St. Petersburg State University
Graduate School of Management
Master in Corporate Finance Program
THE DETERMINANTS OF THE ASYMMETRY
IN CASH CURRENCY EXCHANGE RATES
Master’s Thesis by the 2nd year student
Concentration – Corporate Finance
Kondor Pavel Dmitrievich
Research advisor:
Okulov Vitaliy Leonidovich,
Associate Professor
St. Petersburg
2016
ЗАЯВЛЕНИЕ О САМОСТОЯТЕЛЬНОМ ХАРАКТЕРЕ ВЫПОЛНЕНИЯ
ВЫПУСКНОЙ КВАЛИФИКАЦИОННОЙ РАБОТЫ
Я, Кондор Павел Дмитриевич, студент второго курса магистратуры направления
«Менеджмент», заявляю, что в моей магистерской диссертации на тему «Изучение факторов,
влияющих на асимметрию курсов наличного обмена валюты», представленной в службу
обеспечения программ магистратуры для последующей передачи в государственную
аттестационную комиссию для публичной защиты, не содержится элементов плагиата.
Все прямые заимствования из печатных и электронных источников, а также из
защищенных ранее выпускных квалификационных работ, кандидатских и докторских
диссертаций имеют соответствующие ссылки.
Мне известно содержание п. 9.7.1 Правил обучения по основным образовательным
программам высшего и среднего профессионального образования в СПбГУ о том, что «ВКР
выполняется индивидуально каждым студентом под руководством назначенного ему научного
руководителя», и п. 51 Устава федерального государственного бюджетного образовательного
учреждения высшего образования «Санкт-Петербургский государственный университет» о
том, что «студент подлежит отчислению из Санкт-Петербургского университета за
представление курсовой или выпускной квалификационной работы, выполненной другим
лицом (лицами)».
_______________________________________________ (Подпись студента)
_______________25.05.2016_________________________________ (Дата)
STATEMENT ABOUT THE INDEPENDENT CHARACTER OF
THE MASTER THESIS
I, Kondor Pavel Dmitrievich, (second) year master student, program «Management», state that
my master thesis on the topic «Determinants of the asymmetry in cash currency exchange rates»,
which is presented to the Master Office to be submitted to the Official Defense Committee for the
public defense, does not contain any elements of plagiarism.
All direct borrowings from printed and electronic sources, as well as from master theses, PhD
and doctorate theses which were defended earlier, have appropriate references.
I am aware that according to paragraph 9.7.1. of Guidelines for instruction in major
curriculum programs of higher and secondary professional education at St.Petersburg University «A
master thesis must be completed by each of the degree candidates individually under the supervision
of his or her advisor», and according to paragraph 51 of Charter of the Federal State Institution of
Higher Education Saint-Petersburg State University «a student can be expelled from St.Petersburg
University for submitting of the course or graduation qualification work developed by other person
(persons)».
________________________________________________(Student’s signature)
____________________25.05.2016____________________________ (Date)
2
Аннотация
Автор
Название
магистерской
диссертации
Факультет
Направление
подготовки
Год
Научный
руководитель
Цель, задачи и
основные
результаты
Ключевые слова
Кондор Павел Дмитриевич
«Изучение факторов, влияющих на асимметрию курсов наличного обмена
валюты»
Высшая Школа Менеджмента
Корпоративные финансы
2016
Окулов Виталий Леонидович
Цель: Выявить факторы, определяющие асимметрию в курсах наличного обмена
валют.
Задачи: Изучить существующие исследования в этой области и создать
теоретическое обоснование асимметрии; Собрать данные для проведения
анализа; Провести эконометрический и статистический анализ данных и сделать
выводы.
Выводы: Асимметрия связана с колебаниями на биржевом валютном рынке и с
изменениями в спросе на обмен валют со стороны населения.
Валютный рынок, Розничный обмен валют, Микроструктура рынка, Асимметрия
спрэда Продажа/Покупка
Abstract
Author
Kondor Pavel Dmitrievich
Master thesis title
«Determinants of the asymmetry in cash currency exchange rates»
Faculty
Graduate School of Management
Main field of study
Corporate finance
Year
2016
Scientific advisor
Okulov Vitaliy Leonidovich
Research goal,
objectives and key
results
Goal: Find the factors that determine the asymmetry in cash currency exchange rates
Objectives: Study the existing academic researches in this field and create a
theoretical framework; Collect the necessary data; Carry out statistical and
econometrical analysis and make conclusions.
Conclusions: The asymmetry is closely connected with FX market price shifts and
trends and with the customer demand for currency conversion.
FX market, Retail currency exchange, Market microstructure, Bid-ask spread
asymmetry.
Keywords
3
Contents
Introduction ............................................................................................................................................................. 5
Chapter 1. Research Background ............................................................................................................................ 9
Paragraph 1. Research Goal and Objectives ....................................................................................................... 9
Paragraph 2. Theoretical review ....................................................................................................................... 10
Paragraph 3. Russian Foreign Exchange Market Overview ............................................................................. 14
Paragraph 4. Theoretical Framework for the Research ..................................................................................... 16
Paragraph 5. Managerial Application ............................................................................................................... 18
Paragraph 5. Research Methodology ................................................................................................................ 19
Chapter 2. Dataset Overview ................................................................................................................................ 21
Paragraph 1. Dataset Origins ............................................................................................................................ 21
Paragraph 2. Variables’ Description ................................................................................................................. 23
Paragraph 3. Dataset Adjustments .................................................................................................................... 25
Chapter 3. Dataset Statistical Analysis ................................................................................................................. 32
Paragraph 1. Asymmetry Analysis.................................................................................................................... 32
Paragraph 2. Determinants Analysis ................................................................................................................. 33
Chapter 4. The Results .......................................................................................................................................... 38
Paragraph 1. Demand Aspect ............................................................................................................................ 38
Paragraph 2. Supply Aspect .............................................................................................................................. 40
Paragraph 3. Asymmetry Regression Model .................................................................................................... 42
Paragraph 4. Conclusion ................................................................................................................................... 46
List of References ................................................................................................................................................. 50
Appendices............................................................................................................................................................ 52
Appendix 1. ....................................................................................................................................................... 52
Appendix 2 ........................................................................................................................................................ 56
USDRUB ...................................................................................................................................................... 56
EURRUB ...................................................................................................................................................... 58
4
Introduction
The research focuses on the asymmetry of quotes or bid-ask spreads that the retail banks set
forth to make the market of retail cash foreign currency exchange. The cash currency market is distinct
from the regular interbank FX market. It is a dealer made market, where the retail banks act as the
dealers. The bid-ask spreads are not always in equilibrium with the “major” rate (set either by the
Central Bank or quoted from FX market). In other words, the medium point between bid and ask price
is not always the same as the major quote from the FX market or Central Banks official rate. The
research analyzes the asymmetry in the spread and suggests a model that quantitatively describes the
asymmetry.
The
Asymmetry =
asymmetry
in
percentage
points
(Ask Price−Major Rate)−(Major Rate−Bid Price)
Major Rate
was
formulated
as
follows:
∗ 100%. The ask price is the price the banks
ask to exchange their currency for customers’ rubles, the bid price is the price the banks pay in rubles
for customers’ currency, the major rate is the rate on the interbank market.
The banks and exchange agencies offer bilateral services in cash exchange: sell foreign for the
domestic currency and buy foreign for the domestic currency. Furthermore, many retail banks also
offer the cross currency operations (e.g. exchange JPY for CHF). The volume of the latter type
conversions is much lower than the direct domestic-to-foreign and foreign-to-domestic operations.
Limited information and low liquidity cause market inefficiencies and potential flaws in samples. To
reduce biases and inconsistencies in the samples, only the direct operations will be considered in the
research.
Retail exchangers have direct access to the global FX market. The population and small-tomedium enterprises usually avoid brokerage fees and have no direct access to the global FX market.
The retail exchange houses create a separate retail foreign exchange market. It is closely connected to
the regular FX, but has different buyers, sellers and specificities.
The market can be described as a dealer market with considerable conversion commissions.
The base conversion rate is the rate quoted on the FX market. The rate for reporting and accounting for
the financial operations is the official rate published by the Central Bank of Russia. The retail
exchangers ask for a premium while selling and bid for a discount while buying foreign currency from
its customers.
5
However, it turns out that these bids and asks are not ultimately fixed to the major rate: neither
to FX market rates, nor to the quotes published by the central bank. Furthermore, these spreads even
happen to be asymmetrical in respect to the major rates mentioned both Central Bank rates and FX
quotes.
The main goal of the research is to analyze the asymmetry in quotations over the latest years
for the set of banks. The result of the analysis is the econometrical model that explains the asymmetry
in the bid-ask spread quotation in comparison to the “major” rate quotation.
The research uses a solid block of literature overviewing several aspects of the problem.
Firstly, the main question is the FX market itself, therefore a significant part of the literature
references is devoted to foreign exchange market for currency pairs either including Russian Ruble or
not. The latter financial markets – though different nominally: neither USDRUB, nor EURRUB, have
a lot in common. They all are integral part of a closely interconnected global FX market and therefore
have similar peculiarities.
The second very important aspect of the research problem is the phenomenon of the dealer
markets and the bid-ask spreads on the traded assets. It is crucial to get a good understanding of what
drives the dealers on the market, when they set their rates. The rate-margins must neither be too wide
to drive away the customers, nor too narrow not to lose profits. Therefore, another block of literature
reference list is devoted to the bid-ask spreads on the dealer markets.
Thirdly, the question is how to obtain the data to be analyzed. Most of the banks store their
quotes online but the format of the information makes it a complicated problem to download the data.
Fortunately, there exist the quotation comparison web-services, which accumulate the exchange rates
over time.
Finally, an econometric model was to be built. That is the reason why many research works
with sophisticated models are included into the literature reference list. These works do not contribute
to theoretical perspective in terms of analyzing the quotes, they rather support the practical aspect of
analysis. They give useful insights into dealing with the data and discovering dependencies.
The data set obtained describes the quotes for EUR and USD against RUB in Moscow retail
banks. These two currency pairs and, consequently four transaction types (buy and sell both
currencies), were chosen because they happen to be the transactions with the highest volumes in
comparison to other currencies exchanges (Central Bank of Russia, 2016).
6
The Moscow banks are chosen due to the two following reasons. Firstly, the number of banks
in Moscow is the largest as compared to other cities in Russia (2). Secondly, Moscow is the center of
Russian financial markets and taking Moscow banks as a benchmark means eliminating cash
transportation costs from the cash currency retail prices. Foreign cash gets into Russia mainly through
Moscow and delivering it into other cities requires additional spending which transfers into the
commission for cash currency exchange: into the bid-ask spread.
The data set gives the quotes for the banks over the time period starting in early June 2014 and
ending in March 2016. June 2014 is chosen as a starting point because it is the beginning of the
negative trend in oil prices: the crucial factor for Russian currency. (8, 9, 10). March 2016 is the time
when the final data mining started, after the literature analysis and model creation.
The data contains information for a large set of banks over a 2 year period. Some of the banks
did not succeed in overcoming the crisis and disappeared. Other banks did not publish their rates
regularly. These banks were eliminated from the analyzed data set, to eliminate inconsistencies in the
sample.
Cash currency exchange market, although specific, has the traits of any regular market.
Theoretical approach suggests there must be two main groups of factors determining the price (the
commission in the terms of the retail market): supply and demand factors. The crucial supply factor is
the ability of the retail currency exchangers get the currency. The banks offering cash currency
exchange rates have access to global FX market. In other words, the global FX market and its trends
influence the supply side significantly. The demand factor is the willingness of the citizens and
companies to make exchange transactions. This consumer behavior may be an object for analysis as
well. However, that question requires another separate research.
The econometrical model tries to take into account retail bankers’ expectations of the future
movements of the FX rates. The future price paths are often predicted by focusing on the past
performance of the assets prices (in our terms, foreign currency). That is why the trend direction and
velocity were chosen as the key indicators.
Open and close prices were chosen as proxies for determining the overall daily trends on the
FX market. These or any other arbitrarily chosen parameters from the price-line appear to be
representative indicators, because the market is highly liquid and does not have many outlying prices.
The supply side study shows that the asymmetry is stable when there is a stable trend on the
market. When the foreign currency appreciates against Russian ruble, the ask commissions are
7
consistently higher than the bid commissions. Apparently, the banks take into account the direction of
the FX price movements to fix profit for the future.
The two regression models describing the asymmetry on USDRUB and EURRUB bid-ask
retail quotes were built. The independent variables for these models were the historic price movements
over the latest trading sessions before each observation. The regression analysis has shown, that the
asymmetry is closely related to the historic price changes. Furthermore, it turned out that the older the
history the lower impact on the asymmetry is evident.
The demand side analysis has demonstrated strong correlation between volumes of retail
foreign currency conversions and span of asymmetry in bid-ask spreads. When the population is
buying more foreign currency than sells, the retail ask quotes are more distant from the FX rates than
bid quotes.
The overall result of the research shows that there is a strong relation between the differences
in premiums and discounts for selling and buying foreign currency and trends in FX market quotes.
The changes in net retail currency conversion operations are also strongly correlated with the
asymmetry. These facts support the hypothesis that the determinants of cash currency exchange rates
asymmetry originate from supply and demand side factors.
8
Chapter 1. Research Background
This chapter gives the academic description of the research. Its goal and objectives are stated.
After that, the review of the research works on the relevant topics are presented and the research gap is
demonstrated. The managerial application is also discussed here. The chapter continues with the
research hypothesis and questions to be answered in the research. The conclusion of the chapter is the
methodology of the research. The research techniques and methods are described in the end of the
chapter.
Paragraph 1. Research Goal and Objectives
The first issue to discuss is the goal of the research and its objectives. The goal is to analyze
the quotes data, FX rates and market aggregated statistics to find the interdependencies that determine
the asymmetry of the bid ask spreads that make of the cash currency exchange market. In other words,
the goal is to find the factors for the rates asymmetry.
The research goal stated requires that a number of research objectives be stated and solved
consequently. The research strategy is to prove the factors existence, spot the factors and analyze them
in terms of degree of their influence.
The first objective of the research is to carry out a thorough study of the existing theoretical
papers and empirical models concerning the topic of the research. This step is aimed at summarizing
what has been done already and what is still needed to be done. This objective is fulfilled by studying
the literature and publications in the academic journals. The information was outlined and summarized
in order to give a brief overview of the research gap.
The second objective was to suggest a full set of factors which could potentially determine the
bid-ask spread asymmetry. The extensive list was to be created in order to make tests for the elements
of the list and determine their importance. To complete this objective, the theoretical literature and FX
market analytics was used.
The third objective was to collect the data for the research. To analyze the asymmetry in cash
currency exchange rates three milestone datasets were to be collected. These are the bid and ask quotes
that the banks publish, the “major” rate data set for determining and calculating the asymmetry and the
quantitative information about the influencing factors. The most difficult set to collect was the sample
of bid and ask prices published by the banks. The banks publish the information of their historical
quotes partially which made it a complicated problem to get a regular standardized dataset. The major
9
rates are freely available in the public access. The quantitative data about the influencing factors was
also difficult to collect. In fact, a lot of specific factors the differ from bank to bank are impossible to
collect.
Fourthly, the final objective was to make the data analysis and suggest a model that describes
the asymmetry. This part of the research was structured according to the split of the overall influencing
factors on the retail foreign exchange market. The demand – the population – side of the market was
studies and the supply – the retail banks – was analyzed.
To sum up the overall composition of the research is the following. We start with the overview
of the theoretical background in order to determine the unexplored fields in this topic. We then use the
results of the theoretical studies to create a theoretical basement for the future analysis: we suggest a
complete set of various potential factors that are connected to the retail FX market. The next step is to
collect the extensive dataset: the retail banks bid-ask quotes, the major exchange rate for calculating
the asymmetry and the quantitative information about the factors that influence the market. The final
objective is to conduct a thorough analysis of the data and to determine statistically the tendencies and
relations.
Paragraph 2. Theoretical review
In this section, the existing research papers are discussed in order to describe the current situation
in the academic research of the field of cash currency exchange rates. A number of academic research
papers were studied and summarized. The short descriptions and the relevance of the papers are also
discussed in this section. The summary of the existing findings is given. Most of the books, reviews
and articles analyze the topic only in limited aspects, thus creating a research gap, which is covered in
this work.
The research works that are related to the topic of cash currency exchange rates asymmetry focus
on fundamental analysis of theoretical bid-ask spreads, empirical studies of interbank FX markets and
on specific retail foreign currency markets characteristics. The works that are most relevant to our
research of Russian retail foreign exchange market are stated in the list.
(1) Copeland, T. E., & Galai, D. (1983). Information Effects on the Bid-Ask Spread. The Journal
of Finance, 38(5), 1457-1469.
(2) Saida Gtifa, Samir Maktouf and Nadia Labidi (2015). Dealer Behavior and Price Strategy in
the Foreign Exchange Market: Evidence from FX Tunisian Market Dealer. Journal of Business
Studies Quarterly 2015, Volume 7, Number 2.
10
(3) Enn Listra, Niina Vaiser and Katrin Rahu (2011). Systematic Differencies of Retail Exchange
Spreads in Some EU Counties: The Banks Against Financial Integration. Discussions on
Estonian Economic Policy No. 2/2011.
(4) Hussain S. (2011). The Intraday Behaviour of Bid-Ask Spreads, Trading Volume and Return
Volatility: Evidence from DAX30. International Journal of Economics and Finance, Vol.3,
No.1.
(5) Kyle, A. S., & Obizhaeva, A. A. (2004). Market Microstructure Invariance: Empirical
Hypotheses.
SSRN
Electronic
Journal
SSRN
Journal.
Available
at
SSRN:
http://ssrn.com/abstract=1687965
The first article from the list is (Copeland and Galai, 1983). The authors focus on determining the
bid-ask spreads in the organized dealer-made markets. They start with creating a framework for
analyzing the spread: they estimate premiums and discounts from a profit maximization approach as a
trade-off between liquidity and earnings. The key factor in adjusting the rates over time is the
information: just as in the cash FX market making. The risk is in the information asymmetry. If the
customer of the dealer gets the information about the changes in fair price of the traded asset earlier
than the dealer, the customer can sometimes make an arbitrage and make the dealer lose on the
transaction.
The approach of (Copeland and Galai, 1983) to the problem is to view the bid-ask spread as a call and
put options combination: as a straddle. The exchange dealer gives the buyer the call option to call
currency at the buy price (with a premium) and to put currency at the sell price (with a discount). The
research creates a model for calculating the fair value of the bid-ask spread for the dealer. The research
results are based on the pricing and, what is equal, bid-ask spread measured according to the geometric
Brownian motion process.
This is a highly theoretical study. The authors focus on creating a model, which simulates the
appearance of new relevant information effects on the prices and on the bid-ask spreads on the
organized markets. The authors describe the assumptions they have for bid-ask spreads modelling. For
example, they state why exactly bid ask spreads should be neither too short nor too narrow and offer
the way to quantify these extremes and find the optimal trade off. The overly wide spread leads to
losses of liquidity trader which would avoid greater commissions, while overly narrow spreads would
let the dealer lose to better informed traders.
11
This paper is relevant for the research at hand. This fact can be backed by the argument that the
bid-ask spread analysis is key for the asymmetry valuation. Before any asymmetry is analyzed one
should first focus on the bid-ask spread itself to better understand the problem and to know the
approaches already used in such analysis. Furthermore, the factors that determine the bids and asks are
highly likely to influence them not equally in both directions in reality, thus creating asymmetry.
The next article to be described is (Gtifa, Maktouf and Labidi, 2015). This article analyzes the
dynamics of Tunisian Dinar (TND) price in EUR before and during the period of Tunisian FX market
turmoil. The authors test different factors influencing the exchange rate, such as trading volume, bidask spreads and volatility.
It demonstrates empirical research results. Although this article is about a market other than
Russian currency: Tunisian Dinar, the circumstances are quite similar. The study considers the period
of low volatility and a period of greater uncertainty and market turmoil. This reflects the Russian FX
market situation.
This article is useful for the study of Determinants of the cash currency exchange rates
asymmetry, because of several factors. Firstly, the research describes a method used for estimation of
the implied volatility on the FX market. This is a helpful instrument to use in the research, because
volatility and the way it is perceived by the dealers is could be among the key factors influencing the
bid-ask spreads. Secondly, the methods used in the study (Gtifa, Maktouf and Labidi, 2015) were
relevant for the research on Russian markets, because the circumstances in Tunisia were similar to
what Russian market faced during the latest time. In other words, this piece of literature is of help to
the research both from the technical point of view and as a theoretical foundation for the research.
(Listra, Vaiser and Rahu, 2011) is the third important research work to be studied. This article
refers to the differences the same banking groups have in retail currency exchange spreads in various
countries of presence. One of the important aspects the authors study are the factors influencing the
spreads: macroeconomic issues and internal banking features of the banks’ operations.
The article mainly focuses on the internal issues and discusses the price discrimination options
the banks have in different countries of presence. This article is important for the research on cash
currency exchange rates asymmetry, because it covers another aspect of the problem. The methods
used in the study by (Listra, Rahu and Vaiser, 2011) are of greater importance as well as the outcomes
of their study. They use such analysis instruments as regression analyses, statistical significance tests
12
etc. The results built are attributable to the European market and do not include the peculiarities of
Russian market, however their methods are of great importance.
The authors consider the factors that influence the spread difference across the different
countries in European Union. They elaborate on the spreads for the same large banking corporations.
For this reason, the set of influencing factors is limited. They automatically exclude from the scope the
factors that change across banks and which are caused by differences in internal procedures. So the
result is more focused on the market itself and not overwhelmed by the different players’ differences.
The outcome of the research (Listra, Rahu and Vaiser, 2011) is that the difference in the
exchange rates commissions across different countries of EU are significant. The authors demonstrate
that the spreads are narrower in the West-European countries and wider in the East-European.
However, the authors do not provide a theoretical basement for this discrepancy. Furthermore, they do
not discuss asymmetry in the rates.
The study (Hussain, 2011) analyzes the connection between the bid-ask spreads, trading
volumes, returns and volatility. The author creates a model, which shows the connection between the
key parameters of the market trading: spreads, trading volume and return volatility. The research
studies regularities and analyses the degrees of significance for each parameter.
This work is not directly about the FX market, but it describes a fundamental work concerning
organized dealer-made markets and is considered an important source of knowledge for creating an
analytical model for the hypothesis testing. The paper is a powerful fundamental basement for the
research on cash currency exchange rates asymmetry. This is a solid theoretical analysis of the dealermarkets and of the microstructures on such markets. This is, perhaps, one of the most relevant sources
of information for the present research, because it gives a well-developed theoretical background and
can be used as a basement for all the findings about the determinants of the asymmetry of the cash
currency exchange rates.
The research paper (Kyle and Obizhaeva, 2004) provides a powerful fundamental basement for
my research. This is a solid theoretical analysis of the dealer-markets and of the microstructures on
such markets. This is, perhaps, one of the most relevant sources of information for my research,
because it gives a well-developed theoretical background and can be used as a basement for all the
findings about the determinants of the asymmetry of the cash currency exchange rates.
The overall academic findings relevant for the research describe the following aspects of the
cash currency exchange rates asymmetry. The articles focus either on the spread size and its changes
13
across distinct markets, or on the parameters that explain the bid and ask quotes. (Listra, Vaiser and
Rahu, 2011) as well as (Hussain, 2011) describe the spread size and the significance of the spread
difference across different markets. The papers (Kyle and Obizhaeva, 2004), (Hussain, 2011) and
(Copeland and Galai, 1983) analyze the theoretical influencing factors on the size of the bid-ask
spread and discuss the potential determinants for the bid and ask quotes separately.
There have been no papers that treat the asymmetry as the major object of the research. Most
of the papers focus on the size of the bid-ask spread, volatility and liquidity. They mention the
asymmetry only indirectly; discussing that bid and ask prices may be sometimes determined by
different factors. As the papers (Copeland and Galai, 1983) and (Hussain, 2011) suggest, higher
volatility usually causes wider spreads. When the bid-ask spread becomes large enough in respect to
the major FX quote the presence of the asymmetry becomes evident and rather significant.
Paragraph 3. Russian Foreign Exchange Market Overview
According to Central Bank publications (Central Bank of Russia, 2016) and (9, 10) Russian
foreign exchange market has become very volatile in the recent years. This was determined by the
significant shifts in the oil prices and the volatility of returns on energy assets, tougher geopolitical
stakes and partial isolation of Russia from the global financial markets.
The cash currency exchange market is separate from the interbank FX trading floors. The
players and buyers, the liquidity and the speed of information flow are different on the retail exchange
market. That is why the main characteristics of the market: the prices are different from the regular
interbank FX prices.
Retail FX market for Russian population is extremely important. Russia, as well as many other
developing countries has an unstable economy. The high risk of a local financial crisis and, as a
consequence, currency devaluation makes people seek the more safe storage of value than the local
unstable currency (Savenkov, 2015). One of the most straightforward of them would be the foreign
currency in cash. The retail cash currency exchangers offer this service plentifully. That is why the
study of this market is of high interest nowadays.
The Russian retail FX market can be described as a highly concentrated market in terms of
currency choice. The most popular currency for conversion operations is the United States Dollar,
around 73-75% of all operations. The second most popular one is Euro, 23-25 of the overall retail FX
conversions. The other 1-2% account for all the other currencies against Ruble.
14
Aggregated Supply of Foreign Currency, 2015
1%
25%
74%
USD
EUR
Others
Figure 1. The relative portions of currencies in aggregated supply statistics for Russian retail FX
market 2015. Most of foreign currency sold to the banks is in USD, 74%. The next most popular
currency for sale among the population is EUR, 25%. Only 1% of all the foreign currency sold to the
dealers were neither USD, nor EUR: all the other currencies.
Aggregated Demand for Foreign Currency, 2015
1%
24%
75%
USD
EUR
Others
Figure 2. The pie chart of the portions in the aggregated demand for foreign currency in Russia 2015.
Most of foreign currency bought from the banks is USD, 75%. The next most popular currency for
purchase among the Russian population is EUR, 24%. Only 1% of all the foreign currency purchased
from the dealers were neither USD, nor EUR: all the other currencies.
15
Recently there have been research papers, concerning the retail bid-ask spread size on the
Russian market. However, the focus was once again more on the size of the spread, not on its shifts in
relation to the major FX rates.
The asymmetry has become evident on the retail market in the recent years. There have been
even cases, when the FX rate exceeded the ask prices and or sank below the bid prices (17). Apart
from the single events of out-of-range retail FX quotes, in the existing terms of volatility on the
market, some banks publish the consistently asymmetrical exchange rates.
These events have been taking place throughout the latest two years. However, they are out of
the regular scope of the financial market analysis. In fact, the asymmetry in the bid-ask spreads has
never been researched yet. That is why the topic is relevant, because it covers the research gap. The
uncommon in regular circumstances events, which cannot be explained by the existing theories, have
been present on the Russian retail foreign exchange market in the last two years.
The chosen research topic for the master thesis is “Determinants of asymmetry in cash
currency exchange rates”. The topic is of high importance to the financial theory. Firstly, it will give
an insight into the field of financial markets, which is rarely covered by academic research papers.
Most researches of the financial markets are rather broad and do not focus on specific, narrow
markets. Instead, the researches often generalize markets and describe certain relationships, which are
similar on different trading floors independently of their particular characteristics. The research at
hand will provide the analysis of several closely related markets with their peculiarities and will try to
spot the factors, which are specifically important for this particular market. Secondly, the research will
be up-to-date, thus relevant. The analysis will use the latest market quotes and trends and will base on
the current factors influencing the retail FX market, thus making it useful.
Paragraph 4. Theoretical Framework for the Research
The general microeconomics theory states that a market is comprised of the two sides: the
supply side and the demand side. These two parties interact and make their deals. The result of their
interaction is the market itself with its natural characteristics: the volumes traded and the price on the
market.
The analyzed market is the cash currency exchange market. We focus only on USDRUB and
EURRUB operations because these are the most frequent ones and account for 95-99% of the market.
The other currency pairs are much less liquid and require a separate research. There are 4 types of
transactions we analyze: USD to RUB, RUB to USD, EUR to RUB and RUB to EUR. In all of the
16
operations, the participants are similar: the dealers or the retail banks with cash currency exchange
services and the customers – the general population who want to convert their cash from one currency
to the other. In this paper the dealers will be referred to as the supply side, or the market makers. The
general public, who pays for the service of the exchange, will be called the demand side of the market.
In other words, the supply offers services of converting cash and has a range of four distinct
services. The demand side purchases the services from the demand side and chooses the suitable
service from the list of four direct and indirect transactions with EUR or USD against ruble.
The research of factors that influence the spread, or the price for the service, was structured in
accordance with demand and supply market segregation. The both sides were analyzed in respect to
the driving factors for them to enter the deals and to the means they use for the transaction. The set of
potential factors, influencing the asymmetry was created.
The market makers, or the supply side, operate in the following way. They buy the foreign cash
currency from the population with a discount and sell it to them with a premium. They clear their
balances on the interbank FX market, which is much more liquid. The cash FX market is much less
liquid and informationally-efficient, that is why the different prices are available at different banks at
the same time. At the same time the spread should not be too wide to cast away the customers. In other
words, the two main factors that are important for the dealers are the interbank FX quotes, because
they make profits by adding margins to it and the competitive factors, which determine the number of
transactions a market maker services (Listra, Vaiser and Rahu, 2011).
The retail FX market is heavily dependent on the interbank FX market. That is why this factor
influence will be primarily analyzed. The rivalry across retail banks issue includes not only
competitive pricing, but also differentiation strategies and locality factors. These issues are important
to the pricing, but are out of the scope of this research, because they need to be studied from bank to
bank individually. This research however is focused on the industry analysis and on the peculiarities
that are attributable to most of the players on the market.
As for the demand side, there are two crucial reasons for the public to buy or sell the foreign
currency: the first is the need to make transactions with foreign parties. The customers buy foreign
currency to pay for something abroad or sell foreign currency to convert their proceeds from foreign
parties into rubles. The second important reason is the expectations of inflation and/or devaluation. In
the periods of local currency depreciation, the public seeks a safe haven to protect their savings. When
17
ruble appreciates, the population decides to take advantage of it and buys rubles back for foreign
currency (15).
The first option, transactions with foreign parties, is assumed to happen on a rolling basis,
without any significant shifts over time. Except for the seasonal summer shifts in demand for foreign
currency, which occur due to the vacation schedule and travelling. The changes in consumer behavior
during ruble appreciation or depreciation against the basket of American and European currencies are
also the important factor for the research of the asymmetry on the retail FX market.
Paragraph 5. Managerial Application
As far as managerial applications are concerned, the research sets its goal in creating an
econometrical model for calculation of the fair value of the fee for cash currency conversion: the bid
discounts and ask premiums. The research is aimed to help the retail bank managers, who are
responsible for setting up the daily bid and ask rates.
In the periods of low volatility and of the stable currency quotes on the interbank markets, the
retail FX market is in equilibrium and the competition among banks drives the prices to the level of
transactional expenses (Savenkov, 2015). During the calm periods, the banks get their ask price by
adding a transaction fee and a margin to the interbank FX rate and get their bid prices by subtracting
instead of adding.
On the other hand, the FX market turmoil drives the spreads wide. The retail bank manager
responsible for retail FX transactions should not simply add for ask and subtract for bid the transaction
costs and margins to the base rate, but also consider the potential movements of the interbank FX price
and the existing rivalry. So the bank manager has to take into account profitability, risks of sudden
currency shifts and of the competitors, who can offer better prices. The bankers have to find a tradeoff
between casting away the customers by the enormous bid-ask spread and losing money in case of a
sudden price shift. The potential remedy to this problem is to establish an asymmetrical pricing for
selling and for buying. This allows the banks hedge their risks and still lets them stay competitive.
There are models that describe the dependence of the spread on the market volatility (Gtifa,
Maktouf and Labidi, 2015) and (Savenkov, 2015), however, none of them describes the amount of the
asymmetry. The current research gives a model that can be used by a bank manager as a benchmarking
tool. The bank manager will be able to see the asymmetry amount that theoretically corresponds to the
current market conditions. The bank manager will thus be able to compare his FX pricing to the
18
general market pricing to see the competitiveness level. Furthermore, the managers will be able to
analyze their internal mechanisms for setting the quotes.
Paragraph 5. Research Methodology
The research is focusing on the empirical sets of data and gives a description of the
determinants that influence the cash currency exchange spreads on the specific market. That is why the
results are not aimed to be universally applicable for cash currency markets but work for the particular
circumstances at hand. The research methodology corresponds to the goal and objective stated and is
designed to get practical numerical solution to the problem stated. Therefore, the main research
method used is data collection and data analysis. The analysis of the data is both quantitative and
qualitative.
First, the description of the quantitative analysis methods will be given. At the first stage, the
data about the banks’ exchange rates was collected and primarily analyzed. Here the sample
visualization methods were used (bid-ask spreads over time visualization, frequency histograms,
volatility graphs). This kind of analysis helps to check for any peculiarities that might lead to a
conclusion of one or the other factor may potentially be the determinant of asymmetry. Then the
descriptive statistics were calculated for the periods, where the specific patterns are present. The
results of these calculations lead to a number of formal conclusions. The next stage was running the
regression models and analyzing the levels of significance for the different factors. The data set in the
analysis is a list of quotes that different banks publish over almost two years. For each set of quotes,
there are two corresponding navigators: the date and time of publishing the quote and the publishing
bank, In other words, the data is both cross-sectional: ~200 banks, and time-series: there is data for
almost every day from June 2014 till March 2016. Therefore, panel data analysis toolset was
exploited. The data set was tested to have random or fixed effects. Hausman test suggested, the data
should be analyzed with a fixed effects approximation.
Second, the qualitative analysis of the data itself and of the findings of the quantitative section
was performed. The qualitative analysis is the general analysis of the data in accordance with
theoretical frameworks. The key influencing factors from theoretical perspective have been discovered
and structured. The formalized groups of factors were then explained in accordance to the economical
and financial theory. Then their influence and the degree of their influence was summarized too in
order to give a formal output of the research applicable to the banks management.
19
The previous studies that use the methodology described above are the articles mentioned in
the literature review section of this research paper. Especially those, which are highly empirical
studies, because their goals are rather close to what is dealt with in this research. These studies are
(Gtifa, Maktouf and Labidi, 2015), (Listra, Vaiser and Rahu, 2011) and (Hussain, 2011).
All in all, in the first chapter the holistic analysis of the research was presented. It was shown
that there is a research gap to be covered by the research. It was discussed that the research goals and
objectives are relevant to both academic research in the field of finance and to the managerial
application of the outcomes of the research. The literature review was given in order to provide an
overview of the existing situation in the field of the research and to show what had been covered so far
by researchers. The methodology section elaborated on the ways to solve the problems. The research
goal is to determine the factors to explain the asymmetry in cash currency exchange rates. The
literature used concentrates on the FX market analysis and on the market study methods. The major
methods are the statistical data analysis methods and the regression analysis methods.
20
Chapter 2. Dataset Overview
This chapter provides an overview of the data set used in the research work. Due to the volume
of the data set, only the key information of interest will be summarized in this chapter. The following
features of the data set will be disclosed: the origin and the description of importance of each of the
variables, the explanation of the frames used to limit the information needed, the key statistics and
further limitations imposed on the data set and the data required for the deeper research will also be
discussed here.
Paragraph 1. Dataset Origins
To analyze the problem of the asymmetry determinants, four major sets of data were used.
These data sets describe the asymmetry and the influencing factors, suggested by the theoretical
analysis:
(1) The cash currency exchange rates published by the banks,
(2) FX market quotes for EUR/RUB currency pair,
(3) FX market quotes for USD/RUB currency pair
(4) Russian Central Bank statistics of the shifts in demand for foreign currency among Russian
population.
The first dataset was treated as the dependent variable set, while the latter three sets from the
FX market and from Central Bank of Russia were used as the independent sets of variables.
In accordance with (Savenkov, 2015) during the low volatility periods, the spread is
determined mainly by the internal transactional costs. The spreads are narrow and the asymmetry is
not significant. Starting from the time, when oil prices peaked, the volatility increased, the ruble
depreciation trend became stong and the asymmetry emerged. The time period chosen for the research
starts in the end of May 2014 and ends in the end of March 2016. This time frame is the most
descriptive, because it starts before the oil market turmoil (see Figure 3) and spans until the latest
dates.
21
Figure 3. Historical Brent Oil prices. The historical prices graph shows the evolution of the oil prices
over the latest 2 years. Oil prices and Russian Ruble rates are highly correlated. With the fall in oil
quotes Russian Ruble market turmoil started. This is the most interesting period for research, because
of the considerable bid-ask spreads set by the retail banks: the scope of the spreads demonstrated the
asymmetry. The observation set for this research begins in summer of 2014, when the oil prices were
high and stable (8).
22
Figure 4. Historical USDRUB and EURRUB quotes. The graph shows the appearance of the positive
trends in US Dollar and Euro direct quotes in the summer 2014. This is the beginning of the FX
market turmoil, the most interesting period for the research, because the huge changes in the market
rates made the retail bankers impose high bid-ask spreads for cash currency conversion transactions
(9, 10).
Oil price turmoil began in summer 2014. The high correlation of oil prices and Russian Ruble
prices in US Dollars and in Euros lead to the turmoil in Russian FX market. The core shifts in the FX
market made the retail bankers worried about their profits and they started adjusting to the changing
FX quotes. High volatility on the market has led to the high fees on cash foreign currency exchange
transactions. That is why this period is the most interesting for the research.
Paragraph 2. Variables’ Description
The problems to be solved by data set collecting were the following:
(1) In order to analyze the retail banks’ cash currency exchange rates, the rates must found
first.
(2) To analyze the asymmetry, one should determine the central rate to calculate the
asymmetry.
23
(3) Influencing factors study requires having the data for a quantitative description of these
factors
The problem (1) stated above: collecting the banks’ cash foreign exchange rates [8], was the
most challenging task to carry out. The first and the most straightforward way to collect the
information was to go to Central Bank web page to get a list of banks offering cash currency exchange
services. Then, using the list of the banks, go to their respective websites and collect the data over the
last years. Here the main hindering obstacles appeared.
Most of the banks do publish their rates publicly on their websites. However, the data on the
websites is difficult to download. That is to say, the information is publicly available and one can read
it or otherwise get the information on whichever date one desires. Nevertheless, it is practically
impossible to download the data for a deep quantitative analysis.
For example, PAO Sberbank of Russia on its web site (12) has an archive of the foreign
exchange rates. Each update is offered in a separate excel file available for downloading. PAO
Sberbank of Russia updates its rates several times a day, which makes it necessary to download and
aggregate at least a thousand of separate files only for Sberbank. Another negative example is Russian
Standard Bank. This bank also stores the information publicly, but the format of a daily updated graph
with historical rates (13).
The solution to this problem was to use the online foreign exchange rates aggregators
platforms. Their websites, for example (11) is analogous to skyscanner.com for airlines tickets or
booking.com for hotels. The service gives comparison of the banks exchange rates and stores the
historical rates in a table format. The rates downloaded from the website (11) were used as a primary
source of data about the market makers’ quotes.
However, the website (11) is not the official source of information. For this reason, the data
obtained required checking for inconsistences with the official rates, published by the banks
themselves. Checking the whole data set was impossible due to the same reasons as collecting the
whole set. There have were approximately 170 thousand of observations, so direct checking would be
also impossible. Therefore, the sampling method was used. Following the sampling method, we took
arbitrarily chosen observations and compared the observations with the official information from the
official banks websites. Some banks store the historical data; the other banks do not, so for the banks
without the historical rates, the latest rates were checked. It turned out that all the banks checked did
publish the same quotes as the aggregator (11) claimed.
24
Paragraph 3. Dataset Adjustments
During the data checking, the next problem aroused. Not all of the banks could be checked:
some of the banks lost their licenses over the course of the dataset period and their official webpages
lost their authenticity. That is why a number of banks had to be excluded from the list.
There has been a total of 197 different banks in the data set: 197 banks published their retail
foreign exchange rates during the period from June 2014 till March 2016 in Moscow. However, not all
of the banks published their results regularly and during the whole time period. That is why 123 banks
from the initial list were excluded from the further consideration. The full list of the banks and the list
of the excluded banks can be found in the appendix 1.
In order to get the most relevant banks for further consideration, Pivot table technology in MS
Excel was used. The pivot table was created. The major column was with the bank names and the two
supporting columns were with the first and last publishing date of cash currency conversion rates. If
the first and last rate publishing dates were not within a week from starting and ending period, it was
assumed that the bank was either inactively participating in retail foreign exchange or was not
operating during the entire period of the research.
Figure 5. During the crisis period Russian banking sector underwent serious changes. Many banks
lost their licenses or went bankrupt. The research used the quotes only of those banks, who have been
publishing their rates consistently throughout the time of analysis without periods of no rates
updating. The pie chart demonstrates, that of 197 bank that were publishing retail FX rates, only 74
did it uninterruptedly.
25
Figure 6. This pie chart shows that the excluded 2/3 of the banks accounted for 1/3 of all the
observations. This supports the argument that the excluded banks did not publish their retail FX rates
regularly. Apparently that means these banks were not engaged into cash currency exchange
operations.
After the list of banks was narrowed down approximately threefold, the number of
observations decreased approximately by 35% (from 170 thousand to 109 thousand of observations).
This result supports the idea of decreasing the data set. The large number of banks (123) that did not
meet the requirements did not publish their quotes regularly.
As the adjusted data set of the bank exchange rates was obtained, the next problem (2) had to
be solved: how to determine the asymmetry in cash currency exchange rates. The asymmetry consists
of two parts. One is the set of rates themselves. The other is the benchmark, which serves as a
comparison milestone for the rates.
The first idea was to use the officially published by Central Bank of Russia foreign currency
rates. The idea seemed perfect. However, it turned out that the Central Bank Rates are published for
other than transaction making purposes. These rates serve as an accounting reference, but not as the
real exchange rate. For that reason, another rate was taken.
This alternative rate is the rate used by the banks in their operational transactions: the FX rate
from the interbank market. The retail banks make conversion operations on the FX. At the same time,
they buy and sell cash currency on the retail market, thus becoming intermediaries between the two
markets. That is why not the Central Bank official rate, but the FX market rates were used.
26
The interbank FX in fact is also a dealer made market and there are the bids and asks as well.
However, the bid-ask spreads on FX are hundreds to thousands times smaller than the retail spreads.
These two magnitudes are incomparable and thus the interbank FX spreads can be ignored, by using
the mid quotes only. This was what had been done and that is why the FX rates for EUR/RUB and
USD/RUB currency pairs were taken (9, 10).
When the data from the FX market was collected, the asymmetry was determined as follows.
The discounts to buying foreign currency from customers was determined to be open price on that day
on the MICEX stock exchange minus the bid rates set by the banks: Bid. Discount = Open. FX − Bid.
The premiums to sell foreign currency to customers was determined as the ask rate published by the
bank minus the open price quote from the MICEX Moscow stock exchange: Ask. Premium = Ask −
Open. FX. Then the absolute asymmetry equals the value of the ask premium minus bid discount:
Asymmetry = Ask. Premium − Bid. Discount . The price of USD and EUR in rubles changed
significantly over the period of observations. Therefore, it was considered more accurate to use a
relative scale to describe the asymmetry in percentage points. The base for the percentage asymmetry
calculation was the open quote on the FX market: Asymmetry =
Ask.Premium−Bid.Discount
Open.FX
.
Overall, the asymmetry formula takes the following form:
Asymmetry =
(Ask − Open. FX) − (Open. FX − Bid)
Open. FX
The results of the descriptive statistics present the general information about asymmetry in the
sample of banks that published the rates regularly.
Asymmetry EUR
Mean
0,34%
Standard Error
8,64E-05
Median
0,26%
Standard Deviation
2,89%
Sample Variance
0,0008
Kurtosis
21,83818
Skewness
1,561392
Range
89,98%
Minimum
-25,98%
27
Maximum
64,02%
Count
111982
Figure 7. Descriptive statistics for the asymmetry in EUR against RUB rates. The central tendency
results are close to zero, however the standard deviation is large as well as the range of the sample.
The offset of positive and negative asymmetry periods.
Asymmetry USD
Mean
0,67%
Standard Error
0,01%
Median
0,50%
Standard Deviation 4,94%
Sample Variance
0,0024
Kurtosis
24,46
Skewness
2,39
Range
104,14%
Minimum
-29,16%
Maximum
74,98%
Count
109760
Figure 8. Descriptive statistics for USDRUB rates. The central tendency results are close to zero,
however the standard deviation is large as well as the range of the sample. This could be explained by
the offset of positive and negative asymmetry periods.
The descriptive statistics tables for both US Dollar against Ruble and Euro against Ruble show
rather similar conclusions. On average the bias for both currency pairs is directed towards the ask
price: the ask price premium is generally higher than the bid price discount. However, the ranges of
the asymmetries are very broad and cover both directions. Although the mean and median are rather
small, the standard deviations and range demonstrate the existence of the nonzero asymmetry in the
sample. The explanation is that there are periods of positive and of negative asymmetry on the market.
These two types of periods offset each other, bringing the total average close to zero.
The central tendency indicators imply the greater positive (towards the ask price) asymmetry in
the dollar quotes. However, the standard deviation measures are also higher for dollars. This is
28
determined by the greater volume of USD/RUB transactions as compared with EUR/RUB. Larger
volumes of transactions correspond to the greater demand for the product, greater margins: bid-ask
spreads and greater volatility consequently.
As the descriptive statistics tables show, the asymmetry exists and needs to be explained. The
last question (3) for data collecting is about the influencing factors. It is desirable to have data about
all the possible factors that could determine the asymmetry.
As discussed in the first chapter, the retail cash foreign exchange is a market with the buyers
and the sellers. When an active market exists, then there always are the demand and supply sides, both
having their needs and being subject to a range of factors. We structure the potential determinants in
the following way. The supply side is directly related to the FX market price movements and to the
market competitive factors. The latter are specific for every market maker and are not considered in
this research. So the FX market data becomes the source of the information for the supply side
analysis. The demand is driven by the transactions with foreign parties and by the need to store value
in foreign currency. The transactions with foreign parties are impossible to account for in this research.
These both factors are included into the statistics by Central Bank of Russia. The statistical report
contains information about the volumes of the cash FX conversions in Russia on a monthly basis.
29
Figure 9. Central Bank of Russia monthly aggregated statistics of total conversion operations made by
individuals. The total numbers are in US Dollars millions. The graph shows that most of the time the
net conversion operations account is positive. The population in general buys more foreign currency
than sells.
The Central Bank Statistics is presented on the figure above. The monthly sums of purchased
and sold foreign currency in US dollars mln. demonstrate increased volume of FX market activity in
December 2014 during the FX crisis and market panic. The seasonal shifts in demand happen
primarily in the middle of Spring and in the middle of autumn.
To make a conclusion, the second chapter gave an overview of the data set. Such issues as
timing and key data sources were described. The challenges and hurdling circumstances in collecting
the data were discussed and the solutions to the problems were described. The outcomes of the dataset
adjustments were the decrease of the number of observations and number of retail banks offering cash
currency exchange services.
The asymmetry itself was determined as the difference between ask premiums and bid
discounts divided by the open price on the MICEX USDRUB. The relevant descriptive statistics and
the conclusions drawn from the statistics were stated. The major fact is that the asymmetry exists and
the bias is aimed at the increased ask prices as compared to the FX market. Furthermore, the range of
30
the asymmetry is surprisingly high. Sometimes the asymmetry gets as great as approximately 30% of
the FX price. The potential influencing factors were preliminarily analyzed and some conclusions
were made. The supply side factors were analyzed with historical interbank FX quotes. The demand
side factors were studied with Central Bank statistics for aggregated monthly customer spending on
retail FX transactions.
31
Chapter 3. Dataset Statistical Analysis
This chapter provides the overview of the analytical research steps. The analysis was structured
in the following way. At first we study the asymmetry in general and try to analyze its peculiarities.
Then we switch to the factors datasets analysis and making conclusions. We find out there are strong
similarities in the datasets for the asymmetry and of the factors. Then we arrive at studying how the
datasets are related to one another. Finally, we create a focused model that quantitatively describes the
asymmetry.
Paragraph 1. Asymmetry Analysis
As the descriptive statistics tables state, the asymmetries on average are positive and are biased
towards the ask quotes. The asymmetry is not exclusively positive and has high standard deviations.
The maximal ranges of the asymmetries in samples are around 90% for Euro and around 104% for US
Dollars. This suggests that the asymmetry direction changes over time making it close to zero on the
cumulative basis. In order to visualize the asymmetry two graphs are given: one for the asymmetry in
US Dollars against ruble quotes and the other one for Euro against ruble prices.
Figure 10. The time series graph of the asymmetries in USD/RUB retail prices. The graph shows high
volatility and period of positive and negative medians.
32
Figure 11. The time series graph of the asymmetry in EUR/rub retail prices. The graph shows high
volatility and period of positive and negative medians.
The visual analysis suggests the asymmetry takes place on a regular basis. Either the ask
premiums or the bid discounts are greater during the certain time periods. The positive asymmetry
directed at the ask premiums implies that that higher demand of the population for cash Dollars and
Euros outweighs the supply of it.
Paragraph 2. Determinants Analysis
In this section, the analysis of the influencing factors dataset is described. We begin with the
supply aspect of our investigation and study the behavior of the FX market indicators during the
observation period. Then we turn to the demand aspect and analyze the statistical data over the period.
Then we compare the results of the analyses of both aspects and make inferences about that.
33
I
II
III
IV
V
VI
VII
Figure 12. The candlestick chat of USDRUB instrument. The chart shows seven distinct areas with
different price behavior. These areas are market with roman letters from one to seven. The periods
differ in the trend direction and price volatility.
34
I
II
III
IV
V
VI
VII
Figure 13. The candlestick diagram shows the behavior of EURRUB instrument on the interbank FX
market. The graph shows seven different periods of price movements. They include Euro appreciation,
depreciation and stagnation. The periods are different in terms of their volatility or trend direction.
The supply aspect analysis leads us to a set of seven periods with distinct price behavior
patterns. The first period starts in June 2014 with the beginning of the observations. It lasts till the first
of November 2014, when the first FX market panic signs appeared. This period can be characterized
as a smooth uptrend with relatively low volatility. The period number two starts on the 1st of
November 2014 and spans over the FX market panic period through to the end of January 2015. This
period demonstrated high volatility and steep trend of ruble depreciation. Third period begins in
February 2015. This was the first reversal trend. Ruble started appreciating. The appreciation was
volatile and lasted till the mid May 2015. Throughout the Summer 2015, the period IV, the ruble went
down again. The volatility was lower that before, but still greater than during the first period. In
September – October 2015, section number V, there was a medium-volatile side trend present, which
35
turned to strong upsloping trend in the period VI: from mid October 2015 till mod January 2016. The
final period, section number VII demonstrated a downsloping trend with high volatility.
Both instruments, USDRUB and EURRUB, show almost identical performance. The more
accurate analysis results are presented in the tables below.
USDRUB Quotes Analysis
Beginning
End
of Period
Period
of Visual Characteristics
Volatility
Monthly
Change
average:
over the period
(High-Low)/
open
1 1-Jun-14
1-Nov-14
Smooth&slow uptrend
1,07%
4,64%
2 1-Nov-14
31-Jan-15
volatile uptrend
4,95%
20,47%
3 1-Feb-15
20-May-15
Volatile downtrend
3,12%
-7,88%
4 20-May-15
31-Aug-15
Medium-Volatile uptrend
2,39%
9,46%
5 1-Sep-15
15-Oct-15
Medium-Volatile stagnation
2,37%
-1,27%
6 16-Oct-15
20-Jan-16
Medium-Volatile uptrend
2,07%
7,64%
7 21-Jan-16
22-Mar-16
Volatile downtrend
2,74%
-5,54%
Figure 14. The table summarizes the analysis of USDRUB quotes from the MICEX market. There are
7 periods that are characterized by volatility, direction and steepness of the trend.
36
EURRUB Quotes Analysis
Beginning
End
of Period
Period
of Visual Characteristics
Volatility
Monthly
Price
average:
Change over the
(High-low)/
period
open
1 1-Jun-14
1-Nov-14
Smooth&slow uptrend
1,13%
2,03%
2 1-Nov-14
31-Jan-15
volatile uptrend
5,26%
14,95%
3 1-Feb-15
20-May-15
Volatile downtrend
3,10%
-8,12%
4 20-May-15
31-Aug-15
Medium-Volatile uptrend
2,71%
9,74%
5 1-Sep-15
15-Oct-15
Medium-Volatile stagnation
2,71%
-1,58%
6 16-Oct-15
20-Jan-16
Medium-Volatile uptrend
2,38%
7,66%
7 21-Jan-16
22-Mar-16
Volatile downtrend
3,25%
-5,20%
Figure 15. The table summarizes the analysis of USDRUB quotes from the MICEX market. There are
7 periods that are characterized by volatility, direction and steepness of the trend.
If we compare the FX market quotes and the asymmetry chart period by period, we see there is
a certain pattern. The steeper upsloping trends take place at the same time periods as the positive
median asymmetries. The downsloping trends occur simultaneously with the negative average
asymmetry periods.
37
Chapter 4. The Results
This chapter presents the results of the research. The statistical analysis and the econometrical
model are provided here. After that, the conclusion is drawn. The strengths and the weaknesses are
described. The potential areas for improvement are also presented. As discussed in the previous
chapters, the determinants are structured in accordance to the parties on the market: the demand – the
general public, and the supply – the dealers or the banks.
Paragraph 1. Demand Aspect
We begin our analysis with the demand side study. The hypothesis is that the biased consumer
FX conversion operations provoke the dealers impose higher tariffs in the more demanded types of
transactions. For example, if the public prefers buying Dollars to selling dollars, the banks are
expected to set higher ask premium than the bid discount for operation with US Dollars.
The statistical data about the demand, provided by the Central Bank of Russia, as well as the
asymmetries in pricing USD and EUR are presented in the chart below.
38
Figure 16. The chart represents monthly net FX conversion operations with cash RUB in USD
millions and average monthly asymmetries for USD and EUR cash exchange rates in percentage
points.
The chart represents the relation between the average asymmetry in the cash currency
exchange rates and the net account of currency conversion operations. These two pairs of elements:
USDRUB asymmetry with net conversion and EURRUB asymmetry with net conversion demonstrate
the strong connection.
The degree of their relation was measured by the correlation metrics. The correlation matrix for
the three variables is presented on the next figure.
Asymmetry in Asymmetry in Net Conversion
USD
EUR
Balance
Asymmetry in USD
100,0%
-
-
Asymmetry in EUR
69,7%
100,0%
-
51,8%
100,0%
Net Conversion Balance 70,5%
Figure 17. Correlation matrix for monthly average asymmetries in USD and EUR retail prices and for
the net retail FX conversion. It turns out, that the correlation with net conversions balance coefficient
39
for USD is higher than for the EUR. This can be explained by the fact, that USD transactions volume
is much greater than EUR. Therefore, the retail USD market is more liquid and better reflects the
shifts in demand.
The statistical analysis demonstrates a strong connection between the asymmetry magnitudes
and the net conversion balance for both currency pairs. The conversion balance has higher correlation
coefficient with asymmetries in US Dollar prices that with the Euro rate asymmetries. The reason
behind that effect is the difference in volumes of the respective transactions. The transaction type
volumes mean different liquidity levels of the markets. USD retail market is more liquid and better
reflects the changes in demand.
The analysis of demand side shifts and changes in asymmetry patterns has demonstrated a
strong positive connection. The higher the demand for foreign currency from the population, the larger
is the ask premium than the bid discount for a unit of foreign currency. And vice versa: the higher the
supply of foreign currency from the population, the greater the bid discounts as compared to the ask
premiums. This proves the hypothesis, that the higher demand for certain financial products leads to
higher transaction costs for that operation.
Paragraph 2. Supply Aspect
The other side of our analysis focuses on the supply side of the retail foreign exchange market:
on the market-makers and the issues relevant to them. As mentioned above, there are two types of
factors, which influence the market makers. These are the interbank FX market price, which is equally
important for all the dealers, and the competitive factors, which are specific for every single dealer and
are therefore out of the scope of this research. So the focus in this research is made on the influence
the FX market has on the retail bid and ask quotes.
As mentioned above, there are 7 consecutive periods on the FX market with different trends
and characteristics. These periods were analyzed in terms of the bid-ask spread asymmetry. Our sets of
data for USDRUB and EURRUB were split into 7 parts each. The rates for each of the periods were
analyzed for each bank. The averages, medians and standard deviations over periods were calculated
for each of the banks. After that, the frequency histograms for average asymmetries in different banks
for each period were created.
The hypothesis here is that the dealers, who see movements in the FX rates set higher
commissions in the direction of the trend.
40
The histograms (See Appendix 2) demonstrated narrow peaks of average asymmetry. Positive
median asymmetry was present when the uptrend was on the market. The asymmetry was close to zero
when there was a side trend or stagnation on the market. The asymmetry was negative, when Dollar
and Euro sank against Ruble.
In other words, when the US Dollar and Euro appreciated against Ruble, most of the banks
required higher margin for selling the currency to the public. When the Ruble strengthened, the banks
required higher margins for selling Rubles to the public. When there was stagnation on the FX market,
the banks published equal margins: bid discounts equal to ask premiums, for selling and for buying the
cash foreign currency.
Beginning
of Period
1-Jun-14
1
2 1-Nov-14
3 1-Feb-15
4 20-May-15
5 1-Sep-15
6 16-Oct-15
7 21-Jan-16
Figure 18. The
End of
Period
USDRUB Asymmetry Analysis
Volatility average: Monthly Change
(High-Low)/ open
over the period
Average asymmetry
published by the greatest
number of banks
1-Nov-14
1,07%
4,64 %
0,51 %
31-Jan-15
4,95%
20,47%
4,51 %
20-May-15
3,12%
-7,88%
-1,67%
31-Aug-15
2,39%
9,46 %
2,25 %
15-Oct-15
2,37%
-1,27%
-0,03%
20-Jan-16
2,07%
7,64 %
1,63 %
22-Mar-16
2,74%
-5,54%
-1,06%
chart of the USDRUB retail quotes asymmetry. The periods of stronger trends on the
market correspond to more significant average asymmetry.
EURRUB Asymmetry Analysis
Beginning
of Period
End of
Period
Volatility average:
(High-Low)/ Open
1-Jun-14
1-Nov-14
1
31-Jan-15
2 1-Nov-14
20-May-15
3 1-Feb-15
4 20-May-15 31-Aug-15
15-Oct-15
5 1-Sep-15
20-Jan-16
6 16-Oct-15
22-Mar-16
7 21-Jan-16
Figure 19. The chart of the EURRUB
Monthly Price
Change over the
period
Average asymmetry
published by the greatest
number of banks
1,13%
2,03 %
0,54 %
5,26%
14,95%
2,54 %
3,10%
-8,12%
-0,61%
2,71%
9,74 %
0,95 %
2,71%
-1,58%
-0,15%
2,38%
7,66 %
0,77 %
3,25%
-5,20%
-0,38%
retail quotes asymmetry. The periods of stronger trends on the
market correspond to the more significant average asymmetry.
The two tables above describe the general patterns of retail banks pricing policy for currency
conversion services. It is evident from the table, that during stable trends of dollar strengthening, the
41
asymmetry is positive. During the reversal trend directions, the asymmetry is negative on average. The
general result is clear and corresponds to the expectations of the demand and supply analysis. This
supports the hypothesis, that the banks set up asymmetries in the direction of the FX trends. However,
it is desirable to have a more precise model for the retail FX transactions pricing.
Paragraph 3. Asymmetry Regression Model
Due to the dataset limitations, we assume here that the FX market price movements have the
most significant influence on the market makers. The model is based solely on the data about the FX
market prices. We have seen so far that the dealers apparently take the FX market trends into account
for publishing the retail exchange quotes.
It is also assumed that the banks which offer retail foreign exchange services exploit only the
historical data from the market. In other words, on the date publishing the bid-ask spread asymmetry
depends only on the previous to that date FX price movements. The FX market is very liquid and the
price changes are very frequent. It is practically impossible to base the model on all of the price
changes. Furthermore, the scope of the research is different. Most of the banks publish their rates once
a day or even more rarely. That is why it is probable that the banks focus on the price changes over the
full trading sessions.
The results we have so far are: the descriptive statistics for USDRUB and EURRUB
asymmetry in retail exchange rates, the market trends over the observation period, the correlation of
the average monthly asymmetry and the net conversion balance. In all of the studies, the results for
USD and for EUR were similar. That is why, we build the model symmetrically for both of the
currency pairs.
The hypothesis on this stage is that on average the banks determine the asymmetry in their bidask spreads according to the most recent history of the interbank FX market price movements.
We use the asymmetry as a dependent variable and the historical price changes as the
independent ones. We start with the FX price change during the date of publishing the retail rate, say
day [0], then we include the overnight change from the end of trading session a day ago [-1] and the
current day [0]. We look at the price change during the regular trading session the previous trading day
[-1]. The fourth point is the night change between the day [-1] and the day [-2]. The price change
during the trading session two days ago [-2] is the deepest point into the history.
42
The list of the independent variables for the asymmetries in cash foreign exchange rates are:
(1) Trend of Today (retail rate publishing date)
Trend.TDY = Close.FX.DAY0 – Open.FX.DAY0
(2) Trend of Today Night
Trend.TDYNGHT = Open.FX.DAY0 – Close.FX.DAY(3) Trend of Yesterday
Trend.YSTD = Close.FX.DAY-1 – Open.FX.DAY-1
(4) Trend of Yesterday Night
Trend.YSTDNGHT = Open.FX.DAY-1 – Close.FX.DAY-2
(5) Trend of the Day Before Yesterday
Trend.DbfYSTD = Close.FX.DAY-2 – Open.FX.DAY-2
We take USD FX rates for asymmetry in US Dollar cash exchange quotes, and EUR FX rates
for Euro respectively.
The econometric model is formulated the following way for the USDRUB cash currency
exchange rates asymmetry:
𝐴𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑦. 𝑈𝑆𝐷 = 𝑎 + 𝑝1 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝑇𝐷𝑌 +
+𝑝2 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝑇𝑅𝑌𝑁𝐺𝐻𝑇 + 𝑝3 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝑌𝑆𝑇𝐷 +
+ 𝑝4 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝑌𝑆𝑇𝐷𝑁𝐺𝐻𝑇 + 𝑝5 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝐷𝑏𝑓𝑌𝑆𝑇𝐷 + 𝜉
Where,
a – is the intercept term,
𝑝𝑖 – the regression coefficients,
𝜉 – the error term.
The model for the EURRUB retail conversion rates asymmetry takes the following form:
𝐴𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑦. 𝐸𝑈𝑅 = 𝑏 + 𝑞1 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝑇𝐷𝑌 +
+𝑞2 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝑇𝑅𝑌𝑁𝐺𝐻𝑇 + 𝑞3 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝑌𝑆𝑇𝐷 +
+ 𝑞4 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝑌𝑆𝑇𝐷𝑁𝐺𝐻𝑇 + 𝑞5 ∙ 𝑇𝑟𝑒𝑛𝑑. 𝐷𝑏𝑓𝑌𝑆𝑇𝐷 + 𝜁
Where,
b – is the intercept term,
43
𝑞𝑖 – the regression coefficients,
𝜁 – the error term
The data we are dealing with is panel data. There are quotes for each date since June 2014 till
March 2016 issued by each of the 74 banks during the observation period. A set of one bank rates over
time is time series dataset, a set of rates of different banks for each particular date is cross-sectional
date. The data is both cross-sectional and time-series, so the panel data analysis methods was applied.
The both datasets were analyzed on be subject to fixed or random effects analysis methods with
Hausman’s tests. The results stated clearly, that the fixed effects approximation must be used for both
samples.
Fixed-effects (within) regression
Group variable: BankNumber
Number of obs
Number of groups
=
=
43351
74
R-sq:
Obs per group: min =
avg =
max =
150
471.2
660
within = 0.1510
between = 0.0668
overall = 0.1449
corr(u_i, Xb)
= 0.0041
F(5,43254)
Prob > F
=
=
1538.61
0.0000
----------------------------------------------------------------------------------------DependentAsymmetry |
Coef.
Std. Err.
t
P>|t|
[95% Conf. Interval]
------------------------+---------------------------------------------------------------trendtoday |
.0063349
.0001161
54.54
0.000
.0061072
.0065625
trendtodaynight |
-.018564
.0003962
-46.85
0.000
-.0193406
-.0177874
yesterdaytrend | -.0026916
.0001408
-19.11
0.000
-.0029676
-.0024156
trendYesterdaynight |
.0002942
.0003042
9.67
0.000
.0023458
.0035383
TrendDaybefore |
.0008638
.0001294
6.68
0.000
.0006102
.0011174
_cons |
.0014959
.0001169
12.79
0.000
.0012667
.0017251
------------------------+---------------------------------------------------------------sigma_u | .00603323
sigma_e | .02392302
rho | .05979824
(fraction of variance due to u_i)
----------------------------------------------------------------------------------------F test that all u_i=0:
F(91, 43254) =
26.55
Prob > F = 0.0000
Figure 20. Panel fixed effects regression analysis results for USDRUB retail rates asymmetry as a
function of FX price trends during the tree previous days and two previous nights.
44
Fixed-effects (within) regression
Group variable: Bank
Number of obs
Number of groups
=
=
40525
89
R-sq:
Obs per group: min =
avg =
max =
146
455.3
660
within = 0.2534
between = 0.1074
overall = 0.2428
corr(u_i, Xb)
= 0.0048
F(5,40431)
Prob > F
=
=
2745.01
0.0000
-------------------------------------------------------------------------------AsymmetryEUR |
Coef.
Std. Err.
t
P>|t|
[95% Conf. Interval]
---------------+---------------------------------------------------------------TrendTDY |
.0064936
.0000759
85.55
0.000
.0063449
.0066424
TrendTDYnight | -.0113774
.0002656
-42.83
0.000
-.011898
-.0108568
TrendYSTD | -.0036187
.0000817
-44.28
0.000
-.0037789
-.0034585
TrendYSTDnight | -.0009618
.0002072
-4.64
0.000
-.001368
-.0005557
TrendBfYSTD | -.0001344
.0000797
-1.69
0.092
-.0002906
.0000218
_cons |
.0028065
.0001043
26.91
0.000
.0026021
.003011
---------------+---------------------------------------------------------------sigma_u | .00553349
sigma_e | .02073307
rho | .06649474
(fraction of variance due to u_i)
-------------------------------------------------------------------------------F test that all u_i=0:
F(88, 40431) =
29.21
Prob > F = 0.0000
Figure 21. Panel fixed effects regression analysis results for EURRUB retail rates asymmetry as a
function of FX price trends during the tree previous days and two previous nights.
The regression analysis for both of the variables shows high degree of significance of both of
the models. The probabilities that the coefficients are statistically insignificant are close to zero. This
means that the hypotheses of statistically insignificant regression coefficients for both models are
rejected. It should be mentioned that the asymmetry is measured in percentage points, so the results
should be multiplied by 100, to arrive at percentage form.
The coefficients become closer to zero, as the number of period they are responsible for goes
deeper in the history. This means that the linear process is dependent less of the information about the
older periods. To make things simple, our process has memory and that memory is poor: newer
information is stored better than the older.
If we translate it into the real terms, it means that the managers of the retail banks, who are
responsible for assigning the bid and ask rates for cash currency exchange, pay more attention to the
most recent information and care less about the older data. In general, the asymmetry in cash foreign
exchange rates is measured as a weighted sum of changes in trends of previous periods.
45
USDRUB asymmetry (%) EURRUB asymmetry (%)
Trend TODAY
0.63
0.65
Trend TODAY night
-1.86
-1.14
Trend YESTERDAY
-0.27
-0.36
Trend YESTERDAY night
0.03
-0.09
Trend the DAY before YESTERDAY
0.09
-0.01
Figure 22. The summary table of the regression models for asymmetry in USDRUB and EURRUB
currency pairs. The similar features are that the latest trend is used with a positive sign, what was
expected. The two previous periods coefficients are used with the negative sign. The greatest impact
on the calculated asymmetry, in terms of the coefficient magnitude, have the overnight changes in the
FX price, off the regular trading sessions.
The results of the econometrical modelling are summarized in a table below. Generally, the
banks follow the logic of imposing positive asymmetry, if the price is going up in the current moment
currently. The also use the contrarian strategy against the price changes during the previous night and
previous trading session. So they react in an opposite way to the information from the previous night
and the previous trading day.
The low magnitudes of P>|t| values and decreasing weight of trends in historical depth
supports the hypothesis, that the most recent changes on the FX market influence the bid-ask spread
asymmetry most.
The main strength of the finding is that it gives a clear picture of the factors, influencing the
asymmetry in bid-ask spreads. It covers both aspects of the market: the supply and demand and
focuses on 99% of all transactions: USD and EUR transactions. The empirical results meet the
theoretical expectations.
The key area for improvement is the deeper analysis of the consumer behavior patterns on the
market and finding the ways to predict them. This would create a fundamental forecasting basis for the
asymmetry model. Another path for deeper analysis of the topic would be to find the fundamental
reasons behind the different signs in accounting for historical trends on the market to calculate the
asymmetry.
Paragraph 4. Conclusion
The research provides the analysis of the asymmetrical bid-ask spreads published by the
Moscow retail banks, which make the market of cash currency exchange. The bid and ask quotes turn
out not to be equidistant from the major reference rate. In the periods of FX uptrend the dealers ask for
46
higher margin of selling foreign cash to the public and in the periods of downtrend, the margins on
buying foreign currency is higher.
The asymmetry was defined as a function of bid and ask prices and of the open price on the FX
market for each currency pair:
Asymmetry =
(Ask − Open. FX) − (Open. FX − Bid)
∗ 100%
Open. FX
The relative formulation is preferred to the absolute, because the base rate changed significantly over
the course of observations.
The main goal of the research was to find the factors that determine the asymmetry in pricing
the cash exchange services. The objectives were:
(1) Study the existing academic papers about FX dealer market microstructure
(2) Outline a complete set of factors, that may influence the asymmetry in bid-ask spreads
(3) Collect the necessary data: the retail bank quotes, FX historical rates and the statistics about the
consumer trends
(4) Analyze the data and offer a model for calculating the market-average asymmetry
The academic research about market microstructure in the retail FX conversion market is far
from plentiful. There have been papers about the bid-ask spread in general and about the spreads on
different local markets. However, the topic of bid-ask asymmetry was unexplored yet.
The research gap is worsened by the latest events on the Russian retail FX market. During the
last two years, the FX quotes were highly volatile and the retail bid-ask spread sometimes even
enabled arbitrage transactions. This proves the academic relevance of the research.
The factors influencing the difference in bid and ask price formulation were split into demand
(the public) and supply (the banks) sides. The first group was determined to be influenced by Ruble
devaluation expectations. The latter were identified to be subject to the FX market trend changes.
The data required for the research was (a) the retail quotes from the banks, (b) the FX market
data and (c) the statistics about the consumer spending on the retail FX market.
The results of the analysis revealed that all the factors, that were predicted by theory to be
strongly connected to the asymmetry and which were analyzed, are actually important for the general
asymmetry on the retail FX market.
Managerial application of the research is aimed at the retail bank officers, who are responsible
for calculating the bid and ask quotes. The research sheds light on how the asymmetry can be used to
47
make the spread narrow to attract the customers and to still make profit on the transactions. The
regression model presented enables the management to calculate the average asymmetry present on the
market with the given FX market history.
As the research demonstrated, there is a strong association of the net consumer foreign
exchange transactions and the asymmetry in the cash currency bid-ask spreads. However, the
correlation for asymmetries in USD is higher than for EUR. Perhaps, the greater transactions volume
helps make the market more informationally efficient.
Manager of a retail bank can use the research to compare the bank’s rates he publishes with the
fair market rates policy. The banks are advised to impose positive asymmetry, or higher ask premium
that bid discount during the bullish trends on the market, symmetrical quotes during the stagnation and
negative asymmetry, greater bid discounts that ask premiums, in the periods of bearish market trends.1
The bank managers may also take advantage of the fact, that the retail FX market takes into
account only the current trend direction: for the day of retail FX rates publishing, of the previous
overnight change in price and of the change in price on the previous trading day.
The research has also shown that the asymmetry is determined by the power of the trend on the
interbank FX market. The strengthening currency basket provokes higher commissions for purchasing
foreign currency than for selling it. On the contrary, the negative trend on FX market, leads to higher
commissions for selling foreign cash.
The econometrical model revealed the bid and ask asymmetry depends on the FX market
history. The asymmetry calculation is a model with memory. This memory is poor. The average bank
manager cares more of the recent market changes than of the older history.
The future research in this field could further study two groups of factors: the demand and the
supply. The demand analysis would require better understanding the consumer behavior patterns and
ways to predict them. The supply study could reveal the reasons behind the contrarian strategy of the
average asymmetry: why the signs in previous periods coefficients are negative.
All in all, the paper presents the first academic research analyzing the asymmetries on the retail
FX market. The goal was to analyze the data and find the influencing factors. The factors were
inferred theoretically and then were supported empirically. The main factors are the following: (1) the
consumer temporary preferences drive the demand up and shift the asymmetry, making the most
1
The trends are referred to as bullish or bearish in the direct FX quotation for Ruble.
48
popular transaction the most expensive, (2) the FX market changes make the bank managers set up
higher margins in the direction of price changes for both purposes: risk hedging and consumer surplus
extracting.
49
List of References
(1) Copeland, T. E., & Galai, D. (1983). Information Effects on the Bid-Ask Spread. The
Journal of Finance, 38(5), 1457-1469.
(2) Enn Listra, Niina Vaiser and Katrin Rahu (2011). Systematic Differencies of Retail
Exchange Spreads in Some EU Counties: The Banks Against Financial Integration.
Discussions on Estonian Economic Policy No. 2/2011.
(3) EUR/RUB Historical Data (retrieved on 2016, March 22), retrieved from
http://www.investing.com/currencies/eur-rub-historical-data
(4) Foreign Exchange Rates Archive, (retrieved on 2016, March 10), retrieved from
http://data.sberbank.ru/moscow/ru/quotes/archivecurrencies/index.php?all_data115=1
(5) Galati G. (2001). Trading Volumes, Volatility and Spreads in FX Markets: Evidence
from Emerging Market Countries. Bank for International Settlements, 2001. BIS Papers, №2,
pp.197–225. Available at: http://www.bis.org/publ/work93.htm
(6) Garman M. and Klass M. (1980), Estimation of Security Price Volatility from
Historical Data. Journal of Business, 1980, vol. 53, 1, pp. 67-78.
(7) Historical currency exchange rates in Moscow banks (retrieved on 2016, March 22),
retrieved from http://kovalut.ru/history.php?kod=7701
(8) Historical Oil prices Brent and WTI (retrieved on 2016, March 15), retrieved from
http://www.investing.com/commodities/brent-oil-streaming-chart
(9) Hussain S. (2011). The Intraday Behaviour of Bid-Ask Spreads, Trading Volume and
Return Volatility: Evidence from DAX30. International Journal of Economics and Finance,
Vol.3, No.1.
(10) Internal foreign exchange market review by Central Bank of Russia on January 2015
(retrieved on 2016, April 10), retrieved from
http://www.cbr.ru/analytics/bank_system/Exp15_01.pdf
(11) Internal foreign exchange market review by Central Bank of Russia on January 2016
(retrieved on 2016, April 10), retrieved from
http://www.cbr.ru/analytics/bank_system/Exp16_01.pdf
50
(12) Kyle, A. S., & Obizhaeva, A. A. (2004). Market Microstructure Invariance: Empirical
Hypotheses. SSRN Electronic Journal SSRN Journal. Available at SSRN:
http://ssrn.com/abstract=1687965
(13) Mankiw, N. G. (1998). Principles of economics (4th ed.). Fort Worth, TX: Dryden
Press.
(14) Retail FX market review by Vedomosti (retrieved on 2016, April 7), retrieved from
https://www.vedomosti.ru/finance/articles/2016/02/26/631516-pokupat-valyutu-birzhevsegda-vigodno
(15) Russian commercial banks comparison table. (retrieved on 2016, April 18), retrieved
from http://www.banki.ru/banks/?order=fin_rating
(16) Russian holiday making statistics by Rossiyskaya Gazeta (retrieved on 2016, April 12),
retrieved from http://rg.ru/2010/07/20/otpusk.html
(17) Russian Standard Bank Historical Foreign Exchange Rates (retrieved on 2016, March
28), retrieved from www.rsb.ru/courses/
(18) Saida Gtifa, Samir Maktouf and Nadia Labidi (2015). Dealer Behavior and Price
Strategy in the Foreign Exchange Market: Evidence from FX Tunisian Market Dealer.
Journal of Business Studies Quarterly 2015, Volume 7, Number 2.
(19) Sarno L. And Taylor M. (2003). Economics of Exchange Rates. Cambridge University
Press. Cambridge, 2003. doi:10.2277/0521485843.
(20) Savenkov P.G. (2015). The research on bid-ask quotes on Russian cash-currency
market (Master dissertation, Saint Petersburg University, Graduate School of Management,
Master in Corporate Finance Program).
(21) USD/RUB Historical Data (retrieved on 2016, March 22), retrieved from
http://www.investing.com/currencies/usd-rub-historical-data
51
Appendices
Appendix 1.
The list of the banks in the full sample
and the list of banks, subject to further analysis.
All the banks names collected
«АК БАРС» БАНК
Абсолют Банк
Азиатско-Тихоокеанский Банк
Азия-Инвест Банк
АИКБ «Татфондбанк»
АйМаниБанк
АКБ "СТРАТЕГИЯ" (ПАО)
АктивКапитал Банк
Алеф-Банк
АЛЬФА-БАНК
АО «РОСТ БАНК»
АО КБ «АГРОПРОМКРЕДИТ»
Арксбанк
БайкалИнвестБанк
БАЛТИНВЕСТБАНК
Банк "Агророс"
Банк "АЗИМУТ"
Банк "АКРОПОЛЬ"
Банк "Балтика"
Банк "Легион"
Банк "Мегаполис"
Банк "ПЛАТИНА"
Банк "Пойдём!"
Банк "Пурпе"
Банк "РБР"
Банк "Таврический"
БАНК "ЮГРА"
БАНК “МОСКВА-СИТИ”
Банк «Богородский»
Банк «ВБРР»
БАНК «ВЕК»
Банк «Возрождение»
Банк «Восточный»* *(комиссия от 0 до 45 руб)
Банк «ВПБ»
Банк «ГЛОБУС»
Банк «ГЛОБЭКС»
Банк «ДАЛЕНА»
Банк «Камский горизонт»
Банк «Кредит-Москва»
Банк «ЛОГОС»
БАНК «МБФИ»
Банк «Новопокровский»
Банк «Новый век»
Банк «Развитие-Столица»
Банк «РЕСО Кредит»
Банк «РТБК»
Банк «РУБЛЕВ»
Банк «Северный Кредит»
Банк «Советский»* *(взимается комиссия 75
руб.)
Банк «Таатта»
Банк «ТРАСТ»
Банк «ФК Открытие»
Банк «Экспресс-кредит»
Банк АВБ
Банк Город
Банк ЗЕНИТ
Банк Инноваций и Развития
БАНК ИТБ
Банк МКБ
Банк Москвы
Банк Оранжевый
Банк Развития Технологий
Банк Российский Кредит
Банк РСИ
БАНК СГБ
Банк СОЮЗ
БАНК УРАЛСИБ* * (взимается комиссия 30
рублей)
Банк ФИНАМ
Банк Финсервис
Банк Экономический Союз
Банк24.ру
Банкхаус Эрбе
Бенифит-банк
БИНБАНК
БКС Банк* *(взимается комиссия от 0% до
2%)
Владпромбанк
ВЛБАНК
Внешпромбанк
ВНЕШФИНБАНК
ВТБ 24* *(взимается комиссия от 0 до 150
руб)
Выборг-банк
Гагаринский
Газпромбанк
Геленджик-Банк
ГЕНБАНК
Гринфилдбанк
ГУТА-БАНК
Дил-банк
Евразийский банк
Евроазиатский Инвестиционный Банк
ЕВРОКОММЕРЦ
Еврокредит
ЕВРОСИБ БАНК
ЕвроситиБанк
Запсибкомбанк
Заубер Банк
Златкомбанк
ИК Банк
Инвестторгбанк
Интерпрогрессбанк
ИНТЕРПРОМБАНК
ИнтрастБанк
ИпоТек Банк
КБР БАНК
Кредпромбанк
Крокус-Банк
Ланта - Банк
ЛОКО-Банк
МАСТ-Банк
МДМ Банк
Международный акционерный банк
Межрегионбанк
Металлинвестбанк
МЕТРОБАНК
МОРСКОЙ БАНК
Мосводоканалбанк
Московский Индустриальный банк
Московский Нефтехимический банк
Москомприватбанк
Мосстройэкономбанк
МТС-Банк
МФБанк
Народный кредит
НБК-Банк
Невский банк
Нефтепромбанк
НЗБанк
НоваховКапиталБанк
Новый Промышленный Банк
НРБанк
НС Банк
ОРБАНК
ОРГБАНК
ОТП Банк
ОФК Банк
ПАО МОСОБЛБАНК
Первобанк
Плюс Банк
ПРЕОДОЛЕНИЕ
Примсоцбанк
ПРИСКО КАПИТАЛ БАНК,
Пробизнесбанк
Проинвестбанк
ПромТрансБанк
ПЧРБ
Райффайзенбанк
Региональный коммерческий банк
РИАБАНК
РИАЛ-КРЕДИТ
Ринвестбанк
РОСАВТОБАНК
РОСБАНК
Росгосстрах Банк
Росинтербанк
РОСПРОМБАНК
53
Россельхозбанк
Росэнергобанк
РТС-Банк
РУСНАРБАНК
РФИ БАНК
САРОВБИЗНЕСБАНК
Сбербанк России
СКБ-банк
СМП Банк
Собинбанк
Солид Банк
Старый Кремль
ТАТАГРОПРОМБАНК
Темпбанк
Тимер Банк
Транскапиталбанк
Транснациональный банк
Трансстройбанк
Траст Капитал Банк
Трастовый Республиканский Банк
ТЭМБР-БАНК
Тюменьагропромбанк
Унифин
Уральский банк реконструкции и развития
Финанс Бизнес Банк
ФИНИНВЕСТ
ФОРА-БАНК
ФорБанк
Ханты-Мансийский банк Открытие
Центркомбанк
ЦентроКредит
Чувашкредитпромбанк
Экономикс-Банк
Эксперт Банк
Экспобанк
ЭРГОБАНК
ЮНИАСТРУМ БАНК
ЮниКредит Банк
Япы Креди Банк Москва
ЯР-Банк
The banks for further analysis
«АК БАРС» БАНК
Абсолют Банк
Азиатско-Тихоокеанский Банк
Азия-Инвест Банк
Алеф-Банк
АЛЬФА-БАНК
АО КБ «АГРОПРОМКРЕДИТ»
Банк "АКРОПОЛЬ"
Банк "Легион"
Банк "Мегаполис"
БАНК “МОСКВА-СИТИ”
Банк «Богородский»
Банк «ГЛОБУС»
Банк «ГЛОБЭКС»
Банк «ДАЛЕНА»
Банк «Кредит-Москва»
БАНК «МБФИ»
Банк «РЕСО Кредит»
Банк «ФК Открытие»
Банк «Экспресс-кредит»
Банк Оранжевый
Банк РСИ
Банк Экономический Союз
Банкхаус Эрбе
БИНБАНК
Газпромбанк
ГУТА-БАНК
Евроазиатский Инвестиционный Банк
Заубер Банк
Златкомбанк
Инвестторгбанк
ИНТЕРПРОМБАНК
Кредпромбанк
Крокус-Банк
Ланта - Банк
Металлинвестбанк
Мосводоканалбанк
Московский Нефтехимический банк
МФБанк
НБК-Банк
54
НЗБанк
НоваховКапиталБанк
Новый Промышленный Банк
НРБанк
ОРБАНК
ПАО МОСОБЛБАНК
ПРЕОДОЛЕНИЕ
ПРИСКО КАПИТАЛ БАНК,
ПромТрансБанк
ПЧРБ
Райффайзенбанк
Региональный коммерческий банк
РИАЛ-КРЕДИТ
Росгосстрах Банк
Росинтербанк
РОСПРОМБАНК
Россельхозбанк
Росэнергобанк
РУСНАРБАНК
Сбербанк России
СДМ-БАНК
СКБ-банк
Собинбанк
Темпбанк
Тимер Банк
Трансстройбанк
ТЭМБР-БАНК
Уральский банк реконструкции и развития
ФорБанк
Ханты-Мансийский банк Открытие
ЦентроКредит
Эксперт Банк
ЮНИАСТРУМ БАНК
Япы Креди Банк Москва
55
Appendix 2
The histograms of frequency distribution of average over time asymmetries across banks.
USDRUB
1Feb15-20May15
1Jun14-31Oct14
60
70
50
60
50
40
40
30
30
20
20
10
10
0
0
21May15-31Aug15
70
60
50
40
30
20
10
0
16Oct15-20Jan16
50
45
40
35
30
25
20
15
10
5
0
1Nov14-31Jan15
21Jan16-22Mar16
70
40
60
35
50
30
25
40
20
30
15
20
10
10
5
0
0
1Sep15-15Oct15
60
50
40
30
20
10
0
57
EURRUB
1Jun14-31Oct14
60
50
40
30
20
10
0
1Nov14-31Jan15
70
60
50
40
30
20
10
0
1Feb15-20May15
21May15-31Aug15
70
70
60
60
50
50
40
40
30
30
20
20
10
10
0
0
58
16Oct15-20Jan16
1Sep15-15Oct15
80
70
60
50
40
30
20
10
0
60
50
40
30
20
10
0
21Jan16-22Mar16
70
60
50
40
30
20
10
0
59
Отзывы:
Авторизуйтесь, чтобы оставить отзыв