure. We just sort it in ascending order. Then we cumulative the land size cumulative and then do the step easily. It's not any problem for me. I think any of our participants do it easily but in practical in our data set some figure, some variable is not present. Due to absent, we cannot find the SBZ 2.3.1 and 2.3.2. If we get the data available, then I think it will be easy for us to calculate SBZ. It's 2.3.1. Thank you. Very good. Okay. Anybody else want to intervene before I maybe? On another hand raised. Go ahead. Yes, I have partially complicated the task set but I'm facing some problem and revenue threshold. Revenue threshold between the problem. Can you specify the problem you had with the revenue? That's what I'm finding. This is revenue data. Revenue data which I'm thinking that I could not find out the species data 3.2.1 and forest data. I could not find the species and forest data. That's why I couldn't complete the task. Okay. Okay. Okay. If there is no more question, maybe Aida, you want to say something out of this? Yes, I'm not sure. I understood what was the problem that was found in the revenues. I think he's saying there was no data on the fishery and forestry, which is the usual problems in many countries. Okay. Yes. I think in the fictional file that I've sent, there were data for all the sectors. So crop, livestock, fishery and forestry. But as it was mentioned yesterday, whenever you don't have data on one particular sector, especially fishery or forestry, you can still compute the indicator and identify small holders by looking at the crop revenues and livestock revenues. So if some of the components are missing, you can still produce the indicator, but it's important to flag, let's say, the components that are missing. For example, in the metadata of the indicator that you compute. Okay. If there is no more question, I think those missing, exactly, you are putting the finger on the difficulty with this indicator. So I think we'll have the opportunity to discuss more on this missing data in most of the agriculture surveys and how to try to calculate the indicators. So we'll have the time to discuss that and maybe tomorrow also we can go through the surveys conducted by BBS and see eventually what variables could be added to this survey in order to facilitate the computation of the indicators. So there is another question. Yes. Okay. I have another question. The tax you have given us is mainly to find out the threshold. And I think it is easy for us and not revenue, it's not our tax. The tax you have given, find out the threshold. And I think it will be easy for us to find out. Okay. Good. Thank you. Yeah. Okay. So let's go start the sessions of today. As yesterday, the presentation will be divided in two. One will be a kind of theoretical methodology presentation. And after that, just before the break, we will start the practical exercise. So we want to put more emphasis on the practical exercise where we really will see working with the data how the process works. So I will try to be brief in the presentation of the methodology. As you said, the calculations and the computations are not very complicated. Most of them are straightforward. But the difficulty lies with the manipulation of the data and all the issues related to the data. So I will share my presentation. Go please, work. Can you see it? Can you see my question? Yes. Okay. So I will go briefly on this methodology and steps for computing the indicators 231 and 232. And the total duration is about one hour. But I will try to be shorter so that we have a few time for the questions. Faridun, please let me know if I am close to the time so that I can speed up a little bit. Okay. And then after that, we will go straight to the practical exercises which will be introduced by Jacob and with the involvement of Max and Audrey. So just recall, these are two slides that I took from the presentation of AIDA yesterday. We are still in the target 2.3, the goal 2 and the target 2.3 by 2030 to double the agriculture productivity and incomes of small-scale food producers. Yesterday we focused on the definition and identification of the small-scale food producers. Today we will focus more on the agriculture productivity and the income. So the indicators 231, productivity, and the indicator 232, the income. Now these indicators in the beginning, there was really no internationally agreed definition and methodology established. But now we have such a methodological document, there is a methodological document which is, I think the link is provided in the agenda. And this methodology as AIDA presented yesterday involved the following steps. First of all, to define and identify the target population, namely the small-scale food producer. So all yesterday we spent the time trying to do that. And then after identifying the target population on that target population to compute the indicator 231 and to compute the indicator 232, that's what we are going to do today. So first of all, the indicator 231. Now, we have seen that in the target, they talk about agriculture productivity, but which, you know, there can be many ways of measuring productivity. Basically, productivity is defined as a volume measure of output divided by a volume measure of input. That's taken from the OECD manual for measuring productivity published in 2001. So the productivity measures the amount of output produced by an economic unit could be country, industry, sector, farm, or over-economic operators given a set of resources and inputs. So the productivity can be measured for a single economic entity, such as commodity, a group of farms, and geographical scale, depending on the purpose of the inquiry. So for more methodological discussion on the productivity, there is a publication by FAO in the framework of the Global Strategy, which is indicated here. And at the end, the Global Strategy publishes the guidelines for measuring productivity and efficiency in agriculture. It can be accessed on the web. And the general productivity concept was adapted for the agriculture sector. As it follows, agriculture productivity is a ratio of output to the input. And this simple equation indicates how this works. So it's the output divided by the inputs. Now, this is easy to say, but in practice, we'll see that it's not so easy. So the indicator 231 monitors productivity as a volume of production per labor unit by classes of farming, pastoral, forestry, et cetera. And basically, the formula is the volume of production divided by labor inputs. And I think this started touching some of the difficulties that was raised a few minutes ago. But first of all, let's look at the numerator. In order to standardize and aggregate different agriculture activities, the volume of production is quantified by the monetary value of the agriculture output expressed in constant PPPs. Of course, we cannot add the kilos of rice to the kilos of apple, et cetera, et cetera. So we need to convert all the production in monetary value so that we can be able to summarize them. Now, when we are doing this, this is the conceptual discussion, but when we go to the reality, we see that the data could be coming from different sources. So when we are dealing with a census or surveys, there could be two types of census or surveys. One is when it's a complete enumeration census. So all the holdings in the population are interviewed. So that's a complete enumeration. In this case, there is no sampling rate because all sampling rates are equal to one. So we don't have to deal with to bother with the occasion of the sampling rate when it's a complete enumeration. The other type is when we are using self-weighted design. Typically, I think many counties are using this. So the PSUs, the primary sampling units are selected with probability proportional to size. And then at the second stage, a fixed number of units are selected. So if you combine those two, you get a constant weight. So when the weight is constant also, I think the formula for computing the indicators can be simplified. And as I said, this type of design is used in many, many developing countries where you have, for example, enumeration areas that are selected with probability proportional to size. And within each enumeration area, you select a fixed number of holdings, for example. Now in that case, the formula is very simple. So if you have high agricultural activities, in the ideal case, including crops, livestock, fisheries, and forestry production, and small food producers defined as it was defined by Ida in the session of yesterday. So the SDG indicator 231 can be computed by just averaging. So you calculate the indicator for each holding. You sum it and you average it. So basically, the formula is as it is shown. So the VIJ is the physical volume of the agricultural product. And this must be multiplied by the corresponding price, the PIJ. I forget for women the T, all this should be referred to the same time period, of course. So you see here that we are calculating the revenue. So you multiply the production in quantity by the corresponding price. And then the LDTJ is a problematic one is the number of labor days utilized by the small scale food producer J during the year T. This is also very clear. And there is no complication in it. So N is the number of small scale food producers. So as you see here, you just compute the ratio for each holding. You sum it and you average it by the number of producers. So this case applies when you don't have to apply any weight, complete enumeration or complete enumeration senses or self-weighted samples. Now when we, of course, all the value should be converted into, so all the calculations are done in a local currency, but they need to be converted into PPP dollars. And the information can be found in the link indicated here, was indicated also by AIDA yesterday. So now when we are not in the situation of complete enumeration or self-weighted the sample, so we need to bring in the formula, the weight, the sampling weight. We need to apply the sampling weight and one possibility of applying this weight is indicated here below in this formula. It's the same as previously. The only difference is the introduction of the weight, the WGJ in the formula. So you see that basically the computation is very simple, but as it is said, the devil is in the details and we'll see that the big problem will be the availability of the data required to do these simple calculations. Now first of all, the agricultural activities that are of concern, this is the same as was shown yesterday when AIDA was talking about the revenue. So maybe I will not spend too much time on that because we have seen that yesterday the crops sold, the consumption, self-consumption, etc. For the livestock, the livestock sold, so the live animals that are sold, the other products of the livestock, etc. The same also for the fishery products, the captured fish, fresh fish sold, the fish process and sold, the fish for own consumption and the captured and processed fish and traded fish and processed fish sold also. The same for the forestry. All the value should be expressed in PPP. Now, so that's for the numerator. Now when it comes to the denominator, this is the labor days that has been considered. So different types of labor should be considered, namely all forms of paid and unpaid labor, including family labor, hired labor, temporary and permanent workers, and exchange of labor. So these three categories of labor should be considered when we are dealing with the labor days. Now why the labor day has been selected? Because for computing the labor input, different approaches could be considered. First of all, we could just take the number of workers but you can imagine that this is really too much a simplification because a worker can be working six months while another one is working just one month or a few hours. So the number of workers really is not a good variable to consider. The second possibility is the number of days worked. And we'll discuss more on that. And the best one would have been the number of hours worked. So that would be much more accurate to measure the labor input. However, the problem of data availability makes it difficult to measure the number of hours worked. So the compromise solution has been to select the number of days worked as the input labor input. So the denominator will be the annual number of working days which can be computed directly or sometimes indirectly in some cases. But we will see again what are the challenges with these variables. Now as it was said just a few minutes ago, even this information is not widely available as we'll discuss later on. So for the 2-3-1 and because of the challenges that we'll discuss in detail, the problem has been so far that a few countries has been really able to compute the 2-3-1 because of the challenges related to the data. So if you look at the numerator, as it was stressed in both Jacob and Ida's presentation, one of the difficulty is that all the variables need to be aggregated as at a holding level, so at farm level. So in most agricultural surveys, you can have different types of levels of collecting the data. It could be plot level, it could be farm level, etc. So all these need to be aggregated at holding level. And all the variables ideally should be coming from a single survey so that in one survey you get all the variables and you are able to aggregate them for each holding. So you have a line for a holding as we have seen in the table shown yesterday. So you have one line for each holding where you get all the variables of interest in order to compute the indicator. And for that, you need some kind of integrated surveys. Examples could be the FAO Agris. You don't know if you are familiar with this, but you can maybe talk more. I think Ida will talk more about that just tomorrow. Or the World Bank LSMS ISA, which also is an integrated survey or any kind of integrated survey where you have the unit is holding and you collect and aggregate all the variables at the holding level. Or it could be also administrative sources in some advanced countries. They have a farm registry where you have for the entry is holding and you have all the variables that are recorded in the registry. To give an example of the challenges related to the aggregation at the holding level, you know in many countries and I think Bangladesh is one of the countries also where crop cutting is used to estimate the production. So the production is derived from the yield, which is coming from crop cutting. But this crop cutting is done at plot level with a sample of plots. It's not for each individual farm that you do, but you put the squares 2 by 2 or 5 by 5 in a number of plots and you compute the yield at the domain level. So now how do you get the production for the individual holder? That's one of the challenges. So in some countries they have the possibility of asking the farmer directly the amount of his production and this can be reliable in some countries it is reliable. In other countries it's not reliable and that's why the crop cutting is used. So in case you have the possibility of asking directly the farmer his production, the problem is solved because you can record the production of the farmer for each one of the crops. In case you don't have that and you use the crop cutting, you need to find a way of deriving the farmer's production out of this crop cutting. Now this is an open to discussion. So one possibility could be to once you have the yield for a domain to affect that yield to the farmer because you know the area of the farmer. This area is known. What is not known is individual yield so you can affect the yield for the domain in which the holding is to affect this to all the holding's concern. So you will have a production by multiplying the yield by the area of the holding. The other problem has to do with the prices and the ideal situation would have been to have the price for each commodity, each crop if possible farm gate prices. Now this also is a difficulty in some countries and we need to use some proxies. It can be rural market prices if it is available or overprices in the country. If all these are not available, there is also FAO database, the farm start where you can look at the producer prices for each country. So these are some of the difficulties, the data difficulties that we have when we want to compute the indicators and the other problem has to do with the coverage of the agriculture surveys, especially the agriculture production surveys. Usually they focus, many of the surveys focus on the crops and you have seen that we are talking about the revenue of the small holder. If you want to be really exhaustive, you should include not only the crop, but also the livestock and the fishery and the forestry. So you should cover all these activities because you can underestimate the revenue. We have seen that, for example, in the case of Calvary. When you calculate the revenue only on the crop, you may end up with zero revenue because if the reigning season is not good, they don't produce anything and if you were able to get information on the livestock, I'm sure that the revenue will certainly be positive. They can sell some milk, some meat, etc. All these can be additional revenues, not only the crop. So that's one difficulty and even if some of these activities are available, for example livestock, usually you can have only the number of livestock, but you need also the production of the livestock. So you need also the production of the fishery and the forestry and in many cases those are missing in the agriculture production surveys. So you can see that the point is here that we need maybe to revise our agriculture production surveys and to see how to what extent we can introduce some of these variables in order to be able to compute the indicators easily. Now it's not that if you don't have those, you should not calculate the indicators, but you need to report which activities are included in the indicators, number one. The second, you need to keep consistency between the numerator and the denominator. For example, if you have only crop, you should not include in the denominator the labor days allocated to the livestock or the fishery or the forestry. So it should be consistent with the numerator. If it is only crop, it should be livestock days for the crop activities. If it is crop plus livestock, the denominator should be labor days for crop and livestock, etc. So it's important to maintain consistency between the numerator and the denominator and importantly to indicate this in the metadata that for this country, these are the activities included in the computation because of the activities the data is missing. So these are some of the challenges and I hope you can have some discussion on those challenges and see how we can overcome them in the case, particularly in the case of Bangladesh in order to facilitate the computation of the indicators. Now I come to the denominator which is, you know, the labor is a difficult variable to measure in agriculture surveys because of the high presence of seasonal and part-time workers because of the, especially for the small farm, you have a large number of proportion of family members used as workers and the long time that is needed for collecting the data on the labor in the surveys. And also you need to be very careful because this variable is very sensitive to non-sampling errors. The measurement errors could lead to some inconsistencies in the data. So that's why at this point that we are making this training, there is the discussions going on within FEO to see first of all how we can help countries to introduce the labor variable in their agriculture surveys, particularly by adopting the agris and also in parallel to see if over factor productivity can be considered. Of course, any factor productivity adopted, the formula will remain exactly the same. You just need to change the denominator into the factor that has been adopted. So the process of identifying the small order of computing the indicator will be exactly the same. The only change will be that the denominator will be the factor which will replace eventually the labor. But in the meantime, we are trying to help countries particularly through the agris survey and the over surveys to be able to collect also data in labor days. And we'll see that there could be some indirect solutions to estimating the labor days. This will be shown by Max in the presentation on Calvary. So this is just an example of agriculture output per labor day. So the number of dollars per day for a worker, that's for different countries. You can see that most of these countries are really included in the agris in the LSMS, World Bank LSMS survey. The number is not very high, as you can see. You have less than 20 countries that we have the data on. Fortunately, this year with the EU, the number of reporting countries has jumped. We have more than 20 countries for the EU. So the number has increased substantially for reporting on this indicator. So I don't know if I stop here. We have a short discussion or I go directly to the second indicator. Kamrul, what do you think? I think let us discuss all together. You finish the second part and we discuss. Okay. Now the second is more or less the same type of presentation because the indicator 232. So the 231 is about labor productivity. And the 232 is about the average income of the small-scale food producers. Of course, that should be disaggregated by sex and indigenous status. So the computation of the farm income of the agricultural holding adopted by FAO includes, again, the four activities ideally should be included, the cropping activities, the livestock activities, fishery and forestry. And in the previous presentation, we are talking about the revenue. Here we are talking about the income. So the difference is that we are talking about the gross income, which is computed by the revenue minus the cost. Normally, we should also take into account the depreciation of the assets. But usually information on that is not available, the stock variation. So we end up by calculating the income with revenue minus the cost. All has to be expressed again in PPP, of course. So here again is the same as previously. If the data is coming from complete enumeration or self-weighted sample survey, no weight comes into the formula. And the formula is straightforward. So as we have seen before, so it's the revenue minus the cost for each activity. And then this is averaged by the number of holdings. The notations are the same as previously, except the cost, which is the production cost of each agriculture product. So for each, if you take rice, what is the production cost of the rice? What is the production cost of the, I don't know, the maize, et cetera. So, and then we compute the income from each one. We sum everything and we divide by the number of producers. Now, when the weights, so when we are in a situation where we need to introduce the weight, the possible formula is the one shown here also. Now, it comes to what we mean by revenue. What should be included and what should be considered as costs and what should be included in the cost? So if we take the crop, so we need to include in the revenues with the plus, the crop which is sold, the crop for own consumption, the crop used as feed, the crop used, which is stored, the crop used as byproduct, the crops given as gift, the crop saved as a seed, the crop used for paying the labor, the crop used for paying rent, the crop used for paying inputs, and the crop given out in share cropping agreement, and the crop wasted. You will see that some of the items are in both sides. And then on the cost side, we have the inputs paid in cash. Here is not shown, but it includes also the inputs we use for the material like tractors and things like that. So, all the payment in cash for the input, the land which is rented, so rented from someone, you take his land, you pay him back. So that's included in the cost, the extension cost, the crop saves for the seed, the crop used for paying the labor, the crop used for paying the rent, the crop used for paying inputs, and the crop given as a share cropping agreement, and the crop wasted. So that's for the crop. And the byproducts of the crop should be also included, including the product that are sold, the products used for income payment used for own consumption. Here we are talking about the byproducts. So the product which has been transformed somehow, not the direct production. And on the other side, we have the crop used for the byproducts and the total value of the input purchase. And we have also the crop received in share cropping agreement. So that's in the revenue side. For the livestock, we have the livestock sold alive, the livestock given away as a revenue. And we have for the livestock activities, the change in cash value of the stock, and we have the livestock bought, the livestock additional expenditures, the crop used as a feed, and the extension cost. Maybe it's not necessary to go through all of this line by line. You can see it in the presentation. The same should apply for the fishery, where we see here the different items that should be included in the revenue and the cost. The same also for the forestry. Now, as we have seen again, the formula in itself is simple. But the problem is to get the data for the indicator. First of all, decide the difficulties with the revenue. We have here the additional difficulty with the cost of production. So we need cost of production data for each one of the items commodity that is included in the revenue. And I think that in Bangladesh, you have a lot of some surveys on the cost of production. If my memory is good, last year I visited Bangladesh, but so cost of production surveys are very, very important for the computation of 232. And in many countries, they don't have the data on the cost of production. And this is absolutely necessary in order to compute the 232, because the income is revenue minus the cost. The other difficulty is the lack of a detailed price data for the calculation of a holding level revenue and income. So again, in order to compute all this, you need the price data, not only globally, but a detailed price data for each commodity. And sometimes this information is missing and it makes the calculation of a revenue income 232 for some countries, including Bangladesh here, Bangladesh for 2010. So again, same question. What is the situation in Bangladesh regarding these challenges? And how can we discuss possible solutions for overcoming the challenges and competing the SDG 2.3.2? That's all for the presentation. And the floor to the participants for questions or clarifications. Thanks a lot, Mr. Naan. Before anyone from BBS and other agencies raised their hand, let me just raise some issues for you to respond on. Say, for example, you have no considered labor and categorized into different categories, family labor and then higher permanent temporary exchange labor, something like that. But in Bangladesh, we have experienced that we have male laborer and female laborer and they are not paid equally and the expectations from them are not equal. So I think in this one, to calculate this one, we have to use some sort of weight for women and then finally get one. So for Bangladesh, when you will discuss, we have to keep in mind. And for the second indicator, you indicated cost of production. Cost of production services are there are many in Bangladesh, total 13 or something, but these are not regular. So if in the survey, we can ask the farmer about their production costs, which are the components they are producing and then side by side their production costs. In that case, directly, we can get the actual production cost. Otherwise, production cost varies across the country. Sometimes labor is cheap, sometimes input is cheap, something like that or sometimes it is costly. So it should be built in the survey so that there is no other data missing or underestimation of the indicators. This is my first observation on the methodology. Now you can expect questions or suggestions from BBS, but I would request many of your time know how much time you allocate for this discussion. It is up to you. Thank you. Okay. I think Farid and how much time do we have? 15 minutes. 15 minutes. Yeah. Let's use the 15 minutes. In that case, I would request BBS, one person, first one person take the lead and respond and others may supplement. Okay. So who will take the lead? Mr. Kamdol or anyone? Yeah, we have a hand, Akhtar Hassan. Please hand. Yes, Akhtar Hassan. Okay. Okay. Thank you very much, Naman. Yeah. I have a question. Not a question, it's a problem. We say this is this indicator is individual holding data collection, but when we go for the cop cutting method individually, it is very difficult, but BBS collecting the cop cutting method and by hello with interview method. We are following two ways. One is cutting method and the interview method. Do we go for collecting the data in individual two, three, one? For interview method, it is easy able to contain the indirect. Otherwise, using the cop cutting instrument where calculating the lead is very difficult for the server. Okay. But the firm has declaration, is it reliable? Reliable. Reliable. In the server period, the server will need the cops. Otherwise, we can't use the cop cutting method. That's why we go for the interview method after harvesting holders easily understand about the data. Okay. In that case, I think simplifies the issue. But Mr. Naman, in addition to Mr. Hassan has just said, this survey is with a minimum set of questions and these are asked to sample the commerce. So the other related information for the indicator. So you have to plan one single survey that will incorporate all the necessary variables in one. And again, the sampling frame will be a little bit different because this time you are incorporating fishery sector, livestock, land also. So I think the survey Mr. Hassan is talking about it is only crop. And again, crop based only for some major crops and minor crops as well, not covering all everything. So a farmer should ask whatever the properties and even livestock and forestry and fisheries they have, you know, elevating. So one in the independent service needed. The BS cannot support with any of the existing service. That is my understanding. Okay. Anyone else? Can I interview Naman? Yes, yes. Yeah, go ahead. Okay. On this issue, I mean, one thing which I a little bit differ from Amirul is like, you know, we, it's too difficult for to impose countries for to conduct surveys for each and every indicator as much as possible. It will be really good to address, you know, an integrated approach therefore to integrate in the existing system rather than proposing, you know, a new survey methodology. You know, therefore, we are going to discuss tomorrow. What is the existing system so in which in type of things that can really incorporate the collect data that's imported that's needed for SDG 231232. I think that should be the cost effective approach. Otherwise, it will be very costly to propose, you know, you know, stand alone survey for each and every indicator. That's one, one point I want to make. The second point which I wanted to make is on the crop cutting thing. We need to be very consistent in that thing. You know, in my opinion, the, which method are you using to report the volume of production? You know, you report volume of production, not total production in this year in Bangladesh and for rice or whatever, whatever. So is it based on the crop cutting result or is it from the farmer's declaration data collected? Therefore, we need to be very consistent. Therefore, if the volume of production that you are reporting is based on the farmer's declaration, we can still use the farmer's declaration for computing this indicator. If the volume of production is being used based on the crop cutting, then it might be too difficult to use two different approaches in the same country in terms of you know, computing the volume of the production. Therefore, it depends. Therefore, if the volume production you are reporting is based on the farmer's declaration, we can consistently use that one for this 231232 as well. These are the two points which I wanted to make. Thank you. Can I step in as well, Naman? Yes, is that Arbab? Yes. Okay, so you are silent. Go ahead, please. So actually, I have two points to make. First one is related to this concept of revenue that you just explained, right? So the revenue and the cost of production. You know, it's obvious from the presentation that from revenue here, we mean value of production or value of output or output value. Because you know, this concept of revenue gives the impression that we are partially taking into account the production which is sold in the market and the piece which is not sold in the market, which is not part of the revenue, but self-consumed is somehow out of the scope. So this value of production concept is all encompassing whether the product produced is for market or is for self-consumption. Secondly, I agree with Yaku that we shouldn't be proposing a new service for the indicators. Now, from this perspective, the denominator of 231, which is on labor productivity, relies on the same kind of information that is collected for the numerator of SG241 under the land productivity sub-indicator. So there, we are concerned with the value of output produced by the agriculture holding in a given year, and the denominator is agriculture land area. In this case, again, I mean the numerator is the value of output generated by the agriculture holding and the denominator is labor. So from this perspective, I mean, once the information for 241 is collected using this dedicated server which the BBS is planning on, which we have discussed thoroughly in our part of the present training, then perhaps that the same information can be utilized to inform the numerator of 231 as well. So this is the point that I wanted to get across. Yeah. Okay. Participants and Kamrul, you know our colleague, Arbab. He conducted the training for 241. Arbab, you know, welcome, Arbab. Thank you for your, you know, may I add something here, you know, as Arbab and Yaku just mentioned something here. Say for example, you know, as I was involved with, you know, BBS for two years in NAMI's project, you know, as the international consultant, we have, you know, once all of the, you know, possible surveys, available surveys, their methodologies. And, you know, one thing is the problem with your equation. So we have supplementary information from other surveys, but your equation suggests that for single household level, we need to have information for all. So even if we have information in other survey, we cannot fix one common household for that we can array all of the information. That will be a problem. It is not the national level figure you calculate and then you divide something with someone else. It is not that it is it you have to, you have to array all of the information against all the households. So how can you ensure that households should be common in different services? Because different services have their own purposes. Some you may find some, some commonality, but that won't represent the entire country. And this is one. And another point was raised by Yaku. Constant production, you know, sorry, the crop cutting method, it is not for all the crops, mainly some measure crops and mostly other for others, it is face-to-face interview. And again, the government of Bangladesh, they report yield on the basis of the crop cutting. Where is crop cutting? Where crop cutting is available. Otherwise, they based based on or rely on the face-to-face interview. So to do information, I thought before we further, you know, you know, proceed with the discussion, you need to know. So the whole idea of, you know, integrating different surveys, I think we need your assistance or suggestion on how to incorporate the point I have just raised. Otherwise, we have several surveys. If you can give some clear-cut idea, how to incorporate those variables from different service, then BBS can try. Otherwise, I don't find any amicable solution right at this point. But in here, we can discuss it further. Thank you for my segment. Thank you very much. I think tomorrow also we'll have the opportunity to have more discussion on this very, very, very important issues. Because tomorrow really we'll be discussing the sources of the data. We'll look at the surveys conducted by BBS and to what extent those surveys could be amended in order to include variables of interest for computing these indicators. Of course, I think as it was said, it would be more cost-effective if the ongoing surveys could be amended in order to include variables. Because one of the very challenging requirements of these indicators is that all the variables should be coming from a real survey so that you can aggregate everything at holding level. Now, we have two international initiatives, or maybe three. One is the LSMS ISA. Some counties are doing that, and this is providing some useful information. We have the AGRIS, FAO AGRIS. Hopefully tomorrow we'll talk more about that. And there is this new initiative, the 50 by 30, that is coming on. All these are possibilities of assisting the countries in order to expand their ongoing surveys to include the variables that are required for the computation of the indicators. So I hope that tomorrow we'll have really, particularly with the presentation of the BBS surveys to see concretely what are the variables included, what are the variables that are missing, to what extent we can try to include some of the relevant variables into these surveys. Regarding the source of the production information, yes, I remember that exactly the two methods are used. The major crops, if I understand, are using the crop cutting. And the minor crops are using farmer's declaration. So the official production information is based on these two sources of information. Now, the challenge is when for the major crops, when you use crop cutting to know exactly how the methodology is working and how the aggregation is done, and if there is any possibility to derive the holding level production using the crop cutting. Regarding the revenue, this brings me to one observation I wanted to make at the end of this presentation. In FAO, within FAO, the methodological note is undergoing a revision, actually. So we are in the middle of the revision. So all the questions that you will raise here, we'll take them and see also how we can improve the methodological note, taking into account those observations, including what Abab just said about the revenue. So shall we replace the term revenue by over things? If it makes it more clear, maybe that could be something to consider. But even for the factor productivity regarding the labor with all the difficulties with the labor, discussions are going on that is there any simpler way of computing the productivity, but no decision has been made. For so far, we are using the labor productivity and assisting countries to include the labor variable in their surveys. Now, also we are discussing the proper way of including the sampling weight in the formulas. So we have shown one example here that we'll demonstrate. But just to say that we are in the phase of revising the methodological note and finalizing it. It will not be major changes, but some small changes may happen in the methodological note. So for the moment, I think that's what I have to say regarding the Amrull's comment about the male and female. Since we are, what do you think? Because here we are concerned with the amount. So we are considering the number of days worked by the worker, be it a male or a female. So do you think we should discriminate between these two? Now regarding the asking the farmer the cost of production, yes, if this can be done for each survey, that would be very good, of course. That would be much more actual. But even a simple cost of production in some countries, they don't have that. That's the problem. But I know that in Bangladesh, even if it is not so frequent, but at least data on cost of production is available. So if we can improve by asking the farmer for each survey, his cost of production, that would be much better, of course. So I don't know if we are still in the time, if we have exceeded the time, I don't know. Maybe we have Khatum raising his hand. Maybe, can we give a chance for one more question? I don't know. Okay. More questions? Khatum? Yeah. Thanks for the questions. The possible sources for SDG-221 and SDG-221, I think there are only two surveys at Khatum in India. One is household income and exchange of the survey. And another one is agricultural sample centers. Our regular activities only cover the production of selective crops on it. There is no livestock information. Also, there is no forestry and future information. And also, the crop information is known as the household-level information. We just have two processes. One is for the calculation of fields. That is, we have some select seed clusters where crops are only for the seed crops. Other crops, we only use the farmer's infrastructure. We don't have household-level information. Also, we don't have how much they store, how much they store, and the other even parts. So only from production, we kind of measure 2.3.1 and 2.3.2. And for 80 years data, 80 years data, they don't have the labor work that is oriented, a number of labor days. So this is acceptable. And another problem is we have post-production survey. But that is not regular basis and that is not for all the crops. So we have selective post-production survey and also selective crops. So there are difficulties in this whole process. Maybe the good source of 2.3.1 and 2.3.2. So we are looking for that. And another observation is for Mr. Arbaq. He mentioned about land productivity. But here we see labor productivity. So land productivity and labor productivity, I think there are some problems. Thank you. Okay. Thank you. I think as I said, probably tomorrow we'll try to have a more detailed discussion on the data availability and how we can work together in order to introduce at least a minimum set of variables in the ongoing surveys. So because we know that if you introduce more variables, it has a cost. So but can we see how with the ongoing surveys, we can modify slightly some of the things or the way questions are asked in order to get to orient, to make it more SDG friendly in a way. And examples are in the agris because there are some questions, questionnaires and etc developed by agris. We can see to what extent some of these ideas could be introduced in the ongoing surveys in order to make the surveys more SDG friendly, particularly 2.3.1 and 2.3.2. And maybe over indicators. So are there other questions or shall we move to the next? Please check the chat. There is two more questions. Okay. Can someone check for me? Because it's more a consideration than a question. So Amirul was mentioning the problem of finding, you know, a unique something wait when you combine different sources to compute an indicator. And also he mentioned that the sampling should be modified to cover crop, livestock, fishery and forestry. Because perhaps, yeah, they have data sources that cover only one sector. Yeah, I think that's probably a very good, particularly the second question. Certainly depending on how these activities are distributed within the population, you may need to revise your sampling design in order to ensure that they are well represented in the sample. So certainly now the first question to combine data from different sources, ideally it should be all variables coming from an integrated survey. Of course, you have, you know, link survey, there is a whole literature on the linkage, record linkage, etc. But this is very, very, very complicated. Some countries are more advanced in that than others. But it requires a lot of technical work to try to link the different surveys and to get them integrated in one and to analyze the variables. So what is advice is the approach of integrated surveys, meaning that in one surveys, you try to collect most of the data that you need. This is the same type as the LSMS, as the agris, and maybe some countries also are doing different type of integrated surveys. But so you go to the farmer, you ask all the questions that you need in order to do your computation, basically to simplify the approach. Of course, it has also some constraints because the questionnaire may be too big, etc., etc. That's why the approach in the agris is going by rounds. So you are not collecting everything at once, but there is a rotation in the modules. So it's a modular approach. But I think I don't need to expand on this now. Maybe tomorrow we can talk more about those things. Okay, any more questions? No, Manifai, you can add something on what you were saying about record linkage and other methods about integrating data. One of the difficulties there is that they need to be planned at the design stage. So when the single surveys are designed, if then they need to be integrated, these should be planned. So when you design the sample, I don't know, perhaps use similar sampling frames or sampling frames that are connected. So perhaps after getting into the details of your data sources, we may be able to say more on what can be integrated and how. Yeah, and you need a single identifier for all the Yeah, okay. Any other question? Or we have finished all the questions, we can move to the next session. Kamrul, can we move to the next session? Yeah, there is no other question. But probably main concerns are already raised, you know, from our discussion. So if there is any tomorrow, that may be raised. No problem, just possibly the next presentation. Okay, you can start thinking of what we have talked about many problems. Tomorrow, we hope we will talk also about possible solutions. So think of the solutions tomorrow. So we come up with some plan of action. What the FAO could do in order to help Bangladesh to integrate some of these variables in order to be able to facilitate the calculation of the indicators. Now we move to the next session. And this is very important also. As yesterday you have seen, you know, the computation themselves are easy, but the problem is the data. First of all, how you manipulate the data, how you put them in the proper format, and then once they are in the proper format, calculation is not complicated. So today we'll see from the presentations, from an example, a concrete example, how you go from the questionnaires, the plot data, et cetera, how do you combine them up to getting at the point where you get this nice table that I showed yesterday. And then from that, how do you compute the indicator themselves? So this session will be led by Jakob and our two colleagues, Max and Audrey will present the computation that they have done, and they will demonstrate how it works. Jakob, you have the floor. Okay. Thank you, Naaman. Very briefly, like as has been said, you know, this computation of these indicators in my opinion can be subdivided into three stages. One is having the data sets, you know, having surveys which can really generate the variables that are required. That's one thing. Second is organizing the data so that it will be easier to go straight to computation. The second one is just apply the formulas and compute the indicators. So the most of the discussions that we have this morning, we're focusing on the first part, like can we have, you know, the survey data itself, you know, which type of surveys, how can we integrate, we'll use an integrated approach to make sure that all the variables that are required for this type of, for this indicator, the computation are available. That's one thing we are going to discuss in more details in the Bangladesh context tomorrow. But the important thing which I wanted to inform you is Bangladesh is not exceptional. I mean, the reporting level for 231232 is really, really very low because of all the issues that we have been discussing this morning, getting all the required data. You know, in terms of, you know, the variables look like simple, like, you know, it's volume of production, prices, then labor, you know, mostly these are the variables. But, you know, under each component, it's more detailed. How do you get the production? How do you get the price? How do you get the number of working days in different, you know, production units in crop, livestock, fishers? These are the challenging things. Therefore, how can we really make sure that we can really be able to generate this data? We'll discuss it tomorrow based on your system. That's one thing. Second thing is, okay, we have the data, we conducted this as a survey and we are promoting this type of data should be collected through farm-level surveys. Okay, so once you have this farm-level surveys, how can you just bring this data into the format that Ida was showing yesterday? That's what we are going to see this morning so that we will be able to compute the indicator using the formulas that Namanu was showing. Therefore, Max will really take an example of Kevverde data. For this demonstration purpose, we took data, I mean, micro-level data, farm-level data from Kevverde. Therefore, he will walk us through all the steps until we get the structured data sets which will be used to compute the indicators by applying the formulas. Then, ODR will take that output from Max and show us in Excel again how to compute these two indicators in a very simple way. These are the things that we are going to do. Once they finish those things, here at FAO, we try to develop an application which is going to be a very simple thing. Once we really structure the data as Max is going to show us, there is a small application that we can demonstrate. It is not like officially launched but it is a potential utility. We will demonstrate or we can quickly show that application as well at the end of the session. Therefore, it is more practical. I will leave the floor to Max to start discussing the whole procedure using sample data from Kevverde. Max, you have the floor. I'm sorry, Faridun, where are we with the time? We should be on the break now. Let's take 10-15 minutes for the presentation of Max. I won't take long. It's about 10-15 minutes. My presentation will cover the steps from survey to organize the structure of the dataset. To calculate the indicators, we mainly need full parts of data, the revenue, the input of labor and land and cost and sampling weight. Unfortunately here, I cannot show the cost because the cost of data is not available in Kevverde. I started with here showing the revenue. It's straightforward. We collect all the output from each plot owned by the household times the price, sum them up and then convert the revenue in local currencies into PPP dollars. The link here will take you to the World Bank PPP conversion factor. The Kevverde structured the questionnaire in this way. First, you ask the household how many plots you have. Here they have three plots. Then you ask the details of each plot. For plot one, what is the plot size? What's the main crop? Then you ask them the production. What is the quantity sold? What is the price sold? What is the quantity for self-consumption and for other purpose? Finally, we collect the unit of production because some of them measure the output by weight. Some of them measure the output by volume. We need to convert all the volume into liters and all the weight into kilograms. Once we've done that, the collected data set production would be like this. We have the ID column, which is the household ID. The plot ID will identify for one household which plot is the data coming from. Then the plot size, which will all be converted into hectares. Then the crop. For the household number one, on their first plot, which is two hectares, they produced 120 kilograms of corn, out of which 50 kilograms are sold at two. Then the other 50 is for self-consumption with the last 20 for other purpose. That's the clean data set for production. As for price, some of them were reported price, but because they didn't sell any of their output, so price is not available. In this case, the price is taken from the FAO step producer prices with the link showing on the slide. Here is the price is organized in this way. We have the crop ID, which is important because we need this crop ID to link price with the output. Then we have the crop name, the price, if it's measured in kilogram, or the price, if it's measured in volume. Sometimes in very rare situations, they measure the same output, let's say corn. Some of the households that measure the output in kilograms, some of them measure 18 liters. Then we needed to have a conversion so the same price can be expressed in both weight and volume. After that, it's the section for labor input. The labor is divided by types first, so we have a temporary workers and household members also capability includes teenagers. After that, we know the labor type, we may ask them the labor activities, which is divided into four main categories. The first one is land preparation, next one is sowing, weeding, and harvest. For each main activities, there are specific activities. For example, land preparations, the labor input on clearing the land, applying the manure, transportation, all that labor input is recorded. The final labor input would be the total labor days in all these activities and all labor types summed together. This would be the survey, we ask the household what is the activity. If they identify the main activity as preparing the land, we ask them how many people are involved in these activities, and then they report the average hours of each worker. So we can time the number of people with the average hour of each worker to get the number of hours of input. Of course, we also know the labor type and the specific activities. So once we get all that organized, the labor input can be organized into a table like this one showing on the slide. We have the household ID, plot ID, so we know all the labor input is on which plot of land, and then we record the labor type, what's the specific activities, how many people are involved in these activities, and what is the average hour for each person. We can time the last two columns to get the labor hours and then convert the labor hours into labor days by assuming that one labor day equals eight hours. And then for land area, it's pretty straightforward. The main thing to keep in mind is we need to exclude land rented to others, but add the land rented from others. So in this case, for this example, we have this household, their own plot one and plot two, the rented plot three from others, but they rented the plot four out to some people. So the total land area is the size of plot one plus two plus three, and then convert that size into hectares. Then for the collected data, it would be organized into this table with the ID, the plot ID, and then the cultivated area. Sampling weight is straightforward. We have the household ID, and then the sampling weight for each household, which can be organized as a table. After we collect each separated table, then we can aggregate each specific table into the household label. So for example, for the revenue, we needed to add all the revenue from all the crops, from all the plots of land into one single number and assign that to the household. The same thing applies to the labor input, the land, and the cost of the data if it's available. So we will have four main tables. For table one, we have the household ID and revenue. For table two, we have the household ID, land area, labor days. For table three, we have the ID and the cost. For table four, we have the ID and the sampling weight. So for all four tables, share the same ID, which is unique to each household. Then we can join these four tables together to form the final clean data set. Each row will be one household, and the main columns would be the revenue, cost, land area, labor, and the sampling weight. After we organize the data setting in this way, then we can go to the next step to identify which household is considered as a smallholder and then calculate the two indicators. That's the end of my presentation. Okay, Jacob. Okay, this is the end of the presentation, as Max was saying, but he will have some more details on explaining the codes in producing each of the tables that he explained and the code that he was using to come up to generate these tables and gives you the final data sets in Excel file after following all the steps. Then we can do that one after the break, maybe. Okay, so yeah. Sorry, we have a question in the chat. I don't know if we want to address it now or later. Is the auto compute considered a land size for shared cropping? So maybe, yeah. I guess this shared cropping is preferring to mixed cropping, and so you have several crops in the same plots. No, there are two things. I think one is a mixed crop, as you said, but that one can easily be calculated by completing the percentage. I think this question is like shared cropping, because you have one plot, and then that plot is being used for shared cropping. Two households may use that particular plot for a kind of shared cropping. They share the output at the end of the day. Can you specify your question, please? Yeah, it was my question. So for example, a piece of land, as Yaku was telling, a piece of land is cultivated from investment from two to person. So kind of rent in that type of situation. So this same piece of land may be considered in two households. Will you split into half or I don't know there should be one mechanism? I suppose there is a way of dividing the production between the two, no? In this case? Yeah, here it says 70-30. So we can consider 70% of the land in one household. As a proportion, we can take the proportion. That's to raise the issue so that we don't miss anything. Sure. I mean, if it is managed by 2% 70-30, we use that proportion to come up with the revenue for this particular household. As Naamanu is saying, if that particular crop has two crops, mixed crops, like maize and barley for instance, and if maize is 20% and barley is 80%, we just calculate that percentage for each particular crop when we multiply by the cost in the production. Yeah, yeah, that is also one option. Okay, so let's take a 15-minute break and Faridun, at what time should we be back? Ideally, we should come back at 11.50, but we can come back at 11.55 if you announce 15 minutes break. This is the time in the rooms. There's more 55 in Bangladesh, 455 in Bangladesh. 15.55, 355. Okay, thanks. Okay, and as there are few end rates or there were, maybe after the break, we can address other questions. So, there was Mr. Taidul Islam that asked a question. So, perhaps when we come back, we can, before going back to presentations, we can give some space. Okay. Odri, if I'm speaking from a audition, please join us at 3.55. Please don't go out from the webinar, just being here and you can have a rest. Yeah, we will stay connected. Yeah, maybe Naaman, you raise a... Before we go to the next presentation, I have a question to Max. Max mentioned in slide number seven, land, your labor activities included. He mentioned here, land preparation between waiting and harvesting. And in the prospect of Bangladesh, there is two gaps. One is seed bed preparation and there is tracing and cleaning of crops and irrigation also. These activities involve in labor. So, it is, it is, it is depend on cost of production on seed bed preparation, tracing and cleaning and irrigation. They will be included here for calculating the cost of production. Irrigation and what, what is the second activity? Irrigation and seed bed preparation. Our holder, they have two types. One is purchasing plant and another is planting the seed and another is tracing the crops and cleaning and another is irrigation. Irrigation needs a leaf labor. Okay. Let me clarify. He indicated that in Bangladesh, as indicated by Max, there are three or four activities during the field activities. But in Bangladesh, there will be more three categories of labor in inclusion. So, in the questionnaire, there may be three more variables and in the calculation, we can have those. Yeah, you need to include all the costs. So, everything, even if I think the covered example is maybe a simplified one in Bangladesh where maybe one example farmer are purchasing the seed and they, they prepare the plant and they need a, they need a seed bed preparation. The seed bed preparation needs a labor and another is after harvesting, they need a labor for tracing and cleaning the crops and in standing crops, they need a labor for irrigation. Yeah, I think you should count all the labor work for the activity. Yes. Shall we continue now? One question. Mr. Tavit, Mr. Tavit, you wanted to ask a question. Mr. Tavit, can you raise your hand? Yeah, yeah, yeah, yeah. Raise your hand. Can you hear me? Yeah, you can hear me. My question is, in the case of joint landowner, in that case, what will be the weight and what will be the number of size? That means N. Is there any chance? So, can you explain better? In the case of joint landowner, landowner is joint. In that case, is there any chance in the weight or in the size of N, number of denominator, what will be the denominator? You know, if you don't get the point, let me just, you know, reiterate it again. Say, in Bangladesh, there are so many cases that one piece of land is owned by two person. So, and then similarly, like share crop, that piece of land is used for cultivation. Then how to incorporate this, you know, land size in the equation? Okay, so it's a kind of, it's a one holding with two, let's say two owners. Because the definition of the holding is like an enterprise. So it's, the holding is considered as a unit. If they don't have other activities aside of, if they are only doing this one plot activity, but if they are doing over, if you go back to the definition of the holding, you know, in the FAO publication on agriculture census, for example, you will find the definition there. And if holder A and holder B have separate activities, agricultural activities, and only they are working on one land as a, that they have over land or over livestock, et cetera. So in this case, probably there are two holdings, but you need to find a way of dividing the activity on the plot. Now, if they are only like a cooperative or like a joint, people are only doing this activity on the plot, this normally should be considered as a one holding. I don't know if it is clear enough or but those cases. Is it clear? I think it's clear. No man, you explained it very, very well. Is it like, you know, what is the purpose of that? If it is an enterprise type of thing, it's just one. If it is like, you know, both are earning their living from this piece of land, then this is going to be counted as a stool. I can send you the link to the publication of FAO on the concepts. So you will find there are discussions on this same issue in that publication of the agriculture census. You know, you have different types of holdings. And the key point is agriculture activity, not selling products and things like that, but involved in agriculture activity, crop activity or livestock, et cetera. So this is, if they are combined and doing that together, like an enterprise, that should be one holding. Now, if they have separate activities, it should be two holdings in that case. Max, can we continue? Before I start, can we have a show of hands so I can have an idea of how many our users are in the participants? If you are our user or have some experience with our language, can you please raise your hand? Is it possible to count the hands raised? Yes, but for the moment, there are no hands raised. No, I'm sure there are people in BBS who knows very well R. Kamrul, your people, I think there are certainly data people, IT people who knows R, no? Hello. Kamrul, hello, hello. Yes, maybe Max, you repeat the question. Can you repeat the question? Yes, before I start, I just want to have an idea of how many participants are familiar with the R language, so I can decide how deep I go into the details of the R code. So for most of them are used to data, but are there some? No, not all, not all. But you can go logically so that whenever they are, so that they can follow. But allow a number of questions when they just find any problem. Okay, I will focus more on the logic behind the code. Because if you understand the logic, they can also replicate in the status. That's right. So this document contains all the steps from the survey datasets and all the steps, clean the dataset, organize the dataset. The output is the clean dataset as I just presented in the last slide. So first step is I loaded the packages that I will be needed for the clearing the dataset. And then I begin organizing the revenue dataset. So the keyword dataset for the revenue is saved in this file. The first step I replaced all the NAs. Some of the NAs is represented by negative 9999. Some of them are represented by this symbol. So I replaced all that by NAs. And then I identified the cultivated land area, which is this variable S2 area. But the unit is in square meters. So I divided the land area by 10,000 to convert all the land areas into hectares. The next step I did is to convert all the units. So some of them, they measure the weight by grams. Some of them measure them by 25 kilograms, some of them are in 50 kilograms. So I converted all that unit into kilograms. If it's measured in liters, I also did the same conversion. So after these steps, I will have the output measured in kilograms, all liters, just these two units. So then after that, I combined the dataset together. Because in the main dataset, all of the crops are identified. In the last step, they ask, do you have other crops? And that is saved in a different dataset. I call it the other. So I combine the main dataset with the common crops with the other dataset, which have some rarely grown crops. In this case, it's tomato in Cape Verde. So I combined the dataset together with the full join function. Then I calculated the revenue. So in this case, the crop revenue equals to the output if it's measured in kilograms times the price in kilograms plus the output measuring volume in liters times the price in liters. And then I sum it all up. So group the revenue by the household ID, the parcel or plot ID, and then crop ID. And then I summarize the crop revenue divided the total revenue by the PPP factors to convert all the revenue into PPP dollars. And then the last step here, I group all the revenue by their household ID, and then summarize the crop revenue. So all the crop revenue within that household is summarized into one single number. The output is this table. I have the household ID, and then I have the revenue for the entire household. In this case here, you can see there are some zero revenues for the entire household. It's because in this specific year, it's a very dry year, and a lot of the households, they simply don't have any revenue. Next step is the code to cleaning the labor input. So mainly the labor input is saved in four parts as I presented. So the labor used for preparing the land, the labor used for sowing, weeding, and harvest. Again, the first step is replacing the missing values by NA. And then to calculate the total hours for each specific activities here, to calculate the total hours for preparing the land, it is the product of this variable, which is the total number of workers. And then the second variable is the average hour spent by each worker. So the product will tell me the total hours on preparing the land for a specific plot of land. I did the same thing for sowing, weeding, and harvest. And then I summarized each activities by the household ID. So the sow data set here is the total hours spent on sowing for the entire household. The same thing applied to weeding, harvest, and then I joined the data sets. All the hours spent on preparing the land sowing, weeding, harvest, I joined them together. And then I have the total hours of all activities, which is the sum of total hours on preparing the land sowing, weeding, harvest with the NA variables removed. So then the labor hour hold this data set contains the household level, labor input. And then the steps here is adding variable levels. The important part is here, I divided all the total hours by eight. So the hours are converted into labor days by assuming one labor days equals to eight hours. And then I will get the data set for labor days, which is the household ID, and then the labor input measured in days for that household. You can see here for the first household, the labor input for that entire day is only 2.5 days. The reason is that for Cape Word, the land is really limited. So for some households, they are only working with half or even a quarter hectares. So in this case, the 2.5 days may be only sowing and then after that, they have a very dry year. So then there's no additional labor input. After that, I work on the land input. So read the land data, replace all the missing data with NA. And then I summarize that to make sure that for each household, which is identified by the ID, and then this ID identifies each specific plot. So once I group the data set by the household ID and the plot ID, I should only have one, which is in this case two, because I don't want to double count when I calculate the total area. In this case, once I grouped the data set by the household ID, I can just summarize the area of each specific plot. That the total number will be the total land owned by that household or cultivated by that household. So here you can see the minimum household plot of land owned by the household is one. For some households, they own up to 14 plots. So that's a very big household owning lots of land. And then here the data set not only have the area, it also have the sampling weight. So I added the sampling weight, which is the last column. And then the cultivated area for that household is the summation of the size of each specific plot owned by the household. So in this case, for the first household, they own 0.5 hectares of land. And also you can see here, a lot of the household, they have less than one hectares of land. So up to this point, we have the data set on revenues, the data set on labor input, and then the data set on areas. Then the next step would be join the data set. Because all the data set that share the same ID household ID, I can join the data set by the household ID. Before, before doing that, I added another column, which is the county where the household is located in. This step can be skipped if you don't want to group the indicators by counties. So here is the summary of the data set. The first column, I have the household ID, the county code specifying where the household is located. And then the revenue of the household with the minimum household, they have zero revenue. And then the maximum household with the $10,000 of revenue. The labor days minimum input is zero, which I will filter in the next step. And then the maximum labor day input is 82 days. The county weighted area minimum is zero, which also needs to be filtered in the next step. And then the maximum is 4.9 hectares. So here the last step, I filtered to only keep the complete cases. And then I filtered to make sure all the labor days is greater than zero. Because to plant anything, you have to have a positive input on labor. And then the cultivated area for the entire household needs to be greater than zero. After that filter, that will give me the final clean data set, which is presented here. You have the household ID, the first column, the county where that household is located in, the revenue for that entire household. And labor days cultivated the land area, the last column is sampling weight. So in this case, for Cape Word, we have 883 unique household. And then once we get the data into the form like this last table, the next step would be easy. Jacob, do I present the app now or later? Maybe later. Later. Okay. So then that's the end of my presentation showing how I take the survey data set, organize them to the clean format. And this data set will be ready for the next step to identify smallholders and calculate all the indicators. Yeah. Okay. I don't know if there are very burning questions now, or if we leave Audrey to present the second path in the data. Cameroon? I will seek any kind of suggestion or question from the BBS counterpart, because they are the people who will finally calculate this indicator. So if anything, Confucian is there for the timing, they can raise question. But I understand most of them are very good with these kind of quotes, especially in Stata, but our quotes are very much similar to now. Okay. So maybe let's go to the next one and at the end we can raise this. Okay. So do you hear? Yeah. Okay. So I will present the practical exercise we're using Excel file. I'm going to share my screen. Okay. Can you see Excel file? Yeah. Make it a little bit bigger in our context. Yeah, that will be fine. Yeah. Okay. Okay. So the output coming from the process described by Max. Yeah. Yes. As you can see here, we have four columns. The first one is for the measurement of a lent labor, the second one, cost of production, the third one, and the sampling weight, the last one. So this data is coming from Cape Verde. The cost of production was just generated by Max for the exercise purpose. But over real data from Cape Verde. So the first thing when we have this data is to identify the small scale food production for the target 2.3. So for the two indicator, 2.31 and 2.32. So as you can see here, we have an exit sheet here named threshold for identification. So the first step is to identify small scale food production. For Cape Verde, we have just data for land and for review new. We don't have livestock. If you remember, on past presentation, we have for identification free variable revenue lined in livestock. But for this example, we don't have livestock. So we will do the identification using only revenue and land. So the first thing is to identify the threshold for land. Here you can see I can make it bigger. We have the land size for each household here. We need to complete the cumulative distribution of land. The first thing is to order the household decreasing land size. So we can select the data file and sort using the variable end from the smallest to the largest. So it is okay. And then you can compute the cumulative distribution of land. So I will do it right now. And then this plus this. And I compute the formula until the end. And here the maximum value of the cumulative distribution equal to the last digit here. And I can compute 0.40 times the maximum value of the cumulative value. So I have 14.71. So you have to look for the value of a land that correspond to this 40%. You can see here we have 0.9. So for land we have I can put it in yellow to remember we have 0.9 here. So if I go back to the data file you can see here I can put the digit here. I go back here and select and I have the first value of the threshold for land size. So I go back to the threshold sheet and we will determine the threshold for the revenue. It is the same process we have to order from the smallest to the largest and compute the cumulative distribution. So you can just show the results here. Okay. So we have already the result here. It is here in yellow. So I will do it quickly. So it is automatically computed here. So I will select your research here as well. Okay. So this is okay for the threshold. And here we will compute a dichotomy value one for small scale production and zero for overs. So I will you can see the formula here. So we need for the land is here less or equal to the land size. I think there is a L missing L for land. What? If it is land no? Yes. No it is before we like it is a function and to have a junction it is a function excel function it is not the name. Okay. Okay. And the second one it is the revenue here. You see here the cell L3 you need to fix it because it is the same pressure for all the thresholds. So I use the sub function with dollar at the left and at the right. And I go for the revenue less or equal to the threshold here. It is the same. I need to fix this value. And when we have one for small and zero for overs. So I need to have a formula. So we can see we have 27 small holders on this data file. So the next column is the computation for the numerator. If you remember the formula for 231. In the case we have something we applied we need to multiply the weight by the revenue divided by the labor per day. So this is the first column for the first and equal to 231. In order to compute the 232 we need to have a revenue minus the cost of production because we don't have stock variation here. So we have just revenue minus the cost of production. As we have sampling we have to multiply by the weight. So it is a simple formula. So we can compute here. So the sample weight is applied by the revenue divided by the labor. So I can apply the formula. And for the last column we have a sample weight multiplied by the revenue minus the cost of production. So I can apply the formula. Here we have the total. You can see here the total. And here as well the total. We have the total of the sample weight. It is the total the size of the population. So it is within a minute of the both formulas. So for the first indicator we have just to divide this total by the total of sample weight. And for the second indicator 232 we have to divide this total by total of the population. But the next step is to filter and to get all small orders and to filter again to get all small orders. So it is just to go through data and filter. I will keep only the small order here. You can see. So we have 27 as I said. So I have just to copy and pass to the next Excel file. So here we have we need to compute the total. Yeah, total for the sample weight, total for this and this. So you can see for the SDG 231 we have to divide the colon H with times revenue divided by the sum divided by the total of something weight. So we have the result for the first indicator. And the second one here we have the weight multiplied by the subtraction revenue by cost divided by the total of something weight. So we have the SDG 232 result for small orders. So I can go back here and filter again to take no small order and it is the same process. Copy and pass here and we will get the result. We have the total here. So we have the result here. And here I can put the summary of result just for presentation. So labor productivity for small order. I go here. I have a non-small order labor productivity here. And for HGG 232 with him, I go to small order file and non-small order SDG 232. And that is the result we have for everyday exercise. So we can see here we have the result of HGG 231 for both small order and non-small order here. And the SDG 232 as well. So that is for this computation using Excel 5. I don't know if you have some questions. Okay. Maybe Max can also talk about the application now. Yeah, Max can quickly show the application and then summarize. Sure. I will demonstrate the app. Let me share my screen first. So this would be the final data set. I have a little bit more columns here. This will be illustrating the ideal case. Here we have bigger. So I have the household ID, the land area, a livestock unit, the revenue, the cost, sampling weight, and then the gender of the household head, whether that household is in rural area or urban area. And then the last one is whether it's indigenous or non-indigenous. This data is missing. So I have put it here to illustrate what happens if we have missing data. Once we prepare the data like this, then it's easy to upload all the data set into this app here. It's an R Shiny app to calculate the indicators automatically. I just need to browse the data set and then it's done. So this is the original data set as we just saw it. The next tab will automatically identify the smallholders by the land, the livestock unit, and revenue. The third tab calculates the indicators for smallholders, non-smallholders, and then group it by the gender of the household head, the location, and the indigenous. So here because we have all the missing data, so the indicator is for the entire sample. The last tab will automatically calculate the SDG 232, also group it by smallholders, non-smallholders, and the gender of the household head and the location. So once we organize the data set in a clean format, you can upload it into this app and then the app will do the rest of the work. Jacob, you wanted to say something before you end the floor to the questions? Yeah, I think this session was made to show you how we can really practically compute this indicator starting from the raw data, from the survey data, and clean data from Cape Verde. We took, I think, one district data as a sample and then we just tried to produce the whole step. Therefore, ODI shows the Excel computation, but the app can also produce the same result. But the important thing, in my opinion, is how to get the data in that straight format, which Max was really showing the whole logical step by creating about four tables from revenue, land, weight, labor, and merge them together so that you know how to have a structured data sense. After that, either you apply Excel or use the application that the app that has been developed, it will really simplify the whole thing. Therefore, as I said, the first problem we are going to discuss tomorrow, can we get such a data in Bangladesh context? That's one. The second and the third problems are, I think, resolved. There are the how to connect the data and how to compute the indicators. And if there are any questions, I think we can leave the floor for discussion. Yeah, so up to you. If you have any questions, clarifications, or comments, feel free to raise your hand. I should add that, of course, the case presented for Cape Verde is very, very simple because, as you have seen, it's only crop. There is no livestock. But we are trying to work with Cape Verde in order to amend their survey to see if also livestock will be added. That's one. The second is that this is a rain-fed crop only, but they do have some irrigated crop also. So we'll work with them to see if they can include also that because this makes a big change in the revenue of the holder. Rain-fed is very, very random. If there's not rain, there is no production. So it may have some irrigated land, may have some livestock, et cetera. So in order really to not underestimate the revenue of the holding, you need to make some effort to include a minimum set of agricultural activities, not just the crop. And even for the labor also, for Cape Verde, it's broken down into four types of, but in Bangladesh, it's much, much more certainly as someone was saying. There are many other activities that require labor, so we need to take those into account. And we need to adapt what has been shown here to adapt to the context of the country. This was just presenting the logic on how to go about it, but it has to be customized and adapted to the specific case of the country. I don't know if anybody has any comment or questions at this stage. Mr. Naman, I think the way the slides were presented, the logical sequence is already understood by the participants. And again, certainly this dataset is a little bit easier than the complexity that may arise in Bangladesh context. So I think from the presentation, I understand finding the appropriate information either from one survey or other survey. It is one important one. And cleaning the data to fit into this assigned shape as Excel file, that is also kind of cumbersome. But the way it was presented, probably the logical sequence is already understood by the participants. When they will have their own data, they will find their ways to how to gain this data. Because at the end, we need some assigned columns for this calculation. And as the program is there, very much easier. This program is there, so Excel file is there, so we do not need to go for the calculation of the equation. So equation is in building the system. So once data is ready, VBS will be able to, without understanding the internal mechanism of how to write those equations, it will be easier for them to calculate this one. But the main concern is to find appropriate dataset and covering all the aspects of the indicator. And just if that is there, clean the data accordingly. This is my opinion. But I think any of the participants have any opinion or any question, because they are the persons, they will at the end calculate this indicator and report to FAO. So I will request anyone, you know, if Mr. Tahitul Islam or other, you know, they are very, you know, actually active, you know, in this kind of an exercise. I will request who just respond to the next. Mr. Tahitul Islam, you may allow him. Yeah, I understand the logic. It's okay. But it would be good if you send us the do file. If you send us the data with the help of a data, it will be good for us. Okay. Sorry, may I take this request? So as we said yesterday, we will share a stata file, a template, but the template is generic. It's not, you know, you will have to adapt it to your specific case and your specific data. Yeah, yeah, I understand. So yes, we can share a template with you. And then, yes, you will have to adapt it to your data. Yeah, it would be good. Thank you. Thank you. Yeah, as for this application that Max just showed, I think after this meeting, probably we'll have a technical meeting within FAO in order to go through the application. And once it is validated by FAO, it can be freely distributed to all countries if they want to use it. So in the next maybe weeks or months, certainly that application also will be available for anybody who wants to use it. Yeah, that is fine. And I think, you know, there is no hands raised. But even if there is any, we have, you know, tomorrow to just, you know, discuss these issues. Do we have anything left in the agenda today? No, can we just do the same as we did yesterday to see if, since you all have the data now, if you can kind of, as a homework, look at the, go through the process yourself. And tomorrow, as we did today, so tomorrow, you can ask if some of you have been able to go through the whole process and get the results. And if there is any additional questions, and we can respond to those questions. No, this is a good proposition. I think, you know, like today, tomorrow, we can also do the same. And I will request all of the participants to try at home. But individually, I want to request Mr. Tahid al-Islam as he is found to be very much, you know, involved in our question and discussion. So I request Mr. Tahid and other colleagues if possible, at least go, you know, go through this process and try yourself if you can find any appropriate solution. And by doing that, if you find any problem, we can also discuss this on tomorrow. Yeah, this will be very helpful even for us also. Because if you find some problems, maybe we can improve the applications or the way we do the calculations. So your input could be also interesting in that way. Yeah, yeah. And by this time, I saw Saleha send the PowerPoint to you. You can also, you know, share with others, you know, the PowerPoint slide. And if not Mr. Saddam, if he doesn't have, you know, has not yet paid the same PowerPoint files, I'll request, you know, do that one so that, you know, the organizers had those slides beforehand. So that will ease our investigations. Yeah, that would be very nice. Yes. So, if there is no more. So regarding maybe this is some common, send this presentation to our email. So if I do, you have all the presentations for all the sessions, so that can be shared, right? To all the participants. I don't know exactly which one I need to send. If somebody will give me this, the name of the presentations which I need to send, I will do it, no problems. But, you know, you can share your PowerPoints that you have, you know, shared with us to all of the participants, methodological issues and then you have shared the Excel files and other data files, but the PowerPoints if you share, for future reference actually. Yeah. Okay, so I will tell you which ones should be sent to all participants after this meeting. Also please include when you send those documents. Yeah, now I, as you said, it would be nice if we have also the two remaining PowerPoints, but even if we don't have that, we will send already the presentations that we had. Okay. Okay, no more points, no communication pharism from you regarding the organization. So we do as we, we do tomorrow as we did today. Yeah. Okay, so today basically, as you have seen, we have gone through the the calculation of the indicators, which is not really the most challenging, but the most challenging is a data issue. And hopefully tomorrow we'll have a really good discussion and good proposals on how to overcome the gaps, data gaps, at least to have a minimum set of data than that can allow to compete those two indicators. And as I said, FEO has some initiatives going on. Probably they can work with Bangladesh as they are doing with other countries to help them into getting the ongoing surveys more SDG friendly. It's better to, if possible, to work with the ongoing surveys because there is a infrastructure in place done to try to, to, to invent a new survey and all the logistics that goes with that. And we have models that agrees the LSNS and OBS that certainly we can borrow some ideas from those and the, including the ongoing surveys. So with that, I think we, we can stop here for today and resume the, yes, someone wants to. No, just I was saying, telling me yes that I was agreeing with you actually, we can say no quality today actually. Okay, so we stop the session for today and tomorrow morning we start again at two o'clock Bangladesh time. Thank you to everybody and see you tomorrow. Thank you very much.