The Unbundling of Dealing Commissions
Pepijn Kluin
Pepijn Kluin
10386513
28-‐06-‐2015
Bachelor thesis Economics and Finance
Universiteit van Amsterdam
Supervisor UvA: Patrick Tuijp
Supervisor Kempen & Co: Mark van der Plas
Statement of Originality
This document is written by Student Pepijn Kluin who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents
1. Introduction
This thesis examines the effects of the unbundling of dealing commission. The commission and service payments of the last three years from clients of a full service broker are observed. Because there are already clients who are paying an unbundled commission tariff, a comparison can be made with the clients who did not. This gives a first impression of the potential effects of a switch of the industry to unbundling dealing
commissions.
In previous studies performed by Edmister and Subramanian (1982) the
commission fees are taken as a constant across brokers because of their inflexibility. In other studies of Keim and Madhaven (1997), Jones (2000) and Goldstein et al. (2009) the service provided isn’t taken into account as a determinant. The unbundling of these commissions should give more insight on the determinants of the commission fee. Further there not many studies that make a distinction in low cost and full service brokers. As this study is done at full service broker this contributes to a more profound view on the brokerage commissions. Finally the study also shows whether the unbundling of commission addresses the problem of research service payments being
linked to volume. This is a relatively new field due to regulation.
For a long time low cost brokers and full-‐service brokers co-‐exist. Hence the full-‐
service brokers offer much higher dealing commissions, then the low cost broker. The reason for this is that the commission of the low cost broker only exists of execution cost whereas by the full-‐service broker the execution cost and services are bundled in the commission. (Goldstein et al., 2009).
Therefore commission prices that institutional investors pay at full service brokers are often inflexible and negotiated as a package deal upfront. As consequence of the inflexible tariff the service cost the institutional inventors make increase with the traded volume. If the research costs are for instance bundled in the dealing commission then they pay more for research when they are trading more as an investor.
This leads to service costs not being transparent (Financial Conduct Authority,
2014). Given this lack of transparency the Financial Conduct Authority (FCA) noted a conflict of interest between the clients of the institutional investors, who bear the transaction cost, and the investment managers. The investment managers take less scrutiny for services paid by management fee then for services paid by their own
resources (Financial Conduct Authority, 2014).
To address these problems the regulation for research payments by investment
funds is being revised. The FCA and ESMA are both advising to unbundle the payments for services from dealing commission. The European Commission (2015) has also implemented this advice in the first draft for MiFID II.
Through regression analyses is shown that from the clients who have unbundled their commission payments the service payments are still linked to traded volume. Further the clients with an execution only tariff paid significantly less for research than the client with a bundled tariff this. They paid € 75,000 more in terms of cheque payments and 5.7 basis points more for research services. Moreover the size of the clients was only an important determinant of the amount of research paid if measured in basis points. Small size clients paid on average 4.7 basis points more for research.
For execution there is no significant difference between clients who unbundled their commissions paid and clients with a bundled fee. This is corrected for the size of the client in terms of order flow, which is negatively correlated with the fee on a 5% significance level. Also the volume traded was an important determinant. The liquidity provision did not play a significant role in the determination of the execution tariff while this should according to previous research. However the broker confirmed they intend to start using it as an indicator.
The results show that the research and service payments after unbundling the
dealing commissions are still linked to traded volume. However the new regulation does address some other concerns of the FCA (2014) and the ESMA (2015). Table 2 and Table 3 show that the bundled clients pay more for research services than the clients who unbundled their dealing commissions. Thus it could be that the new regulation does improve the scrutiny on the payments for research. Further small clients do not pay more for research when the dealing commissions are unbundled.
2. Literature Review
2.1 History
Dealing commission is the amount of money paid by investors to their broker for doing a transaction for them. Before 1975, dealing commissions were highly regulated. The commission consisted of a fixed price. This way it was difficult to compete for brokers among each other on price so they began offering extra services besides the execution of a trade. This way the commission paid became a bundled cost, which included the
transaction cost as well as services (Goldstein et al., 2009).
After the deregulation of the fixed price for commission cost in 1975 the commission costs lowered substantially. As can be seen from Figure 1 in the study of Goldstein et al. (2009) the average commission paid in the late 70’s was around 1.2% where at the beginning of 2000’s it was around 0.2%. Second low cost brokers arose because institutional investors were able to unbundle the execution cost from other services provided by the brokers. From a sample of 25,643,364 NYSE listed trades by institutional investors Goldstein et al. (2009) found that 40% was executed by discount brokers.
This means that the largest part of trades is still done by full-‐service brokers who bundle their cost. The FCA estimates that of the three billion dollar dealing commission paid by institutional investors half is spend on research (2014). These cost are directly borne by the costumers of the investment funds. This customer is already paying for the in-‐house research of the investment fund through the management fee. Bergowitz (1988) proved that the impact of commission costs are significant and thus need to be
taken into account by the investment manager.
Because the research and transaction cost are bundled, the price for research is unclear. Here arises a conflict of interest between the investment manager and his customer. The FCA found from research in 2011 and 2012 that only a few firms preserved the same standards of control for the research paid through dealing commission as for the research paid by their own resources. Further firms paid for research on trade-‐by-‐trade basis what would mean the amount spend on research would depend on the trade volume. That’s why they want to unbundle the research cost from the execution part. Besides this conflict of interest the FCA thinks that the market
for research is inefficient and lacks competition. This is also caused by the unclear price for research. By implementing new regulations the FCA wants to address this problem.
2.2 Regulation
On 2 June 2014 the Financial Conduct Authority published the new guidelines on the use of dealing commission. These guidelines the FCA are only intentional for investment
managers settled in the United Kingdom. But because the FCA is also designated with
the design on inducements for MiFID II, the FCA guidelines will be a base for MiFID II. Research is part of inducements so the FCA is deciding on what to do with the bundled
dealing commission.
The new rules in the UK demand investment managers to use dealing commission costs, paid by clients through the management fee, only for execution and substantive research. Where substantive research is research that provides meaningful conclusions based on analysis or manipulation of data. And for all research service payments investment managers should be able to justify the decision to acquire the service. This would improve the accountability of the investment managers by their
clients.
Further the unbundling of the commission is required. The new rules direct how asset managers should address mixed-‐use assessments. In mixed-‐use assessments goods and services that cannot be purchased through dealing commission, such as corporate access or research that is not substantive, are bundled with execution and substantive research goods and services. In such circumstances an asset manager should unbundle these elements so that its clients only bear the costs of execution and substantive research. Subsequently the investment manager pays separately for the corporate access and research. This way there will be a clearer price for research and as a consequence a more efficient market.
On 19 December 2014 The European Securities and Market Authority (ESMA) wrote their advice on MiFID II. The ESMA acknowledges just like they FCA the conflict of interest between the investment manager and his customer. Further it also recognizes the lack of efficiency in the market for research. It only addresses the problem in a
The ESMA suggests that research must be paid either directly. Or the investment fund has to set up a separate research account funded by specific charge to their clients. The direct payment would require the investment fund to pay out of its own resources by either absorbing the cost themselves or increasing their management fee.
If research is paid through the specific research account it has to meet the
following conditions. The account is only funded by the specific research charge to the client. Only be based on the research budget set by the investment firm for the purpose of determining the need for outside research. Further the payments should not be linked to trading volume and should not exceed the predefined amount of the account.
The allocation of the research budget to purchase third party research should be
subject to appropriate controls and senior management oversight to ensure it is managed and used in the best interests of the firm’s clients. Such controls include a clear audit trail of payments made to research providers and how the amounts paid were determined with reference to quality criteria. The allocation of the research account
should be explained to their clients on the hand of these quality criteria.
Though the ESMA admits that under the current legislation research qualifies as
inducement. Under their proposed new rules it would not. This is in contrast with the FCA who does see research as an inducement. Nevertheless the want clearer conditions
under which the receipt of research would not qualify as inducement.
On 19 February 2015 the FCA reacted on the ESMA’s advice for MiFID II. The think that the ESMA’s final proposals will lead to increased accountability over the expenditure on third party research by the portfolio manager. ESMA’s reform proposal is based on the principle of requiring a full separation of portfolio managers’ payments for third-‐party research (and how they account for these costs to their customers) from the execution of orders with brokers. ESMA’s final proposals will remove the inducement and conflicts of interest that can otherwise arise for portfolio managers in their arrangements with brokers.
Further the FCA believes that the separation of costs will improve the ability to monitor best execution, and given the existing competition between brokers and venues for execution-‐only business, they think it is unlikely that firms could maintain higher
transaction fees to subsidize other services.
At the 16 April of 2015 the European Commission (EC) issued a draft for the new
conditions the research account must meet. Investment managers can increase the fees through client notification instead of a written agreement of the client. Further there is no change in policy of banning the linkage between research payments and commission. For instance the commission sharing agreement is still to be forbidden in the draft. The EC has intends to publicize the final rules in June so it’s not foreseen that there hard line on research payments will change.
2.3 Implications 2.3.1. Dealing commission
First of all it should be clear that the average price of dealing commission should decline. Now the regulators demand to pay for research services separately the dealing commission is only presented by their execution cost. The level of these execution cost
could be determined by the difficulty of the trade.
Edmister and Subramanian (1982) investigated the determinants of the difficulty
of a trade. The found several factors having negative correlation with the amount of commission per share paid. Share price a higher share price usually causes an order where fewer shares are needed to buy. Second the size of the order; due to economies of scale a larger order lowers the cost. On the other hand a bigger order size will require
more capital requirements of the broker (Goldstein et al., 2009).
Third liquidity, higher liquidity in a stock makes the price effect, which is trying
to be minimized, smaller. This liquidity provision plays a crucial role in the added value of high touch brokers against the discount brokers. The high touch brokers are characterized by the doing orders via the phone. This in contrast to low touch brokers who do all their orders electronic. The advantage of telephone calls is that the investor
can be kept up to date of the latest news, order flows and insights.
Trading a block, which the broker takes at his own book or wherefore the broker
finds a counterpart, is valuable for the institutional investor because there is no price influence this way. Loeb (1983) also showed that the cost of trading increase with the size of the fill of the trade. The bigger the block relative to the companies market capitalization the bigger the spread/price cost. Even under the firms with the largest market capitalization a block of $5.000 had a cost of 1.1%. Also the speed of execution,
which is a lot faster than with a market order, is valuable to the investors.
Schwartz and Steil (2002) investigated what the Chief Investment Officers (CIO’s)
of the seventy-‐two funds surveyed valued most. Little or no market impact, speed of execution, not revealing the full size of the order and not revealing the identity of the company were the most important factors. This indicates that block trades over the counter are of added value to the investors. Blocks as mentioned do fulfil all the characteristics. Randi and Odegaard (2006) also prove that trading cost on crossing networks are lower than trading cost on exchanges. Therefore the broker should receive more commission on trades where this kind of liquidity is provided. The investor makes use of the specific network of the broker or the broker takes on the risk by doing the trade against his own book. Herewith the broker is minimizing the market impact and
execution time.
Later study by Goldstein et al. (2009) defines other important determinants.
Their findings are that size of the investment fund and the previous commission per share paid are important factors determining the commission per share paid. Size is significant because larger parties have more bargaining power because they could bring more order inflow (Goldstein et al., 2009). Not only because more order flow assures more commission but also because other flow attracts order flow from other clients.
This way the ability to do over the counter trades rises.
This implies that the big institutional investors pay lower commission cost
and use more brokers because it’s easier for them to be relevant for their broker. The smaller investors will need to concentrate more flow to acquire premium services (Goldstein et al., 2009). Previous commission price is significant because opportunities come and go in between seconds. Therefore it’s easier for the broker and the investment
fund to have long-‐term contracts on the commission per share paid.
Due to these long term contracts between the broker and the institutional
investor, changing market conditions don’t play a role in the determination of amount of basis points. Therefore market volatility, market liquidity and interest rates all relevant factors determining market liquidity will not be accounted for in this study (Chordia et all., 2001). However firm specific factors as mentioned above do have an impact on the
basis points paid per trade.
Concluding under the new regulations the predictions are lower basis points paid
it’s still efficient for both parties to set up a long term contract. On the other hand the commission don’t have to be as low as the tariff of the discount brokers. Because in the negotiations the broker could argue that specialised execution in illiquid funds and executing large blocks OTC is value adding. Further the basis points paid could vary
across size of the investors.
2.3.2. Research market
First of all there are different kinds of research services. Maber et al. (2014) identified three kinds of research services: (i) published research i.e., widely spread written communications without personal contact; (ii) concierge services i.e., scheduled communications that arise from the co-‐location client investors and corporate managers; (iii) high touch services i.e., private, personalized, and interactive communication between analyst and client investor.
The price for these research services is not clear in the old situation. However research should have a price because it does at value. Green (2006) found that the client who had first access to the brokerage firms new research report got an outperformance. For roughly two hours after the report the outperformance was about 1% for buy recommendations and 1.5% for sells. However this alpha measure is one way to value the research. Each investor attaches value to different services. Some are concerned with generating alpha while for others corporate access is important. Besides the timing and time horizon of the investor are also important factors.
The FCA estimated the price for all the research services to be half of the bundled three billion of dealing commission. However under the unbundling of dealing commission investment fund needs to value the research on the hand of quality measures. These quality measures could follow the same process that results in a broker. Or could the institutional investor would require research providers to set an upfront price for an agreed level and quality of goods and services supplied, which they
would review on an ongoing basis. A broker vote is a rating of the value of the brokers
services (Maber et al., 2014). In the process investment managers, analysts and sometimes traders of the investment firm are surveyed on the value of the broker’s research services. On the basis of the number of votes received the research provider gets an amount assigned from the research account. This needs to make sure that the
spend research isn’t linked to volume.
However it could be that the price for research is best determined if the research
payments are linked to traded volume. Brennan and Chordia (1993) found that under some conditions the price for research is better determined by selling research through dealing commission then through a direct payment upfront. For example on a good buy advice, where the research found a large difference between the current and target price, the institutional investor will buy more shares. Because it takes longer before the current price has reached the target price. This way larger trading volume is coupled with good advices. And so the two parties share the risk.
Maybe that’s why a study released by TABB Group on 5 May 2015 showed that 86% of the buy side heads of trading think that their research payments will decline under the new regulations. TABB surveyed heads of trading of 50 European investment fund managing €25.5 trillion in assets. Also a Although they are going to cut on research spending this will mostly be done at brokers where there is a lot of low touch flow linked to research. The high touch broker where specialized order flow is allocated to research isn’t so likely to be overpaid.
Another reason could be that the bigger investment funds are going to insource their research. They already have their own analyst and could expand this section. However the smaller investment funds have not enough money to hire analysts. And under the proposal of the ESMA setting up a separate research account would be too much of an administrative burden (ESMA, 2014). Therefore they are likely to cut on their research spending now they are forced to pay out of their own resources.
There are also concerns that due to this decline, especially the small and
medium sized enterprises (enterprises with a market cap of less than 200mln SMEs) will be negatively affected. It will be no longer profitable to coverage research for SMEs. Therefore the SMEs will have more trouble raising capital (FCA 2014, ESMA 2014). However the FCA (2014) thinks that also under the new regulations there is research on SMEs. The increased transparency will increase competition and specialization. This would favour the smaller research firms and brokers who supply more research on SMEs.
Finally there are concerns about the value added tax. These taxes are not levied on the bundled dealing commission. However separate research payments to third parties are subject to value added tax. This would cut research spending even more. On
the other hand it creates a level playing field for specialized parties who only deliver research.
Concluding the spending on research is going to decline under the new regulations. The unbundling will lead to more price transparency and greater accountability of the portfolio managers. This will lead to investment managers applying more scrutiny to research spending. On the other hand the increased price transparency will lead to more specialization and improved quality (integrity research, 2015). The broker is already highly specialized and covers a lot research on SMEs and high touch services. This would mean they are well positioned and the impact of the decline in research payments will be small
3. Data
I received all the data of the brokers’ clients. From these clients a random selection was made from the clients who had a minimum tiering of gold. Clients receive at the broker a tiering of platinum, gold, silver or bronze. This tiering is based on the amount of revenue a client had last year or the potential revenue. This resulted in a selection of 27 clients. All the transaction from 2012 until 10 April 2015 are retrieved. These transaction data contain the following specifics:, Date, ISIN Code, Buy/Sell, Volume, Average price and turnover.
The venue on which a trade is done is coming from another system and is only available for 2014. From each trade is sorted out on which platform the trade is done. These platforms are grouped in four different categories. Primary market, these are the markets where the stocks are initially issued and listed. Multilateral trading facilities (MTF), these platforms are an alternative to the primary market and are subjected to less regulation. Dark pools, these are also alternative trading platforms where the order book isn’t visible. Even after the trade the counterparty isn’t discovered. And last over the counter (OTC Cross). These are trades against the own book of the traders or directly to other parties without intervention of a trading platform.
All trades other than equity are deleted. Also the trades who cancelled and therefore did not earn any commission are deleted. All amounts are converted to euros with the matching end of the date quote in the currency pair. After these corrections 10,985 trades remain. All the cheque payments are also converted to euros. These payments are direct payments, which are not in the commission fee paid on an annual basis.
From Bloomberg all the daily volumes of all the 445 stocks traded in are retrieved. I got all the daily volumes from 01-01-2012 until 10-04-2015. The volumes are of the primary
listing of the stock.
The size of a client and level of research services provided to a client are also important. These factors are used in a several hypothesis. To determine the size clients are sorted into three classes. According to their traded volume over the year clients are small, medium or big. To determine the research level the broker provided the amount of calls made to clients by analysts or sales marketing people. Further the number of niches where research reports are provided, number events, marketing events and roadshows are retrieved.
Figure 1 shows that 87% of all the commission tariffs paid on trades are between the 5 and 20 basis points. These basis points tariffs are in line with the findings of Goldstein et al. (2009), Jones (2000) and Keim and Madhaven (1997). Goldstein et al. (2009) reports that 90% of the commissions paid on trades are between the 5 and 15 basis points. Jones (2000) finds an average commission tariff of 12 basis points. Keim and Madhaven (1997) reported average commissions tariffs of 20 basis points. Therefore the tariffs used in this thesis are for representative for the brokerage industry.
Further Table 1 shows that the average share price of 30,00 is a bit lower than 40,00 reported in Goldstein et al. (2009). However the average amount of shares traded is much higher than in Goldstein et al. (2009). Where Goldstein reports an average amount of shares of 2,327 the average amount of share traded in this sample is 93,812. This could be due to the nature of the broker who is specialized in block trading. The average volume traded per client of 118,117 million is line with the categories one and two in terms of size in the study of Goldstein et al. (2009).
Figure 1: The amount of basis points paid per trade.
This figure shows the frequency of the amount of basis points paid per trade. Those are both bundled and unbundled commission tariffs.
0 500 1000 1500 2000 2500 0.0001 0.0003 0.0005 0.0007 0.0009 0.0011 0.0013 0.0015 0.0017 0.0019 0.0021 0.0023 0.0025 0.0027 0.0029 0.0031 0.0033 0.0035 0.0037 0.0039 0.0041 0.0043 0.0045 0.0047
basis points tariff
Table 1: Descriptive statistics
This table shows the number, mean, median, standard deviation, minimum and maximum of the variables used in the regressions. The research payments of clients paying by cheque measured in euros. The average amount of shares traded per year. The average percentage of daily volume traded per year. The average stock price traded per year. Average volume traded measured in euros. The average volume traded by bundled + cheque account measured in euros. The average amount of basis points paid for research per year. The average volume traded by bundled account measured in euros. The amount of cross trades over the counter as percentage of traded volume. Service provided to the client measured with the dummy variables medium service and premium service. The size of the investors measured with the dummy variables small or medium. The sort account measured with the dummy variables bundled or bundled +cheque.
Variable N mean median st dev min max
Research payments of clients paying by
cheque € 46 112,004.51 100,687.73 67,424.40 0 278,172.50 Shares 106 93,812 70,359 78,230 10,567 532,368 %Mkt 106 50.5% 31.98% 119% 1% 1223% Price 106 29.84 27.4 10.9 14.77 84.18 Volume € 76 118,107,863 97,054,285 87,488,188 9,690,735 483,636,373 Bundled + cheque * Volume 32 100,962,337 81,063,131 73,323,730 9,690,735 329,517,884 Execution basis points 106 0.073% 0.074% 0.021% 0.021% 0.120%
Research basis points 76 0.107% 0.080% 0.076% 0.000% 0.414%
Bundled*volume € 30 131,676,971 106,017,981 103,801,708 19,735,880 483,636,373 Liquidity provision 27 65% 65% 22% 21% 98% Dummies Medium service 27 30% 0 0.460 0 1 Premium service 27 37% 0 0.488 0 1 Small size 27 33% 0 0.476 0 1 Bundled +Cheque 76 41% 0 0.495 0 1 Medium size 27 33% 0 0.476 0 1 Bundled account 76 79% 0 0.410 0 1
4. Empirical Method
It would be wise for the industry to anticipate on the unbundling of dealing commission. Given that the FCA as well as the ESMA is supporting unbundling and this will become the standard for the coming years. By already getting an insight in the current execution and research tariffs the clients are paying. The broker could take initiative in the
negotiations with clients.
To get insight this paper examines the current payments structures, examination
of 27 big clients. There are currently five possible payment structures for investment funds. (i) Execution only: the basis points paid on a trade are only represented by the execution cost, no other services are provided. (ii) Execution only + check payments: these funds have already unbundled their commission payments. They pay an execution only tariff per trade and pay for research separately in form of a check. (iii) Bundled tariff: The commission tariff is represented by the execution cost as well as the research services provided. Sometimes also check payments are done along with the bundled fee. (iv) Bundled + cheque: the client is paying a bundled tariff and pays by cheque for the research. (v) Dual tariff: clients pay a bundled fee until the research budget is reached and then switches to an execution only tariff.
First the execution only tariff needs to be determined. Long-‐term contracts are made for the amount of basis points paid per trade. For this reason the execution only fee doesn’t per trade. Thus assumptions have to be made how much of the fixed fee is for execution. The clients who already have an execution only fee are a good measure of what the fee could be. Further there are a few clients who have a dual tariff and one client who made break-‐up in the bundled fee. I will use all of these as a base for the execution only tariff plus the client with the dual tariff as basis for the execution only fee in the bundled tariff.
Then the research payments will be determined. For the clients paying by cheques this is straightforward. For the bundled tariff I’m using the cheque payments and the execution only fee, after subtracting the execution only fee from the bundled fee and multiplication of the remaining fee with the volume the research payment in the bundled fee are derived. After this the cheque payments are added and the research payments of the bundled clients are complete.
According to the ESMA and FCA volume shouldn’t be a determinant for the clients who are paying by cheque for their research. Cheque payments indicate that client has made valuation of the research and pay according to that. This is one of the main reasons for the unbundling of commissions. To check whether this is true a regression is performed on cheque payments of the execution only accounts and the account who pay bundled tariff + cheques. An additional factor ”service level” is added because clients can receive different service level at the broker. There are three service levels: premium, medium and low. A higher service level should result in higher research payments.
Clients receiving premium service should pay more for research.
The service level is determined on the amount of telephone calls from sales
marketing, the number of telephone calls from analysts, number of road shows participated, number of events participated and the numbers of niches in which research reports are provided. In each of these categories clients are ranked according to their use of the specific service. The ranking of the research reports weighs twice as
heavy as the others factors.
Also the size medium or small is taking into the regression. The size of a client is determined on amount of order flow brought in over the last three years. This way the potential of the client, which is also important, is clearer relative to only last year’s volume. The clients will be split up in three classes, big medium and small. The dummy variables medium and small will be taken into the regression. Herewith can be seen whether small clients pay relatively more for research because they don’t have the opportunity to insource their research (Markit, 2015., Schwartz and Steil, 2002., TABB Group, 2014). Finally to see whether there is a difference between clients who have an execution only tariff and who have a bundled + cheque account the factor is added. The regression equation is given by,
Research payments by clients paying by cheque it = α + β1 traded volumeit + β2 medium size i (1)
+ β3 small size i β4 premium service i + β5 medium service i β6 bundled + chequeit +
ε
it ,
where i identifies the company and t identifies the year.
The second model investigates the second concern of the FCA and the ESMA. According to the regulators the clients who are paying with a bundled fee are paying significantly
more for research than the clients with an execution only fee.
So unlike the previous regression the basis points paid for research and services instead of the euro amount paid are used as dependent variable. As it is known that the payments of the bundled clients are linked to volume the relative price, the amount of basis points is a better measure. By using “bundled” as a dummy variable in the regression a comparison can be made between the bundled account and the execution only account. Other factors used in the model will be the service level and the size of the company. This again will show whether the concern that the small firms will pay relatively more for research and services than the large firms is true (TABB Group, 2014., MARKIT 2015., Schwartz and Steil, 2002.). Further to see whether the volume has a different effect for bundled clients than for the execution only clients the interaction
variable “bundled*volume is added. The regression equation is given by,
Basis points paid for researchit = α + β1 bundled account it + β2 volume it + β3
(2) Bundledit*volume it + β4 premium service i + β5 medium service i +
β6 medium size i + β7 small size i + β8 bundled + chequeit +β9 (bundled + chequeit)
*Volume +
ε
it,where i identifies the company and t identifies the year.
The third regression model investigates which factors determine the execution tariff. Just as in the studies of Golstein et al (2009) and Edminster and Subramanian (1982) the price volume and amount of shares are added. Further the liquidity provision should be in the model. This is measured in two ways, first the volume traded over the counter against another party or against the book of the broker (OTC Cross) as percentage of the total volume traded. The studies of Loeb (1983), Randi and Odegaard (2006) and Schwartz and Steil (2002) confirm this as an important measure of providing liquidity. Second the average percentage of daily volume traded. This is also taken as a determinant in the study of Goldstein et al. (2009). Also like the study of Goldstein et al. (2009) the size is taken into the model. This is measured the same as in the above regressions. Finally to see whether the small brokers like the broker where the research is performed would be hit harder if all clients would switch to execution only the factor execution only will be added. This is predicted by surveys of TABB Group (2015) and Markit (2015). The regression equation is given by,
Execution only tariff it= α + β1 execution only accountit + β2 big sizei + (3)
β3 medium sizei + β4 otc crossi+ β5 Mkt%it + β6 volume + β7 price +