Banks in Tax Havens: Evidence from the 2013-2019 Country-by-Country Reporting Data

• Název práce v češtině: Banky v daňových rájích: Výsledky na základě dat z podávání zpráv podle zemí mezi roky 2013 a 2019
• Název v anglickém jazyce: Banks in Tax Havens: Evidence from the 2013-2019 Country-by-Country Reporting Data
• Klíčová slova anglicky: country-by-country reporting, banks, tax havens, gravity model, Poisson pseudo-maximum likelihood estimator
• Akademický rok vypsání: 2017/2018
• Typ práce: bakalářská práce
• Jazyk práce: angličtina
• Ústav: Institut ekonomických studií (23-IES)
• Vedoucí / školitel: doc. Petr Janský, M.Sc., Ph.D.
• Řešitel: skrytý - zadáno vedoucím/školitelem
• Datum přihlášení: 30.05.2019
• Datum zadání: 30.05.2019
• Datum a čas obhajoby: 09.09.2020 09:00
• Místo konání obhajoby: Opletalova - Opletalova 26, O206, Opletalova - místn. č. 206
• Datum odevzdání elektronické podoby: 27.07.2020
• Datum proběhlé obhajoby: 09.09.2020
• Oponenti: Sarah Godar, M.A., Ph.D.

Seznam odborné literatury

List of academic literature:
Bibliography
• Vincent Bouvatier, Gunther Capelle-Blancard & Anne-Laure Delatte (2017) Banks in Tax Havens: First Evidence based on Country-by-Country Reporting
• doc. Petr Janský Ph.D (2018) European Banks and Tax Havens: Evidence from Country-by-Country Reporting
• OECD (2018) Country-by-Country Reporting – Compilation of Peer Review Reports (Phase 1)
• Witada Anukoonwattaka (2016) Introduction to the basic gravity model
• OECD (2016) Tax transparency and country by country reporting BEPS and beyond

Předběžná náplň práce v anglickém jazyce

Preliminary scope of work:
Research question and motivation
Motivation: In 2013 the Organization for Economic Co-operation and Development (OECD) introduced the OECD/G20 BEPS project to address Base erosion and profit shifting (BEPS). BEPS refers to tax avoidance where multinational companies use different strategies to shift their profit from higher-tax jurisdictions to lower-tax jurisdictions. Example of those strategies is Treaty Shopping, Special Purpose Entities, Transfer Pricing etc. The project is divided into fifteen “actions” that address different problems. I will specifically be interested in action number 13 which addresses transfer pricing. In order to increase transparency of multinational enterprises, action 13 obliges them to provide Country-by-country report (CBCR) about allocation of profit, sales, employees, assets and where taxes are paid.

Research question: In my thesis I will collect data from CBC report for European banks and I will analyze them. I will analyze whether banks tend to report profit in tax havens. I will mainly use papers on the same topic: one from P. Janský (2018), second from Bouvatier et al. (2017). Those papers analyzed CBCR data for years 2013-2017, my goal is to extend this literature for years 2018 and 2019. As was proceeded in paper from Bouvatier et al. I will estimate standard gravity model showing relation of turnover in partner countries where branch offices are situated and GDP per capita in partner country, population in partner country, distance between the two countries and dummy variables indicating whether the two countries have a common border, a common language, a colonial or territorial relationship and whether they signed agreement about regional trade. This way I will have controlled for all important standard factors influencing turnover of each partner country (a partner country is such country where branch office of an institution, with headquarters in another country, is located). To this model I will add additional dummy variable indicating whether the country is considered as tax haven or not. I will estimate the new model and compare it to the previous one, this way I will estimate the effect of country being a tax haven.
Subsequently, I will try to break down the effect of tax haven into four categories that should drive foreign activities of banks (hence, driving turnover specified above):
• Clearly, first factor is tax in partner countries, this should have negative effect as lower tax clearly attracts new investment.
• Financial secrecy may tend to positively affect the dependent variable (turnover) because it enables individuals or corporates to escape from laws and rules that are effective in their headquarter location. I will use AML index (Anti-Money Laundering index, which is also used in Bouvatier et al. paper) to control for financial secrecy.
• Level of regulations should be also included as higher level of regulations might repel investors.
• Last variable I will put into consideration is quality of governance controlled by Worldwide governance indicator (WGI) made by Kaufman et al. (2011).

I would like to highlight that this process was done in paper by Bouvatier et al. This paper however covers data only up to 2017. My goal is to evaluate data for 2018 and 2019.
Furthermore, I will try to prove the following hypotheses:
1. The effect of being tax haven is higher number of branch offices in the region.
2. The bigger and more profitable the bank is (in terms of number of workers, branch offices, sales, profits) the more likely it is that the bank uses tax avoidance strategies and profit shifting.
More information about hypotheses and their testing is located in “methodology”.


Banks are institutions where majority of people deposit their money, many of them have more than one branch offices in several countries. They may report few or no worker in one branch office and lots of them in another one, they can shift their profits from one branch to another one and so on, thus it is important to monitor their economic activity, profits, taxes they pay, number of workers etc. This is why I think that this topic is interesting and important.

Contribution
As far as I am concerned there is yet no interpretation of CBCR data for years 2018 and 2019. Thus, my contribution will be to the literature about CBCR data for years 2018 and 2019. Data for year 2019 are not available but I will include them in my thesis when they are released.


Methodology
As I said before, I will use country-by-country data reported by European banks, I will put them together and by excel I will prepare them for analyzing. To interpret the data I collect, I will use econometrics. I will start with standard gravity equation:
Turnover(k,i,j) = C* GDP(j)^b(1)*Population(j)^b(2)*v(k,I,j)/Distance(I,j)^b(3)


Where k is bank k, i a is country with headquarters and j is a partner country. Therefore, Turnover(k,i,1) refers to turnover reported by bank k, with headquarters in country i, in country j. From this we can get: Turnover(k,i,j) = exp[b(0) + b(1)log(GDP(j)) + b(2)log(Poupulation(j)) + b(3)log(Distance(i,j))]+ u(k,i,j)
Finally, I will add dummy variables which I introduced before and my base gravity equation will look like this: Turnover(k,i,j) = exp[b(0) + b(1)log(GDP(j)) + b(3)log(Distance(i,j)) + b(4)Comboard(i,j) + b(5)Comlang(i,j) + b(6)Colony(i,j) + b(7)Territory(i,j) + b(8)Agreement(i,j)] + u(k,i,j)

Using Poisson pseudo-maximum likelihood estimator (PPML) I will estimate this model and analyze estimates in terms of testing whether the estimates are significant. Subsequently I will construct new model that includes dummy variable about country being tax haven estimate the model again (using PPML) then I will compare the two models. Essentially, I will observe the sign of effect of tax haven and its size, I will test whether the value is significant using t-test, by this test the first hypothesis (formulated in “Research question”) is tested because if the estimate will be proved to be significant and positive (as I expect it to be) it means that the fact that a country is tax haven leads to (ceteris paribus) increase in foreign activity in this country, hence larger number of banks operate in the country. From this analysis I will make first statements (about effect of tax haven on foreign activities of banks).
Afterwards, as described in “Research question”, I will include the four new variables (about tax, financial secrecy, regulations and quality of governance) to conclude which features of tax havens are the most crucial.
For the second hypothesis I will estimate OLS model: The independent variables will be number of workers, number of branch offices, sales and profit before tax (as indicators of size and profitability of the bank). The dependent variable will be dummy variable (of value 0 or 1) indicating whether bank report profit in tax haven (represented by 1) or not (represented by 0). The expectation of such a variable will be the probability (given all independent variables) that a bank reports its profit in tax haven. Then I will test if parameters of the model are significant, using t-test, or jointly significant, using F-test or LM-test. I expect that the effect of the independent variables will be positive, but the explanation might be ambiguous because the model could be wrong (in terms of omitting important variables).

Outline
1. Introduction
2. Theoretical background
2.1 Introduction of papers from this field, introducing their results
2.2 Introducing data for previous years
2.3 Literature review
3. Methodology
4. Presenting data
5. Data analyzing
6. Discussing results
7. Conclusion