Popular in Economics

Economics

Gender and Economics

University of California Santa Barbara Shelly Lundberg

Banks and Policy Organisations

Greening the financial sector

European Investment Bank Matteo Ferrazzi

Economic History

Gravity with Gravitas: A Solution to the Border Puzzle

Boston College James E. Anderson

Economic Inequalities

Visions of Inequality

City University of New York Branko Milanovic

Economics

School Finance Equalization Increases Intergenerational Mobility

Yale University Barbara Biasi

Economics

Olivier Blanchard on U.S. Pandemic-Era Inflation

Peterson Institute for International Economics Olivier Blanchard

The State Capacity Ceiling on Tax Rates

View more insights in Social Sciences - Economics - Development Economics

View more figures from University of Southern California

View the page of Augustin Bergeron

Read the Study

How can developing countries increase the tax revenue they collect? Augustin Bergeron discusses research in the D.R. Congo and provides evidence that low government enforcement capacity sets a binding ceiling on the Laffer tax rate in some developing countries, thereby demonstrating the value of increasing tax enforcement in tandem with tax rates to expand fiscal capacity.

Image courtesy of interviewee. May 16, 2024

Generating Workspace

Export as PDF

The speaker discusses their research on the fiscal capacity of low-income countries and its relationship to economic development. They highlight two main motivations: first, the crucial role of tax revenue in funding public services and infrastructure; and second, the significant disparity in tax collection between low-income (about 10% of GDP) and wealthy countries (about 40% of GDP). The speaker, along with their co-authors, investigates two critical questions: the optimal setting of tax rates in low-capacity countries and the impact of tax enforcement on tax compliance and revenue.

They describe a study conducted in Canga, Democratic Republic of the Congo, focusing on property tax collection, which represents a significant portion of government revenue. The research utilized a randomized controlled trial (RCT) to understand the effects of varying tax rates on compliance and revenue. Key findings showed that lowering tax rates significantly increased taxpayer compliance, even surpassing the rate of decrease due to reduced rates, ultimately leading to higher overall tax revenue.

Furthermore, the research examined the influence of enforcement on taxpayer behavior, finding that weak enforcement led to a greater response to tax rate changes. Through additional RCTs, the researchers tested different enforcement strategies, confirming that low enforcement fuels compliance responsiveness to tax changes.

From these findings, they deduce that adjusting tax rates and enforcement strategies should be complementary, suggesting that governments with limited enforcement capacity might benefit from lowering tax rates, while those capable of improving enforcement should consider raising tax rates alongside stronger compliance measures.

Looking ahead, the researchers plan to analyze the impact of implementing a progressive property tax system in Canga, using advanced methods such as drone imagery for property assessments. They aim to compare the effects of a progressive tax system against a flat tax system, contributing further insights into tax policy effectiveness in low-income contexts.

Language:

Export as PDF

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.000 --> 00:02.605
Essentially the, the, the motivation for kind

00:02.605 --> 00:05.125
of my overall research comes from two main facts.

00:05.785 --> 00:07.405
Uh, the first one is that, you know,

00:07.825 --> 00:10.405
fiscal capacity has been shown to be essential

00:10.405 --> 00:11.525
for economic development.

00:12.105 --> 00:13.765
Uh, and, and typically

00:13.765 --> 00:15.925
because government need tax revenue very intuitively

00:15.925 --> 00:17.125
to provide public service

00:17.665 --> 00:20.805
or infrastructure, so road hospital schools,

00:20.805 --> 00:23.085
that are necessary for the development of the population.

00:23.545 --> 00:26.605
Uh, but despite this, what we see across the world is that,

00:26.865 --> 00:28.725
low income countries typically have

00:28.755 --> 00:30.005
very limited tax revenue.

00:30.265 --> 00:32.445
Uh, they collect, you know, if you wanna put number on it,

00:32.825 --> 00:35.725
they collect only 10% of their GDP and taxes.

00:36.105 --> 00:38.725
Uh, and if you want to compare it with rich countries,

00:38.725 --> 00:40.445
that number would be about 40%.

00:41.625 --> 00:44.405
Uh, and so, you know, there's, that's,

00:44.405 --> 00:45.605
that's kind of the status quo.

00:45.745 --> 00:48.725
Uh, and then you asked me, what, what can governments do

00:48.745 --> 00:50.965
to increase the tax, the revenue that they collect?

00:51.195 --> 00:52.485
Well, there's many things that they can do.

00:53.145 --> 00:56.565
Uh, and essentially what my co-authors

00:56.565 --> 00:58.725
and I, so my co-authors are Gabe to rec at the University

00:58.725 --> 01:00.885
of Pittsburgh and John Vial at uc, Berkeley.

01:01.265 --> 01:03.005
Uh, what we decided to do,

01:03.095 --> 01:05.365
we're all PhD student when we started this series

01:05.385 --> 01:07.005
of project, was

01:07.005 --> 01:09.925
to focus on the most fundamental tools you can think of

01:09.995 --> 01:13.405
that government have access, to, to raise, revenue.

01:14.025 --> 01:17.245
And those are essentially tax rates and tax enforcements.

01:18.025 --> 01:19.325
And so what we ask in this,

01:19.345 --> 01:21.885
in this project is essentially two things,

01:22.465 --> 01:24.765
is whether tax revenue in low capacity

01:24.825 --> 01:26.645
or low income countries are low

01:26.715 --> 01:28.765
because tax rates are not set optimally.

01:29.385 --> 01:30.405
Uh, so maybe,

01:30.585 --> 01:31.725
and you know, there's two things that

01:31.725 --> 01:32.845
could go be going on here.

01:32.935 --> 01:34.845
Maybe tax rates are set too low.

01:35.445 --> 01:36.525
Countries are not ambitious enough.

01:36.525 --> 01:38.325
They set low tax rate,

01:38.345 --> 01:41.045
and so they generate too little tax revenue per taxpayer.

01:41.985 --> 01:44.365
But you could imagine that actually the opposite is true.

01:44.695 --> 01:46.725
Maybe tax rates are actually set too high

01:46.745 --> 01:48.325
by those governments, and

01:48.325 --> 01:50.965
that generates too few taxpayers in the first place.

01:51.555 --> 01:53.845
Okay. So that's the first question we're interested in.

01:54.425 --> 01:56.445
Uh, and then the second one is more to deal

01:56.445 --> 01:58.365
with the second tool with tax enforcement.

01:58.465 --> 02:00.805
And what we're trying to understand is maybe

02:01.185 --> 02:03.845
how government should set their tax rate is actually going

02:03.845 --> 02:06.685
to depend on the government's capacity to enforce taxes.

02:06.785 --> 02:10.005
So in their enforcement capacity, their capacity to detect,

02:10.425 --> 02:12.765
tax delinquent, their capacities to sanction,

02:12.815 --> 02:14.805
those were, tax delinquent.

02:15.305 --> 02:18.165
And, and this is very intuitive, intuitively, you know,

02:18.165 --> 02:20.125
if, if enforcement is, is very low,

02:20.715 --> 02:22.685
then maybe if you raise tax rates, you're just going

02:22.685 --> 02:23.765
to have more tax delinquents.

02:23.945 --> 02:26.365
Uh, that's how people respond to tax rate changes.

02:27.065 --> 02:28.805
Uh, and if that's the case, maybe

02:28.805 --> 02:30.645
that won't affect tax revenue.

02:30.945 --> 02:32.365
You get some more revenue,

02:32.625 --> 02:34.285
but you get some more tax delinquent.

02:34.505 --> 02:37.165
But it could even need to lower tax revenue if there's

02:37.165 --> 02:39.285
really a lot of delinquents when you raise your tax rate.

02:40.145 --> 02:41.365
So that's, that's kind of

02:41.365 --> 02:43.245
what might happen if tax enforcement is low.

02:43.665 --> 02:46.045
Uh, but by contrast, if you were

02:46.045 --> 02:47.445
to increase tax enforcement,

02:47.815 --> 02:49.725
maybe suddenly things would look different.

02:49.985 --> 02:51.765
Uh, if you increase tax enforcement

02:52.105 --> 02:53.285
and raise your tax rates,

02:53.835 --> 02:55.045
then maybe you won't have this

02:55.045 --> 02:56.565
delinquency response to higher rates.

02:56.945 --> 02:58.725
Uh, and so you want to have a change

02:58.725 --> 02:59.805
In tax compliance.

03:00.305 --> 03:02.605
All you'll get is that some people will pay more taxes,

03:03.105 --> 03:04.685
and so then you'll get more tax revenue.

03:05.145 --> 03:07.365
And so we wanted to understand kind of both questions.

03:07.465 --> 03:10.485
Uh, how should you set tax rates, and,

03:10.505 --> 03:12.405
and in particular, how should you set tax rates when

03:12.405 --> 03:14.245
enforcement is low versus when enforcement is higher?

03:14.425 --> 03:17.085
An interesting moment for, I think, the type

03:17.285 --> 03:19.445
of studies my coauthor and I do,

03:20.065 --> 03:23.045
and we really build, like we're on the shoulder of giants.

03:23.065 --> 03:25.565
We build on a fantastic kind of trend

03:25.565 --> 03:27.245
that has happened in the past 10 years

03:27.945 --> 03:30.565
in development economics, in the field, in its entirety

03:31.035 --> 03:32.085
with, there's been more

03:32.085 --> 03:34.725
and more focus on studying,

03:34.845 --> 03:38.605
taxation in low income countries in the past 10 years.

03:39.185 --> 03:40.725
Uh, and this is for two reasons.

03:40.905 --> 03:43.325
The, those countries, have been more willing

03:43.345 --> 03:44.885
to share administrative data.

03:45.545 --> 03:48.925
And so researchers have had more access to tax data.

03:49.025 --> 03:51.525
So information that comes directly from the tax

03:51.525 --> 03:53.885
administration rather than relying on surveys

03:53.885 --> 03:55.965
where you can ask people whether they paid tax or not,

03:55.965 --> 03:57.165
but that's not going to be perfect.

03:57.315 --> 04:00.485
It's much better if you have the actual information from the

04:00.485 --> 04:03.845
tax administration about who paid, and what amount.

04:04.385 --> 04:05.605
So that's the first thing,

04:05.605 --> 04:07.205
that has been happening in this literature.

04:07.585 --> 04:10.365
Uh, and the second one is governments have been more willing

04:10.365 --> 04:11.925
to, experiment

04:12.345 --> 04:16.285
and to work closely with researchers on trying to understand

04:16.795 --> 04:19.085
what can they do to improve their tax capacity

04:19.085 --> 04:20.085
and their fiscal capacity.

04:20.085 --> 04:22.565
And in particular, and our study is an example of that,

04:22.565 --> 04:24.445
governments have been increasingly willing

04:24.505 --> 04:26.725
to rely on randomized control trials.

04:27.545 --> 04:30.005
Uh, so this is a method that's in part from,

04:30.065 --> 04:31.285
the literature in medicine

04:31.405 --> 04:33.685
that consistent in studying the impact of a policy

04:34.185 --> 04:35.405
by assigning it randomly.

04:35.425 --> 04:39.005
So you have a control group that receives the status quo,

04:39.185 --> 04:41.805
and then treatment groups that receive an intervention.

04:42.265 --> 04:44.605
And the decision between to allocate an individual

04:44.825 --> 04:46.085
or, a neighborhoods

04:46.085 --> 04:49.325
or village to each group is, is random.

04:49.665 --> 04:51.365
Uh, so we can compare those two groups

04:51.385 --> 04:53.725
to understand the impact of the policy we're interested in.

04:54.025 --> 04:55.405
And so, governments,

04:55.625 --> 04:58.245
in low income countries have been increasingly willing

04:58.305 --> 05:01.365
to experiment on taxation using those kind

05:01.365 --> 05:02.765
of randomized control trials.

05:03.185 --> 05:04.805
And so more concretely, you know, what,

05:04.805 --> 05:06.165
what does it mean in terms of studies?

05:06.165 --> 05:07.965
There's been kind of a lot of studies

05:07.965 --> 05:10.845
that have been very influential for, my co-authors

05:10.845 --> 05:13.605
and I, and we probably wouldn't be, doing the type

05:13.605 --> 05:15.085
of things we, we do without them.

05:15.505 --> 05:18.125
Uh, so for example, Dina Pomerance, who's at the University

05:18.125 --> 05:20.565
of Zurich, kind of pioneered this literature.

05:20.785 --> 05:23.805
She did an amazing kind of study, using both

05:23.805 --> 05:27.045
of those kind of modern things, the administrative tax data

05:27.425 --> 05:31.085
and the randomized controlled trial to understand the role

05:31.085 --> 05:33.885
of enforcement in the context of,

05:33.945 --> 05:36.085
the value added tax in, Chile.

05:36.505 --> 05:37.565
Uh, but there's more

05:37.565 --> 05:40.085
and more of those studies, for example,

05:40.185 --> 05:42.485
an Brooke Meyer and, and Juan Carlos Serato.

05:42.905 --> 05:47.045
Um, I've, I've done excellent work on trying to understand,

05:47.265 --> 05:49.245
property taxation in Mexico.

05:49.865 --> 05:50.965
Uh, and more recently,

05:50.965 --> 05:54.005
someone who's just on the market this year, Justine Belman,

05:54.305 --> 05:57.285
also did very impressive work in Senegal on trying

05:57.285 --> 05:59.245
to understand how do we assess properties?

05:59.385 --> 06:01.485
How do we assess values in order to tax them?

06:02.145 --> 06:04.765
Uh, so I think it's very exciting time for this literature.

06:05.225 --> 06:07.685
Uh, and, and so that's where we, we blend in.

06:07.685 --> 06:09.525
But, but actually I think what's interesting is

06:10.115 --> 06:12.765
even though it's a, it's a kind of booming moment

06:12.825 --> 06:15.085
and exciting moment for this literature, there's still a lot

06:15.365 --> 06:17.085
of questions that are quite fundamental

06:17.085 --> 06:18.765
that have not yet been answered.

06:19.385 --> 06:21.565
Uh, and that's, I think, where we try

06:21.565 --> 06:22.605
to make a contribution.

06:22.665 --> 06:24.925
So when you think of like questions

06:25.115 --> 06:27.965
that are very fundamental, such as how do you set tax rates?

06:28.065 --> 06:29.125
How do you enforce access?

06:29.555 --> 06:33.285
It's not that we were in kind of unchartered territory,

06:33.345 --> 06:35.565
but it's mostly that those questions have been typically,

06:35.905 --> 06:39.765
studied by public economists that focused on kind

06:39.765 --> 06:42.445
of high income countries, so the us, Europe, et cetera.

06:42.785 --> 06:44.445
Uh, and so we had a good understanding of

06:44.705 --> 06:46.405
how do you set tax rates in those setting.

06:46.945 --> 06:48.965
Uh, but what we wanted to understand as well,

06:49.315 --> 06:52.165
with those tools that are novel, how, you know,

06:52.785 --> 06:55.725
do our recommendation translate to low income countries?

06:55.725 --> 06:57.285
And Accenture? There's reasons to think

06:57.285 --> 06:58.725
that those things might be different,

06:58.885 --> 07:01.485
particularly if you think very intuitively about tax rates.

07:01.865 --> 07:04.565
The way you set tax rates might be very different in,

07:04.585 --> 07:08.085
in high income countries where you have information about,

07:08.305 --> 07:09.525
you know, the income

07:09.585 --> 07:12.965
and the assets of, of individuals versus in low income

07:13.205 --> 07:14.565
countries where you don't have information

07:14.705 --> 07:16.765
and you don't have capacity to enforce things.

07:17.145 --> 07:18.645
And so that's what we wanted to study,

07:18.785 --> 07:21.805
is those very fundamental tools, and,

07:21.825 --> 07:24.205
and how we could kind of understand,

07:24.205 --> 07:26.405
whether the policy recommendation we might give

07:26.405 --> 07:29.485
to a government would be drastically different in low income

07:29.965 --> 07:32.205
countries and low capacity settings, than

07:32.205 --> 07:33.325
what has been studied before.

07:33.595 --> 07:36.405
It's a story dear to me. I, I, I love working there.

07:36.545 --> 07:39.605
Uh, and so essentially, you know, we started all this as,

07:39.625 --> 07:41.165
as grad students,

07:41.705 --> 07:44.525
and there was a big opportunity, a big momentum.

07:44.665 --> 07:46.285
So that's more the anecdote of the paper.

07:46.285 --> 07:48.965
There was a big momentum in the, the place

07:48.965 --> 07:50.045
where we started working.

07:50.405 --> 07:52.205
'cause essentially they had,

07:52.295 --> 07:54.725
there has been an administrative split,

07:54.785 --> 07:56.725
and the province where we worked had lost a lot

07:56.725 --> 07:57.805
of natural resources.

07:57.995 --> 07:59.885
Diamond and gold went to another province.

07:59.885 --> 08:02.325
So they were, cash constrained,

08:02.385 --> 08:04.125
and they were working with the World Bank,

08:04.465 --> 08:07.685
and they talked to us about essentially kind of,

08:07.685 --> 08:10.485
working more on, on taxation, in particular,

08:10.665 --> 08:11.925
property tax.

08:11.985 --> 08:14.365
So more concretely, what we do is we,

08:14.425 --> 08:16.925
we study our questions, you know,

08:17.185 --> 08:20.165
in a particular setting that I just described, which is,

08:20.305 --> 08:22.445
cas the province of Cassai Sanal in the

08:22.445 --> 08:24.245
Democratic Republic of the Congo.

08:24.705 --> 08:27.685
Uh, and more kind of zooming in further, we work in the city

08:27.685 --> 08:31.405
of Canga, in, in that province, it's a large city.

08:31.545 --> 08:33.765
Um, it's like the fourth biggest in the country.

08:34.025 --> 08:36.365
And its population is, you know, to give you a rough sense,

08:36.395 --> 08:39.405
it's between 1.6 and 1.8 million inhabitants.

08:39.405 --> 08:42.925
So, a large kind of, a city in this context.

08:43.625 --> 08:45.925
Uh, however, it's a, you know, despite being a city,

08:45.955 --> 08:48.645
it's a context where people are relatively impoverished.

08:48.905 --> 08:52.125
Um, using our survey, what we estimate is,

08:52.125 --> 08:54.805
the median household income is about a hundred dollars,

08:54.865 --> 08:55.865
Per month. Uh, so

08:55.865 --> 08:57.125
still not very high.

08:58.115 --> 09:00.615
Uh, and, you know, kind of, as I discussed

09:00.615 --> 09:02.775
before, the government splits lost a lot

09:02.775 --> 09:05.215
of natural resources,

09:05.235 --> 09:08.135
and so as, as a result, it's, it's revenue,

09:08.505 --> 09:11.215
which is kind of its budget to provide infrastructure

09:11.235 --> 09:13.095
and public goods is, is fairly limited.

09:13.595 --> 09:17.295
Uh, we've done some work to try to estimate what's kind

09:17.295 --> 09:19.735
of the amount that the government can spend on each

09:19.735 --> 09:21.375
individual, in this context.

09:21.835 --> 09:23.895
And it's less than a dollar per year.

09:24.195 --> 09:26.935
Uh, so very low, the, the appropriate,

09:26.935 --> 09:28.695
like the similar numbers in the US would be,

09:28.695 --> 09:30.535
about $20,000 per individual.

09:30.795 --> 09:33.615
Uh, so, you know, quite a big contrast here.

09:34.195 --> 09:36.975
Um, however, that doesn't mean that the government's kind

09:36.975 --> 09:38.695
of, not trying to do better.

09:39.435 --> 09:42.775
Uh, and so in this context, what the government wants

09:42.775 --> 09:46.015
to do is, is essentially kind of revitalize some of its,

09:46.355 --> 09:47.375
tax collection effort.

09:47.635 --> 09:50.375
Uh, as I said, as a result of that administrative split

09:50.375 --> 09:52.855
where it lost resources, it kind

09:52.855 --> 09:53.855
of approached the World Bank

09:53.855 --> 09:56.935
and us to try to understand, you know, what can they do,

09:56.955 --> 09:59.415
better to, raise tax revenue.

09:59.875 --> 10:03.335
In particular, kind of their, intuition, has been

10:03.335 --> 10:05.215
to rely increasingly on the property tax.

10:05.285 --> 10:06.975
It's about a third of their revenue right now.

10:07.795 --> 10:09.455
Uh, and why the property tax?

10:09.455 --> 10:11.575
Well, you know, if you're studying from scratch

10:11.595 --> 10:13.855
and you don't have a lot of information,

10:14.055 --> 10:17.455
a really big appealing, aspect of the property tax is

10:17.455 --> 10:18.455
that it's easy to tax.

10:18.475 --> 10:19.935
You can observe properties,

10:20.395 --> 10:21.975
you can measure the characteristics.

10:22.395 --> 10:24.935
Uh, there's other reasons why property tax is appealing,

10:24.935 --> 10:27.255
but that's kind of the, big one.

10:27.715 --> 10:30.335
And so what we've done is we worked on the government,

10:30.335 --> 10:32.535
with the government, sorry, in the context

10:32.635 --> 10:36.655
of the 2018 property tax campaign, in Canga,

10:37.435 --> 10:39.975
and so that was in partnership with the tax ministry

10:39.975 --> 10:42.135
and the provincial government of Kasai Soha.

10:42.275 --> 10:43.615
And so that's, that's the context.

10:44.195 --> 10:47.975
Uh, but now kind of diving into the, like more details of,

10:47.995 --> 10:51.135
of how do we do it as scientists, and as economists.

10:51.195 --> 10:53.655
So what we did is essentially, as I mentioned

10:53.655 --> 10:55.455
before, we, we, leveraged two things.

10:55.455 --> 10:58.975
We have, we'll have administrative data on tax compliance

10:58.975 --> 11:00.775
and tax revenue that the government shared with us.

11:01.235 --> 11:02.415
Uh, and then on top of that,

11:02.415 --> 11:04.975
we implemented a randomized controlled trial.

11:05.515 --> 11:07.895
Uh, and so what, what did it consistent?

11:08.435 --> 11:10.415
It consisted in, in essentially,

11:12.085 --> 11:14.015
assigning all the properties in the city.

11:14.035 --> 11:16.535
So it was a very large scale randomized control trial.

11:16.835 --> 11:19.975
Uh, and we as we work with the government,

11:19.975 --> 11:21.655
to assign all properties in the city

11:21.675 --> 11:23.455
to different, tax rates.

11:24.125 --> 11:27.055
Okay? And so essentially some properties in the city were

11:27.255 --> 11:30.095
randomly assigned to receive the status quo tax amount.

11:30.195 --> 11:33.735
So you were assigned to kind of, what the, the,

11:33.735 --> 11:36.135
the tax amount should be, in the absence

11:36.135 --> 11:37.535
of the randomized control trial.

11:38.075 --> 11:39.415
Uh, but some other properties

11:39.415 --> 11:42.095
that were randomly selected received lower amounts.

11:42.535 --> 11:44.695
Actually, they received tax discounts.

11:45.235 --> 11:47.535
Uh, so tax reductions of different magnitude.

11:47.595 --> 11:48.855
We had three treatment groups.

11:49.475 --> 11:53.205
Uh, so either a 17% reduction in the tax amount,

11:53.205 --> 11:56.765
you face a 33% reduction or a 50% reduction.

11:57.745 --> 11:59.005
Uh, and so that,

11:59.005 --> 12:01.645
that was directly printed on the, tax letter.

12:01.745 --> 12:04.845
So it's a context where property owners receive a tax letter

12:05.115 --> 12:07.205
that features the amount that they are assigned to.

12:07.205 --> 12:09.365
And so you should think of an individual randomly selected.

12:09.575 --> 12:11.405
Again, it's a randomized control trial.

12:11.945 --> 12:14.845
And what they see on their tax letter is, you know,

12:14.845 --> 12:19.005
3000 Congolese francs versus 2,500 versus 2000

12:19.005 --> 12:20.365
versus 1,500.

12:20.745 --> 12:22.845
So that's in practice what happened,

12:22.945 --> 12:24.285
for those, property owners.

12:24.585 --> 12:26.205
And as I said, the beauty of this kind

12:26.205 --> 12:28.325
of randomized control trial is

12:28.475 --> 12:30.285
because the assignment is random.

12:30.555 --> 12:31.765
That means that individuals

12:32.145 --> 12:34.645
who receive different tax rates are on average going

12:34.645 --> 12:37.285
to be similar on all observable characteristics.

12:37.825 --> 12:39.645
The only difference between them is going to be,

12:39.705 --> 12:40.885
what tax rate they face.

12:40.945 --> 12:42.085
And so we can compare them

12:42.545 --> 12:44.845
and we'll be sure that when we compare them,

12:44.985 --> 12:47.805
the only difference, between them is going to be exactly

12:47.805 --> 12:49.405
what we varied, which is the tax rate.

12:49.425 --> 12:51.045
So we can compare these individuals

12:51.045 --> 12:53.325
and get a sense of what's the causal impact

12:53.465 --> 12:55.925
of this change in tax rate on the,

12:56.345 --> 12:58.125
outcomes we care about.

12:58.125 --> 13:00.965
And in this context, it's, you know, tax related, so

13:01.305 --> 13:03.325
who pays taxes and how much they paid.

13:03.705 --> 13:06.885
So that's, that's kind of the, the, the, the scope

13:06.885 --> 13:07.965
of, of what we did there.

13:08.355 --> 13:09.365
It's very interesting.

13:09.505 --> 13:11.925
So, you know, we were actually surprised,

13:12.065 --> 13:13.405
a little bit by the finding.

13:13.505 --> 13:15.685
So what we find is, is a few things.

13:16.065 --> 13:19.245
Um, we find that that's maybe the not

13:19.245 --> 13:21.765
so surprising aspect is we find that tax compliance is,

13:21.765 --> 13:22.765
is low in canga.

13:22.785 --> 13:26.445
So context with enforcement is limited,

13:26.625 --> 13:29.405
and there's not a lot of sanctions for not paying the tax.

13:29.505 --> 13:32.765
So, you know, on average across neighborhoods of canga,

13:32.945 --> 13:35.765
you know, about 10%, a little bit less than 10%

13:36.305 --> 13:38.285
of property owners, pay the tax.

13:38.285 --> 13:39.765
So that, that's maybe not surprising.

13:39.765 --> 13:42.245
It's on the low end, but it's relatively comparable to

13:43.025 --> 13:46.405
tax compliance and similar low capacity settings, so hight,

13:46.905 --> 13:48.405
Senegal, et cetera.

13:48.945 --> 13:50.685
So that part was not so surprising.

13:50.685 --> 13:52.405
The, the part that to us was,

13:52.665 --> 13:56.005
extent they may be more surprising is, is we find that,

13:56.245 --> 13:59.885
lowering tax rates actually generated huge responses

13:59.885 --> 14:02.925
or very large responses in terms of,

14:02.985 --> 14:04.445
tax compliance and tax revenue.

14:04.985 --> 14:09.245
Uh, and in particular when, you know, when we lower tax,

14:09.545 --> 14:11.525
rates by 33%

14:11.545 --> 14:16.085
or 50%, what we find is that the compliance response,

14:16.085 --> 14:17.965
and so essentially when we lower tax rate,

14:17.965 --> 14:19.365
we find more tax compliance.

14:19.395 --> 14:20.765
There's more people who are willing

14:20.765 --> 14:22.925
to pay at a lower amount than at a higher amount.

14:22.925 --> 14:24.445
That's, that's not the surprising part.

14:24.785 --> 14:27.005
The surprising part is the magnitude of this response.

14:27.355 --> 14:29.285
When you lower tax rate by 50%,

14:29.285 --> 14:33.245
you get a more than 50% increase in the, the percentage

14:33.245 --> 14:34.485
of, of people that pay.

14:34.825 --> 14:37.045
Uh, when you lower tax rates by 33%,

14:37.045 --> 14:40.565
you get more than 33% increase, in,

14:40.565 --> 14:42.085
compliance, in tax compliance.

14:42.585 --> 14:44.445
Uh, and so that means very drastic,

14:44.635 --> 14:46.885
very important response in terms of this kind

14:46.885 --> 14:50.525
of tax compliance margin, to tax rate, changes.

14:51.305 --> 14:54.365
Uh, and, and, and kind of if we translate that into

14:54.365 --> 14:55.605
what happens to tax revenue.

14:56.225 --> 14:58.685
So when you lower tax rates, you get two effects, right?

14:58.705 --> 15:01.165
You get, an increase in compliance.

15:01.265 --> 15:02.885
That's what we see, and it's very large,

15:03.105 --> 15:06.325
but you get a decrease in, in tax revenue for each payer

15:06.475 --> 15:08.325
because they're facing a, a lower amount.

15:08.465 --> 15:10.165
But what we find is that the first effect

15:10.405 --> 15:11.685
actually dominates the second effect.

15:11.865 --> 15:14.925
Uh, there's so many more tax co compliers at lower rates

15:15.035 --> 15:16.565
that tax revenue actually increase.

15:16.875 --> 15:19.245
Okay? So essentially what we find is, you know,

15:19.245 --> 15:23.125
in the setting, lowering tax rates, lead to,

15:23.305 --> 15:24.405
to more tax revenue.

15:24.985 --> 15:28.565
Uh, and, and so that was kind of the first result we found.

15:28.785 --> 15:32.085
And that let us, let, led us a little bit, surprised,

15:32.545 --> 15:35.245
because typically what we think is, you know,

15:35.345 --> 15:37.405
of course there is a revenue maximizing rates.

15:37.785 --> 15:42.325
And when we increase tax rates, it, past that point,

15:42.945 --> 15:44.445
we would get lower revenue.

15:44.665 --> 15:47.525
But typically in our minds, I think, as economists,

15:47.525 --> 15:49.605
it's, it's a much higher level of taxation.

15:49.625 --> 15:53.965
We think, well, of course if you tax me 60, 80%,

15:54.145 --> 15:57.485
of my income of my property, of course I'm going to respond

15:57.485 --> 15:58.845
and I'm not going to pay, et cetera.

15:58.845 --> 16:00.925
But here we find that that was the case at very low level

16:00.945 --> 16:04.405
of, of tax rates, at a much lower level of tax rate.

16:04.985 --> 16:07.085
Uh, and so that gave us pause,

16:07.145 --> 16:11.125
and then that kind of pushed us, towards trying

16:11.125 --> 16:13.365
to understand this a little bit more.

16:14.065 --> 16:16.925
And, and what we ask ourself is, well, is it

16:16.925 --> 16:18.285
because enforcement is very low?

16:18.665 --> 16:21.765
Uh, and so intuitively, you know, might think that,

16:23.135 --> 16:25.865
essentially, if the government were

16:25.865 --> 16:28.305
to increase its enforcement effort, if it were

16:28.305 --> 16:30.065
to threaten tax delinquents

16:30.485 --> 16:31.785
and to impose sanctions,

16:31.795 --> 16:34.905
maybe this large compliance respond we observe,

16:35.035 --> 16:36.945
would not, manifest itself.

16:36.945 --> 16:38.705
Essentially, if enforcement is very high,

16:39.005 --> 16:40.825
if there's sanction for not paying,

16:41.495 --> 16:42.825
when you change the tax rate,

16:42.975 --> 16:44.225
your compliance wouldn't change.

16:44.645 --> 16:47.825
Uh, people are not responding to the change in tax rate

16:47.825 --> 16:49.665
because at any kind of price point,

16:50.055 --> 16:53.545
they're equally fearing some consequence, from evading.

16:53.765 --> 16:55.425
So that's kind of what we wanted to understand.

16:55.555 --> 16:59.145
Maybe the response, we find is simply due to the low,

16:59.245 --> 17:01.145
enforcement capacity of the government.

17:01.565 --> 17:05.145
And so what we did is, as, as, economists

17:05.285 --> 17:07.305
and kind of, not scared

17:07.305 --> 17:08.985
of having done a first randomized control

17:08.985 --> 17:10.185
trial, we did a second one.

17:10.285 --> 17:14.465
Uh, and essentially, we, implemented,

17:14.465 --> 17:18.225
an additional randomized control trial where on the tax bill

17:18.225 --> 17:19.865
that the property owners received

17:19.975 --> 17:22.785
that indicated the tax liability that they faced.

17:22.805 --> 17:26.145
Uh, as I described before, we added an additional,

17:26.365 --> 17:27.745
aspect, which is,

17:27.745 --> 17:30.585
for some randomly selected property owners, they were,

17:30.695 --> 17:32.345
they were shown an enforcement message,

17:32.365 --> 17:34.825
and that message was read by the tax collector to kind of,

17:35.165 --> 17:36.705
amplify its effect.

17:37.165 --> 17:39.545
And basically that message said something like, you know,

17:39.545 --> 17:41.505
if you, if you refuse to pay the tax,

17:41.525 --> 17:44.385
you might face an audit, investigation

17:44.445 --> 17:46.745
and sanction by the provincial tax ministry.

17:47.645 --> 17:49.805
And so what do we find? Well, we find actually

17:50.035 --> 17:51.685
that our interpretation was correct.

17:51.945 --> 17:55.005
Uh, property owners who receive that enforcement message

17:55.515 --> 17:58.165
respond way less to, tax rate changes.

17:58.545 --> 18:01.485
Uh, there, we see way less response,

18:01.485 --> 18:03.605
to tax rate changes, when

18:03.745 --> 18:05.245
for individuals who receive this message.

18:05.865 --> 18:07.525
Uh, and so in other words, that means

18:07.525 --> 18:10.925
that our interpretation was correct, that indeed,

18:11.105 --> 18:13.725
the response to lower tax rate is a function of,

18:13.795 --> 18:14.965
enforcement environment.

18:15.225 --> 18:17.245
So when enforcement is very low, yes,

18:17.345 --> 18:19.845
you should maybe reduce tax rate if you can't really,

18:20.085 --> 18:21.565
increase your enforcement,

18:21.565 --> 18:23.005
because increasing tax rate will

18:23.005 --> 18:24.085
lead to a lot of delinquency.

18:24.105 --> 18:27.245
But in, if you have capacity to, increase enforcement,

18:27.715 --> 18:30.685
then you might be able to, get more revenue by,

18:30.685 --> 18:32.845
increasing tax rates rather than lowering them.

18:33.225 --> 18:37.085
Uh, and so that was kind of the, the, the gamut

18:37.085 --> 18:38.205
of, of results we found.

18:38.385 --> 18:41.125
Uh, and, and again, kind of being surprised by the first,

18:41.625 --> 18:42.925
fact, but then kind of

18:43.485 --> 18:45.605
thinking a little bit better about why

18:45.605 --> 18:46.965
that might be happening and, and,

18:46.985 --> 18:50.205
and kind of, building on that, finding answers that are,

18:50.305 --> 18:52.845
are, relevant for developing countries in general.

18:53.065 --> 18:55.205
That's a good point. So I think it truly depends a little

18:55.205 --> 18:56.965
bit on, on what the government can do,

18:57.025 --> 19:00.765
but I think what's nice is, we can kind of talk to kind

19:00.765 --> 19:02.165
of, all aspects.

19:02.825 --> 19:05.765
Uh, and so essentially what we find is, you know,

19:05.765 --> 19:07.125
our findings show that tax rates

19:07.125 --> 19:09.445
and enforcement are complimentary policy levers.

19:09.945 --> 19:14.205
So that means that you should kind of adjust tax rates

19:14.305 --> 19:15.725
and enforcement together.

19:16.225 --> 19:18.645
Um, and so our conclusion,

19:18.665 --> 19:21.085
the conclusion will depend a little bit on on what's the

19:21.285 --> 19:22.365
capacity of the government

19:22.465 --> 19:24.205
to actually change its enforcement.

19:24.825 --> 19:27.725
So in other words, if, if the government is very limited in

19:27.745 --> 19:29.245
its ability to raise enforcement,

19:29.955 --> 19:32.525
then maybe lowering tax rates is a good strategy.

19:33.065 --> 19:34.365
Um, and again, that's

19:34.365 --> 19:36.205
because if, if you lower tax rates, you'll get more revenue.

19:36.865 --> 19:39.045
Uh, but if you increase your tax rates, you'll get

19:39.065 --> 19:40.485
so much delinquent that you

19:40.565 --> 19:41.645
know you're going to lose revenue.

19:42.345 --> 19:45.365
So if you can't do anything on enforcement, then

19:45.365 --> 19:48.365
that's probably a kind of, a right solution.

19:48.555 --> 19:50.605
However, if you can do,

19:50.605 --> 19:53.245
if you can increase enforcement capacity through any means,

19:53.305 --> 19:54.965
it can be kind of the message we study.

19:54.985 --> 19:57.485
It can be recruiting, better tax, tax collector.

19:57.485 --> 19:59.925
It can be, many other things,

19:59.925 --> 20:02.205
improving tax administration,

20:02.205 --> 20:05.125
getting more information about individuals so

20:05.125 --> 20:06.485
that you can trace how much they should

20:06.505 --> 20:07.685
pay and their ability to pay.

20:07.705 --> 20:09.205
So all this, if you can do any of this,

20:09.795 --> 20:11.565
then the conclusion change.

20:11.785 --> 20:14.845
Uh, essentially what you should do is you should,

20:15.285 --> 20:16.525
increase enforcement,

20:16.545 --> 20:18.565
and you should increase tax rates once

20:18.565 --> 20:19.645
you've increased enforcement.

20:20.145 --> 20:22.765
Uh, and so that's, that's quite interesting.

20:22.765 --> 20:25.005
And again, like, I think what's nice as well is

20:25.005 --> 20:28.205
that in our study, what we did in terms of enforcement,

20:28.225 --> 20:29.885
we had two variation.

20:29.905 --> 20:32.645
The one I I mentioned, which is thisfor enforcement message,

20:32.905 --> 20:35.205
and then we did one, additional one

20:35.205 --> 20:37.485
that's more about the type of tax collectors

20:37.485 --> 20:39.005
that are recruited,

20:39.465 --> 20:42.485
and both kind of result, point in the same direction.

20:42.865 --> 20:45.285
But what's nice, in terms of having both is,

20:45.285 --> 20:46.565
we have one that's very light touch.

20:46.635 --> 20:49.085
It's just kind of threatening, sanctions

20:49.265 --> 20:50.485
and one that's more like heavy.

20:50.665 --> 20:53.445
Um, 'cause it, it it's about like, you know,

20:53.445 --> 20:54.885
recruiting better tax collector.

20:55.345 --> 20:57.325
Uh, the fact that they both work is kind of nice.

20:57.565 --> 21:00.085
'cause that means it's, you know, it doesn't mean

21:00.085 --> 21:01.725
that the government has to do drastic things

21:01.745 --> 21:03.085
to increase its enforcements.

21:03.085 --> 21:05.365
They can start by like doing small reform

21:05.365 --> 21:07.565
and then in the longer run, do things

21:07.565 --> 21:09.885
that are maybe a bit more costly and a bit more difficult.

21:10.545 --> 21:13.005
Um, so that's, that's kind of, you know,

21:13.005 --> 21:15.925
the policy implication, that we see, in our study.

21:16.145 --> 21:19.445
Um, in terms of, of kind of what does it mean

21:19.445 --> 21:21.405
for future research, I mean there's,

21:21.405 --> 21:23.005
there's many aspects,

21:24.025 --> 21:26.685
and there's things we're working on, right now that I'm,

21:26.745 --> 21:27.885
I'm quite excited about.

21:27.985 --> 21:30.645
So, essentially what we,

21:30.945 --> 21:34.165
and what we found when digging a little bit deeper in our

21:34.165 --> 21:38.725
data is that the reason why those response to tax rate are

21:38.725 --> 21:42.445
so pronounced in, in the setting is

21:42.645 --> 21:45.685
because of, property owners cash constraint.

21:45.705 --> 21:47.445
Uh, so it's a setting where people are poor,

21:47.445 --> 21:48.725
they have limited ability to pay.

21:48.725 --> 21:52.845
And what happens is some property owners only pay if the tax

21:52.845 --> 21:55.085
rates that they face is sufficiently low.

21:55.585 --> 21:57.485
Uh, and, and so this made us pause and,

21:57.485 --> 21:59.885
and wonder, you know, maybe that means that

22:00.675 --> 22:03.485
what the government should do is impose a progressivity,

22:03.565 --> 22:06.845
a progressive tax system, inject progressivity and, and,

22:06.865 --> 22:08.405
and its tax system in order

22:08.405 --> 22:10.245
to better match household's ability to pay.

22:10.785 --> 22:13.485
So we're currently working with a provincial government,

22:13.745 --> 22:16.685
on the implementation of, of such a property tax system

22:16.685 --> 22:18.845
that would be, fully progressive.

22:18.905 --> 22:20.605
So this had two steps.

22:20.905 --> 22:22.965
Uh, one that we worked on in 2020,

22:23.065 --> 22:27.405
in 2023 was flying drones in Canga to get a sense of,

22:27.745 --> 22:29.605
of the features of, of properties.

22:29.605 --> 22:31.245
So that's kind of a, a fun aspect.

22:31.745 --> 22:34.525
And then we work with the government tax agents to kind of,

22:34.525 --> 22:36.645
they did a survey of all properties in the city

22:36.705 --> 22:39.325
to compliment the drone imagery with,

22:39.325 --> 22:40.645
survey features of the property.

22:41.345 --> 22:44.085
And the goal is going to be to essentially,

22:44.085 --> 22:45.685
predict the property values,

22:45.685 --> 22:48.565
of all properties in the city using a mass appraisal type

22:48.565 --> 22:49.845
of model,

22:49.985 --> 22:53.045
and then, randomly roll out progressive taxation,

22:53.425 --> 22:54.525
in, in the city.

22:54.705 --> 22:55.845
So in the first year, we'll be able

22:55.845 --> 22:58.245
to compare tax compliance, tax revenue,

22:58.345 --> 23:01.045
and many other things in neighborhoods that face,

23:01.325 --> 23:02.765
a flat tax system or a not

23:02.765 --> 23:05.725
so progressive tax system versus a neighborhoods where,

23:05.825 --> 23:08.205
there is a much more progressive, tax system.

23:08.625 --> 23:10.565
Uh, and so that's, that's kind of, you know,

23:10.565 --> 23:14.365
something we're very excited to see, for two reasons.

23:14.585 --> 23:17.525
One is, you know, there's a lot of, there's a lot

23:17.525 --> 23:18.765
of emphasis on the importance

23:18.765 --> 23:21.085
of progressivity across the board in,

23:21.145 --> 23:24.765
in explaining the emergence of, of tax states and,

23:24.785 --> 23:26.405
and taxation in the 20th century.

23:26.865 --> 23:28.885
Uh, but there's not been an effort to,

23:28.885 --> 23:31.045
implement a randomized control trial on,

23:31.325 --> 23:32.565
progressivity of, of tax system.

23:32.565 --> 23:35.805
So, that's, that's kind of, why we think it's,

23:35.805 --> 23:36.965
it's exciting,

23:37.065 --> 23:38.805
and hopefully we'll have the results,

23:39.065 --> 23:40.365
by the end of 2024.

23:40.465 --> 23:42.645
So maybe we should do another, episode

23:42.645 --> 23:43.605
and talk about that paper.

00:00 - Essentially the, the, the motivation for kind of my overall research comes from two main facts. the first one is that, you know, fiscal capacity has been shown to be essential for economic development. and, and typically

00:13 - because government need tax revenue very intuitively to provide public service or infrastructure, so road hospital schools, that are necessary for the development of the population. but despite this, what we see across the world is that, low income countries typically have very limited tax revenue. they collect, you know, if you wanna put number on it, they collect only 10% of their GDP and taxes. and if you want to compare it with rich countries,

00:38 - that number would be about 40%. and so, you know, there's, that's, that's kind of the status quo. and then you asked me, what, what can governments do to increase the tax, the revenue that they collect? Well, there's many things that they can do. and essentially what my co-authors and I, so my co-authors are Gabe to rec at the University of Pittsburgh and John Vial at uc, Berkeley. what we decided to do,

01:03 - we're all PhD student when we started this series of project, was to focus on the most fundamental tools you can think of that government have access, to, to raise, revenue. And those are essentially tax rates and tax enforcements. And so what we ask in this, in this project is essentially two things, is whether tax revenue in low capacity or low income countries are low because tax rates are not set optimally.

01:29 - so maybe, and you know, there's two things that could go be going on here. Maybe tax rates are set too low. Countries are not ambitious enough. They set low tax rate, and so they generate too little tax revenue per taxpayer. But you could imagine that actually the opposite is true. Maybe tax rates are actually set too high by those governments, and

01:48 - that generates too few taxpayers in the first place. Okay. So that's the first question we're interested in. and then the second one is more to deal with the second tool with tax enforcement. And what we're trying to understand is maybe how government should set their tax rate is actually going to depend on the government's capacity to enforce taxes. So in their enforcement capacity, their capacity to detect, tax delinquent, their capacities to sanction, those were, tax delinquent.

02:15 - And, and this is very intuitive, intuitively, you know, if, if enforcement is, is very low, then maybe if you raise tax rates, you're just going to have more tax delinquents. that's how people respond to tax rate changes. and if that's the case, maybe that won't affect tax revenue. You get some more revenue, but you get some more tax delinquent. But it could even need to lower tax revenue if there's

02:37 - really a lot of delinquents when you raise your tax rate. So that's, that's kind of what might happen if tax enforcement is low. but by contrast, if you were to increase tax enforcement, maybe suddenly things would look different. if you increase tax enforcement and raise your tax rates, then maybe you won't have this delinquency response to higher rates.

02:56 - and so you want to have a change In tax compliance. All you'll get is that some people will pay more taxes, and so then you'll get more tax revenue. And so we wanted to understand kind of both questions. how should you set tax rates, and, and in particular, how should you set tax rates when enforcement is low versus when enforcement is higher? An interesting moment for, I think, the type of studies my coauthor and I do,

03:20 - and we really build, like we're on the shoulder of giants. We build on a fantastic kind of trend that has happened in the past 10 years in development economics, in the field, in its entirety with, there's been more and more focus on studying, taxation in low income countries in the past 10 years. and this is for two reasons. The, those countries, have been more willing to share administrative data.

03:45 - And so researchers have had more access to tax data. So information that comes directly from the tax administration rather than relying on surveys where you can ask people whether they paid tax or not, but that's not going to be perfect. It's much better if you have the actual information from the tax administration about who paid, and what amount. So that's the first thing, that has been happening in this literature. and the second one is governments have been more willing

04:10 - to, experiment and to work closely with researchers on trying to understand what can they do to improve their tax capacity and their fiscal capacity. And in particular, and our study is an example of that, governments have been increasingly willing to rely on randomized control trials. so this is a method that's in part from, the literature in medicine that consistent in studying the impact of a policy

04:34 - by assigning it randomly. So you have a control group that receives the status quo, and then treatment groups that receive an intervention. And the decision between to allocate an individual or, a neighborhoods or village to each group is, is random. so we can compare those two groups to understand the impact of the policy we're interested in. And so, governments, in low income countries have been increasingly willing

04:58 - to experiment on taxation using those kind of randomized control trials. And so more concretely, you know, what, what does it mean in terms of studies? There's been kind of a lot of studies that have been very influential for, my co-authors and I, and we probably wouldn't be, doing the type of things we, we do without them. so for example, Dina Pomerance, who's at the University of Zurich, kind of pioneered this literature.

05:20 - She did an amazing kind of study, using both of those kind of modern things, the administrative tax data and the randomized controlled trial to understand the role of enforcement in the context of, the value added tax in, Chile. but there's more and more of those studies, for example, an Brooke Meyer and, and Juan Carlos Serato. I've, I've done excellent work on trying to understand, property taxation in Mexico.

05:49 - and more recently, someone who's just on the market this year, Justine Belman, also did very impressive work in Senegal on trying to understand how do we assess properties? How do we assess values in order to tax them? so I think it's very exciting time for this literature. and, and so that's where we, we blend in. But, but actually I think what's interesting is even though it's a, it's a kind of booming moment and exciting moment for this literature, there's still a lot

06:15 - of questions that are quite fundamental that have not yet been answered. and that's, I think, where we try to make a contribution. So when you think of like questions that are very fundamental, such as how do you set tax rates? How do you enforce access? It's not that we were in kind of unchartered territory, but it's mostly that those questions have been typically, studied by public economists that focused on kind

06:39 - of high income countries, so the us, Europe, et cetera. and so we had a good understanding of how do you set tax rates in those setting. but what we wanted to understand as well, with those tools that are novel, how, you know, do our recommendation translate to low income countries? And Accenture? There's reasons to think that those things might be different, particularly if you think very intuitively about tax rates. The way you set tax rates might be very different in,

07:04 - in high income countries where you have information about, you know, the income and the assets of, of individuals versus in low income countries where you don't have information and you don't have capacity to enforce things. And so that's what we wanted to study, is those very fundamental tools, and, and how we could kind of understand, whether the policy recommendation we might give to a government would be drastically different in low income

07:29 - countries and low capacity settings, than what has been studied before. It's a story dear to me. I, I, I love working there. and so essentially, you know, we started all this as, as grad students, and there was a big opportunity, a big momentum. So that's more the anecdote of the paper. There was a big momentum in the, the place where we started working. 'cause essentially they had,

07:52 - there has been an administrative split, and the province where we worked had lost a lot of natural resources. Diamond and gold went to another province. So they were, cash constrained, and they were working with the World Bank, and they talked to us about essentially kind of, working more on, on taxation, in particular, property tax. So more concretely, what we do is we,

08:14 - we study our questions, you know, in a particular setting that I just described, which is, cas the province of Cassai Sanal in the Democratic Republic of the Congo. and more kind of zooming in further, we work in the city of Canga, in, in that province, it's a large city. it's like the fourth biggest in the country. And its population is, you know, to give you a rough sense, it's between 1.6 and 1.8 million inhabitants. So, a large kind of, a city in this context.

08:43 - however, it's a, you know, despite being a city, it's a context where people are relatively impoverished. using our survey, what we estimate is, the median household income is about a hundred dollars, Per month. so still not very high. and, you know, kind of, as I discussed before, the government splits lost a lot of natural resources, and so as, as a result, it's, it's revenue,

09:08 - which is kind of its budget to provide infrastructure and public goods is, is fairly limited. we've done some work to try to estimate what's kind of the amount that the government can spend on each individual, in this context. And it's less than a dollar per year. so very low, the, the appropriate, like the similar numbers in the US would be, about $20,000 per individual. so, you know, quite a big contrast here.

09:34 - however, that doesn't mean that the government's kind of, not trying to do better. and so in this context, what the government wants to do is, is essentially kind of revitalize some of its, tax collection effort. as I said, as a result of that administrative split where it lost resources, it kind of approached the World Bank and us to try to understand, you know, what can they do, better to, raise tax revenue.

09:59 - In particular, kind of their, intuition, has been to rely increasingly on the property tax. It's about a third of their revenue right now. and why the property tax? Well, you know, if you're studying from scratch and you don't have a lot of information, a really big appealing, aspect of the property tax is that it's easy to tax. You can observe properties, you can measure the characteristics.

10:22 - there's other reasons why property tax is appealing, but that's kind of the, big one. And so what we've done is we worked on the government, with the government, sorry, in the context of the 2018 property tax campaign, in Canga, and so that was in partnership with the tax ministry and the provincial government of Kasai Soha. And so that's, that's the context. but now kind of diving into the, like more details of, of how do we do it as scientists, and as economists.

10:51 - So what we did is essentially, as I mentioned before, we, we, leveraged two things. We have, we'll have administrative data on tax compliance and tax revenue that the government shared with us. and then on top of that, we implemented a randomized controlled trial. and so what, what did it consistent? It consisted in, in essentially, assigning all the properties in the city. So it was a very large scale randomized control trial.

11:16 - and we as we work with the government, to assign all properties in the city to different, tax rates. Okay? And so essentially some properties in the city were randomly assigned to receive the status quo tax amount. So you were assigned to kind of, what the, the, the tax amount should be, in the absence of the randomized control trial. but some other properties that were randomly selected received lower amounts.

11:42 - Actually, they received tax discounts. so tax reductions of different magnitude. We had three treatment groups. so either a 17% reduction in the tax amount, you face a 33% reduction or a 50% reduction. and so that, that was directly printed on the, tax letter. So it's a context where property owners receive a tax letter that features the amount that they are assigned to. And so you should think of an individual randomly selected.

12:09 - Again, it's a randomized control trial. And what they see on their tax letter is, you know, 3000 Congolese francs versus 2,500 versus 2000 versus 1,500. So that's in practice what happened, for those, property owners. And as I said, the beauty of this kind of randomized control trial is because the assignment is random. That means that individuals

12:32 - who receive different tax rates are on average going to be similar on all observable characteristics. The only difference between them is going to be, what tax rate they face. And so we can compare them and we'll be sure that when we compare them, the only difference, between them is going to be exactly what we varied, which is the tax rate. So we can compare these individuals and get a sense of what's the causal impact

12:53 - of this change in tax rate on the, outcomes we care about. And in this context, it's, you know, tax related, so who pays taxes and how much they paid. So that's, that's kind of the, the, the, the scope of, of what we did there. It's very interesting. So, you know, we were actually surprised, a little bit by the finding. So what we find is, is a few things.

13:16 - we find that that's maybe the not so surprising aspect is we find that tax compliance is, is low in canga. So context with enforcement is limited, and there's not a lot of sanctions for not paying the tax. So, you know, on average across neighborhoods of canga, you know, about 10%, a little bit less than 10% of property owners, pay the tax. So that, that's maybe not surprising. It's on the low end, but it's relatively comparable to

13:43 - tax compliance and similar low capacity settings, so hight, Senegal, et cetera. So that part was not so surprising. The, the part that to us was, extent they may be more surprising is, is we find that, lowering tax rates actually generated huge responses or very large responses in terms of, tax compliance and tax revenue. and in particular when, you know, when we lower tax, rates by 33%

14:11 - or 50%, what we find is that the compliance response, and so essentially when we lower tax rate, we find more tax compliance. There's more people who are willing to pay at a lower amount than at a higher amount. That's, that's not the surprising part. The surprising part is the magnitude of this response. When you lower tax rate by 50%, you get a more than 50% increase in the, the percentage of, of people that pay.

14:34 - when you lower tax rates by 33%, you get more than 33% increase, in, compliance, in tax compliance. and so that means very drastic, very important response in terms of this kind of tax compliance margin, to tax rate, changes. and, and, and kind of if we translate that into what happens to tax revenue. So when you lower tax rates, you get two effects, right? You get, an increase in compliance.

15:01 - That's what we see, and it's very large, but you get a decrease in, in tax revenue for each payer because they're facing a, a lower amount. But what we find is that the first effect actually dominates the second effect. there's so many more tax co compliers at lower rates that tax revenue actually increase. Okay? So essentially what we find is, you know, in the setting, lowering tax rates, lead to, to more tax revenue.

15:24 - and, and so that was kind of the first result we found. And that let us, let, led us a little bit, surprised, because typically what we think is, you know, of course there is a revenue maximizing rates. And when we increase tax rates, it, past that point, we would get lower revenue. But typically in our minds, I think, as economists, it's, it's a much higher level of taxation. We think, well, of course if you tax me 60, 80%, of my income of my property, of course I'm going to respond

15:57 - and I'm not going to pay, et cetera. But here we find that that was the case at very low level of, of tax rates, at a much lower level of tax rate. and so that gave us pause, and then that kind of pushed us, towards trying to understand this a little bit more. And, and what we ask ourself is, well, is it because enforcement is very low? and so intuitively, you know, might think that, essentially, if the government were

16:25 - to increase its enforcement effort, if it were to threaten tax delinquents and to impose sanctions, maybe this large compliance respond we observe, would not, manifest itself. Essentially, if enforcement is very high, if there's sanction for not paying, when you change the tax rate, your compliance wouldn't change. people are not responding to the change in tax rate

16:47 - because at any kind of price point, they're equally fearing some consequence, from evading. So that's kind of what we wanted to understand. Maybe the response, we find is simply due to the low, enforcement capacity of the government. And so what we did is, as, as, economists and kind of, not scared of having done a first randomized control trial, we did a second one. and essentially, we, implemented,

17:14 - an additional randomized control trial where on the tax bill that the property owners received that indicated the tax liability that they faced. as I described before, we added an additional, aspect, which is, for some randomly selected property owners, they were, they were shown an enforcement message, and that message was read by the tax collector to kind of, amplify its effect. And basically that message said something like, you know,

17:39 - if you, if you refuse to pay the tax, you might face an audit, investigation and sanction by the provincial tax ministry. And so what do we find? Well, we find actually that our interpretation was correct. property owners who receive that enforcement message respond way less to, tax rate changes. there, we see way less response, to tax rate changes, when for individuals who receive this message.

18:05 - and so in other words, that means that our interpretation was correct, that indeed, the response to lower tax rate is a function of, enforcement environment. So when enforcement is very low, yes, you should maybe reduce tax rate if you can't really, increase your enforcement, because increasing tax rate will lead to a lot of delinquency. But in, if you have capacity to, increase enforcement,

18:27 - then you might be able to, get more revenue by, increasing tax rates rather than lowering them. and so that was kind of the, the, the gamut of, of results we found. and, and again, kind of being surprised by the first, fact, but then kind of thinking a little bit better about why that might be happening and, and, and kind of, building on that, finding answers that are, are, relevant for developing countries in general.

18:53 - That's a good point. So I think it truly depends a little bit on, on what the government can do, but I think what's nice is, we can kind of talk to kind of, all aspects. and so essentially what we find is, you know, our findings show that tax rates and enforcement are complimentary policy levers. So that means that you should kind of adjust tax rates and enforcement together. and so our conclusion,

19:18 - the conclusion will depend a little bit on on what's the capacity of the government to actually change its enforcement. So in other words, if, if the government is very limited in its ability to raise enforcement, then maybe lowering tax rates is a good strategy. and again, that's because if, if you lower tax rates, you'll get more revenue. but if you increase your tax rates, you'll get so much delinquent that you

19:40 - know you're going to lose revenue. So if you can't do anything on enforcement, then that's probably a kind of, a right solution. However, if you can do, if you can increase enforcement capacity through any means, it can be kind of the message we study. It can be recruiting, better tax, tax collector. It can be, many other things, improving tax administration, getting more information about individuals so

20:05 - that you can trace how much they should pay and their ability to pay. So all this, if you can do any of this, then the conclusion change. essentially what you should do is you should, increase enforcement, and you should increase tax rates once you've increased enforcement. and so that's, that's quite interesting. And again, like, I think what's nice as well is

20:25 - that in our study, what we did in terms of enforcement, we had two variation. The one I I mentioned, which is thisfor enforcement message, and then we did one, additional one that's more about the type of tax collectors that are recruited, and both kind of result, point in the same direction. But what's nice, in terms of having both is, we have one that's very light touch. It's just kind of threatening, sanctions

20:49 - and one that's more like heavy. 'cause it, it it's about like, you know, recruiting better tax collector. the fact that they both work is kind of nice. 'cause that means it's, you know, it doesn't mean that the government has to do drastic things to increase its enforcements. They can start by like doing small reform and then in the longer run, do things that are maybe a bit more costly and a bit more difficult.

21:10 - so that's, that's kind of, you know, the policy implication, that we see, in our study. in terms of, of kind of what does it mean for future research, I mean there's, there's many aspects, and there's things we're working on, right now that I'm, I'm quite excited about. So, essentially what we, and what we found when digging a little bit deeper in our data is that the reason why those response to tax rate are

21:38 - so pronounced in, in the setting is because of, property owners cash constraint. so it's a setting where people are poor, they have limited ability to pay. And what happens is some property owners only pay if the tax rates that they face is sufficiently low. and, and so this made us pause and, and wonder, you know, maybe that means that what the government should do is impose a progressivity, a progressive tax system, inject progressivity and, and,

22:06 - and its tax system in order to better match household's ability to pay. So we're currently working with a provincial government, on the implementation of, of such a property tax system that would be, fully progressive. So this had two steps. one that we worked on in 2020, in 2023 was flying drones in Canga to get a sense of, of the features of, of properties. So that's kind of a, a fun aspect.

22:31 - And then we work with the government tax agents to kind of, they did a survey of all properties in the city to compliment the drone imagery with, survey features of the property. And the goal is going to be to essentially, predict the property values, of all properties in the city using a mass appraisal type of model, and then, randomly roll out progressive taxation, in, in the city.

22:54 - So in the first year, we'll be able to compare tax compliance, tax revenue, and many other things in neighborhoods that face, a flat tax system or a not so progressive tax system versus a neighborhoods where, there is a much more progressive, tax system. and so that's, that's kind of, you know, something we're very excited to see, for two reasons. One is, you know, there's a lot of, there's a lot of emphasis on the importance

23:18 - of progressivity across the board in, in explaining the emergence of, of tax states and, and taxation in the 20th century. but there's not been an effort to, implement a randomized control trial on, progressivity of, of tax system. So, that's, that's kind of, why we think it's, it's exciting, and hopefully we'll have the results, by the end of 2024.

23:40 - So maybe we should do another, episode and talk about that paper.

Faculti AI can make mistakes