Taxation, Inequality, and Development

videoType

dynamic-media

videoDmUrl

keyFrameImage

timestamp

00:00 Um

00:01 Welcome everybody,

00:02 um,

00:03 to this,

00:03 uh,

00:04 policy research talk.

00:06 As you know,

00:06 these talks provide us an opportunity to present work

00:09 coming out of the World Bank's research department.

00:12 With the goal of sharing the findings

00:14 with colleagues inside and outside of our department as

00:17 well as with people outside the World Bank.

00:19 Now we have a platform for doing that easily.

00:22 Um,

00:23 I'd like to welcome our online audience,

00:24 both on Webex as well as on YouTube.

00:28 Today's talk is going to focus on tax policy

00:31 and its impact on efficiency and redistribution.

00:35 Low income countries only raise a fraction of the tax revenue

00:38 that rich countries collect.

00:41 Which limits our ability to provide their ability to

00:43 provide public goods and invest in human capital,

00:46 redistribute income

00:47 and insure against shocks.

00:49 So we're going to start with my colleague,

00:51 Pierre Bakas,

00:51 who will discuss how the difference in economic structure

00:55 and state capacity can lead to markedly different tax policies across countries.

01:00 He's going to illustrate some implications of these differences for growth

01:04 and inequalities using results from his recent work,

01:07 and he'll conclude by discussing some of the challenges

01:09 to make tax systems more efficient and redistributed.

01:13 Pierre is an economist in the macro and

01:14 growth team of the Development Research Group.

01:18 We're also grateful to have today,

01:20 um,

01:20 Marcelo Estevao

01:23 as a discussion today.

01:24 Marcelo is the global director of the World Bank Group's macroeconomics Trade

01:29 and Investment Global Practice,

01:31 MTI.

01:32 He leads a large team of country economists,

01:34 macroeconomists,

01:35 and fiscal policy,

01:36 debt and macro modeling experts.

01:39 He is responsible for overseeing the delivery of

01:41 the global analytical work on fiscal policy,

01:44 debt policy,

01:45 and economics of climate change.

01:47 For coordinating the strategic direction of MTI

01:50 and implementing it,

01:52 for helping to shape

01:53 and oversee MTI's country,

01:55 regional and regional programs,

01:57 and for mobilizing staff to work more effectively

02:00 across the EFI and other global practices.

02:05 So Marcella,

02:05 I hope I got your last name right,

02:07 because

02:08 on my,

02:09 on my notes,

02:09 it's much longer,

02:10 on the screen,

02:11 it's much shorter,

02:12 so tell me if I got it wrong.

02:16 You did well.

02:17 You did well.

02:18 I don't want to impose the

02:20 You know,

02:21 Brazilian slash Portuguese last names to anyone.

02:25 Thank you.

02:27 Um,

02:27 so,

02:28 OK,

02:28 so,

02:28 um,

02:29 with that,

02:29 I'll turn it over to Pierre,

02:30 who will talk for about 45 minutes,

02:32 after which we'll hear from Marcelo for about 10 to 15 minutes.

02:36 And then we'll conclude with the Q&A session,

02:39 um,

02:40 from questions from the audience.

02:42 If you have questions,

02:43 please submit them in the chat function,

02:45 either in Webex or YouTube because we'll be monitoring both of those chats.

02:49 Uh,

02:50 so with that over to you,

02:51 Pierre.

02:53 Thank you,

02:53 Dean,

02:54 let me share my screen.

03:00 OK.

03:01 Thank you,

03:02 Dion for the introduction and for inviting me to present my work.

03:05 It's a pleasure to be here and so today I'll tell

03:08 you about my field of focus which is on taxes,

03:10 inequality and development

03:12 and I'm very excited to do so as you'll see it's an active field

03:15 of research which I hope I will convince you has some practical and useful

03:19 uh lessons for the work we do at the bank in terms of policy making.

03:24 The talk will follow the following structure.

03:25 I'm gonna give you a bit of an overview of tax policy,

03:28 especially how it changes across development in the 1st 1012 minutes,

03:31 and then I'm gonna enter into more depth into two of my recent papers,

03:34 one which with one which focuses on the role of taxes

03:38 and its impact on efficiency and the other one on equity.

03:43 So let me start with this overview of tax systems

03:45 and development and I'm gonna take first a very big

03:48 bird eye view of taxes in particular what are

03:51 the key functions of the tax and transfer system.

03:54 One way to categorize them are those four maybe in order of kind of

03:58 maybe not importance but the order in which they follow with development.

04:02 The first one of course and a lot of our work at the bank is

04:05 this,

04:05 is to invest in public goods,

04:07 infrastructure,

04:08 health,

04:08 and education.

04:10 The second one,

04:11 which is the 2nd pillar of the bank,

04:13 is to help reduce market inequality,

04:15 and this can be done in two ways through taxes,

04:18 in particular,

04:18 to impose progressive taxes which increases

04:22 with household incomes,

04:24 and the second way is through transfers,

04:26 in particular the well targeted and reach the poorest households

04:29 in the country.

04:31 The third role is to provide insurance,

04:33 in particular in the context where

04:35 uh market insurance

04:37 might unravel,

04:38 and so that's the case for example,

04:39 for old age,

04:40 for health,

04:41 for unemployment,

04:42 and as we've seen more recently,

04:43 there's also a big demand for

04:45 insurance from the state

04:46 in the context of systemic crisis like the current pandemic.

04:50 And the 4th goal that we're getting increasingly involved in

04:53 the climate emergency is to correct for externalities such that,

04:57 reflect the true social cost,

04:59 uh,

05:00 of goods.

05:01 Now against the backdrop.

05:04 A fact maybe you have seen or relation you have seen is how does tax

05:07 revenue measured as a percentage of a country's

05:09 GDP correlate with log GDP per capita.

05:14 And you can see that there's this strong positive correlation

05:17 that is richest

05:18 countries in the world,

05:19 countries in the OECD collect on average 35% of the GDP in taxes.

05:24 Well if you look at the bottom

05:25 corner of this graph,

05:26 you see that poorest countries in the world collect 10,

05:28 maybe 15% of the GDP in taxes.

05:32 There is some

05:34 extent to which this varies a little bit across regions

05:37 and so here I'm showing you the same figure but averaging

05:40 across regions

05:42 and you can see for example that the South Asia is a little bit richer,

05:45 that's sub-Saharan Africa,

05:46 there's a very low tax revenue collection in these countries,

05:49 similarly for the Middle East and North

05:51 Africa.

05:53 Those differences in tax,

05:55 total tax revenue

05:57 can be decomposed

05:58 with the tax instruments that are used in each country.

06:01 So let me start here.

06:02 So now I've put groups of countries on the x-axis starting

06:06 from the lower income countries going all the way to,

06:08 to high-income countries

06:09 and I'm showing you two types of taxes,

06:11 taxes on trade,

06:13 that's tariffs and taxes on corporations.

06:15 You can see that if you look at this relation,

06:17 well,

06:18 if anything this looks pretty flat or maybe even

06:20 a decreasing with,

06:21 with countries' income,

06:23 right?

06:23 Now,

06:23 taxes on trade and corporations often thought to be

06:26 Easy to enforce taxes.

06:27 In the case of trade,

06:29 for example,

06:29 all you have to do is observe a few big ports or in

06:32 the case of corporations is focus enforcement on some of the large,

06:35 uh,

06:35 corporations in your countries.

06:38 The next step we can add is that on consumption.

06:40 It's often a big share of total revenue and you can see that's the

06:42 case everywhere and now we start observing a little bit of a relation with,

06:46 with income though you can still see that upper middle income countries,

06:49 once you,

06:49 when you only have those taxes collect a bit more than high-income countries.

06:53 And so where does the big gap arise?

06:55 Well,

06:55 it's in those next

06:57 set of taxes that I'm adding here

06:59 which are taxes on personal income,

07:01 on payroll,

07:01 and on property and what I called other which often taxes on wealth,

07:05 inheritance,

07:06 and the likes.

07:08 And so you can see that it's those taxes,

07:10 those last three taxes that really explain the

07:12 big gap we see between poor and rich countries

07:16 and that is worrying for two reasons.

07:17 First,

07:17 it's because personal income.

07:21 One of the key instruments to redistribute.

07:22 So when we saw the 2nd role of the state to redistribute in rich countries,

07:25 this is done with the personal income tax and

07:27 increasing marginal tax rate on personal income

07:31 and the 3rd role of the state,

07:32 that of social insurance is usually achieved with revenue that comes

07:35 from payroll taxes and so you can see that the lack

07:38 of existence of,

07:39 of those taxes in poor countries

07:41 probably limits uh what the state can do.

07:45 Now before

07:46 I enter into,

07:46 into the meat of the subject there's a question

07:49 I often get is should governments try to,

07:52 to actually increase the tax revenue?

07:53 Is,

07:53 is that a good objective to have?

07:56 And I'm not going to provide a definitive answer because it probably depends on,

07:59 on each country's circumstances,

08:00 but I think in general,

08:02 given where we are currently,

08:03 the answer is on average,

08:04 yes,

08:05 and that's for a few reasons.

08:06 First,

08:07 many countries still collect today under 15% of the GDP in taxes,

08:10 which is really an insufficient level

08:13 to meet

08:14 the,

08:14 the sustainable

08:16 goals and more generally provide enough of even that basic infrastructure,

08:19 health,

08:20 and education.

08:21 Second,

08:22 the current crisis is putting a lot of

08:23 demand for the government to provide social insurance

08:26 but also it's deteriorating its fiscal position,

08:29 and so I think there's little doubt that in this,

08:31 the aftermath of this crisis,

08:33 there's gonna be

08:34 uh

08:35 a large push for governments to try to increase their tax revenue.

08:39 Now it's important to design taxes which should not impede growth.

08:43 But keeping that in

08:45 mind,

08:45 these few gross success stories that have happened with that,

08:48 the size of the state increasing quite a

08:50 bit and providing good infrastructure and insurance.

08:52 So there's a bit of a chicken and egg problem if you want in any case.

08:56 Secondly,

08:56 it's clear that in some cases this

08:59 And taxes,

09:00 you know,

09:00 can be inefficient,

09:02 but even these inefficiencies,

09:03 they have to be balanced against the fact that it's often

09:06 very high returns to investment and to redistribution in context,

09:10 uh,

09:10 with low investment and a lot of inequality.

09:13 And finally,

09:13 taxation is more

09:15 than just an instrument,

09:17 it's also a way through which cities and the state interact

09:20 and so tax can spur,

09:22 can spur accountability and better governance.

09:25 Now,

09:26 one of the key issues that taxation is constrained by what governments can observe.

09:30 Indeed,

09:30 modern tax system rely on self-reported activity which is cross-validated then

09:34 with a lot of third-party reported data from different sources.

09:38 So let me give you the typical example you're probably familiar with

09:41 which is the relation between employer and employee.

09:44 If you try to report your salaried income differently than

09:47 what it really was in a lot of rich countries,

09:50 you'd have a problem because your uh your employer would have sent the,

09:54 the same slip that it sent to you to the tax authority and automatically you,

09:57 you,

09:57 you get caught,

09:58 right?

10:00 Well,

10:00 the problems in poorer countries information is limited and

10:03 that's the case because there's a lot of self-employment,

10:05 small-scale productions,

10:07 production chains are small,

10:08 so there's not a lot of interaction between firms

10:11 and even

10:12 uh

10:13 when there are interactions there's often incomplete accounting right.

10:17 So the first big constraint is that of information.

10:20 The second one is even once you have information,

10:22 you need a given level of administrative capacity

10:25 to process this information.

10:27 Modern tax administrations use efficient databases

10:30 and risk algorithm to detect tax frauds

10:33 and have incentives to actually detect tax evasion

10:35 for tax inspectors.

10:37 Now not to pick on this,

10:39 on this tax inspector,

10:40 but you can see in this kind of messy office that this is something we observe often

10:44 when we're working in country with tax administration that

10:46 the systems and departments are not always integrated,

10:49 for example,

10:49 maybe the customs doesn't communicate with the domestic revenue agency.

10:54 There are few incentives for tax inspectors to detect frauds

10:57 and even once they detect fraud,

10:58 it's often quite hard to apply the law and actually collect

11:02 the,

11:02 the,

11:02 the taxes that are.

11:06 So

11:07 what this shows is that there's some trade-offs in tax policy design.

11:10 When we often study this in graduate school for example and study public finance,

11:15 the textbook case we see is that of a

11:17 trade-off between equity and efficiency.

11:20 Poorer countries also need to consider

11:22 that there's disinformation and capacity constraints

11:24 and then move us from the world of second best tax policy to that of 3rd best policy.

11:31 So let me provide you with two examples and then I'll

11:33 delve into the two papers that I will,

11:35 uh,

11:36 I will discuss today.

11:37 So the first one is that of consumption taxes.

11:40 So in the textbook case when you have no information capacity,

11:45 You actually do not care if you are,

11:47 if you levy a tax on value added or a final sales tax.

11:51 In practice,

11:52 once you consider these information constraints,

11:55 Value-added tax are to be preferred because they collect

11:58 taxes at each

11:59 level of production

12:01 and

12:02 this creates incentives for sellers and buyers to cross-report on each other.

12:07 Now what's interesting that for example Dina Portman showed that this is the case in

12:10 Chile and that indeed the value-added tax is to be preferred to the sales tax,

12:13 but Chile is a fairly high

12:15 capacity country

12:16 and so recently there's been some research,

12:18 for example,

12:19 by Mara Wasim in Pakistan

12:21 showing that in countries with weak administrative capacity,

12:24 VAT.

12:26 And there's a lot of these bogus firms that get created that

12:28 create fake invoices and then it becomes a nightmare for the tax administration

12:33 to try to levy anything.

12:35 And that's interesting because actually if you see developments in the last few

12:38 years it's the first time ever that a few countries have returned,

12:40 have left the value-added tax and gone back to a sales tax and

12:43 so it shows the importance of thinking of both information capacity constraint.

12:47 Another example

12:50 With work by Michael Best and co-authors and some of my work in Costa Rica

12:54 is that

12:55 of tax corporate

12:57 profits.

12:58 So the efficient thing to do is to tax profits that

13:00 is revenue minus all deductible costs that have entered into production.

13:05 In practice,

13:06 costs are quite easy to fake and are very responsive to the tax rate

13:10 and so these two papers show that a little efficiency cost.

13:14 In terms of revenue,

13:15 it's preferable to tax small and medium enterprises on the turnover,

13:18 the revenue rather than on their profits.

13:21 So those were examples of kind of constrained

13:23 policies and how policies in poor countries might differ

13:26 quite a bit from the optimal if you want

13:28 policy that is often advised for OECD country.

13:31 So today I'm gonna take this as a starting

13:33 point and I'll give you two examples from some,

13:35 from some of my recent work.

13:37 The first one is gonna focus on efficiency and it's gonna ask,

13:40 given limited resources of the audit department,

13:44 countries may often focus on only the largest firms.

13:48 What does that mean for aggregate production?

13:50 And the second paper is going to ask,

13:53 in a world where there's a

13:56 What are the equity properties of consumption taxes?

14:00 So let me dive into the,

14:01 into the first statement.

14:03 So this paper is joint work with Roberto Fatal,

14:05 who's my colleague in the research department,

14:07 and Anders Jensen

14:09 at uh Harvard Kennedy School.

14:13 The context is the following,

14:14 administrative constraints lead to governments often

14:17 only enforcing taxes on large firms.

14:20 This produces a size-dependent tax.

14:23 Which

14:24 can have,

14:25 as you can imagine,

14:26 negative properties on growth.

14:29 That is the effective tax rate faced by firms as they grow increases

14:32 and therefore you might want to invest less than the optimal amount.

14:36 Uh,

14:37 and the allocation of resources,

14:38 for example,

14:39 labor is not optimal because often very productive firms do not hire

14:43 enough labor compared to what they should do while less productive firms

14:46 have too much labor.

14:49 I have two questions.

14:50 First,

14:50 to what extent is tax enforcement size dependent

14:53 and how does that differ across countries?

14:55 And secondly,

14:56 what are the,

14:57 what are the implications of this size dependence for aggregate productivity and

15:00 in explaining the gaps in productivity between rich and poorer countries.

15:06 We're gonna use data that are collected by our colleagues from the

15:08 World Bank Enterprise survey teams and so I thank them for this

15:11 amazing data efforts

15:13 and so today the data you'll see has data from over 140 countries.

15:17 We're gonna use specifically the question that

15:19 asked about tax inspection and tax compliance.

15:23 Take averages for everything you see at narrow industry

15:27 levels and I will explain why we do this

15:29 in a second.

15:32 So first let me show you the relation kind of in the raw data if you want

15:36 between the average size of an industry

15:38 measured with measured with its number of workers

15:41 and the likelihood of tax inspection that you see on the

15:44 vertical axis.

15:46 I'm showing this for the 6 most populous countries that we have in

15:49 the data and you can see that everywhere this relation is positive.

15:52 For example,

15:52 in Mexico,

15:54 uh,

15:54 a firm at the bottom of the size distribution has a

15:57 likelihood of tax inspection of 20% while one at the top

16:00 at around 60%.

16:05 So what we actually do in the prac in,

16:07 in the paper is use an instrumental variable.

16:09 So why is that?

16:09 The issue with measuring the OLS is that observed firm size in the

16:13 data might be distorted by the enforcement practice of the tax administration.

16:18 So the solution we'll use is to proxy for the optimal size of the firm.

16:23 In a context with few distortions,

16:25 so what we're really saying is that take the US and assume it is indeed

16:28 an economy without distortions.

16:31 Then the average size

16:32 of an industry of a permanent industry in the US reflects its true optimal

16:37 technology size.

16:38 So what the idea is gonna concretely does do is

16:42 Use US industry size to predict the size of the

16:45 same industry in the World Bank enterprise survey countries.

16:49 These predicted values in the 2nd stage are then going to be regressed on

16:52 average inspection of that industry in each of the World Bank and enterprise.

16:58 We find that a firm in an industry

17:01 with an average 25 workers

17:03 compared to a firm in an industry with an average of 50 workers

17:06 faces a 6% higher probability of tax inspection

17:10 over a mean of 61%.

17:12 And that firm also

17:15 Sorry,

17:16 I put it the wrong way around.

17:18 It's a firm in the,

17:19 it's the larger firms,

17:20 right,

17:20 that report 6% higher than the tax inspection

17:23 and also reports 5.5% more sales to the tax authority.

17:27 And these results are robust to many specifications that we show in the paper.

17:30 In particular,

17:31 they hold even when you only compare very narrow industries

17:35 with each other.

17:36 So for example,

17:36 take manufacturing of rubber,

17:38 rubber and plastic products

17:40 and

17:41 that relation holds when you compare

17:42 rubber manufacturing to plastic manufacturers.

17:45 It also holds if you only use the panel dimension of the data.

17:49 Which we don't have for all countries,

17:50 but some countries have repeated surveys and we can look at when industry

17:54 average firm size grows,

17:56 what happens to tax,

17:57 tax inspection and tax compliment.

18:00 Now,

18:00 for our purposes,

18:01 what we really care about is how does that

18:03 Resolve changes with development.

18:07 And this is what this figure shows you,

18:09 it shows you the coefficient from this IV regression,

18:11 those are the ones with the triangles,

18:13 4

18:14 different income groups,

18:15 5 groups of

18:17 uh countries's income.

18:19 What you can see

18:21 is that if you go to the bottom right,

18:22 you can see that in the richest countries in the sample,

18:24 basically OECD countries,

18:26 there's actually a zero-size gradient in tax enforcement.

18:31 As you go to increasingly poorer countries.

18:34 Increasing and indeed in the poorest countries in

18:36 the sample you can see that the tax,

18:38 the,

18:38 the tax gradient is reaches its maximum value and it's for example

18:42 twice as large as what it is in upper middle income country.

18:47 So with these results,

18:49 Now,

18:50 what I want to ask what are the impacts on efficiency and aggregate productivity.

18:54 So we're going to take this result and use a firm dynamics model.

18:58 Uh,

18:59 to

19:00 ask

19:01 what is the effect of having those size dependent taxes,

19:04 that is effective tax rate that increases the firm size

19:07 on firms and in the model firms of different productivity

19:11 are gonna take into account that the size dependent taxes

19:14 and they're gonna choose the production number of workers,

19:17 their investment intensity,

19:19 and if they want to enter

19:21 or exit

19:22 the economy.

19:24 The question we ask is what happens

19:25 to aggregate productivity which we measure as TFP

19:28 if we remove the size dependent tax.

19:31 So what does it mean to remove a size-dependent tax?

19:33 Well,

19:34 I want you to think of the following counterfactuals that now all

19:36 firms are going to face the median enforcement intensity in the economy,

19:40 right?

19:41 So instead of this profile that you get a higher effective tax rate

19:44 which you get larger,

19:45 now everyone faces the same set of the median level.

19:48 We're going to calibrate this using the results you've seen

19:51 and then technology parameters to match from level properties.

19:56 That is standard in the literature.

19:59 So let me show you some of the mechanisms.

20:00 What happens when you remove

20:03 the size dependent taxes.

20:04 Those figures group

20:06 on the x-axis different levels of economic

20:09 development where 5 are the richest countries,

20:11 the OECD countries if you want.

20:13 As we've seen in this country,

20:14 there's no size dependent taxes,

20:15 so nothing happens when you remove size dependent.

20:19 There's two outcomes I'm showing you here.

20:21 To the left,

20:22 you can see average innovation intensity or investment levels if you want.

20:26 So if you remove the size dependence in tax

20:28 and you equalize everyone at the average level,

20:30 now firms have incentives to invest.

20:33 That makes them grow,

20:34 but as they grow,

20:34 they don't face higher tax rate and so they invest quite a bit more.

20:37 You can see that investment levels in the poorest countries in

20:39 the sample increased by 10% in this kind of fact.

20:42 On the other hand,

20:43 everything is not positive because what happens is that for

20:46 smaller firms,

20:47 less productive firms if you want,

20:48 they were actually facing a very low tax rate,

20:51 right?

20:51 That was below the medium we set.

20:53 And so

20:54 some firms now will exit.

20:57 will decide not to enter and you can see that

20:59 levels of entry compared to before and now at 80%.

21:03 So those are two of the mechanisms.

21:05 What happens then

21:06 you pass to average firm size.

21:08 Well,

21:08 if you combine those two mechanisms,

21:10 the fact

21:12 that there's more investment,

21:13 therefore firms grow

21:14 and small firms tend to not enter or exit

21:17 the economy,

21:18 well you see that big increase in firm size,

21:20 right?

21:20 And that matches quite well something we know that poorer

21:23 countries have firms that are too small compared to,

21:26 to,

21:26 to,

21:27 to richer countries,

21:28 right?

21:28 And you see that this is not trivial,

21:29 this is about a 30% increase in the average firm size.

21:34 right,

21:35 a measure of aggregate productivity.

21:38 And here you see that in the poorest country,

21:41 removing size dependence would increase

21:43 TFP by a little bit less than 1%.

21:47 So this is not huge,

21:48 which,

21:48 which is interesting,

21:49 right?

21:49 Because we have this effect on investment.

21:51 There's also an effect that I haven't shown you on static allocation.

21:54 You allocate

21:55 labor better

21:57 between uh productive and less productive firms.

22:00 But you see that that impact,

22:01 the dynamic impact of exit and entry actually limits

22:03 a little bit the gains from TFP right?

22:05 And so we get

22:06 an effect on TFP of about a bit 7%.

22:10 So what I've shown you here is that size-dependent enforcement

22:12 is actually fairly strong and it's stronger in lower-income countries

22:16 and then we've combined this data in a model to

22:18 shed light on aggregate implication of removing that size-dependent taxation.

22:22 We see that average from size increases substantially

22:25 but aggregate productivity gains are more moderate.

22:28 Uh,

22:28 and you know,

22:29 now this literature that actually studies

22:31 in those distortions

22:34 often argues that each single distortion is not that big for TFP and you have to

22:37 take them all into account and actually if you look in the paper we have a,

22:41 we have a,

22:42 an extension where we think of removing size dependence when you already have

22:46 a set of other distortion,

22:48 for example,

22:48 financial constraints in the economy

22:50 and there we find that the aggregate productivity gains are a little bit larger.

22:55 So this

22:56 focused on,

22:58 on,

22:58 on how constrained policies lead to maybe some inefficiencies.

23:02 The second paper is gonna think of how

23:05 a constraint,

23:06 in this case the existence of an informal sector is going to impact equity

23:10 and that's going to be the equity of consumption tax.

23:13 This paper is joint work with Lucy Gadel,

23:15 the University of Warwick and Anders Jensen and currently is

23:18 a working paper in the NBR working paper series.

23:24 So for the second paper let me give you a bit of

23:26 the backdrop which is a little bit of a bleak view of

23:29 what we think are the potential for tax

23:32 systems to reduce inequality in poorer countries.

23:34 And that's because of 22 characteristics of those tax systems.

23:38 First,

23:39 personal income taxes collection that we've discussed is limited

23:42 and usually personal income taxes are the key instruments

23:46 through which taxes can be progressed.

23:49 The second one is that consumption taxes,

23:51 which we often think of as a regressive or neutral tax instrument,

23:55 is a really important part of the tax mix.

23:57 It's actually

23:58 for poor countries,

23:59 50% of all their revenue.

24:01 So if you start from that picture,

24:02 you,

24:03 you know,

24:03 the consensus view with that it's actually quite hard

24:05 for taxis to perform redistribution in this country.

24:09 We're gonna challenge a little bit this view and ask the

24:11 following question can tax on consumption actually redistribute and if so,

24:15 how?

24:16 And I want you to think of two potential channels.

24:20 The first one I'm going to call the old channel,

24:22 all because it's been discussed for a long time in the literature,

24:25 is that of rate differentiation.

24:27 Indeed,

24:27 you could

24:28 exempt or set a reduced rate

24:30 on necessity products.

24:32 In particular,

24:33 a lot of countries' tax policy actually exempts food items

24:37 or sets a reduced rate.

24:40 The second channel we'll study.

24:42 is the more novel informality channel.

24:45 And though you know that has been discussed in the literature,

24:47 I think we're going to bring the first systematic evidence of this channel.

24:50 That is that consumption in the informal untaxed

24:53 sector may actually vary a lot with income.

24:57 So how are we going to go about this?

24:59 So we're going to use household expenditure surveys

25:02 from 31 countries that we've assembled together.

25:06 That's microdata

25:07 at

25:07 the,

25:08 for each purchase done by these households and that covers

25:11 a bit under half a million households across these countries.

25:15 The income levels go from some very poor sub-Saharan

25:17 African countries such as Burundi to some of the upper

25:20 middle income countries in Latin America such as Chile or Uruguay.

25:25 Those data I want you to think are open diaries of

25:27 consumption with coverage of all products that can be consumed.

25:31 And the key restriction and the reason why we

25:33 only have 31 countries and not let's say 100,

25:37 is that we required

25:38 that for each item purchased,

25:41 there's a detailed information on the place of purchase,

25:45 where this item from.

25:47 So think of place of purchases,

25:48 for example,

25:48 the supermarkets versus street sellers.

25:53 So why do we do this?

25:55 Well,

25:55 as you can imagine,

25:56 taxes paid on purchases are not observed in expenditure service,

26:00 at least in none of the ones I have.

26:04 So we're gonna rely on this place of purchase variable to try to purchases

26:09 if they belong to the formal or the informal sector.

26:12 So we use this place of purchase variable as a proxy.

26:16 What we're going to do is

26:18 develop a taxonomy of places of purchases that is

26:20 going to follow the work of David Lagakos.

26:23 We're going to define two sectors,

26:25 the traditional sector,

26:26 so that's going to be goods

26:28 that are home produced,

26:29 that are going to be goods from non-brick and mortar stores.

26:32 So think of open markets,

26:33 street sellers and the likes.

26:35 Goods purchased from convenience stores or from individual providers.

26:40 And that's going to be the definition of the traditional sector.

26:42 And then we have to define also the modern

26:44 sector that's going to have specialized branded stores,

26:46 for example,

26:47 large stores and then institutional services.

26:50 And the assumption we'll make is that purchases in

26:52 small-scale traditional sector do not pay consumption taxes.

26:56 Now that can happen either the jury by law or de facto.

26:59 So the jury could in some countries actually small sellers

27:03 do not have to register for the land tax.

27:05 That is quite a common policy.

27:07 That policy is often motivated by administrative reasons that they think that

27:11 getting after those small firms is too costly for the administration.

27:14 De facto means that those firms,

27:16 even when they should be registering,

27:18 stay under the radar of the tax administration

27:21 and then do not pay

27:23 taxes.

27:23 In the paper we bring quite a bit of evidence that this categorization we do

27:27 actually makes sense

27:28 uh using census data in a few countries.

27:32 The note,

27:33 and then if I have time I will come back to this later.

27:36 The traditional sector could still pay taxes through the production chain,

27:39 especially in the value-added tax

27:41 system,

27:42 and there are some adjustments we made in the paper

27:44 based on data to try to account for it.

27:48 With this measure now,

27:49 right?

27:49 That is we assigned the traditional sector to,

27:52 to be informal,

27:53 we can now measure the total share of

27:55 consumption in an economy that is done informal.

27:58 And that is the relation you see here,

28:00 which is the total informal consumption share at the

28:03 country level on a country's log GDP per capita.

28:06 There's not too many surprises here.

28:08 You can see that strong negative relation.

28:11 Countries in Africa that are to the left of this graph

28:14 on average consume 70 or 80% of total consumption informally,

28:18 while some of the richest countries you can see here

28:20 at the bottom,

28:21 for example,

28:22 Chile,

28:22 Uruguay,

28:22 Costa Rica,

28:23 that level falls more around 20%.

28:26 Now remember what we're after in this paper is not that aggregate relation,

28:29 but the one within countries.

28:31 What happens across households.

28:33 Get a bit richer.

28:35 So that's what I'm going to turn to next and I'm going to take an example

28:39 using Mexico.

28:41 What you see here is a fitted graph,

28:43 right?

28:43 So imagine there's 20,000 households behind this graph and I fitted you,

28:47 uh,

28:48 the best fit line

28:50 and we see the informal budget share

28:53 as a function of household log expenditure

28:56 purpose.

28:57 So the way to read this graph is at the bottom left,

29:00 you have household in the bottom D side.

29:02 Of the income distribution.

29:04 And you see that they consume 55% of all the

29:07 consumption from the informal sector or the traditional sector.

29:10 As you go to a household of the median where the gray vertical bar is,

29:14 that falls to 40% and then for household in the top side,

29:17 that falls to 20,

29:18 25%.

29:19 OK,

29:20 so you see the strong slopes.

29:22 The second thing to be observed is that this relation is actually very

29:25 linear,

29:25 right?

29:25 So we can summarize it quite well with just the slope

29:28 of this line.

29:30 In the paper you can see those for the 31

29:32 countries we have in the sample if you're interested,

29:34 but I'm gonna summarize

29:36 this

29:37 with

29:38 uh on,

29:39 on this graph,

29:39 right?

29:40 So I'm just showing you the slopes for the 31 countries we have in the sample.

29:44 Again,

29:44 the X axis is now the log GDP per capita of the country.

29:48 So two things to take away from this figure.

29:50 The first one is you can see that in all these countries,

29:52 those slopes are negatives,

29:53 right?

29:54 The,

29:54 the relation is downward sloping.

29:57 And the substantial,

29:58 right,

29:58 the average slope is -10.

30:00 So that means that as the household becomes 10% richer,

30:03 it consumes 1% point less

30:06 from the informal sector.

30:08 Maybe a second observation that is not.

30:11 That you observe a little bit of a U shape that is in poorer countries,

30:14 the slope is a little bit lower,

30:16 it increases for middle-income countries and maybe

30:19 decreases again as countries get richer.

30:23 What are the implications of these patterns

30:25 for the progressivity of consumption tax?

30:27 So remember,

30:28 progressive tax is a tax for which the rate increases

30:31 with households.

30:33 Total income or total expenditure.

30:35 Now consumption taxes are progressive if items

30:38 consumed disproportionate by the poor are exempted.

30:41 For example,

30:42 the traditional sector is exempt or

30:44 the food sector is exempt.

30:47 So in this first set of exercise I'm going to do,

30:48 I'm going to show you some mechanical simulation

30:50 where I'm going to assume that the government has

30:52 to collect 10% of its GDP in taxes and we're going to think of three scenarios.

30:56 The first scenario is going to apply a uniform tax rate on all goods.

31:00 That can only tax the modern sector

31:02 as is probably realistic.

31:05 The second scenario is a bit of a counterfactual,

31:07 counterfactual scenario

31:09 that I'm gonna call the naive policymaker,

31:11 right?

31:11 It's someone who

31:13 wants to create progressive tax,

31:14 so he's gonna exempt

31:16 food

31:17 from the tax base.

31:19 We think that it's possible to tax both modern and traditional goods,

31:22 right?

31:22 To actually tax the informants.

31:24 And then the third scenario is going to be one where food is exempted

31:28 and only the modern sector is going to be taxed,

31:30 so maybe that's the closest to the actual policy that you might observe uh

31:34 in a country.

31:37 So let's look at the average progressivity of consumption taxes,

31:40 and for this we're gonna use a,

31:42 look at the tax budget share under each of those scenarios.

31:45 On the X axis you have the sizes of the expenditure distribution.

31:48 I'm gonna take the average

31:50 across the 31 countries.

31:53 Uniform rate and there was no informal sector,

31:55 mechanically everyone would be paying 10% by assumption,

31:58 right?

31:58 So that,

31:58 that's black line which you can just think of is the,

32:01 what sometimes people think about is the neutral

32:04 uh

32:05 uh progressivity of consumption tax.

32:08 Now,

32:08 let's look at the scenario where we now

32:10 take seriously the existence of the informal sector,

32:12 right?

32:12 And so you can only tax the modern sector.

32:15 And so here you can see that now you get this fairly

32:18 A steep line that reflects the figures you saw before,

32:21 where on average a household in the bottom diesel

32:24 now pays less than 6% of its budget in taxes

32:27 while a household in the top diesels pays around 15% of its budget in tax,

32:32 OK.

32:34 The second thing we can do is now go back to,

32:36 to,

32:36 to.

32:39 Uh,

32:39 counterfactual

32:41 where we exempt food but think that we can tax both the modern traditional sector,

32:44 right?

32:44 And that gives you that green line.

32:46 So that green line is,

32:47 you know,

32:47 you can see it makes it somewhat progressive compared to the,

32:50 to the neutral scenario,

32:52 a bit less so

32:53 than just

32:54 accounting for the informal sector.

32:56 In and of itself,

32:57 the green line is not that interesting.

32:58 What is interesting is to compare it to what actually happens

33:01 in real life,

33:02 which is this

33:03 orange line,

33:04 right?

33:04 So now food is exempt

33:06 and the modern sector is taxed,

33:08 right?

33:08 So you cannot tax the informal.

33:10 And

33:10 so the key message I want you to get

33:12 from this new orange line

33:14 is to compare what one might think naively is happening when they exempt food that is

33:18 moving from the black line to the green

33:20 line which does increase quite a bit progressivity

33:23 to what actually happens in practice when one

33:26 exempts food and the informal sector exists,

33:29 which is actually to move from that

33:31 red line to the orange line and you can see that the marginal gain in progressivity

33:36 is actually very small,

33:37 right,

33:38 compared to

33:39 Uh,

33:40 to what you might have thought you were doing,

33:41 uh,

33:42 on that green.

33:44 That's one of one of the key messages.

33:47 Now,

33:49 I'm going to go quickly over this,

33:50 but you might wonder,

33:52 right now I'm showing you this for the average country in the sample,

33:54 how does progressivity changes

33:57 with countries that the function of the income level.

34:00 So,

34:02 A graph that has on the x axis of the log GDP per capita of the country

34:06 and show you

34:07 the effective tax rate paid at the top 20%,

34:10 the top two sides

34:12 over that of the bottom two sides,

34:14 OK?

34:14 And then you can see that that's a strongly downward sloping relation.

34:17 That means that those

34:18 taxes,

34:19 you know,

34:19 including the informal sector

34:21 in your accounting

34:22 implied that taxes on consumption are quite a bit

34:25 more progressive and particularly so in the poorest countries.

34:28 And the logic for that is that.

34:31 In a context where the informal sector is very large,

34:34 such as is the case in the poor countries.

34:38 Taxing consumption,

34:39 formal consumption is a really good tag of income if you want.

34:42 I know with almost certainty in some countries that if I observe

34:46 consumption in the formal sector,

34:47 this has to happen from a rich individual.

34:50 As the informal sector decreases in size,

34:52 the tagging mechanism weakens,

34:54 right?

34:54 And now

34:55 it's still the case that it's majority from richer people,

34:58 but some consumption in the formal sector also happens

35:00 from poorer people.

35:04 So what you've seen is fairly simplistic and the key

35:07 reason is that there's no behavioral responses right now,

35:10 right?

35:10 So what households and consumers are not allowed

35:13 to respond to the level of taxes which is

35:16 of course unrealistic.

35:17 So what we do in the second part of the paper

35:20 is adapt a commodity tax model of diamond

35:23 and

35:24 To have two different varieties,

35:26 a modern and a traditional varieties,

35:28 and that's the case for each good,

35:29 right?

35:29 So you can buy your milk from the informal sector or from the formal sector.

35:34 As consumers get richer,

35:35 they want to consume more of the modern variety that is tax and

35:38 you can think of maybe that's because of varieties of a higher quality,

35:41 for example,

35:42 so there's a taste for quality as you get richer.

35:45 In this model now consumers are going to be able to respond to taxes.

35:48 When you set higher tax rates,

35:49 that's going to lead to substitution

35:51 away

35:52 from the modern varieties and towards

35:54 the traditional varieties that aren't taxed.

35:58 To

35:59 produce a new type of equity efficiency trade-off.

36:04 The efficiency logic is going to push tax rates down

36:08 to prevent too much substitution occurring.

36:11 The equity reason those we've seen want to

36:13 have high tax rate because actually taxing formal consumption

36:16 is good for redistribution.

36:19 And so the question armed with that model we're

36:21 gonna ask is how do optimal tax rates and inequality

36:25 change with development.

36:28 We use for calibration the data patterns you saw

36:30 before and then we need to have some elasticities,

36:33 both product of good elasticities,

36:35 but

36:36 something that's harder to find which is

36:37 substitution elasticities between modern traditional varieties.

36:41 In practice,

36:42 I'll be honest,

36:42 we do not have

36:43 a lot of estimates in the literature and something that comes close

36:46 is the elasticity between

36:49 big brand stores and smaller

36:51 brand stores,

36:52 right?

36:52 And so we'll use this elasticity that is not exactly the

36:54 one we need but that is somewhat akin to it,

36:57 uh,

36:57 to calibrate our results.

37:00 And there's 22 outcomes I want to show you.

37:03 The first one is how much would you subsidize,

37:05 and I'm going to say subsidize,

37:06 there's still tax,

37:07 but you know how,

37:08 how much lower should the rate

37:10 on foods be compared to the general tax rate.

37:14 I'm going to come that poor

37:16 naive policymaker.

37:18 Uh,

37:18 that unfortunate naive policymaker that hasn't

37:20 realized that there's an informal sector.

37:22 He lives in the world of figure

37:25 8 to the left.

37:27 So if you don't account for the informal sector,

37:30 you see that on average you actually

37:32 want to set a rate on food that is about half

37:34 the rate on all other goods.

37:37 And if anything,

37:39 The preferential treatment

37:41 of food items should be more pronounced in poorer countries.

37:46 As you now go to the more realistic

37:47 world where you take into account the formal sector

37:49 and so you move from figure A to figure B,

37:51 you can see that this relation is actually overturned.

37:54 And if anything,

37:55 in some of the poorest countries we see in our data,

37:58 The preferential treatment you want to give to food is actually very small.

38:02 And a fraction of what it was before.

38:05 The second result is what happens to inequality reduction.

38:09 So here we're gonna look at the change in the Gini coefficient in before and after

38:14 consumption taxes.

38:16 Something important is we're not doing

38:18 anything with the revenue collected,

38:19 right?

38:19 So do not think of transfers,

38:21 revenue collected as

38:22 what I'm showing you basically.

38:25 And there's a few messages I want to show you here.

38:28 The first one is that that red dot

38:31 is the percentage change in Gini.

38:34 From applying a uniform rate to all goods

38:37 to when you can only tax,

38:38 only the formal sectors taxed

38:40 and you see that even this very coarse policy.

38:44 Produces a drop in Gi of,

38:46 of 2.3%,

38:48 right?

38:48 So that's,

38:49 that's,

38:49 that's not

38:50 nothing.

38:51 If you take the standard view

38:53 that often you know you,

38:54 you,

38:55 you see.

38:56 And here I'm using data from some of our colleagues from the commitment to

38:59 equity and a lot of thank you to Gabriela and Chalte and her colleagues.

39:04 The standard view you can see is that first black dot,

39:06 right?

39:07 That's the effect when you don't

39:08 take into account

39:10 uh the informal sector

39:11 of consumption and excise taxes on inequality and you can see

39:15 that take into account the informal sector leads to quite a bit

39:17 higher redistribution.

39:20 The second message

39:22 to highlight

39:24 is that you can see from the green dot,

39:26 that's the policy again the naive policy

39:29 where

39:30 you

39:30 can differentiate food,

39:32 apply a lower rate to food but you are,

39:33 you tax both sectors,

39:34 right?

39:35 And so if

39:35 the naive policymaker against might think he's getting

39:38 a big drop in GDP of 3% points.

39:41 Again,

39:41 that drop is actually a lot too

39:43 because the true drop that you're getting is going from that red dot to that orange.

39:48 OK.

39:49 Now the optimal policy you can see

39:51 is that orange dot that leads you to,

39:53 to a drop in Gini of around 3%.

39:55 That's when you actually optimally differentiate food and non-food items

39:59 and only in the formal sector is taxed.

40:02 Now,

40:02 how big is 3% of a drop in GDP?

40:05 So there's two ways to see this.

40:07 One is that it's a lot better than what we thought and

40:09 it's actually doing better than what

40:10 currently personal income tax and social securities

40:14 do in lower and middle-income countries.

40:16 There's a more negative view which is to say that we know that in rich countries,

40:19 personal income tax and social security can get to

40:21 drops on the Gini coefficient of 89 or 10%,

40:25 right?

40:26 So that shows you,

40:26 you know,

40:27 that there's some

40:27 moderate redistribution that's occurring but there's still a long way to go and

40:32 Very efficient,

40:33 very equitable tax system are gonna need

40:35 uh the development of,

40:36 of personal income tax.

40:40 I have a few minutes,

40:41 so I,

40:42 I'll tell you about some of the extensions

40:44 and limitations,

40:44 and a lot of these are developed in the paper,

40:47 some of the,

40:47 of which we're still working on.

40:50 The first one is that the pass through of

40:52 taxes to the traditional sector may not be zero.

40:55 In particular,

40:56 when you apply a value-added,

40:57 it's possible that some of the goods have

40:58 paid taxes earlier in the production chain.

41:01 In the paper,

41:02 we actually do an adjustment

41:05 using data from Mexico where all firms report the VAT they paid on

41:08 both inputs and output so that you can see in the paper.

41:11 We also have an ongoing micros study that uses a quasi experiment

41:16 to measure the pass-through

41:18 of

41:20 VAT

41:21 onto the prices

41:23 in informal retailers

41:24 and so for this we use the fact that in Mexico

41:27 municipalities at the border saw a big change in the value-added tax.

41:31 We merge the data on prices with data from the census

41:34 and in the census we can see the status in terms of paying the VATR.

41:39 The second limitation is that we assume

41:41 that incidents is entirely borne by consumers.

41:43 In practice it might fall partially on workers and on profits of those formal firms.

41:48 And so complete analysis we need to know who are the workers

41:51 and who are the owners of those formal firms versus informal firms.

41:55 In fact,

41:55 that's actually something we can often do

41:58 and so we'll,

41:58 we'll try to,

41:59 we'll try to do this extension.

42:06 implication of this work,

42:07 well,

42:08 first,

42:08 consumption taxes perform a non-trivial redistribution

42:11 in lower and middle-income countries.

42:13 That does not mean,

42:14 and I

42:14 hope that's not the message you're getting out of this,

42:16 that enforcement should stop focusing on small firms and on the informal sector.

42:21 There's a lot of reasons,

42:22 in particular production efficiency and fairness concerns,

42:25 competition

42:26 that you should still be trying to,

42:28 to,

42:28 to level the playing field and,

42:30 and,

42:30 and formalize.

42:31 But there's a strong equity case to exempt small firms,

42:34 the jury by law firm taxes,

42:37 and that case might become more and more important as technology is allowing

42:40 us now to bring smaller and smaller firms into the tax net,

42:43 and we might really want to think of what is the equity consequences of,

42:47 of,

42:47 of newer technologies and expanding the tax net.

42:50 The last

42:51 policy recommendation is that should food and

42:54 more general necessities be exempted from taxes.

42:56 What we show that's actually very hard to justify on

42:59 equity grounds in poor countries once you take into account

43:02 the informal sector.

43:03 There's a stronger case

43:05 for exempted,

43:06 exempting food

43:07 in

43:08 middle-income countries,

43:09 but then what I hope is that actually what should really happen in those

43:11 countries that the personal income tax should

43:13 be developed and performed more redistribution.

43:16 So in the 5 minutes I have left,

43:18 I'm gonna use those to hopefully,

43:20 you know,

43:20 to taret to tell you about more text research.

43:23 Uh,

43:23 and

43:24 a little bit on how to try to impact policy.

43:28 The two papers I showed you have the same structure

43:31 that they use open source data from a lot of countries

43:34 and then

43:35 describe some patterns which then they read through a model to

43:38 try to measure their impact on efficiency and on equity.

43:41 There's a fast developing complementary approach that is very exciting.

43:46 That use that idea of

43:47 the economist as a plumber and often uses RCTs and the idea

43:50 is to try to improve taxation by tweaking policies and incentives.

43:55 The method is also very exciting because often they work with

43:57 digitized administrative data and directly work with the tax administration.

44:01 So I'm going to take a few selected examples,

44:03 but I'm happy for anyone interested to,

44:05 to talk more.

44:07 For example,

44:08 Harry told me

44:09 shows that tax lotteries in Brazil really help

44:12 with the final stage reporting that is final sellers

44:16 increased by quite a bit the reported

44:18 uh sales toities

44:21 when consumers are incentivized to ask for receipts through those tax lotteries.

44:25 However,

44:25 on net,

44:26 she finds that because you have to redistribute

44:28 some of that money through the lotteries,

44:30 there's a large cost and the cost-benefit analysis is

44:32 not very clear on how efficient those policies.

44:36 There's also some exciting work by uh my colleague

44:39 in the recent group of Yebola Kunoe and,

44:41 and Victor Pullia

44:43 that show that electronic filing,

44:45 a policy that has become ubiquitous across countries.

44:48 Has no impact actually on average reported taxes,

44:51 which was interesting.

44:53 It has very heterogeneous impact as a function of the evasion risk levels of firms.

44:58 Indeed,

44:59 high risk evasion firms

45:02 increase their reported taxes

45:03 and presumably because now they pay less bribes,

45:07 which was shielding them from audits.

45:09 On the opposite,

45:10 low evasion risk firms.

45:12 Actually decrease the tax payments and that's probably for the same

45:15 reason because before they were overpaying potentially to avoid getting into

45:19 uh,

45:20 uh,

45:20 into a confrontation with the tax inspector.

45:24 And

45:25 a 4th paper

45:27 which

45:27 links to that idea of taxation as being more than just collecting revenue

45:32 is the work by Jonathan Weigel that shows in

45:34 the Democratic Republic of the Congo that broadening the

45:36 property tax base has some really nice externalities in

45:39 that it encourages citizens to participate and monitor local governments

45:43 and then enter into decisions of what public goods to provide.

45:51 I have left

45:53 I want to discuss a little bit how research on taxes and

45:57 at the World Bank

45:58 can

45:59 shape a bit what tax policies are.

46:01 The first one I hope that I conveyed is that

46:04 in lower and middle-income countries tax

46:05 policies face information capacity constraints,

46:08 and research is important to help tailor tax design to those constraints,

46:13 right?

46:13 And so that.

46:14 That means that of course one size doesn't fit all

46:16 but we can go a bit further than that.

46:17 This might be a correct or an appropriate tax policy given

46:21 the level of,

46:22 of uh

46:24 information and capacity constraint.

46:26 Data and technology really promised to help in

46:28 terms of relieving some of these constraints,

46:29 but again they have to be well used,

46:32 uh,

46:32 well integrated,

46:33 and so there's a lot of new challenges that also arising

46:36 with data and as I highlighted,

46:37 for example,

46:38 there's some new questions around equity.

46:41 So let me conclude by saying a little bit of where,

46:43 where we are and how I see the,

46:44 the,

46:44 the,

46:45 the next few years.

46:46 We're obviously in a,

46:48 in a very big crisis where I still think that

46:51 administrative data and data on tax can play a big role

46:54 and for example in some work we've done is using

46:57 administrative data on firms that are kind of live data,

46:59 for example,

46:59 the VAT is reported every month by firms to

47:02 try to track in real-time the economic situation.

47:05 Uh.

47:06 To the extent that

47:07 the reports are correct.

47:09 The second thing is a lot of countries are

47:11 currently designing emergency tax relief and social transfer measures

47:15 and we both have a role to advise,

47:17 but then later on to also evaluate

47:19 what was the impact of these policies.

47:21 Without a doubt,

47:21 I think the aftermath is going to put a lot of

47:23 pressure on public finances and it's going to require a lot

47:25 of countries to raise revenue once we move from the problematic

47:28 of debt towards that of kind of paying that debt.

47:32 And

47:33 it's gonna be hard,

47:34 uh,

47:34 a lot of hard work and

47:36 there's gonna be a lot of demand to achieve this equitably.

47:38 And so I,

47:38 I highlighted three I think areas where currently there's actually not a

47:41 lot of research in particular because it's hard to get the right data

47:45 and

47:45 we have a lot of work done to,

47:47 to be done to set the right standards and

47:49 provide evidence and that's on the international tax architecture,

47:52 how to adapt to those multinational corporations,

47:55 particularly the digital ones

47:57 where the current architecture is really inappropriate to tax them.

48:01 There's going to be the issues of taxes on income and wealth,

48:03 especially in a globalized economy with mobile taxpayers

48:07 and

48:08 to meet one of the pressing challenge of our times,

48:10 there's going to be issues of environmental taxation.

48:13 And so those are really interesting challenges because all of them have

48:15 some degree of coordination that is going to be required across countries

48:18 and what we're gonna require a lot more evidence

48:20 and I hope to be able to contribute and uh

48:23 and to bring data

48:25 to these questions.

48:27 So thank you very much for listening to me.

48:28 I'm looking forward to Marcelo's comments and I will

48:32 make the slides available also for uh anyone interested.

48:35 Thank you.

48:38 Thanks,

48:38 Pierre.

48:39 Um,

48:41 I just want to encourage people,

48:42 we have a couple of questions that are coming to the Webex chat.

48:45 I want to encourage people to add more questions on either the Webex or the,

48:49 or the YouTube chat,

48:51 but,

48:51 um,

48:52 that over to you,

48:53 Marcella.

48:55 Thanks so much,

48:55 Dion,

48:56 and um

48:57 it's a great pleasure to be here and I would like to thank uh Pierre for,

49:01 for a very interesting uh presentation.

49:04 Before I delve a bit into the presentation,

49:07 let me emphasize the importance of strengthening domestic revenue mobilization

49:12 and improving capacity to,

49:13 to collect taxes and,

49:14 and other revenues in a

49:17 of say in a transparent,

49:18 accountable,

49:19 and equitable manner.

49:20 I mean,

49:20 this is very central.

49:22 To the work program,

49:23 have a World Bank,

49:24 you know,

49:24 to

49:25 reduce poverty and,

49:26 and,

49:27 and,

49:27 and reduce inequality.

49:29 Um,

49:30 I mean,

49:31 domestic revenues from tax and,

49:32 and non-taxed sources are the most reliable and,

49:35 and sustainable way for a country

49:37 to pursue,

49:38 to pursue its objective of development

49:41 and shared prosperity.

49:42 That's particularly true.

49:44 And Pierre mentioned that

49:45 when we see the problems that excessive indebtedness

49:48 has caused to developing countries,

49:50 a lot of my time and the

49:51 time of my team

49:53 is on,

49:53 on that topics,

49:54 and that includes the G20,

49:56 debt service suspension initiative,

49:58 the new wider policy,

50:00 sustainable development financing policy.

50:02 So,

50:02 there's a lot of work

50:03 dealing

50:04 with the amount of debt accumulated and,

50:07 you know,

50:07 indeed,

50:08 I mean,

50:08 DRM,

50:09 the mass revenue mobilization.

50:11 Um,

50:12 it's even more relevant now that the

50:13 COVID-19 pandemic has increased the gap between

50:17 domestic revenues

50:18 and financing needs.

50:20 In most countries,

50:21 economic growth is declining and tax bases are shrinking

50:24 while spending pressure is increasing.

50:26 So the pandemic has elevated the need for,

50:28 for domestic revenue as a

50:30 reliable source to strengthen resilience and response of

50:34 countries in the light of the crisis.

50:36 So I really welcome all,

50:37 all these research efforts

50:39 and these events.

50:41 In particular,

50:41 I mean,

50:41 we are putting a lot of resources and time

50:45 and effort in thinking about how to improve domestic revenue mobilization

50:49 in a moment when it comes to getting out of a crisis.

50:51 So,

50:52 basically how to broaden the tax base

50:54 without

50:55 raising the tax burden on individuals

50:58 that,

50:58 uh,

50:59 you know,

50:59 already paying

51:00 enough taxes.

51:02 So,

51:02 let me start actually by offering a few thoughts on

51:05 how I view the tax and development research field.

51:08 Uh,

51:08 SPR also highlighted,

51:10 effective taxation

51:13 is constrained by what governments can observe,

51:16 all the information they have,

51:17 as well as the administrative capacity to implement policies.

51:21 These are indeed essential elements for strengthening revenue mobilization,

51:25 and most of the World Bank operations and assistance in developing countries

51:30 do focus on improving

51:31 revenue authority's administrative capacity and

51:35 And the ability

51:36 to access and make use of third party information.

51:39 What I would like to add to this equation is the need to build citizen trust.

51:45 Not only to increase revenue collection but also

51:47 to improve the quality and transparency of spending.

51:51 Trust between citizens and their governments can be strengthened by,

51:54 for example,

51:55 demonstrating that their hard-earned resources are being used wisely

51:59 or by simplifying tax systems

52:02 or making the administration less arbitrary.

52:05 You know why these elements can be

52:07 regarded as the extensions of administrative capacity,

52:10 for example,

52:11 a capable government builds trust and keeps its tax system simple.

52:16 I like to think of them as compliments,

52:18 uh,

52:18 you know,

52:19 and that,

52:19 that we should explicitly analyze and consider in our systems.

52:23 So I'll come back to this point at the end,

52:25 at the end of my discussion,

52:27 and there's

52:28 some resources being put out on this question of trust,

52:31 how to build trust.

52:32 The idea that paying tax is the most patriotic thing you can do or one of the most.

52:37 So that's,

52:38 I think it's also an important active,

52:39 uh,

52:40 agenda.

52:41 So if I turn on to Pierre's presentation

52:43 per se.

52:45 My take from it is the role of 3rd best policies

52:48 and the importance of microdata for studying questions of optimal taxation

52:52 in specific situations.

52:55 Third best assumes that developing

52:58 country tax developing country tax authorities

53:01 face severe information barriers

53:03 and serious informational constraints.

53:07 They actually showed us the unintended consequence of falling

53:10 quote unquote best practice.

53:13 Best practice often make implicit assumptions about the

53:16 conditions under which these practices are implemented,

53:19 such as government capacity

53:21 and the structure of the economy and agent's behavior.

53:24 So when these conditions are not present,

53:27 best practice may have unintended consequences.

53:30 In such cases,

53:31 it is important to recognize that there are trade-offs

53:33 in choice involved

53:35 and that there may not be a perfect solution.

53:38 Now that we have more and more access to micro-level data,

53:41 uh such as administrative data which includes digitalized tax returns,

53:47 we can obtain one's response and insights to these trade-offs.

53:51 Ultimately,

53:52 it allows us to better tailor policy and I told Pierre before,

53:55 this is really research that you can,

53:57 you can kind of

53:58 sink your teeth into it.

53:59 It's not

54:00 abstract even when you're doing empirical as a macroeconomist,

54:04 I've written many papers.

54:06 Often,

54:06 I mean,

54:07 we have abstract policy uh suggestion,

54:10 but this type of research agenda,

54:11 have very specific policy suggestion,

54:14 and that is very helpful.

54:16 In his first presentation,

54:17 Pierre implicitly described the effect of the best practice

54:21 to establish large taxpayer units,

54:24 uh,

54:24 enabling more effective enforcement.

54:26 So he finds that in low and middle-income countries

54:29 with limited capacity environment,

54:32 environments,

54:32 I mean,

54:33 tax enforcement tends to focus predominantly on,

54:35 on,

54:36 on large firms.

54:38 This is

54:39 understandable.

54:40 Uh,

54:41 if administration do not have the resource

54:42 to enforce tax compliance across all taxpayers,

54:46 then it is better to focus on the quote unquote big fish.

54:51 Now,

54:51 this may,

54:52 however,

54:52 have unintended consequence on productivity of these firms,

54:56 just innovation,

54:57 you know,

54:57 it,

54:57 it,

54:57 it will suppress innovation

55:00 and maybe lead to resource misallocation.

55:01 Several papers have

55:03 been

55:04 written on that.

55:06 I find that this effect disappears when tax audit becomes

55:09 more evenly distributed across firm sizes for rich countries.

55:13 Um,

55:15 You know,

55:16 uh,

55:16 um,

55:17 suggested that with increased capacity,

55:19 these unintended consequences recede.

55:23 Um,

55:23 it,

55:23 it's,

55:24 there,

55:24 there's an interesting aspect to this problem.

55:26 There's the dynamics between,

55:27 uh,

55:28 the tax policy that you choose

55:30 and how firms decide to allocate themselves,

55:32 including,

55:33 uh,

55:33 in terms of size.

55:34 So you've,

55:35 you've seen countries that actually have these special regimes for small firms.

55:39 Actually creating threshold effects

55:41 and firms not being able,

55:43 not,

55:43 not want to be detected or not want to,

55:45 you know,

55:45 break the law,

55:47 want to remain

55:48 small

55:48 or informal

55:50 to avoid detection and,

55:51 and,

55:52 uh,

55:52 and

55:53 sometimes they are even breaking the law,

55:54 but

55:55 they have,

55:56 stressful effects.

55:58 The act doesn't develop that as much,

55:59 but that's an effect that is important.

56:02 And in his chart of,

56:03 uh,

56:04 you know,

56:04 this kind of relationship between

56:06 being more developed and,

56:07 and having

56:08 less of a focus on,

56:10 on,

56:10 on,

56:10 on firm sizes

56:12 for,

56:12 to,

56:12 to,

56:13 to,

56:13 to get,

56:14 uh,

56:14 uh,

56:15 tax revenues.

56:16 Um,

56:17 it's,

56:17 it's actually codifying in many countries.

56:19 It's not just like,

56:21 uh,

56:21 uh,

56:22 operational,

56:23 you know,

56:24 direction to,

56:25 to individuals who work in tax

56:28 administration,

56:28 but it's actually in law.

56:30 And there is a big discussion of uh

56:32 some of these countries have too high of a tax burden given their development,

56:37 uh,

56:37 you know,

56:38 development,

56:38 um,

56:39 in,

56:40 in terms of GDP per capita,

56:42 income per capita.

56:43 But they have to,

56:44 you know,

56:44 they have too much tax,

56:45 so you tend to kind of try to help

56:47 the smaller firms by creating this,

56:50 this kind of special,

56:52 um,

56:53 special,

56:53 uh,

56:54 programs,

56:54 but that's

56:55 Counter-producing Pre produce

56:58 some evidence on

56:59 why it can be counter-producing

57:01 and uh um

57:03 there's research showing that there's an,

57:04 a negative impact on productivity and all that.

57:07 Um,

57:08 anyway,

57:08 so this dynamic relationship between taxation,

57:11 firm size,

57:11 and

57:12 it's quite interesting,

57:13 something that we should continue working on it,

57:15 despite already some results uh being uh

57:18 Available.

57:20 In his second presentation,

57:21 Pierre described the best practice of redistributive

57:24 policies in the form of sales,

57:26 tax exemptions,

57:27 or necessity goods

57:29 that is proportionately consumed by the poor.

57:32 Typically,

57:33 these exemptions apply to food.

57:35 However,

57:35 he finds that because they put them to purchase,

57:38 many of them assets in the informal sector,

57:40 such redistributive policies have a limited impact on inequality.

57:44 Formal food consumption not correlated with income

57:47 in,

57:47 in developing countries.

57:49 They may therefore be more effective instruments in this case

57:52 to supplement the income of the poor rather than costly,

57:56 uh,

57:56 possibly mistargeted tax exemptions.

57:59 So defines the phasizing importance of looking not

58:01 only at the jury but also the fact.

58:03 Uh,

58:04 de facto tax structures.

58:05 Now,

58:06 I'm not saying,

58:07 and I do not think they are saying

58:09 they should abandon

58:10 large taxpayer units or that you should have a blanket ban

58:14 on tax exemptions for goods predominantly.

58:17 Consumed by the poor.

58:18 But Pen's work

58:20 encourages us to think carefully

58:22 through the assumption consequences of our policy advice.

58:25 Going back to the point I made about the first part of the presentation,

58:29 I also

58:30 want you to think about

58:31 how having this particular

58:33 uh exemptions also affect the decisions of firms

58:36 to work in particular sectors.

58:38 Um,

58:39 It's a little bit different because here I talk about different products,

58:42 so food production,

58:43 but going back to the first part of the presentation about size

58:47 independently of what type of product you produce,

58:51 again,

58:51 the special tax systems

58:54 and programs

58:55 uh can have um

58:57 bad dynamic effects.

58:59 Uh,

59:00 now,

59:00 I want to take a step back and reflect on what this means,

59:02 um,

59:03 all,

59:03 all this means for the broader work that we do here at the World Bank.

59:07 As you may know,

59:08 the World Bank is one of the

59:09 largest providers of technical assistance and concession financing

59:13 for building capacity,

59:14 tax policy formulation,

59:16 tax administration.

59:17 So as such,

59:18 it is important that our device is underpinned by strong analytical work

59:21 like the work presented by Pierre today.

59:24 Uh,

59:25 application of such research to concrete country cases can enhance the support

59:29 the regional teams bring to our client countries,

59:32 and the global units in the research department

59:34 are really putting a lot of effort to understand

59:37 of providing these strong analytical underpinnings for policy work.

59:42 The link between research and practice is essential.

59:44 It's very important that our research,

59:47 researchers and operational teams continue to work together

59:49 and learn from each other,

59:51 take on board the latest scientific insights,

59:53 but also the lessons coming from the field.

59:56 We are uniquely positioned to facilitate this exchange.

59:59 And I believe actually the 2nd tax conference on tax and personal income

1:00:03 and wealth in developing countries,

1:00:04 which we have,

1:00:05 we organized just a few weeks ago,

1:00:08 is a good example of these joint efforts.

1:00:11 So let me end by identifying some questions for future research.

1:00:15 I will raise 3 areas that I believe will be of critical importance

1:00:18 to improve our understanding of how the tax system

1:00:22 works.

1:00:23 But first,

1:00:24 it is great to see a growing

1:00:25 body of research recognize the differences among countries

1:00:29 at different levels of income

1:00:31 and being able to derive meaningful policy implications

1:00:34 that reflect these differences.

1:00:37 Looking ahead,

1:00:37 I believe we need a framework that allows us to customize efforts

1:00:41 to different types of countries,

1:00:43 and that actually would apply

1:00:45 even

1:00:46 when they have the same level of income.

1:00:49 Fragile and conflicted affected countries are a good example and we put a lot of,

1:00:54 we are putting more and more resources

1:00:56 into studying these countries.

1:00:58 They face a unique set of challenges,

1:01:00 and nearly 2/3 of these countries

1:01:03 struggle to mobilize significant revenue.

1:01:06 Uh,

1:01:06 Although the World Bank is working with governments

1:01:08 in FCD countries to improve revenue mobilization,

1:01:12 often in very innovative ways,

1:01:14 more research and deeper understanding of various policy scenarios

1:01:18 and the impact

1:01:19 would be of great value.

1:01:21 Second,

1:01:22 moving

1:01:22 to tax administration,

1:01:24 doing his talk,

1:01:25 they observed the taxation constrained by what governments can observe.

1:01:30 This implies a focus on tax enforcement.

1:01:33 While enforcement is a key determinant of tax compliance,

1:01:37 there's some recent World Bank research

1:01:40 working paper

1:01:41 on innovations in tax compliance

1:01:43 which highlights the complementary roles of facilitation

1:01:47 and tax morale

1:01:48 by improving tax compliance.

1:01:51 Tax morale reflects individual ethics and values,

1:01:53 social norms,

1:01:55 and the extent of trust in tax systems,

1:01:58 with the latter

1:01:59 offering the most immediate target for prospective reformers.

1:02:03 So,

1:02:04 uh,

1:02:04 the World Bank has supported many reforms in,

1:02:06 uh,

1:02:07 aimed at facilitating tax compliance

1:02:09 in recent years

1:02:10 and is now increasingly exploring the impact of taxpayer trust

1:02:14 on tax morale and tax compliance,

1:02:17 you know,

1:02:17 as well as,

1:02:18 you know,

1:02:18 its potential to contribute to

1:02:20 political support for reforms.

1:02:23 So,

1:02:23 this said,

1:02:24 although research tells us that tax morale and trust matter,

1:02:28 it is hard to formulate specific policies

1:02:31 that would raise tax morale and trust.

1:02:34 In addition,

1:02:35 further research is needed to illustrate how tax morale and trust

1:02:39 impact tax compliance

1:02:41 and how trust-related interventions can

1:02:43 and should be sequenced

1:02:45 with other reform areas,

1:02:47 notably enforcement and facilitation.

1:02:50 For example,

1:02:51 when resourcing reform opportunities are limited,

1:02:53 should the tax authority focus

1:02:56 all its efforts

1:02:57 on improving,

1:02:58 uh,

1:02:58 audit strategies?

1:03:00 Or should Fox Mo say it's redress and appeal system

1:03:04 to ensure

1:03:05 taxpayers feel treated fairly and equitably.

1:03:08 So,

1:03:08 these are the type of funds again,

1:03:10 you can sink your teeth

1:03:12 on if,

1:03:12 if you get the answers and that's very helpful.

1:03:15 Finally,

1:03:16 if you turn to tax policy,

1:03:18 further research into the option mix of tax policies.

1:03:22 In different contexts remains much needed.

1:03:26 The results from PF's second paper are very important in this respect

1:03:30 and illustrate how innovative research can

1:03:32 yield new and important policy recommendations,

1:03:35 so I would encourage practitioners

1:03:37 to take note of the findings of how consumption tax rates,

1:03:40 informality,

1:03:41 and equity considerations interact.

1:03:43 I hope to see more research in this direction

1:03:46 as more microdata become available.

1:03:49 The evaluation of distribution impact of tax

1:03:51 policies returning to fashion and its relevance

1:03:53 has increased as we emerge from the pandemic.

1:03:57 Moreover,

1:03:57 increased attention is paid to different tax policy mix

1:04:01 which would include green taxes,

1:04:03 which is an area that

1:04:04 we in the global practice put a lot of emphasis on.

1:04:07 The current price may provide a window of opportunity

1:04:11 to develop a tool on carbon price taxation assessment.

1:04:15 We should use this tool

1:04:17 and you know,

1:04:18 and,

1:04:18 and by the way,

1:04:19 we are,

1:04:21 this tool is being developed together with IMF and,

1:04:23 and other colleagues at the World Bank

1:04:26 and we should use this tool and not

1:04:27 analytical device to choose the best tax policy mix

1:04:31 to address revenue mobilization

1:04:33 and reversing carbon emissions within more equitable.

1:04:37 You know,

1:04:37 tax systems.

1:04:38 So let me end on that note.

1:04:40 Look forward to continued discussion on how new research

1:04:43 strands can lead to innovative solutions

1:04:45 in countries where lack capacity,

1:04:47 low income,

1:04:48 and larger formal sectors present

1:04:50 severe constraints

1:04:52 that need to be overcome

1:04:53 to improve revenue collection service delivery,

1:04:56 and the name of the game is broadening tax

1:04:59 basis.

1:05:00 Uh,

1:05:00 and think about where the tax system's not going,

1:05:04 how that part of the economy can be

1:05:06 reached

1:05:07 without really raising the burden,

1:05:09 uh,

1:05:10 on people that are already paying a lot of tax and then maybe killing the recovery

1:05:14 that we hope is coming very soon.

1:05:16 Thank you.

1:05:17 Thank you very much for the work,

1:05:18 uh,

1:05:19 Pierre.

1:05:20 Creon.

1:05:23 Thanks,

1:05:23 Marcella,

1:05:24 both for the,

1:05:24 the reflections,

1:05:25 but then also the the very specific sort of guidance

1:05:28 on where you think the research should be going.

1:05:31 I think it's very helpful for us to hear,

1:05:33 to hear that.

1:05:34 Um,

1:05:35 Pierre,

1:05:35 before turning it back to you,

1:05:36 we've got a number of questions on,

1:05:38 on Webex.

1:05:38 I'm gonna just read out the 1st 3 and then maybe you could respond,

1:05:42 you could react to Marcella and maybe

1:05:44 address the 1st 3 and then we have,

1:05:46 if we have time,

1:05:46 we can go to the

1:05:48 subsequent questions.

1:05:50 Um,

1:05:51 so the first one comes from Richard.

1:05:53 And he asks,

1:05:54 does removing size dependent taxes affect aggregate revenue,

1:05:59 and if so,

1:05:59 by how much?

1:06:00 What are the implications of the lost or gain

1:06:04 in revenue for poor countries?

1:06:07 That's the first question.

1:06:10 Um,

1:06:11 Isidro asks,

1:06:13 Could informality be a subsidy to informal firms

1:06:17 instead of a subsidy to poor consumers?

1:06:20 Could informality be a subsidy to informal firms

1:06:23 instead of a subsidy to poor consumers?

1:06:26 And then Ishani asks,

1:06:28 how should we think about the potential endogeneity of size-based inspection

1:06:33 or enforcement.

1:06:38 So you've got a mix of sort of general and some very specific,

1:06:41 so I don't know how you wanna handle that,

1:06:42 but I wish you to,

1:06:43 um,

1:06:45 thank you

1:06:46 Marcelo for this excellent talk.

1:06:48 I,

1:06:49 you know,

1:06:49 I don't think I have,

1:06:50 there was no specific questions.

1:06:52 I,

1:06:52 I thank you for putting some pointers of where you see,

1:06:55 you know,

1:06:55 research being,

1:06:56 being useful.

1:06:58 Um,

1:06:59 I totally.

1:07:00 You said about,

1:07:01 you know,

1:07:01 the importance of tax morale and so on,

1:07:03 and actually,

1:07:04 you know,

1:07:04 as you know,

1:07:05 there's some exciting work being done,

1:07:06 uh,

1:07:07 including at the bank,

1:07:08 for example,

1:07:08 in our group,

1:07:09 there's Mavi Shokat who recently joined us and who's doing

1:07:12 excellent work exactly on this.

1:07:13 I'm trying to link

1:07:14 more closely,

1:07:15 for example,

1:07:15 public goods provision

1:07:17 with tax payments,

1:07:18 right,

1:07:18 and make people realize that the taxes are going to very specific uh uses.

1:07:23 Um,

1:07:24 thanks also for mentioning that

1:07:26 this question about the optimal tax mix.

1:07:28 I think that's a very interesting one and you know,

1:07:30 one of the things that

1:07:31 I think came out of today we focus

1:07:33 a lot of consumption taxes because in practice they're

1:07:35 overused if you want,

1:07:37 but one of the real question and I think probably the only way

1:07:39 to make tax systems more progressive will be to develop personal income taxes,

1:07:43 right?

1:07:43 And the question is

1:07:44 how do we,

1:07:45 how do we do this,

1:07:46 and that's really,

1:07:46 you know,

1:07:46 a question about the optimal tax mix.

1:07:49 Um,

1:07:51 And uh yeah so thank,

1:07:53 thank you Marcelo.

1:07:55 So let me take the questions in order.

1:07:57 So the first one I think is on the

1:07:58 revenue uh consequences of removing size dependent taxes.

1:08:02 And so that's an excellent point and I don't think,

1:08:04 I hope I,

1:08:04 I didn't convey that

1:08:06 this is a policy that should be done removing uh size dependent taxes.

1:08:09 Indeed if you look

1:08:11 at what happened the last 20 or 30 years,

1:08:13 most countries adopted those large taxpayer units and now actually

1:08:15 they're adopting medium taxpayer units and we're going towards increasing,

1:08:19 uh,

1:08:19 segmentation.

1:08:20 There's actually a nice paper out now by Basri,

1:08:24 Ben Olken,

1:08:24 and some co-authors that actually shows that.

1:08:27 Uh,

1:08:27 opening a large taxpayer unit leads to a lot more revenue collection.

1:08:31 Uh,

1:08:32 so

1:08:33 what we did in this paper is actually take the,

1:08:35 you know,

1:08:36 we,

1:08:36 we stay diagnostic about the revenue side and I think it's

1:08:39 great that there's evidence in other papers that tell us that,

1:08:41 you know,

1:08:41 these policies are good for revenue,

1:08:43 but we more wanted to ask is how large are the distortions created.

1:08:47 Actually,

1:08:47 I think the message is that those distortions are potentially not that large,

1:08:50 right?

1:08:50 It's about 1%

1:08:52 uh in terms.

1:08:53 Uh,

1:08:53 of TFP and so I think the total kind of evaluation

1:08:56 of this policy would have to bring those two sides together,

1:08:59 you know,

1:08:59 the distortions and the inefficiencies with the

1:09:01 revenue and depending on how you value

1:09:04 revenue,

1:09:04 you know,

1:09:04 you might think that the policy is optimal or not.

1:09:08 Um,

1:09:09 Thank you for these comments.

1:09:10 You know,

1:09:11 the reason why we didn't go about revenue is,

1:09:12 I think the data we had to actually

1:09:14 measure accurately revenue or the course for this.

1:09:18 The second question was on

1:09:21 informality,

1:09:22 is that right?

1:09:22 And this is rather a subsidy.

1:09:25 To informal sector.

1:09:26 OK,

1:09:27 so that's a very good

1:09:28 point and I think you know that's something I mentioned at the end which

1:09:32 again I don't think a message

1:09:34 or we don't have enough to say that

1:09:36 the informal sector should be,

1:09:38 uh,

1:09:39 that,

1:09:39 that,

1:09:39 that formalization should stop,

1:09:41 right?

1:09:41 And one of the good

1:09:42 reason to keep on formalizing is,

1:09:44 is exactly the question which is that

1:09:47 There's unfair competition between formal and informal firms

1:09:51 and so that's,

1:09:51 you know,

1:09:51 taking it from the production side.

1:09:53 What's quite new I think in our paper and no one had I

1:09:55 think looked at informality from that angle is taking more the consumption view,

1:09:58 right?

1:09:58 And so,

1:09:59 you know,

1:09:59 the two can coexist and I think there's an interesting question of how do they,

1:10:02 they all come together and that's not what we did here.

1:10:04 Here we only focus on the consumption side

1:10:06 and show that there's this very strong correlation

1:10:09 um

1:10:10 between

1:10:11 One's income level and the share consuming the informal sector

1:10:14 and I think you know that more raises some alarm bells

1:10:17 that

1:10:18 equity has to be part

1:10:20 of our thinking about formalization,

1:10:22 about thresholds,

1:10:24 about how technology is changing all that,

1:10:25 right?

1:10:25 And so,

1:10:26 you know,

1:10:27 point well taken,

1:10:28 I think that's an important

1:10:29 reason why we formalize in practice

1:10:31 and I think what we're doing here is bringing a,

1:10:33 a novel view.

1:10:35 The third paper is more specific.

1:10:37 It's about endogeneity of,

1:10:39 of firm size with respect to the level of taxes.

1:10:43 That's a,

1:10:43 a very

1:10:44 good point.

1:10:45 You know what we try to do in this paper is do this instrumental strategy,

1:10:49 these IV strategies,

1:10:50 assuming that in the US,

1:10:52 you know,

1:10:52 it's not a first order concern if you want that

1:10:55 that taxes are maybe more flat with respect to firm size

1:10:58 and that firms don't change their size,

1:11:00 uh,

1:11:01 thinking of tax enforcement.

1:11:03 You know,

1:11:03 in practice,

1:11:04 we,

1:11:04 you know,

1:11:04 that's an assumption that

1:11:06 could,

1:11:06 could,

1:11:06 you know,

1:11:07 could be false,

1:11:08 uh,

1:11:09 and that's some of the limitation likely my

1:11:10 tuition is that this paper is already published,

1:11:13 so we're not gonna go back on it,

1:11:15 but that's a point well taken,

1:11:16 and you know that we discussed with the referee.

1:11:18 The paper has

1:11:19 some robustness to changing that instrument and so on,

1:11:21 but,

1:11:21 uh,

1:11:22 not fully solved.

1:11:26 OK,

1:11:26 uh,

1:11:27 thanks,

1:11:27 Pierre.

1:11:28 Marcelo,

1:11:29 before going to the next set of questions,

1:11:30 I just wanna

1:11:31 give you a chance to come in if you wanted to say anything at this point.

1:11:34 It's fine if you want to just keep going.

1:11:36 I think actually,

1:11:37 um,

1:11:37 uh,

1:11:38 one of the questions touched the point that I

1:11:40 made about dynamic effects of taxation from size and

1:11:44 the decision to become informal and formal.

1:11:46 You know,

1:11:46 people doesn't quite

1:11:48 certain the sequences doesn't quite address that because

1:11:50 you are looking at the consumption side,

1:11:52 like you said,

1:11:53 um.

1:11:54 But there is a whole discussion,

1:11:55 this informality literature that basically tries to understand informality,

1:12:00 uh,

1:12:01 a result

1:12:02 of kind of too demanding institutions

1:12:05 given the level of development of our country.

1:12:07 And you do see some evidence in countries

1:12:09 that have very high tax to GDP ratios,

1:12:12 say,

1:12:13 Brazil,

1:12:14 that,

1:12:14 uh,

1:12:15 firms kind of become informal.

1:12:18 As a way to kind of

1:12:20 lower labor costs because they wouldn't even exist

1:12:23 if that,

1:12:23 and then if you,

1:12:24 if you put on top of it,

1:12:26 a,

1:12:26 a,

1:12:26 a tax system that basically says you don't need to pay tax if you're very little.

1:12:31 And there is a program like that in,

1:12:33 in Brazil that you pay tax,

1:12:35 but you pay less tax,

1:12:36 it's simpler and all that.

1:12:38 What is happen is that you have

1:12:40 more firms migrate to that particular program

1:12:43 without addressing the core problem,

1:12:45 which is labor costs are too high,

1:12:47 given productivity,

1:12:48 of course.

1:12:49 So,

1:12:49 um,

1:12:50 I mean,

1:12:50 think about this dynamic

1:12:52 effects.

1:12:53 I think you mentioned some of that

1:12:55 in your first presentation.

1:12:57 About

1:12:57 size and,

1:12:58 and this has an impact on TFP

1:13:00 which is very significant and several research attributes

1:13:04 this kind of bad incentive to become formal.

1:13:07 In Brazil,

1:13:08 uh,

1:13:10 because of the tax system,

1:13:11 and,

1:13:12 and,

1:13:12 and the solution was,

1:13:13 let me make it very simple for some and then everybody migrate to that

1:13:17 and some

1:13:18 people,

1:13:18 you know,

1:13:20 they don't have 50 employees,

1:13:22 they have 49,

1:13:23 so they fall into that to others.

1:13:25 So threshold effects are,

1:13:26 are very important there.

1:13:28 So anyway,

1:13:29 so that's something also that would be interesting to think about the,

1:13:32 the behavioral effect that you,

1:13:34 you mentioned,

1:13:34 I think in your presentation.

1:13:36 Thanks.

1:13:39 Thanks.

1:13:39 OK,

1:13:39 with that,

1:13:40 I'll,

1:13:40 I'll go to uh the question from Roman.

1:13:43 Who I think he's referring to the second paper you were

1:13:46 describing,

1:13:47 um,

1:13:47 are the preferences non-hommothetic

1:13:50 or,

1:13:51 as in

1:13:52 Faber and Fly,

1:13:53 the expenditure shares are fixed in each quintile

1:13:56 of the expenditure distribution within a country?

1:14:02 Then Maria asks,

1:14:04 Is there convincing evidence on how to improve information and

1:14:08 capacity constraints so that these are not taken as given?

1:14:12 Going back to the very

1:14:13 first premise

1:14:14 that you started off with.

1:14:16 Um

1:14:19 And then

1:14:21 Um,

1:14:21 a question from Oyebola.

1:14:24 Where she says,

1:14:25 curious about the no pass-through result of the VAT increase in Mexico.

1:14:32 Don't the informal firms buy from the formal sector,

1:14:35 for example,

1:14:36 processed goods?

1:14:37 What do you think explains the results?

1:14:45 Thank you,

1:14:45 Dan.

1:14:46 And then,

1:14:46 OK,

1:14:47 so.

1:14:51 Sorry,

1:14:51 um,

1:14:52 Mike Thoman sent a question which is pretty long,

1:14:54 so I'm gonna try to

1:14:55 shorten it slightly,

1:14:57 um.

1:14:58 Thank you for mentioning environmental taxation

1:15:01 and,

1:15:01 and we've been looking recently at the

1:15:04 effects of increasing excise tax on fossil energy to reduce emissions and

1:15:10 with effects on climate change.

1:15:12 This can increase the tax base

1:15:14 and reduce high marginal rates on other form of tax.

1:15:18 Other formal sector taxes.

1:15:21 Um

1:15:23 On the other hand,

1:15:24 energy expenditure increases with income.

1:15:27 In some cases with income elasticity greater than one,

1:15:29 so energy taxes can be progressive.

1:15:32 So then his question is,

1:15:33 can tax revenue.

1:15:35 Uh

1:15:37 Sorry.

1:15:40 Tax revenues that

1:15:41 can be used to further increase progressivity

1:15:45 through income transfers and or to reduce other taxes.

1:15:49 Sorry,

1:15:49 I'm

1:15:51 Maybe I'll ask

1:15:52 Mike if he wants to rephrase his question,

1:15:54 because I think something got a little bit lost in the last sentence there and um

1:15:59 Could

1:16:00 I,

1:16:00 I can also take that offline if you want it.

1:16:03 OK,

1:16:04 sorry about that.

1:16:06 Uh,

1:16:06 uh,

1:16:07 OK,

1:16:07 OK,

1:16:08 so we have the first three,

1:16:10 let me,

1:16:11 so,

1:16:11 so first,

1:16:12 to,

1:16:12 to Marcelo's point,

1:16:13 you know that that's an excellent one about dynamic incentives of,

1:16:16 of informality.

1:16:17 There's a nice paper actually in Marcelo Yo Brazilian by Gabriello

1:16:21 Lia at UCL which maybe you know that really tries to.

1:16:23 Quantify a bit those two views of informality,

1:16:26 right?

1:16:26 Those two high costs and you're forcing firms

1:16:29 to formalize when they shouldn't and the fact that we still have firms that are

1:16:31 informal when really they're fairly productive firms

1:16:33 and they really gaming the system and so

1:16:35 that hopefully,

1:16:36 you know,

1:16:36 that does provide some answers into a dynamic model to some of those questions.

1:16:41 Um

1:16:42 So,

1:16:42 so,

1:16:43 so that was 2.1.

1:16:44 On,

1:16:44 on Roman,

1:16:45 that's a fairly technical question.

1:16:47 Uh,

1:16:47 we allow for non-homothetic preferences,

1:16:50 but,

1:16:51 uh,

1:16:52 I would be happy to have a discussion with you

1:16:54 on if we should follow Faber and Fai,

1:16:56 uh,

1:16:57 instead,

1:16:58 and,

1:16:58 uh,

1:16:58 you know,

1:16:59 what's your recommendation,

1:17:00 you know,

1:17:00 I'll be happy to tell you more how we actually calibrated,

1:17:02 uh,

1:17:03 calibrated the model.

1:17:05 Uh

1:17:07 On the more general question of

1:17:10 what are we doing or you know,

1:17:11 is there

1:17:12 any

1:17:13 any research on how to improve information capacity constraints.

1:17:16 So I think that's a great point because

1:17:18 the two papers I showed takes more the structural view,

1:17:21 right,

1:17:21 which is you take this information capacity constraints as given

1:17:24 and you say how do we adapt policy to them.

1:17:26 The other view is to say,

1:17:27 you know,

1:17:27 let's try to improve them so that we don't,

1:17:29 we,

1:17:30 we get rid of them.

1:17:31 And maybe that

1:17:32 slide one before last that I showed started answering some of that,

1:17:35 right?

1:17:36 It's often,

1:17:36 you know,

1:17:37 we try to improve those at the margin.

1:17:38 That's an exciting literature because we try to use quasi experiments or RCT.

1:17:43 And

1:17:43 there's a lot going on in the

1:17:45 Philippines.

1:17:46 It's quite new.

1:17:46 I think the reason that this research,

1:17:48 research is fairly new is that before administrative tax data

1:17:51 was often hard to get and sometimes not digitized,

1:17:54 right?

1:17:54 And so in the last 5 years we've seen a

1:17:56 lot of new research and I think that's gonna continue

1:17:58 and that hopefully will have precise answers to the type

1:18:01 of uh technologies or interventions that can help.

1:18:05 With getting more information.

1:18:06 I can tell you for example myself with uh Anne Brockmeyer also from the World Bank,

1:18:10 we've been working in Senegal

1:18:11 on how to design risk scoring algorithm for risk for,

1:18:15 uh,

1:18:15 for firms

1:18:16 and then how to use this information because you can have

1:18:19 a good system but then if it's not being used,

1:18:21 there's not much point how to use this to select optimally

1:18:24 taxpayers and uh to incentivize uh tax inspectors to do so.

1:18:29 Uh,

1:18:29 so the short answer is yes,

1:18:30 there's a lot going on.

1:18:31 I'm happy to answer more by email and uh to give an update on this in,

1:18:35 in a year or two also.

1:18:37 Uh,

1:18:38 on Oyebola's question,

1:18:39 so we have indeed this study in Mexico that finds very

1:18:42 limited pass-through of an increase in taxes onto the informal sector.

1:18:47 We're,

1:18:47 to be honest,

1:18:48 we're currently investigating this and work is going slower with the

1:18:51 pandemic where we have to send two files to the to,

1:18:54 to the statistical office.

1:18:56 Uh,

1:18:56 it seems to be one,

1:18:58 you know,

1:18:58 that markets,

1:18:59 the formal informal sectors seem to be fairly segmented

1:19:03 in this case,

1:19:03 and so that could be one way to think of this.

1:19:06 There's actually some papers that you will know,

1:19:07 Ebola by Luci and another showing that

1:19:10 even within production chains,

1:19:11 informal firms tend to interact more with each other where formal firms

1:19:15 interact,

1:19:15 uh among formal firms,

1:19:18 and it could be the same way that consumers,

1:19:20 uh,

1:19:20 of,

1:19:21 um,

1:19:22 the consumers are very segmented,

1:19:24 uh.

1:19:25 In the consumption,

1:19:25 but

1:19:26 you know we don't have a,

1:19:27 a,

1:19:27 a,

1:19:28 a full,

1:19:28 full result on this

1:19:30 and maybe something relevant is that we

1:19:31 find very limited pass-through on the informal

1:19:33 sector but we do not find full pass through onto the formal sector.

1:19:36 We only find about 50% pass through onto the formal sector.

1:19:39 So that means that you know a lot of things are happening

1:19:42 away from prices that we should try to address.

1:19:45 Uh,

1:19:46 yeah,

1:19:46 I think that's,

1:19:48 that's the question,

1:19:49 right?

1:19:49 Yeah.

1:19:50 May I,

1:19:51 may I add one thing,

1:19:52 a quick thing.

1:19:53 Um,

1:19:54 it's,

1:19:54 uh,

1:19:55 uh,

1:19:55 uh,

1:19:56 your papers don't say anything about,

1:19:58 you know,

1:19:58 the first batch.

1:19:59 The first batch is still like backs of your body.

1:20:02 Flat tax,

1:20:03 but then you have perfect information,

1:20:05 perfect administration,

1:20:06 you go back through the income side

1:20:09 and just redistribute,

1:20:11 um,

1:20:11 say,

1:20:12 all the tax paid by the poor,

1:20:13 you just redistribute to them and all that.

1:20:16 That would be interesting,

1:20:17 and I don't know how you do this paper,

1:20:19 but

1:20:20 to kind of

1:20:21 think about the capacity of

1:20:23 say key,

1:20:24 let's talk about either countries,

1:20:25 so the poorest countries that we serve.

1:20:28 Uh,

1:20:30 some index of,

1:20:31 uh,

1:20:31 administrative capacity to redistributive policies

1:20:35 and,

1:20:35 and interact that with your results.

1:20:38 I mean,

1:20:39 in what

1:20:39 cases,

1:20:40 you know,

1:20:41 don't even go there because it's just going to be so regressive

1:20:44 and let's,

1:20:45 you know,

1:20:46 but in some cases,

1:20:47 you see,

1:20:47 there is some capacity there

1:20:49 and,

1:20:50 and maybe there has a trade-off between uh

1:20:53 uh.

1:20:54 You know,

1:20:54 uh,

1:20:55 equity and revenue,

1:20:57 uh,

1:20:57 capacity to,

1:20:57 to mobilize revenue,

1:20:59 and then for each country you think,

1:21:01 you know,

1:21:01 maybe this country should have more kind of

1:21:04 uniform

1:21:05 tax rate and use the capacity

1:21:07 to redistribute.

1:21:08 I mean,

1:21:09 if I talk to particular experts from particular countries,

1:21:11 they will have views,

1:21:12 but I'm wondering if there's a kind of a more general.

1:21:15 Kind of

1:21:16 paper that you can look

1:21:18 and,

1:21:18 and even if it's just internal,

1:21:20 you know,

1:21:21 that you can look at kind of how

1:21:22 you see countries in this in this capacity distribution

1:21:26 and how they interact

1:21:27 with the distributive issues.

1:21:30 That you mentioned in your paper.

1:21:32 Um,

1:21:32 anyway.

1:21:33 Thanks.

1:21:39 Uh,

1:21:39 thanks Marcello.

1:21:40 Maybe,

1:21:40 uh,

1:21:40 a quick answer to this.

1:21:42 I,

1:21:42 to my knowledge,

1:21:43 there's not such an index

1:21:45 currently.

1:21:46 Um,

1:21:46 I think it's one about statistical capacity,

1:21:48 right?

1:21:49 The capacity of the statistical agency that would be an interesting,

1:21:53 uh,

1:21:53 one to,

1:21:54 that's something that's we probably could do with the bank because

1:21:57 in a lot of cases we are the ones interacting with,

1:21:59 uh,

1:21:59 with them,

1:22:00 and you know,

1:22:00 the diamond tools maybe if you aggregated all those

1:22:04 information,

1:22:04 so just for everyone,

1:22:05 the diamond tool is something developed by the bank that tries to assess.

1:22:08 Among a set of metrics,

1:22:10 uh,

1:22:10 the practices of tax administration,

1:22:13 uh,

1:22:13 and the IMF has also a similar tool called TDA.

1:22:15 So,

1:22:16 so,

1:22:16 so there is information that could be,

1:22:18 that could be used for that.

1:22:19 I think what there isn't necessarily is a mapping

1:22:21 from that towards which policies can be implemented,

1:22:23 which is what you suggest.

1:22:25 There is a project again jointly led by DMF and the

1:22:28 World Bank to try to a little bit those tax guides.

1:22:30 Now they're not as precise maybe as what you suggest,

1:22:33 but they do try to link,

1:22:35 um.

1:22:36 You know,

1:22:36 to think of this idea of

1:22:37 appropriate tax policies given,

1:22:39 given the constraints.

1:22:41 Um,

1:22:42 in the more research side,

1:22:44 I think what happens is we often try to

1:22:45 find a proxy or sufficient statistic if you want for

1:22:49 a given level,

1:22:50 you know,

1:22:50 there's something interesting,

1:22:51 you know,

1:22:51 I work a lot with my co-author Anders Jensen has a really nice paper showing how

1:22:55 he tries to explain why the personal income tax is so weak.

1:22:59 Um,

1:22:59 in a lot of poorer countries and he directly

1:23:01 relates with the share of self-employed in the economy,

1:23:03 right?

1:23:04 And you can see that as this develops,

1:23:05 you see the thresholds moving and policies developing and so on.

1:23:09 So in that sense you can think of for some specific policies,

1:23:11 you know,

1:23:12 you might have

1:23:13 some of the sufficient statistics,

1:23:14 for example,

1:23:15 the share of self-employed

1:23:16 or we could think,

1:23:17 you know,

1:23:17 for example,

1:23:18 Harvard University also has developed a nice index of complexity of an economy.

1:23:22 I think that's a very interesting variable for something like the value-added tax,

1:23:25 you know,

1:23:25 the more links you have across firms,

1:23:27 probably the

1:23:28 more information you're able to generate,

1:23:30 right?

1:23:30 So more complex economies and a better position

1:23:34 for a given level of state capacity the less complex ones,

1:23:36 right,

1:23:37 so,

1:23:38 so,

1:23:38 yeah.

1:23:44 Well,

1:23:44 thank you,

1:23:44 everyone.

1:23:45 I think we,

1:23:45 we,

1:23:46 we don't have any,

1:23:46 we've exhausted all the questions in the Webex and,

1:23:48 and the,

1:23:49 and the YouTube and,

1:23:50 and just to let you know,

1:23:51 we had a,

1:23:51 a,

1:23:51 a sizable following on both platforms,

1:23:54 uh,

1:23:54 just attesting to the interest that everybody had in this topic.

1:23:58 Um,

1:23:58 I want to thank you,

1:23:59 Pierre,

1:23:59 for the presentation and Marcelo for,

1:24:02 for making time to come and discuss

1:24:03 and provide some really thoughtful reactions and,

1:24:05 and,

1:24:06 and,

1:24:06 and,

1:24:06 and good specific ideas for,

1:24:08 for moving forward.

1:24:10 Um,

1:24:11 So with that,

1:24:12 just thank you,

1:24:12 everybody,

1:24:13 and um

1:24:14 unless

1:24:15 Pierre Marcello,

1:24:16 you want to say again one last word,

1:24:18 I think we'll call this,

1:24:19 uh,

1:24:20 call this to an end.

1:24:21 Um,

1:24:22 thank you very much.

1:24:24 Bye bye everybody.

1:24:25 Thank you very much.

1:24:26 Great job,

1:24:27 yeah.

1:24:27 Thank you,

1:24:28 Marcella.

1:24:28 Thanks everyone.

showAllTimestamps

no

transcript

Um Welcome everybody, um, to this, uh, policy research talk. As you know, these talks provide us an opportunity to present work coming out of the World Bank's research department. With the goal of sharing the findings with colleagues inside and outside of our department as well as with people outside the World Bank. Now we have a platform for doing that easily. Um, I'd like to welcome our online audience, both on Webex as well as on YouTube. Today's talk is going to focus on tax policy and its impact on efficiency and redistribution. Low income countries only raise a fraction of the tax revenue that rich countries collect. Which limits our ability to provide their ability to provide public goods and invest in human capital, redistribute income and insure against shocks. So we're going to start with my colleague, Pierre Bakas, who will discuss how the difference in economic structure and state capacity can lead to markedly different tax policies across countries. He's going to illustrate some implications of these differences for growth and inequalities using results from his recent work, and he'll conclude by discussing some of the challenges to make tax systems more efficient and redistributed. Pierre is an economist in the macro and growth team of the Development Research Group. We're also grateful to have today, um, Marcelo Estevao as a discussion today. Marcelo is the global director of the World Bank Group's macroeconomics Trade and Investment Global Practice, MTI. He leads a large team of country economists, macroeconomists, and fiscal policy, debt and macro modeling experts. He is responsible for overseeing the delivery of the global analytical work on fiscal policy, debt policy, and economics of climate change. For coordinating the strategic direction of MTI and implementing it, for helping to shape and oversee MTI's country, regional and regional programs, and for mobilizing staff to work more effectively across the EFI and other global practices. So Marcella, I hope I got your last name right, because on my, on my notes, it's much longer, on the screen, it's much shorter, so tell me if I got it wrong. You did well. You did well. I don't want to impose the You know, Brazilian slash Portuguese last names to anyone. Thank you. Um, so, OK, so, um, with that, I'll turn it over to Pierre, who will talk for about 45 minutes, after which we'll hear from Marcelo for about 10 to 15 minutes. And then we'll conclude with the Q&A session, um, from questions from the audience. If you have questions, please submit them in the chat function, either in Webex or YouTube because we'll be monitoring both of those chats. Uh, so with that over to you, Pierre. Thank you, Dean, let me share my screen. OK. Thank you, Dion for the introduction and for inviting me to present my work. It's a pleasure to be here and so today I'll tell you about my field of focus which is on taxes, inequality and development and I'm very excited to do so as you'll see it's an active field of research which I hope I will convince you has some practical and useful uh lessons for the work we do at the bank in terms of policy making. The talk will follow the following structure. I'm gonna give you a bit of an overview of tax policy, especially how it changes across development in the 1st 1012 minutes, and then I'm gonna enter into more depth into two of my recent papers, one which with one which focuses on the role of taxes and its impact on efficiency and the other one on equity. So let me start with this overview of tax systems and development and I'm gonna take first a very big bird eye view of taxes in particular what are the key functions of the tax and transfer system. One way to categorize them are those four maybe in order of kind of maybe not importance but the order in which they follow with development. The first one of course and a lot of our work at the bank is this, is to invest in public goods, infrastructure, health, and education. The second one, which is the 2nd pillar of the bank, is to help reduce market inequality, and this can be done in two ways through taxes, in particular, to impose progressive taxes which increases with household incomes, and the second way is through transfers, in particular the well targeted and reach the poorest households in the country. The third role is to provide insurance, in particular in the context where uh market insurance might unravel, and so that's the case for example, for old age, for health, for unemployment, and as we've seen more recently, there's also a big demand for insurance from the state in the context of systemic crisis like the current pandemic. And the 4th goal that we're getting increasingly involved in the climate emergency is to correct for externalities such that, reflect the true social cost, uh, of goods. Now against the backdrop. A fact maybe you have seen or relation you have seen is how does tax revenue measured as a percentage of a country's GDP correlate with log GDP per capita. And you can see that there's this strong positive correlation that is richest countries in the world, countries in the OECD collect on average 35% of the GDP in taxes. Well if you look at the bottom corner of this graph, you see that poorest countries in the world collect 10, maybe 15% of the GDP in taxes. There is some extent to which this varies a little bit across regions and so here I'm showing you the same figure but averaging across regions and you can see for example that the South Asia is a little bit richer, that's sub-Saharan Africa, there's a very low tax revenue collection in these countries, similarly for the Middle East and North Africa. Those differences in tax, total tax revenue can be decomposed with the tax instruments that are used in each country. So let me start here. So now I've put groups of countries on the x-axis starting from the lower income countries going all the way to, to high-income countries and I'm showing you two types of taxes, taxes on trade, that's tariffs and taxes on corporations. You can see that if you look at this relation, well, if anything this looks pretty flat or maybe even a decreasing with, with countries' income, right? Now, taxes on trade and corporations often thought to be Easy to enforce taxes. In the case of trade, for example, all you have to do is observe a few big ports or in the case of corporations is focus enforcement on some of the large, uh, corporations in your countries. The next step we can add is that on consumption. It's often a big share of total revenue and you can see that's the case everywhere and now we start observing a little bit of a relation with, with income though you can still see that upper middle income countries, once you, when you only have those taxes collect a bit more than high-income countries. And so where does the big gap arise? Well, it's in those next set of taxes that I'm adding here which are taxes on personal income, on payroll, and on property and what I called other which often taxes on wealth, inheritance, and the likes. And so you can see that it's those taxes, those last three taxes that really explain the big gap we see between poor and rich countries and that is worrying for two reasons. First, it's because personal income. One of the key instruments to redistribute. So when we saw the 2nd role of the state to redistribute in rich countries, this is done with the personal income tax and increasing marginal tax rate on personal income and the 3rd role of the state, that of social insurance is usually achieved with revenue that comes from payroll taxes and so you can see that the lack of existence of, of those taxes in poor countries probably limits uh what the state can do. Now before I enter into, into the meat of the subject there's a question I often get is should governments try to, to actually increase the tax revenue? Is, is that a good objective to have? And I'm not going to provide a definitive answer because it probably depends on, on each country's circumstances, but I think in general, given where we are currently, the answer is on average, yes, and that's for a few reasons. First, many countries still collect today under 15% of the GDP in taxes, which is really an insufficient level to meet the, the sustainable goals and more generally provide enough of even that basic infrastructure, health, and education. Second, the current crisis is putting a lot of demand for the government to provide social insurance but also it's deteriorating its fiscal position, and so I think there's little doubt that in this, the aftermath of this crisis, there's gonna be uh a large push for governments to try to increase their tax revenue. Now it's important to design taxes which should not impede growth. But keeping that in mind, these few gross success stories that have happened with that, the size of the state increasing quite a bit and providing good infrastructure and insurance. So there's a bit of a chicken and egg problem if you want in any case. Secondly, it's clear that in some cases this And taxes, you know, can be inefficient, but even these inefficiencies, they have to be balanced against the fact that it's often very high returns to investment and to redistribution in context, uh, with low investment and a lot of inequality. And finally, taxation is more than just an instrument, it's also a way through which cities and the state interact and so tax can spur, can spur accountability and better governance. Now, one of the key issues that taxation is constrained by what governments can observe. Indeed, modern tax system rely on self-reported activity which is cross-validated then with a lot of third-party reported data from different sources. So let me give you the typical example you're probably familiar with which is the relation between employer and employee. If you try to report your salaried income differently than what it really was in a lot of rich countries, you'd have a problem because your uh your employer would have sent the, the same slip that it sent to you to the tax authority and automatically you, you, you get caught, right? Well, the problems in poorer countries information is limited and that's the case because there's a lot of self-employment, small-scale productions, production chains are small, so there's not a lot of interaction between firms and even uh when there are interactions there's often incomplete accounting right. So the first big constraint is that of information. The second one is even once you have information, you need a given level of administrative capacity to process this information. Modern tax administrations use efficient databases and risk algorithm to detect tax frauds and have incentives to actually detect tax evasion for tax inspectors. Now not to pick on this, on this tax inspector, but you can see in this kind of messy office that this is something we observe often when we're working in country with tax administration that the systems and departments are not always integrated, for example, maybe the customs doesn't communicate with the domestic revenue agency. There are few incentives for tax inspectors to detect frauds and even once they detect fraud, it's often quite hard to apply the law and actually collect the, the, the taxes that are. So what this shows is that there's some trade-offs in tax policy design. When we often study this in graduate school for example and study public finance, the textbook case we see is that of a trade-off between equity and efficiency. Poorer countries also need to consider that there's disinformation and capacity constraints and then move us from the world of second best tax policy to that of 3rd best policy. So let me provide you with two examples and then I'll delve into the two papers that I will, uh, I will discuss today. So the first one is that of consumption taxes. So in the textbook case when you have no information capacity, You actually do not care if you are, if you levy a tax on value added or a final sales tax. In practice, once you consider these information constraints, Value-added tax are to be preferred because they collect taxes at each level of production and this creates incentives for sellers and buyers to cross-report on each other. Now what's interesting that for example Dina Portman showed that this is the case in Chile and that indeed the value-added tax is to be preferred to the sales tax, but Chile is a fairly high capacity country and so recently there's been some research, for example, by Mara Wasim in Pakistan showing that in countries with weak administrative capacity, VAT. And there's a lot of these bogus firms that get created that create fake invoices and then it becomes a nightmare for the tax administration to try to levy anything. And that's interesting because actually if you see developments in the last few years it's the first time ever that a few countries have returned, have left the value-added tax and gone back to a sales tax and so it shows the importance of thinking of both information capacity constraint. Another example With work by Michael Best and co-authors and some of my work in Costa Rica is that of tax corporate profits. So the efficient thing to do is to tax profits that is revenue minus all deductible costs that have entered into production. In practice, costs are quite easy to fake and are very responsive to the tax rate and so these two papers show that a little efficiency cost. In terms of revenue, it's preferable to tax small and medium enterprises on the turnover, the revenue rather than on their profits. So those were examples of kind of constrained policies and how policies in poor countries might differ quite a bit from the optimal if you want policy that is often advised for OECD country. So today I'm gonna take this as a starting point and I'll give you two examples from some, from some of my recent work. The first one is gonna focus on efficiency and it's gonna ask, given limited resources of the audit department, countries may often focus on only the largest firms. What does that mean for aggregate production? And the second paper is going to ask, in a world where there's a What are the equity properties of consumption taxes? So let me dive into the, into the first statement. So this paper is joint work with Roberto Fatal, who's my colleague in the research department, and Anders Jensen at uh Harvard Kennedy School. The context is the following, administrative constraints lead to governments often only enforcing taxes on large firms. This produces a size-dependent tax. Which can have, as you can imagine, negative properties on growth. That is the effective tax rate faced by firms as they grow increases and therefore you might want to invest less than the optimal amount. Uh, and the allocation of resources, for example, labor is not optimal because often very productive firms do not hire enough labor compared to what they should do while less productive firms have too much labor. I have two questions. First, to what extent is tax enforcement size dependent and how does that differ across countries? And secondly, what are the, what are the implications of this size dependence for aggregate productivity and in explaining the gaps in productivity between rich and poorer countries. We're gonna use data that are collected by our colleagues from the World Bank Enterprise survey teams and so I thank them for this amazing data efforts and so today the data you'll see has data from over 140 countries. We're gonna use specifically the question that asked about tax inspection and tax compliance. Take averages for everything you see at narrow industry levels and I will explain why we do this in a second. So first let me show you the relation kind of in the raw data if you want between the average size of an industry measured with measured with its number of workers and the likelihood of tax inspection that you see on the vertical axis. I'm showing this for the 6 most populous countries that we have in the data and you can see that everywhere this relation is positive. For example, in Mexico, uh, a firm at the bottom of the size distribution has a likelihood of tax inspection of 20% while one at the top at around 60%. So what we actually do in the prac in, in the paper is use an instrumental variable. So why is that? The issue with measuring the OLS is that observed firm size in the data might be distorted by the enforcement practice of the tax administration. So the solution we'll use is to proxy for the optimal size of the firm. In a context with few distortions, so what we're really saying is that take the US and assume it is indeed an economy without distortions. Then the average size of an industry of a permanent industry in the US reflects its true optimal technology size. So what the idea is gonna concretely does do is Use US industry size to predict the size of the same industry in the World Bank enterprise survey countries. These predicted values in the 2nd stage are then going to be regressed on average inspection of that industry in each of the World Bank and enterprise. We find that a firm in an industry with an average 25 workers compared to a firm in an industry with an average of 50 workers faces a 6% higher probability of tax inspection over a mean of 61%. And that firm also Sorry, I put it the wrong way around. It's a firm in the, it's the larger firms, right, that report 6% higher than the tax inspection and also reports 5.5% more sales to the tax authority. And these results are robust to many specifications that we show in the paper. In particular, they hold even when you only compare very narrow industries with each other. So for example, take manufacturing of rubber, rubber and plastic products and that relation holds when you compare rubber manufacturing to plastic manufacturers. It also holds if you only use the panel dimension of the data. Which we don't have for all countries, but some countries have repeated surveys and we can look at when industry average firm size grows, what happens to tax, tax inspection and tax compliment. Now, for our purposes, what we really care about is how does that Resolve changes with development. And this is what this figure shows you, it shows you the coefficient from this IV regression, those are the ones with the triangles, 4 different income groups, 5 groups of uh countries's income. What you can see is that if you go to the bottom right, you can see that in the richest countries in the sample, basically OECD countries, there's actually a zero-size gradient in tax enforcement. As you go to increasingly poorer countries. Increasing and indeed in the poorest countries in the sample you can see that the tax, the, the tax gradient is reaches its maximum value and it's for example twice as large as what it is in upper middle income country. So with these results, Now, what I want to ask what are the impacts on efficiency and aggregate productivity. So we're going to take this result and use a firm dynamics model. Uh, to ask what is the effect of having those size dependent taxes, that is effective tax rate that increases the firm size on firms and in the model firms of different productivity are gonna take into account that the size dependent taxes and they're gonna choose the production number of workers, their investment intensity, and if they want to enter or exit the economy. The question we ask is what happens to aggregate productivity which we measure as TFP if we remove the size dependent tax. So what does it mean to remove a size-dependent tax? Well, I want you to think of the following counterfactuals that now all firms are going to face the median enforcement intensity in the economy, right? So instead of this profile that you get a higher effective tax rate which you get larger, now everyone faces the same set of the median level. We're going to calibrate this using the results you've seen and then technology parameters to match from level properties. That is standard in the literature. So let me show you some of the mechanisms. What happens when you remove the size dependent taxes. Those figures group on the x-axis different levels of economic development where 5 are the richest countries, the OECD countries if you want. As we've seen in this country, there's no size dependent taxes, so nothing happens when you remove size dependent. There's two outcomes I'm showing you here. To the left, you can see average innovation intensity or investment levels if you want. So if you remove the size dependence in tax and you equalize everyone at the average level, now firms have incentives to invest. That makes them grow, but as they grow, they don't face higher tax rate and so they invest quite a bit more. You can see that investment levels in the poorest countries in the sample increased by 10% in this kind of fact. On the other hand, everything is not positive because what happens is that for smaller firms, less productive firms if you want, they were actually facing a very low tax rate, right? That was below the medium we set. And so some firms now will exit. will decide not to enter and you can see that levels of entry compared to before and now at 80%. So those are two of the mechanisms. What happens then you pass to average firm size. Well, if you combine those two mechanisms, the fact that there's more investment, therefore firms grow and small firms tend to not enter or exit the economy, well you see that big increase in firm size, right? And that matches quite well something we know that poorer countries have firms that are too small compared to, to, to, to richer countries, right? And you see that this is not trivial, this is about a 30% increase in the average firm size. right, a measure of aggregate productivity. And here you see that in the poorest country, removing size dependence would increase TFP by a little bit less than 1%. So this is not huge, which, which is interesting, right? Because we have this effect on investment. There's also an effect that I haven't shown you on static allocation. You allocate labor better between uh productive and less productive firms. But you see that that impact, the dynamic impact of exit and entry actually limits a little bit the gains from TFP right? And so we get an effect on TFP of about a bit 7%. So what I've shown you here is that size-dependent enforcement is actually fairly strong and it's stronger in lower-income countries and then we've combined this data in a model to shed light on aggregate implication of removing that size-dependent taxation. We see that average from size increases substantially but aggregate productivity gains are more moderate. Uh, and you know, now this literature that actually studies in those distortions often argues that each single distortion is not that big for TFP and you have to take them all into account and actually if you look in the paper we have a, we have a, an extension where we think of removing size dependence when you already have a set of other distortion, for example, financial constraints in the economy and there we find that the aggregate productivity gains are a little bit larger. So this focused on, on, on how constrained policies lead to maybe some inefficiencies. The second paper is gonna think of how a constraint, in this case the existence of an informal sector is going to impact equity and that's going to be the equity of consumption tax. This paper is joint work with Lucy Gadel, the University of Warwick and Anders Jensen and currently is a working paper in the NBR working paper series. So for the second paper let me give you a bit of the backdrop which is a little bit of a bleak view of what we think are the potential for tax systems to reduce inequality in poorer countries. And that's because of 22 characteristics of those tax systems. First, personal income taxes collection that we've discussed is limited and usually personal income taxes are the key instruments through which taxes can be progressed. The second one is that consumption taxes, which we often think of as a regressive or neutral tax instrument, is a really important part of the tax mix. It's actually for poor countries, 50% of all their revenue. So if you start from that picture, you, you know, the consensus view with that it's actually quite hard for taxis to perform redistribution in this country. We're gonna challenge a little bit this view and ask the following question can tax on consumption actually redistribute and if so, how? And I want you to think of two potential channels. The first one I'm going to call the old channel, all because it's been discussed for a long time in the literature, is that of rate differentiation. Indeed, you could exempt or set a reduced rate on necessity products. In particular, a lot of countries' tax policy actually exempts food items or sets a reduced rate. The second channel we'll study. is the more novel informality channel. And though you know that has been discussed in the literature, I think we're going to bring the first systematic evidence of this channel. That is that consumption in the informal untaxed sector may actually vary a lot with income. So how are we going to go about this? So we're going to use household expenditure surveys from 31 countries that we've assembled together. That's microdata at the, for each purchase done by these households and that covers a bit under half a million households across these countries. The income levels go from some very poor sub-Saharan African countries such as Burundi to some of the upper middle income countries in Latin America such as Chile or Uruguay. Those data I want you to think are open diaries of consumption with coverage of all products that can be consumed. And the key restriction and the reason why we only have 31 countries and not let's say 100, is that we required that for each item purchased, there's a detailed information on the place of purchase, where this item from. So think of place of purchases, for example, the supermarkets versus street sellers. So why do we do this? Well, as you can imagine, taxes paid on purchases are not observed in expenditure service, at least in none of the ones I have. So we're gonna rely on this place of purchase variable to try to purchases if they belong to the formal or the informal sector. So we use this place of purchase variable as a proxy. What we're going to do is develop a taxonomy of places of purchases that is going to follow the work of David Lagakos. We're going to define two sectors, the traditional sector, so that's going to be goods that are home produced, that are going to be goods from non-brick and mortar stores. So think of open markets, street sellers and the likes. Goods purchased from convenience stores or from individual providers. And that's going to be the definition of the traditional sector. And then we have to define also the modern sector that's going to have specialized branded stores, for example, large stores and then institutional services. And the assumption we'll make is that purchases in small-scale traditional sector do not pay consumption taxes. Now that can happen either the jury by law or de facto. So the jury could in some countries actually small sellers do not have to register for the land tax. That is quite a common policy. That policy is often motivated by administrative reasons that they think that getting after those small firms is too costly for the administration. De facto means that those firms, even when they should be registering, stay under the radar of the tax administration and then do not pay taxes. In the paper we bring quite a bit of evidence that this categorization we do actually makes sense uh using census data in a few countries. The note, and then if I have time I will come back to this later. The traditional sector could still pay taxes through the production chain, especially in the value-added tax system, and there are some adjustments we made in the paper based on data to try to account for it. With this measure now, right? That is we assigned the traditional sector to, to be informal, we can now measure the total share of consumption in an economy that is done informal. And that is the relation you see here, which is the total informal consumption share at the country level on a country's log GDP per capita. There's not too many surprises here. You can see that strong negative relation. Countries in Africa that are to the left of this graph on average consume 70 or 80% of total consumption informally, while some of the richest countries you can see here at the bottom, for example, Chile, Uruguay, Costa Rica, that level falls more around 20%. Now remember what we're after in this paper is not that aggregate relation, but the one within countries. What happens across households. Get a bit richer. So that's what I'm going to turn to next and I'm going to take an example using Mexico. What you see here is a fitted graph, right? So imagine there's 20,000 households behind this graph and I fitted you, uh, the best fit line and we see the informal budget share as a function of household log expenditure purpose. So the way to read this graph is at the bottom left, you have household in the bottom D side. Of the income distribution. And you see that they consume 55% of all the consumption from the informal sector or the traditional sector. As you go to a household of the median where the gray vertical bar is, that falls to 40% and then for household in the top side, that falls to 20, 25%. OK, so you see the strong slopes. The second thing to be observed is that this relation is actually very linear, right? So we can summarize it quite well with just the slope of this line. In the paper you can see those for the 31 countries we have in the sample if you're interested, but I'm gonna summarize this with uh on, on this graph, right? So I'm just showing you the slopes for the 31 countries we have in the sample. Again, the X axis is now the log GDP per capita of the country. So two things to take away from this figure. The first one is you can see that in all these countries, those slopes are negatives, right? The, the relation is downward sloping. And the substantial, right, the average slope is -10. So that means that as the household becomes 10% richer, it consumes 1% point less from the informal sector. Maybe a second observation that is not. That you observe a little bit of a U shape that is in poorer countries, the slope is a little bit lower, it increases for middle-income countries and maybe decreases again as countries get richer. What are the implications of these patterns for the progressivity of consumption tax? So remember, progressive tax is a tax for which the rate increases with households. Total income or total expenditure. Now consumption taxes are progressive if items consumed disproportionate by the poor are exempted. For example, the traditional sector is exempt or the food sector is exempt. So in this first set of exercise I'm going to do, I'm going to show you some mechanical simulation where I'm going to assume that the government has to collect 10% of its GDP in taxes and we're going to think of three scenarios. The first scenario is going to apply a uniform tax rate on all goods. That can only tax the modern sector as is probably realistic. The second scenario is a bit of a counterfactual, counterfactual scenario that I'm gonna call the naive policymaker, right? It's someone who wants to create progressive tax, so he's gonna exempt food from the tax base. We think that it's possible to tax both modern and traditional goods, right? To actually tax the informants. And then the third scenario is going to be one where food is exempted and only the modern sector is going to be taxed, so maybe that's the closest to the actual policy that you might observe uh in a country. So let's look at the average progressivity of consumption taxes, and for this we're gonna use a, look at the tax budget share under each of those scenarios. On the X axis you have the sizes of the expenditure distribution. I'm gonna take the average across the 31 countries. Uniform rate and there was no informal sector, mechanically everyone would be paying 10% by assumption, right? So that, that's black line which you can just think of is the, what sometimes people think about is the neutral uh uh progressivity of consumption tax. Now, let's look at the scenario where we now take seriously the existence of the informal sector, right? And so you can only tax the modern sector. And so here you can see that now you get this fairly A steep line that reflects the figures you saw before, where on average a household in the bottom diesel now pays less than 6% of its budget in taxes while a household in the top diesels pays around 15% of its budget in tax, OK. The second thing we can do is now go back to, to, to. Uh, counterfactual where we exempt food but think that we can tax both the modern traditional sector, right? And that gives you that green line. So that green line is, you know, you can see it makes it somewhat progressive compared to the, to the neutral scenario, a bit less so than just accounting for the informal sector. In and of itself, the green line is not that interesting. What is interesting is to compare it to what actually happens in real life, which is this orange line, right? So now food is exempt and the modern sector is taxed, right? So you cannot tax the informal. And so the key message I want you to get from this new orange line is to compare what one might think naively is happening when they exempt food that is moving from the black line to the green line which does increase quite a bit progressivity to what actually happens in practice when one exempts food and the informal sector exists, which is actually to move from that red line to the orange line and you can see that the marginal gain in progressivity is actually very small, right, compared to Uh, to what you might have thought you were doing, uh, on that green. That's one of one of the key messages. Now, I'm going to go quickly over this, but you might wonder, right now I'm showing you this for the average country in the sample, how does progressivity changes with countries that the function of the income level. So, A graph that has on the x axis of the log GDP per capita of the country and show you the effective tax rate paid at the top 20%, the top two sides over that of the bottom two sides, OK? And then you can see that that's a strongly downward sloping relation. That means that those taxes, you know, including the informal sector in your accounting implied that taxes on consumption are quite a bit more progressive and particularly so in the poorest countries. And the logic for that is that. In a context where the informal sector is very large, such as is the case in the poor countries. Taxing consumption, formal consumption is a really good tag of income if you want. I know with almost certainty in some countries that if I observe consumption in the formal sector, this has to happen from a rich individual. As the informal sector decreases in size, the tagging mechanism weakens, right? And now it's still the case that it's majority from richer people, but some consumption in the formal sector also happens from poorer people. So what you've seen is fairly simplistic and the key reason is that there's no behavioral responses right now, right? So what households and consumers are not allowed to respond to the level of taxes which is of course unrealistic. So what we do in the second part of the paper is adapt a commodity tax model of diamond and To have two different varieties, a modern and a traditional varieties, and that's the case for each good, right? So you can buy your milk from the informal sector or from the formal sector. As consumers get richer, they want to consume more of the modern variety that is tax and you can think of maybe that's because of varieties of a higher quality, for example, so there's a taste for quality as you get richer. In this model now consumers are going to be able to respond to taxes. When you set higher tax rates, that's going to lead to substitution away from the modern varieties and towards the traditional varieties that aren't taxed. To produce a new type of equity efficiency trade-off. The efficiency logic is going to push tax rates down to prevent too much substitution occurring. The equity reason those we've seen want to have high tax rate because actually taxing formal consumption is good for redistribution. And so the question armed with that model we're gonna ask is how do optimal tax rates and inequality change with development. We use for calibration the data patterns you saw before and then we need to have some elasticities, both product of good elasticities, but something that's harder to find which is substitution elasticities between modern traditional varieties. In practice, I'll be honest, we do not have a lot of estimates in the literature and something that comes close is the elasticity between big brand stores and smaller brand stores, right? And so we'll use this elasticity that is not exactly the one we need but that is somewhat akin to it, uh, to calibrate our results. And there's 22 outcomes I want to show you. The first one is how much would you subsidize, and I'm going to say subsidize, there's still tax, but you know how, how much lower should the rate on foods be compared to the general tax rate. I'm going to come that poor naive policymaker. Uh, that unfortunate naive policymaker that hasn't realized that there's an informal sector. He lives in the world of figure 8 to the left. So if you don't account for the informal sector, you see that on average you actually want to set a rate on food that is about half the rate on all other goods. And if anything, The preferential treatment of food items should be more pronounced in poorer countries. As you now go to the more realistic world where you take into account the formal sector and so you move from figure A to figure B, you can see that this relation is actually overturned. And if anything, in some of the poorest countries we see in our data, The preferential treatment you want to give to food is actually very small. And a fraction of what it was before. The second result is what happens to inequality reduction. So here we're gonna look at the change in the Gini coefficient in before and after consumption taxes. Something important is we're not doing anything with the revenue collected, right? So do not think of transfers, revenue collected as what I'm showing you basically. And there's a few messages I want to show you here. The first one is that that red dot is the percentage change in Gini. From applying a uniform rate to all goods to when you can only tax, only the formal sectors taxed and you see that even this very coarse policy. Produces a drop in Gi of, of 2.3%, right? So that's, that's, that's not nothing. If you take the standard view that often you know you, you, you see. And here I'm using data from some of our colleagues from the commitment to equity and a lot of thank you to Gabriela and Chalte and her colleagues. The standard view you can see is that first black dot, right? That's the effect when you don't take into account uh the informal sector of consumption and excise taxes on inequality and you can see that take into account the informal sector leads to quite a bit higher redistribution. The second message to highlight is that you can see from the green dot, that's the policy again the naive policy where you can differentiate food, apply a lower rate to food but you are, you tax both sectors, right? And so if the naive policymaker against might think he's getting a big drop in GDP of 3% points. Again, that drop is actually a lot too because the true drop that you're getting is going from that red dot to that orange. OK. Now the optimal policy you can see is that orange dot that leads you to, to a drop in Gini of around 3%. That's when you actually optimally differentiate food and non-food items and only in the formal sector is taxed. Now, how big is 3% of a drop in GDP? So there's two ways to see this. One is that it's a lot better than what we thought and it's actually doing better than what currently personal income tax and social securities do in lower and middle-income countries. There's a more negative view which is to say that we know that in rich countries, personal income tax and social security can get to drops on the Gini coefficient of 89 or 10%, right? So that shows you, you know, that there's some moderate redistribution that's occurring but there's still a long way to go and Very efficient, very equitable tax system are gonna need uh the development of, of personal income tax. I have a few minutes, so I, I'll tell you about some of the extensions and limitations, and a lot of these are developed in the paper, some of the, of which we're still working on. The first one is that the pass through of taxes to the traditional sector may not be zero. In particular, when you apply a value-added, it's possible that some of the goods have paid taxes earlier in the production chain. In the paper, we actually do an adjustment using data from Mexico where all firms report the VAT they paid on both inputs and output so that you can see in the paper. We also have an ongoing micros study that uses a quasi experiment to measure the pass-through of VAT onto the prices in informal retailers and so for this we use the fact that in Mexico municipalities at the border saw a big change in the value-added tax. We merge the data on prices with data from the census and in the census we can see the status in terms of paying the VATR. The second limitation is that we assume that incidents is entirely borne by consumers. In practice it might fall partially on workers and on profits of those formal firms. And so complete analysis we need to know who are the workers and who are the owners of those formal firms versus informal firms. In fact, that's actually something we can often do and so we'll, we'll try to, we'll try to do this extension. implication of this work, well, first, consumption taxes perform a non-trivial redistribution in lower and middle-income countries. That does not mean, and I hope that's not the message you're getting out of this, that enforcement should stop focusing on small firms and on the informal sector. There's a lot of reasons, in particular production efficiency and fairness concerns, competition that you should still be trying to, to, to level the playing field and, and, and formalize. But there's a strong equity case to exempt small firms, the jury by law firm taxes, and that case might become more and more important as technology is allowing us now to bring smaller and smaller firms into the tax net, and we might really want to think of what is the equity consequences of, of, of newer technologies and expanding the tax net. The last policy recommendation is that should food and more general necessities be exempted from taxes. What we show that's actually very hard to justify on equity grounds in poor countries once you take into account the informal sector. There's a stronger case for exempted, exempting food in middle-income countries, but then what I hope is that actually what should really happen in those countries that the personal income tax should be developed and performed more redistribution. So in the 5 minutes I have left, I'm gonna use those to hopefully, you know, to taret to tell you about more text research. Uh, and a little bit on how to try to impact policy. The two papers I showed you have the same structure that they use open source data from a lot of countries and then describe some patterns which then they read through a model to try to measure their impact on efficiency and on equity. There's a fast developing complementary approach that is very exciting. That use that idea of the economist as a plumber and often uses RCTs and the idea is to try to improve taxation by tweaking policies and incentives. The method is also very exciting because often they work with digitized administrative data and directly work with the tax administration. So I'm going to take a few selected examples, but I'm happy for anyone interested to, to talk more. For example, Harry told me shows that tax lotteries in Brazil really help with the final stage reporting that is final sellers increased by quite a bit the reported uh sales toities when consumers are incentivized to ask for receipts through those tax lotteries. However, on net, she finds that because you have to redistribute some of that money through the lotteries, there's a large cost and the cost-benefit analysis is not very clear on how efficient those policies. There's also some exciting work by uh my colleague in the recent group of Yebola Kunoe and, and Victor Pullia that show that electronic filing, a policy that has become ubiquitous across countries. Has no impact actually on average reported taxes, which was interesting. It has very heterogeneous impact as a function of the evasion risk levels of firms. Indeed, high risk evasion firms increase their reported taxes and presumably because now they pay less bribes, which was shielding them from audits. On the opposite, low evasion risk firms. Actually decrease the tax payments and that's probably for the same reason because before they were overpaying potentially to avoid getting into uh, uh, into a confrontation with the tax inspector. And a 4th paper which links to that idea of taxation as being more than just collecting revenue is the work by Jonathan Weigel that shows in the Democratic Republic of the Congo that broadening the property tax base has some really nice externalities in that it encourages citizens to participate and monitor local governments and then enter into decisions of what public goods to provide. I have left I want to discuss a little bit how research on taxes and at the World Bank can shape a bit what tax policies are. The first one I hope that I conveyed is that in lower and middle-income countries tax policies face information capacity constraints, and research is important to help tailor tax design to those constraints, right? And so that. That means that of course one size doesn't fit all but we can go a bit further than that. This might be a correct or an appropriate tax policy given the level of, of uh information and capacity constraint. Data and technology really promised to help in terms of relieving some of these constraints, but again they have to be well used, uh, well integrated, and so there's a lot of new challenges that also arising with data and as I highlighted, for example, there's some new questions around equity. So let me conclude by saying a little bit of where, where we are and how I see the, the, the, the next few years. We're obviously in a, in a very big crisis where I still think that administrative data and data on tax can play a big role and for example in some work we've done is using administrative data on firms that are kind of live data, for example, the VAT is reported every month by firms to try to track in real-time the economic situation. Uh. To the extent that the reports are correct. The second thing is a lot of countries are currently designing emergency tax relief and social transfer measures and we both have a role to advise, but then later on to also evaluate what was the impact of these policies. Without a doubt, I think the aftermath is going to put a lot of pressure on public finances and it's going to require a lot of countries to raise revenue once we move from the problematic of debt towards that of kind of paying that debt. And it's gonna be hard, uh, a lot of hard work and there's gonna be a lot of demand to achieve this equitably. And so I, I highlighted three I think areas where currently there's actually not a lot of research in particular because it's hard to get the right data and we have a lot of work done to, to be done to set the right standards and provide evidence and that's on the international tax architecture, how to adapt to those multinational corporations, particularly the digital ones where the current architecture is really inappropriate to tax them. There's going to be the issues of taxes on income and wealth, especially in a globalized economy with mobile taxpayers and to meet one of the pressing challenge of our times, there's going to be issues of environmental taxation. And so those are really interesting challenges because all of them have some degree of coordination that is going to be required across countries and what we're gonna require a lot more evidence and I hope to be able to contribute and uh and to bring data to these questions. So thank you very much for listening to me. I'm looking forward to Marcelo's comments and I will make the slides available also for uh anyone interested. Thank you. Thanks, Pierre. Um, I just want to encourage people, we have a couple of questions that are coming to the Webex chat. I want to encourage people to add more questions on either the Webex or the, or the YouTube chat, but, um, that over to you, Marcella. Thanks so much, Dion, and um it's a great pleasure to be here and I would like to thank uh Pierre for, for a very interesting uh presentation. Before I delve a bit into the presentation, let me emphasize the importance of strengthening domestic revenue mobilization and improving capacity to, to collect taxes and, and other revenues in a of say in a transparent, accountable, and equitable manner. I mean, this is very central. To the work program, have a World Bank, you know, to reduce poverty and, and, and, and reduce inequality. Um, I mean, domestic revenues from tax and, and non-taxed sources are the most reliable and, and sustainable way for a country to pursue, to pursue its objective of development and shared prosperity. That's particularly true. And Pierre mentioned that when we see the problems that excessive indebtedness has caused to developing countries, a lot of my time and the time of my team is on, on that topics, and that includes the G20, debt service suspension initiative, the new wider policy, sustainable development financing policy. So, there's a lot of work dealing with the amount of debt accumulated and, you know, indeed, I mean, DRM, the mass revenue mobilization. Um, it's even more relevant now that the COVID-19 pandemic has increased the gap between domestic revenues and financing needs. In most countries, economic growth is declining and tax bases are shrinking while spending pressure is increasing. So the pandemic has elevated the need for, for domestic revenue as a reliable source to strengthen resilience and response of countries in the light of the crisis. So I really welcome all, all these research efforts and these events. In particular, I mean, we are putting a lot of resources and time and effort in thinking about how to improve domestic revenue mobilization in a moment when it comes to getting out of a crisis. So, basically how to broaden the tax base without raising the tax burden on individuals that, uh, you know, already paying enough taxes. So, let me start actually by offering a few thoughts on how I view the tax and development research field. Uh, SPR also highlighted, effective taxation is constrained by what governments can observe, all the information they have, as well as the administrative capacity to implement policies. These are indeed essential elements for strengthening revenue mobilization, and most of the World Bank operations and assistance in developing countries do focus on improving revenue authority's administrative capacity and And the ability to access and make use of third party information. What I would like to add to this equation is the need to build citizen trust. Not only to increase revenue collection but also to improve the quality and transparency of spending. Trust between citizens and their governments can be strengthened by, for example, demonstrating that their hard-earned resources are being used wisely or by simplifying tax systems or making the administration less arbitrary. You know why these elements can be regarded as the extensions of administrative capacity, for example, a capable government builds trust and keeps its tax system simple. I like to think of them as compliments, uh, you know, and that, that we should explicitly analyze and consider in our systems. So I'll come back to this point at the end, at the end of my discussion, and there's some resources being put out on this question of trust, how to build trust. The idea that paying tax is the most patriotic thing you can do or one of the most. So that's, I think it's also an important active, uh, agenda. So if I turn on to Pierre's presentation per se. My take from it is the role of 3rd best policies and the importance of microdata for studying questions of optimal taxation in specific situations. Third best assumes that developing country tax developing country tax authorities face severe information barriers and serious informational constraints. They actually showed us the unintended consequence of falling quote unquote best practice. Best practice often make implicit assumptions about the conditions under which these practices are implemented, such as government capacity and the structure of the economy and agent's behavior. So when these conditions are not present, best practice may have unintended consequences. In such cases, it is important to recognize that there are trade-offs in choice involved and that there may not be a perfect solution. Now that we have more and more access to micro-level data, uh such as administrative data which includes digitalized tax returns, we can obtain one's response and insights to these trade-offs. Ultimately, it allows us to better tailor policy and I told Pierre before, this is really research that you can, you can kind of sink your teeth into it. It's not abstract even when you're doing empirical as a macroeconomist, I've written many papers. Often, I mean, we have abstract policy uh suggestion, but this type of research agenda, have very specific policy suggestion, and that is very helpful. In his first presentation, Pierre implicitly described the effect of the best practice to establish large taxpayer units, uh, enabling more effective enforcement. So he finds that in low and middle-income countries with limited capacity environment, environments, I mean, tax enforcement tends to focus predominantly on, on, on large firms. This is understandable. Uh, if administration do not have the resource to enforce tax compliance across all taxpayers, then it is better to focus on the quote unquote big fish. Now, this may, however, have unintended consequence on productivity of these firms, just innovation, you know, it, it, it will suppress innovation and maybe lead to resource misallocation. Several papers have been written on that. I find that this effect disappears when tax audit becomes more evenly distributed across firm sizes for rich countries. Um, You know, uh, um, suggested that with increased capacity, these unintended consequences recede. Um, it, it's, there, there's an interesting aspect to this problem. There's the dynamics between, uh, the tax policy that you choose and how firms decide to allocate themselves, including, uh, in terms of size. So you've, you've seen countries that actually have these special regimes for small firms. Actually creating threshold effects and firms not being able, not, not want to be detected or not want to, you know, break the law, want to remain small or informal to avoid detection and, and, uh, and sometimes they are even breaking the law, but they have, stressful effects. The act doesn't develop that as much, but that's an effect that is important. And in his chart of, uh, you know, this kind of relationship between being more developed and, and having less of a focus on, on, on, on firm sizes for, to, to, to, to get, uh, uh, tax revenues. Um, it's, it's actually codifying in many countries. It's not just like, uh, uh, operational, you know, direction to, to individuals who work in tax administration, but it's actually in law. And there is a big discussion of uh some of these countries have too high of a tax burden given their development, uh, you know, development, um, in, in terms of GDP per capita, income per capita. But they have to, you know, they have too much tax, so you tend to kind of try to help the smaller firms by creating this, this kind of special, um, special, uh, programs, but that's Counter-producing Pre produce some evidence on why it can be counter-producing and uh um there's research showing that there's an, a negative impact on productivity and all that. Um, anyway, so this dynamic relationship between taxation, firm size, and it's quite interesting, something that we should continue working on it, despite already some results uh being uh Available. In his second presentation, Pierre described the best practice of redistributive policies in the form of sales, tax exemptions, or necessity goods that is proportionately consumed by the poor. Typically, these exemptions apply to food. However, he finds that because they put them to purchase, many of them assets in the informal sector, such redistributive policies have a limited impact on inequality. Formal food consumption not correlated with income in, in developing countries. They may therefore be more effective instruments in this case to supplement the income of the poor rather than costly, uh, possibly mistargeted tax exemptions. So defines the phasizing importance of looking not only at the jury but also the fact. Uh, de facto tax structures. Now, I'm not saying, and I do not think they are saying they should abandon large taxpayer units or that you should have a blanket ban on tax exemptions for goods predominantly. Consumed by the poor. But Pen's work encourages us to think carefully through the assumption consequences of our policy advice. Going back to the point I made about the first part of the presentation, I also want you to think about how having this particular uh exemptions also affect the decisions of firms to work in particular sectors. Um, It's a little bit different because here I talk about different products, so food production, but going back to the first part of the presentation about size independently of what type of product you produce, again, the special tax systems and programs uh can have um bad dynamic effects. Uh, now, I want to take a step back and reflect on what this means, um, all, all this means for the broader work that we do here at the World Bank. As you may know, the World Bank is one of the largest providers of technical assistance and concession financing for building capacity, tax policy formulation, tax administration. So as such, it is important that our device is underpinned by strong analytical work like the work presented by Pierre today. Uh, application of such research to concrete country cases can enhance the support the regional teams bring to our client countries, and the global units in the research department are really putting a lot of effort to understand of providing these strong analytical underpinnings for policy work. The link between research and practice is essential. It's very important that our research, researchers and operational teams continue to work together and learn from each other, take on board the latest scientific insights, but also the lessons coming from the field. We are uniquely positioned to facilitate this exchange. And I believe actually the 2nd tax conference on tax and personal income and wealth in developing countries, which we have, we organized just a few weeks ago, is a good example of these joint efforts. So let me end by identifying some questions for future research. I will raise 3 areas that I believe will be of critical importance to improve our understanding of how the tax system works. But first, it is great to see a growing body of research recognize the differences among countries at different levels of income and being able to derive meaningful policy implications that reflect these differences. Looking ahead, I believe we need a framework that allows us to customize efforts to different types of countries, and that actually would apply even when they have the same level of income. Fragile and conflicted affected countries are a good example and we put a lot of, we are putting more and more resources into studying these countries. They face a unique set of challenges, and nearly 2/3 of these countries struggle to mobilize significant revenue. Uh, Although the World Bank is working with governments in FCD countries to improve revenue mobilization, often in very innovative ways, more research and deeper understanding of various policy scenarios and the impact would be of great value. Second, moving to tax administration, doing his talk, they observed the taxation constrained by what governments can observe. This implies a focus on tax enforcement. While enforcement is a key determinant of tax compliance, there's some recent World Bank research working paper on innovations in tax compliance which highlights the complementary roles of facilitation and tax morale by improving tax compliance. Tax morale reflects individual ethics and values, social norms, and the extent of trust in tax systems, with the latter offering the most immediate target for prospective reformers. So, uh, the World Bank has supported many reforms in, uh, aimed at facilitating tax compliance in recent years and is now increasingly exploring the impact of taxpayer trust on tax morale and tax compliance, you know, as well as, you know, its potential to contribute to political support for reforms. So, this said, although research tells us that tax morale and trust matter, it is hard to formulate specific policies that would raise tax morale and trust. In addition, further research is needed to illustrate how tax morale and trust impact tax compliance and how trust-related interventions can and should be sequenced with other reform areas, notably enforcement and facilitation. For example, when resourcing reform opportunities are limited, should the tax authority focus all its efforts on improving, uh, audit strategies? Or should Fox Mo say it's redress and appeal system to ensure taxpayers feel treated fairly and equitably. So, these are the type of funds again, you can sink your teeth on if, if you get the answers and that's very helpful. Finally, if you turn to tax policy, further research into the option mix of tax policies. In different contexts remains much needed. The results from PF's second paper are very important in this respect and illustrate how innovative research can yield new and important policy recommendations, so I would encourage practitioners to take note of the findings of how consumption tax rates, informality, and equity considerations interact. I hope to see more research in this direction as more microdata become available. The evaluation of distribution impact of tax policies returning to fashion and its relevance has increased as we emerge from the pandemic. Moreover, increased attention is paid to different tax policy mix which would include green taxes, which is an area that we in the global practice put a lot of emphasis on. The current price may provide a window of opportunity to develop a tool on carbon price taxation assessment. We should use this tool and you know, and, and by the way, we are, this tool is being developed together with IMF and, and other colleagues at the World Bank and we should use this tool and not analytical device to choose the best tax policy mix to address revenue mobilization and reversing carbon emissions within more equitable. You know, tax systems. So let me end on that note. Look forward to continued discussion on how new research strands can lead to innovative solutions in countries where lack capacity, low income, and larger formal sectors present severe constraints that need to be overcome to improve revenue collection service delivery, and the name of the game is broadening tax basis. Uh, and think about where the tax system's not going, how that part of the economy can be reached without really raising the burden, uh, on people that are already paying a lot of tax and then maybe killing the recovery that we hope is coming very soon. Thank you. Thank you very much for the work, uh, Pierre. Creon. Thanks, Marcella, both for the, the reflections, but then also the the very specific sort of guidance on where you think the research should be going. I think it's very helpful for us to hear, to hear that. Um, Pierre, before turning it back to you, we've got a number of questions on, on Webex. I'm gonna just read out the 1st 3 and then maybe you could respond, you could react to Marcella and maybe address the 1st 3 and then we have, if we have time, we can go to the subsequent questions. Um, so the first one comes from Richard. And he asks, does removing size dependent taxes affect aggregate revenue, and if so, by how much? What are the implications of the lost or gain in revenue for poor countries? That's the first question. Um, Isidro asks, Could informality be a subsidy to informal firms instead of a subsidy to poor consumers? Could informality be a subsidy to informal firms instead of a subsidy to poor consumers? And then Ishani asks, how should we think about the potential endogeneity of size-based inspection or enforcement. So you've got a mix of sort of general and some very specific, so I don't know how you wanna handle that, but I wish you to, um, thank you Marcelo for this excellent talk. I, you know, I don't think I have, there was no specific questions. I, I thank you for putting some pointers of where you see, you know, research being, being useful. Um, I totally. You said about, you know, the importance of tax morale and so on, and actually, you know, as you know, there's some exciting work being done, uh, including at the bank, for example, in our group, there's Mavi Shokat who recently joined us and who's doing excellent work exactly on this. I'm trying to link more closely, for example, public goods provision with tax payments, right, and make people realize that the taxes are going to very specific uh uses. Um, thanks also for mentioning that this question about the optimal tax mix. I think that's a very interesting one and you know, one of the things that I think came out of today we focus a lot of consumption taxes because in practice they're overused if you want, but one of the real question and I think probably the only way to make tax systems more progressive will be to develop personal income taxes, right? And the question is how do we, how do we do this, and that's really, you know, a question about the optimal tax mix. Um, And uh yeah so thank, thank you Marcelo. So let me take the questions in order. So the first one I think is on the revenue uh consequences of removing size dependent taxes. And so that's an excellent point and I don't think, I hope I, I didn't convey that this is a policy that should be done removing uh size dependent taxes. Indeed if you look at what happened the last 20 or 30 years, most countries adopted those large taxpayer units and now actually they're adopting medium taxpayer units and we're going towards increasing, uh, segmentation. There's actually a nice paper out now by Basri, Ben Olken, and some co-authors that actually shows that. Uh, opening a large taxpayer unit leads to a lot more revenue collection. Uh, so what we did in this paper is actually take the, you know, we, we stay diagnostic about the revenue side and I think it's great that there's evidence in other papers that tell us that, you know, these policies are good for revenue, but we more wanted to ask is how large are the distortions created. Actually, I think the message is that those distortions are potentially not that large, right? It's about 1% uh in terms. Uh, of TFP and so I think the total kind of evaluation of this policy would have to bring those two sides together, you know, the distortions and the inefficiencies with the revenue and depending on how you value revenue, you know, you might think that the policy is optimal or not. Um, Thank you for these comments. You know, the reason why we didn't go about revenue is, I think the data we had to actually measure accurately revenue or the course for this. The second question was on informality, is that right? And this is rather a subsidy. To informal sector. OK, so that's a very good point and I think you know that's something I mentioned at the end which again I don't think a message or we don't have enough to say that the informal sector should be, uh, that, that, that formalization should stop, right? And one of the good reason to keep on formalizing is, is exactly the question which is that There's unfair competition between formal and informal firms and so that's, you know, taking it from the production side. What's quite new I think in our paper and no one had I think looked at informality from that angle is taking more the consumption view, right? And so, you know, the two can coexist and I think there's an interesting question of how do they, they all come together and that's not what we did here. Here we only focus on the consumption side and show that there's this very strong correlation um between One's income level and the share consuming the informal sector and I think you know that more raises some alarm bells that equity has to be part of our thinking about formalization, about thresholds, about how technology is changing all that, right? And so, you know, point well taken, I think that's an important reason why we formalize in practice and I think what we're doing here is bringing a, a novel view. The third paper is more specific. It's about endogeneity of, of firm size with respect to the level of taxes. That's a, a very good point. You know what we try to do in this paper is do this instrumental strategy, these IV strategies, assuming that in the US, you know, it's not a first order concern if you want that that taxes are maybe more flat with respect to firm size and that firms don't change their size, uh, thinking of tax enforcement. You know, in practice, we, you know, that's an assumption that could, could, you know, could be false, uh, and that's some of the limitation likely my tuition is that this paper is already published, so we're not gonna go back on it, but that's a point well taken, and you know that we discussed with the referee. The paper has some robustness to changing that instrument and so on, but, uh, not fully solved. OK, uh, thanks, Pierre. Marcelo, before going to the next set of questions, I just wanna give you a chance to come in if you wanted to say anything at this point. It's fine if you want to just keep going. I think actually, um, uh, one of the questions touched the point that I made about dynamic effects of taxation from size and the decision to become informal and formal. You know, people doesn't quite certain the sequences doesn't quite address that because you are looking at the consumption side, like you said, um. But there is a whole discussion, this informality literature that basically tries to understand informality, uh, a result of kind of too demanding institutions given the level of development of our country. And you do see some evidence in countries that have very high tax to GDP ratios, say, Brazil, that, uh, firms kind of become informal. As a way to kind of lower labor costs because they wouldn't even exist if that, and then if you, if you put on top of it, a, a, a tax system that basically says you don't need to pay tax if you're very little. And there is a program like that in, in Brazil that you pay tax, but you pay less tax, it's simpler and all that. What is happen is that you have more firms migrate to that particular program without addressing the core problem, which is labor costs are too high, given productivity, of course. So, um, I mean, think about this dynamic effects. I think you mentioned some of that in your first presentation. About size and, and this has an impact on TFP which is very significant and several research attributes this kind of bad incentive to become formal. In Brazil, uh, because of the tax system, and, and, and the solution was, let me make it very simple for some and then everybody migrate to that and some people, you know, they don't have 50 employees, they have 49, so they fall into that to others. So threshold effects are, are very important there. So anyway, so that's something also that would be interesting to think about the, the behavioral effect that you, you mentioned, I think in your presentation. Thanks. Thanks. OK, with that, I'll, I'll go to uh the question from Roman. Who I think he's referring to the second paper you were describing, um, are the preferences non-hommothetic or, as in Faber and Fly, the expenditure shares are fixed in each quintile of the expenditure distribution within a country? Then Maria asks, Is there convincing evidence on how to improve information and capacity constraints so that these are not taken as given? Going back to the very first premise that you started off with. Um And then Um, a question from Oyebola. Where she says, curious about the no pass-through result of the VAT increase in Mexico. Don't the informal firms buy from the formal sector, for example, processed goods? What do you think explains the results? Thank you, Dan. And then, OK, so. Sorry, um, Mike Thoman sent a question which is pretty long, so I'm gonna try to shorten it slightly, um. Thank you for mentioning environmental taxation and, and we've been looking recently at the effects of increasing excise tax on fossil energy to reduce emissions and with effects on climate change. This can increase the tax base and reduce high marginal rates on other form of tax. Other formal sector taxes. Um On the other hand, energy expenditure increases with income. In some cases with income elasticity greater than one, so energy taxes can be progressive. So then his question is, can tax revenue. Uh Sorry. Tax revenues that can be used to further increase progressivity through income transfers and or to reduce other taxes. Sorry, I'm Maybe I'll ask Mike if he wants to rephrase his question, because I think something got a little bit lost in the last sentence there and um Could I, I can also take that offline if you want it. OK, sorry about that. Uh, uh, OK, OK, so we have the first three, let me, so, so first, to, to Marcelo's point, you know that that's an excellent one about dynamic incentives of, of informality. There's a nice paper actually in Marcelo Yo Brazilian by Gabriello Lia at UCL which maybe you know that really tries to. Quantify a bit those two views of informality, right? Those two high costs and you're forcing firms to formalize when they shouldn't and the fact that we still have firms that are informal when really they're fairly productive firms and they really gaming the system and so that hopefully, you know, that does provide some answers into a dynamic model to some of those questions. Um So, so, so that was 2.1. On, on Roman, that's a fairly technical question. Uh, we allow for non-homothetic preferences, but, uh, I would be happy to have a discussion with you on if we should follow Faber and Fai, uh, instead, and, uh, you know, what's your recommendation, you know, I'll be happy to tell you more how we actually calibrated, uh, calibrated the model. Uh On the more general question of what are we doing or you know, is there any any research on how to improve information capacity constraints. So I think that's a great point because the two papers I showed takes more the structural view, right, which is you take this information capacity constraints as given and you say how do we adapt policy to them. The other view is to say, you know, let's try to improve them so that we don't, we, we get rid of them. And maybe that slide one before last that I showed started answering some of that, right? It's often, you know, we try to improve those at the margin. That's an exciting literature because we try to use quasi experiments or RCT. And there's a lot going on in the Philippines. It's quite new. I think the reason that this research, research is fairly new is that before administrative tax data was often hard to get and sometimes not digitized, right? And so in the last 5 years we've seen a lot of new research and I think that's gonna continue and that hopefully will have precise answers to the type of uh technologies or interventions that can help. With getting more information. I can tell you for example myself with uh Anne Brockmeyer also from the World Bank, we've been working in Senegal on how to design risk scoring algorithm for risk for, uh, for firms and then how to use this information because you can have a good system but then if it's not being used, there's not much point how to use this to select optimally taxpayers and uh to incentivize uh tax inspectors to do so. Uh, so the short answer is yes, there's a lot going on. I'm happy to answer more by email and uh to give an update on this in, in a year or two also. Uh, on Oyebola's question, so we have indeed this study in Mexico that finds very limited pass-through of an increase in taxes onto the informal sector. We're, to be honest, we're currently investigating this and work is going slower with the pandemic where we have to send two files to the to, to the statistical office. Uh, it seems to be one, you know, that markets, the formal informal sectors seem to be fairly segmented in this case, and so that could be one way to think of this. There's actually some papers that you will know, Ebola by Luci and another showing that even within production chains, informal firms tend to interact more with each other where formal firms interact, uh among formal firms, and it could be the same way that consumers, uh, of, um, the consumers are very segmented, uh. In the consumption, but you know we don't have a, a, a, a full, full result on this and maybe something relevant is that we find very limited pass-through on the informal sector but we do not find full pass through onto the formal sector. We only find about 50% pass through onto the formal sector. So that means that you know a lot of things are happening away from prices that we should try to address. Uh, yeah, I think that's, that's the question, right? Yeah. May I, may I add one thing, a quick thing. Um, it's, uh, uh, uh, your papers don't say anything about, you know, the first batch. The first batch is still like backs of your body. Flat tax, but then you have perfect information, perfect administration, you go back through the income side and just redistribute, um, say, all the tax paid by the poor, you just redistribute to them and all that. That would be interesting, and I don't know how you do this paper, but to kind of think about the capacity of say key, let's talk about either countries, so the poorest countries that we serve. Uh, some index of, uh, administrative capacity to redistributive policies and, and interact that with your results. I mean, in what cases, you know, don't even go there because it's just going to be so regressive and let's, you know, but in some cases, you see, there is some capacity there and, and maybe there has a trade-off between uh uh. You know, uh, equity and revenue, uh, capacity to, to mobilize revenue, and then for each country you think, you know, maybe this country should have more kind of uniform tax rate and use the capacity to redistribute. I mean, if I talk to particular experts from particular countries, they will have views, but I'm wondering if there's a kind of a more general. Kind of paper that you can look and, and even if it's just internal, you know, that you can look at kind of how you see countries in this in this capacity distribution and how they interact with the distributive issues. That you mentioned in your paper. Um, anyway. Thanks. Uh, thanks Marcello. Maybe, uh, a quick answer to this. I, to my knowledge, there's not such an index currently. Um, I think it's one about statistical capacity, right? The capacity of the statistical agency that would be an interesting, uh, one to, that's something that's we probably could do with the bank because in a lot of cases we are the ones interacting with, uh, with them, and you know, the diamond tools maybe if you aggregated all those information, so just for everyone, the diamond tool is something developed by the bank that tries to assess. Among a set of metrics, uh, the practices of tax administration, uh, and the IMF has also a similar tool called TDA. So, so, so there is information that could be, that could be used for that. I think what there isn't necessarily is a mapping from that towards which policies can be implemented, which is what you suggest. There is a project again jointly led by DMF and the World Bank to try to a little bit those tax guides. Now they're not as precise maybe as what you suggest, but they do try to link, um. You know, to think of this idea of appropriate tax policies given, given the constraints. Um, in the more research side, I think what happens is we often try to find a proxy or sufficient statistic if you want for a given level, you know, there's something interesting, you know, I work a lot with my co-author Anders Jensen has a really nice paper showing how he tries to explain why the personal income tax is so weak. Um, in a lot of poorer countries and he directly relates with the share of self-employed in the economy, right? And you can see that as this develops, you see the thresholds moving and policies developing and so on. So in that sense you can think of for some specific policies, you know, you might have some of the sufficient statistics, for example, the share of self-employed or we could think, you know, for example, Harvard University also has developed a nice index of complexity of an economy. I think that's a very interesting variable for something like the value-added tax, you know, the more links you have across firms, probably the more information you're able to generate, right? So more complex economies and a better position for a given level of state capacity the less complex ones, right, so, so, yeah. Well, thank you, everyone. I think we, we, we don't have any, we've exhausted all the questions in the Webex and, and the, and the YouTube and, and just to let you know, we had a, a, a sizable following on both platforms, uh, just attesting to the interest that everybody had in this topic. Um, I want to thank you, Pierre, for the presentation and Marcelo for, for making time to come and discuss and provide some really thoughtful reactions and, and, and, and, and good specific ideas for, for moving forward. Um, So with that, just thank you, everybody, and um unless Pierre Marcello, you want to say again one last word, I think we'll call this, uh, call this to an end. Um, thank you very much. Bye bye everybody. Thank you very much. Great job, yeah. Thank you, Marcella. Thanks everyone.

showAllTranscripts

no

duration

PT1H24M31S

scene7File

worldbank/TaxationInequalityDevelopmentPRT

scene7Domain

scene7FileAvs

worldbank/TaxationInequalityDevelopmentPRT-AVS

title

TaxationInequalityDevelopmentPRT

description

TaxationInequalityDevelopmentPRT

showTimestampAndTranscript

yes