THE EFFECT OF BONUS CAP
ANNOUNCEMENT IN SYSTEMIC RISK: A
SYNTHETIC CONTROL METHOD
APPROACH
Bachelor Thesis
August 2020
Author: Jairo Steven Asuncion Murillo
Student Number: 11382767
BSc Economics and Business Economics
Supervisor: Ekaterina Seregina
1
Statement of Originality
This document is written by Student Jairo Steven Asuncion Murillo who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
2
ACKNOWLEGMENTS
I would like to acknowledge the financial support received from the Secretariat of Higher Education, Science, Technology, and Innovation (SENESCYT) to study my Bachelor in the University of Amsterdam.
I would also like to thank my supervisor, Ekaterina Seregina for her guidance to successfully write my thesis.
In addition, I would like to thank to my parents for their unconditional support, and for understanding and accepting the decisions I take to develop my career.
Finally, I would like to thank to my friend Kevin Alejandro Paredes Yepez and my
grandmother Otilde Bohorquez for being my motivation to not give up in difficult times. Rest in Peace.
3
ABSTRACT
This research aims to study the announcement of the United Kingdom potentially abandoning the bonus cap when the Brexit takes place. For this research, data from 9 non-European OECD countries was retrieved. A sophisticated approach such as the synthetic control method is used to assess the change in systemic risk if the Brexit referendum had not taken place. By April 2017, the bonus cap announcement effect was mixed, and it was not clear whether it had a steady positive or negative impact in systemic risk for the United Kingdom. Further robustness checks were performed to support the validity of these findings.
4
Table of content
1. Introduction……….. 5
1.1. Bonus cap and United Kingdom……… 8
2. Literature Review………. 10
2.1 Systemic Risk Measures………. 12
2.2 SRISK……… 13
3. Data and Methodology……… 14
3.1 Control Units………. 14
3.2 Synthetic Control Method……… 14
3.3 Methodology………. 15 4. Results………. 16 5. Conclusion……… 18 5.1 Further recommendations...……….. 19 References………. 20 Appendix………. 22
5
1. INTRODUCTION
Since the financial crisis of 2008-2009, the compensation methods among financial institutions have become a relevant factor related to financial stability of the monetary system. Regulatory authorities began to consider that the remuneration to bank executives influences the risk-taking incentives from banks. As a result, a stage for a reform strategy on executive pay in the banking sector was set by The Financial Stability Forum in 2009 which proposed that the contribution of financial institutions to systemic risk may be influenced by their incentives to take risk, and fully argued that “The lack of attention to risk also contributed to the large, in some cases extreme absolute level of compensation in the industry.”1
Subsequently, The Financial Stability Board (FSB) Guidelines for Fair Pay Policies and their Compliance Requirements (Principles and Standards, P&S) were adopted internationally in 2011. The P&S required to align employee’s incentives with the long-term profitability of banks through compensation practices. In addition, the P&S advocate for good regulation of compensation and for compensation to be tailored for all forms of risk, to be symmetrical with risk consequences and to be responsive to the time frame of risk. In particular, the P&S aimed to limit skewed incentives which exacerbated the reckless risk-taking that seriously disrupted the global financial system in 2008, leaving companies with less capital to bear losses as threats materialized.2
At this time, The British Prime Minister Gordon Brown, among many other countries, pushed for changes in the “questionable behaviour” inside the worldwide finance industry since struggling financial firms remained with high bonuses while they were receiving bailouts due to the financial crisis.3
Previously, The United Kingdom Financial Services Authority (FSA) adopted the remuneration code, after a review process and having accepted the suggestions of the Turner Report4 and
1 The G-7 finance officials created The Financial Stability Forum (FSF) in 1999 to act as supervisor of the global financial integration.
2 The Financial Stability Board (FSB) is the successor of the FSF created in 2009 to strengthen global financial supervision and regulation by the G-20.
3 Monaghan, “Brown sets out plan for bonus clampdown,” The Daily Telegraph (2009). 4 Turner (2009)
6
the Walker reports5, making the United Kingdom the firs jurisdiction to control payment
policies in 2009. In the same way as the P&S, The United Kingdom Remuneration Code aimed to improve the decision-making process and the reduce the bank executives’ motives to take risk.
Above all, all members of the European Union had to follow the legislations agreed by the executives. Thus, even though the United Kingdom did not agree with a strict compensation regime arguing that bonus cap would have a negative impact in the financial system, it followed the financial regulations stablished by the European Union including the bonus limitation with the purpose of enhancing the framework around incentives to take risk for financial institutions.
Murphy, K. J. (2013) demonstrated that incentives to take risk in the banking sector actually increased because of the Bonus cap proposed by the European Union. Although, Murphy, K. J. (2013) focused only on risk taken by employees (Bank CEOs) which can be interpreted as potential losses for a single Institution or Bank. Losses spread across financial institutions in times of financial crisis, endangering the entire financial system. The propagation of instability generates a structural threat to the system: the risk of impairing the strength of the financial system as a whole, with possibly negative implications for the actual economy.6 The negative
effect of each bank on the financial network will be identified as its contribution to systemic risk.
Moreover, during periods with low asset market uncertainty, financial threats usually develop and materialize through crises. The development of these threats should be captured by a proper systemic measure of risk. High-frequency risk assessments that are mainly relying on contemporary fluctuations in asset prices are inherently deceptive.
It should be noted that the implementation of financial legislation focused exclusively on the vulnerability of an individual entity could not be enough to insulate the financial industry from systemic risk.7 Although, the arrangements of bank managers could be ideally compatible
5 Walker (2009a, 2009b)
6 The distress that spread from structured investment vehicles to commercial banks, to investment banks and hedge funds during 2007-2009
7Adrian, T., & Brunnermeier, M. K. (2016) showed that only a very weak connection exists between the value of a firm at risk and its input to systemic risk.
7
with the desires of stakeholders, they may also produce an inefficient potential danger on society as a whole.8
Hence, Compensation policy is an alternative framework above and above the current legislation to tackle extremely powerful opportunities for financial institutions management to take risks that might not be socially desirable. In fact, Traditional Banking legislation only limit the decision-making of financial companies by enforcing borrowing and lending restrictions.
Furthermore, Kleymenova and Tuna (2018) presented an article more aligned to what this paper is aimed to study. They chose the United Kingdom for their research and found out that after the British legislation that regulates remuneration in financial institutions, Banks in the United Kingdom contributed less to systemic risk. This research was taken as a model to perform the current study and the results are expected to support their findings and extend the analysis to the Brexit event. Hence, the aim of this paper is studying the impact of the Brexit announcement and the potential abandoning of the bonus cap.
Kleymenova and Tuna (2018) performed a Diff-in-Diff methodology and used CoVar to measure systemic risk. In this paper, Aggregate SRISK will be the measure for systemic risk, and it will be assumed that the aggregate systemic risk represents the banking system. Moreover, the synthetic control method which incorporates elements of difference-in-difference and matching methodology will be used to create a synthetic unit from selected OECD countries that will act as the counterfactual scenario of the bonus cap announcements.9
The United Kingdom has shown interesting responses when the European Union introduced legislations into financial markets and has been one of the countries with great toughness to agree with Bonus cap regulation. Furthermore, it became the first country where a referendum to leave the European Union resulted in the decision to resign from the European Block. Therefore, this research will study the United Kingdom which faced the Bonus cap
8 Anginer et al. (2018) proposed that also in the absence of principal-agent disputes at financial firms, there may always be scope for legislation.
9 In the SCM, weights are assigned systematically to create a synthetic control unit, and as in difference-in-difference this control unit will be used to compare to the treated unit and measure the effect of the treatment.
8
introduction announcement in February 201310 and the Brexit announcement with a
potential abandon of the Bonus cap in June 2016.11
The remaining part of the essay will be divided by the progress of bonus cap legislation in the United Kingdom. Followed by the literature review, where the measures of systemic risk will be discussed and the relationship of bank bonus announcement and the risk measure that will be used on this paper. After that, in data and methodology, data will explain how the dataset was set up and the variables used to build a reasonable environment for the event that will be assessed. In the methodology, the procedure to construct the control units based on the Synthetic Control Method and the econometric details will be presented. Finally, the results will be interpreted, followed by the conclusion which includes further discussion and suggestions that will be proposed.
1.1 BONUS CAP AND UNITED KINGDOM
In 2009, The president of France Sarkozy proposed in a “G20” meeting to further strengthen the banking incentive cap. However, President Sarkozy’s expectations that the G20 would commit to a worldwide limit on financial incentive collapsed when the United Kingdom and the United States suggested that the new proposal was too rigorous.12 Nonetheless, the
United Kingdom introduced the modern remuneration structures in accordance with FSB guidelines.13
In 2013, the European Union (EU) established a temporary agreement to restrict the sum of bankers rewards to the level of fixed remuneration, the so call one-to-one ratio; the cap may be raised to 2:1 with the approval of the plurality of stakeholders.14
British bankers did not consider the new regulation was an efficient strategy. It was argued that bonuses are a mechanism in which banks monitor their payroll bills. For instance, during recessions they do not have to remove workers, they only reduce the salaries of the
10 European Union, 2013, Presidency Flash Note: CRD4/CRR, Results of the Trilogue of 27 February 2013 (February 27)
11 The Governor of the Bank of England, a critic of the compensation rules, mentioned the possible scrapping of the bonus cap after Brexit. Traenor, J. (2020, February 3)
12 Jagger and Frean, “Sarkozy back-pedals over his demands for worldwide cap on banking bonuses,” The Times (2009).
13 Exclusively in the financial sector the bonus payments will be deferred over a number of years, and half payment would be in restricted shares.
9
employees by significantly decreasing bonus rewards. The CEO of British Bankers’ Association argued that this went contrary to the recommendations of the Financial Stability Board which suggested that compensation policy ought to be counter-cyclical rather than pro-cyclical.15
Since the Brexit announcement, the compensation limit on bonuses is considered as one of the changes that may be made to the financial legislation after the United Kingdom exists the European Union in March 2019. For the Governor of the Bank of England, one of the reasons is that current regulation makes more difficult to reduce incentives for disappointing results or inappropriate behaviour. 16 The banks reacted to the compensation limitation by providing
“allowances” in addition to wages and rewards to cover for the loss in remuneration induced by the threshold.
Furthermore, in case compensation limit is loosened or withdrawn, British banks could consider it easier to compete with American or Asian financial firms than their European competitors. In addition, London could flourish as a non-EU bank workforce place relative to other EU locations where compensation restrictions apply. However, in order to maintain entry to the European common market, the United Kingdom would undoubtedly need to enforce finance sector regulations that are approximately similar to those in Europe. Any adjustment to the incentive limit could place that position in jeopardy.17
With respect to bonus compensation and competitiveness on the talent market, it is argued that British banks are finding it more challenging to attract and retain top talent to London because of remuneration-level issues in the United Kingdom and factors like salary limit.18 In
addition to this, the regulatory approach in the financial industry has become increasingly relevant since it is expected to grow rapidly in the future.19
15 World Finance Videos. (2013, December 4). Anthony Browne on the EU banker bonus cap | British Bankers'
Association [video]. Retrieved from https://youtu.be/tFtOMfHo4Kg
16 The Governor of the Bank of England, a critic of the compensation rules, mentioned the possible scrapping of the bonus cap after Brexit. Traenor, J. (2020, February 3)
17 Reuters Editorial. (2016, June 28)
18 Brown stated that foreign banks have a competitive advantage over banks under EU bonus cap. World Finance Videos. (2013, December 4).
19 The Governor of the Bank of England foresees that the financial sector would increase in size to almost 20 times the GDP of UK. Traenor, J. (2020, February 3)
10
2. LITERATURE REVIEW
The aim of this paper is to study the potential effects on systemic risk if abolishing bonus cap from the proposals on Brexit with respect to financial regulation takes place. There are several researches on the correlation between bank bonus regulations and how banks react to these regulations, and the spill over effects on the financial system when these reactions take place. The most similar research this paper can be related and aims to complement used an event study methodology, Kleymenova and Tuna (2018) examine how financial markets react to the news of imposing a bonus cap on executive remuneration in the United Kingdom and in the European Union. Their paper performed and analytical study of the new legislation using the cross-country variability in the implementation of the rule, and it further indicates the effect of the new British legislation was greater than in the European Union’s corresponding compensation restriction, which could have represented an unintended aspect of the second policy package. They further examine the response of British banks in terms of executive remuneration, recruitment and risk-taking to a legislation that required salary deferment and linked it to performance-based vestments. The bonus cap helped to minimize structural risk as UK banks are significantly less likely to contribute in and individual way to systemic risk and less vulnerable to systemic risk in the United Kingdom compared to other major UK businesses. In addition, with a robustness procedure it also reported that British Financial Institutions allocate considerably less to British systemic risk than similar American banks did to the American systemic risk. However, it potentially undermined the ability of banks to maintain their CEOs, according to their conclusions.
Further related papers to this research includes Harris, Mercieca, Soane, and Tanaka (2018) laboratory-based experimental research which indicates that bonus cap is strongly successful in reducing risk-taking if and only if the incentive is not conditional on fulfilling a performance goal. Nevertheless, a research at bank level done by Colonnello et al. (2019) shows that banks which compensation is significantly strict show lower adjusted returns and increased risk-taking tendencies. The later report is compatible with the hypothesis that a rise in the fixed-to-variable compensation ratio causes risk-averse administrators to accept greater risks. Moreover, their results do not support the suggestion that after adopting the bonus limitation, banks sacrifice their ability to maintain their most talented executives as the
11
analytical findings actually demonstrate that banks comply with the legislation by providing greater fixed pay for their managers and lower average variable remuneration.
Murphy (2013) address the rise in the degree of fixed compensation as the most evident and unquestionable consequence of capping the variable-to-fixed compensation. He further mentions that risk-taking motivation increases rather than get eliminated when the level of fixed compensation increases because a guaranteed income reduces the sanction attributable to disappointing performance. Lastly and more significantly, he suggests that the rise of the fixed remuneration element raises the probability of bank failure because the market of financial services is highly cyclical.
Likewise, It is argued that the original aim of the year-end cash rewards was not to offer benefits per se, but instead to guarantee that the expense of remuneration will be small in years with poor performance and moderate in years of high profit margins. This pay versatility became especially relevant in the extremely cyclical financial services industry. Although, year-end rewards have recently been specifically used as rewards to compensate workers on the basis of individual, team and business performance.
The later suggestion is consistent as well with Colonnello et al. (2019) findings that the bank’s reaction to the legislation to increased fixed pay was enough to retain their executives. As a consequence, the risk-adjusted bank efficiency decreased aligns the fixed income with diminished motivation to exercise responsibility on executives’ duty.
Furthermore, commercial bank executives justify high bonus strategies because commercial banks provide a growing variety of services typically aligned with investment banks. Thus, in order to become competitive on the same market with investment banks, commercial banks often needed to contend for investment bankers on the job market, which required delivering remuneration deals to line with those of investment banking.
In financial services companies, the best performers usually have limited and extremely advance expertise unique to their sector but not generally to their boss. As a result, financial sector staff become exceptionally flexible at both domestic and foreign markets. This flexibility has raised overall remuneration rates, as commercial and investment banks contend with each other and simultaneously with private equity and hedge funds for limited skills on a worldwide level.
12
To sum up, the effects of ceiling on bonus compensation is debatable and changes among different research. With this in mind, the objective of this research is to contribute to the wide studies on bank bonus regulations, specifically to the effect on systemic risk.
2.1 SISTEMYC RISK MEASURES
Adrian and Brunnermeier (2016) highlight the need for countercyclical prudential supervision, which is establishing a supervisory structure focused on Conditional Value at Risk (CoVaR), a metric that represents the Value at Risk (VaR) of the financial system dependent on an entity in distress. CoVaR evaluates the contagion impact of a bank on the market by the VaR of the network whether each bank is at its own VaR point.
The Value at Risk, financial institutions most famous risk measure, represents the highest potential loss (as a percentage of overall market equity) that a single bank or system may report for a defined confidence level α (i.e. 1 percent) for a determined period of time. It focuses on a particular institution’s vulnerability in isolation. Nonetheless, the risk of metric of a particular financial institution does not inherently indicates its relation to the entire systemic risk; some institutions are systemic on and individual basis: Due to the interconnectedness of these institutions negative spillovers can be produced for other institutions. Therefore, the development of an enhanced metric for the contribution of each financial institution to the systemic risk (CoVaR) is calculated, CoVar is the difference between the VaR of the entire system depending on the event that the bank hits the lowest level of confidence (α) and the VaR of the entire system conditioned on the event that the bank hits the median return.20
In addition, further applied measures of systemic risk such as long-run marginal expected shortfall (LRMES) which calculates how much capital an enterprise requires to survive a structural occurrence (Acharya et al. [2017]) and the expected capital shortfall (SRISK) described as a financial entity's expected capital shortfall conditional on a sustained downturn in the economy (Brownlees and Engle [2017]).
Systemic risk has a time-series and a cross-sectional aspect. Financial firms endogenously take unnecessary risk in time-series where currently calculated volatility is small, generating
13
procyclicality or what is defined as a “volatility paradox”.21 The cross-sectional aspect of
systemic risk relates the spillovers that exacerbate initial adverse reactions. The spillovers may be immediate, via contractual arrangements between financial institutions. However, indirect spillovers are more significant in quantity. The sale-off of properties for all market players with equivalent exposures will contribute to mark-to-market losses.
2.2 SRISK
We are using a measure named SRISK described as the expected capital shortfall of a financial institution subject to a sustained market downturn. SRISK relies on the volume of the company, its degree of leverage, and its anticipated capital loss on the business downturn. The rise in SRISK forecasts potential decreases in industrial productivity and rises in unemployment, and the predictive power of aggregate SRISK is greater with longer horizons. Brownlees and Engle (2017) developed SRISK to create lists of institutions with systemic risk: organizations with the strongest SRISK are the main reason for the financial system to be undercapitalized in periods of crisis. The amount of SRISK among all companies is seen as an indicator of aggregate systemic risk in the financial sector as a whole.22
SRISK assumes that a sustained price downturn is the leading structural incident of the financial crisis. It is normal to correlate the fragility of the financial sector with the contractual capital deficit that the market will experience in periods of crisis according to Brownlees and Engle (2017). The financial sector is especially susceptible to downward price fluctuations due to excessive usage of debt.23
The key distinction between the systemic risk metric that we would use and the majority of market-based systemic risk indexes is that SRISK incorporates equity and balance-sheet data to design a market-based indicator of financial instability, which is the expected capital shortfall of a financial institution on the basis of a systemic occurrence.24
21 Markus K. Brunnermeier and Yuliy Sannikov (2014) termed this phenomenon as the “volatility paradox”. 22 Aggregate SRISK may be thought of as the overall sum of money that the government will have to raise to bail out the financial sector on the premise of the systemic case.
23 The more popular narrative is that if the finance sector is undercapitalized and cannot withstand a substantial slowdown in the economy, it will, in itself, precipitate a decline.
24 The capital shortfall may also be determined by using only the financial valuation of the assets and liabilities. Nonetheless, the estimated valuation of the company's shares represents a reasonable assessment of the company's potential worth, which can vary from the financial value because the assets or liabilities are
14
3. DATA AND METHODOLOGY
3.1 CONTROL UNITS
The impact of the Brexit on the Systemic Risk (SRISK) will be analysed with a data set built from variables retrieved from the Organization for Economic Co-operation and Development (OECD) database. Australia, Canada, Chile, Israel, Japan, Korea, Mexico, New Zealand, and the United States shape the OECD-members donor pool. As Abadie, Diamond, and Hainmueller (2013) indicate, the study will exclude countries which may be influenced by interference in the "treated" region. Thus, European countries are not considered to be part of the donor pool. Moreover, in further SCM analysis the countries which represent the greatest weights are excluded. (table 2)
The systemic risk, SRIKS aggregate for countries, is selected as the outcome variable and it is assumed for simplicity that the aggregate systemic risk represents the bank industry risk.25
The covariates selected adequately represent the national monetary and financial system, as well as the regional macroeconomic growth: the 10-year bond yield, 90-days interbank rate, 24-hour interest rate, the Composed Leading Indicator (CLI), GDP ratio to trend, Consumer Price Index (CPI), harmonized unemployment, import and exports values, and total production value. Such factors describe the existing status of the economies of every nation in the donor pool. (table 1)
3.2 SYNTHETIC CONTROL METHOD
Abadie and Gardeazabal (2003), Abadie, Diamond, and Hainmueller (2010) and Abadie, Diamond, and Hainmueller (2015) developed the Synthetic Method Control to tackle the difficulties of seeking the counterfactual design of a treated unit. The Synthetic Control Method assigns weights to the control units such that these units represent the ideal match to the pre-treatment features of the treatment unit. The inference procedure introduced by Abadie, Diamond, and Hainmueller (2010) and Abadie, Diamond, and Hainmueller (2015) consist of p-value assessment by permutation examination. Using this method, the null hypothesis that the intervention has no effect can be tested. We use the synthetic control
calculated in a different way from the accounting estimates. Moreover, the market perception is forward-looking and can take into consideration circumstances that have not yet happened.
15
method of Abadie, Diamond, and Hainmueller (2015) to show the effect of the Brexit vote and Bonus cap on Systemic risk.
3.3 METHODOLOGY
Following Abaddie et al. (2010) paper, it is assumed that from a data sample which consist of information about J + 1 countries, only the first country faces the intervention of the treatment from period 𝑡0 ∈ {1, . . . , 𝐽 + 1}. Consequently, remaining countries serve as potential monitoring units which are not affected by intervention. Let 𝑌𝑖𝑡𝑁 indicate the potential consequence of concern in the absence of treatment for country 𝑖 in period 𝑡, where 𝑖 ∈ {1, . . ., 𝐽 + 1} and 𝑡 ∈ {1, . . . , 𝑇}. Let 𝑌𝑖𝑡𝐼 indicate the consequence of the treatment for country 𝑖 in period 𝑡 ∈ {1, . . . , 𝑇}. Let 𝑇0, where 1 ≤ 𝑇0 ≤ 𝑇, be the number of pre-treatment
periods. Depending on the anticipation effect, it is possible to reset 𝑇0 to the period when the
first effect of the intervention is assumed to be appreciable, Abaddie et al. (2015). The outcome of the treatment with 𝑡 > 𝑇0 is defined as follow:
𝑣𝑖𝑡 = 𝑌𝑖𝑡𝐼 − 𝑌𝑖𝑡𝑁
𝑌𝑖𝑡𝑁 must be estimated because 𝑌𝑖𝑡𝐼 is observed in the previous equation. The weighted average of the control units with weights 𝑤 = {𝑤2, . . . , 𝑤𝐽+1} with 0 ≤ 𝑤𝑗 ≤ 1 for 𝑗 = 2, . . . , 𝐽 + 1 and
-
∑ 𝑤𝑗 𝑗+1 𝑗=2
= 1
-were defined by Abadie and Gardeazabal (2003), such restrictions help to avoid extrapolation. By means of the given weights {𝑤2, . . . , 𝑤𝐽+1}, the synthetic control estimators
of 𝑌𝑖𝑡𝑁 and 𝑣𝑖𝑡 are:
𝑌̂𝑖𝑡𝑁= 𝑤2𝑌2𝑡+ ⋯ + 𝑤𝐽+1𝑌𝐽+1,𝑡 𝑣̂𝑖𝑡 = 𝑌𝑖𝑡𝐼 − 𝑌̂𝑖𝑡𝑁
Next, the chosen weights that ideally reflect the pre-treatment characteristics of the treated unit and that follow Abadie et al. (2010) criteria are defined as 𝑤∗ = {𝑤𝑤∗, . . . , 𝑤
𝐽+1∗ } and
16
𝑣1(𝑥11− 𝑤2𝑥12− ⋯ − 𝑤𝐽+1𝑥1,𝐽+1)2+ ⋯ + 𝑣
𝑘(𝑥𝑘1− 𝑤2𝑥𝑘2− ⋯ − 𝑤𝐽+1𝑥𝑘,𝐽+1)2
Where the importance of the synthetic control assigned predictors {𝑥11, . . . , 𝑥𝑘,𝐽+1} are
represented by {𝑣1, . . . , 𝑣𝑘} and their weights are chosen to minimize the size of the prediction
error. The value of the root mean square predicted error (RMSPE) changes depending on the country weight 𝑤(𝑣) for a chosen 𝑣, where:
𝑅𝑀𝑆𝑃𝐸 = (1 𝑇0∑ (𝑌1𝑡− ∑ 𝑤𝑗 ∗𝑌 𝑗𝑡) 𝐽+1 𝑗=2 2 𝑇0 𝑡=1 ) 1 2
4. RESULTS
In Fig. 1 it can be seen that the synthetic output follows a more prominent path compared to the real output that converges at the end of 2016: The real output follows a more steady downward trend compare to the synthetic output which shows an opposite direction from 2017 with respect to the SRISK pattern after the event of Brexit referendum. The Aggregate SRISK comparison between the United Kingdom and the synthetic unit seems to be mixed with the Aggregate SRISK of the treated unit being slightly lower than the synthetic unit during most of the period (table 2). Our findings are structured primarily by Canada and Japan, with weights of 0.524 and 0.476, respectively. Whereas the other units were not taken into account by the SCM method. (table 3)
Overall, there is a more noticeable lower aggregate systemic risk on the real output which could be related to the fact that the Brexit referendum did not materialize right after it took place. Thus, United Kingdom bonus regulation was still bound to the European decision. Knowing this, the downward trend of SRISK shown by the United Kingdom is aligned with the Capital Requirements Directive (CRD VI) European Commission, the regulatory basis for regulation of financial agencies, investment funds and their parent companies in all Member States of the European Union and of the EEA.26 Moreover, the remarkable difference between
the synthetic and the real output shown in 2017, where the synthetic results show an upward direction while real results remain going down, can be related to the European Banking Authority to extend the bonus cap to investment firm from January 2017 (This might impact
17
the assumption that aggregate SRISK represents only financial institutions such as banks).27
(table 2)
To verify the robustness of the control group countries with the highest weights were excluded. The outcome is shown in figure 2 and table 3. It can be seen that when restricting the countries with the highest weights, the ups and downs are more pronounced. Surprisingly, when both control units which drive the original synthetic unit are excluded, the new synthetic output also take only two control units into account. Even though, this new output shows huge differences from the real output during the pre-treatment period and also the after-treatment period, the path seems to be similar with respect to increases or decreases in SRISK before the treatment until 2016, and from there onwards the real output and the synthetic output follow a mixed path between them. However, the outcome without Canada is the most similar to the one with the full control units from donor pool. The fact that removing one control unit changes just slightly the pre-intervention fit provides a robustness check conditional to one change, and further restricting control units show a greater impact on the pre-intervention fit.28
Furthermore, we run the synthetic control method including European OECD countries with a high variable-to-fixed remuneration and the synthetic outcome shows a similar trend compared to the original SCM but with a less prominent difference with respect to the real output (figure 3). Although, Abadie et al. (2015) suggested to exclude to exclude from the donor pool countries that may be affected by the intervention of the treated. There was only one country included in the unit weights which is France with 0.344; the remaining was filled by Canada with 0.393 and Japan with 0.263 (table 4). This weights further support the original findings where the main weights to construct the synthetic unit lies on Canada and Japan. Lastly in this robustness check, France was excluded from the donor pool and the unit weights were exactly the same as the original findings with 0.524 from Canada and 0.476 from Japan (table 4). Thus, this augmented SCM are not different from the original findings.
In addition, the SCM was applied to a low variable-to-fixed remuneration country such as the Netherlands and the results show that the aggregate SRISK in the Netherlands performed a
27 Bank of England reiterates concerns over EU bonus cap plans. (n.d.).
28 Since the synthetic unit is built from only two units, it is not abnormal that excluding them the new output would be become biased.
18
steady downward path whereas the synthetic outcome maintain a similar level after the treated period. Such outcome suggests that if a low variable-to-fixed remuneration country such as the Netherlands experimented an event like the Brexit and an announcement of abandoning the bonus cap, its systemic risk would have followed a steady path and not a downward trend which represent a lower systemic risk (figure 5).
Finally, a SCM analysis around the bonus cap introduction announcement in February 2013 was performed and the result suggests that the United Kingdom and its respective synthetic unit followed a mixed path between them (figure 4). Whereas, in the Netherlands after the bonus introduction announcement the aggregate systemic risk was lower compared to the synthetic output (figure 6).
5. CONCLUSION
We examine the impact of bank bonus regulation on systemic risk through the event of United Kingdom announcement of leaving the bonus cap introduced by the European commission after the crisis of 2008. An alternate world built using the synthetic control method proposed by Abadie and Gardeazabal (2003), shows how a specific variable would have developed if there had not been Brexit. Hence, no bonus cap legislation change announcement. This paper aims to contribute to the analysis of systemic risk changes followed after a bonus cap announcement using a sophisticated quantitative analysis tool.
The results are not clear to conclude whether the systemic risk in the United Kingdom increased after the Brexit announcement because real and synthetic outcomes seems to be mixed in the United Kingdom. Nonetheless, a remarkable change comes in 2017 when the British systemic risk maintains a downward pace whereas the synthetic risk presented an upward change. This observation is present in the different SCM robustness check where the units with the highest weights are excluded. However, the SCM performance becomes weaker when performing some of the robustness check.
Although, it cannot be presented a one side conclusion from the results. The SCM output seems to be consistent after a few the robustness checks. And, when applied to a low variable-to-fixed country the treatment seems to be negative as the aggregate SRISK in the synthetic remains close to a horizontal line whereas the real systemic risk shows a solid downward trend.
19
In summary, it was estimated the impact of Bonus cap announcement on systemic risk using the Brexit announcement of United Kingdom abandoning the European bonus cap and the presented results showed a mixed outcome. Nevertheless, the Brexit conditions have not been decided nor taken place, therefore, an accurate empirical analysis of the systemic risk could not be estimated.
5.1 FURTHER RECOMMENDATIONS
The conclusion of this paper relies on the assumption that the variable aggregate SRISK represents only the banking system. However, as it was discussed the variable take into account all financial institutions not only banks. Thus, further work with SRISK separating only Bank data could enhance the validity of this quantitative work. Moreover, the Brexit bonus cap announcement is still in debate since it is a condition for the British banks to be able to operate in European countries. Hence, this work is based on the announcement and its potential materialization but not empirically in the bonus cap change taking place.
20
REFERENCES
Abadie, A., & Gardeazabal, J. (2003). The economic costs of conflict: A case study of the Basque Country. American economic review, 93(1), 113-132.
Abadie, A., Diamond, A., & Hainmueller, J. (2010). Synthetic control methods for
comparative case studies: Estimating the effect of California’s tobacco control program. Journal of the American statistical Association, 105(490), 493-505.
Abadie, A., Diamond, A., & Hainmueller, J. (2015). Comparative politics and the synthetic control method. American Journal of Political Science, 59(2), 495-510.
Acharya, V., L. Pedersen, T. Philippon, and M. Richardson. “Measuring Systemic Risk.” Review of Financial Studies 30 (2017): 2–47
Adrian, T., Brunnermeier, M.K., 2016. CoVaR. American Economic Review 106(7), 1705-1741.
Anginer, D., A. Demirgüç-Kunt, H. Huizinga, and K. Ma. “Corporate Governance of Banks and Financial Stability.” Journal of Financial Economics 130 (2018): 327-346.
Bank of England reiterates concerns over EU bonus cap plans. (n.d.). Pinsent Masons. Retrieved August 2, 2020, from https://www.pinsentmasons.com/out-law/news/bank-of-england-reiterates-concerns-over-eu-bonus-cap-plans
Brownlees, C., and R. Engle. “SRISK: A Conditional Capital Shortfall Measure of Systemic Risk.” Review of Financial Studies 30 (2017): 48-79
Brunnermeier, Markus K., and Yuliy Sannikov. 2014. “A Macroeconomic Model with a Financial Sector.” American Economic Review, 104(2): 379–421.
Cerasi, V., Deininger, S. M., Gambacorta, L., & Oliviero, T. (2020). How post-crisis regulation has affected bank CEO compensation. Journal of International Money and Finance, 102153. Colonnello, S., Koetter, M., & Wagner, K. (2019). Compensation Regulation in Banking: Executive Director Behavior and Bank Performance after the EU Bonus Cap. Available at
21
CRD IV - Capital Requirements Regulation (CRR) - 575/2013 - Open Boek Toezicht. (n.d.). www.Toezicht.Dnb.Nl. Retrieved August 3, 2020, from https://www.toezicht.dnb.nl/5/50-228261.jsp
European Union, 2013, Presidency Flash Note: CRD4/CRR, Results of the Trilogue of 27
February 2013 (February 27)
Financial Stability Forum (2009). Principles for Sound Compensation.
Financial Stability Board. Implementing the FSB Principles for Sound Compensation Practices and their Implementation Standards. 2012-2015.
Jagger, Suzy, and Alexandra Frean, 2009, “Sarkozy back-pedals over his demands for worldwide cap on banking bonuses,” The Times (September 25)
Kleymenova, A., & Tuna, A. (2018). Regulation of Compensation and Systemic Risk: Evidence from the UK. Chicago Booth Research Paper, (16-07).
Lombardi, D. (2011). The governance of the financial stability board. Issue paper,
Washington, DC: Brookings.
Monaghan, Angela, 2009, “Brown sets out plan for bonus clampdown,” The Daily Telegraph (September 4).
Murphy, K. J. (2013). Regulating banking bonuses in the European Union: a case study in unintended consequences. European Financial Management, 19(4), 631-657.
Nistor Mutu, S., & Ongena, S. (2017). The impact of policy interventions on systemic risk across banks. Available at SSRN 2580791.
Reuters Editorial. (2016, June 28). CORRECTED-Brexit could give UK banks bonus flexibility. Retrieved from https://de.reuters.com/article/banks-regulations-bonuses/corrected-brexit-could-give-uk-banks-bonus-flexibility-idUKL1N19G1RV
Treanor, J. (2020, February 3). EU rule capping bankers’ bonuses “could be scrapped after Brexit.” Retrieved from https://www.theguardian.com/business/2017/nov/29/eu-rule-capping-bankers-bonuses-could-be-scrapped-after-brexit-says-bank-boss
22
Turner, A. “A Regulatory Response to the Global Banking Crisis.” Financial Services Authority, March 2009.
Vlab.stern.nyu.edu. 2020. V-Lab: Systemic Risk Analysis Summary. [online] Available at: <https://vlab.stern.nyu.edu/welcome/srisk> [Accessed 13 August 2020].
Walker, D. “A Review of Corporate Governance in UK Banks and Other Financial Industry Entities.” Financial Services Authority, July, 2009a.
Walker, D. “A Review of Corporate Governance in UK Banks and Other Financial Industry Entities: Final Recommendations.” Financial Services Authority, November, 2009b
World Finance Videos. (2013, December 4). Anthony Browne on the EU banker bonus cap | British Bankers' Association [video]. Retrieved from https://youtu.be/tFtOMfHo4Kg
APPENDIX
APPENDIX 1: DESCRIPTIVE STATISTISCS
Table 1. Descriptive Statistics of the variables used for the SCM computation. Source: Author’s computation.
Variables Obs Mean Std.Dev. Min Max p1 p99 Skew. Kurt. date 400 667.5 11.558 648 687 648 687 0 1.798 srisk 400 144000 179000 0 681000 0 628000 1.125 3.003 bondyield 388 2.661 1.549 -.23 7.6 -.085 7.24 .665 3.318 interbank3~h 383 1.599 1.432 .056 6.93 .056 6.28 .9 3.365 leadingInd 400 100.04 .634 98.365 101.806 98.64 101.736 .12 3.098 priceIndex 348 1.633 1.498 -1.004 6.068 -.905 5.387 .656 2.877
exports 400 5.54e+15 1.67e+16 2.33e+09 6.56e+16 2.54e+09 6.14e+16 2.703 8.386
GDPtoTrend 400 100.045 .525 98.173 101.433 98.475 101.308 -.408 3.856
Unemployme nt
374 5.165 1.218 2.8 7.3 3 7.1 -.135 1.824
Overnightr 400 1.401 1.218 -.059 4.49 -.05 4.035 .473 1.921
Imports 400 4.73e+15 1.42e+16 2.54e+09 5.48e+16 2.76e+09 5.19e+16 2.696 8.336
totalProduct 322 100.591 2.294 92.03 106.691 94.535 106.508 .086 3.842
23
APPENDIX 2: OUTCOME RESULTS
Table 2. Synthetic outcome results after Brexit Referendum (Bonus cap announcement). Source: Author’s computation.
Date UK Synth UK – Full Sample
Jun-16 362894 370980 Jul-16 411441 392015 Aug-16 400748 412010 Sep-16 337490 376068 Oct-16 330244 375029 Nov-16 321261 371874 Dec-16 311698 336272 Jan-17 309718 294670 Feb-17 300249 302440 Mar-17 271603 303152 Apr-17 257844 331885
Figure 1. SRISK – Full Sample. Source: Author’s computation based on SCM
150000 200000 250000 300000 350000 400000 450000
Dec-13 Jun-14 Dec-14 Jun-15 Dec-15 Jun-16 Dec-16 May-17
Aggre
gat
e
SRISK
Month - Year
SRISK - Treatment period set on June
24
APPENDIX 3: ROBUSTNESS CHECKS
Figure 2. SRISK – changing control group. Source: Author’s computation based on SCM
Table 3. Country weights computed by SCM - SRISK. Source: Author’s computation based on SCM.
Country Synth UK Canada out Canada and Japan out
Australia 0.000 0.000 0.000 Canada 0.524 - - Chile 0.000 0.001 0.000 Israel 0.000 0.210 0.000 Japan 0.476 0.369 - Korea 0.000 0.000 0.225 Mexico 0.000 0.106 0.000 New Zealand 0.000 0.000 0.000 USA 0.000 0.314 0.775 150000 200000 250000 300000 350000 400000 450000
Dec-13 Jun-14 Dec-14 Jun-15 Dec-15 Jun-16 Dec-16 May-17
Aggre
gat
e
SRISK
Month-Year
SRISK - Treatment period set on June
25
Table 4. Country weights computed by SCM - SRISK. Source: Author’s computation based on SCM.
Country Synth UK France out
Australia 0.000 0.000 Canada 0.393 0.524 Chile 0.000 0.000 France 0.344 - Greece 0.000 0.000 Hungary 0.000 0.000 Ireland 0.000 0.000 Israel 0.000 0.000 Japan 0.263 0.476 Korea 0.000 0.000 Mexico 0.000 0.000 New Zealand 0.000 0.000 Norway 0.000 0.000 Poland 0.000 0.000 Spain 0.000 0.000 USA 0.000 0.000
Figure 3. SRISK – Treatment including European countries. Source: Author’s computation based on SCM 150000 200000 250000 300000 350000 400000 450000
Dec-13 Jun-14 Dec-14 Jun-15 Dec-15 Jun-16 Dec-16 May-17
Aggre
gat
e
SRISK
Month - Year
SRISK - Treatment including European countries
26
Figure 4. SRISK – Bonus Cap announcement for UK. Source: Author’s computation based on SCM
Figure 5. SRISK – Full Sample for The Netherlands. Source: Author’s computation based on SCM 150000 200000 250000 300000 350000 400000 450000 500000
Oct-10 May-11 Dec-11 Jul-12 Feb-13 Sep-13 Apr-14
Aggre
gare
SRISK
Month -Year
UK - Bonus cap announcement
UK Synth UK 0 10000 20000 30000 40000 50000 60000 70000 80000 90000
Dec-13 Jun-14 Dec-14 Jun-15 Dec-15 Jun-16 Dec-16 May-17
Aggre
gat
e
SRISK
Month - Year
SRISK - Treatment for Netherlands
27
Figure 6. SRISK – Bonus Cap announcement for The Netherlands. Source: Author’s computation based on SCM 50000 60000 70000 80000 90000 100000 110000 120000 130000
Oct-10 May-11 Dec-11 Jul-12 Feb-13 Sep-13 Apr-14
Aggre
gat
e
SRISK
Month - Year
