Anni Tilli 

Curbing Abnormal Trading: The Effectiveness of 
TRACE in Combating Withholding Tax Schemes in 

Finland 
 

 

 

 

 

 

 
  

Vaasa 2025 

School of Accounting and Finance 
Master’s Thesis in Finance  

Master’s Degree Programme in Finance 



2 

UNIVERSITY OF VAASA 
School of Accounting and Finance 
Author:    Anni Tilli 
Title of the Thesis:  Curbing Abnormal Trading: The Effectiveness of TRACE in Com-
bating Withholding Tax Schemes in Finland 
Degree:    Master’s Degree in Finance 
Programme:   Finance 
Supervisor:   Tatiana King 
Year:    2025 Number of Pages: 81 

ABSTRACT: 
Dividend withholding tax (DWT) schemes, such as cum-cum and cum-ex, have exploited regula-
tory loopholes across Europe, leading to significant tax revenue losses. In response to these is-
sues, Finland implemented the TRACE policy to combat DWT schemes. This thesis examines the 
effectiveness of TRACE by investigating its impact on trading volumes around ex-dividend dates 
and assessing whether the policy has successfully reduced abnormal trading activity linked to 
these schemes. Taxation has long faced challenges adapting to situations where legal and eco-
nomic ownership are not fully aligned, such as in the cum-cum and cum-ex schemes mentioned 
above. 
 
Using an event study methodology, this thesis examines daily trading volume data from 2013 to 
2023 of 100 of the most actively traded dividend-paying companies in Finland. This thesis fo-
cuses on two big events, a share's ex-dividend date and the introduction of TRACE. Abnormal 
trading volumes are calculated by contrasting trading activity within event windows with base-
line activity external to these windows. Statistical methods, including t-tests, are employed to 
estimate the success of TRACE at lowering abnormal trading volumes. 
 
Consistent with past studies and data of DWT schemes, this thesis's results show notable abnor-
mal trading volumes around ex-dividend dates in the pre-policy period. A significant decline in 
abnormal trading volumes following the TRACE policy’s implementation indicates the policy's 
effectiveness. TRACE effectively targeted DWT schemes and other abusive behaviours without 
interfering with or interrupting regular market activity, as seen by the stable trading volumes 
outside the event window.   
 
This thesis highlights the importance of TRACE in reducing DWT schemes in Finland and supports 
findings from similar reforms in countries like Germany and Denmark. These results highlight 
the possibility for targeted regulatory interventions to reduce aggressive tax planning and im-
prove the integrity of tax systems. These findings also further promote the importance of TRACE 
as a regulatory tool for addressing DWT schemes and support the broader adoption of similar 
measures, such as the FASTER directive, across Europe. Lastly, this thesis contributes to the on-
going efforts to create a more transparent and equitable financial system by offering new in-
sights into the effectiveness of targeted tax reforms.  
 

KEYWORDS: Dividend withholding scheme, Withholding tax, Dividend withholding tax, Cum-
ex, Cum-cum, Dividend taxation, Tax planning 

  



3 

VAASAN YLIOPISTO 
School of Accounting and Finance  
Tekijä:    Anni Tilli 
Tutkielman nimi:  Curbing Abnormal Trading: The Effectiveness of TRACE in Com-
bating Withholding Tax Schemes in Finland 
Tutkinto:    Master’s Degree in Finance 
Oppiaine:   Finance 
Työn ohjaaja:   Tatiana King 
Vuosi:    2025 Sivumäärä: 81 

TIIVISTELMÄ: 
Osinkojen lähdeverojärjestelmiin liittyvät järjestelyt, kuten cum-cum ja cum-ex, ovat merkittävä 
haaste Euroopan verojärjestelmille, sillä ne hyödyntävät sääntelyyn liittyviä heikkouksia ja ai-
heuttavat merkittäviä verotulojen menetyksiä. Nämä järjestelyt hyödyntävät lainsäädännön 
porsaanreikiä ja luovat keinotekoisia rakenteita, joiden avulla verovelvolliset voivat saada esi-
merkiksi perusteettomia veronpalautuksia. Verotus on jo pitkään kohdannut haasteita mukau-
tua tilanteisiin, joissa oikeudellinen ja taloudellinen omistus eivät ole täysin linjassa keskenään, 
kuten näissä edellä mainituissa cum-cum ja cum-ex järjestelyissä. 
 
Tässä pro gradu -tutkielmassa tarkastellaan Suomessa vuonna 2021 käyttöön otetun TRACE-di-
rektiivin vaikutuksia näiden järjestelyiden torjumiseksi. Tämän tutkielman tavoitteena on selvit-
tää, vaikuttaako TRACE osakkeiden kaupankäyntivolyymeihin osingon irtoamispäivän ympärillä 
ja onko poikkeuksellista kaupankäyntiä havaittavissa ennen ja jälkeen TRACE-direktiivin käyt-
töönoton.  
 
Tämä tutkielma perustuu tapahtumatutkimukseen ja analysoi päivittäisiä kaupankäyntivolyy-
meja vuosilta 2013–2023. Tutkielmassa tarkastellaan 100 aktiivisimmin kaupankäydyn ja osin-
koja maksavan suomalaisen yhtiön dataa. Pro gradu -tutkielman keskiössä ovat kaksi keskeistä 
tapahtumaa, jotka ovat osakkeen irtoamispäivä sekä TRACE:n käyttöönotto Suomessa. Tilastol-
lisina menetelminä käytetään muun muassa t-testiä, joka auttaa mittaamaan TRACE:n tehok-
kuuden epänormaalin kaupankäynnin vähentämisessä.  
 
Tulokset osoittavat, että ennen TRACE:n käyttöönottoa osingon irtoamispäivän ympärillä esiin-
tyi merkittävää, tavallisesta poikkeavan suurta kaupankäyntiä, mikä viittaa osinkojen lähdevero-
petoksien esiintyvyyteen. TRACE:n käyttöönoton jälkeen tavallisesta poikkeavan suuri kaupan-
käynti väheni huomattavasti, mikä osoittaa TRACE:n onnistuneen osinkojen lähdeveropetoksien 
torjumisessa. Samalla kaupankäyntivolyymit pysyivät vakaana tapahtumaikkunan ulkopuolella, 
mikä viittaa siihen, että TRACE kohdistui nimenomaan osinkojen lähdeveropetoksiin vaikutta-
matta normaaliin markkinatoimintaan.  
 
Tämän tutkielman tulokset korostavat TRACE:n merkitystä tehokkaana sääntelykeinona osinko-
jen lähdeveropetoksien torjunnassa ja tukevat vastaavien toimien, kuten FASTER-direktiivin, laa-
jempaa käyttöönottoa koko Euroopassa. Tämä tukee aiempia havaintoja tutkimuksissa, joissa 
vastaavat lainsäädännölliset uudistukset Saksassa ja Tanskassa ovat osoittautuneet tehokkaiksi. 
Tulokset vahvistavat, että kohdennetut verouudistukset voivat vähentää aggressiivista vero-
suunnittelua ja parantaa verotusjärjestelmien läpinäkyvyyttä. Tämä tutkielma tarjoaa myös uu-
sia näkökulmia kohdennettujen verouudistusten tehokkuudesta. 

AVAINSANAT: Dividend withholding scheme, Withholding tax, Dividend withholding tax, 
Cum-ex, Cum-cum, Dividend taxation, Tax planning 

  



4 

 

Contents 

1 Introduction 7 

1.1 Background 7 

1.2 Literature review 9 

1.3 Purpose of the study 10 

1.4 Research question and hypothesis 12 

1.5 Structure of the Study 15 

2 Theoretical framework 17 

2.1 The differences between tax planning and tax evasion 17 

2.2 Dividend taxation 19 

2.2.1 Using withholding taxes to combat tax avoidance 20 

2.2.2 Withholding tax refunds and relief at source 21 

2.3 Exploitation of tax treaties 23 

2.4 Trade settlement process 24 

2.5 Ex-date and record date 25 

2.6 Trading spikes near the ex-dividend date 26 

3 Different dividend withholding tax schemes 29 

3.1 Cum-cum 29 

3.2 Cum-ex 34 

3.2.1 Cum-ex in Germany 35 

3.2.2 Evidence of cum-ex in the rest of Europe 38 

4 Research approach 43 

4.1 The data collection 43 

4.1.1 The data samples 43 

4.2 Statistical analysis 46 

5 Results 49 

5.1 Key findings 49 

5.2 Legislative measures to combat DWT schemes 54 



5 

5.3 FASTER directive 60 

5.4 Possible ways to combat DWT schemes in Finland 64 

6 Conclusions 69 

References 74 

  



6 

 

Figures 
 
Figure 1. Stock lending (Based on Veikkola, 2022). ........................................................ 30 

Figure 2. Dividend payments paid abroad from Finland by country 2010-2018 (Tokola/ 

Verohallinto, Harmaan talouden selvitysyksikkö, 2019). ................................................ 34 

Figure 3. T-test results as p-values. ................................................................................. 51 

 
 

Tables 
 
Table 1. Cum-cum trading (Based on Schreiber, 2013). ................................................. 31 

Table 2. Cum-ex scheme (Based on Correctiv, 2018). .................................................... 36 

Table 3. Results on share turnover.................................................................................. 49 

 
 
 

Abbreviations 
ATAD      Anti-Tax Avoidance Directive 

DWT      Dividend Withholding Tax 

EU      European Union 

FASTER     Framework for Authorised Systematic and  

Transparent Exchange of Relief 

GAAR      General Anti-Avoidance Rule 

LOB      Limitation of Benefits 

MNE      Multinational Enterprise 

PPT      Principal Purpose Test 

TRACE      Treaty Relief and Compliance Enhancement 

TRS      Total Return Swap 

  



7 

1 Introduction 

The mobile capital tax base is leaking in Finland and across Europe, which poses a serious 

threat to economic stability and equity. One example is the phenomenon commonly re-

ferred to as withholding tax schemes. As a result of leaks such as the Panama Papers 

(OECD, 2023), tax evasion and aggressive tax planning have been revealed as widely 

adopted business practices worldwide. The complexity and poorly managed tax systems 

allow wealthy individuals and global enterprises to reduce their tax burden by taking 

advantage of loopholes in the tax legislation. This thesis investigates trading schemes 

associated with dividend withholding tax (DWT), which have incurred significant costs 

for European taxpayers in recent decades.  

 

An OECD report on dividend tax schemes released in 2023 describes dividend withhold-

ing tax schemes as a process in which shares are traded, sold, and repurchased to avoid 

paying dividend taxes or claiming refunds for unjustified taxes. 

 

Withholding tax schemes and other abuses related to withholding tax are typically based 

on the abuse of benefits based on tax treaties and directives (Finnwatch, 2024). The ra-

tionale for arrangements lies in that dividends distributed to different countries are 

treated differently in terms of withholding tax: for example, the tax agreements between 

Finland and Great Britain, Finland and Ireland, and Finland and the United Arab Emirates 

exclude dividends distributed to the tax treaty partner country completely from with-

holding tax (Verohallinto, 2023). This creates a motive to artificially direct dividends ac-

crued in Finland by the countries in question in situations where the actual recipient of 

the dividend is beyond the countries in question, and withholding tax would be charged 

on the dividends when paid out to the country in question.  

 

1.1 Background 

Dividend taxation has become an international topic of discussion following the revela-

tions of dividend tax refund schemes in the media. Tax avoidance related to dividend 



8 

payments made it into public discussion in 2018 when an international group of journal-

ists published the so-called CumEx Files data leak documents. According to Yle's MOT 

program, the Finnish government lost an estimated 100 million euros annually due to 

the laundering of dividends, or a total of 1.3 billion euros between 2005 and 2017 (Sokala, 

2018a; Sokala, 2018b). 

 

Informal calculations suggest that cum-ex transactions alone have led to revenue losses 

totaling 69.2 billion euros across the EU (Correctiv, 2018). However, subsequent investi-

gations have shown even higher losses surpassing €100 billion across a broader geo-

graphical scope than initially revealed in the Cum-Ex Files (ESMA, 2020). According to 

Correctiv (2021), when cum-ex and cum-cum transactions in 12 countries, including the 

United States, are factored in, the losses are estimated at 150 billion euros. Therefore, 

based on the estimates mentioned above, this form of dividend tax arbitrage stands as 

the most extensive tax arbitrage scheme ever revealed globally. 

 

These schemes are facilitated by major European banks exploiting legal and regulatory 

loopholes to evade taxes on behalf of certain investors, and as a result, transferring the 

tax burden onto the society, according to an article by the New York Times (2020). Some 

banks also received bailouts from treasuries during the 2008 financial crisis. These 

schemes yielded almost risk-free profits from European treasuries, with one former par-

ticipant aptly describing them as "The Devil’s Machine" in an interview with Segal (2020) 

of the New York Times. 

 

In the Cum-Ex Files, financial sector agents employed intricate strategies involving own-

ership changes around ex-dates. Through technically legal means, these schemes were 

used to evade taxes and, in more severe instances, to claim multiple tax refunds from a 

single tax payment (Correctiv, 2018). These methods are called cum-cum and cum-ex, 

and they are discussed in the third chapter. These trading schemes are collectively re-

ferred to as dividend withholding tax (DWT) schemes targeted at minimizing or avoiding 

dividend withholding tax. 



9 

 

The first DWT schemes were initiated in 2001 by German state tax auditor Hanno Berger, 

also known as Mr. Cum-Ex (Doctorow, 2020). In many jurisdictions, determining whether 

DWT schemes are legal is challenging, as they operate at the crossing of financial markets 

and tax laws. In addition, communication among financial and tax authorities both do-

mestically and internationally is imperfect. This institutional failure has, however, hin-

dered efforts to address DWT schemes effectively (ESMA, 2020). Hence, several Euro-

pean countries have introduced legal measures to combat DWT schemes. 

 

However, there are still serious challenges to monitoring and dealing with collusive ac-

tion, in part because of the complex and sometimes hidden patterns of modern tax eva-

sion activity. New legislative efforts, such as the European Union's FASTER directive pro-

posal, are some of the measures directed toward the development of greater transpar-

ency and enforcement on cross-border financial flows. 

 

1.2 Literature review 

Following the revelations from the CumEx-files, numerous studies have investigated 

DWT schemes across Europe. A consistent finding among these studies is evidence sug-

gesting the widespread use of DWT schemes in various European countries. Many of 

these analyses rely on the distinct trading patterns that DWT schemes generate near ex-

dividend dates to identify the presence of extensive cum-cum or cum-ex trading activity. 

 

Buettner, Holzmann, Kreidl, and Scholz, in their 2020 paper “Withholding-tax non-com-

pliance: the case of cum-ex stock-market transactions, ” studied Germany’s reform in 

2012, which attempted to combat cum-ex schemes. They analyzed publicly traded com-

panies in Germany from 2009 to 2015. The findings of the study show that the reform 

was successful in reducing the spike in the trading volume on the German stock exchange 

during the dividend payment period, which implies that cum-ex trading was being effec-

tively addressed.  

 



10 

Casi, Gavrilova, Murphy, and Zoutman in 2022 assessed the impact of Denmark's 2016 

reform which was implemented to ensure that dividend withholding tax (DWT) was not 

misused. They compared Denmark to other Nordic countries and evaluated them before 

and after Denmark’s 2016 reform. The study revealed that the reform effectively elimi-

nated the known patterns of the DWT scheme trading. Denmark was able to collect an 

additional $1.3 billion as tax revenue every year. The study also found that there was 

evidence that Germany’s 2016 reforms were able to stop cum-cum trading, and it also 

pointed out that DWT schemes are still a big issue in several other European countries. 

 

Laturnus et al. (2022) used different sources of data than Buettner et al. (2020) and Casi 

et al. (2022) and still came to the same conclusion that Germany’s measures were effec-

tive. Moreover, Wagner and Wei (2022) investigated several financial markets in Western 

Europe and found abnormal trading activities in every country on the ex-dividend date. 

They also estimated the amount of loss for the countries involved.  

 

Haesner and Schanz (2013) discuss the German context where there are huge trading 

volumes on ex-dividend dates.  While there have been several other studies, which have 

provided different explanations for these patterns in anomalous trading activity sur-

rounding the record date, including Lakonishok and Vermaelen (1986), Karpof and 

Walkling (1990), Michaely and Vila (1995), Dhaliwal and Li (2006), Akhmedov and Jakob 

(2010), Haesner and Schanz (2013), Hartzmark and Solomon (2013), and Henry and Koski 

(2017), this thesis argues that non-compliance with withholding tax legislation provides 

an additional explanation for these patterns. Later, in Chapter 2.6, this thesis will explore 

studies with alternative explanations for trading spikes near the ex-date.   

 

1.3 Purpose of the study 

The goal of this master's thesis is to make a comprehensive and quantitative analysis of 

dividend withholding tax (DWT) schemes, such as cum-cum and cum-ex, focusing on 

their exploitation of weaknesses in tax systems. Finland has become the first EU member 

state to accept the TRACE policy to fight DWT schemes. Based on this, the EU is preparing 



11 

to introduce the FASTER directive to other EU member states. The FASTER directive is 

modelled after Finland’s TRACE policy. This thesis examines the impact of TRACE to eval-

uate whether implementing the FASTER directive across the rest of Europe would be a 

beneficial and effective measure. 

 

This thesis describes how these DWT schemes work, how often they occur in European 

countries while focusing on Finland, and how TRACE works to curb these schemes. To 

get a sense of the scope and trend of these schemes, this thesis looks at trading volumes 

daily and pays particular attention to trading on ex-dividend days. Anomalies in the data 

may be related to cum-cum and cum-ex schemes. This thesis also assesses the efficacy 

of the TRACE policy by comparing trading activities at the ex-dividend date before and 

after the implementation of the regulatory changes made by the TRACE. By focusing on 

trading volume data, this thesis also bridges a gap in existing literature, where previous 

research has predominantly relied on share lending or derivatives data, thereby adding 

a novel dimension to the analysis. 

 

The prevalence of DWT schemes in Europe has been well-documented, as discussed in 

the literature review chapter. However, there is limited research on the measures taken 

to curb these schemes and the effectiveness of such interventions. Notable exceptions 

include studies on Germany and Denmark, where reforms have shown significant suc-

cess. However, there has been no evaluation of the impact of the TRACE policy imple-

mentation in Finland, which is the focus of this thesis. Through this analysis, the thesis 

aims to provide valuable insights for current and future policymakers. 

 

International investors and cross-border trading have become widespread in the last few 

decades due to the rapid development of the capital markets around the world. For ex-

ample, between 1980 and 2017, foreign ownership of publicly traded equities in the UK 

increased from 4% to 54% (Tang et al., 2019), a trend also observable in other European 

markets. Examining dividend withholding tax schemes is important because of the size 

of the capital market in Europe. 



12 

To the best of my knowledge, there is no previous legal research that looks at DWT 

schemes and their application of intervention models from European and Finnish inter-

national tax law (EU tax law, tax treaty law, and Finnish national tax legislation) in cases 

of dividend withholding tax schemes. Furthermore, this thesis discusses the new Euro-

pean Union’s so-called FASTER directive proposal (COM (2023) 324 final) and attempts 

to assess its potential future impacts. 

 

However, it is worth noting that the study has some limitations. First, the analysis is 

based on the assumption that the trading behaviour is abnormal when there are cum-

cum and cum-ex schemes, which may not always be the case in practice. In addition, the 

study only examines the impact of the TRACE policy on trading activities and does not 

consider other factors that may influence market behaviour. Despite these limitations, 

this thesis provides valuable insights into the working of cum-cum and cum-ex schemes 

and the effectiveness of the TRACE policy in combating these schemes. The findings of 

the study have implications for policymakers, regulators, banks, and other financial in-

stitutions to come up with and implement effective measures to prevent tax evasion and 

money laundering activities in the capital market. 

 

1.4 Research question and hypothesis 

The research question examined in this thesis is: 

 

"Does the implementation of the TRACE policy in Finland influence trading volumes 

around the ex-dividend date, and is there evidence of abnormal trading activity before 

and after its introduction?" 

 

Two hypotheses are formulated to explore this question: 

1. 𝐻0𝐴: Share trading volumes near the ex-dividend date remain consistent with 

typical trading volumes. 



13 

2. 𝐻0𝐵: The implementation of the TRACE policy does not impact trading volumes 

near the ex-dividend date, nor does it affect patterns associated with DWT 

schemes. 

 

The primary focus of the thesis lies in testing the second hypothesis, as the core objective 

is to evaluate how the TRACE policy influences DWT schemes. Nonetheless, addressing 

𝐻0𝐴 is necessary for achieving this goal. 

 

To examine these hypotheses, the thesis conducts a comparison event study, focusing 

on two key events: 

1. A company's stock going ex-dividend. 

2. The implementation of the TRACE policy. 

 

This study takes as its empirical setting the daily share trading volume of up to one hun-

dred of the most actively traded dividend-paying companies in Finland. It also looks at 

the change in the volume of trading and the difference in the abnormal volumes before 

and after the policy change. 

 

Based on previous research, the thesis anticipates identifying abnormal trading volumes 

in Finland before TRACE, leading to the rejection of hypothesis 𝐻0𝐴. Furthermore, it is 

expected that the findings will show the same results as previous research, which ob-

served a decline in abnormal trading after the reforms in Germany and Denmark. If sim-

ilar patterns are observed, hypothesis 𝐻0𝐵 will also be rejected. 

 

However, this thesis distinguishes itself by utilizing trading volume data instead of share 

lending data or single stock futures, as employed in studies like Casi et al. (2022). More-

over, the analysis focuses on the period from 2020 to 2023, a timeframe that has not 

been extensively explored in previous research. As a result, the findings of this thesis 

may present some new ideas and may lead to some contrasting conclusions.  

 



14 

Therefore, based on the identified research problem, the quantitative empirical analysis 

with the event study method is considered appropriate. This is because an event study 

is a technique particularly suited for analyzing changes in the volume of trading activity, 

such as those that occur on and after ex-dividend dates or on and after the implemen-

tation of the TRACE policy. Daily trading volume data enables this approach to identify, 

with high accuracy, periods of abnormally high or low trading activity—i.e., the kind of 

activity that may be associated with DWT schemes. 

 

In addition, statistical analysis techniques, including t-tests, assist in examining the sig-

nificance of the differences in trading volume between the periods before and after the 

policy change. This method was chosen because it best fits the study's objective of pre-

senting empirical data on how effectively TRACE works in preventing DWT strategies, 

with results that can be generalized and replicated. In addition, the study focuses on 

trading volume data, which is not commonly used in existing literature, and has mainly 

concentrated on share lending or derivatives data, thus providing a new perspective to 

the analysis. 

 

Furthermore, the thesis uses a legal dogmatic approach as the secondary method. The 

legal dogmatic method, in particular, is significant for addressing the intricate legal struc-

tures and principles underlying DWT schemes. This approach provides a systematic ex-

amination, interpretation, and critical analysis of legal provisions, principles, and con-

cepts about statutes, regulations, and court decisions. It focuses on organizing and cat-

egorizing these norms into a coherent framework, evaluating their consistency, effec-

tiveness, and fairness in their practical application. 

 

With the legal dogmatic approach, the thesis analyzes the effectiveness of existing legal 

instruments in combating DWT schemes such as cum-ex and cum-cum. It criticizes 

whether current legislation is adequate to prevent such schemes and recommends how 

they can be improved. This is also a foundation upon which the thesis presents legal 

reform proposals to make withholding tax regimes more effective and fairer. Through 



15 

empirical analysis and legal dogmatic approach, the thesis fills the gap between the the-

oretical and practical perspectives and the economic and legal dimensions of DWT 

schemes. 

 

1.5 Structure of the Study 

This chapter will present the structure of this thesis. First, this study introduces the topic, 

research, and hypothesis in the first chapter. The second chapter will go through im-

portant concepts and definitions, laying the foundation for the third chapter.  

 

The third chapter introduces the concept of DWT schemes, and everything related to it, 

including different types of schemes. Chapter 3 discusses the existing research on divi-

dend withholding tax schemes, primarily in Germany, focusing on identifying dividend 

arbitrage collusion and assessing the effectiveness of legislative interventions. Chapter 

4 explains and goes through the statistical methods used in the thesis.  Chapter 4 lays 

down the context for the results and analysis, wherein the outcome of the t-tests and 

share turnover tests will be provided and explained. 

 

The fifth chapter shows the analysis of share turnover data, focusing on trading patterns 

before and after the TRACE policy. It also shows the findings from the t-tests conducted 

to assess differences in share turnover across event and non-event windows, as well as 

between pre-policy and post-policy periods. The t-tests serve to statistically validate the 

observations regarding trading activity and the impact of the TRACE policy on abnormal 

trading volumes. 

 

The fifth chapter also provides different legislative measures to combat dividend with-

holding schemes. The fifth chapter goes through what has already been done and how 

various regulations have worked on combating DWT schemes.  This chapter focuses on 

different legislation in Europe, especially in Finland. The fourth chapter also discusses 

whether the legislation could support access to information even better than at present. 

 



16 

The final chapter, conclusions, summarizes the key findings of the thesis, addressing the 

research questions introduced at the beginning of the thesis. The current regulatory and 

legislative measures are examined, and possible solutions to the threats posed by such 

schemes are presented. The significance of the results in a larger context is examined for 

policymakers, tax administrations, and legislators, with recommendations for making the 

regulatory environment better in dealing with DWT schemes in the future. Suggestions 

for future research are also made in the final chapter.  



17 

2 Theoretical framework 

This chapter provides the theoretical introduction to the rest of the thesis. First, it ad-

dresses some basic notions of different types of tax avoidance and what is legal and ille-

gal tax avoidance. Second, it provides an overview of how dividends are generally taxed 

in Europe, with a particular emphasis on withholding taxes, which are both a component 

of dividend taxation and the DWT schemes and are discussed in detail. Thirdly, it briefly 

discusses two key concepts that are needed to comprehend the cum-ex scheme: share 

settlement and dividend entitlement. 

 

2.1 The differences between tax planning and tax evasion 

Since the thesis deals with withholding tax schemes on dividends and the interface of 

these schemes between tax planning and tax evasion, it is justified to first examine some 

important concepts related to the subject area, to clarify on a general level what type of 

activity has been addressed by the general tax evasion regulations. 

 

Understanding the distinction between tax planning and tax evasion is key to formulating 

the research question in this thesis. Tax planning is the legal steps taken to minimize tax 

burdens within the bounds of legal provision, whereas tax evasion comprises illegal ac-

tivities that actively avoid tax payment. This is significant in identifying the machinery of 

how DWT schemes operate because they tend to exploit the gray area between aggres-

sive planning and tax evasion. Some schemes, like cum-cum and cum-ex, exploit tax 

loopholes as well as tactics in timing their claims for exorbitant tax rebates or benefits. 

By going through these differences, this section offers the context to comprehend how 

DWT schemes function and why TRACE policies need to be implemented to counter such 

exploitative activities.  This theoretical foundation directly supports the research hypoth-

esis, which seeks to evaluate the impact of TRACE on abnormal trading volumes linked 

to these schemes. 

 



18 

The terms “tax planning”, “tax avoidance” and “tax evasion” are not defined and applied 

uniformly across all countries and are therefore interpreted in a different way. In Finland 

these concepts are divided into three levels (Knuutinen, 2012). Complying with the leg-

islator’s intent, tax planning is the use of lawful ways to minimize tax liability; for example, 

using tax incentives to encourage development of science, technology, and specific ac-

tivities like R&D. Tax avoidance, which is within the law but amounts to adopting artificial 

arrangements to secure unintended tax benefits which are not in the spirit of the tax 

laws, is in contrast to tax evasion which involves illegal activities like fraud or legal closure 

of the eyes to tax default. The Finnish criminal law has a strict definition of tax evasion 

(Juusela, 2008). 

 

In general, tax planning is accepted as long as it does not go against the laws of the 

country, both in Finnish domestic tax law and EU regulations (Ossa, 2020). For instance, 

businesses are permitted to organize their affairs so as to minimize their tax liabilities so 

long as the methods employed are consistent with the provisions of the tax laws. How-

ever, this is where the distinction between what is permissible tax planning and what 

amounts to tax avoidance can be rather foggy. Aggressive tax planning, which is often 

complex and involves things like tax havens or exploiting weaknesses in international tax 

rules, only serves to muddy the waters further (Knuutinen, 2014). While not illegal, ag-

gressive tax planning is morally questionable and is, therefore, a focus of the OECD and 

EU (Dourado, 2015). 

 

To prevent tax avoidance and aggressive tax planning, regulatory tools like GAARs (Gen-

eral Anti Abuse Rule) have been put in place. GAARs allow tax authorities to disregard 

arrangements designed primarily to obtain tax benefits contrary to the intent of tax leg-

islation (Knuutinen, 2009). In the EU, the Anti-Tax Avoidance Directive (ATAD) strength-

ens this approach by codifying anti-abuse provisions to protect member states’ tax bases. 

However, GAARs also raise concerns about legal certainty and the balance between en-

forcement and the protection of taxpayers’ rights (COM 136, 2015).  

 



19 

In Finland, measures to prevent tax evasion are set forth in Section 28 of the Taxation 

Procedure Act (VML). This provision allows the tax authorities to ignore the legal form of 

a transaction when the transaction does not reflect the real economic substance of the 

transaction, especially in the case of artificial arrangements (Knuutinen, 2006). The ap-

plication of Section 28 to both domestic and cross-border arrangements guarantees that 

the taxes are properly levied, even when the benefits of a treaty are being misused (Veik-

kola, 2020a). This application is also supported by Finnish legal literature as relevant to 

the elaboration of the issues arising from DWT schemes (Ryynänen, 2003; Helminen, 

2016).  

 

While tax planning is an important and lawful activity for individuals and businesses, 

their intention is often the factor that most clearly distinguishes between tax planning 

and tax avoidance. Effective anti-avoidance rules must address the challenges posed by 

aggressive tax planning without undermining the principles of legality and fairness. In 

the context of DWT schemes, Finland’s regulatory framework, supported by EU direc-

tives, aims to strike this balance by countering artificial arrangements and ensuring com-

pliance with tax laws. This thesis examines how these legal measures can be most 

properly used to stop DWT schemes without producing either relative legal uncertainty 

or tax injustice. 

 

2.2 Dividend taxation 

The mechanics of dividend taxation play an important role in understanding the exploi-

tation of withholding tax systems. Withholding taxes are designed to ensure that gov-

ernments collect tax revenue on dividend income before it reaches investors. However, 

as seen in various European countries, these systems have been manipulated to enable 

tax refunds on payments that were never made, often through complex schemes. This 

section explains the purpose of withholding taxes and their role in preventing tax avoid-

ance while highlighting how the refund and relief processes can be exploited. Under-

standing these mechanisms is critical to framing the research problem, as the TRACE 

policy directly targets these loopholes. By linking the principles of dividend taxation to 



20 

the research question, this section provides the theoretical context for analyzing the 

trading behaviours around ex-dividend dates and the effectiveness of policy interven-

tions. 

 

 Dividends can be taxed in two ways: through income taxes or withholding taxes. Gener-

ally, when companies pay dividends to individual shareholders, the shareholders must 

pay income tax on those dividends (Schreiber, 2013). Similarly, dividend income can be 

treated as part of a business's taxable income and subject to corporate taxation. There 

is, however, a tendency for dividends between corporations to escape taxation. 

 

Only this type of taxation applies to shareholders residing in the country where the div-

idends are generated. Instead, according to Schreiber, a withholding tax may be imposed 

on non-resident individuals and businesses, which cannot be taxed directly by the coun-

try distributing the dividends. The distributing company deducts the tax before distrib-

uting the net amount to shareholders. As an alternative, financial institutions, such as 

depositary banks or account operators, can carry out this process on behalf of a company 

(Buettner et al., 2018). At the end of the day, shareholders are responsible for meeting 

their tax obligations. 

 

2.2.1 Using withholding taxes to combat tax avoidance 

Withholding taxes also promote tax compliance. Suppose taxpayers are subjected to the 

withholding tax but report the income to the relevant tax authorities; they may receive 

a refund for the difference between this high withholding tax rate and the (usually lower) 

national tax rate. As a result, the elevated withholding tax encourages tax compliance by 

acting as a deterrent for evaders while providing a financial benefit to those who truth-

fully declare their earnings. This mechanism effectively penalizes non-compliance while 

rewarding transparency (European Commission, 2014). 

 

Johansson et al. (2016) found that withholding taxes also serve to curtail international 

tax planning. By imposing withholding taxes on interest payments, Johansson et al. 



21 

believe that patents, trademarks, and other intangible assets are less likely to be trans-

ferred to low-tax jurisdictions. Profits are commonly transferred from high-tax countries 

to low-tax countries using these tactics (Schreiber, 2013). Profits can be effectively taxed 

in high-tax countries by subjecting interest and intangible asset payments to withholding 

taxes, removing the incentive to shift profits. Likewise, withholding taxes on dividends 

discourages investors and MNEs from structuring their ownership to receive dividends 

in jurisdictions that treat dividends favourably. In essence, withholding taxes is seen as a 

vital tool in combating tax avoidance. 

 

The withholding tax on dividends is also affected by whether the share is registered as a 

trustee. A trustee-registered share means security in the value share system, which is 

owned by a foreigner or a foreign entity or foundation and which is managed by a Finnish 

custodian entity, i.e. a custodian bank (Nykänen, 2015). Based on a publicly listed share 

registered as a trustee, the dividend to be paid is subject to special regulation. In this 

case, also taking into account the applicable tax treaty regulations, the amount of the 

dividend to be collected can vary between 0% and 35%, regardless of whether the taxa-

ble person is a natural person or an entity. 

 

As said by Buettner et al., through tax withholding systems, taxpayers can claim refunds, 

but this can lead to the potential for fraudulent refund claims. Therefore, government 

revenue may be at risk of loss in the form of refunds of taxes that were never actually 

withheld, or from accidental double refunds for a single tax through schemes like cum-

ex trades. 

 

2.2.2 Withholding tax refunds and relief at source 

Refunding withholding taxes plays a major role in correcting situations in which excessive 

amounts of tax have been withheld due to a variety of factors, including differential 

treatment based on shareholder status or different tax rates for different investors. In 

some countries, for example, dividends distributed to foreign shareholders are subject 

to withholding taxes, while certain types of investors are exempt from withholding taxes. 



22 

This applies especially to inter-corporate dividends within the European Economic Area 

governed by the Parent-Subsidiary Directive (Schreiber, 2013). In addition, many coun-

tries have entered into bilateral tax treaties with the objective of eliminating or reducing 

the withholding tax on dividends paid to other countries, and, as such, different tax rates 

are incurred based on the shareholder's country of residence (Schreiber, 2013). 

 

These complexities require countries to implement systems to ensure shareholders re-

ceive accurate tax rates. When individuals or businesses earn income in multiple coun-

tries and are subject to taxation in each, they may be eligible for reduced tax obligations 

to avoid double taxation. One of the ways this burden can be eased is through the relief 

from withholding tax on sources, which reduces the net burden of taxation. The amount 

of such relief depends on several factors, including the nature and location of the in-

come-generating activity and any relevant international tax agreements (IMSG, 2010). 

 

Taxpayers must provide proof that they have already paid tax on their income in the host 

country for such relief to apply (IMSG, 2010). This may involve the production of docu-

ments such as tax receipts, bank transfer documents, or other official documentation of 

tax payment. In some cases, additional documentation, for example, a certificate from 

the foreign tax authority, may be required to establish entitlement to withholding tax 

relief. Once all requisite documents are lodged and cleared with the concerned authori-

ties, taxpayers stand to benefit by paying a lowered rate of taxation on their income in 

the second country. Such relief is tremendous in some cases, bringing the withholding 

tax rate down to nil, thus practically abolishing double taxation (IMSG, 2010). 

 

Briefly, withholding tax refund is an essential adjustment mechanism for over-withhold-

ing because of differential treatment and varying tax rates among shareholders. Cross-

border investment arrangements are endowed with this provision to offer fairness and 

compliance, particularly by addressing the challenges that may be encountered by for-

eign investors in recovering excessive taxes withheld. 

 



23 

2.3 Exploitation of tax treaties 

A notable feature of Finnish withholding tax is section 1 subsection 3 of the Withholding 

Tax Act, which subordinates withholding tax to tax agreements that Finland has made 

with different countries. In other words, the withholding tax law applies only to the ex-

tent that tax treaties do not apply. This leads to the fact that international tax treaties 

limit Finland's right to tax at the points specified in the tax treaties. In addition, the mere 

existence of a tax treaty can affect the taxation of national legislation, regardless of its 

content (Peltomäki, 2020). 

 

Situations where the taxpayer seeks benefits based on a tax treaty are called tax treaty 

abuse. In connection with tax treaties, it is natural to talk about abuse, because the ob-

ject of abuse is specifically the benefits based on the tax treaty. As said by Veikkola (2022), 

this is because by abusing tax treaty benefits, the taxpayer evades the tax based on na-

tional tax legislation. 

 

One form of tax treaty abuse can be considered treaty shopping, although it is a broader 

concept than abuse. Treaty shopping is not a legal concept and can be understood as an 

attempt to obtain benefits that persons living in another country receive based on a tax 

treaty concluded by this country. The concept differs from the misuse of tax treaties in 

that treaty shopping can also involve bona fide business arrangements (Veikkola, 2022). 

 

Tax treaty abuse and treaty shopping are strategies employed by global corporations to 

reduce their global tax obligations, providing them with an unfair advantage over do-

mestic companies that lack the scale or resources to participate in international tax 

avoidance. Essentially, these tactics allow global corporations to move income from 

higher-tax regions to lower-tax ones, thereby decreasing their overall tax burden. 

 

For illustration purposes, imagine a scenario where Country X and Country Y have a tax 

treaty that reduces withholding taxes on dividends. Company A, based in Country Z, 

wants to benefit from this reduced tax rate in Country X. To do so, Company A establishes 



24 

a shell company in Country Y, which has a favourable treaty with Country X. This setup 

enables Company A to route its income through the entity in Country Y, thereby qualify-

ing for the lower withholding tax rate in Country X under the treaty between Country X 

and Country Y. In this way, Company A is engaging in “treaty shopping,” structuring its 

operations to exploit the most advantageous tax treaty, despite lacking substantial busi-

ness activities in either Country X or Country Y. This manipulation undermines the 

treaty’s intent, allowing the company to cherry-pick favourable provisions without a gen-

uine economic presence in the involved countries. 

 

2.4 Trade settlement process 

To understand how a cum-ex scheme works and why this study uses the “5 days before 

and after the ex-date” in its analysis, it is first necessary to understand what the settle-

ment process means. Cum-ex trading capitalizes on the delayed settlement of share 

trades, exploiting the period between the trade date and the settlement date when own-

ership of shares is transferred. This settlement cycle, typically spanning a few trading 

days, is generally called a settlement rule or settlement period (Yadav, 2017). For a secu-

rity to move from a seller to a buyer, it must go through a settlement process, which 

creates a delay between the time of trade (T) and its settlement.  

 

For instance, a trade that settles two days after the trade date is said to follow a T+2 

cycle; in like manner, a settlement after three days is referred to as a T+3 cycle (Yadav, 

2017). It is important to understand the settlement process to understand the timing 

manipulations that occur in cum-ex transactions. 

 

While the settlement period is typically standardized within regulated markets of a coun-

try (Clearstream, 2014), international standardization has been lacking until recently. 

However, there has been a shift towards shorter settlement cycles internationally. Ac-

cording to Clearstream (2014), the EU adopted T+2 as the standard settlement cycle in 

2014. Meanwhile, both the U.S. and Canada transitioned from T+3 to T+2 in 2017 (Clear-

stream, 2017; U.S. Securities and Exchange Commission, 2017). Prior to the change to 



25 

T+3, the markets operated on a T+5 settlement cycle (U.S. Securities and Exchange Com-

mission, 2024). 

 

Since May 28, 2024, with the advances in technology and electronic trading, most stock 

trades settle in just one business day (T+1) (U.S. Securities and Exchange Commission, 

2024). The U.S., Canada, Mexico, and Argentina have switched to the T+1 settlement 

cycle. The European Commission is proposing 11 October 2027 as the appropriate date 

for the move to the T+1 settlement (European Commission, 2025).  

 

To make sure cum-ex schemes are captured for the whole period in the analysis, T+5 is 

used near the ex-date. Therefore, cum-ex schemes will fully be captured if they are pre-

sent during the settlement cycle and near ex-date together. In the next chapter, ex-date 

and record date are explained, so the basis for this thesis’ analysis (and for cum-ex 

schemes) can be fully grasped.  

 

2.5 Ex-date and record date 

One significant aspect of dividend payments is specific dates, and the ex-date is of major 

significance. An ex-dividend date is a date on which stock is sold without the benefit of 

the next following dividend payment. Instead, the dividend accrues to the previous 

owner. In other words, if an investor has the objective to purchase shares bearing divi-

dend entitlement, or "cum-dividend," they will need to make a purchase before the ex-

date. The day before the ex-date is commonly known as the cum-date. 

 

Additionally, the record date is important in dividend processes.  A company's board of 

directors decides the record date, and this is when a company compiles a list of the 

shareholders of the stock on which a company has announced a dividend. The list serves 

to determine the shareholders who are entitled to receive the dividend (Zoutman, 2019). 

Since the shareholder register on a settlement of transactions basis, only transactions 

settled on or earlier than this date qualify for dividend payment. But the entitlement to 

the dividend depends on when the transaction is done. Since transactions tend to settle 



26 

a few days post-execution, there is a time lag between shareholder registration and div-

idend entitlement. This inconsistency has the potential to lead to complications, such as 

when favored investors are not recorded as shareholders. As will be examined in more 

detail below, this imbalance is a central pillar of the cum-ex scheme, with room for divi-

dends to flow to the incorrect investors. 

 

As of May 2024, the ex-date and record date are the same in the U.S., Canada, Mexico, 

and Argentina (U.S. Securities and Exchange Commission, 2024). This change in the reg-

ulations will curb the challenges of preventing cum-ex and cum-cum schemes. As the 

European Commission has proposed to change the settlement cycle for EU securities 

from two days to one, the ex-date and record date would, therefore, also be the same 

day in the EU.  

 

2.6 Trading spikes near the ex-dividend date 

The thesis will attempt to determine the patterns of cum-cum and cum-ex trading by 

analyzing the trading volume around the ex-dividend date. As expected, there should be 

spikes in trading around the ex-dividend date, particularly in the pre-TRACE implemen-

tation era. However, as noted in the literature review, there are trading spikes around ex-

dividend dates, which can also occur in the absence of DWT schemes. Therefore, it is 

necessary that the thesis also examines other potential reasons for this increased trading 

volume. 

 

In a Walrasian ideal capital market with the absence of transaction costs and taxes, the 

stock price is expected to decrease precisely by the value of the dividend distributed per 

share (Miller and Modigliani, 1961). However, as empirical evidence shows by Frank and 

Jagannathan (1998), stock prices typically fall short and do not decrease by the value of 

the dividend distributed, a phenomenon primarily attributed to tax implications. When 

investors face varying effective tax rates on dividend income, arbitrage opportunities 

arise. Elton and Gruber (1970) were the first to justify the ex-dividend phenomenon and 

introduced the tax-driven clientele effect. They pointed out that the price drop ratio is 



27 

highly affected by specific tax clientele who receive dividends. Previous research by Wag-

ner and Wei (2022) also insists that trading spikes near ex-dividend dates are largely 

driven by tax heterogeneity among investors. 

 

Lakonishok and Vermaelen in 1986 in their study “Tax-induced trading around ex-divi-

dend days” investigated trading volume around the ex-dividend dates for US companies 

from 1970 to 1980. Lakonishok and Vermaelen found increased trading activity both be-

fore and after the ex-dividend date. Similar findings were reported by Michaely and Vila 

(1995) and Henry and Koski (2017), with the latter highlighting that skilled institutional 

investors use dividend capture strategies to capitalize on their lower transaction costs. 

 

Rantapuska (2008) investigates the Finnish market and discovers that different tax ben-

efits on dividends attract investors to trade around the ex-dividend day, but he finds no 

evidence that the tax-induced clientele effect is reflected in stock prices. While Chen et 

al. (2013) show evidence from the Taiwanese market that ownership structure has a sig-

nificant effect on price and trading around the ex-dividend day, they also find evidence 

that contradicts the dynamic dividend clientele hypothesis. 

 

It should also be acknowledged that tax heterogeneity is not the only factor causing 

spikes in trading around ex-dividend dates. In their paper from 2001 “The ex-dividend 

day stock price behavior in the Athens Stock Exchange” Milanos and Travlos investigated 

trading spikes near ex-dividend dates on the Athens Stock Exchange. They found a sig-

nificant increase in trading, even in 2001 when Greece had no taxation on dividends or 

capital gains. They attributed this to price discrepancies, with cum-dividend transactions 

being executed at the bid price and the ex-dividend transactions at the asking price. This 

price differential gives investors reasons to purchase stocks on the ex-dividend day and 

sell on the cum-dividend day, while market makers exploit these dynamics to increase 

trading volumes. These findings align with Dasilas (2007) and Al-Yahyaee, Pham, and 

Walter (2011), who observed similar trends in markets such as Oman. 

 



28 

The analysis of the thesis is also based on these factors for DWT schemes. However, by 

comparing trading patterns before and after the legislative changes, the analysis as-

sumes that most of these factors are time-invariant, and hence, the effects of DWT 

schemes on trading activity are isolated. 



29 

3 Different dividend withholding tax schemes 

Many factors are involved in dividend withholding tax schemes, making it difficult to 

summarize them. This chapter examines three examples of schemes that differ from one 

another. It should be noted that similar schemes may operate differently depending on 

the local laws and regulations. 

 

The focus turns to tax-motivated trading after laying the foundational knowledge in the 

previous chapter, beginning with a concise description of the mechanism and objectives 

of cum-cum trading. Afterward, the thesis turns to its biggest scheme - cum-ex trading. 

To provide a detailed examination of cum-ex transactions, an in-depth assessment of the 

cum-ex scandal in Germany is presented, followed by an overview of known details con-

cerning cum-ex activities in other countries.  

 

3.1 Cum-cum 

According to Buettner et al., cum-cum is a form of arbitrage that takes advantage of var-

iations among tax burdens on investors. Cum-cum is designed to take advantage of dif-

ferent tax rates and tax treaties applicable to different types of investors. In contrast to 

cum-ex, the trading is much simpler and straightforwardly near the ex-date. This scheme 

requires trading securities and transferring securities, both with dividend entitlements, 

to another party with a lower tax liability on dividends than the original owner, thus 

reducing the investor's tax burden. It is possible to minimize or avoid dividend withhold-

ing taxes through cum-cum trading; however, unlike cum-ex, it does not allow for the 

reimbursement of multiple taxes at once during cum-cum trading. Due to legal complex-

ity, curbing cum-cum schemes can be a challenge, since legislation targeting such 

schemes is often difficult to enforce (ESMA, 2020). 

 

As quoted by Wigan (2019), cum-cum trades have a broadly similar structure in most 

countries. Such trades involve a transfer of shares' ownership before and after the share 

ex-dividend date, with either share lending or outright purchase being the method to 



30 

achieve this. Transaction volume data are employed throughout this thesis, so cum-cum 

transactions executed with the use of share lending will be excluded. Whereas cum-ex 

schemes are time-based and thus may be vaguer to distinguish at times, cum-cum 

schemes both demand an issue prior to the ex-dividend date and a subsequent buy-back. 

Thus, cum-cum schemes will always manifest in the data as two discernible events: ini-

tially, the sale, followed by the buy-back of the shares. 

 

 

 

 

Figure 1. Stock lending (Based on Veikkola, 2022). 

 

According to Knuutinen (2009), a situation in a share lending arrangement can arise 

where the ownership in the financial sense remains, but in the legal and formal sense, it 

changes. That is, the formal ownership of the share can be transferred to the borrower 

(B), but its financial ownership remains with the lender (A). Taxation has had challenges 

in adapting to these situations where there is tension between legal and financial own-

ership. 

 

Share lending often has a connection point to financial institutions' cash flow financing 

means, in which case the loan is part of a secured debt arrangement. In the loan agree-

ment, the lender (A) transfers the shares to the borrower (B), who has the obligation to 

return the shares at the end of the loan period. In a simple arrangement, the borrower 

(B) receives the dividend produced by the stock, while the lender (A) can receive 



31 

payment in the form of a manufactured dividend. In the arrangement, the lender (A) 

basically has no risk of the share falling in value, because the borrower (B) is responsible 

for this (Pankakoski, 2020). Share lending arrangements are made, for example, for se-

curities trading. There is also typically an agreement in which the borrower (B) pays mon-

etary compensation for the shares (i.e. pays, as it were, rent) and can pledge the shares 

as security at the market value of the shares (Collin, 2020).  

 

Table 1. Cum-cum trading (Based on Schreiber, 2013). 

 

 

In Table 1, two investors are involved in a cum-cum trade. When investors have different 

dividend withholding tax rates, cum-cum trades are generally profitable. The disparity 

allows cum-cum trades to occur even with domestic investors or those who qualify for a 

more favourable reimbursement. Without considering transaction costs, investor A and 

investor B achieve a profit compared to the passive alternative of non-collusion. Without 

collusion, investor A would have to pay more taxes, such as €34,000. When colliding with 

investor B, they made a €9,000 profit (€34,000 - €25,000). Using cum-cum trades, the 

tax authority (and society) incurs the same loss as the investor gains. The cum-cum 

scheme uses treaty shopping to its advantage.  



32 

 

With a total return swap (TRS), it is possible to transfer the financial risk of the agreed 

target asset to a party that does not actually own the target asset. The use of TRS agree-

ments in the withholding tax planning of dividends is based on the exploitation of asym-

metries in the tax system, i.e. so-called tax arbitrage: either on the fact that dividends 

and payments based on derivatives are treated differently in international tax law (divi-

dends are subject to withholding tax more widely than income from derivatives) or on 

the fact that the recipient of the dividend is “switched”  to a recipient residing in a coun-

try that is exempt from dividend withholding tax based on a tax treaty. Practically getting 

perfect results often requires the existence of arbitrage opportunities when looking at 

the taxation of the source country of both the dividend and the derivative income. 

 

The example above was a simplified version of cum-cum scheme. Next, this thesis shows 

another version of how tax treaties and derivatives are used in cum-cum schemes. Con-

sider the following scenario: a company based in Germany (Company A) holds shares in 

a Finnish company (Company X), where a valid tax treaty exists between Finland and 

Germany (SopS 85–86/2017138). This tax treaty, modeled after the OECD framework, 

allows Finland, the source country of the dividends, to impose a 15% withholding tax on 

portfolio dividends. It is important to clarify that the German company (A) does not 

maintain a permanent establishment in Finland, and the investment qualifies as a port-

folio investment, meaning that it does not grant controlling power over the Finnish com-

pany (X). Consequently, under the provisions of the tax treaty, Finland is entitled to im-

pose a 15% withholding tax on the dividends paid by Company X to Company A. 

 

However, Company A may engage in a strategic arrangement with a British company 

(Company B), by utilizing share or derivative agreements to circumvent Finland's with-

holding tax obligations. In this arrangement, Finland effectively forfeits its right to levy 

the withholding tax, as the dividend recipient is nominally recognized as Company B un-

der established tax practices. Due to the tax treaty between Finland and Great Britain, 

dividends paid to British residents are subject to a 0% withholding tax. Moreover, Finland 



33 

is unable to tax the derivative (or substitute dividend) payments made between the com-

panies, as these payments lack a sufficient nexus to Finland under § 10 of the Finnish 

Income Tax Act. Compounding this issue, Great Britain's domestic tax legislation broadly 

exempts dividends from taxation (HMRC, 2016; Veikkola, 2022). 

 

This situation is however, exacerbated by the fact that Great Britain does not impose 

withholding taxes on substitute dividends paid to Germany. According to Veikkola (2022), 

the OECD model tax treaty prevents the taxation of substitute dividends based on share 

lending agreements in the residence country of the payer (in this case Great Britain) as 

for instance such income is considered as “other income” under Article 21. Therefore, 

this arrangement raises significant concerns, particularly given the disproportionate al-

location of dividend flows to jurisdictions with low or zero tax regimes as illustrated in 

the next Figure 2. As highlighted, the practice of using such arrangements to avoid with-

holding tax on dividends remains a relevant issue, and tax authorities are well aware of 

the methods used in these cases. 

 

This scenario shows a present and sophisticated problem in the field of international 

taxation, which highlights the difficulties that tax authorities have in their governments. 

From the point of view of international withholding tax planning, activities directed to 

Great Britain, in particular, should be critically examined, as the second largest number 

of dividends from Finland are paid there, but Finland does not have the right to withhold 

tax on them (Tokola, 2019). 



34 

 

Figure 2. Dividend payments paid abroad from Finland by country 2010-2018 (Tokola/ 
Verohallinto, Harmaan talouden selvitysyksikkö, 2019). 

 

Annually, significant amounts of dividends are paid abroad from Finland, and the Finnish 

Tax Administration's point of view is to try to protect Finland's right to tax as a source 

country as much as possible. When looking at the statistics, however, a worrying phe-

nomenon can be observed that a significant part of the dividend flows paid from Finland 

are directed to countries with which Finland has a tax treaty with which either all or 

some groups of recipients apply a zero-tax rate. Today, those zero-tax treaties are still in 

force between Great Britain, Ireland, Mexico, and France. 

 

3.2 Cum-ex  

According to Spengel (2016), the purpose of cum-ex trading is not to avoid taxes, but 

rather to obtain multiple refunds for withholding taxes only paid once. Since cum-ex lit-

erature is very limited and largely based on the German cum-ex scandal, this explanation 

of cum-ex is derived from the method utilized there. Nevertheless, cum-ex has been de-

tected in other countries, and similar schemes have likely been implemented or with 

similar characteristics. It is therefore deemed necessary to describe in detail everything 



35 

behind cum-ex trading in Germany, as well as the weaknesses in the German withholding 

tax system that led to the development of this scheme. In Germany, the scheme involved 

a series of complex trades executed within a brief timeframe near dividend distributions. 

As said by Buettner et al. (2018), these trades were structured to benefit from the way 

dividends and withholding taxes were processed.  

 

3.2.1 Cum-ex in Germany 

These frauds take advantage of weaknesses in how dividend withholding tax refunds are 

handled. The following example illustrates how cum-ex was used in Germany. In Latin 

terminology, "cum" refers to securities with dividend entitlements, while "ex" refers to 

securities settled without dividend entitlements. An important feature of cum-ex is that 

it involves the trading of securities first with dividend entitlements, and then without 

them. The scheme is designed to appear sophisticated and complex, creating the im-

pression of legitimate claims, allowing multiple refunds on taxes withheld only once. This 

section outlines a variety of cum-ex approaches, some of which perhaps could be uni-

dentified. Table 2 is based on the Cum-Ex Files method that Correctiv (2018) revealed on 

the German market.  



36 

 

Table 2. Cum-ex scheme (Based on Correctiv, 2018). 

 

 

Table 2 outlines the cum-ex scheme steps as depicted in the investigation by Correctiv 

(2018). Investor A in this scheme sells the shares to Investor B just before the ex-divi-

dend date, that is, the day on which rights to the dividends are determined. Though 

the sale is made, Investor A and Investor B both believe that they will be entitled to the 

dividend. Even though Investor B is not receiving the dividend itself, both investors file 

refund claims on withholding tax. This results in both Investor A and Investor B getting 

a refund of the same payment of tax. 

 

But because of weaknesses in the system, tax authorities inadvertently refund the tax 

to both parties, refunding more tax than was withheld. The two investors then split the 

proceeds of these duplicate refunds between them, profiting from the fraud. The tax 

authority, and therefore society, loses out financially, with more tax being refunded 

than was initially paid. Such abuse of the system demonstrates the impact of the cum-

ex scheme on public revenue. 



37 

 

There are typically multiple parties involved in cum-ex trading. As stated by Correctiv 

(2018), this process involves the original shareholder, financial institutions or brokers 

who engage in borrowing and short selling of the shares, and a secondary investor who 

acquires the shares shortly before the ex-dividend date, frequently relying on bank 

loans to leverage their investment position. Generally, security transactions are settled 

within two or three business days, otherwise known as T+2 or T+3. On the settlement 

day, payment is required, and the shares must be handed over to the buyer. When a 

transaction is made, for instance, two business days before the ex-dividend day, the 

settlement may coincide with the ex-dividend day, resulting in the purchase of shares 

cum dividend but delivery of shares ex-dividend. 

 

The typical clearing process ensures that the buyer receives the stock at its ex-dividend 

price, a dividend compensation from the seller, and a tax certificate from their custo-

dian bank. In the case of cum-ex trading, although the seller does not actually possess 

the stocks; instead, they engage in short selling. Short selling of this form is allowed if 

the seller has made arrangements to borrow the securities or has a contract with a 

third party securing the availability of the securities at the time of settlement. Since the 

depository bank does not recognize the transactions as short selling, it issues a tax cer-

tificate, leading to a double tax refund. Ultimately, the buyer resells the shares to the 

original owner, who redistributes the additional tax refunds to the parties involved. 

 

Most cum-ex transactions can be easily recognizable through their pattern of stock 

transfer around the ex-dividend date and thus as trading strategies. In the case of cum-

ex trading, the short sale occurs one or two days before the ex-dividend date, leading 

to a noticeable increase in trading activity. This type of short sale is characterized by 

the fact that it is executed before the ex-dividend date, but the actual delivery of the 

shares is made on or after the ex-dividend date. 



38 

3.2.2 Evidence of cum-ex in the rest of Europe 

So far, most of the research has been on methods used in Germany. It is important to 

note that the cum-ex strategies that may be in use in other nations are not exactly similar 

to the German model. Since there are likely differences in how dividends and dividend 

withholding taxes are handled across different countries, it is reasonable to conclude 

that cum-ex operators, when operating across borders, have had to change their meth-

ods. The Danish cum-ex case is a good example of such adaptation. 

 

However, it has been noted that Denmark is one of the countries most affected by cum-

ex activities. This is also supported by media coverage, as can be seen in the CumEx-Files. 

Denmark is one of the few European nations, along with Germany, where cum-ex has 

been mentioned as a problem by several reliable sources. During a hearing in the EU 

Parliament in 2018, it was explicitly acknowledged that the occurrence of cum-ex trades 

in Denmark (Spengel & Schick, 2018). Additionally, as previously mentioned, Buettner et 

al. (2018) asserted that Denmark has been affected by cum-ex activities. 

 

Danish tax authorities uncovered in 2015 extensive dividend arbitrage transactions that 

exploited loopholes in the DWT reimbursement system, which were estimated to yield 

revenues of approximately $1.3 billion annually. In response, the authorities imposed 

stricter documentation requirements to discourage nonqualified parties from claiming 

DWT refunds. Casi et al. (2022) report that this policy change led to a 130% increase in 

annual DWT revenue. 

 

Casi et al. in their 2022 work examine the effect of tax reform on DWT arbitrage, tax 

collection by Denmark, and investors' and Danish companies' behavior. From their study, 

they are in a position to establish that the reform was highly effective in preventing ar-

bitrage, particularly in cum-cum and cum-ex transactions, with a remarkable increase in 

revenue for DWT. Additionally, the reform only briefly negatively affected Danish com-

panies. 

 



39 

Finland introduced significant changes in 2021, the TRACE policy, to the withholding tax 

policies. The Finnish regulatory framework was not very restrictive before these changes 

regarding the imposition of cum-ex schemes, and such tax arbitrage techniques could be 

implemented with few problems. Through January 1, 2021, Finland had a simplified 

framework for the withholding taxes on dividends distributed to nominee shareholders. 

According to Joiniemi (2021), this system resulted in dividends being taxed at the rate of 

15% by the relevant tax treaty without the need for detailed information on the final 

beneficiary of the dividend. If a higher tax rate was available under the treaty, then that 

was used, and in the event of dividends paid to jurisdictions with no tax treaty, the stand-

ard withholding tax rate was used. 

 

The simplified system required the dividend payer to provide evidence that the recipient 

was entitled to tax treaty benefits. Typically, an agreement between the domestic ac-

count manager and a foreign asset manager was deemed sufficient, provided they en-

sured compliance with the tax treaty and reported any relevant changes in the recipi-

ent’s tax status (Nykänen, 2015). However, the process did not demand detailed infor-

mation about the final recipient of the dividend, leading to gaps in oversight. 

 

To enhance Finland's position in international capital markets, a simplified procedure has 

been introduced to ensure competitiveness by clarifying the withholding tax process and 

liability. However, this has also raised issues, particularly for the Finnish Tax Administra-

tion, which has had difficulty in verifying whether beneficiaries were properly entitled 

to tax treaty protections. Limited access to information on foreign recipients and the 

potential for providing inaccurate data hampered effective monitoring of the system, 

making it easier for cum-ex schemes. 

 

In January 2013, the OECD published a model called Treaty Relief and Compliance En-

hancement- Implementation (TRACE) package for the adoption of the Authorized inter-

mediary system, the purpose of which is to facilitate the submission of information re-

quired for withholding tax and to enable the application of tax treaty benefits in the 



40 

correct amounts already during the tax year. The TRACE implementation package was a 

very detailed and independent entity that contained several standardized applications, 

contracts, forms and descriptions, with the help of which the TRACE model was sup-

posed to be introduced in as many countries as possible with as little modification as 

possible (OECD, 2013). 

 

In the spring of 2019, with the approval of the legislative amendment package in parlia-

ment, Finland adopted the OECD's TRACE model in connection with the imposition of 

withholding tax on nominee-registered shares according to the 2018 proposal of the 

government (Joiniemi, 2021). The law change entered into force on 01.01.2021 and cov-

ered dividends paid out after the above date. The TRACE model was adopted in an un-

revised manner. 

 

Registered custodians are banks or institutions that hold securities (such as stocks or 

bonds) on behalf of investors, including nominee-registered shares. Registered custodi-

ans are intermediaries in asset management in that they hold securities securely and 

settle trades, for instance, receive interest or dividend payments and address the corre-

sponding tax liabilities. 

 

The major role of registered custodians is, in terms of the TRACE model, to investigate 

and decide the final residence state of the recipient of the dividend received on the 

shares subject to custody, with the aid of which the custodian ensures the validity of any 

advantages of tax treaty. The Finnish Tax Administration explains how sufficient due dil-

igence can be evidenced, in terms of the recipient country of residence and accuracy of 

tax treaty benefits, by a withholding tax card of the Finnish Tax Administration, a certifi-

cate issued by the tax authority in the recipient country of residence, and self-declara-

tion of the recipient by the TRACE model. The tax authority emphasizes that the inde-

pendent declaration by the recipient of the dividend must be trustworthy and encom-

passing enough for the declaration to be accepted. 



41 

A notable aspect of the 2019 legal amendment was the introduction of the actual and 

primary tax liability for a registered custodian under the TRACE model in cases where 

insufficient withholding tax was collected. As per Section 10c(2) of the Withholding Tax 

Act, a registered custodian is liable for unpaid taxes due to an incorrect annual report 

“as if it were their own”. This liability only applies to the custodian's own incorrect re-

porting, and not to incorrect information received through the custody chain (Verohal-

linto, 2020). However, under Section 10b(3) of the Withholding Tax Act, the burden of 

proof regarding the involvement of other parties in the custody chain in the incorrect 

annual report lies with the registered custodian. If the custodian can demonstrate that 

the error was caused by another registered custodian in the chain and that they acted 

with due diligence, the tax liability shifts to the previously registered custodian. Thus, 

the registered custodian closest to the dividend recipient is always responsible for the 

tax. The liability also only applies in situations where the custodians themselves caused 

the under-withholding (Verohallinto, 2020). 

 

The simple procedure, which was in use before 2021, did not offer the Tax Administration 

tools to find out the final beneficiary of nominee-registered shares. Thus, it was relatively 

easy for generally taxable and limitedly taxable dividend recipients to evade taxes by 

declaring the home country of the final recipient of the dividend as a country with which 

Finland has a valid tax treaty, which, when applied, reduces the amount of withholding 

tax levied by Finland. The information obtained with the help of the TRACE policy about 

the chain of custody and the final beneficiary contributes to the Tax Administration's 

ability to monitor and find entities that evade withholding tax using trustee-registered 

shares.  

 

According to the new TRACE policy, the primary tax liability of the registered custodian 

for uncollected withholding tax also increases the custodian's risk of withholding tax eva-

sion being discovered. This, in turn, may reduce the willingness to withhold the tax. The 

procedure seems to have produced results, since after the introduction of the TRACE 



42 

policy, several companies have been caught in Finland engaging in withholding tax eva-

sion or trading services that include withholding tax evasion (Verohallinto, 2022c). 

 

Although the amount of dividends paid to non-resident taxpayers has remained rela-

tively stable, the increase in withholding tax revenue collected by the tax administration 

has been a trend in Finland over time. This is because the Finnish Tax Administration has 

moved to improve the monitoring of withholding tax, which has contributed to the 

growth in revenue. In the Tax Administration’s statistics on withholding tax, there is a 

significant, nearly twofold increase between 2020 and 2021, coinciding with the imple-

mentation of the TRACE policy, as noted by the Tax Administration. 

 

The clear increase in withholding tax revenue can be attributed at least in part to the 

implementation of the TRACE policy. The stricter oversight introduced with the TRACE 

policy may have ensured the rightful application of tax treaty benefits, thereby reducing 

the improper use of these benefits (Verohallinto, 2022d). 

 

 

 



43 

4 Research approach 

This chapter outlines the approach taken to investigate trading volumes around the ex-

dividend date, focusing on identifying patterns indicative of dividend withholding tax 

schemes. The analysis is intended for identifying, especially abnormal trading volume 

and evaluating the impact of TRACE on such activity. The methodology is structured 

around the following key steps: data collection, event definition, trading volume compu-

tation, and statistical analysis. 

 

4.1 The data collection 

The dataset includes the daily trading volumes of dividend-paying stocks listed on the 

Nasdaq Helsinki exchange. The analysis focuses on the 100 most active dividend-paying 

companies on Finland’s biggest market, the Nasdaq Helsinki exchange, in the period 

from January 1, 2013 to December 31, 2023 to investigate: (1) whether there are abnor-

mal trading volumes at the ex-dividend date; (2) whether there are changes in abnormal 

trading volume after the implementation of TRACE. 

 

Other information includes dividend-related information such as the ex-dividend and 

record dates. The data was obtained from the Thomson Reuters database to ensure that 

the information received is accurate and comprehensive from the market. The ex-divi-

dend date serves as the main point for the analysis. An event window is established to 

capture trading activity immediately before and after the ex-dividend date, typically 

spanning several days. For this thesis, the event window is defined as [T−5, T+5], where 

T represents the ex-dividend date. This 11-day window allows for the detection of trad-

ing volume patterns that may arise due to DWT schemes. 

 

4.1.1 The data samples 

The data is carefully selected for analysis with two primary objectives: to ensure that the 

data is suitable for statistical analysis and to focus on securities likely to be affected by 



44 

potential DWT schemes. The sample is restricted in three ways to ensure it is comparable 

across different contexts. 

 

The first limitation involves limiting transaction data to trades only conducted on the 

main stock exchange, Nasdaq Helsinki. This decision addresses the lack of detailed infor-

mation about smaller and specialized exchanges, which often consist of volatile or newly 

listed companies with different regulatory environments. It is unlikely for these compa-

nies to distribute consistent dividends, and they are less appealing for DWT schemes due 

to their unpredictability. Furthermore, data from these exchanges is often incomplete 

and unsuitable for statistical analysis. Restricting the data to Nasdaq Helsinki also mini-

mizes the risk of duplicating securities listed on multiple trading platforms. 

 

The second limitation targets firms that issued cash dividends at least once before and 

once after the implementation of TRACE, requiring a minimum of two payments for in-

clusion. This approach allows for within-firm comparisons of trading volumes before and 

after the policy changes. Non-dividend-paying companies are excluded, as they do not 

provide relevant information, and the analysis is limited to cash dividends, as other 

forms of dividends are not known to be associated with DWT schemes. While analyzing 

all dividends in the pre-policy period could offer insights into the extent of DWT schemes, 

the main objective of this thesis is to assess the impact of the TRACE. Therefore, pre- 

and post-policy dividend data from the same companies are prioritized to enable mean-

ingful comparisons. 

 

The third limitation reduces the dataset to the 100 most traded dividend-paying compa-

nies to manage computational challenges. A baseline average trading volume is calcu-

lated using trading volume data from a pre-event period, excluding the event window. 

This baseline serves as a reference point to identify deviations. The formula for average 

trading volume is: 

 

𝐴𝑣𝑎𝑟𝑎𝑔𝑒 𝑇𝑟𝑎𝑑𝑖𝑛𝑔 𝑉𝑜𝑙𝑢𝑚𝑒 =  
∑ 𝑇𝑟𝑎𝑑𝑖𝑛𝑔 𝑉𝑜𝑙𝑢𝑚𝑒𝑡

𝑇𝑝𝑜𝑠𝑡
𝑡=𝑇𝑝𝑟𝑒

𝑁
,     (1) 



45 

 

Where 𝑇𝑝𝑟𝑒 and 𝑇𝑝𝑜𝑠𝑡 define the baseline period, and N is the number of days in that 

period. 

 

Investors participating in DWT schemes are likely to seek minimal attention. Larger com-

panies with consistent dividend payments are considered the most plausible choices for 

such schemes, as recurring dividends enable repeated transactions with minimal plan-

ning, and substantial trades can blend into the overall market activity without raising 

suspicion. In contrast, similar trades in smaller companies might attract attention due to 

their infrequent nature. Therefore, this limitation is unlikely to compromise the validity 

of the analysis significantly. 

 

It is important to acknowledge that these limitations may introduce survivorship bias 

into the samples, making them purposive and non-random. However, given the rationale 

for these limitations, this bias is deemed acceptable. Since the focus of the thesis is on 

evaluating the TRACE policy change, these effects will only be observable where DWT 

schemes have been present. Consequently, the sample is designed to prioritize identify-

ing instances where such schemes are likely to occur if they exist. 

 

In cases where companies have more than one share class, each class is treated as its 

own entity, comparable to individual companies. This approach reflects the fact that dif-

ferent share classes are associated with varying rights and are traded at different prices 

and conditions. Putting the share classes together is avoided to ensure the accuracy of 

the analysis, even though they represent ownership in the same company.  

 

After applying these restrictions, the final dataset consists of daily transaction data for 

the 100 most traded companies listed on Nasdaq Helsinki during the period from January 

1, 2013, to December 31, 2023. 

 



46 

4.2 Statistical analysis 

The primary variable used in the analysis is share turnover, which measures the relative 

trading activity of a stock. Share turnover is defined as the daily trading volume of shares 

traded divided by the total shares outstanding for a given company: 

 

𝑆ℎ𝑎𝑟𝑒 𝑇𝑢𝑟𝑛𝑜𝑣𝑒𝑟𝑖,𝑡 =
𝑆ℎ𝑎𝑟𝑒𝑠 𝑇𝑟𝑎𝑑𝑒𝑑𝑖,𝑡

𝑆ℎ𝑎𝑟𝑒𝑠 𝑂𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔𝑖,𝑡
 ,      (2) 

 

Where 𝑖 represents the company, and 𝑡 represents the specific trading day. Share turn-

over is an effective measure of market activity as it normalizes trading volumes across 

companies of different sizes and adjusts for events such as stock splits or reverse stock 

splits. By expressing trading activity as a percentage of total shares outstanding, share 

turnover provides a consistent and comparable metric for analyzing trading patterns. 

 

This is particularly useful when trying to identify the cum-ex and cum-cum schemes, 

which are characterized by unusual trading activity on the ex-dividend date. An increase 

in the turnover of the shares around this event window may be a sign of increased trad-

ing activity related to cum-ex or cum-cum trading. The focus on share turnover ensures 

that market activity is examined in a way that accounts for differences in company size 

and stock liquidity. 

 

First, the share turnover for all companies in the dataset is calculated. Event windows 

and non-event windows are also divided into trading volumes before and after the im-

plementation of the TRACE policy. The means of the groups are compared using t-tests, 

and the results are used in identifying the existence of abnormal trading activity as well 

as the efficacy of the TRACE policy. 

 

T-tests are statistical tools used to determine whether two groups have significantly dif-

ferent means; this thesis employs t-tests to evaluate the presence of abnormal trading 



47 

volumes and the effectiveness of the TRACE policy. Two primary applications of t-tests 

are implemented: 

 

1. Event window analysis: pre-policy trading patterns 

• The objective is to examine whether there are abnormal trading volumes 

during the event window (5 days before and 5 days after the ex-dividend 

date) or not, compared to non-event periods. 

 

• The t-test is used to compare the mean share turnover in the event win-

dow to the mean share turnover outside the event window. A significant 

difference supports the presence of abnormal trading activity, potentially 

linked to DWT schemes. 

 

2. Pre- and post-TRACE policy analysis 

• The objective is to examine whether the implementation of the TRACE 

policy has reduced abnormal trading volumes during the event window. 

 

• The t-test is used to compare the mean share turnover in the event win-

dow before TRACE to the mean share turnover after TRACE. This analysis 

identifies whether or not TRACE has been effective in preventing DWT 

schemes. 

 

Share turnover and t-tests allow for a systematic method of analyzing trading patterns 

and measuring policy efficacy. The analysis is therefore focused on share turnover to 

identify instances of unusual trading activity, which are more likely to be associated with 

DWT schemes. This allows measurement of abnormal trading volumes around the ex-

dividend date. The t-tests help to compare trading patterns across time periods (pre- and 

post-TRACE) and trading windows (event and non-event windows). Comparisons are 

necessary for identifying anomalies as well as evaluating policy changes. The t-tests are 



48 

employed in order to establish that the differences observed in the trading volumes are 

statistically significant, which helps in eliminating random shifts in the data. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



49 

5 Results 

This chapter presents the results obtained from applying the statistical equations to the 

data sample. It then examines the impact of the TRACE policy, supported by visual rep-

resentations that highlight patterns in share turnover around the ex-dividend date. Fi-

nally, the chapter provides a comprehensive analysis of various legislative measures 

aimed at addressing dividend withholding tax (DWT) schemes. 

 

5.1 Key findings 

This section presents the results from the statistical analysis, focusing on the abnormal 

trading volumes around ex-dividend dates and the impact of the TRACE policy on reduc-

ing DWT schemes. The findings are derived based on the statistical analysis explained in 

Chapter 4.2.  

 

Table 3. Results on share turnover. 

  

 

The mean share turnover in the event window (5 days before to 5 days after the ex-

dividend date) is as follows: 

• Pre-policy period: 1.759% 

• Post-policy period: 1.395% 

 

The decrease in mean share turnover in the event window post-TRACE shows that there 

is decreased trading activity near ex-dividend dates. These findings refute the 𝐻0B hy-

pothesis because the findings show that the TRACE policy effectively reduced the DWT 

schemes, as these schemes often involve heightened trading activity during the event 

window. 



50 

 

The mean share turnover outside the event window, representing normal trading activity, 

as shown in Table 3: 

• Pre-policy period: 1.431 % 

• Post-policy period: 1.203 % 

 

A similar decline is observed in non-event trading volumes post-TRACE, indicating that 

overall trading activity decreased after the policy implementation. While this reduction 

could reflect broader market trends, it also suggests a normalization of trading behaviour 

in the absence of DWT-driven transactions. 

 

The mean share turnover during the event window (1.759%) is higher than outside the 

event window (1.431%), indicating abnormal trading activity that can be linked to DWT 

schemes. The mean share turnover during the event window (1.395%) and outside the 

event window (1.203%) has a smaller difference, indicating that trading activity around 

ex-dividend dates has become more aligned with normal trading activity. The narrowing 

of the gap between non-event and event trading volumes post-TRACE reflects the poli-

cy's likely effectiveness in discouraging abnormal trading. The findings are in line with 

rejecting the 𝐻0B hypothesis because the findings show that TRACE successfully re-

moved the incentives and mechanisms underlying DWT schemes. This also supports the 

fact that certain trading behaviours around ex-dividend dates may be linked to DWT 

schemes, such as cum-cum and cum-ex transactions.  

 

The results of the t-tests provide statistical validation for the observed differences in 

share turnover across event and non-event windows, as well as between pre-policy and 

post-policy periods. These tests were to see whether the implementation of the TRACE 

policy had an impact on trading patterns associated with DWT schemes. Four t-tests 

were done: the first compared share turnover within and outside the event window be-

fore the implementation of TRACE, while the second examined the same comparison 

after TRACE. The third t-test assessed the differences in share turnover during the event 



51 

window before and after TRACE, and the fourth compared share turnover outside the 

event window across the two periods. 

 

 

Figure 3. T-test results as p-values. 

 

P-value is a statistical value which is contrasted with the null hypothesis (𝐻0). It is the 

chance of getting results as extreme as or more extreme than the observed data, given 

that the null hypothesis is true. A smaller p-value implies greater evidence against the 

null hypothesis, i.e., the results observed are not likely to occur by chance. For example, 

p < 0.05 is usually considered to be statistically significant; if p-value > 0.05, then there 

is no sufficient evidence to reject the null hypothesis, and results obtained may be due 

to chance. The p-value does not provide the probability of the null hypothesis or the 

magnitude of effect but estimates the compatibility of data with the null hypothesis. In 

this thesis, the p-values are employed in order to determine the significance of the dif-

ferences between the trading volumes to identify abnormal activity and determine the 

effect of the TRACE policy. 



52 

 

The t-test for share turnover in and out of the event window before TRACE was statisti-

cally significant (p = 0.0026) and demonstrated that there existed abnormal trading 

around ex-dividend dates. This is in line with the hypothesis that DWT schemes explained 

higher trading in the event window of the pre-policy period. The post-TRACE t-test also 

exhibited a significant difference (p = 0.0009) between event and non-event trade vol-

ume, with even the decline in mean share turnover reflecting lower levels of abnormal 

trading after TRACE implementation. 

 

The before-after policy comparison of event-window trading volumes produced a large 

effect (p = 0.0049) that favored a reduction in abnormal activity on ex-dividend days fol-

lowing TRACE. This finding is consistent with the hypothesis that the policy was effective 

in deterring DWT schemes and removing incentives to engage in such activity. But the t-

test for non-event window share turnover before and after TRACE was not significant (p 

= 0.2000), meaning normal tracking patterns were stable. Stability of non-event turnover 

volume means the policy was targeted and did not influence normal market activity. 

 

Overall, the t-tests' results emphasize the effectiveness of TRACE in dampening abnor-

mal event-window trading for DWT schemes. The dramatic reduction in event-window 

trading after TRACE attests to the policy's effectiveness in alleviating the problem. In 

addition, the fact that there were no significant changes in the volume of trading outside 

the event window implies that the policy did not exert other undesirable effects on the 

market. However, the fact that differences in event-window versus trading volumes out-

side of the event-window remain after TRACE implies that abnormal trading activity 

might still be present and may call for further regulation. These findings suggest that the 

TRACE policy targeted efficiently the DWT scheme mechanisms and was aligned with its 

intended outcome. Reduction of anomalous trading volumes is best observable for the 

biggest and most widely traded dividend stocks, commonly being the most immediate 



53 

concern of such schemes. The same fact is supported also by Table 3, representing a 

clear volume reduction after implementing the policy. 

 

The results of this thesis are very similar to previous studies on DWT schemes and their 

impact on trading volumes. Thus, helps in validating the theoretical model of abnormal 

trading patterns around ex-dividend dates. As expected, this thesis found similar results 

of high trading activity in the event window in the pre-policy period as in the studies of 

Buettner et al. (2020) and Casi et al. (2022). These spikes in trading activity are a sign 

that cum-cum and cum-ex transactions based on tax system loopholes are widespread, 

representing DWT schemes. Similarly, the findings in Finland are similar to those in Ger-

many and Denmark, which experienced legislative reforms that effectively reduced 

these patterns. Hence, the presence of such schemes calls for targeted intervention, as 

well as the need to build on previous efforts. 

 

The significant reduction in abnormal trading volumes during the event window post-

TRACE implementation mirrors the outcomes of Germany's 2012 and 2016 reforms, as 

well as Denmark's 2016 measures. TRACE, like these earlier reforms, appears to have 

effectively curbed the abnormal trading activity associated with DWT schemes. This 

aligns with the conclusions of Laturnus et al. (2022) and Wagner and Wei (2022), who 

emphasized the effectiveness of legislative measures in reducing abusive trading prac-

tices. 

 

Furthermore, the fact that there were no significant changes in trading volumes outside 

of the event window in Finland further supports the targeted success of TRACE in ad-

dressing DWT schemes without affecting normal market behaviour. This stability sup-

ports the argument by Casi et al. (2022) that meaningful reforms can reduce DWT 

scheme activity without compromising the overall financial market. This thesis contrib-

utes to the growing evidence base on the effectiveness of legislative measures in com-

bating tax avoidance schemes across Europe by identifying these patterns. 

 



54 

5.2 Legislative measures to combat DWT schemes 

The legality of dividend withholding tax schemes is difficult to determine due to the com-

plexity of the issue and similar problems in many countries. There is a need to clarify 

which institutions should be liable for the loss caused by these schemes in many coun-

tries. The issue blurs the line between financial market concerns and taxation legislation, 

involving both financial and tax authorities. An inability to assign responsibility and com-

munication gaps further complicate efforts to define legality and prosecute such 

schemes. Following the recognition of dividend withholding tax schemes as a genuine 

threat to market integrity in Denmark (ESMA, 2020), there was enhanced communica-

tion between tax and financial authorities. Similarly, until 2016, cum-cum schemes in 

Germany were deemed legal unless they could prove their non-tax nature definitively 

(Büettner et al., 2020; Podkul, 2016). 

 

DWT schemes have been restricted or outlawed by various legislative measures, some 

enacted by the EU (European Banking Authority, 2020) and others by individual member 

states (ESMA, 2020). The legal status of DWT schemes remains complex due to differ-

ences between national legislative systems. Further complicating matters is collusion 

among colluders from different jurisdictions. For example, Germany, Denmark, Austria, 

Belgium, France, Switzerland, and Norway have amended their legislation to address 

DWT schemes. Additionally, increased surveillance and market monitoring have been 

deployed to curb these schemes (ESMA, 2020). 

 

Although Denmark has not aimed directly at the DWT schemes, it has, nonetheless, en-

hanced its examination of dividend refunds greatly (ESMA, 2020). As outlined by Casi et 

al. (2022), these reforms are generally divided into two categories. The first set of 

measures seeks to tighten the documentation requirements to address the loopholes 

employed by cum-ex schemes. The second set of measures seeks to establish a minimum 

holding period for the purpose of obtaining a tax reimbursement certificate so as to pre-

vent short-term transactions that are intended to capture tax benefits from periods 

around ex-dividend dates, such as cum-ex and cum-cum schemes. 



55 

 

In 2021, Finland was the first country to adopt the OECD’s TRACE model (Isomaa and 

Viitala, 2023). The TRACE model is a procedure for the withholding tax of dividends ac-

cording to tax treaties, i.e., how tax treaty benefits can be granted at the time of payment. 

The process, according to the TRACE model, was implemented at the beginning of 2021, 

and its main objective was to transfer the risk of wrongful collection of withholding tax 

from the dividend payers to the shareholders. TRACE seeks to streamline processes by 

eliminating administrative hurdles, reducing costs for all parties, and protecting data pri-

vacy by minimizing the number of parties involved in the dividend withholding tax re-

claim process. 

 

In the early years, the application of the TRACE policy has led to a sharp increase in the 

gross withholding tax collection, as registered custodians have not desired to assume the 

tax risk incurred while claiming tax treaty benefits in doubtful situations (Isomaa and 

Viitala, 2023). Accordingly, there has been an increase in refund applications for with-

holding taxes (Isomaa and Viitala, 2023). As can be seen from the result of the statistical 

analysis, the TRACE has declined significantly DWT schemes. 

 

TRACE also helps to stop the abuse of dividend withholding tax (Isomaa and Viitala, 

2023). With the help of TRACE, the Finnish Tax Administration has more information 

about companies’ share ownership, so the information available to the d