Business School at University of Gothenburg
Industrial and financial management
Fall 2020
Exchange rate determinations
- A multiple linear regression analysis on the correlation between exchange rate,
interest rate and inflation rate
Examinator:
Marta Gonzalez-Aregall
Authors:
William Lönnegren - 970128
Robin Sälgfors - 980201
1
Abstract
The exchange rate market is a financial market with very high volatility and unpredictability.
Even though there does exist some key economic theories on what influences the exchange
rate market, the high volatility makes predictions as hard as for any other financial market.
Inflation, interest rate, current account deficits, public debt, terms of trade, economic
performance and future expectations are seven main factors that affect the exchange rate.
Interest rate and inflation are the two known key variables that affect exchange rate
fluctuations, but the correlation between the two variables and the exchange rate is uncertain.
Thus, these two variables will be further examined in this report. The aim of this report is to
provide further understanding to the strength of a correlation between exchange rate, interest
rates and inflation rates in Sweden, and investigate the effects on the correlation by a global
financial crisis.
In order to investigate the correlation, a multiple linear regression analysis was conducted, by
putting the exchange rate as the dependent variable and putting inflation rate and interest rate
as the independent variables. The data used in the multiple linear regression analysis are all in
percentage points and the sample size contains 120 different values since the time period was
10 years, with one value from each month. Furthermore, a second multiple linear regression
was conducted with annual data, which resulted in a sample size of 10 values. This was done
in order to investigate the significance of a large sample size, and to investigate the
sticky-price theory. There is a strong correlation between exchange rates and inflation
rate/interest rate and the strength is affected by the global financial crisis.
Keywords: Exchange rate, inflation, interest rate, multiple linear regression analysis.
2
Table of content
1. Introduction 5
1.1 Background 6
1.1.1 Exchange rates 6
1.1.2 Financial crisis 2008 8
1.2 Problem discussion 9
1.3 Purpose 10
1.4 Research questions 11
1.5 Hypothesis 11
1.6 Limitations 11
2. Comprehensive literature review 12
2.1 Introduction 12
2.1.1 The Purchasing Power Parity 12
2.1.2 The Balance of Payments Theory 13
2.1.3 The Fisher theory 14
2.1.4 Sticky - price theory 15
2.2 Determinants of Exchange Rates 15
2.2.1 Interest Rate 15
2.2.2 Inflation 16
3. Method 17
3.1 Methodology 17
3.2 Research Strategy 18
3.3 Collection of empirical data 19
3.4 Analysis method 20
4. Empirical findings 23
4.1 Empirical data 23
4.2 Empirical analysis 28
3
4.2.1 Multiple linear regression analysis monthly figures January 2003-December 2012 28
4.2.2 Regression analysis annual figures 2003-2012 32
4.2.3 Regression analysis monthly figures 2003-3007 and 2008-2012 34
5. Analysis 36
5.1 Results multiple linear regression analyses 1 and 2 37
5.1.1 Mathematical point of view 37
5.1.2 Economical point of view 38
5.2 Results multiple linear regression analyses 3 and 4 42
6. Conclusion 43
6.1 Further studies in this subject 45
Reference list: 46
4
1. Introduction
In this chapter the paper's background, problem discussion, purpose, research question and
hypothesis will be presented. This chapter aims to provide a clear overview of the targets and
characteristics of the report. The subject of the report is to investigate the correlation between
exchange rate and interest/inflation rate in Sweden. The basis of the report is to analyze a
variety of theories involving exchange rate determinations, and try to apply the theories on
real historical data. The intention of the report is to find evidence of a correlation that
strengthens the theories, as well as putting the theories into further context. Furthermore, this
report hopes to acknowledge additional understanding for the strength of the correlation, and
the behaviour of the relationship between exchange rate and interest/inflation rate. In order
to investigate the correlation between the variables, data has been analyzed over a 10 year
period of time ranging from 2003-2012.
Further, to examine the variables the researchers have chosen to do a multiple linear
regression analysis between the different factors, putting exchange rates as the dependent
variable and putting both inflation rate and interest rate as independent variables over a time
period of 10 years. Moreover, in 2008 in the middle of the investigated time period between
2003-2012, a global financial crisis broke out. Due to that, this report aims to additionally
analyze the relationship between exchange rate and inflation/interest rate during the crisis to
provide further evidence to how the correlation between the three variables were affected by
a global financial crisis. In order to investigate the effects of the crisis, the time frame was
split into two, and two separate multiple linear regressions were conducted on 2003-2007 and
2008-2012. By doing this, the strength of the correlation pre- and post crisis could be
investigated
If the interest rate increases in a monetary policy, theoretically it will result in an appreciation
of the domestic currency, which in the case of Sweden means the SEK will strengten its value
against foreign currencies. This makes it cheaper to import goods since the same basket of
products can be bought for less SEK. Ultimately, foreign currency will depreciate because of
the supply - demand concept of currency applies on the exchange rate market. This will
ultimately lead to a decreasing inflation pressure in the domestic country. The same concept
works if it is displayed the other way around. Lower interest rates decrease the demand of
domestic currency which will lead to inflation (Riksbanken 2018a). The relationship between
5
exchange rate, inflation rate and interest rate is clear, but the theoretical guidelines do not
alway apply to reality. Therefore, this report aims to provide further evidence to how the
relationship works in reality.
1.1 Background
In this chapter the basics of exchange rate movements will be discussed and the seven main
factors affecting exchange rates will be mentioned (Twin, 2020). There will also be a general
explanation on how exchange rates work, followed by a short overview of the financial crisis
that started in 2008. The financial crisis that started in 2008 has been the crisis of choice since
it is considered to be the biggest one since the great depression (Ohlin, 2018). It is also the
most recent one and the only financial crisis in modern time that has all the data needed for
further investigations of exchange rate, interest rate and inflation available. Further, a
problem discussion will take place followed by the purpose of this report. Lastly, three
research questions and a hypothesis will be formed.
1.1.1 Exchange rates
An easy explanation on what exchange rates are and why we have them, is that they tell us
how much foreign currency you can obtain for a certain amount of the domestic currency.
The exchange rate is determined by supply and demand of the currency of a nation. A high
demand for domestic currency in a certain country will ultimately appreciate the currency
relative to other currencies. This will strengthen the exchange rate for the domestic currency
relative foreign currency. On the other hand, if the currency has an aggregated supply that is
greater than the demand, the currency will depreciate relative to other currencies (Gibson &
Thirlwall, 1992). This paper will focus on the Swedish kronor relative to the Euro.
(SEK/EUR). When SEK appreciates relative to the Euro, the value of Swedish kronor
increases, and less SEK is needed in order to buy one unit Euro, thus, a depreciation of SEK
relative to the euro will result in more required SEK in order to buy one unit of the Euro.
Therefore, throughout this paper the phrase “increasing exchange rate” refers to an increasing
amount of SEK needed to buy one Euro, which indicates a depreciation of SEK. Furthermore,
6
a decreasing exchange rate refers to a decreasing amount of SEK needed to buy one Euro,
which indicates an appreciation of SEK.
There are countries that are very dependent on the exchange rate since it affects the balance
of payments between a country and its partners. There are two different kinds of exchange
rates, first there is “free floating” which is the most common, and fluctuates due to several
different factors that affect exchange rates, these factors will be explained further. Sweden
has a floating exchange rate, therefore, the empirics and analysis of this report is based on
floating exchange rate. The second one is “fixed” and this means that the government in a
country has decided that their exchange rate is going to be tied up with another country's
exchange rate (Riksbanken 2019).
Exchange rates are affected by many different determinants and they are expressed as a
comparison between the value of different currencies. According to Twin (2020) the most
common determining  factors are:
● Inflation
● Interest rate
● Current account deficits
● Public debt
● Terms of trade
● Economic performance
● Expectations
An in-depth explanation to the underlying theories involving exchange rate determination
will be further conducted in the literature review chapter. There is no clear and obvious
answer to what variables affect exchange rate the most. Two major economic theories about
exchange rate determination are the Purchasing power parity and the fisher theory. Sahisahli
and Ho (2002) explains that these two theories are closely correlated, and works to explain
the exchange rate fluctuations based on interest rate and inflation rate. The PPP theory uses
inflation as a way to explain exchange rate fluctuations, while the fisher theory explains that
expected inflation is reflected by nominal interest rates, therefore, interest rates are a key
factor in exchange rate determinations (Sahisahli & Ho, 2002). The correlation of the two
theories creates an interesting starting point for further investigations on how interest rate and
inflation rate correlates with exchange rate. Therefore, this report is going to focus on
7
interest and inflation rates as the two key variables for exchange rate movements, and an
in-depth explanation of these two variables will take place in chapter 2.
1.1.2 Financial crisis 2008
According to Ohlin (2018) the financial crisis that took place in 2008 was initiated in The
UnitedStates because consumers had been taking bigger loans than they could afford, which
led to a collapse of the housing market.. The background check on loan takers had not been
done properly which led to higher risk. To avoid leakage of this information, the banks
combined bad rated loans with a lot of good rated loans, which meant that the package
containing both good and bad loans got a higher rating than it deserved. This had been going
on for a long time and when house prices began to fall, no one knew who was at risk. The
market collapsed after this due to the fact that none of the banks bought the packages
anymore, in the beginning the US Federal Reserve helped out the banks that were close to
bankruptcy (Ohlin, 2018). The whole world was affected by the bankruptcy of major US
banks. In Sweden the stock market plummeted 60% all within a year. Large Swedish
companies that usually exported all over the world could no longer sell their goods.
Unemployment rose markedly and one of Sweden's largest banks was close to bankruptcy. If
we look on how this affected the world Sweden managed to get away quite easily (Ohlin,
2018).
When national banks are at risk of bankruptcy, inflation tends to increase. A decreasing trust
in the banking system lowers the incitament to keep savings within the banking system.
During a time when inflation is rapidly increasing, central banks tend to increase the policy
rate in order to manipulate the market rates in the same direction. Eventually, this will restore
the incitament to save money due to higher returns. Both interest rates and inflation rate
affect the exchange rate. Luckily for Sweden, the inflation rates were restored by
governmental influence of increasing interest rates, and the exchange rate movements were
stabilized (Riksbanken, 2018a). Even though the exchange rate and inflation rate were
stabilized by the governmental interference in the economy, Sweden did experience a rapid
increase of the inflation during a short period of time, which did affect the exchange rate
relative to the Euro. The rapid changes due to the financial crisis provides interesting data
that will be analyzed in this report.
8
1.2 Problem discussion
The exchange rate market is a highly volatile financial market where the value of multiple
corresponding currencies are constantly moving relative to each other. Future predictions of
the exchange rate movements are as hard to predict as for any other financial market. Even
though it is practically impossible to predict exchange rate movements, there are a variety of
key factors that influence the exchange rate. If a reliable correlation were to be found in
movements of the key variables in relation to the exchange rate, more accurate predictions of
future exchange rates may be possible.
Devereux and Engel (2003) uses the purchasing power parity theory to state that inflation and
interest rate are highly influential in the outcome of exchange rate movements. Increasing
inflation will result in increasing price levels, ultimately leading to a depreciation of the
domestic currency in order for the exchange rate to return to PPP. Simultaneously, Increasing
interest rates theoretically should decrease inflation and appreciate the domestic currency due
to it being more attractive for foreign investors. Contradictional, Jonsson and Reslow (2015)
states that interest rate and inflation tends to have a positive correlation in the short term. The
contradictions, as well as the obvious connection that interest rate and inflation rate has with
exchange rate sparks an interesting subject for further research on the subject.
Furthermore, the literature review provides a few key theories but there has not been much
further comprehensive testing on the theories. There have been very few similar studies
conducted where the influence of both interest rate and inflation is measured relative to
exchange rate. Even though there does not exist much previous work on the relationship
between the three variables, there do exist some papers where only one of the two
independent variables has been tested relative to exchange rate in a linear regression model
with only one independent variable. In the paper “Relationship between exchange rates and
interest rates: Case of albania '' posted in the medeteranian journal of social sciences, Tafa
(2015) examined the relationship between exchange rate and interest rate during a 12 year
period in albania. Tafa (2015) used the domestic interest rates in Albania as her independent
variable, and conducted two separate regression models with exchange rate as the dependent
variable. One where the Ablanian domestic currency was used against the Euro, and one
where it was used against USD. The results of the fist model where Euro is used is of interest
for this report since this report uses SEK relative to Euro. Tafa (2015) states that her
9
regression model is statistically insignificant, and that it shows a small correlation between
the variables. The results of her study are relevant to this study, and can be used as a way to
compare the results and therefore, this study hopes to bring further understanding to the
thesis.
1.3 Purpose
The following report aims to investigate the correlation between the exchange rate and
interest rate as well as the correlation between the exchange rate and inflation in Sweden.
Furthermore, to what extent movements in interest rate and inflation rate influence the
exchange rate.The results will hopefully also provide answers to how the exchange rate is
affected by a financial crisis.. Hence, the purpose of the report is to give further
understanding and support to theories that implies dependence between foregoing variables.
In addition, our results will provide significant information to exchange rate operators and
currency traders who benefit from acknowledging following data.Our findings may be
beneficial for commercial banks who will be able to make more accurate forecasts of the
exchange rate fluctuations, and offer loans with a well calculated interest rate and
information. It may also improve central and commercial banks in their analysis on how to
handle a future financial crisis. Ultimately, our prospect is to provide relevant knowledge for
further studies made in the same field.
1.4 Research questions
● Has there been a correlation between the exchange rate movements and the interest
rate/inflation movements in Sweden during a 10 year period, ranging from 2003 until
2012. If so, how strong is the correlation?
● In what directions does interest rate and inflation rate influence exchange rate?
● Will the exchange rate correlation with interest and inflation rate be strengthened or
weakened during a financial crisis?
10
1.5 Hypothesis
An increase in interest rates of domestic currency will cause the domestic currency to
appreciate against other foreign currencies, while an increased inflation will depreciate the
domestic currency. Further, a global financial crisis will decrease the exchange rate
correlation with interest rate and inflation rate due to uncertainty.
1.6 Limitations
This study is solely focusing on the Swedish exchange rate relative to the Euro, which should
be considered as a limitation to the study. The results will provide a clear overview of how
the swedish exchange rate is affected by interest rate and inflation, but the single focus on
one specific country makes the results insignificant for the universal idea of how the variables
affect exchange rate. However, the results can be used as a guideline for future studies and
research for other countries, which might ultimately provide enough evidence for a universal
explanation of how exchange rate is affected by inflation and interest rate. Furthermore, the
remaining five factors previously mentioned have not been taken into count. Also the time
frame that we choose to examine is a limitation. In order to get an even better result on how
exchange rates are affected by changes in inflation and interest rate during a global financial
crisis the period examined and the factors taken into account should be increased. Even
though the results of the study shows a clear correlation between the variables, the limitations
of the time period makes it impossible to draw a definitive conclusion that the correlation
exists at all times. The results of this study merely provides strong evidence of an existing
correlation at all times, since the correlation exists during the 10 years examined.
11
2. Comprehensive literature review
2.1 Introduction
Following sections of the report contains a literature review on the theories applied in the
report. Various factors that affect exchange rates will be addressed and explained through
theories which we consider applies to the subject matter.
2.1.1 The Purchasing Power Parity
The theory explaining exchange rate determination is called purchasing power parity (PPP)
and is used to compare the prices on services and goods between countries. The theory is
based on the fact that the actions made by both importers and exporters will affect the spot
exchange rate due to price differences between different countries. PPP also means that
transactions affect a country’s current account by influencing the exchange rate on the foreign
exchange market. The interest rate parity theory is in contrast with this due to it assuming
that the investors actions will result in changes in the exchange rate. The PPP theory is based
on a variation and extension of the “law of one price” that is applied on the aggregated
economy today. The theory states that the exchange rate between two currencies should
reflect the common price of a fixed basket of goods. In reality this means that an increase of
the inflation rate on one certain country must lead to a depreciation of the currency of that
country to maintain PPP (Devereux & Engel, 2003).
The purchasing power parity is partly a tool that makes it possible to improve the accuracy
concerning data analyzes between countries, but it is also a theory about exchange rate
determination. Unfortunately, this theory does not apply as well in reality because of the fact
that it is based on several assumptions which of those rarely reflects the present. However,
this theory is still of importance in terms of providing a background for its use as a tool to
compare wages and income in a country. International organizations, for example the World
Bank, are using this tool to present a lot of their international data.
12
2.1.2 The Balance of Payments Theory
The balance of payments theory is also called the demand and supply theory of exchange rate
and it is the most modern exchange rate determination. The theory states that the foreign
exchange rate market determines the exchange rate, it does so by using supply and demand
on the foreign exchange market. When the demand on a country’s currency drops to a certain
exchange rate, it will result in a deficit in the balance of payments and a deficit in the balance
of payments will lead to a fall in the external value of a country’s currency. If the demand on
a country’s currency instead is appreciated to a certain exchange rate, it will lead to a surplus
in the balance of payments that will lead to an increase in the external value of the country’s
currency (Stern, 2017).
The theory says that a deficit in the balance of payments will result in the exchange rate
depreciating and a surplus in the balance will instead make the foreign exchange reserves
stronger and this will lead the domestic currency to appreciation. If a country has a deficit in
their balance of payments it means that the demand on foreign currency is larger than the
supply.
2.1.3 The Fisher theory
According to Lagerwall (2008) nominal interest rate is the rate before taking inflation into
account. When referred to as the interest rate on a loan we usually in this case address the
nominal interest rate. The rate after taking inflation into account entitles what is actually
called the real interest rate. In the previous example, the real interest rate of a loan is
therefore the actual cost of the loan even though it is rarely acknowledged.
The fisher theory is named after the American economist Irving Fisher. In the theory nominal
and real interest rates are related and the Fisher equation follows:
Source: Lagerwall (2008).
13
In the equation above i
t
equals the nominal interest rate and t equals time, r
t
is the real interest
rate and again the t stands for time. Lastly E
t
𝝿
t+1
refers to the expected inflation rate between t
and t+1. This equation tells us that by adding the real interest rate and expected inflation rate
we can submit the nominal interest rate. In practice we find a hindrance in calculating the real
interest rate due to the difficulty to find suitable measures to the expected inflation for the
period of time (Lagerwall, 2008).
The Fisher effect is one part of the theory, the hypothesis tells us that the nominal interest rate
follows the expected inflation rate which leads to that the real interest rate will be kept
constant. In this case, if the nominal interest rate increases it would result in the expected
inflation to increase with the same percentage point.
2.1.4 Sticky - price theory
The sticky prices theory refers to a slow and steady adjustment in prices. This has been a
bigger issue in the past when companies set their price range beforehand, for example
catalogues that were made once or twice a year. Because of the fact that it was expensive
printing new catalogues, the expense of reprinting would surpass the proceeds of
augmentation of the prices. This is called a menu cost. In addition to this there is something
called a contractual cost. As an example, let’s say a company has set a price for a service and
signed a contract that is valid for a longer period of time. This means that the company can
not change the price until the assignment is completed. This is the explanation behind the
concept of sticky prices. They move against the equilibrium slowly and never actually hits it,
since the equilibrium changes faster due to aggregate demand and supply (Ball & Mankiw,
1994).
2.2 Determinants of Exchange Rates
There are a number of factors that affect the exchange rate between countries, but ultimately
it is demand and supply of a certain currency compared to other currencies that determines
the exchange rate. In chapter one of this study an overview of the determination factors of
exchange rate were listed.
14
2.2.1 Interest Rate
There is a strong correlation between inflation and interest rates. In general, a higher inflation
will result in higher interest rates, which is why high interest rates will affect movements in
exchange rates. The central bank uses the interest rate as a tool to manipulate big fluctuations
in the exchange rate. The central bank can also try to keep the exchange rate at a certain level
by manipulating the interest rates. A higher interest rate will result in more foreign investors
investing their money in the concerned country, due to higher returns (Brigo & Mercurio,
2007).
According to the Mundell-Fleming model the interest rate has to be increased in order to
stabilize the depreciation in exchange rates to dampen the inflation pressure and by doing this
being able to avoid big economic consequences (Fregert & Jonung, 2014). There are three
reasons for why a high interest rate policy is considered. The first thing is that it provides the
market with information about the authorities not allowing the exchange rate to fluctuate
sharply given a certain economic situation and by doing this, reducing the expected
fluctuation. The second thing is that it will attract more investors for financial domestic assets
which will limit the exchange rate depreciation. The third one is that by lowering the import
level it will improve the position in the balance of payment and it will also reduce the
domestic aggregate demand (Devereux & Engel, 2003).
2.2.2 Inflation
One of the biggest factors that affect the exchange rate is inflation. In theory a low inflation
rate will result in the exchange rate increasing, because the purchasing power of the currency
increases compared to other currencies (Duarte & Stockman, 2002). The definition of
inflation is the depreciating value of domestic currency. Hence, the amount of money needed
to buy one unit of a certain good increases over a period of time. There are a lot of
determination factors that lead to a depreciation in domestic currency. The main reason for a
depreciation is an increasing amount of money. If the money supply in a certain country
exceeds the demand, the value of the money will automatically decrease. The money supply
of a country is determined by a central bank. Therefore, the central bank has the ability to
15
manipulate inflation through money supply. Another way of manipulating the inflation rate of
a country is to manipulate market rates. Once again, this is done by the central bank, by
manipulating the policy rate. As mentioned in the previous section, inflation and interest rate
therefore are highly correlated. An implementation of a higher policy rate will be reflected by
increasing market rate. Thus, higher market rates increase the incitament to save money
rather than spending it. Further, higher market rates attract foreign investors due to a higher
return in the concerned country.
Inflation is commonly used to measure price stability in an economy. Inflation can be divided
into two aspects; we have the demand side also called demand inflation and we have the
supply side also called cost push inflation. In countries with open economies, inflation comes
due to the domestic and foreign factors. The external factors could, for example, be a price
increase on materials in the world. The exchange rate system is important, if used correctly it
can minimize the risk of fluctuations in the exchange rate (Totonchi, 2011).
In a system with floating exchange rate, the fluctuations might have a strong impact on the
price level through aggregated demand (AD) and the aggregate supply (AS). On the
aggregated supply, the effect of the domestic currency depreciates could affect the price level
by domestic consumers buying imported goods. But this will only happen if the country is the
recipient country of the international prices. The indirect effect of depreciation of currencies
(devaluation) on the price level in a country can be seen by the prices of capital goods that the
manufacturers state as input goods. A weak exchange rate will lead to the price of inputs
becoming more expensive, leading to higher production costs (Engle, 2002).
16
3. Method
3.1 Methodology
This section of the paper aims to give a further understanding of what methods were used to
gather empirical data about past financial crises, and why these methods were chosen.
Furthermore, it discusses what methods were used to research the effect and outcomes of the
fluctuating exchange rates during these crises, and how the choice of method is motivated by
our research question.
3.2 Research Strategy
Our choice of methodology and research strategies are based on the literature “An
introduction to business research methods” by Greener and Martelli (2018). There are two
methodologies to use when gathering data, quantitative and qualitative methodology. The two
methods demonstrate two different ways to gather and analyse data. The quantitative
approach relies on already existing data and is often associated with a deductive approach,
while a qualitative method reflects an inductive approach to data (Greener & Martelli, 2008).
This paper is analysing the correlation between exchange rate and interest/inflation rate over
a 10 year period. Further, the variables are split up into two different time periods in order to
examine the effects on the correlation during a financial crisis. The paper is based on existing
data and written theories explaining the three given variables. Therefore, we are applying a
quantitative method of research. We use existing theories and methods to analyse our
empirical data, to find correlations that support our research question which implies a
deductive approach.
A deductive approach is based on existing data and theories. The idea of a deductive
approach is to start off with an already existing theory and form a hypothesis based on it. The
hypothesis is then tested by using quantitative data to match the initial theory. By doing this,
a conclusion can be drawn from the empirical data to accept or dismiss the hypothesis. A
deductive reasoning moves from a general point of view with a known theory to a specific
point of view, where the known theory is used to examine the hypothesis and the data. Thus
A deductive approach moves from theory to data (Greener & Martelli, 2008).
17
An inductive approach is a rearrangement of the order compared to a deductive approach.
The inductive approach uses empirical data as its initial standpoint rather than theories.
Through investigation, the data is being analysed in order to form general theories. Inductive
research moves from a specific focus to a general focus (Greener & Martelli, 2008). The
hypothesis is based on empirical data rather than existing theories which results in a more
observational result. Therefore, rather than accepting or dismissing a theory, the inductive
approach aims to develop a new theory of frame.
3.3 Collection of empirical data
The empirical data used in this paper consists of secondary data gathered from articles and
scientific reports, as well as statistical data from various institutions. The vast majority of our
data is gathered from online sources. The data is analysed based on existing theories found in
literature. The aim of the report is to analyze the effects interest rate and inflation has on the
exchange rate in sweden. Since the report is solely focused on sweden, the empirical data
used in the regression analysis is mostly collected from reliable swedish institutions. The vast
majority of data used in this report were gathered from SCB (swedish bureau of statistics)
which is a Swedish state administrative authority which falls under the ministry of finance.
SCB is a gathering place for official statistical reports, and other public statistical data about
Sweden (SCB, n.d.). Since SCB is a swedish state administrative authority we believe their
data is reliable and trustworthy. Further, in the beginning of the thesis a literature scanning
was conducted in order to find a variation of sources with different approaches. During the
literature scanning the reliability of the sources were taken into account. The importance of
creating a wide view of the subject in addition to finding reliable sources were key in order to
form a literature review. The theories used in the paper were chosen based on the problem
discussion and research questions.
The majority of sources used are gathered from scientific articles and conventional economic
literature. While conducting the literature review, the importance of using published articles,
and well-cited articles were taken into account to provide reliability to the thesis.
Furthermore, articles and data published by state-governed sources such as the central bank
of Sweden and the Swedish bureau of statistics were preferred, to provide additional
credibility to the paper. The results of the paper are based on data collected from the Swedish
18
bureau of statistics (SCB) and OFX. As mentioned, SCB is a swedish state administrative
authority. Thus, the credibility is solid. OFX were used to gather the SEK/EUR exchange rate
data. There are various sources that provide documented exchange rate data, and the data
seem to differ depending on the source. The decision to use OFX as a source in this paper
was based on its size and credibility. OFX is a public company traded on the Australian
securities exchange, as well as being one of the biggest money transfer companies in the
world (Black, 2013). Lastly,as mentioned in section 1.2 there exists very few similar studies
on the subject, which restricts the ability to compare sources and results to similar work.
Thus, the credibility of the results are affected since they cant be compared.
3.4 Analysis method
A multiple linear regression analysis has been used to investigate the relationship between
exchange rate and inflation/interest rate in Sweden during a 10 year period. The aim of the
analysis is to establish a correlation between two independent variables and one dependent
variable. The choice of method was based on previous studies where exchange rate has been
investigated. As mentioned in section 1.2 there exists very few studies where both interest
rate and inflation rate has been used to investigate exchange rate determination, but similar
studies have been conducted where one of the two variables has been investigated. Tafa
(2015) uses a linear regression model to establish a correlation between interest rate and
exchange rate in Albania, which proved to work. The study conducted by Tafa (2015), along
with useful information about the fundamentals of how regression analyses works, and what
results to expect provided by Chatterjee and Hadi (2012) motivated the method of choice for this
paper.
As mentioned in section 1.1.1 the exchange rate of a currency is constantly in fluctuation
relative to other currencies. The changing currency value is a result of a lot of different
factors. Two key factors whose influence plays a great role in the exchange rate movements
are interest rate and inflation (Riksbanken, 2018a). Therefore, the correlation between interest
rate/inflation and exchange rate will be investigated with inflation and interest rate as
independent variables and exchange rate as a dependent variable. The multiple linear
regression analysis was conducted in Excel using the data analysis tool-pack. The aim of a
multiple linear regression analysis is to use multiple independent variables to predict the
19
outcome of a dependent variable, in order to establish the relationship between the variables
(Chatterjee & Hadi, 2012). By implementing a multiple linear regression analysis on a
number of variables, the strength of the correlation will be examined through a linear
relationship.
Monthly exchange rate figures over a 10 year period starting in January 2003 were used as
the dependent variable. The same time period was used for interest rate and inflation as two
independent variables. The numbers were listed in excel where the regression model was
conducted. A regression model will provide a coefficient of multiple determination value (
. is a percentage value that determines to what extent the dependent variable variations𝑅
2
) 𝑅
2
can be explained by the variating independent variables. This value will provide a measure to
how well correlated the variables are. Furthermore the regression model generates a P-value.
The P-value is an indicator to how significant the analysis is, and a way to determine if the
null hypothesis can be rejected. A significant P-value is crucial in order to determine the
findings of the analysis as trustworthy and reliable. Additionally, the regression model will
provide coefficients for each variable. These coefficients dictate to what extent an increase or
decrease of the independent variables will affect the constant (Chatterjee & Hadi 2012).
The equation used in the regression analysis model is:
Y= β
0
+  β
1
𝑋
1
+ β
2
𝑋
2
+... + β
𝑖
𝑋
𝑖
Y = Dependent variable (Exchange rate SEK/EUR)
= Intercept (dependent variable)β
0
= slopes (coefficients for each independent variable)β
1
, β
2
= Independent variables (Inflation rate, interest rate)𝑋
1
, 𝑋
2
Further, a second multiple linear regression analysis was made with annual data instead of
monthly data. This was done to give further insight to the relationship between the variables,
and provide an overall view. It is a way to find weaknesses in the method of using monthly
figures, such as the sticky price theory. Lastly, two separate multiple linear regression
analyses were made where the timeframe was split into January 2003-December 2007 and
January 2008-December 2012. This was done to examine the correlation of the variables
20
before- and after the crisis of 2008. By investigating the two time frames separately
conclusions can be drawn whether the correlation is strengthened or weakened during an
unstable economy.
Ultimately, a Pearson correlation test was conducted to analyze the correlation between each
variable individually. A Pearson correlation test establishes the strength and direction of a
correlation between two variables, but does not take into account how the variables affect
each other. A regression analysis will provide a more in depth understanding of how the
relationship between variables work and how they influence each other (SOS, 2015). Even
though a multiple linear regression analysis provides more evidence on the relationship
between the variables we believe that a Pearson correlation test is a relevant supplement to
our analysis because it provides a clear visible representation on the correlation between the
numbers.
As mentioned, the strength of a multiple linear regression analysis is the ability to determine
how one or more variables influence a dependent variable. By determining how well
correlated two variables are and how they influence each other, future events and fluctuations
will be easier to understand and therefore easier to cope with. In the case of our analysis, an
understanding of how interest rate and inflation influence exchange rate might provide
additional insight on how to balance future exchange rates by influencing interest rates and
inflation.
Even though a regression analysis is a great tool to establish correlations, there are still
important weaknesses to note. Flom (2018) discusses the weakness of a regression analysis in
his article “The disadvantages of linear regression”. Flom (2018) discusses the importance of
being cautious of “outliers”. The author defines “outliers” as surprising data that does not
represent the entirety of the population. Outliers may have a great impact on the outcome of
the model, and thus, the result should be interpreted with caution (Flom, 2018). Outliers is a
relevant weakness in the model conducted in this report. As Mentioned in section 1.1.1
exchange rate is influenced by various different variables, and the regression analysis in this
report only focuses on two of those variables. Therefore, the influence the two independent
variables have on exchange rate can not be 100% proven. Another limitation of the regression
model in this report is the timeframe. The sticky price theory states that variables might be
resistant to change. Changing variables does not always influence each other right away (Ball
21
& Mankiw, 1994). To tackle the possibility of this theory, as mentioned a separate regression
analysis was made where annual data was used instead of monthly data. The results of the
analysis on annual data provides further information about the relationship between the
variables. It is a way to provide a second result to match against the monthly data in order to
find additional strengths and weaknesses to the model.
4. Empirical findings
This section will present and explain how the data was collected and used to perform the
regression analysis. The first part of this chapter will clarify how the data was converted to fit
the model that was conducted in order to answer the research questions asked in chapter one.
Furthermore, the second part of this chapter will present the findings and results of the
regression model.
4.1 Empirical data
This report focuses solely on the exchange rate of the Swedish SEK relative to the Euro.
Since the aim of this report is to investigate the Swedish currency rate, the independent
variables in the regression model (interest rate/inflation rate) are data for Sweden alone.
Thus, the importance of accumulating data from reliable swedish sources has been
prioritized. The book “Regression analysis by example” by Chatterjee and Hadi (2012) was
used as the main source of information for the regression analysis in this report. In addition,
the article “How to perform regression analysis using Excel” by Frost (2021) was used to
prepend additional knowledge on how to perform a regression analysis in excel. The article
“how to perform regression analysis using excel” is written by Frost (2021) on a private
unauthorized website. Thus, the reliability of the article must be handled with precaution. In
order to fortify the reliability of the source, and insure the work in this report to be correct
additional sources were used to verify the information. Additional sources used were the
article “linear regression analysis in excel” by Cheusheva (2020) and “Multiple regression
analysis using excel” by Zaionts (2020).
22
As mentioned in chapter 3 the multiple regression model in this report is based on one
dependent variable and two independent variables. The three variables used are as follows:
● Exchange rate
● Inflation rate
● Interest rate
Historical Swedish exchange rates relative to the Euro has been used as the dependent
variable in the analysis. The exchange rate numbers were gathered from OFX . Historical
Exchange rate data stretching back to 2003 has been proven to be somewhat hard to
congregate due to the fact that most services only stretch ten years back in time. OFX along
with a few other currency data services provides historical data stretching back to 1999. OFX
was chosen as the main source of Exchange rate data due to it being a credible source of
information. OFX is a public company traded on the Australian securities exchange (Black,
2013), as well as being one of the biggest money transfer companies in the world, transfering
to 55 countries (OFX n.d.).
The first independent variable in the model used in this report is inflation rate. As mentioned
in chapter 2 inflation is used to measure price stability in an economy. A higher price for the
same amount of consumption indicates a higher inflation rate, which leads to a depreciation
of the currency. Inflation in Sweden is regulated by the central bank of Sweden by
influencing the policy rate. The changing policy rate will ultimately affect market rates, thus,
the incentive to borrow or save money increases depending on which direction market rates
are affected (Riksbanken, 2018a). The most common way to measure inflation is to use an
index to calculate the average price development for private consumption in an economy. The
central bank of Sweden is using the consumer price index (CPI) as their most viable index
(Riksbanken, 2018a).
In theory, interest rate and inflation rate should have an inverse relationship, but in reality this
is not always the case. This can depend on various reasons. Jonsson and Reslow (2015)
illustrated in their article “interest and inflation rates through the lens of the theory of Irving
Fisher'' that interest rate and inflation rate in the short term are positively correlated, which
contradicts the theory that inflation and interest rates have an inverse relationship (Jonsson
23
and Reslow, 2015). The authors explain this outcome using the fisher theory. An increased
policy rate signals that the inflation is on its way up, thus, the expected inflation increases
which promotes economic growth. The Authors also use the sticky price theory to explain the
positive correlation. Governed economic changes tend to take some time to be reflected in the
overall economic system, therefore, the inverse relationship might not show in the short term.
It is worth noting that Jonsson and Reslow (2015) use CPI as their measure of inflation, as
well as the central bank of Sweden, but CPI is not the only measure used to measure
inflation.
In the article “Hur mäts inflationen?” the central bank of Sweden employee Johansson
(2015) introduces consumer price index with a fixed interest rate (CPIF) as a measurement of
inflation. CPIF is used the same way as CPI but with a fixed interest rate, thus, the index is
not affected by the household mortgage rates. Johansson (2015) opines that interest and
inflation rates often are illustrated in a misleading way where they show a positive correlation
in the short term, just as Jonsson and Reslow (2015) explains. Johansson (2015) further
explains that the swedish central bank complements their inflation reports with CPIF, in order
to provide a more accurate view of the reality of the relationship. Hence, in order to illustrate
the correlation in the regression analysis, this paper uses CPIF as a measure of inflation rather
than CPI. Historical data of monthly CPIF was collected from SCB. There are a number of
ways to illustrate CPI and CPIF, and a number of ways to convert the figures. CPIF (as well
as CPI). The most common way to interpret CPIF is by using a fixed year with a fixed value
in history as a startpoint to measure the forward increase or decrease from. SCB is using the
year 1989 as a starting year, where the fixed value is set to 100. This value indicates that in
1980 the total value for consumption of a certain unit of goods equals 100 (SCB, 2020a).
By valuing the consumption with an index, the monthly or annual increase is easy to
calculate. This way, the monthly increase or decrease of inflation will have a fixed value
instead of a procentual value. The other way of calculating the increase and decrease of
inflation is by calculating the procentual change for a certain month relative to the same
month a year before (SCB, 2020a). By doing this the procentual inflation change from month
to month is illustrated by a procentual value, which is the most common way to illustrate
inflation. By doing this, the inflation rate is illustrated instead of the CPIF value.
24
In this model the inflation rate will be used, with the procentual change as a value for the
independent variable. The reason to use the procentual inflation rate instead of a fixed CPIF
value is because the second independent variable, interest rate, is also based on a procentual
value. Further, the procentual inflation values are usually ranging from 1-3%, backed up by
the fact that Sweden has an inflation target of 2% (Riksbanken, 2018b). The interest rates are
fluctuating around 0-5%, therefore, the inflation rate as a procentual value of 1-3% will work
better with the regression models, rather than having CPIF as the independent variable with a
3-figure value. Ultimately, when inflation rate is discussed further in this report, it is referring
to the procentual change of the CPIF. The diagram below shows an illustration of the monthly
CPIF in Sweden starting January 2003 until december 2012 (SCB 2020b).
Diagram 1: monthly CPIF in sweden January 2003 - December 2012:
Source: Own elaboration with data from SCB (2020b).
The second independent variable is interest rate. As mentioned, the fluctuating market rates
are controlled by the central bank of Sweden by influencing the policy rate (Riksbanken,
2018a). This report uses the historical short-term rates in Sweden, the yield on a three month
treasury bill as a measure for interest rate. The reason why the short-term rate was used
instead of the policy rate is because the policy rate in Sweden tends to have a steady value for
a longer amount of time due to it being decided by the central bank directly. This would give
a false representation of the correlation with an exchange rate that is fluctuating constantly.
Additionally, the short-rate was used rather than the long-term rate because the long-term rate
is influenced by the expected inflation, and since inflation is the first independent variable in
the analysis, the results would be misleading if the values of the independent variables would
25
affect each other directly. Monthly short-term rates were gathered from SCB. The diagram
below shows the interest rate fluctuations in Sweden from January 2003 until December 2012
(SCB 2021).
Diagram 1: Monthly Interest rate in Sweden January 2003-December 2012:
Source: Own elaboration with data from SCB (2021).
4.2 Empirical analysis
This section will present the results generated by the regression analysis. Our Empirical
analysis will be presented in 3 parts. Part 1 will present our findings when monthly data
between January 2003 – December 2012 were used. Part 2 will present our findings after
using annual data where each year represents the mean for all our monthly values during that
year. The third part will present our results where the timeframe is split into two sections,
2003-2007 and 2008-2012. This was made in order to investigate the relationship between
our variables before- and after the financial crisis in 2008. This way, the results from both
timeframes can be compared, and the effects of the crisis in 2008 can be examined. The
results will be explained and put into context. An in-depth analysis and discussion of the
results will continue in section 5.
26
4.2.1 Multiple linear regression analysis monthly figures January 2003-December 2012
As mentioned in section 3 a multiple linear regression analysis was conducted in order to
investigate how well correlated Exchange rate fluctuations are with changes in interest rate
and inflation. Therefore, the model used Exchange rate as the dependent variable, while
interest rate and inflation were used as independent variables. The model uses SEK/EUR
exchange rate collected from OFX. The inflation rate and interest rate are collected from the
Swedish central bureau of statistics.
The model was conducted using CPIF (Consumer price index with fixed rate) as a measure
for the monthly average inflation rate percentage. The monthly average interest rate
percentage is a measure of the average short rate in Sweden. The model was run in Excel
using the regression analysis data tool, and the table below presents the structure of the model
in Excel.
Table 1: Multiple linear regression analysis monthly data:
Source: Own elaboration, data from SCB.
The results from the model are presented in table 2. The model generated a coefficient of
multiple determination ( ) of 0,277 = 28%. This indicates that 28% of the monthly change𝑅
2
in Exchange rate (dependent variable) can be explained by changes in the independent
variables. Furthermore, The ANOVA table shows a P-value of 5,3115E-09. Since the p-value
is smaller than 0.05, a conclusion can be drawn that the model is significant. The individual
p-values for each independent variable is also < 0.05, this provides enough evidence to
determine that the overall model is of significance. (Chatterjee and Hadi, 2012).
Table 2: Multiple linear regression analysis statistics, ANOVA table monthly data:
27
Source: Multiple linear Regression analysis Excel, data from SCB.
Table 3 shows further results from the multiple linear regression analysis. The coefficients for
each variable are shown below. The constant (Exchange rate) has a coefficient of 9,22, The
inflation rate has a coefficient of 0,36 and the interest rate has a coefficient of -0,22. The
coefficients indicate how much an increase/decrease of each of the independent variables
affects the constant. In this model inflation rate and interest rate are measured in percentage,
which indicates that when the inflation increases with one percentage point, the constant
value (exchange rate) will increase with 0,36 units. The coefficient for the dependent variable
is 9,22, which indicates that when the two independent variables are equal to zero, the
dependent variable, the exchange rate, will be equal to 9,22 SEK/EUR (Cheusheva, 2020).
Table 3: Multiple linear regression analysis statistics (coefficients) monthly data:
Source: Multiple linear regression analysis monthly data Excel data from SCB.
Table 2 and 3 provides the data necessary to calculate the regression analysis model using the
regression analysis equation (Chatterjee and Hadi, 2012):
Y = β
0
+  β
1
𝑋
1
+ β
2
𝑋
2
+... + β
𝑖
𝑋
𝑖
Y = Dependent variable (Exchange rate SEK/EUR)
= Intercept (dependent variable)β
0
= slopes (coefficients for each independent variable)β
1
, β
2
= Independent variables (Inflation rate, interest rate)𝑋
1
, 𝑋
2
Thus, the regression analysis equation for this model equals:
28
Exchange rate SEK/EUR = 9,2211 + (0,3649 Inflation rate) + (-0,2234 Interest rate).
After finishing the regression analysis, the Pearson correlation coefficient test was conducted
on our data in order to provide additional evidence for our hypothesis. As mentioned in
chapter 3 a Pearson correlation coefficient test does not provide any estimates for slope and
intercept, which are used to analyse predictions on the outcome variable (Y) based on the
predictor variables (X) in a linear regression (SOS). The Pearson test only generates a value
between -1 – 1 that indicates the direction and the strength of two variables without splitting
the variables into dependent/independent. Therefore, the results of the Pearson test does not
provide any evidence to how much the variables affect each other, it will only show the
correlation between x, y in terms of direction. The Pearson test was done merely to
strengthen the results from the linear regression, and to further clarify the level of association
between exchange rate and interest rate/inflation. The table below shows the results of the
Pearson correlation coefficient test.
Table 4: Pearson correlation coefficient monthly data:
Source: Pearson correlation test monthly data Excel, data from SCB.
As seen above the Pearson test indicates a correlation of ~ 0,26 between Exchange rate and
inflation and a correlation of ~ -0,32 between exchange rate and interest rate. The positive
correlation between inflation rate and exchange rate indicates that an increase in inflation will
most likely lead to an increase in Exchange rate, thus the cost for one euro will increase in
SEK, which means that SEK will deprecate in value. The negative correlation between
interest rate and exchange rate indicates that a decrease in the domestic interest rate will
decrease the SEK/EUR exchange rate, meaning that the SEK will be worth more in relation
to the Euro, thus the SEK will undergo an appreciation. The results of the multiple linear
regression and the Pearson test indicates that the hypothesis formed in chapter 2 is correct.
When inflation is increasing, the cost of goods will increase. When the cost of goods is
increasing, more money is required in order to buy the same goods. Therefore, the currency
value will decrease. This leads to an increased SEK in relation to EUR, which indicates a
depreciation in the domestic currency. As for the negative correlation with interest rate, when
29
interest rates are increasing, foreign investors will look to invest in Sweden due to a higher
rate of return than other countries. Increased foreign investments will increase the demand of
Swedish currency which will lead to an appreciation. When SEK is appreciating in relation to
EUR, less SEK is needed to buy one euro and the SEK/EUR exchange rate will decrease.
Hence the negative correlation. The results will be further analysed in chapter 5, followed by
a discussion.
4.2.2 Regression analysis annual figures 2003-2012
Table 5 shows how a regression analysis on annual figures was conducted. Annual figures
were analysed subsequent to the monthly regression analysis to investigate whether or not the
correlation between the variables would strengthen when annuals figures are examined. The
sticky price theory states that a theoretical lag exists in the economy, which means that
central and governmental influences on the economy do not inflict the numbers right away, in
theory governmental interference usually takes some time to notice in the numbers, which in
this case is the exchange rate (Ball & Mankiw, 1994). To further investigate this, a regression
model was conducted when the monthly figures from the previous model were added up for
each individual year, and divided by 12 to get a mean value for each year. This was done for
each of the variables and can be seen in the table below.
Table 5: Multiple linear regression analysis structure annual data:
Source: Own elaboration, data from SCB.
The regression analysis results with annual figures are presented in table 6. The model
generated a coefficient of multiple determination ( ) of 0,386 = ~39% compared to our𝑅
2
monthly observations of = 28%. The results indicate that 39% of the annual change in
exchange rate can be explained by changes on the two independent variables. The increased
30
-value can not be explicitly derived to only one cause, such as the theory explaining a lag𝑅
2
in the economy in the last paragraph. Other factors that might have participated in the
increase are for example the decreased amount of observations in this regression model
compared to the previous one.
Furthermore, the p-value for the model is 0,18, significantly higher than our monthly
regression. The significance is > 0,05 which means that the evidence of a changing exchange
rate due to changing independent variables are not strong enough to rely on. Again, the high
significance value can not be derived from one certain cause. The high p-value might be
caused by a small sample size, or because the effect size is too small (Deziel, 2018).
Table 6: Multiple linear regression analysis statistics, ANOVA table annual data:
Source: Multiple linear regression analysis annual data excel, data from SCB.
Table 7 below shows the coefficients for each variable. One thing to notice is that the
coefficients for each individual variable have increased compared to the monthly regression,
just like the value. Inflation increased to 0,53, an increase with 0,17. Interest rate only𝑅
2
increased by 0,01. Just like the p-value for our model, the individual p-values for each
variable are not significant. They both have a p-value > 0,05 which proves that we can't rely
on them to be significantly accurate. This provides useful information for the authors to draw
a conclusion that even though the is higher in the annual regression, the monthly𝑅
2
regression is a better method to use for the entire population, and a better measure to draw
conclusions from because the significance level is statistically more accurate (Zaionts, 2020).
Shown below:
Table 7: Multiple linear regression analysis statistics (coefficients) annual data:
31
Source: Multiple linear regression analysis annual data Excel,data from SCB.
Even though the annual model is not of any significant value to draw conclusions from, it can
still work as an indicator to show how Exchange rate behaves annually compared to monthly,
when time has been given to the market to react to governmental financial interference such
as changing policy rates. Therefore, the table below shows a Pearson Correlation test done on
the annual figures to illustrate the Strength and direction of the correlation. The results show
that interest rate has a slightly less weakened correlation with Exchange rate, a decrease by
0,015. On the other hand, inflation correlation with exchange rate strengthened with 0.11.
Shown below:
Table 8: Pearson correlation coefficient annual data:
Source: Pearson correlation coefficient test annual data Excel, data SCB.
4.2.3 Regression analysis monthly figures 2003-3007 and 2008-2012
This part of our empirical analysis was done in order to investigate whether or not the
correlation between exchange rate and inflation/interest rate as independent variables is
strengthened or weakened during a financial crisis. Previous regression models have shown
that the monthly figures are of more significance than annual figures, therefore monthly
figures have been used in these following regression analysis. To answer the research
question the timeframe was split up into two parts, the first regression was run on data
stretching between January 2003 and December 2007. The second regression analyses data
from January 2008 to December 2012. The same data has been used in these regressions as
the first regression where monthly figures were analysed over the entire ten-year period. The
results of the two regressions are shown below.
Table 9:  Multiple linear regression analyses monthly data 2003-2007, 2008-2012:
32
Source: Multiple linear regression analysis split data Excel, data from SCB.
The two Multiple linear regression analyses shows that the -value is significantly higher𝑅
2
during- and after the financial crisis of 2008. The timeframe for the two models are the same,
but the -value has increased with approximately 0,349 = 34,9 Percentage points. The table𝑅
2
below illustrates the ANOVA tables for the two models:
Table 10:  Multiple linear regression analysis statistics, ANOVA table 2003-2007, 2008-2012:
Source: Anova table split data  Excel,data from SCB.
As seen in the anova tables, the second period of time (2008-2012) is significantly reliable
while the 2003-2007 timeframe is not. Surprisingly, just as the -value the p-value is very𝑅
2
different in the two models. This shows that the second period of time is a reliable model to
draw conclusions from while the first is not. Lastly, the Pearson correlation coefficient was
applied to the two models, shown below in table 11. The Pearson correlation once again
shows that the correlation between the variables is stronger for the second time frame.
Table 11:  Pearson correlation coefficients monthly data 2003-2007, 2008, 2012:
33
Source: Pearson Correlation Coefficient split data Excel, data from SCB.
5. Analysis
This chapter will provide a further analysis to the results in chapter 4. The results will be
discussed and put into context using the theories established in chapter 2. Moreover, the
research questions will be discussed and answered using the results and theories from
previous chapters.
The main goal of this report is to investigate the relationship and strength of the correlation
between Exchange rate and interest/inflation rate in Sweden. Additionally, the report expects
to provide further evidence to how the relationship between the variables are affected by a
global financial crisis. Four different multiple linear regression analyses were made in order
to provide enough evidence to answer the research questions asked in chapter 1.
5.1 Results multiple linear regression analyses 1 and 2
The first two models were made with monthly and annual data over a time period of 10 years.
The first notable difference is the -value. Monthly data provided a value of 0,277 = 28%,𝑅
2
while annual data resulted in a value of 0,386 = 39%. Moreover, the p-value of the monthly
data was 5,3115E-09 and the p-value for annual data was 0,18. A further analysis of the
numbers can be made from two different points of view. A mathematical point of view and an
economical point of view.
34
5.1.1 Mathematical point of view
The first way to approach the results is from a mathematical point of view using statistics as a
guideline. Deziel (2018) states that a small sample size reduced the power of a study and
increases the likelihood of type II error. The monthly data consists of 120 observations
compared to the annual 10 observations. The extensive difference in sample size between the
two models might contribute to the different results. As mentioned in chapter 3, Flom (2018)
addresses the disadvantages of outliers in a regression model. This correlates to what Deziel
(2018) writes about sample size. A small sample size will naturally be more exposed to
outliers, meaning that the impact of outliers will show a greater effect on the results of a
small sample size. Furthermore, Deziel (2018) mentions that a small sample size increases
the likelihood of a type II error. This statement is proven right by our results, since the annual
data with a sample size of 10 resulted in a p- value of 0,18 compared to 5,3115E-09 for the
large monthly sample size, proving the insignificance of the high p-value. Ultimately the null
hypothesis can not be rejected for our annual regression analysis.
Lastly, the coefficients for our monthly data were 9,22 0,36 -0,22 (in the order exchange
rate/inflation rate/interest rate). compared to 8,97 -0,23 0,53 for annual data. The coefficient
shows that the exchange rate in the monthly data model increases with 1 unit when the
inflation rate is increased by 36%, while it increases with 1 unit when the interest rate has
decreased with 22%. The relationship is switched in the annual data model since the
coefficients have switched directions. Once again, the p-value for interest rate and inflation in
our annual data are insignificant, while every coefficient in our monthly data is significant.
This provides further evidence that the annual data multiple linear regression analysis is
inconsistent and not reliable enough to draw conclusions from (Zaionts, 2020).
5.1.2 Economical point of view
The second way to approach the results is from an economical point of view. The -value𝑅
2
indicates to what degree movements in the dependent variable (exchange rate) depend on
movements in the independent variables. Therefore, the value is used to predict the𝑅
2
outcome of the exchange rate based on changes in inflation and interest rate. values of𝑅
2
0,277 (monthly data) and 0,386 (Annual data) indicates that 28% and 39% of the changes in
35
exchange rate can be explained by interest rate and inflation rate. This is in line with the PPP
theory which states that exchange rates are affected by inflation. The results also confirm
Brigo and Mercurios (2007) claim that central banks are able to manipulate exchange rates by
increasing or decreasing interest rates. A further extension of the PPP theory is the balance of
pay theory, which states that the exchange rate of the domestic currency is determined by its
supply and demand. Ultimately, the theory explains that supply and demand leads to either a
deficit or a surplus in the balance of trade. An increasing demand on foreign currency relative
to domestic will lead to a deficit in the balance of trade, eventually leading to a depreciation
of domestic currency (Stern, 2017). The theory is backed up by the results of the study, and is
closely correlated with the PPP theory. Even though the theory explains exchange rate
movements solely by supply and demand, the supply and demand of a currency is, according
to the PPP theory, explained by inflation rates (Devereux & Engel, 2003).
Further, the increasing -value for the annual data indicates that the sticky price theory𝑅
2
might apply to the results. When annual data were investigated relative to monthly data,
movements in the exchange rate had a longer time to adjust to changing independent
variables. The increased value indicates that the correlation is strengthened when the sticky𝑅
2
price theory has been counted for. Even though the value increased with annual data, the𝑅
2
model is insignificant due to its high p-value. Therefore, the increasing correlation can not be
used for conclusions, and ultimately, the monthly data with a lower correlation determination
coefficient is still a more reliable model. As mentioned in chapter 2, the fisher theory states
that an increase in nominal interest rate should urge the expected inflation to increase as well
(Jonsson & Reslow, 2015). This theory contradicts the general view that an increased interest
rate will lead to decreased inflation and appreciation of the domestic currency because of an
increasing demand on the currency. Because of this, it is important to note that the fisher
theory only tries to predict the expected inflation, not the actual inflation rates (Jonsson &
Reslow, 2015).
In our models the CPIF has been used as a measure of inflation, which reflects the actual
inflation in Sweden. Worth noting is that the inflation rate coefficient in the annual data is
negative while the interest rate is positive, and the other way around for the monthly data. A
negative coefficient indicates an inverse correlation between the variables meaning that when
the independent variable increases the dependent variable decreases. Theoretically, a direct
36
relationship should be displayed between inflation rate and exchange rate in our models,
meaning that they should both increase and decrease at the same time, not inversely. The PPP
theory states that inflation leads the domestic price level to rise, thus, the exchange rate will
depreciate in order to return to PPP. Our models are built on SEK/EUR, meaning that a
depreciation of the swedish currency in our models will increase our exchange rate numbers.
Hence, more SEK will be used to convert into 1 Euro (SEK will be worth less) (Devereux &
Engel, 2003). The high p-value for annual data, along with the contradicting direction of the
coefficients provides further evidence that the model is not reliable. On the other hand, the
direction of the coefficients for monthly data were as expected in the hypothesis. And due to
the significance of the models p-value, the model is reliable.
A further analysis of the model with a significant p-value (model with monthly data over a 10
year period) shows that 28% of the exchange rate movement can be explained by interest rate
and inflation rate movements. The Pearson correlation test for the model shows that the
correlation between exchange rate and inflation is 26% and the correlation between exchange
rate and interest rate is -32%. This proves that the direction of the variables are in line with
the hypothesis in chapter 1. The results also provides further credibility to the multiple linear
regression analysis, since the strength of the correlations are very similar to the value in𝑅
2
the analysis. Ultimately, the results of the multiple linear regression analysis provide a clear
answer to the first research question in this study:
There has been a correlation of 28% between the variables during a 10 year period between
2003 and 2012. Further, the results of the coefficients provide answers to the second research
question as well:
The results show that there has been a direct relationship between inflation and exchange
rate, and an inverse relationship between interest rate and exchange rate. This indicates that
an increased inflation will increase the exchange rate (SEK/Euro, hence, a depreciation of
SEK), while a decreased interest rate will increase the exchange rate (appreciation) with a
28% certainty. This is in line with the hypothesis formed in chapter 1.
The timeframe and the lack of similar studies to match the results makes it rather difficult to
put the results into further context, but the directions of the coefficients in the results are as
expected, and the significance of the model proves that the model is dependable, and works.
The correlation between the variables is weaker than expected, though. There is a clear
37
enough correlation to prove that the independent variables used in this model are in fact a
great influence on the exchange rate, but not enough correlation to prove that the independent
variables in the model are the main forces that decide exchange rate. The limitations of the
timeframe is due to the fact that only a 10 year period was investigated. The results show a
correlation of 28% certainty between the variables during the time period of 10 years that
were investigated, but the model lacks evidence to suggest that the results were to be similar
for any given period of time. This report does not prove a definitive correlation between the
variables, it only provides evidence for the time period that was analyzed. Furthermore, even
though the results do not conclude a correlation for any given time, the fact that the
timeframe that was investigated resulted in a clear correlation provides enough support to
form an additional hypothesis. The results provide a foundation for an additional thesis to be
made for future research, stating that the correlation between the variables are correlated with
similar degree for any given time.
As mentioned in section 1.2 a similar study was posted by Tafa (2015) in the medeteranian
journal of social sciences. Tafa (2015) conducted a linear regression analysis with the
albanian exchange rate relative to the Euro used as the dependent variable, and the albanian
interest rate over a 12 year period as the independent variable. The results of the study
showed a of 0,003 and a p-value of 0.4, indicating that the study is statistically𝑅
2
insignificant. Compared to the study made by Tafa (2015), the results of this study show a
way stronger correlation, and is statistically significant. There are a lot of different reasons
why the studies show such different results. The biggest reason is that there are countless
socioeconomic differences between Sweden and Albania, and the political influence on the
interest rate might be handled differently. Even though the socioeconomic differences
between the two countries makes it hard to compare the results, there are some similarities
worth mentioning. Tafa (2015) used a similar timeframe as this study does. (12 years
compared to 10 years). Additionally, the studies are investigating a correlation during the
same historical time period (before and after the crisis of 2008). The differences of the studies
are too great to draw any conclusions from, but at the same time the similarities provide an
interesting area of comparison.
The fact that aproxiately 30% of exchange rate movements are related to interest rate and
inflation shows that the remaining 70% are due to different variables, such as public debt and
terms of trade (additional variables are listed in chapter 1). Ultimately, the human factor
38
plays a role in the expectations of future economical influences, and future exchange rate
movements, which affect the present exchange rate. In the Fischer theory, expected inflation
increases when nominal interest rates increase (Jonsson & Reslow, 2015). Expectations are
also a factor in the determination of exchange rate, just as in other financial markets. This
makes it impossible to determine the exact movement of the exchange rate in present time.
This study has proven that 28% of the exchange rate movements can be explained with two
independent variables. If the study were to be further examined, and more independent
variables were to be used, the probability of an even higher correlation between additional
variables might be proven. Considering the irrational nature of the human factor, and the
difficulties of measuring expectations from it, a 28% proven correlation with 2 independent
variables is a good starting point for future research.
5.2 Results multiple linear regression analyses 3 and 4
The last two models were done with the same time period as before, but split into two
different models investigating the difference of the correlation before and after a global
financial crisis. Monthly data was used for both models since the previous models showed
that the p-value for monthly data is significant while annual data is not. The models were
split into 2003-2007 and 2008-2012. The value for the first time period was 0,025 = 2,5%𝑅
2
and the p-value was 0,484. The value for the second period was 0,374 = 37% and the𝑅
2
p-value was 1,5568E-06. The results are surprising, and shows that the correlation between
the variables are stronger after a financial crisis rather than before. Hence, the correlation
between exchange rate, interest rate and inflation rate are strengthened by a financial crisis.
The models do present some limitations. The models do not show clear evidence that the
relationship between the variables are strengthened solely because of the financial crisis in
2008. Further, it is not possible to tie the correlation to the financial crises. Fluctuations in the
exchange rate happens daily, and with the models that have been conducted there is no way to
connect these fluctuations to the financial crises. This means that even though the value is𝑅
2
substantially bigger in the second model, a conclusion can not be drawn to it being by virtue
of the financial crises. Even though any definite conclusions can not be drawn from the
results, they do emit data that indicates a stronger correlation during and after a financial
crisis.
39
Fratzscher (2009) writes in the report “What explains global exchange rate movements during
the financial crises” that the financial exposure of a country, and countries with a weak
currency account position suffered substantially more in terms of currency depreciation
during the crisis. The figures gathered from OFX shows that the swedish exchange rate
relative to the Euro went from 9,42 to 10,42 from January 2008 - December 2009, hitting its
highest point of 11,19 in March 2008 (OFX n.d.). The model in this paper does not measure
the currency account position of Sweden relative to other countries so there is no way to
prove Fratzschers claims. On the other hand, the interest rate in Sweden dropped from 4,11 to
0,17 during the same time period as the exchange rate rose from 9,42 to 10,42. As mentioned,
the results of the model with monthly data during a 10 year time period confirmed the
hypothesis of an inverse relationship between interest rate and exchange rate. This provides
evidence to why the correlation might have been strengthened during the crises since a sharp
increase in exchange rate took place simountainlesly as a sharp decrease of the interest rate.
Ultimately, the inverse relationship between exchange rate and interest rate, along with the
strengthened correlation during a time when interest rate decreased and exchange rate
increased strengthens the argument made by the central bank of Sweden that an increased
policy rate increases the market rate, which leads to an appreciation of the domestic currency,
because the high interest rates is attractive for foreign investors (Riksbanken 2018a).
6. Conclusion
The purpose with this paper was to see if there exists a correlation between exchange rate and
interest/inflation rate, and to further investigate how strong the correlation is. In order to
investigate this, a timeframe of 10 years was used in a multiple linear regression analysis.
Monthly and annual data for a time period of January 2003 - December 2012 were conducted
into a model in order to answer the research questions. Exchange rate data from OFX and
interest/inflation rates from SCB were used in the analysis. Further, the report expected to
provide evidence to how the correlation of the variables are affected by a global financial
crisis. The crisis of 2008 was used since data for the variables do not exist for financial crises
further back in time. The research question displayed in chapter 1 were as follows:
40
● Has there been a correlation between the exchange rate movements and the interest
rate/inflation movements in Sweden during a 10 year period, ranging from 2003 until
2012. If so, how strong is the correlation?
● In what directions does interest rate and inflation rate influence exchange rate?
● Will the exchange rate correlation with interest and inflation rate be strengthened or
weakened during a financial crisis?
The answers to the research questions are based on the findings of the multiple regression
analysis and are as follows:
There has been a clear correlation between exchange rate movements and the interest
rate/inflation movements in Sweden during a 10 year period, ranging from 2003 until 2012.
The results of the multiple linear regression analysis indicates that 28% of the movements in
exchange rate during the 10 year period can be explained by movements in interest rate and
inflation. Furthermore, the Pearson correlation test shows a correlation of 26% between
exchange rate and inflation rate, and a correlation of -32% between interest rate and inflation.
Moreover, the Pearson test along with the coefficients for each independent variable in the
ANOVA table shows a negative correlation between interest rate and exchange rate. This
provides an answer to the second research question:
A negative correlation indicates an inverse relationship between exchange rate and interest
rate. Hence, when interest rates are increasing, the exchange rate tends to decrease. While an
increased inflation tends to also increase the exchange rate. Ultimately, increased interest
rates decrease the exchange rate (SEK relative to the Euro) leading to an appreciation of the
Swedish currency. On the other hand, an increasing inflation also increases the exchange rate
of SEK relative to the Euro, which indicates a depreciation of the Swedish currency.
Lastly, The results indicate that the correlation between the variables are strengthened during
a financial crisis, though, the model in this report is not advanced enough to draw an
definitive conclusion that the strengthened correlation is due solely to the financial crisis.
With the results of the report, the hypotheses in chapter 1 can be tested. The hypothesis were
as follows:
● An increase in interest rates of domestic currency will cause the domestic currency to
appreciate against other foreign currencies, while an increased inflation will
41
depreciate the domestic currency. Further, a global financial crisis will decrease the
exchange rate correlation with interest rate and inflation rate due to uncertainty.
The Hypotheses of the study shows to be partly true. an increasing interest will, according to
the results, lead to an appreciation of the domestic currency while an increasing inflation will
depreciate the domestic currency. However, a global financial crisis will not decrease the
exchange rate correlation with interest rate and inflation according to the results of the study.
6.1 Further studies in this subject
This study did face some limitations, such as it focusing on how the exchange rate in Sweden
is affected by inflation and interest rate, without taking into account the remaining five
variables. Further, the time frame examined is a limitation. The fact that the investigated time
frame only takes one financial crisis into account limits the results, since they cant be
compared to further results of another crisis. The results of this study merely provides strong
evidence of an existing correlation at all times, since the correlation exists during the 10 years
examined.
The study focuses solely on interest rate and exchange rate. An interesting subject for further
research would be to investigate other contributing variables to get a bigger picture of the
exchange rate movements. In order to elaborate this study further and continue the research, a
recommendation for future researchers would be to take more factors into account, such as
politics and import/export. It would be interesting conduct this study on a bigger scale, by
comparing more countries with more variables taken into account. Further, it would be
interesting to investigate a longer period of time to see if the correlations are affected in the
same way during another global financial crisis. The reason this would be interesting is to see
how big the difference in the correlations would be, since this study is narrowed down to the
correlation in Sweden during a limited period of time.
Also when it comes to the literature there is a very limited selection on previous studies and
articles that take all three variables into account. Instead most of the previous studies focus on
either the effect on exchange rates with inflation rate or they choose to focus on how the
42
exchange rates are affected by interest rate. This report aims to provide a fulfillment of this
research gap. Hopefully, this study will bring knowledge to future researchers.
Reference list:
Back, A. (2013). Admission to official list Australian security exchange
https://www.asx.com.au/asxpdf/20131011/pdf/42jyv7syc05cff.pdf
Ball, L. & Mankiw, N. G (1994). A sticky-price manifesto. NBER.
Brigo, D. & Mercurioth, F. (2007). Interest Rate Models - Theory and Practice: With Smile, Inflation
and Credit (2th). Springer.
Chatterjee, S. & Hadi, A. S. (2012). Regression analysis by example. Wiley
Cheusheva, S. (2020). Linear regression analysis in Excel. Adlebits.
https://www.ablebits.com/office-addins-blog/2018/08/01/linear-regression-analysis-excel/
Devereux, M. B. & Engel, C. (2003). Monetary Policy in the Open Economy Revisited: Price Setting
and Exchange-Rate Flexibility. NBER. Volume 70, number 4, (P. 765-783).
Deziel, C. (2018). The Effects of a Small Sample Size Limitation. Sciencing.
https://sciencing.com/effects-small-sample-size-limitation-8545371.html
Duarte, M. & Stockman, A. C. (2002). Comment on: Exchange rate pass-through, exchange rate
volatility, and exchange rate disconnect. Journal of Monetary Economics, 49, (P. 941-946).
43
Flom, P. (2018). The Disadvantages of Linear Regression. Sciencing.
https://sciencing.com/disadvantages-linear-regression-8562780.html
Fratzscher, M. (2009). What explains global exchange rate Movements During the Financial crisis?.
European Central Bank.
https://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1060.pdf?1dfad27c734f8404869e0ed27eda4fc9
Fregert, K. & Jonung, L. (2014). Makroekonomi: Teori, politik och institutioner. Studentlitteratur.
Frost, J. (2020). How to Perform Regression Analysis using Excel. statistics by jim.
https://statisticsbyjim.com/regression/regression-analysis-excel/
Gibson, H. D & Thirlwall, A. P. (1992). Exchange rate determination. Balance-of-payments theory
and the United Kingdom experience (P. 59-128)
Greener, S. & Martelli, J. (2018). An introduction to business research methods. Bookboon.
http://103.5.132.213:8080/jspui/bitstream/123456789/371/1/an-introduction-to-business-research-met
hods.pdf
Johansson, J. (2015). Hur mäts inflation?. (2015:05). Sveriges Riksbank.
Jonsson, M. & Reslow, A. (2015). Interest and inflation rates through the lens of the theory of Irving
Fisher. (2015:02). Sveriges Riksbank.
Lagerwall, B. (2008). Realräntan i Sverige. (2015:05) Sveriges Riksbank.
Mishkin, F. S. (2008). Exchange rate pass-through and monetary policy. NBER.
Ohlin, J. (2018). Finanskrisen - Vad var det som hände?. SVT. Retrieved 2020-11-14
https://www.svt.se/nyheter/ekonomi/finanskrisen-vad-var-det-som-hande
OFX. (n.d.). Historical exchange rates. Retrieved 2020-11-06
https://www.ofx.com/en-au/forex-news/historical-exchange-rates/
Riksbanken, (2018a). Så påverkar penningpolitiken inflationen. Retrieved 2020-12-15
44
https://www.riksbank.se/sv/penningpolitik/vad-ar-penningpolitik/sa-paverkar-penningpolitiken-inflati
onen/
Riksbanken (2018b). Inflationsmålet. Retrieved 2020-12-15
https://www.riksbank.se/sv/penningpolitik/inflationsmalet/
Riksbanken, (2019) Ingves: Inflationsmål och rörlig växelkurs har fungerat väl i Sverige. Retrieved
2020-11-23
https://www.riksbank.se/sv/press-och-publicerat/nyheter-och-pressmeddelanden/debattartiklar/ingves-
inflationsmal-och-rorlig-vaxelkurs-har-fungerat-val-i-sverige/
Sahisahli, M. K. & Ho, J. C. (2002). Inflation, interest rate and exchange rate: What is the
relationship? Journal of economics and economic education research, volume 3, nr 1, 2002
SCB, (n.d.) Vad är SCB? Retrieved 2021-01-02
https://www.scb.se/om-scb/lattlast/vad-ar-scb/
SCB, (2020a). Konsumentprisindex, KPI. Retrieved 2021-01-03
https://www.scb.se/hitta-statistik/statistik-efter-amne/priser-och-konsumtion/konsumentprisindex/kons
umentprisindex-kpi/
SCB, (2020b). Inflationstakten enligt KPI och KPIF. Retrieved 2020-11-16
https://www.scb.se/hitta-statistik/statistik-efter-amne/priser-och-konsumtion/konsumentprisindex/kons
umentprisindex-kpi/pong/tabell-och-diagram/konsumentprisindex-med-fast-ranta-kpif-och-kpif-xe/inf
lationstakten-enligt-kpi-och-kpif/
SCB, (2021). Kort och lång ränta. Retrieved 2020-11-15
https://www.scb.se/hitta-statistik/statistik-efter-amne/ovrigt/allmant/sveriges-ekonomi/pong/tabell-och
-diagram/kort-och-lang-ranta-1989-/
Shambaugh, J. (2004). The Effect of Fixed Exchange Rates on Monetary Policy. The Quarterly
Journal of Economics, Volume 119, (P. 301-352).
SOS. (2015). Pearson Correlation and Linear Regression. The university of Texas at Austin
http://sites.utexas.edu/sos/guided/inferential/numeric/bivariate/cor/
45
Stern, R. (2017). Balance of Payments: Theory and Economic Policy. Routledge.
Tafa, J (2015). Relationship between exchange rates and interest rates: case of Albania.
Medeterrenian journal of social studying
Taylor, J. B. (2001). The Role of the Exchange Rate in Monetary-Policy Rules. The American
economic review, Volume 91, nr 2,  (P. 263-267).
Totonchi, J (2011) Macroeconomic theories of inflation. International conference on Economics and
finance research IPEDR, Volume 4, (P. 459-462) IACSIT Press, singapore
Twin, A. (2020). 6 Factors that influence exchange rates. Investopedia. Retrieved 2020-12-05
https://www.investopedia.com/trading/factors-influence-exchange-rates/
Zaiontz, C. (n.d.). Multiple Regression Analysis in Excel. Real statistics using excel.
https://www.real-statistics.com/multiple-regression/multiple-regression-analysis/multiple-regression-a
nalysis-excel/
46