 It's a great pleasure for the UNI wider staff myself included to wish you all a warm welcome to this July 2017 wider development conference entitled Public Economics for Development. Time to turn to the opening keynote by Professor McKean, who as I've already indicated will share with us his thinking on tax administration. Professor Keane is Deputy Director of the IMS Fiscal Affairs Department and before joining the IMF he was Professor of Economics at the University of Essex and visiting professor at the Kyoto University. He was awarded the Moscow Prize in 2010 and is an honorary President of the International Institute of Public Finance. Professor Keane has led technical assistance missions to well over 30 countries and is co-author of books on the modern BAT, the taxation of petroleum and minerals and changing customs. McKeane is a true expert and in addition a great colleague. It is a privilege and an honor to invite Professor Keane to take the floor. Please, thank you. Thank you very much, Finth and it's a very great pleasure and honor to be here with you. So the context then of what I'm going to be talking about is this wider rejuvenation of public economics for development. Actually it occurs to me as I was listening that in fact a lot of my presentation doesn't particularly distinguish between developing countries and others because I think at the level a lot of the methodological points are going to make are they're really equally applicable to all countries. Clearly the application would vary but I think there's a lot of commonality in the themes and methodological issues that I get to be touching on. So of course we know that we have some key characteristics of the work of the kind of revival, rejuvenation of work on public economics in the development context. We have as Finn mentions this renewed empirical focus with careful attention usually to issues of endogeneity, maybe not quite as much modesty on external validity as one might like nonetheless, clear methodological improvements, use of new data sets including but of course not only large administrative data sets and at the end I'm going to mention a couple of new data sets that I've got I think maybe of interest to researchers and of course there's an increased focus on issues of practicality, issues of administration and I suspect where Finn was thinking about was wondering about research topics I think this is going to become an increasingly important area really because of technological change, technological change it's not my topic today but it's going to change not only how we do what we now do but the things we can do the things we can do can potentially be extremely different in terms of policy design from what they are now we've really hardly begun to think about that I'd say in terms of tax policy we haven't really thought about that at all we focus on some very narrow issues we haven't embraced some of the wide issues I'll say a little bit about one aspect of technological change digitization as I go through the talk so there's clearly again as Finn says a large and still very rich agenda in fact I think there's still a lot more to do in terms of I'd say better integrating theory evidence of practicality certainly on the tax side and perhaps you made it take a little bit of an undertone in my remarks that actually this the rejuvenation has really delivered as much yet as one might hope and that there's scope to do I think a lot better job in terms of this integration but let me then that's the wider context and let me then say a little bit about to offer you some thoughts on tax administration issues in that in that setting so I'm going to start by reflecting a little bit on the state research on tax administration records and policy focus a little bit on what I'm going to call tax gaps and how one might think about integrating those rather better in thinking about tax policy in tax systems this is really an example of how one can I think actually economists can actually help to bring insight into things that administrators worry about then I want to say a little bit about the rather interesting idea of optimal tax administration and then offer some some final remarks including if I have time a little bit on a couple of these data things so what about research on tax administration clearly there's been a bit of an explosion recently before that the Lyftian tax administration was basically boring and basically what the I hope people here audience weren't actually engaged in research at that time but it was not it was not the the most exciting field which are probably now certainly is amongst them so there's a lot of work for example on measuring administration costs administration costs I mean the costs that the tax administration incurs in implementing the tax system and measuring to compliance costs that is the costs that taxpayers incurred in complying with their obligations or perhaps evading their obligations or planning around their obligations so a lot of work on measuring these two things not really much in not much interest in well why should we care about these things so what there was no real answer to the question so what question but a lot of work done on this not as straightforward as it might sound it's actually technically quite difficult conceptually and technically quite difficult but nonetheless it wasn't really clear why we're why we're doing it and on the theory side there was a lot of work built around some of you may know the Allingham Sanmo model but a kind of a standard model of tax evasion which really didn't explain a lot of things we actually observed so there was a lot of kind of theoretical agonizing over how to make models look a bit more like what we observed by going through you know non-expected utility all this kind of stuff to try and make the models match some basic stylized facts rather better so that was what the research was doing wasn't frankly terribly exciting wasn't the kind of area you would necessarily push a bright graduate students to get into notably there was really no established framework by which you could say well what is the optimal level what is the appropriate level of this administrative intervention how do we assess for example the creation of a large taxpayer unit introduction of a new third party withholding nudges of various kinds what is the actual framework within we within which we can evaluate these moreover how does that framework relate to our framework for policy because on the policy side as I'll say in more detail we do have we've had for many years a very well-established framework for thinking about tax policy interventions we really didn't have anything like that on the administrative side by now you'll have guests and I'm going to be giving you a such a framework later on but nonetheless this is the state of state of play until recently so the recently then we've had this explosion of empirical work and there are many very technically excellent papers out there that use experiments natural or otherwise to address various issues of tax compliance that is extent to which the way in which taxpayers meet or not their obligations just to give some of the my kind of somewhat random but these are amongst my favorite examples as an excellent really good paper by Pomerance which looks at compliance in VAT change and this is a this is a very in it in genius study it's kind of standard thing in that what she did was she sent letters to various VAT registered firms in Chile telling them that the chain basically saying your chance of audit has gone up fairly standard thing but the interesting twist was she looked not at what happened to their compliance but what happened to the compliance of the people they bought from and the people they sold to because that's casting light on how this that chain how the chain of VAT obligations works out and I'll say a bit more about that in a moment another nice study looked at the effect of lotteries in Sao Paulo in Brazil so lotteries that these are these arrangements you may be familiar with where essentially retailers are encouraged to give you an invoice VAT or retailers are encouraged to give you an invoice exposing themselves to more close the monitoring and the incentive them to do that comes from you because the invoice becomes basically a lottery ticket so clients have an incentive to request an invoice because it gives them a lottery ticket so it's a very careful experiment of the natural experiment to try and look at the impact of lotteries on tax compliance and then of course there's a whole set of papers that look at nudges the taxpayers of various kinds these are these kind of letters saying well you know your neighbors have paid their tax wider view or this is what your tax revenues used for and looking at the impact of on compliance of these kind of nudges so there's a lot of a lot of very careful high quality work going on but of course the question is well what of tax administrators actually learned about this so where I work at the fund I have about 40 colleagues who are tax administration people this is what they do that's what the careers has been has been advising countries on tax administration what do they actually learn from all this well I think the pomerance study does teach something because one of the things that comes out of the pomerance study which was the bat chain one that I mentioned a moment ago because there's always this question that for those of you know how the VAT works you know why I charge you you take a credit but you charge your customers so on and so on there's always this issue of well if you want to improve compliance in a VAT do you put your enforcement measures at the start of the chain say on imports and big businesses or do you put it at the end on the smaller retails so do you want do you want the compliance to kind of ripple forward or do you want it to ripple back and one of the implications that comes out of her work is a very count to me at least counterintuitive one which says actually start at the end you start with the retailers and try to work back in terms of compliance rather than focusing on importers and big businesses at the beginning of the check I'm not sure I quite believe the result but you can see that's a really useful result for tax administrators that's actually very the kind of thing that really can help perform administrative interventions a number of other studies I have mentioned them I've done things like document the importance of withholding schemes the idea that you know it's important to get hold of the money for the tax authorities get hold of the money before people can spend it and they're very eminent papers on this this is this I would see is rather different if I go to my tax administration colleagues and say do you know the tax withholding and also third-party reporting is really important they will look at me like I'm an idiot that's what they've been doing at least since in the UK at least since the late 1690s the land tax was the first why she wasn't the first since discovered the first withholding in the UK was 1245 so tax administrators don't really learn much means maybe a very nice very well documented study but it's not actually terribly helpful on the lot trees you can and the nudges there is a question well you may be able to show some some potentially some revenue gain from these things but is this really first order importance when you go to her when we think about the revenue mobilization challenges we think about for achieving the sustainable development goals do we really think lot trees on nudges are going to give us enough money frankly and the answer I suspect is actually not that the actual numbers that come out of these studies are really quite small these are not the things that administrators really need to get to grips with you know if they're ready to achieve these revenue mobilization objectives a nice research papers very nicely done but in terms of helping achieve something real in developing comes in elsewhere we still have a lot to go so that's why I think there's still quite a way to go in terms of trying to make theory more useful by providing some practical frameworks in which we can think about administration and also integrate I think about administration with some ideas on policy so what I'm going to do in the rest of the talk is to try and talk through some of those some of those ideas and really taking as my starting point something the idea that I mentioned a moment ago the idea of a tax gap tax gap wanting to explain in more detail what I mean but this is the idea that there is a gap between what the tax law says the government should be collecting and what it is collecting and more precisely I'm going to call that the compliance gap so tax administrators actually care quite a lot about the compliance gap and they've actually started measuring it quite quite carefully in recent years and I'll say a bit more about that too so I want to take this idea of a compliance gap and think about well what is the significance of the of the tax gap when we think about why the tax system reform administrative reform policy reform how do we as economists make sense of this idea that seems to have some appeal to administration people and then I'm going to take up from there the question of well what is the optimal tax gap probably the optimal tax gap is not zero it's probably not optimal to go to world in which everybody always complies with the tax obligations 100% because that's probably going to require quite heavy administrative costs compliance costs it may actually have some adverse effects on behavior this thought to add to the distortions in the tax system and so on so the optimal compliance gap is probably not zero so well what is it how do we think about what an optimal compliance gap is when you think about that question that takes you into a wider question which is why I got to focus on which is well how do you think about optimal tax administration more generally how do you think about what an optimal administrative intervention is and as a quarry of that how do you go about assessing administrative interventions what are the kind of things you need to know in order to decide whether some particular administrative reform is good or bad so those are the kind of issues I want to to focus on so I'm going to say first a little bit about flesh out a little bit more this idea of a tax gap and why it might matter and how these it's a way in which we can think about integrating policy in administration in ways that can be quite useful for first order importance problems of revenue mobilization so I'm going to think about this by in the context of thinking about the VAT so of course for many countries not least developing countries VAT is an important source of revenue making the VAT work properly is going to be a key to achieving any kind of revenue mobilization goal so let's think about VAT performance naturally people like us the fund we get very interested in what happens to VAT revenue so let's think about how we can understand developments in VAT revenue so here I'm simply writing VAT revenue V as a percent of GDP why so the left is that revenue percent of GDP and then simply by multiplying and dividing various stuff you can write the VAT to that revenue ratio in this as a product of three things one is Tau S which is just the standard rate of the most common rate of the value added tax skipping to the end C over Y is just aggregate consumption in GDP and then the thing in the middle of this EC is a concept called C efficiency which is going to be important in in what I discussed this is a common measure that people use to think about really how effective a value added tax is and what it is as you'll see in the definition there C efficiency is basically revenue from the VAT divided by the revenue you would get if you applied the standard rate of VAT to all consumption so for example if you had a VAT that tax all consumption at the same rate doesn't matter what the rate is and there was no avoidance or evasion then C efficiency would be one so it's a you can see it's going to be a kind of measure of departures from a uniform VAT on a broad base I know for this audience there are going to be issues about desirability of a VAT the uniform rate on a broad base but maybe for this analytical purpose we can put that aside I'm quite happy to discuss that in itself so once you've done that you can then think about well okay so what how do we explain what happened in VAT revenue so here for the further some years ago now this basically the dots show what happened to changes in VAT revenue by income group over this particular period so what was driving that well your first guess might be well maybe they've been playing around with the standard rates well actually no the standard rate doesn't explain much of what's actually happened to VAT revenue perhaps surprising what about changes in consumption relative to GDP well no that doesn't really help much either in fact it tends to go the other way in a number of countries so what really explains it is the thing in yellow except in high-income countries particularly in the non-high-income countries is changes in C efficiency so if you want to understand what's happening to VAT revenue C efficiency is the thing you really need to understand standard rates not a big story the real story is all to do with C efficiency so what drives C efficiency well you can do another kind of decomposition and write C efficiency as again this is just multiplying and dividing by various things to get to get it to look nice so what you end up with is C efficiency if you look to the right of the equation is the product of this 1 minus P and 1 minus what's that gamma right capital gamma so gamma is is the compliance gap so gamma for example is the difference between the revenue you should be collecting if the tax system whatever it is uniform rates or not whatever it is that's the revenue difference in the rev you should get if the thing was properly enforced ratio of that to the revenue sorry sorry the other way right so ratio of the revenue actually got to the revenue collective the thing will properly enforce so that's gamma it's a compliance gap the other term P I'm going to call a policy gap subject of qualification I mentioned a moment ago this is basically the difference between the revenue you've got and the revenue you would get if you had a uniform rate on all consumption so P is really picking up the effects of rate differentiation exemptions and so on gamma is picking up the effects of non-compliance so what's quite neat about that well before I spell out that more just to remind you then the compliance gap is this excess of essentially basically telling you how much you're not collecting you should be collecting and as I mentioned this is something that tax administrators have got really interested in measuring recently UK has been doing this for several years it's now done regularly for the European Union one of the things that we do at the fund is we have this thing called the RA gap project which basically goes to helps developing countries in particular to measure the VAT gap we're doing work on the corporate taxes as well but basically getting a numbers for what's going on with the compliance gap as we previous slide implies really what you want is a major compliance gap and combine it with a measure of the policy gap one of the neat things by the way of course is that if you know see efficiency see efficiency is usually easy to get so if you get one of the other two for example if you know the compliance gap then you can figure out what the policy gap is so you don't need to measure all three you can infer one from any two of them so for example I mentioned this is the kind of thing we've been doing in a number of countries don't need to go through the details but on the right Uganda the blue line is an estimated compliance gap the red line is an estimated policy gap South Africa ignore the red line the line at the top is the policy gap and the line at the bottom is the compliance gap so why is this interesting this is really telling you that if you're looking at Uganda the real issue is compliance the policy gaps actually not that bad the real issue is compliance in South Africa on the other hand it's the other way around but in South Africa the real issue is the policy gap it's not so much the compliance gap in terms of thinking about the overall effectiveness of the VAT so this kind of analysis I think bringing together a bit of economics with a bit of administration helps you identify what are the priorities reform so Uganda you can quantify all these things of course you can say well actually in Uganda if you were say to a half the compliance gap that would get you three points of GDP which is a big number so this is kind of when I when you're thinking about making real first-order effects on on on revenue mobilization that's that's a big number and you can go further in this kind of analysis to try and figure out where the compliance gaps are arising so the methodology of this of this gap analysis which I won't go into now does let you identify particular sectors in which the compliance gap arises so on the on the left basically the numbers correspond to different sectors and you can see where the biggest gaps between potential and actual collection are by sector and that then informs that potentially informs your administrative intervention so what I like about this of course and well just to give another example if you think about the the UK which as I mentioned has been doing this for a while you track what's happened to the compliance gap over time and you can think about well how do we explain it well a couple of things come to mind we know 2005 was a big period for carousel fraud and we know that the crisis years compliance always worsens in economic crisis after economic crisis so again this is the kind of things you can try and figure out what's really going on with compliance clearly there's a bunch of modeling potentially to be done on compliance behavior even at an aggregate level so what I like about this is a kind of framework which brings together some economics and administration in thinking about essentially the balance between an appropriate design of administrative and policy interventions figuring out where the kind of bang for the buck is all in a kind of coherent framework that lets you think both about policy and administration but of course there are come some problems with with this and the one question that comes to mind which is a question that I always irritate my tax administration colleagues with because pretty much whatever they figure out the compliance gap is they tend to think it's too big and that's a kind of a natural thing to do but as I mentioned collusion the compliance gap is not costly there are going to be resources there are going to be distortion involved so what actually is the optimal compliance gap how do we know whether the compliance gap is too large or too small more generally as I was mentioning before how do we how do we go about characterizing optimal administrative interventions and that brings me kind of to the second main part of what I want to talk about which is this idea of trying to get towards some notion of optimal tax administration and more precisely sort of there are there are kind of three questions that come to mind that one might want to ask so one is well how should we assess administrative interventions so as I said if someone is I don't know if you're having a kind of functional reorganization of the tax administration if you're thinking of opening up medium some medium taxpayer offices if you're thinking of changing the remuneration policy for tax administrators how would you actually go about assessing whether that policy potentially is or in practice was a good thing or not and you think about and as I said that we don't really have a framework for this whereas we do have a very well defined general framework for thinking about thinking about optimal policy particularly thinking about optimal tax rates many of you will know for example they were in particular of the concept of the elasticity of taxable income so the elasticity of taxable income is basically the responsiveness of the tax base to the tax rate that may reflect both behavioral changes and avoidance evasion decisions and there's this pretty seminal result going back to Feldstein that says really that's a sufficient statistic for thinking about the welfare impact of change in the tax rate the clever thing is you don't need to know what the responsiveness is in terms of work effort relative or compared to the responsiveness in terms of evasion or avoidance all you need to know is this summary statistic of how change in the tax rate affects taxable income and the higher that tax rate is the lower the optimal tax rate tends to be because that means more distortions behavioral responses at the margin so when policy we have a very well-defined sufficient statistic for thinking about tax rate interventions the elasticity of taxable income well what about administration if we think of also administration is something governments can vary just as they vary tax rates is there also a sufficient statistic on that side of things is there some single number that we should be thinking about when we do evaluation work on administrative reforms some sufficient statistic for making normative judgments and evaluating policy so that's one question the second question is the one I've already mentioned which is kind of a corollary of the first one so once we have a theory of what optimal administrative interventions are can we then figure out well what how do we characterize an optimal compliance gap in that setting so that's a kind of a special case of the more general question of assessing administrative interventions so how can we know when a compliance gap is too big or too small and the third is kind of kind of an amazingly basic question suppose your your government you need to raise some revenue you need to raise a whole bunch of revenue to meet the SDGs or whatever well you basically have two ways you can go about it you can think about raising tax rates or doing or it's or policy measures more broadly and or you can think about strengthening enforcement or more generally by undertaking administrative reform so this is a very basic choice that really all policymakers face yet and the question is well what should it be what is the better route to go is it better to to put your resource into closing compliance gaps or into or to raise additional resources by playing around on the policy side very basic question yet we don't have any as far as I know no one has ever given a particularly satisfactory answer to that question or even a framework within which you can think about it so that's another kind of challenge in all this is to think about well how do we know which is better which is better to raise revenue visit administration or policy measures so those are three questions let me kind of set up a little setting a framework to kind of address them so this is based on some work with Joel Slam one so I'm not going to go through the algebra in any detail for those of you who like algebra this will all be very obvious for those you don't like algebra me trying to explain it won't actually help but basically so here's a kind of very standard framework let's just think about the representative individual we can talk about extensions later so what does this individual do well they earn some income they will wage right W they supply some labor L they pay some tax rate with big T tax depends on their wage income minus some amount E that they conceal from the tax authorities so it could be evasion could be avoidance for the moment I don't really care so tax is kind of your is on your declared income on the other hand there are some costs see of either complying or not complying with the tax system excuse me those costs depend on how much you avoid or conceal E but also on this alpha thing this alpha thing is going to be important that some administrative intervention so this is some administrative intervention that works by making clients more or less costly and then there's also the end V of R some valuation attached to the public expenditure that the revenue finances so the revenue little R is simply the tax revenue less some administration costs big A of alpha so those are the administration costs what it actually cost the tax authorities see on the other hand is more like compliance cost cost incurred by the taxpayer so what's going on is the taxpayer chooses labor and how much to avoid or evade the choose L and E the government chooses the tax rate T and is administrative intervention alpha so this is a very this is for those you know it this is basically the same as going to Chetty Sy's framework but with an administrative intervention thrown in on top so not surprisingly when you think about the choice of T there's nothing new here this gives you this the standard result standard and familiar result that the elasticity of taxable income is a sufficient statistic for thinking about the optimal tax rate so the higher that elasticity the lower is the optimal tax rate and as many of you all know we now have a huge empirical literature that tries to estimate this elasticity of taxable income in various circumstances well so nothing new then but the framework does help us to answer those three questions that I mentioned firstly when you think about administrative intervention you get a very simple rule for the choice of alpha which is this there's this thing on the left called phi which is think of it for the moment as a ratio of administration and compliance cost for revenue the interesting thing there is that's pretty much what this boring literature I started off with the outset has focused on and I'll come back to that in a moment so on the left we have cost terms boring stuff but there's been a lot of work estimating them more interesting is this E thing so E big E means an elasticity so this is the elasticity of Z with respect to alpha what is Z Z is taxable income so it's the WL minus E it's the kind of income the tax authorities get to see so this is just the elasticity of taxable income with respect to the enforcement parameter so what's going on here why is this at all interesting because now this elasticity which more generally is an enforcement elasticity of tax revenue is a sufficient statistic but think about administrative interventions so all you need to know on the administration side an exact parallel and in some respects actually more general than the elasticity of taxable income on the policy side is this elasticity of tax revenue it's a basic responsiveness of tax revenue with respect to your administrative intervention so that's the answer to the first question there is a sufficient statistic and it's this enforcement elasticity of revenue and so I won't go through it here time doesn't allow but you can think of you know if you give me numbers for the valuation of public goods you give me numbers for administration costs and compliance costs I wave my hands a little bit but then I can basically tell you whether more or less enforcement is a good idea so basically this gives you a simple rule of thumb for figuring out whether an intervention is good or bad whether you should do more of it or less of it there is no one's ready though as I said there's a huge list on elasticity of taxable income not really anything directly on this enforcement elasticity but you can back out some estimates kind of more or less by serendipity by chance from some things that are out there I won't go through it but so the compliance get measures I mentioned you can back something out the closest is there has there's been some work done at the IRS in the US from which you can infer pretty careful estimates that turn out to be of this enforcement elasticity but I won't go through these numbers here the point is that this is clearly a kind of you know we have an industry on the policy side we're really stuck for numbers on the administration side just to say a bit more on that cost term there's fee the one that was on the left here this fee term here so as I mentioned that's more or less a ratio of compliance costs plus administration costs to revenue which is what that older literature focused on with a couple of wrinkles that I won't go into one is as is well known you have to wait for social evaluation purposes you have to wait administration and compliance costs slightly differently and compliance costs are a little bit cheaper because they're essentially paid directly by the taxpayer as a lump sum not in the form of a distorted taxes and of course not surprisingly much of the literature has focused on averages of these administration compliance costs what you need for this is marginal costs but nonetheless we can think of getting some numbers for that so just to give an example there are there's people this is not hard to apply and people are beginning to apply there's a paper by Meiselman who essentially applies this framework to a kind of a fairly standard thing of sending letters to potential or suspected non-filers in the city of Detroit and basically you can go through the kind of this is just a discrete analog to the framework I was just describing so you can see for example it depends on this valuation of public goods this delta z the change in the taxable income which is where this enforcement elasticity term change the administration costs and compliance so he goes through the exercise of actually putting numbers on all these things and actually turns out that you know this is the kind of fairly standard experiment it turns out of course the letter does leads more compliance but isn't a good policy does it make sense as a policy well it turns out the answer is no when you do this properly I guess we're often tend to think of these netizens generally being useful this one clearly is not is not worth it and importantly just as a kind of an aside the reason it doesn't work is because of the compliance costs and these kind of which turn out to be quite large um people expect that you induce people to fill in reforms fill in forms that take them a lot of money that take them a lot of time we don't actually give you much money the aside there is is simply to note the compliance costs matter a lot in all this people often do these kind of exercises and maybe say a little bit about administrative costs don't think at all about compliance costs but compliance costs can well be the dominant consideration in lots of these exercises and again I think when we think about areas where we need to do more it's clear that we tend to ignore compliance costs much too much question two and I'll be fairly fairly brief you won't be surprised that you can you can basically rearrange that optimality condition to get an expression for the optimal compliance gap it depends on a slightly different elasticity which is not surprising because the compliance gap is not a welfare measure so that elasticity is not sufficient because compliance gap itself is not a welfare indicator but you get a very simple rule that lets you figure out whether compliance gap is too big or too small or not so again at the bottom I just give you some illustrative numbers if you tell me you know if you tell me if the compliance what the compliance gap is then I can figure out what that elasticity has to be in order for it to be worth undertaking a reduction of that of that compliance gap third third question was this one about the administration versus policy and something that drops out very neatly from this framework is really just a very simple condition that tells you whether you want to raise an additional dollar through by raising the tax rate or by administration it's not terribly surprising what turns out to matter these two elasticities tend to matter as well as compliance and administration costs there are some wrinkles but it's not terribly surprising but what's important I think is the thing I'm not showing you which is really just an explicit expression the less you just put numbers in and figure out well this is what we should do we should go with administration or we should go policy once you think about things in this way once you think about administration and policy integrating this way it does actually also raise a number of other questions which I think people haven't actually thought about much take another question so I've been going on about administration integrating with policy well or administration and policy or more precisely administrative interventions and tax rates are they strategic substitutes or are they strategic conflicts that is if for some reason your administrative your level of your administrative spending is fixed too low does that mean you're going to set higher tax rates or lower tax rates than you otherwise would intuitively it's not completely obvious and it turns out that that matters quite a lot just to give you an example of why it matters suppose we think about technology and suppose we think that really what one of the effects of technology is that evasion is going to be make evasion is going to become harder it's going to be easier for tax authorities to detect evasion it's not clear that's true but let's suppose that's the story well you can then ask well actually okay so evasion is harder does that mean we think tax rates are going to be lower because it's easier for us to raise revenue or do we think tax rates are going to be higher because now there are less distorting effects from the tax system well that turns out to depend on whether these things are strategic components or strategic substitutes I won't go through the diagram on the left but basically this is just drawing you know on the bottom is alpha on the left is t so the t alpha shows what is the optimal tax rate conditional on on a level of particular level of administrative intervention similar alpha t play around with these curves and you can see basically if technology makes detecting evasion easier that means the alpha t curve or the t alpha curve shifts to the right and you can see that in this case the tax rate is going to the optimal tax rate is going to fall but not too hard to believe but if these curves were differently shaped it would go the other way tax rates would go up as a consequence of digitization so this is just an example of a kind of a fairly I don't know I've never seen anyone even think about that question at these things components or substitutes and it matters actually quite a lot for things that we're going to be facing in the future there's a whole load of possible extensions I won't go through it turns out to be fairly robust framework easy to have many administrative instruments to think about multiple households all these kind of things straightforward to do gets a little bit more complicated when you allow for slightly more general forms of concealment costs compliance costs but basically you can extend this all fairly straightforwardly so with that then let me let me just let me just conclude so I hope having sort of persuaded you that or tried to persuade you that there is still a lot more scope for addressing first order policy issues by better integrating theory evidence and consideration of practicalities some of you will have seen this juxtaposition of quotes that I've used before I confess so several years ago a slumber of Yadzaki said well basically the theory on tax administration is pretty well developed who is trying to put the rest to rest the claims of all this stuff is understudied on the other hand there's John Haseldine who's a tax administration guy who basically says that this John tax administration is terrible I confess my inclination has been rather more with the second quote than with the first but I think things are getting better with that I take one minute to try and so this is just to mention or think about research a couple of data things that we're involved in that the fund that I hope will soon provide sources of useful data and think about tax administration so one is something called the International Survey of Revenue Administrations which we do now jointly with the OECD IOTA see it which is basically trying to collect comparative or comparable administration data from countries around the world at all levels of development so for example things like on-time filing rates all these kind of magnitudes that tax administrators look to when they assess their performance so those data should be becoming available and there's also something called the Tax Administration Diagnostic Tool which we're actually pretty excited with at the fund which is called TADDAT which basically helps we've done I think now 30 or 30 countries also helps countries assess in a very detailed way aspects of their tax revenues administration's performance you get scores from A to D some of these are public but we hope to make the data available more generally so for example that Zambia this is the kind of breakdown of scores you get this is across all dimensions of the tax administration's performance from registration all the way through to appeals and governance and so on and just on the right there's a kind of little chart that indicates the sort of variation in results that we get across countries so I'm hoping that we're hoping that those two will will start to become regular input into thinking about tax administration issues how it maybe relates to other aspects of policy but I think with that note there's some references again with that I will conclude and thank you for your attention thank you very much I hope you will join me in thanking Mick I mean as researchers of course what Mick has done is being stressing what we do not know and of course that's extremely welcome both there's an opening to this conference but also in general for us to try to get our focus right and start addressing the things we do not know so thank you very much Mick