What causes fluctuations in the Swiss franc?

(1)

Universiteit van Amsterdam

What causes

fluctuations in the

Swiss franc?

Bachelor Thesis

Author: Lucas Grummels Supervisor: Christiaan van der Kwaak February 20, 2014

(2)

1 Supervisor: Christiaan van der Kwaak

INTRODUCTION: ...2

CHAPTER 2: EXCHANGE RATE DETERMINATION MODELS: ...3

2.1INTRODUCTION: ... 3

2.2MONETARY MODELS:... 3

2.2.1 General features: ... 4

2.2.2 Flexible-Price Monetary model: ... 6

2.2.3 The Dornbusch sticky-price monetary model: ... 8

2.2.4 Rational Expectations Present Value Model: ... 11

2.3PRODUCTIVITY BASED MODEL: ... 14

CHAPTER 3: EXCHANGE RATE REGIMES SWITZERLAND USED

SINCE THE BRETTON WOODS SYSTEM. ... 18

3.2SWITZERLAND IN THE BRETTON WOODS SYSTEM: ... 18

3.3FLOATING EXCHANGE RATE REGIME:... 19

3.3.1 Monetary Target Regime: ... 19

3.3.2 Monetary Target Regime from 1970 until now: ... 20

3.4MONETARY POLICY CONSEQUENCES OF THE TEMPORARY PEG TO THE EURO: ... 24

CHAPTER 4: MODEL PERFORMANCE ... 28

4.2FUNCTION AND MALFUNCTION OF EXCHANGE RATE DETERMINATION MODELS: ... 28

4.2.1 Evidence on monetary models: ... 28

4.2.2 Empirical evidence on Productivity Based Models: ... 29

4.3OTHER IMPORTANT FACTORS IN DETERMINING THE EXCHANGE RATE: ... 30

4.3.1 Switzerland’s safe haven currency status: ... 30

4.3.2 The impact of oil prices on exchange rates: ... 32

4.3.3 The introduction of the Euro and its impact on the Swiss franc: ... 33

CHAPTER 5: CONCLUSION: ... 37

APPENDIX: DORNBUSCH MODEL ... 39

(3)

Introduction:

In September 2010, Brazilian Financial minister Guido Mantega warned for a currency war. Several developed countries tried to devalue their currencies by following a series of monetary interventions by their governments (Hotten, 2010). By devaluing their currencies, governments lower the value of their currency and try to boost its exports and employment as a result.

A country that was involved in these monetary interventions is Switzerland. The Swiss franc, which experienced appreciation against the euro since the start of the global financial crisis in 2007, became a threat to the Swiss economy. Swiss firms relying on exports

experienced a decreasing demand since foreigners regarded their products and services as too expensive because of its strong currency. Swiss citizens decided to shop abroad to exploit the lower prices of foreign goods and services (Hofs, 2011). These facts made the Swiss National Bank decide that it will no longer allow the Swiss franc to appreciate against the euro. In September 2011, the Swiss National Bank announced that it was prepared to buy foreign currency if the value of the Swiss franc approaches 1,20 CHF/€. By doing this, the Swiss National Bank set a lower bound for its exchange rate against the euro and attacked the over-valuated Swiss franc, which caused deflationary pressures and was a threat to the Swiss economy (SNB, 2011).

The actions taken by the Swiss National Bank show that the exchange rate is of major importance to the economy. Several economists have studied the exchange rate and created models in which their movements are predicted. In this study I’ll try to overview the literature on exchange rates and see if there are factors that can explain the Swiss franc’s movements. Since an enormous amount of economists have researched exchange rate models and time is a major constraint, I’ve selected literature that is most important in predicting and explaining movements in the Swiss exchange rate. With this method I’ll seek to give an answer to the research question: What causes fluctuations in the Swiss exchange rate?

The paper is structured as follows: In the next chapter, I’ll present several important exchange rate determination models. In chapter 3, I’ll discuss the exchange rate regimes

Switzerland used since the Bretton Woods system. Evidence on the models and other important factors determining the exchange rate are discussed in chapter 4 and chapter 5 concludes.

(4)

CHAPTER 2: Exchange rate determination models:

2.1 Introduction:

Since the break-down of the Bretton Woods exchange rate system in the seventies, foreign exchange markets became more and more important. During the system of Bretton Woods many exchange rates were pegged to the dollar so they were almost fixed. After the breakdown of the system, most exchange rates became floating exchange rates which were determined in the foreign exchange market. Since then, many economists developed models that seek to determine the exchange rate.

In this paper it is impossible to review all exchange rate models so it is important to make choices about which models to use and describe. Some major literature gave me insight in the most important models and to make this chapter more readable, I will make a distinction in these models at first.

The major distinction in exchange rate models is between monetary models and productivity based models. In monetary models there is a major role for money markets while in the productivity based model, the Balassa Samuelson effect is most important. I will first describe the monetary models in subsection 2 and the productivity based model will follow in subsection 3.

There are many more models to describe the fluctuations in the exchange rate but with these models I cover the most important models from recent and more dated literature.

2.2 Monetary Models:

In this section a few monetary models are discussed. In these models exchange rates can be seen as the relative prices of different currencies. These relative prices are determined in the asset market and like in all markets there will be room for expectations about future relative prices. Authorities have the power to control the supply of money and because of that they affect investor’s expectations about future relative prices of currencies. That is why the money supply plays such a crucial role in determining the exchange rate (Mussa, 1979). Although the money supply plays a crucial role, there are more important factors determining the exchange rate (Mussa, 1979). The models in this section will show which factors are important in exchange rate determination.

To understand the monetary models, it’s important to keep in mind some general features. These general features are discussed at first. After that, the flexible-price monetary model, the Dornbusch sticky-price monetary model and the rational expectations present value model will be discussed.

(5)

4 Supervisor: Christiaan van der Kwaak 2.2.1 General features:

To create monetary models, it’s important to start with the general features. Frenkel (1976) uses three building blocks to start with.

2.2.1.1 Money market equilibrium:

In monetary models the equilibrium in the demand and supply for money is the starting point of all other actions.

The demand for money is determined by the price level, the level of real income and the nominal interest rate, which is stated in the following formula:

(1.)

In equilibrium in the money market, money demand equals money supply. Since the money supply is set by the authorities the following equation can be made for the money market equilibrium:

(2.) (3.)

In this equation m is the logarithm of the nominal money supply, p is logarithm of the price level, y is the logarithm of real income and i is the nominal interest rate. Equation 2 describes a domestic demand and supply function and equation 3 describes the foreign demand and supply function.

When there are changes in the factors that determine demand or supply of money, the price level is the factor that should equilibrate demand and supply (Bilson, 1978).

2.2.1.2 Purchasing Power Parity (sometimes abbreviated as PPP):

The second building block in monetary models is the assumption of Purchasing Power Parity (Frenkel, 1976). Although this method is really old, it is still used in many models. In the sixteenth century in Spain, some scholars thought about prices in different countries and they came to the basic idea of the Purchasing Power Parity. Once currencies are converted into the same currency, prices levels from different countries should be equal (Rogoff, 1996).

In our words, this means that the nominal exchange rate is explained by the ratio of domestic price level to foreign price level (Taylor, 2003). This is shown in the following formulas:

(4.) (5.)

(6)

5 Supervisor: Christiaan van der Kwaak Equation 4 shows the exchange rate as the ratio of domestic price level to foreign price level and Equation 5 shows the logarithm of equation 4. This last equation will mostly be used in the models.

PPP relies on the concept of the law of one price which states that in a competitive market, prices of goods should be equal when converted to the same currency with the assumption that transport and transaction costs are equal to zero. In other words the price of a bundle of goods in Switzerland should be equal to that same bundle of goods in Germany when both prices are expressed in francs or in euro’s. By looking at equation 4 and 5 one can easily see that when prices in Switzerland increase compared to prices in Germany, their Swiss franc should depreciate.

Several economists used PPP to explain movements in exchange rates with varying success. Important findings are for example that PPP performs better for countries near to each other and were trade barriers are almost negligible (Frenkel, 1981). In addition, PPP cannot really explain the large fluctuations in the exchange rates. Exchange rates are much more volatile than price levels, which change relatively slow over time (Frenkel and Mussa, 1980). Some

economists state that PPP performs better in the long run than in the short run which is supported by the literature (Rogoff, 1996). However, PPP can still be a useful tool in determining and following the exchange rate and that’s why lots of economists use the assumption of PPP in their models.

2.2.1.3 Uncovered Interest Rate Parity (sometimes abbreviated as UIP):

The last building block in monetary models is about expectations on asset prices (Frenkel, 1979). In capital markets, international investors can switch easily from Swiss francs to euros and other currencies. When investing in international bonds, interest rates on government bonds and expectations about the value of currencies are crucial (Pilbeam, 2006). It is assumed that the capital markets are efficient which means that prices reflect all available information. This theory is captured in the Uncovered Interest Rate Parity which states that an interest rate difference should be offset by an expected rate of depreciation (Taylor, 1995). UIP can be written as follows:

(6.) ̇

In this equation one can see that when the interest rate on Swiss bonds is for example 1% higher than the interest rate on German bonds, the Swiss franc is expected to depreciate by 1%. For UIP to hold, we need to make some crucial extra assumptions. The first one is that investors are indifferent in choosing Swiss or German bonds, which makes these bonds perfect

(7)

6 Supervisor: Christiaan van der Kwaak substitutes. Perfect capital mobility is the second important assumption and the third one states that investors are risk-neutral (Taylor, 1995).

2.2.2 Flexible-Price Monetary model:

All important general features are listed now and by using these building blocks, complete monetary models can be created. The flexible-price monetary model is the first model to discuss. It has been one of the most influential models since the start of floating exchange rates in the seventies. The model captures the exchange rate as the relative price of two different currencies and to establish this relationship it incorporates money markets of different countries (Taylor, 1995).

In the flexible-price monetary model, the equilibrium in the money market is calculated as it is done in section 2.1.1. Thereby the law of one price is assumed to hold continuously, both on commodities and financial assets. This means that PPP and UIP hold continuously in this model (Bilson, 1978). Since all building blocks are met, we can build the model, which starts with rewriting the money market equilibria formulas of equations 2 and 3:

(7.) (8.)

It should be noted that when changes in factors determining the money supply and demand occur, the price levels adjust immediately to equilibrate demand and supply (Bilson, 1978). Since the assumption is made that PPP holds continuously, it’s possible to rewrite this equation by using equation 5.

(9.)

In this equation s, which represents the nominal exchange rate, is positively affected by the relative money supplies and the relative interest rates and negatively affected by the relative levels of output. Since the exchange rate here is defined as domestic currency per unit of foreign currency a positive effect means depreciation and a negative effect means appreciation of the currency. Thereby, the assumption is made that the parameters and are equal for both countries.

Since UIP also holds continuously, the last part of the equation can be rewritten. This will bring us the following equation:

(8)

7 Supervisor: Christiaan van der Kwaak By watching the flexible-price model equations one can predict what happens to the nominal spot exchange rate when differences in the variables of the model occur:

A. Differences in relative money supplies:

In equation 7 one can see that an increase in the money supply will increase prices, while assuming that other factors remain constant. Since this model assumes PPP to hold continuously and assumes prices to immediately equilibrate the supply and demand in the money market, this money supply increase will be matched immediately by an increase in domestic prices and because of that it changes the exchange rate (Bilson, 1978). That’s why in equation 10 an increase in relative money supply by 5% changes the exchange rate positively (depreciates) by 5%.

B. Differences in relative levels of national income:

Levels of income determine the demand for money (Mussa, 1979). In equation 7 you can see that when the national income increases, money demand and money supply can only be equilibrated by a fall in prices. In relative levels this means that an increase in the relative level of national income will lower the relative price level and thus lowers (appreciates) the exchange rate.

C. Differences in relative interest rates:

Higher levels of interest rates lower the demand for money. To equilibrate the money demand and supply, you can see that prices should increase according to equation 7. According to PPP, prices will increase which positively affects (depreciates) the exchange rate.

Pilbeam (2006) uses another way to describe the effect of a difference in interest rates. The nominal interest rate can be written as the real interest rate plus the expected rate of price inflation. In the following formula, the expected rate of price inflation is defined as ̇.

(11.) ̇ (12.) ̇

While assuming real interest rates to be equal for both countries and being at a constant level, a relative increase in nominal interest is explained by a relative increase in

expected price inflation. This leads to an increase in expenditure on goods which increases goods prices. This increase in goods prices is explained with PPP by a depreciation of the exchange rate (Pilbeam, 2006).

The flexible-price model relies fully on its building blocks. In macroeconomics generally six markets are distinguished: The money market, the goods market, the labour market, the foreign

(9)

8 Supervisor: Christiaan van der Kwaak bonds market, the domestic bonds market and the foreign exchange market. In equilibrium in the foreign bonds market and the domestic bonds market, investors are indifferent in investing in foreign bonds or domestic bonds. Following UIP, interest rate differences are offset by an expected depreciation which equilibrates these markets. Foreign exchange markets react

quickly to changes in variables which is reflected in the spot exchange rate. These changes in the spot exchange rate equilibrate these markets too. Prices and wages are assumed to react quickly to new conditions equilibrating the goods and the labour market. Equilibrium in the money market then equilibrates the whole macroeconomic system (Taylor,1995).

We will see that other models review these building blocks. Especially the PPP condition to hold continuously is criticized. Several papers provided evidence for exchange rates to be much more volatile than prices which could be seen as a critic on the condition for PPP to hold continuously (Dornbusch, 1976). In the next model, this critic on the flexible-price monetary model is

incorporated into a new model.

2.2.3 The Dornbusch sticky-price monetary model:

The sticky-price monetary model was originally created by Rudiger Dornbusch. The following is a brief explanation of his work in 1976. Many economists used this model and fine-tuned it, so there have been some variations on the model in later literature.

The model should explain the large fluctuations of exchange rates. To do this it allows goods markets and assets markets to adjust at different speed to changes in variables. Dornbusch (1976) assumes prices and wages to be sticky in the short run so we will see that the goods market adjusts at a lower speed to changes in variables compared to asset markets (Dornbusch, 1976). This fact makes it possible for variables to overshoot their long-run values compensating for the stickiness of others (Taylor, 1995). This is the first difference between the sticky-price models and the flexible-price models from the previous section.

Dornbusch (1976) divides the model into three sections: Capital Mobility and Expectations, the

Money Market and the Goods Market. By following these steps, we go over all assumptions and

equations.

A. Capital Mobility and Expectations:

The first assumption to make is the one concerning the pricing of assets and capital mobility. Just like in the flexible-price model, assets are assumed to be perfect substitutes given a premium to offset a change in the exchange rate. This makes UIP hold continuously:

(1.) ̇ (2.) ̇ ̅

(10)

9 Supervisor: Christiaan van der Kwaak Equation 1 is similar to the UIP condition stated in previous sections. Equation 2 defines expected depreciation as the difference between the long run exchange rate and the spot exchange rate. represents a parameter of adjustment to these changes. The spot exchange rate is defined as domestic currency per unit of foreign currency.

B. The Money Market:

In the money market, again money supply must equal money demand. In a general function it reads as follows:

(3.)

The money supply and the value of income are assumed to be given. Combining and rewriting equations 1, 2 and 3 will give the following formula:

(4.) ̅

A stationary money supply means that the spot exchange rate is equal to its long run level. In equation 1 and 2 one can see that there will be no gap between domestic and foreign interest rates given this situation, because of that can be written as its

stationary interest rate (Dornbusch, 1979). Making these assumptions about the long run allows us to make the following equation for the long run price level:

(5.) ̅

One can see that the money supply and the interest rate have a positive influence on the price level in the long run. National income has a negative impact on the long run price level. By substitution of this long run price level equation in equation 4, it’s possible to isolate the spot exchange rate and form the following formula for it:

(6.) ̅

̅

The spot exchange rate is written here as a function of the long run exchange rate and a gap between the short run and long run price level.

An increase in price level will increase domestic interest rates (watch equation 3 and assume the value of the money supply and income to be constant). This increase in interest rates creates a capital inflow which makes the exchange rate to appreciate immediately. This is what this formula shows; the increase in price level lowers the spot exchange rate immediately. Now return to the asset market in which investors take notice of the increased interest rate. Like what is stated in UIP, investors are indifferent in investing in two different bonds if the interest rate difference is offset by an expected

(11)

10 Supervisor: Christiaan van der Kwaak depreciation of the exchange rate. Since the exchange rate immediately appreciated after the change in prices (See equation 6), investors believe that this exchange rate will depreciate in the future (Dornbusch, 1979).

C. The Goods Market:

The demand for domestic output in the goods market depends on three variables: relative prices of domestic goods, interest rates and real income/output. Dornbusch (1976) denotes relative prices of domestic goods as which is the logarithm of the ratio of the exchange rate to prices. Additional to these variables, u is added as a shift parameter. The following equation represents the logarithm of the demand function: (7.)

From this equation, we can create a formula for the rate of increase in the price level of domestic goods:

(8.) ̇

Here the rate of increase in the price level is represented by a fraction of the gap between demand and supply. Supply represents the y in this formula (Pilbeam, 2006). From here, we go on looking for a way to determine the long run exchange rate. To find this, we assume that in the long run the domestic interest rate is equal to the foreign interest rate and that the rate of increase in prices is equal to 0. The following equation will then show a formula for the long run exchange rate:

(9.) ̅ ̅

In this equation you can clearly see that the long run exchange rate depends on monetary variables as the interest rate and real variables as real income/output (Dornbusch, 1976).

From here Dornbusch uses some mathematics to rewrite the formula for the increase in domestic price level. In the appendix you can find a step by step method to finally reach this following formula:

(10.) ̇ [

] ̅

Here the increase in price level is written as a function of the gap between actual prices and long run prices.

(12)

11 Supervisor: Christiaan van der Kwaak (11.) [ ]

Calling the term in front of the gap between actual price level and long run price level v makes following actions easier. In the following equation Dornbusch (1976) introduces a time dimension to the formula. To do this he writes ̅ as ̅ . By combining and rewriting equations 10 and 11 the following formulas can be reached:

(12.) ̇ ̅

(13.) ̅ ̅

In this last equation you can see what happens with prices. In the long run, ̅ will be 0. At a short term, a gap in ̅ will cause prices to change to later converge to its new long run value at a rate determined by v(t) (Dornbusch, 1976).

Substitution of equation 13 in equation 6 will give the time path of the exchange rate: (14.) ̅

̅

(15.) ̅ ̅

In these last equations one can see that the exchange rate will converge to its long run value. From equation 14 is it clear that a positive gap in price level appreciates the spot exchange rate at that moment. When time passes the second part of the equation becomes less negative and will approach zero at a certain moment in time. At that moment, the exchange rate is fully converged to its long run exchange rate. The

exchange rate first overshoots or undershoots its long run value, to later converge to it (Stockman, 1980).

2.2.4 Rational Expectations Present Value Model:

The Dornbusch model and the flexible-price model state that the spot exchange rate depends on fundamental variables such as money supplies, outputs and interest rates. In the past, many researchers used these kinds of models to predict future values of the exchange rate but they noticed that most of the time these models could not beat the assumption that exchange rates follow a random walk. In Mussa’s work (1979) he states the following: ‘The natural logarithm of the spot exchange rate follows approximately a random walk’. Meese and Rogoff (1983) tested several models and they also did not find evidence for their models to beat the assumption of exchange rates following a random walk. More recent work of Cheung, Chinn and Pascual (2002) concluded that some models are suitable for some exchange rates, but they perform poorly on others.

(13)

12 Supervisor: Christiaan van der Kwaak These results inspired Engel and West (2005) to establish a model in which the spot exchange rate is the discounted value of observable and unobservable shocks in fundamental variables. With such a model, exchange rates can be seen as asset-prices which are determined in

organized markets. Prices in these markets reflect all the information available at that moment. Price changes reflect new information about future economic conditions (Frenkel and Mussa, 1985). Engel and West (2005) use the following formula to start with:

(1)

In this formula, the spot rate is explained by a weighted average of observable fundamentals, unobservable fundamentals and the expected spot rate in one year. The spot rate is defined as

(which is the logarithm of the price of home currency per foreign currency), observable fundamentals as , unobservable fundamentals as and the expected spot rate in one year is defined as .

The observable – and unobservable fundamentals are shocks in factors that are important in the determination of the spot exchange rate. Examples of these shocks are shocks in money

supplies, incomes and interest rates. Unobservable shocks also include measurement errors in fundamentals. Before going further, Engel and West (2005) make the assumption that

expectations are rational and that when the term approaches 0. This last

assumption is known by the ‘No-Bubble Solution’, which is best explained by a state in which no drastic departures from the fundamentals are expected (Taylor, 1995). These assumptions make the following formula possible:

(2) ∑ ( ) ∑ ( )

This is the framework Engel and West (2005) use to explain exchange rate determination. From here it’s important to incorporate some fundamentals into this model. Engel and West (2005) label it as the money-income model and it’s quite similar to the flexible-price model and the Dornbusch model discussed earlier:

Assume the following domestic - (3) and foreign (4) money market equilibrium: (3)

(4)

In these equations is the logarithm of the money supplies, is the logarithm of the price levels, is the logarithm of the levels of output and shows a shock to the money demand.

Then the authors define the nominal spot exchange rate as the difference between domestic prices and foreign prices plus the real exchange rate.

(14)

13 Supervisor: Christiaan van der Kwaak (5)

In this equation is determined by PPP plus the real exchange rate which is defined as . For the Uncovered Interest rate Parity, the following equation is used:

(6)

In this equation the authors deviate from UIP with the term: . This term can be seen as a risk premium or a measurement error.

All the important assumptions and equations are listed now. To write a formula for the spot exchange rate, we need to define the observed and unobserved shocks. In equation 7, 8 and 9 the observable and unobservable fundamentals are defined:

(7)

In this formula, the observable fundamentals are defined as the relative money supplies and the relative real incomes.

(8)

Here, unobservable fundamentals are defined as the real exchange rate and the relative shocks to the money demand.

(9)

In this formula, the unobservable fundamental is defined by the risk premium .

The authors combine and rearrange equations 3, 4, 5 and 6 to reach the following formula:

(10)

( )

This equation can be compared with equation 1 by writing for

and for . Then by

using equations 7, 8 and 9 for the observable and unobservable fundamentals, you can see that tis formula is similar to equation 1.

(15)

14 Supervisor: Christiaan van der Kwaak In this model, when the is high, which represents the discount factor, it is likely that the

nominal exchange rate follows a random walk since it is mostly represented by the

unobservable fundamentals and that’s what these authors try to establish in their research.

The conclusion of Engel and West (2005) is that they could not find enough evidence for the nominal spot exchange rate to follow only observed fundamentals. So under some

circumstances there must be a part in the determination of the spot rate that is explained by unobserved shocks in the money supply or by risk premiums for example. These results are supported by Meese and Rogoff (1983) who stated that their Random Walk Model did not perform worse and even was more accurate than other structural models in estimating the exchange rate. Mussa (1979) found that structural monetary models could help in explaining the exchange rate, but that there will always be a part that is not explained by these models.

2.3 Productivity Based Model:

In section 2 several models seeking to explain the determination of the exchange rate were introduced. In all models, the money market played a central role. In the following model this will not be the case since it is based on productivity differences in the tradable and non-tradable sector in different countries. The tradable sector produces goods and services that can be traded in the international market. The non-tradable goods and services remain within the country. An example of a non-tradable service is a haircut at a local barber-shop (Pilbeam, 2006).

Balassa and Samuelson wrote individual papers in 1964 about productivity and relative prices. The basic ideas they wrote about are now better known as the Balassa-Samuelson model (Asea and Mendoza, 1994).

The Balassa-Samuelson model criticizes the idea of the Purchasing Power Parity. The most important critic is that a distinction should be made between tradable goods and non-tradable goods (Balassa, 1964). With this in mind, the model seeks to explain movements in the real exchange rate.

To construct the model, Egert et al. (2003) start with making a few assumptions.

First they assume a world with two countries and two different types of goods: tradable goods and non-tradable goods. To be able to say something about these goods, Egert et al (2003) create the following two production functions in which one shows the production function for tradable goods and the other the production function for non-tradable goods.

(16)

15 Supervisor: Christiaan van der Kwaak (2.)

In the first equation the production function for tradable goods is given and that’s why all these factors are marked with t. So equation 2, which represents the production function for non-tradable goods, is market with nt. Total factor productivity is given by A.

The second assumption is that prices of tradable goods are determined internationally so these prices are assumed to be given.

Within a country, labour is assumed to be perfectly mobile, which means that labour can move freely from the non-tradables sector to the tradables sector (Faria and Leon-Ledesma, 2003). The third assumption states that labour is only mobile within the country, which means that it is not able to be transferred to the other country.

It is assumed that wages in the tradables sector are determined by the given price level and the marginal productivity of labour. The last assumption to make is that wihtin each country, wages are equal for both sectors due to labour mobility within each country (Faria and Leon-Ledesma, 2003).

Now these assumptions are made, the first step in developing the model is to determine the marginal productivity of labour for both sectors using equations 1 and 2. Egert et al (2003) then compare these functions with a wage to price ratio:

(3.) ( ) (4.) (

)

Now rewriting these functions with the goal to isolate the prices for both goods yield:

(5.) ( ) for the tradables sector (6.) (

)

for the non-tradables sector

For this country, by dividing the price of non-tradables by the price of tradables, a ratio for relative prices is created:

(7.)

In this formula you can see that an increase in the productivity in the tradables sector increases the relative price of non-tradable goods to the price of tradable goods. Because prices of

(17)

16 Supervisor: Christiaan van der Kwaak tradable goods are determined internationally, the relative productivity increase in the

tradables sector does not necessarily imply an increase in prices. When prices in the tradables sector do not move, wages increase. Since wages are approximately equal for both sectors, this increase in wages in the tradable sector raises wages in the non-tradables sector too. In

equation 6 you can see that the price for non-tradable goods will increase as a result. Egert et al (2003) then write the marginal productivity of labour in the open and sheltered sector as average productivity, which looks as follows:

(8.)

(9.)

By using equation 7, 8 and 9, the relative prices of tradable goods and non-tradable goods of one country can be compared with another country. In the following formula this is done and the other country’s ratio is marked with an asterisk.

(10.)

In this formula one can see that when the productivity increases in the tradables sector of the first country, the non-tradables versus tradables ratio increases compared to the other country. Since the Balassa Samuelson model seeks to explain movements in the real exchange rate, we need an expression for it. Egert et al. (2003) use the following expression for the real exchange rate:

(11.)

In which S is the nominal exchange rate, which is expressed as units of domestic currency per units of foreign currency.

In the tradables sector Purchasing Power Parity holds (Balassa, 1964). This fact makes Egert et al (2003) able to create the following and final equation:

(12.)

In this equation the Balassa-Samuelson effect is noticeable. An increase in the marginal productivity in the tradables sector increases relative prices in the non-tradables sector. If an increase in marginal productivity occurs in one country and not or at a lower speed in the other,

(18)

17 Supervisor: Christiaan van der Kwaak the lower part of equation 12 will increase. This negatively affects the real exchange rate and this could be an explanation for the fact that quickly developing countries experience an

appreciating exchange rate relative to slower developing countries (Pilbeam, 2006). The impact of productivity growth on the real exchange rate is affected by the size of the relative sectors. When comparing the two countries, in the country with relatively the largest non-tradables sector, the impact of productivity growth on the real exchange rate will be the greatest (Egert et al., 2003).

(19)

CHAPTER 3: Exchange rate regimes Switzerland used since the Bretton

Woods system.

3.1 Introduction:

In the past, the Swiss National Bank has used several exchange rate regimes. I focus on the exchange rate regimes from the point where Switzerland is a member of the Bretton Woods system. Since exchange rate regimes and monetary policy are closely related to each other, some major decisions in monetary policy are discussed too in this chapter.

3.2 Switzerland in the Bretton Woods System:

The system of Bretton Woods was created to restore international monetary relations after the 1930s. In this war period, a major problem was wildly fluctuating exchange rates. Germany experienced hyperinflation and with the US stockmarket collapse in 1929 a period of worldwide recession started (Pilbeam, 2006). In an attempt to solve a country’s own problems, countries devalued their currencies which in turn increased the economic problems of others. This phenomenon is called ‘beggar thy neighbour’ policy. Finally, this period experienced exchange and trade restrictions (Bordo, 1993). In 1941 initial ideas were formed for an exchange rate system in an attempt to attack these problems and to create a stable monetary environment. The Bretton Woods system finally emerged at a monetary convention in Bretton Woods in 1944 (Pilbeam, 2006).

The ultimate objectives of the Bretton Woods system were: stable exchange rates, national full employment and cooperation. To establish these three goals, the system created a fixed but adjustable exchange rate regime, a union that oversees the functioning of the system and a credit facility that provides assets through which payment deficits could be soothed (Williamson, 1985).

In the fixed but adjustable exchange rate regime currencies of the system’s countries were pegged against the US dollar and currencies were allowed to fluctuate at 1% in either side. The US Treasury pegged the US dollar itself to the price of gold at $35 per ounce (Pilbeam, 2006).

To promote world trade and cooperation and to oversee the workings of the Bretton Woods system, the International Monetary Fund was established. The main goal of the fund was to smooth the workings of the Bretton Woods system. To do this, the fund aimed at minimizing the need for countries to change their exchange rate by following them closely. In case of temporary imbalances, the fund provided countries running deficits with credit to keep these imbalances just temporary. By following their three objectives closely and by using the credit

(20)

19 Supervisor: Christiaan van der Kwaak facility in case of trade imbalances, the IMF created a stable environment for world trade, national full employment and cooperation (Pilbeam, 2006).

Switzerland was involved in the Bretton Woods system and their Swiss franc was valued at a rate of 4.30 CHF/$. The most important fact of the Bretton Woods system for the Swiss franc is that the exchange rate regime the Swiss National Bank followed at that moment was one with fixed but adjustable exchange rates. In graph 1, one can clearly see that the exchange rate of the Swiss Franc to the US dollar remains stable until about 1970. From that point, the Swiss National Bank uses a floating exchange rate regime (SNB, 2014).

Graph 1 Source: www.snb.ch (Swiss National Bank data)

3.3 Floating Exchange Rate Regime:

After the collapse of the Bretton Woods system, exchange rates were allowed to float. It gave the SNB the chance to alter its monetary policy regime. In the years preceding the collapse of the Bretton Woods system, the whole world experienced accelerated inflation. Switzerland decided, just as Germany did, to focus on inflation control with a regime of monetary targeting

(Baltensperger, 2007). Since monetary policy regimes and its reaction to situations can sometimes indicate why exchange rates move in a certain direction, I’ll briefly discuss the monetary target regimes and some situations in chronological order.

3.3.1 Monetary Target Regime:

When a central bank uses monetary targeting as a monetary policy regime, it announces an annual growth rate of the monetary aggregate at first. Then, after this announcement, the central bank is accountable for reaching the target (Guiliodori et al., 2013). One major advantage of monetary targeting is that figures and stats about monetary aggregates are

0 1 2 3 4 5 19 43 10 19 47 05 19 50 12 19 54 07 19 58 02 19 61 09 19 65 04 19 68 11 19 72 06 19 76 01 19 79 08 19 83 03 19 86 10 19 90 05 19 93 12 19 97 07 20 01 02 20 04 09 20 08 04 20 11 11 e xch an ge r ate CH F/$

Swiss franc/US dollar

1943-2013

(21)

20 Supervisor: Christiaan van der Kwaak reported very frequently. These frequent reports leave less room for speculation and

expectations about future inflation rates and as a consequence it lowers the inflation. This advantage of monetary targeting is dependent on two very important factors. Firstly, the relationship between the target variable and the goal variable must be strong and reliable. In this case the goal variable is inflation, because the central bank strives to control this inflation, and the target variable is the monetary aggregate (Mishkin, 1999). If the relation is weak or unreliable, monetary targeting will not have the desired outcome. Secondly, the targeted monetary aggregate must be controlled well by the central bank. Since transparency, reliability and accountability are such important factors in the success of monetary targeting, the control over the values policymakers report to the public must be well controlled. Otherwise

policymakers cannot provide clear signals and information to the public which makes them less accountable (Mishkin, 1999).

There are several countries that failed in adopting monetary targeting, but Germany and Switzerland can be seen as two important countries that had success by using it. Both used the monetary target regime to inform the public about their monetary policy and its long run goals. All important factors as transparency, reliability and accountability were fulfilled. Their success is a reason for the European Central Bank to incorporate some of these ideas in their official policy regime (Guiliodori et al., 2013).

3.3.2 Monetary Target Regime from 1970 until now:

A still high inflationary environment in countries surrounding Switzerland, in France and England for example, led in the late 1970s to a massive capital inflow which in turn caused the Swiss franc to appreciate. In graph 2, where the Swiss franc is computed against the German mark, you can see an appreciation starting in 1977 against the German mark too. In September 1978, the exchange rate of the Swiss franc against the German mark is about 0.79 CHF/DM (Bundesbank, 2014). This exchange rate made the Swiss National Bank decide to switch from a monetary target regime to an exchange rate target regime because the strong Swiss franc threatened the Swiss real economy. The lower limit of its exchange rate against the German mark should ensure the ease of trade with Germany which was the most important trading partner of Switzerland (Baltensperger, 2007). The Swiss National Bank set a lower bound for its exchange rate against the German mark at 80 Swiss franc per 100 German mark (SNB, 2014). After these changes, the exchange rate returned to more natural levels which made the Swiss National Bank to decide to return to the monetary targeting regime. Initially after returning to the monetary targeting regime, Switzerland experienced a heavy inflation. This inflation was due to the expansionary monetary policy of preceding years and the second oil shock. The

(22)

21 Supervisor: Christiaan van der Kwaak strong Swiss franc in the seventies and its threat to the economy made the Swiss National Bank known of the importance of the exchange rate (Baltensperger, 2007).

Graph 2 Source: www.bundesbank.de (Bundesbank data)

After the recovery of the Swiss franc, it remained quite stable for a great part of the 1980s, but hard times began in the late 1980s. The uncertainty in financial markets and the tendency of the Swiss franc to appreciate in these uncertain times made the Swiss National Bank decide to increase the money supply by a greater amount than was stated in the target. In graph 3 you can see an increasing monetary base(in billions of Swiss francs) from 1985 until the end of 1987. The stock market crash in October 1987 made the Swiss National Bank decide to increase the money supply even more (Baltensperger, 2007).

0.7 0.8 0.9 1 1.1 1.2 1.3 19 73-01 19 73-05 19 73-09 19 74-01 19 74-05 19 74 -09 19 75-01 19 75-05 19 75-09 19 76-01 19 76-05 19 76 -09 19 77-01 19 77-05 19 77-09 19 78-01 19 78-05 19 78-09 19 79-01 19 79-05 19 79-09 19 80-01 19 80-05 19 80-09 e xch an ge r ate CH F/G M

Swiss Franc/German Mark for period

1973-1980

(23)

22 Supervisor: Christiaan van der Kwaak Graph 3

Source: www.snb.ch (Swiss National Bank data)

In the fear of a worldwide economic downturn, the Swiss National Bank set the target for the monetary growth in 1988 on 3%, which was higher than the normal target. After noticing that the expected and feared economic downturn did not materialise, the demand for money seemed to be lower than the supply of money. Two other factors also lowered the demand for money: the introduction of the Swiss Interbank Clearing and a change in the liquidity requirements for banks. These factors increased the imbalances between the supply and demand for money which made the Swiss National bank decide to further tighten its monetary policy in an attempt to equilibrate the money supply and the money demand (Baltensperger, 2007). In graph3you can clearly see that the Swiss National Bank decreases the monetary base from the start of 1988. High inflation rates in the late 1980s and begin 1990s made clear that the monetary policy was still too expansionary. The problems with conducting monetary policy in the late 1980s, made the Swiss National Bank decide to relax the current one-year monetary growth target because it did not have the desired effects (Baltensperger, 2007). It replaced monetary targeting regime with a one-year path by a medium-term path of about three to five years for monetary growth. This new approach should be more consistent with the price stability goal of the Swiss National Bank (Mishkin, 1999; Rich, 2003).

The Swiss National Bank and its renewed monetary policy got really tested in the beginning of the 1990s. In the starting years of the 1990s, Switzerland experienced a recession in which uncertainty and monetary policy by the Swiss National bank played a role. The beginning of the 1990s can be signed by two very important European experiences. The first one was the reunification of Germany and the second one was the establishment of a new European Monetary Union in the Maastricht treaty in which the euro was introduced as a

27 29 31 33 35 37 19 85 01 19 85 06 19 85 11 19 86 04 19 86 09 19 87 02 19 87 07 19 87 12 19 88 05 19 88 10 19 89 03 19 89 08 19 90 01 19 90 06 19 90 11 19 91 04 19 91 09 M o n e tar y B ase

Monetary Base in billions of Swiss

francs 1985-1991

(24)

23 Supervisor: Christiaan van der Kwaak currency (Rich, 2003, Pilbeam, 2006). These two events together with a Swiss economy

experiencing a recession created some uncertainty which made conducting monetary policy a challenging task. The fear of inflation made the Swiss National Bank decide to not lower the interest rate although their monetary indicator signalled a need for a lower interest rate. This action delayed the recovery of the Swiss economy. As a result unemployment started to rise which continued until after 1996 (Rich, 2003). Finally after 1996, the Swiss economy returned to normality (Baltenspercher, 2007). Important variables as interest rates, inflation and exchange rates returned to their long run levels. The boom in the European and Worldwide economy caused economic growth and unemployment rates to reach their long run levels at the end of the 1990s too. The events of the previous decade and the struggles in conducting

monetary policy using the current regime, made the Swiss National Bank decide to change the monetary target regime to a policy concept based on inflation forecasts (Baltensperger, 2007; Rich, 2003).

The new inflation based monetary policy regime continues to have the same ultimate goal as the previous one, being price stability. This is the first element of monetary

policymaking from the Swiss National Bank. It strives to a yearly increase in CPI(Consumer Price Index) of less than 2% which is quite similar to the goal of the European Central Bank (Guiliodori et al., 2013). The second important element is a conditional inflation forecast for the upcoming three years. This forecast serves as a tool to communicate economic decisions with the public (SNB, 2013). The last element in monetary policymaking for the Swiss National Bank is setting a three month LIBOR rate. This rate is an important reference because it shows the rate at which banks can borrow and lend from each other. The Swiss National Bank sets a target for its LIBOR rate every quarter (SNB, 2013).

The new concept of monetary policymaking got tested in the years after

implementation. The terrorist attacks on September 11th_{in 2001 caused stock markets}

worldwide to decline (Chen and Siems, 2003). These events made the Swiss National Bank decide to cut LIBOR rates and continue this loosened monetary policy until 2004. The inflation rate remained quite stable for the whole period. Since price stability was one of the most important elements in monetary policy, the new concept of monetary policy has done well for this period (Baltensperger, 2007).

The period that follows is the period we are in right now, experiencing a financial crisis that started in 2007-2008. Graph 4shows the movements of the Swiss franc against the euro from 2006 until now. Since the beginning of the crisis, the Swiss franc appreciated against the euro, which can be explained by the fact that the Swiss franc is seen as a ‘Safe Haven Currency’

(25)

24 Supervisor: Christiaan van der Kwaak (Jordan, 2009). With investors getting more risk averse over time, in these turbulent economic times, they are attracted to the safe character of the Swiss Franc. The appreciating franc and the fact that interest rates and rates of inflation approached zero, feared the Swiss National Bank for decreasing exports and national income. As a result, it decided in September 2011, to set a lower bound for the Swiss franc against the euro at 1,20 CHF/€ (SNB, 2011). Since the

announcement, the Swiss National Bank uses a floating exchange rate with a lower limit. When the exchange rate approaches this level it will take actions to prevent the exchange rate from crossing the lower bound. Since September 2011, this level has not been crossed anymore.

3.4 Monetary policy consequences of the temporary peg to the euro:

Even though the Swiss National Bank kept the interest rates at a low level in 2011, the desired capital outflow and the resulting depreciation did not materialise. Concerns about the financial stability of the euro-area attracted investors to continue placing capital in Switzerland. The severe appreciation of the Swiss franc in 2011, made the Swiss National Bank decide to set the lower limit for the Swiss franc to the euro at 1,20 CHF/€. This is still a high value for the Swiss franc, but it should lose value over time (SNB, 2011).

By setting the lower limit on the Swiss franc against the euro, the Swiss National Bank basically pegs its currency to the euro. Philipp Hildebrand, chairman of the Swiss National Bank, said that a fixed and permanent peg would not be compatible with the constitutional and legal mandate of the Swiss National Bank to conduct an independent monetary and exchange rate policy (Wille and Verschuur, 2011). But although the peg is supposed to be temporary, there are consequences for conducting monetary policy by the Swiss National Bank.

1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 20 06 01 20 06 06 20 06 11 20 07 04 20 07 09 20 08 02 20 08 07 20 08 12 20 09 05 20 09 10 20 10 03 20 10 08 20 11 01 20 11 06 20 11 11 20 12 04 20 12 09 20 13 02 20 13 07 CH F/ €

Exchange Rate Swiss franc against the

euro for 2006-2013

(26)

25 Supervisor: Christiaan van der Kwaak When conducting monetary policy, National Banks face the ‘macroeconomic trilemma’ in which they have to choose two out of three objectives. Although, all three objectives can be desirable for open economies, only two of them can be achieved simultaneously. Pilbeam (2006) writes this ‘impossible trinity’ as follows:

1. Monetary independence 2. Fixed exchange rate

3. Free movement of international capital

The ‘impossible trinity’ is often displayed in a triangle form in which the policy maker can only choose one side of the triangle:

Fixed exchange rate

Monetary independence Free movement of international capital

Figure 1: The macroeconomic trilemma (Pilbeam, 2006)

By looking at the Mundell Fleming model for a small open economy facing free movement of capital, it becomes clear that the three objectives cannot be achieved

simultaneously. When a small country using fixed exchange rates seeks to affect output by a monetary expansion, interest rates drop at first. This leads to a capital outflow which, in case of flexible exchange rates, should result in a depreciation of the exchange rate. In case of fixed exchange rates, monetary authorities have to intervene in the foreign exchange market to keep the exchange rate fixed. To do this, authorities purchase home currency by using reserves which decreases the money supply. This intervention brings the economy back into the old

equilibrium. The goal of increasing output by a monetary expansion is not reached which shows that with free movement of capital, authorities have to give up a fixed exchange rate or

monetary independence (Pilbeam, 2006). interest rate

output

Figure 2: Mundell Fleming model showing the impossible trinity (Pilbeam, 2006) LM1

LM2

IS

The monetary expansion shifts the LM1 curve to LM2. Interest rates drop and capital flows out of the country. When using fixed exchange rates, authorities have to purchase home currency in the foreign exchange market which makes the LM curve shift back to LM1.

(27)

26 Supervisor: Christiaan van der Kwaak In Europe, and hence in the case of Switzerland, there are no capital controls which means that the movement of capital is free. In their mandate, the Swiss National Banks states that it strives to work as an independent central bank that conducts monetary policy

independently (SNB, 2014). As we have seen, the Swiss franc is normally a floating exchange rate but in the case of the temporary peg to the euro in September 2011, the Swiss National Bank has fixed its exchange rate to the euro. By looking at the triangle in figure 1, the Swiss National Bank had to give up its objective of monetary independence since it could only choose two out of the three objectives (Pilbeam, 2006). The peg obligates the Swiss National Bank to intervene in the foreign exchange market to ensure that the exchange rate does not cross the limit of 1.20 Swiss franc per euro (Wille and Verschuur, 2011). The Swiss National Bank announced that it was prepared to buy an unlimited amount of foreign currency reserves. By looking at graph 5, you can see the severe increase in the amount of currency reserves held by the Swiss National Bank which shows that it has bought foreign currency in severe amounts since the announcement in 2011 (Ross, 2012).

In its Quarterly Bulletin published in December 2013, the Swiss National Bank announced that it maintains the minimum exchange rate of 1,20 Swiss franc per euro. By looking at the exchange rates from May 2011 until the end of 2013 in graph 6, you can see movements of the Swiss franc against the euro. Shortly after the announcement in September 2011, the Swiss franc makes a jump against the euro and from the second quarter of 2012 you can see a small depreciation until the end of 2013. This value of the Swiss franc is still

considered strong by the Swiss National Bank, but it should depreciate over time (SNB b, 2013). 0 50000 100000 150000 200000 250000 300000 350000 400000 450000 2004 2005 2006 2007 2008 2009 2010 2011 2012

Foreign Currency Reserves in billions

of Swiss francs

(28)

27 Supervisor: Christiaan van der Kwaak Graph 6

Source: www.snb.ch (Swiss National Bank data) 1.05 1.1 1.15 1.2 1.25 1.3 20 11 05 20 11 07 20 11 09 20 11 11 20 12 01 20 12 03 20 12 05 20 12 07 20 12 09 20 12 11 20 13 01 20 13 03 20 13 05 20 13 07 20 13 09 CH F/ €

(29)

CHAPTER 4: Model Performance

4.1 Introduction:

When looking at the drivers of the fluctuations in the Swiss franc, most people easily expect models to explain all the fluctuations. But these models cannot always explain every movement in the exchange rate. Since the breakdown of the Bretton Woods system, research on exchange rates became very popular and many economists attempted to establish models explaining the movements of exchange rates. Many of them came to the conclusion that these models didn’t perform well, but several economists found evidence for their models to explain movements in the exchange rate.

To explain movements in the Swiss franc, I’ll first discuss the results and the critics to the models I’ve used in chapter 2. In the following section I’ll explain some special features that are ignored in most models, but could matter for determining the fluctuations in the Swiss franc.

4.2 Function and malfunction of Exchange Rate Determination Models:

4.2.1 Evidence on monetary models:

In 1983, Meese and Rogoff discussed the functioning of several exchange rate determination models. These authors used several models to test if they could beat the random walk forecast of the exchange rate. The models Meese and Rogoff used, included the flexible-price monetary model and the Dornbusch sticky-price monetary model for instance. To test the models’ capability in predicting the exchange rate compared to the random walk forecast, the authors created the random walk model in which they assumed the spot exchange rate to be

unpredictable. This random walk model was evaluated against models as the sticky-price monetary model and the Dornbusch model (Meese and Rogoff, 1983). The conclusion of their work was that the monetary models, both sticky-price and flexible-price, could not beat the random walk model. This conclusion was supported by many economists. Mussa (1979) already said that the natural logarithm of the spot exchange rate approximately followed a random walk. Backus (1984) did not find any significant evidence for both monetary models to beat the random walk too. In his research he suggested to incorporate oil prices into the models since in his time of investigating the exchange rate of Canada, the Canadian economy was attacked by oil price changes (Backus, 1984). In a research by Robert Driskill, he attempted to test the

Dornbusch sticky-price model using US-Swiss data. In his research, he concluded that the Dornbusch model on itself performed poorly, but he did find evidence for the overshooting part to exist (Driskill, 1981).

(30)

29 Supervisor: Christiaan van der Kwaak But not all researchers experienced the exact same results. Especially in more recent empirical research, some economists found models and methods that did beat the random walk forecast. Mark (1995) and later Faust et al. (2003) used more or less the same method in which their model was quite similar to the flexible-price monetary model. Their conclusion was that there is enough evidence to believe that in the long run exchange rates do have a significant predictable component. In the short run, this method could not outperform the random walk (Mark, 1995). Lutz and Taylor (2003) came to the same conclusion in their work since they found strong evidence for the predictability of exchange rates at periods of two or three years, but not for shorter periods.

Rapach and Wohar (2004) ran a research in which they first investigated the ability of a monetary model in predicting the spot exchange rate using cointegration tests on a country by country base. In their first research they could not find a clear link between the monetary model used and exchange rate movements and the power of the test they ran was very low. In their second research they used a panel cointegration test to indicate if relative money supplies and relative output levels affect the exchange rate. To run this kind of test, data from several countries were pooled, which increased the power of their test significantly. With this second test they showed that there is much more support for monetary models by using pooled data than by using country by country data. For these results to hold, the authors needed to make an additional assumption concerning the homogeneity of the coefficients, which made some economists felt doubt about the results because the data did not support the homogeneity assumption. Their ultimate conclusion is that the support for monetary models using these panel data is inaccurate (Rapach and Wohar, 2004). Cheung et al. (2005) tested five models against the random walk. In their research, some models did fit some situations but they could not explain other situations. That is why they concluded that ‘perhaps’ models do well for one occasion because they fulfil the exact criteria, but perform poorly for another situation (Cheung et al., 2005).

What these results show is that by using different methods of testing, different

conclusions can be derived. The results inspired Engel and West (2005) to leave some space in their models for the random walk. A part of the exchange rate could be explained by the fundamentals, but there could be a part that is explained by a random walk (Engel and West, 2005).

4.2.2 Empirical evidence on Productivity Based Models:

Along with economists testing the monetary models, there have been several economists investigating the capabilities of the Balassa Samuelson model. Kenneth Rogoff found significant evidence for the Balassa Samuelson to exist in its work for the yen to dollar rate, but not for all

(31)

30 Supervisor: Christiaan van der Kwaak the rich countries as a group or the poorer countries as a group. He may have found evidence for Japan because it experienced a rapid growth of the income per capita. The subsequent rises in productivities could explain the appreciation in the yen against the dollar (Rogoff, 1996). In their work, Asea and Mendoza (1994), found some evidence that long run relative prices are affected by productivity differences. But these long run relative prices could not explain the movements in the real exchange rate (Asea and Mendoza, 1994). De Gregorio et al. (1994) concluded more or less the same in their research. Relative productivities do affect relative prices, but they cannot explain the total movements of the real exchange rate.

Stephen DeLoach was inspired by the work of Asea and Mendoza (1994) and De Gregorio et al. (1994), but he made some changes to their methods. Firstly, he introduced oil prices, because these prices affect the relative prices he wanted to research. Secondly, he used far more observations than Asea and Mendoza and De Gregorio et al. (DeLoach, 2001). One of the outcomes of his work was a significant long-run relationship between relative prices and output. The other important outcome was that real supply shocks and especially real oil prices should be considered and that there may be other factors that need to be included when determining relative prices (DeLoach, 2001).

The models do provide some information about exchange rates, but the fit of a model depends on many different factors and as DeLoach (2001) suggested in his research, there might be factors that are omitted in creating the model.

4.3 Other important factors in determining the exchange rate:

4.3.1 Switzerland’s safe haven currency status:

Models cannot always describe movements in exchange rates. Cheung et al. (2005) concluded that models perform well in some occasions but performed poorly in others. Since different periods in time involve a wide range of situations, modelling can sometimes be extremely difficult. Times of crisis for instance can change the financial situations in countries

substantially. In these times of uncertainty, investors become more risk averse and search for safe investing opportunities. Switzerland and its Swiss franc are historically seen as being safe, according to its safe economy and its safe currency (Melvin and Taylor, 2009; Habib and Stracca, 2012). In times of uncertainty, these characteristics cause the Swiss franc to make movements some people wouldn’t predict (Kohler, 2010).

The Swiss franc can be seen as a safe haven currency, which is a currency that could serve as a hedge to a portfolio of risky assets, conditional on changes in the risk aversion of investors (Habib and Stracca, 2012). Times of financial crisis, the terrorist attacks on 9/11 or for instance an earthquake can be factors causing investors to become more risk averse. They move

(32)

31 Supervisor: Christiaan van der Kwaak their investments from risky assets to safer assets. Safe haven currencies become more popular in these kinds of moments and will appreciate as a result. People may expect exchange rates to depreciate because of the increasing uncertainty, but these currencies appreciate. Ranaldo and Söderlind (2010) show this movement in exchange rates with a reaction to the terrorist attack on 9/11. Swiss franc appreciated for about 3% to the US dollar just after the first airplane attacked the World Trade Centre (Ranaldo and Söderlind, 2010). In their research, they established a model in which ordinary exchange rate modelling factors and factors measuring volatility and liquidity were incorporated. In their research, the Swiss franc fulfils the following criteria for being a safe haven currency: The currency appreciates through (1) negative stock market returns, (2) increases in bond prices, (3) currency markets becoming more volatile. Negative stock market returns increase demand for stable and safe assets, so the demand for safe haven currencies increases. Bonds can be seen as relatively safe assets, so increases in demand for bonds and bond prices occurs as well as increases in the demand for safe haven currencies. Increasing volatility in currency markets increases the risk of most currencies substantially; as a result, the demand for relatively safer currencies increases. All the criteria demonstrate the increasing popularity of the safe haven currency in times of uncertainty and where other currencies depreciate as a result of uncertainty, these currencies experience an appreciation (Ranaldo and Söderlind, 2010).

Kohler (2010) recognizes the same effect but he adds the importance of interest rate differences. High interest rate currencies experience depreciation in times of financial uncertainty and lower interest rate currencies, as Switzerland, experience appreciation. The construction of carry trade, where a long position in a high yielding currency is hedged by a short position in a lower yielding currency, displays what happens when investors become more risk averse. Their capital invested in the high yielding and risky currency is moved to the lower yielding and safe currency. This makes the high yielding currency to depreciate and the lower yielding currency to appreciate (Kohler, 2010). In addition to this, safe haven currencies are likely to be supported by a stable economy and by a developed and liquid financial market. These factors are very important for risk averse investors, because it increases confidence by investors (Habib and Stracca, 2012).

According to Ranaldo and Söderlind (2010) the Swiss franc fulfils the criteria for being a safe haven currency. The Swiss franc appreciates through negative stock market returns, increasing demand for safer assets and an increase in uncertainty in currency markets. Times of crises mostly go hand in hand with these criteria and increase the uncertainty and risk

averseness of invertors. The increase in risk averseness increases the demand for safe haven currencies which as a result appreciates the safe haven currency. So when determining the

(33)

32 Supervisor: Christiaan van der Kwaak movements in the Swiss franc, it is important to study investors’ beliefs and their risk

averseness because these factors may be of importance in determining the exchange rate. 4.3.2 The impact of oil prices on exchange rates:

As we have seen in section 4.1, economists have doubts about the power of exchange rate determination models. DeLoach (2010) and Backus (1984) already introduced that there could be a role for oil price shocks in these models and several other economists agree with this point of view. Amano and Norden (1998) argued that changes in real oil prices alter the terms of trade. These changes could be of importance explaining movements in exchange rates.

In their research, Amano and Norden (1998) consider oil price shocks to be exogenous. Examining the oil price shocks that occurred in the 1970s, 1980s and begin 1990s, the authors explain the oil price movements by political conflicts occurring in the Middle-East.

Macroeconomic models cannot explain these conflicts, which gives the authors a plausible reason to consider oil price shocks as being exogenous (Amano and Norden, 1998). To investigate the impact of oil price shocks on the exchange rates, they used cointegration tests for the effect of oil prices on the effective real exchange rate. Their first observation was that oil prices explain a reasonable part of the terms of trade and their main conclusion was that large and persistent changes in the price of oil can explain movements in the effective real exchange rate in the long run (Amano and Norden, 1998).

Chen and Chen (2007) ran the same research as Amano and Norden (1998) but they took it a little further. In their first section, using panel cointegration test they examined the effect of oil prices on the real exchange rate for G7 countries. This is similar to the research done by Amano and Norden (1998) and they found similar results. In their second section they investigated the power of oil prices to forecast future movements in the exchange rate. They state that a rise in oil prices deteriorates the terms of trade, which leads to a depreciation in the long run. From their regression they conclude that there is strong evidence to assume that real oil prices help predicting future real exchange rates, and thus play an important role in the determination of exchange rates (Chen and Chen, 2007).

To examine the effects of oil price shocks on a small open economy, Atukeren (2003) ran a research on the impact of oil prices on Switzerland, as being a country without oil resources. A large increase in oil prices is often a consequence of war or other factors increasing uncertainty. Thereby, an increase in the price of oil, increases prices of exportable goods as a result. Both factors, an increase in price level and uncertainty, negatively affect the export sector and economic growth (Atukeren, 2003). To rebalance the export sector, currencies of oil importing countries like Switzerland should depreciate when oil prices rise (Lizardo and Mollick, 2010).