The Effects on LIBOR-Based Mortgage Rates from the LIBOR Scandal

The Effects on LIBOR-Based Mortgage Rates from the

LIBOR Scandal

Bachelor Thesis

Specialization: Finance and Organization Field: Finance

Abstract

This thesis is about the impact of the LIBOR Scandal and how it influenced the mortgage market, with consideration of the financial crisis of 2007-2008. It discusses if the scandal

resulted in any real change of individual borrowing for a mortgage in the period of the scandal. It continues on to elaborate on what the impact of the LIBOR had overall on markets and a brief discussion of what could replace LIBOR. In section two, the scandal, its effects and the mortgage market are discussed from an theoretical point of view. Following this, the third section conducts empirical research to provide evidence on the impact of LIBOR on markets, whilst section four analyses and discusses the data. Section five concludes.

Table of Contents

Page

Chapter 1: Introduction

4

Chapter 2: Literature review

7

2.1

The London InterBank Offered Rate (LIBOR)

7

2.2

Mortgage Rates

9

2.3

The Role of LIBOR on Mortgage Rates

10

Chapter 3: Empirical Approach

13

3.1

Methodology

13

3.2

Equations and Approach to the Research

15

Chapter 4: Analysis and Discussion

17

4.1

Results

17

4.2

Discussion

18

Chapter 5: Conclusion

21

References

23

1. Introduction

The financial world had been under a significant amount of stress following the global financial crisis of 2008. We had just experienced some of the worst economic conditions in almost a century and many banks had just received bailouts for their callous actions regarding the subprime mortgage market, which led to this crisis. Profits were a priority for bankers' and new policies (such as the new Basel Agreement) were implemented to prevent malfeasance from bankers. Despite this, banks continued to collude for their own benefit.

The scandal arose when benchmarking malpractice was revealed in 2012 by whistleblowers and later investigated by prosecutorial authorities in the US, UK, EU and

elsewhere (Lejot, 2018). Banks were manipulating their LIBOR publications to profit from trades and misrepresent their creditworthiness. One of the primary factors in the influence of the scandal was that the interbank reference rates are generally not determined through actual market transactions, but rather, based on banks' reports about their perceptions of what should be the interbank interest rate (Hou & Skeie, 2014). Following the claims of manipulation in 2012, an international investigation of LIBOR revealed that several banks had indeed been

manipulating the reference rate.

The objective of this paper is to investigate how this scandal may have impacted individuals, from student loans to mortgage loans and to those retail investors in derivative markets. Due to the wide scope of the scandal, the primary focus is on the mortgage market that had just been a victim of bankers manipulation. Resultantly, as the primary focus is on the impact of LIBOR on mortgages, the following research question was drawn, “Given the London Interbank Offered Rate (LIBOR) is open to manipulation by collusion, as shown through the LIBOR Scandal, did the LIBOR Scandal have a substantial impact on mortgage rates for retail investors in the United Kingdom?”

Due to the international scope of the scandal, it was decided that the paper would focus on the host nation of the LIBOR, the United Kingdom. Therefore, the analysis will focus on the impact of the scandal on the host nation of the LIBOR, the United Kingdom, in hopes that this would reflect the largest extent of the scandal’s impact. The research is to investigate how the manipulation had influenced mortgage loans (predominantly adjustable rate loans based on the LIBOR) before, during and after the scandal.

The presented question aims to examine one aspect of a larger matter at hand, the manipulation of interest rates for the personal gain of bankers. By researching into LIBOR, this paper aims to add more research to the field of regulation and how a lack of stringent regulation can present troubles for those even far outside the scope of the scandal. The relevance of the question is to look at the volatility of the mortgage market that had occured from LIBOR and whether there was truly a negative impact of the scandal on mortgages or whether the volatility during the period had little to no effect on mortgages. With the discontinuation of LIBOR in 2021, the research aims to look whether it was flawed to let banks control a major benchmark of interest rates, question how badly individuals were affected by the scandal and, finally, what alternatives to LIBOR can combat this problem.

With the wide scale impact of LIBOR on many financial markets, the scandal had undoubtedly impacted numerous financial instruments from derivatives to loans. Therefore, this paper will initially hypothesise that the variable mortgage rates linked to LIBOR would differ drastically from the expected rates of return on fixed rates or those pegged to other instruments (such as tracker mortgages based on the Bank of England set base rate). Given influences from factors, such as the Global Financial Crisis of 2008, banks had used the decrease in rates in their favour and manipulated the LIBOR lower in relation to that, with lower interest rates than expected. Therefore, it would not only be a premium that would be expected from the variation, but also a discount. Resultantly it is possible that the discounts in downturns may negate the effects of the premiums in upturns manipulated by the scandal. Therefore, the alternative to hypothesising an overall change in mortgage rates is that there is no true distinguishable difference between mortgage rates of variable rates based on LIBOR and other styles of mortgages (exempting the risk premium on fixed rates).

For testing purposes, the null hypothesis and the alternative hypothesis are stated as follows; H0: The LIBOR Scandal presented no real impact on the variable rates of mortgages. H1: The LIBOR-based mortgage rates would be different from the Tracker mortgage rates offered, to a statistically significant figure.

The most effective way to approach the research question is through the gathering of mortgage data and LIBOR data. A regression is used in order to assess the impact of LIBOR on adjustable rate mortgages based on LIBOR, especially comparing the rates before and after the discovery of the scandal. Following a regression analysis, the paper would be able to draw conclusions about the true extent of the LIBOR scandal on mortgage rates.

The paper will follow on from here to look at previous literature which can help understand and answer the research question. This will be looked at from the perspective of LIBOR itself, the mortgage rates and literature which looks at the effect of LIBOR on mortgage rates. Following on from this, the methodology will be drawn up to find the most effective way to find a conclusive answer for the hypothesis. Once the methodology has been constructed and the regression equations have been determined, the paper will continue on to produce the results and analyse the outcomes. Once the outcomes have been produced, the analysis of what the outcomes mean for the hypothesis will be discussed. Lastly, the conclusion will look at what this research has done for the field and what could be done for further investigations on the scandal as a whole.

2. Literature Review

2.1 The London InterBank Offered Rate (LIBOR)

The London InterBank Offered Rate (LIBOR) was established in 1986 by the British banking association (BBA) and has since been a benchmark interest rate for the pricing of many

derivative financial instruments, loans and mortgages worldwide (Bariviera, Guercio, Martinez & Rosso, 2015). To understand the importance of LIBOR, Ashton & Christophers, (2015) had looked at the importance of LIBOR from the perspective of two foundational aspects of the index; “the legal technology of arbitration” and its status as a commodity. The former of the two concepts is looked at first from the idea of arbitration, the means to mediate and shape the conditions of the interaction between two parties with no greater gain to one party, and more than just represent the market conditions. LIBOR, being a benchmark in the pricing of financial contracts, serves as a legal requirement to situations of uncertainty by attaching a metric against which parties obligations can be measured. By this explanation of LIBOR as a technology of arbitration, Ashton & Christophers (2015) holds it as a critical legal technology, which provides the conventions for future conditions. _{​The emphasis on the importance of LIBOR} here is the usage of this interest rate to provide a contractual basis between two parties, as a large proportion of financial contracts use LIBOR as a basis for the interest rates (Bariviera, Guercio, Martinez & Rosso, 2015). As it is a legal, measurable interest rate that contracting parties may use, the tampering of LIBOR is problematic as it already contradicts the

foundational aspect that Ashton & Christophers describe.

Ashton & Christophers (2015) elaborate on the role of LIBOR with the reference to how financial trade occurs with LIBOR as a neutral and external reference that allows for the

upholding of financial relationships amongst economic agents. In their example, they use the interest rate swaps market which is heavily reliant on LIBOR. One party tends to pay a fixed rate whilst the other receives payments based on a floating rate, whereby the floating rate is

generally indexed by an underlying benchmark, in this case, LIBOR._{​ If the party receives the} payments on a floating rate, different from that of the fixed rate, then the lending party is subject to any volatility in the LIBOR. This volatility makes it enigmatic for the party which incurs the floating rate, as this party must hold a higher risk. This higher risk being manipulated by the scandal puts into question the first foundational aspect discussed by the authors, as it is no

longer a “legal technology of arbitration.”_{​ Another pricing convention, which involves the} arbitration discussed before, is the adjustable rate home mortgage. The Adjustable Rate Mortgage (ARM) is a mortgage where the homeowner pays to their lender a fixed interest rate ‘margin’ on top of a base rate that adjusts with periodic (monthly, quarterly or semi-annual) changes in LIBOR. _{​This concerns the topic at hand directly, as the ARM requires the arbitration} of an interest rate that is neutral and external, but should there be manipulation of the LIBOR, the neutral aspect of the reference rate is dismissed. _{​Bariviera et al. (2015), found that an article} by Mollenkamp and Whitehouse in the Wall Street Journal that casted doubts on the

transparency of LIBOR and implied that published rates were lower than those reported by credit default swaps (CDS). Following investigations by market authorities in 2012, the US Department of Justice and the European Commission detected data manipulation had occurred from banks involved and resultantly imposed severe fines to banks involved in the illegal

manipulation of the rates. _{​These investigations show how LIBOR fails to mediate contractual} agreements as the data manipulation showed rates that were lower than those by credit default swaps. This would imply that the research topic would find some impact on the mortgage rates from LIBOR manipulation. _{​An article by the Federal Reserve Bank of Cleveland had stated that} the influence of manipulation of the LIBOR can underestimate the true lending costs, whilst borrowers gain and lenders lose, and the inverse should the LIBOR overestimate the true lending costs (Schweitzer & Venkatu, 2009). However, the financial literature by Bariviera et al. (2015) had provided no conclusive evidence of manipulation of the LIBOR-fixing process.

The latter foundational aspect Baviviera et al. (2015) refer to in their paper, of its status as a commodity, is reinforced by the US Commodity Futures Trading Commission, which identifies LIBOR as a ‘commodity in the interstate commerce’. If money serves as an universal equivalent for normal, non-financial commodities then the equivalent for financial commodities (that are of a monetary nature) are the interest rate benchmarking technologies. This creates a commodity through the universal referent of an interest rate and not the universal equivalent as with money. When money becomes the foundation of normal markets, LIBOR provides the basis of these markets through the interest rate benchmarking. Derivatives rely upon the LIBOR rate in order to structure and guide international markets. Derivatives provide some of the anchoring functions of currency sovereignty covered by gold and dollar standards and LIBOR is the anchor for such commodities. LIBOR should serve as a neutral market metric,

representative of market conditions. LIBOR here is a calculative legal technology involved in the neutral benchmarking work, but is rooted in the concentrated market power. The nature of

LIBOR as a commodity is exposed by the scandal which erupted in mid 2012. The two qualities of LIBOR as a legal technology of arbitration and as a commodity are both heavily responsible for the attempts to prosecute wrongdoing and re-establish the norms and conventions of pricing in global markets (Ashton & Christophers, 2015).

2.2

Mortgage Rates

Adjustable-rate mortgages are typically tied to one of two indexes, the U.S. treasuries or the London InterBank Offered Rate. The index is used to determine a mortgage’s new interest rate when it is reset and the choice between the two would be relatively arbitrary (Schweitzer & Venkatu, 2009). If the choice of index is arbitrary, then the ARM’s should be the same. The research can then use the treasury rates (or in the case of the UK, the Bank of England Base Rate) as a comparison for the LIBOR-based mortgages. This would help the research

investigate the extent of the scandal relative to a control period. Following 2007, the rates on which the indexes are based have diverged and borrowers with LIBOR-based rate mortgages are likely to pay more than they would have compared to the treasury based mortgages

(Schweitzer & Venkatu, 2009). The rates, having been higher on LIBOR after 2007, would show a higher LIBOR after 2007 than the Bank of England rate. Research conducted by Bible & Joiner (2009) looked into the mortgage lending process of Caddo Parish Louisiana and found that at least half of the adjustable rate mortgages recorded in a three month period had unfavourable terms with having initial rates 50% above those of normal fixed rates. This research reinforces the point made by Schweitzer & Ventaku about the increase in cost of the mortgage for borrowers during the period, and that even fixed rate mortgages would have been more beneficial for borrowers than the adjustable rate mortgages. Additionally, the proportion of LIBOR-based mortgages have increased substantially, in particular with subprime loans

(Schweitzer & Venkatu, 2009).

A report conducted by the National Commission on the Causes of the Financial and Economic Crisis in the United States expressed a strong concern for the existence of fraud in the financial sector during the 2000’s. The report highlights in particular the rising incidence of mortgage fraud which sustained such a rise from a decline in lending standards and lax regulations (Angeletti, 2019). The number of suspicious activity reports related to mortgage fraud grew approximately 20 times in size between 1996 and 2005 and then more than doubled

between the period of 2005 and 2009 (National Commission on the Causes of the Financial and Economic Crisis in the United States [NC]_​ 2011_​: xxii). The importance of stating the increase in mortgage fraud is that it implies that mortgages are susceptible to volatility. As fraud is prevalent in the mortgage market, it may be difficult to assess the impact of the LIBOR scandal. By

considering different time frames, it allows for the research to isolate the LIBOR scandal to each period and therefore minimise the influence of other scandals. With the use of the time frames, variables concerning only the LIBOR time frame should allow for a more accurate measure of the scandal. The error in a regression should absorb the other influences, as a result.

2.3

The Role of LIBOR on Mortgage Rates

Tabb & Grundfest (2013) discuss LIBOR from a different perspective. LIBOR, by their definition, is representative of a hypothetical interest rate, making it relatively inaccurate and unreliable as a measure of the cost of unsecured interbank funding. The illiquid markets that underlie the LIBOR rate-setting processes are “fuzzy objects” which cannot sufficiently describe the amorphous market conditions upon which trillions of dollars of market value depend (Tabb & Grundfest, 2013). The main concern behind LIBOR is that if it is unreliable in measuring the cost of unsecured interbank funding, then it does not provide a clear picture of the adjustable

mortgage market. Ludwig Wittgenstein observes that it can be difficult to paint a clear picture of a fuzzy object, whereby here the clear picture of the mortgage market cannot be described accurately by the “fuzzy” LIBOR (Tabb & Grundfest, 2013). The underlying issue with this is that banks rely now to a much greater extent on shorter term loans and secured funding in their interbank lending transactions. During the financial crisis, banks seized further unsecured lending, which, by extension, reduced the usage of the LIBOR. Following the financial crisis, trading volumes on several currencies and tenors had dropped too low to publicise LIBOR rates (Tabb & Grundfest, 2013). An underestimate of the LIBOR would reduce mortgage rates, which therefore increase the accessibility to mortgages and increase the volume of the real estate market. This analysis shows that mortgage rates are undoubtedly influenced (whether positive or negative is to be decided) by the manipulation of LIBOR.

The embedding of LIBOR in a financial contract explicitly states a transfer of that exposure of the LIBOR risk to either the borrower or lender. Adjustable Rate Mortgages comprised roughly 8.7% of the outstanding mortgages in the United States in July 2008. The

way mortgages were repaid through LIBOR were with the use of a margin on top of a base rate which adjusts periodically (monthly or semi-annually). The variable mortgage rates showed to be of significant concern when related to the global financial crisis of 2008, and may implicate that LIBOR had an influence on the global financial crisis as a whole.

Research conducted by Schweitzer and Venkatu (2008) found that alternatives to LIBOR-based mortgages would have delivered savings of about $25 to $45 per month over those based on LIBOR and substantially more for mortgages that reset in October 2008. Looking at mortgages in October 2008, the six-month LIBOR reached about 4.5 percent

whereby affected borrowers could have saved approximately $87, $94, and $116 per month had their loans been linked to alternatives based on the federal funds rate, the OIS rate, or the U.S. Treasury rate respectively. Looking at an earlier study from the same authors, they found from their data of a sample of Ohio borrowers that subprime mortgages have been based on LIBOR more often than prime mortgages. The two researchers had significant data to show that

mortgages based on other indexes had a smaller repayment than those based on LIBOR during the recession. This helps understand that the U.S. market was affected by LIBOR during the recession, but does not indicate whether the scandal itself had influenced the mortgage market over the whole duration of the scandal. This research will continue to conclusively investigate whether the market was overall worse off due to the scandal, and not solely during the recession period.

The director of the Association of Mortgage Intermediaries had announced in a financial times article in July 2012 that the manipulation of LIBOR is unlikely to have affected the UK mortgage market and that while the manipulation affected banks' bottom lines, the consumer impact would have been minimal. A spokesman for the council of mortgage lenders had stated, on the other hand, that it had been too soon to see the true scope of any detriment to

mortgages. Given the time period since the scandal, there should be far more data available to draw a fairer conclusion.

In a report by Maarten Pieter Schinkel (2015), the impact of monetary policy on collusion in bank cartels was analysed. The Federal Reserve and the European Central Bank responded to the collusion with tighter monetary policy, which caused a steep lowering of funding rates globally to levels around zero in nominal terms and below zero in real terms. This strong monetary policy from central banks created conditions favorable to bank cartels, as low real interest rates in classic cartel theory are associated with reduced incentives to unilaterally deviate and obtain instantaneous defection profits. This analysis showed that it was better for

banks to operate as a cartel in order to maximise profits, which justifies the collusion which occurred during the LIBOR scandal. A higher incidence of a more stable collusion amongst commercial banks raised concerns for the effectiveness of monetary policy (Schinkel, 2015).

Schinkel argues that the assignment of responsibilities between individuals and

organizations plays a decisive role in collusion, using the recent LIBOR scandal as an empirical illustration. Angeletti (2019) emphasised this illustration of the LIBOR scandal as no CEO or board member had received criminal prosecution since the financial crisis in spite of the collusion (Rakoff_​ 2014_​; Pontell, Black and Geis_​ 2014_​). (Angeletti, 2019). This shows that the individuals were willing to take on more risk to act unethically as there was less accountability. The profitability for collusion, however, outweighed the punishments associated with it, which encouraged the scandal to occur.

3. Empirical Approach

3.1 Methodology

To reach a fair conclusion of the answer for the research question, certain factors need to be taken into consideration. The metrics which need to be used for the effect of the LIBOR on mortgages need to be stated. First and foremost, the LIBOR needs to be assessed in relation to LIBOR-based mortgage rates. From this, the mortgage rate will be the dependent variable, which needs to be described from the other independent variables. The more explanatory variables that are found, the more we can understand the impact of LIBOR on mortgages.

Given that the economic conditions would impact the mortgage rate, following the research from Schweitzer & Venkatu (2009), from the willingness of banks and building societies to lend to retailers, the recession period needs to be accounted for. This would be looked at through the use of a dummy variable, which would look at what period the recession occurred and be 1 for the period which it incurred, so that the effects of the recession could be taken into account. In order to use the recession period as a variable, the definition of a recession needs to be accounted for. For this research, a recession is defined as “a period of temporary economic decline during which trade and industrial activity are reduced, generally identified by a fall in GDP in two successive quarters” (House of Commons Library Research, 2020). Therefore, the data for the GDP growth in the UK needs to be taken for the entirety of the period that will be analysed, and when two successive periods of GDP growth are negative, the recession dummy variable will be stated as 1. The recovery of a recession is defined as a period where the GDP has recovered to a state higher than prior to the recession (House of Commons Library Research, 2020). Once GDP has exceeded the value before the recession, the dummy variable would be set at 0. The period analysed as a recession is therefore from the final quarter of 2007 to the third quarter of 2009.

The LIBOR impacted primarily Adjustable Rate Mortgages based on LIBOR and less of an impact on tracker mortgages (generally based on Bank of England rates), leading the research to focus on the effect of the LIBOR on LIBOR-based Mortgages. The LIBOR from the periods prior to, during and following the scandal would need to be looked at. A fair analysis of LIBOR-based Mortgages, that account for economic conditions, would be to look at these mortgages relative to tracker mortgages. Given the literature analysed earlier, the assumption

would be that there would be a premium in the rates households would have paid in

LIBOR-based mortgages relative to tracker mortgages. As the LIBOR scandal encompassed the global recession, we would find the crisis of 2008 impacting LIBOR-based mortgages more heavily than tracker mortgages due to banks unwillingness to lend. This unwillingness to lend would have affected the LIBOR and particularly impacted LIBOR-based mortgages,

predominantly due to the fact that the crisis directly affected the subprime mortgage market, which encompasses a fair amount of adjustable rate mortgages.

To produce a statistical analysis of the two variable mortgage rates in question, the data available from the Bank of England database on the tracker mortgage rates must be gathered from the period affected. For LIBOR-based mortgages, the data needs to be gathered from databases of building societies and banks. One crucial aspect of the research is to maintain a ceteris paribus, so therefore mortgages of the same maturity must be compared. For this research, it is appropriate to choose mortgages with shorter term maturities, as they reflect the interest rate the most accurately. The 3 month LIBOR is taken for the analysis, as it is less volatile to the daily (and short-term) fluctuations that occur in the interbank lending market. A regression equation looking at variables of the regression and LIBOR can fairly assess the dependent variable, which in this case would be the LIBOR-based mortgage rates. The main explanatory variable of interest in this research would be the London Interbank Offered Rate, as the paper is attempting to assess the impact that the scandal had on the LIBOR-based

mortgages.

The most crucial two variables to analyse the effect of the scandal are additional dummy variables; during and after the scandal. Each would be taken differently in order to analyse the effect of LIBOR on each period. The period prior to the scandal would be held as a control and therefore not be measured by a dummy, as the research assumes that the period prior to any manipulation holds the most accuracy in measuring the normal market conditions. By having coefficients which align with the two periods, each dummy variable could be cross-referenced and analysed against each other to see what residual effects there are from the LIBOR on each period. Should there be a substantial measurement from the coefficients from each period, then there would be some evidence to imply that the LIBOR scandal had impacted the mortgage rates. The recession variable should absorb any impact that the mortgage crisis had on the overall economic climate.

3.2 Equations and Approach to the Research

Referring back to the initial hypothesis, it must now be looked at in statistically testable terms. The methodology looks at defining the impact of the LIBOR through the scandal dummy

variable, as that would most accurately define whether there was any real change in the market. Therefore, the hypothesis test would be as follows;

H0: β1(Scandal) = 0 H1: β1(Scandal) ≠ 0

The relevant equations to conduct the research are resultantly as follows; 1. LIBOR Rate Mortgages = _​β0​ + _{​β1(LIBOR) + β2(Recession) + ε}

2. LIBOR Rate Mortgages = β0_​ + _{​β1(LIBOR) + β2(Recession)​​+ β​}3(ScandalPeriod) _​+

β_​4(AfterScandalPeriod)_{​ + ε}

Equations 1 and 2 aim to look at the LIBOR-based mortgages and how this is influenced by the LIBOR and recession. Equation 2 includes the relevant variable for the hypothesis, the ScandalPeriod variable and the AfterScandalPeriod variable. These variables will explain how much the mortgage rates have changed as a result of the scandal and the lasting effects that occurred from the scandal.

3. (LIBOR rate mortgages - Tracker Rate Mortgage) = _​β0​ + _{​β1(LIBOR) + β2(Recession) +}

ε

4. (LIBOR Rate Mortgage - Tracker Rate Mortgage) = β0_​ + _{​β1(LIBOR) + β2(Recession)​​+}

β_​3(ScandalPeriod) _{​+ β​}4(AfterScandalPeriod)_{​ + ε}

Equations 3 and 4 look at the difference between the mortgage rates and how the LIBOR and recession had affected those. The inclusion of the ScandalPeriod and

AfterScandalPeriod are aimed at analysing how the variance in the adjustable mortgage rates are affected from the scandal.

5. (LIBOR Rate Mortgage - Tracker Rate Mortgage) = β0_​ + _​β1​(LIBOR - Base Rate) +

β2_​(Recession) + _​ε

6. (LIBOR Rate Mortgage - Tracker Rate Mortgage) = β0_​ + _​β1​(LIBOR - Base Rate) +

β2_​(Recession) + _​β​3(ScandalPeriod) + _​β​4(AfterScandalPeriod) + _​ε

The last regressions including the LIBOR variables (equations 5 and 6) examine the difference between the LIBOR and Bank of England Base Rate and how they correlate with the difference in the mortgages based on these interest rates. The ScandalPeriod and

AfterScandalPeriod variables are included here to see if there is an influence on the regression from the scandal on the difference. This would examine whether any difference in the interest rates found is consistent in the difference in the mortgage rates and whether this is influenced by the scandal period variables.

By having the above regressions, it would be best to compare the outputs to other regressions, which assess the effects of the coefficients on a control mortgage rate, the tracker rate. This would be tested as follows;

7. Tracker Rate Mortgage = β0_​ + _​β1​(Base Rate) + _​β2​(Recession) + _​ε

8. Tracker Rate Mortgage = β0_​ + _​β1​(Base Rate) + _​β2​(Recession) + _​β​3(ScandalPeriod) +

β_​4(AfterScandalPeriod) + _​ε

For all of the above, the β of the coefficients will be found, where β0 is the margin on the interest rate, β1 is the coefficient of the Bank of England Base Rate, β2 is coefficient of the recession dummy variable, β3 is the coefficient of the variable of the scandal and β4 is the coefficient of the variable after the scandal (being 1 or 0 depending on the period). For the remainder of the research, the variables ScandalPeriod and AfterScandalPeriod refer to the time during the scandal and the time following the exposure of the scandal respectively.

4. Analysis and Discussion

To begin with, the data analysis looks at the impact of the ScandalPeriod variable on the LIBOR based mortgages. Tables 2, 4, 6 and 8 include the ScandalPeriod and AfterScandalPeriod variable and the remaining tables show the regression without the separation of the periods. By looking at the regression without the scandal dummy variables, the research shows the

influence of the interest rates on the mortgages as a whole. The data analysis of table 1 shows a coefficient of 0.551 for the LIBOR, which, for every 0.551% increase in the LIBOR, there is a 1% increase in the mortgage rate. This shows that there is not a strong direct correlation between the LIBOR and the mortgages based on the interest rate. With a ceteris paribus, there is an addition of the scandal dummy variables in order to isolate the impact during the relevant scandal periods.

Table 2 includes the additional variables of the scandal and the post scandal periods, which isolates the periods. By isolating the periods, it allows for the analysis to examine the effects relative to the scandal. The analysis shows a coefficient of 0.00177, which shows that the scandal increased LIBOR-based mortgages by 0.177% for every 1% increase in the

mortgage rate. As the scandal coefficient accounts for the period during to the scandal, it shows that LIBOR better explains the interest on mortgage rates before the scandal and as the

coefficients during and after the scandal have a positive coefficient, the mortgage rates increase when the periods are divided relative to when they are not. The AfterScandalPeriod variable, however, has a higher coefficient than the ScandalPeriod variable, showing that the scandal influenced the period following the scandal more so than during the scandal.

Table 4 shows the inclusion of the scandal period with a coefficient of 0.00493. The period following the scandal, however, shows a significant difference from the period prior to the scandal, with a coefficient of 0.0167, which reflects that this period has a contribution to the dependent variable relative to the period without. This implies that the period after the scandal had an impact on the difference between the mortgages.

Table 6 includes the coefficients of the scandal periods to understand the impact of the differences in the LIBOR and Base Rate with the differences in the mortgages based on the respective interest rates. The coefficient of the LIBOR Base Rate difference from this regression

shows a smaller change than those without the isolated scandal periods, exhibiting that the period prior to the scandal reflected a higher correlation. With the higher correlation, there is a smaller change shown. The scandal periods reveal a positive correlation, highlighting that the periods during the scandal and following the scandal align the interest rate differences better than when the scandal did not occur. This rejects the null hypothesis, as the LIBOR appears to align better with the periods here than it did when there was not a scandal, demonstrating that the scandal may have moved in favour of the retail investors.

4.2 Discussion

The obtained data presents a coefficient of the interest rate between 0.5 and 0.7 on both the LIBOR regressions on the LIBOR based mortgages and on the Bank of England rate regression on the tracker mortgages. This indicates that the interest rates only account for 0.5%-0.7% of a 1% increase on the mortgage rates, leaving the remaining 0.3%-0.5% explained by other factors. The other factors may include the bank's premium for offering a mortgage, the premium for accepting a degree of the volatility and costs associated with the mortgage. As the control for the regressions is the regression of the tracker mortgages, any substantial difference in the LIBOR based mortgages from the tracker mortgages would implicate a potential influence from the scandal. Any changes between the two mortgages, however, may also include the variability in confidence for choosing a riskier peg (such as the LIBOR) over the more stable peg (the Bank of England Base Rate). This is where tables 3 and 4 are taken into account, as they hold the parity of the difference between the mortgage interest rates as a factor of the LIBOR. The data in the two tables varies substantially, from a negative value in a whole period scenario to a positive coefficient in the before scandal scenario, implying that the scandal had an influence on the LIBOR based mortgages. With a negative coefficient over the whole period, it would imply that the LIBOR mortgages were lower than the tracker mortgages, as each percentage change in the difference between the mortgages was influenced negatively by the LIBOR. However, when the scandal period was accounted for, table 4 showed that in each period, the LIBOR had increased the LIBOR mortgage rate relative to the tracker mortgage rate. This data falls in line with the alternative hypothesis that the LIBOR would be different from the tracker mortgages. The statistical relevance must be examined through statistical testing, from T-tests to R-Squared values.

The P-Value obtained for the ScandalPeriod coefficient came to 0.00175 in table 2, which shows that the ScandalPeriod variable was statistically significant throughout the research. Table 4 shows relevance in the answering of the question as it analyses the direct effect of LIBOR on the difference between the two mortgages researched. If the ScandalPeriod variable is statistically significant in relation to the difference between mortgages, the

ScandalPeriod variable shows an effect on the mortgage rate changes. The inclusion of the scandal variables showed a P-value of 0.0000128 and with all of the regressions showing LIBOR as statistically significant to the regression, it would imply that the null hypothesis here would be rejected. To verify that the null hypothesis is indeed rejected, the regression which compares the difference of the mortgage rates with the difference of the interest rates underlying it should be consistent. Table 6 looks at the regression of the aforementioned variables and the p-value of the ScandalPeriod coefficient is 0.00205. The ScandalPeriod variable shows to be statistically significant from this value, and therefore the null hypothesis is again rejected in this analysis.

Following the analysis of variables, the data showed an R-squared value that increased consistently with the use of the scandal dummy variable. This demonstrates that the usage of the Scandal period variables provided a higher explanation for the dependent variable, the mortgage rates. The most significant change on the R-Squared values can be seen in tables 3 to 4, and 5 to 6 with the inclusion of the scandal variables there. Table 3 showed a weak goodness-of-fit for the regression, showing that the LIBOR had provided little explanation for the difference in the mortgage rates. With the introduction of the scandal variables, there is a significantly better explanation of the model than with solely the LIBOR variable. This emphasises that the scandal had been an important factor in the model and validates the rejection of the null hypothesis.

The adjusted R-squared tends to be a more reliable measure of performance as it takes into account the addition of independent variables. In relation to the previous analysis of the R-Squared values, the data remains relatively the same, with a slight drop in correlation across the board with the use of the adjusted R-Squared measure.

The F-statistic calculated across the regressions show a value significantly exceeding the probability that the F-statistic occurs. This rules out the null hypothesis, consistently, over all

of the regressions provided. This implies that the model does not fit the population that it was sampled from, in spite of the fact that the mortgages were based on these interest rates.

The t-test checks the significance of regression coefficients in the linear regression. Should the t-statistic lie outside the acceptance region, then the null hypothesis is rejected. In table 2, the t-statistic for the coefficient of ScandalPeriod is 3.16, which lies outside the acceptance region. As a result, the null hypothesis is rejected and the alternative hypothesis is not rejected. In table 4, the t-statistic is 7.82, lying outside of the rejection region again. Table 6 shows another t-statistic outside the acceptance region, with a value of 3.12. Comparing these values to the control regression, table 8 has a value of 2.62, which in this case is accepted and accepts the null hypothesis. However, table 8 is for the tracker mortgages which should not be influenced by the scandal period, so this aligns with the literature that the null hypothesis should be accepted for the tracker mortgages, but rejected for the mortgages regarding the LIBOR.

Overall, all the statistical testing directs the answer towards a rejection of the null hypothesis, from the numerous t-tests which were conducted on various LIBOR-based mortgages and verified with the control regression. The P-value from each regression was consistently small, emphasising that the variable coefficient was statistically significant, implying an impact on the overall regression. With an impact on the regression, it shows that the null hypothesis, of the scandal having no effect, can no longer be true. The post-scandal period is particularly interesting, as it reflects higher t-statistics and p-values than the scandal itself, implying that it had an effect on the regression more than the scandal period itself. With this being apparent, it is interesting to see that there is more of an impact from the exposure of the scandal than the scandal itself. This could potentially be concluded from the loss in confidence in LIBOR by individuals and banks, as suggested by Tabb & Grundfest (2013), following its exposure.

5. Conclusion

The research conducted looked at the impact of LIBOR on mortgage rates from several

perspectives; directly to the mortgage rate, the impact of the scandal in isolated time frames, in relation to the tracker rate mortgages and with the involvement of the Bank of England Base Rate. The initial assumption was that the LIBOR would be the main influence on the mortgages based on the interest rate and that the manipulation during the scandal would have impacted the mortgage rates. The null hypothesis took the stance that nothing would be changed, but as the research went on, it became abundantly clear that there was a change between the LIBOR and the mortgages based upon it. The interesting realisation from the research was that the LIBOR had, on its own, not correlated strongly with the mortgage rates, even with the omittance of the scandal. The fault in the research is that it should have looked at more variables to find the full explanation of the LIBOR based mortgages, and would have then found a better impact of the scandal on the mortgage rates. Despite this, the research has still found some extent of influence on the mortgage rates from the scandal, especially when comparing the LIBOR based mortgages to similar adjustable rate mortgages.

The interpretation of the results was that the LIBOR was not a largely correlated factor of LIBOR based mortgages. The recession had been accounted for and the results from the

recession showed that there was a negative influence on the adjustable rate mortgages as a result, with a negative coefficient of the recession. Given the overall economic climate during the time of the scandal, it was difficult to confirm that it was solely the scandal and the recession which had impacted the change in the LIBOR rate mortgages relative to other adjustable rate mortgages. Another observation was that the scandal had affected the mortgages and had a stronger influence after the scandal occurred, which aligns with the previous literature that there was less confidence in the LIBOR following the exposure of the scandal in July 2012.

The null hypothesis was rejected consistently throughout the 8 regressions following the analysis. The alternative hypothesis, that the LIBOR scandal had influenced the LIBOR based mortgages, was not rejected due to the evident change in the mortgage rates during and following the scandal, relative to the period prior. This cannot be solely confirmed due to the lack of understanding of other influences on the mortgage market during this time. The collapse of the subprime mortgages, which comprised a large amount of LIBOR based mortgages, was a key factor.

The scandal addresses a particularly relevant topic in this current climate, from banks self regulation in creditworthiness to providing a fair assessment of an internationally pegged interest rate. The termination of LIBOR in 2021 and the replacement of it with the SONIA is a heavily discussed topic in the financial world at present. With this paper in mind, future research could look at the investigation conducted here to help understand the influence of LIBOR on a larger scale. The influence is clearer when understanding that in spite of the weaker correlations of mortgages with their pegs, there is still an underlying influence from the scandal as

demonstrated by the scandal variables. By going forward with a new interest rate, there needs to be a discussion of what can be done to prevent this level of market manipulation from

repeating itself with the new alternative to LIBOR. One particular impact to look at is the defaults of mortgages during this period, as a large amount of subprime mortgages had been based on LIBOR and a large amount had defaulted following 2007.

This research brought light to the fact that mortgages pegged to a certain interest rate or measure are not simply influenced by that measure, irrespective of the scandal. For future research to be conducted on this topic, more variables of what influences mortgages need to be considered and more understanding of the true impact of LIBOR itself needs to be understood.

References

Ashton, P., & Christophers, B. (2015). On arbitration, arbitrage and arbitrariness in financial markets and their governance: unpacking LIBOR and the LIBOR scandal. Economy And Society_​, _​44​(2), 188-217. doi: 10.1080/03085147.2015.1013352

https://www-tandfonline-com.proxy.uba.uva.nl:2443/doi/full/10.1080/03085147.2015.10133 52

Angeletti, T. (2019). The Differential Management of Financial Illegalisms: Assigning Responsibilities in the Libor Scandal. Law & Society Review, 53(4), 1233-1265. doi: 10.1111/lasr.12442

https://onlinelibrary-wiley-com.proxy.uba.uva.nl:2443/doi/full/10.1111/lasr.12442 Bariviera, A., Guercio, M., Martinez, L., & Rosso, O. (2015). The (in)visible hand in the Libor market: an information theory approach. _{​The European Physical Journal B​}, _​88​(8). doi: 10.1140/epjb/e2015-60410-1

https://link-springer-com.proxy.uba.uva.nl:2443/content/pdf/10.1140/epjb/e2015-60410-1.p df

Bible, D., & Joiner, G. (2009). Adjustable rate mortgages and the mortgage crisis. Property Management_​, _​27​(3), 152-162. doi: 10.1108/02637470910964642

https://www-emerald-com.proxy.uba.uva.nl:2443/insight/content/doi/10.1108/02637470910 964642/full/pdf?title=adjustable-rate-mortgages-and-the-mortgage-crisis

Bradshaw, Julia - No evidence of detriment to MMR from Libor fixing (Financial Times July 2012):

https://go-gale-com.proxy.uba.uva.nl:2443/ps/i.do?id=GALE%7CA295371329&v=2.1&u=a mst&it=r&p=AONE&sw=w

Hou, D., & Skeie, D. (2014). LIBOR: Origins, Economics, Crisis, Scandal, and Reform. SSRN Electronic Journal_​. https://doi.org/10.2139/ssrn.2423387

House of Commons Library Research, 2020. _{​Recession And Recovery​}. London: House of Commons Library Research,p.29.

https://www.parliament.uk/documents/commons/lib/research/key_issues/Key-Issues-Rec ession-and-recovery.pdf

Lejot, P. (2018). Misconceptions of Interest Benchmark Misconduct. _{​SSRN Electronic} Journal_​. https://doi.org/10.2139/ssrn.3096898

Mortgage & housing statistics_​. The Building Societies Association: BSA. (2020). Retrieved 2 June 2020, from https://www.bsa.org.uk/statistics/mortgages-housing OECD Statistics_​. Stats.oecd.org. (2020). Retrieved 2 June 2020, from

https://stats.oecd.org/BrandedView.aspx?oecd_bv_id=eo-data-en&doi=c8705713-en#. Schinkel, M. (2018). Coordinated Effects in Monetary Policy. _{​SSRN Electronic Journal​}. doi: 10.2139/ssrn.3186038

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3186038

Standard variable rate mortgage data series | Bank of England | Database. (2020). Retrieved 10 May 2020, from

https://www.bankofengland.co.uk/boeapps/database/FromShowColumns.asp?Travel=NIx AZxI3x&FromCategoryList=Yes&NewMeaningId=RSVRM&CategId=6&HighlightCatValueD isplay=Standard%20variable%20rate%20mortgage

Schweitzer, M., & Venkatu, G. (2009). Adjustable-Rate Mortgages and the Libor Surprise.

Economic Commentary_​.

https://web-b-ebscohost-com.proxy.uba.uva.nl:2443/ehost/pdfviewer/pdfviewer?vid=1&si d=6c5aaa17-400e-4a6a-9f0c-311b01c00222%40sessionmgr103

Schweitzer, M., & Venkatu, G. (2009). Alternatives to Libor in Consumer Mortgages.

Economic Commentary_​.

https://web-b-ebscohost-com.proxy.uba.uva.nl:2443/ehost/pdfviewer/pdfviewer?vid=1&si d=0b4cf944-cc8f-4fcb-a010-21418d99b235%40pdc-v-sessmgr03

Tabb, R., & Grundfest, J. (2013). Alternatives to LIBOR. _{​SSRN Electronic Journal​}. doi: 10.2139/ssrn.2272462

Appendix

Table 1: LIBOR Rate Mortgages = _​β0​ + _{​β1(LIBOR) + β2(Recession) + ε}

Table 1 - Regression of LIBOR-Based Mortgage Rate

Table 2: LIBOR Rate Mortgages = _{​β0 + ​β1(LIBOR) + β2(Recession) ​+ β ​}3(Scandal) _​+ β_​4(After Scandal)_{​ + ε}

Table 3: (LIBOR rate mortgages - Tracker Rate Mortgage) = _{​β0 + ​β1(LIBOR) +} β2(Recession) + ε

Table 3 - Regression of LIBOR-Based Mortgage Rate and Tracker Rate (BoE Base Rate based) difference

Table 4: (LIBOR Rate Mortgage - Tracker Rate Mortgage) = _{​β0 + ​β1(LIBOR) +} β2(Recession)_{​​+ β​}3(Scandal) _{​+ β​}4(After Scandal)_{​ + ε}

Table 4 - Regression of LIBOR-Based Mortgage Rate and Tracker Rate (BoE Base Rate based) difference, including the scandal time frame and its’ effects

Table 5: (LIBOR Rate Mortgage - Tracker Rate Mortgage) = _{​β0 + ​β1 ​}(LIBOR - Base Rate) + β2_​(Recession) + _​ε

Table 5 - Regression of LIBOR-Based Mortgage Rate and Tracker Rate (BoE Base Rate based) difference, including the scandal time frame and its’ effects

Table 6: (LIBOR Rate Mortgage - Tracker Rate Mortgage) = _{​β0 + ​β1 ​}(LIBOR - Base Rate) + β2_​(Recession) + _​β​3(Scandal) + _​β​4(After Scandal) + _​ε

Table 6 - Regression of LIBOR-Based Mortgage Rate and Tracker Rate (BoE Base Rate based) difference, including the scandal time frame and its’ effects

Table 7: Tracker Rate Mortgage = _{​β0 + ​β1 ​}(Base Rate) + _​β2​(Recession) + _​β​3(Scandal) + β_​4(After Scandal) + _​ε

Table 7 - Regression of Tracker Rate (BoE Base Rate based) on the Base Rate

Table 8: Tracker Rate Mortgage = _{​β0 + ​β1 ​}(Base Rate) + _​β2​(Recession) + _​β​3(Scandal) + β_​4(After Scandal) + _​ε