 As I'm here, I've been asked to do my paper, this one will follow. Thank you. This paper that I'm now presenting relates to it a study that we undertook recently for the National Department of Social Development in South Africa. It was completed earlier this year and presented to members of Department of Social Development, South African Social Security Agency, National Treasury and Finance and Fiscal Commission. But other than that presentation that took place, this is its second airing in public and we have special permission to present it here today. Our task for this study was not to explore rationales for universalising, introducing a universal child benefit, but rather to look at the nuts and bolts using a tax benefit micro simulation model to explore the ways of delivering and financing a universal child benefit. I'll say a little about what the current child support grant is and why there's interest in it becoming universal and the two main options that there are for delivering a universal child benefit, then take you through the process that we've underwent in simulating the current grant, the cost of a universal benefit instead of the current means tested grant and the options for financing it. The child support grant is a form of social assistance paid to primary caregivers aged 16 and over and subject to a means test of the caregiver and their spouse if they have one. It's paid for dependent children under the age of 18 who are not in receipt of either the foster child's grant or care dependency grant for children in need of full-time care. In 2016 it was payable at 350 rounds per child per month. There's no limit on the number of biological children who can receive it, but a limit of six of non-biological children. And there's a so-called soft conditionality requiring score attendance. A means test is set simply at 10 times the value of the grant. And why is there an interest in universalising it? Well, 80% of all children in South Africa are currently eligible for the grant and yet about a fifth of those eligible children don't receive it. The means test in numerous studies has been shown to impede take up of the child support grant. It's an onerous process to demonstrate one's income status even if one has an income but even more onerous often if one doesn't. And there are financial and time demands on the applicants. Cultural barriers to early application. The application process itself can be stressful and erosive of dignity. And the means tested element promotes stigma and a majority of attitudes towards recipients. And the goal amongst those interested in the universal child benefit is to move from the CSD as being emblematic of poverty to becoming a child benefit that's a social right of citizenship and an expression of social solidarity. So, whilst it might seem very straightforward to consider a universal child benefit, there are numerous elements that had to be taken into account such as the extent to which it would be compatible with the South African constitution, compatibility with institutional mandates, in particular the South African Social Security Agency and the South African Revenue Service, how we actually define universal as there are in fact lots of different ways. The conditions of entitlements, the age criteria of the child, whether or not we implement the cap still for non-biological children, citizenship status, how it intersects with the other graphs, the amount and who the applicant would be and root on and off of the benefit system. Just on two of those, we simply defined universal for our purposes as meaning that the child benefit is payable for each child in South Africa irrespective of the income status of the child, their caregivers and any present or absent parents or any other person in the household. It's simply the stripping away of the means of the current child support grant. In terms of the intersection with other child grants, as it is becoming universal, it means it can no longer be impossible to claim it as well as the more generously paid foster child grant or care dependency grant. So what we did was to assign a universal child benefit to each child and then top up that amount to the equivalent of what the foster child grant is, what the care dependency grant is, all for the small number of children who are both disabled and fostered, they're entitled to both of those two well paid grants. So I had a bit of a ripple effect in terms of simulating it. Regarding the delivery options, the two main options available are, on the one hand, to implement it as a cash payment through DSD, through SASA, by simply stripping away the means test. The other option would be to introduce it as a tax rebate through the South African Revenue Service, and indeed there did used to be one which was removed a few years ago. There are also possibilities of having both scenarios, so that would be extremely expensive and unnecessary, but in Germany, for example, one can choose to go either through the benefit system or the tax system to attain the child benefit. Our priority was to focus on simplicity, accessibility, especially for poor people and speed, and so on balance, given that SASA already has the records of two-thirds of the children in the country already, we recommended that the universal child benefit should continue to be delivered through DSD. Rather than as a child tax rebate, there's intuitive institutional and legislative appeal in the term retaining the delivery within DSD. It has the mandate to adjust poverty through the social security system, simply involves removing the means test, and the mode of delivery is already in place with an extensive network and put them to cross South Africa to deliver the grants. Remit was to explore financing options using personal income tax as a potential source of income for financing the additional costs of the universal child benefit. And there are numerous ways in which that can be done, including increasing tax rates, restructuring bans, decreasing the minimum tax threshold, though that would be a real no-no. Reducing tax rebates, which couldn't really be considered because they already have been in two parts to contribute to the new national health insurance. Fiscal drag, ideas around hypothecated taxes, and also decisions about whether to tax this new benefit or not at the moment it is excluded from the definition of taxable income. It's a very powerful way of testing out financing options, and as in the merely report, the direct taxes and benefits are the key part of the system for achieving the redistribution that society desires. So we had a free reign to explore different options. The current tax system has six, at the time, six tax bans with primary, secondary, and tertiary rebates. And we were able to use the tax benefits model SA mod to try out various options, first of all to implement it and then to look at ways of raising the cost. The version of SA mod that we used is one that's been developed over the past decade. It's most recently been updated as part of this SA mod programme and is underpinned by the Euro mod software that has been developed by Professor Holly Sutherland and colleagues at the University of Essex. The version that we used for this study was underpinned by the National Income Dynamics study way before. The policies that we simulate within SA mods are those different benefits and also personal income tax. With the NITS version, we don't simulate the indirect taxes that we do with the other version that uses the living conditions service. So simulating the current system and with a situation of full take-up, we estimated that it would cost 72 billion rounds a year to pay the child-related grants, child support grants, foster child grants and care dependency grants. This is 12 billion more than was assigned in DSD's budget and is important because when thinking about universalising the grant, if it was going to all those who were eligible, then already four out of five of the children would be receiving it. In terms of identifying the amount of cost that you need to recoup, you need to identify and be explicit about whether you're looking at the additional cost if there was already full take-up or the cost from the current status quo of non-full take-up. Based on our estimates, if you move from full take-up to a universal benefit, it costs a further 15 billion or a little less if it's made taxable. But moving from non-full take-up, so the current status quo up to full take-up of the child benefits is 27 billion or a little bit less than that if it's taxed. We explored many different possible ways of quickly recouping the money through the personal income tax schedule. The examples which you can see in the accompanying paper of some less progressive approaches that sort of attack each of the bands, through the middle ways that squeeze the higher end of the distribution harder. And then a more progressive place where we add an additional tax bank to pull the higher end, which in fact will subsequently done by the Minister of Finance. Increasing the tax rates in a different way and making the child benefit taxable. Posts of different possible funding scenarios and the great one shaded in grey are all raised enough to finance the extra costs of the universal child benefit starting from a point of full take-up. But the key challenge is that none of those scenarios would finance it in full starting with the current position of partial take-up. And so we resorted to using fiscal drag as a way of bringing in large amounts of revenue going quickly. A quick portion we note on the tickering with band 1, these are the group who identified in our data set that 57% of tax payers fall unsquarelly and solely within the bottom tax bank. And so any adjustment to that first tax add was disproportionately impact on the low income tax care that would be necessary for us. So we tried to avoid that going forward. In order to implement fiscal drag, we created a hypothetical system for 2017. Incleded everything by 5%, which meant we had to raise even more revenue and 19 billion from a position of full take-up or 31 billion from the partial take-up. And it's the string of different scenarios that's really implemented. So for a short poll, the final one is a year of fiscal drag. With salary, another income is inflated at 5%. Tax specials increased by only 2%. Personal rebates increased by only 1%. Incleding the new child benefit of taxable income, increasing the tax rate by 2% at points of band 6, introducing a new band 7, and the billion basication to 5% tax per asset in the size of the billion and cash flow. So it's an example of the flexibility of the model really to explore a host of different possibilities. And this ruthless set of policies generated impact that they needed in every one year. This chart shows two different systems. The dark green colour is child related benefits as a percentage of disposable income in 2016 with partial take-up. And the light green is the situation with the universal child benefit and the implementation of our heavy duty fiscal drag. So it changes the picture, but not as much as one might imagine, compared with the policy that we did. We are confronted with the stark fact that we only need to look at the first column here, that poverty, whilst it reduces from a position of partial take-up to full take-up, there's no additional advantage in terms of poverty alleviation of implementing a universal child benefit and the famous family relation to child poverty results. But the key fact is that that's an artificial comparison because full take-up, one of the main achievements of achieving full take-up is actually the ones taxable income. So just to conclude, and thank you for bearing with me, there are many ways in which income can be raised to fund and finance the universal child benefit. If it is to be a personal income tax, the financing mechanisms would need to be clearly articulated and justified using fiscal finance. There needs to be extensive consultation. We could retain the status of delivery organisation as they are already well connected with the population that would continue to have to push for registration at birth. And there are a number of options that are potentially available that could accomplish it and would depend enormously on when it starts and how it's going down.