The safe haven status of the Swiss franc
Can the Swiss franc still be considered a safe haven currency after the announcements by the Swiss National Bank to adjust its monetary policy? Bachelor thesis Economics and Finance, 12 EC Author: Vivienne Groenewegen UvAnetID: 11034963 Supervisor: ms. Rui Zhuo June 26, 2018Abstract
This paper examines whether the safe haven status of the Swiss franc has altered after a change in monetary policy announced by the Swiss National Bank (SNB) on 03-2009. An empirical analysis using an OLS regression follows, where the variables used to describe the safe haven status are S&P500, Treasury notes and market risk. A VIF test is analyses multicollinearity. Two time periods are compared to each other, namely the period prior and sequent the SNB’s announcement. Besides the regression, fundamentals of a safe haven currency have been considered. The findings of this thesis are ambiguous.2
Statement of originality
This document is written by Student Vivienne Groenewegen who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.3
Table of Contents
1. Introduction ... 4 2. Literature review ... 5 2.1 Safe haven ... 5 2.2 Risk ... 6 2.3 Risk hedging and the Uncovered Interest Parity ... 7 2.4 Currency appreciation ... 8 3. Methodology and data ... 9 3.1 Methodology ... 9 3.2 Assumptions ... 10 3.3 Hypothesis ... 12 3.4 Data ... 13 4. Results and empirical data analysis ... 13 5. Discussion ... 18 6. Conclusion ... 19 Reference list ... 20 Appendix ... 22
4
1. Introduction
According to Nelson and Katzenstein, we live in a world with constant risk and uncertainty and the global financial crisis starting in 2007 reminded us of this once again (2014). Due to the crisis, people became more risk averse (Barrios, Iversen, 2009). They are seeking for less risky investments and a ‘strong’ currency. The financial crisis is not the only period in which signs are shown of increased risk aversion. For example, the attack on the twin towers at 9/11 was another event in which financial uncertainty increased. This event led people to search for a safe currency to invest their money in. One of these currencies was the Swiss franc. In figure 1, a remarkable appreciation of the Swiss franc within an hour of the first plane crash can be observed. The appreciation is exceeding the appreciation of other currencies. The reason for this strong appreciation of the Swiss franc is its safe haven status (e.g. Botman, Carvalho Filho, & Waikei, 2013 and Ranaldo and Söderlind, 2010). Already from WWI onwards, the Swiss currency shows signs of an increasing exchange rate (Latsos & Schnabl, 2008).
Andréa M. Maechler, a member of the governing board of the Swiss National Bank (SNB), explained in a speech in November 2017 that the rapid appreciation of the Swiss franc due to the financial crisis had created a risk for the Swiss franc of deflating. This was due to the drop in price level. In her speech she told that the SNB therefore wants to prevent the Swiss franc from any further appreciation. To do so, the SNB announced and started as of March 2009 to buy foreign currency in order to decrease the demand for Swiss francs.
Tachibana writes in his paper published in 2018 that the Swiss franc is still one of the biggest hedging currencies (2018). However, the SNB did adjust its monetary policy sharply as of 2009. This raises the following research question:
Can the Swiss franc still be considered a safe haven currency after the announcements by the Swiss National Bank to adjust its monetary policy?
Kaul and Sapp show that the U.S dollar used to be a safe haven during the millennium time, but that it no longer is (2006). This confirms that the status of safe haven currency can be altered.
To find an answer to the research question, an empirical research is performed. This research makes use of a time series regression. Also the fundamentals of a safe haven are reviewed. The main finding of this paper is that the results are ambiguous. It does not become clear from this research what the current status of the Swiss franc is.
5
This paper is structured as follows. It starts with a literature review in section 2, followed by a methodology in section 3. Section 4 provides the results. Based on these results, section 5 will provide a discussion. The paper ends with a conclusion in section 6.
FIGURE 1: 9/11 exchange rate appreciations.
In the hours prior to the attack, the Swiss Franc (CHF) had an exchange rate against the U.S. equal the Euro (EUR) and the British Pound (GBP). In the hours after the attack, CHF appreciated significantly, exceeding the exchange rates of the Euro, British Pound and Japanese Yen. Source: Ranaldo & Söderlind, 2010.
2. Literature review
2.1 Safe haven A safe haven asset is defined as an asset that provides a hedge against risk (Ranaldo and Söderlind, 2010). Its term originates from times of stormy weather on sea, where boats and ships had to flee to a harbour to take shelter of the storm. A safe haven asset therefore is an asset that holds its value in times of bad weather (Baur, & Mcdermott, 2010). In the context of this paper, it is an asset that does not lose value in times of increased financial risk. Therefore, the demand for safe haven currencies increases in risky times, as investors are assumed to be risk averse. Ronaldo and Söderlind explain that in times of financial uncertainty and increased exposure to market risk, a safe haven currency will appreciate (2012).Safe haven assets include hard assets, of which the most well-known example is gold. During the financial crisis, the price for gold increased by 42% (Baur, & Mcdermott, 2010). Besides hard assets, currencies can be safe haven assets. There have been many studies performed on the safe status of the
6
Japanese currency the yen. The Japanese yen is sometimes referred to as a safe haven, however not as convincing as the Swiss francs’ safe haven status. The reason for this trust in the Swiss currency is amongst others their stable government and financial system (Zurbrügg, 2012).
Campbell, Serfaty-de Medeiros and Viceira explain that the Swiss franc is negatively correlated to market returns, both the international and their domestic stock returns (2009). Currencies not known as safe haven currencies are positively related to market returns. The Australian dollar and the Canadian dollar are examples of non-safe haven currencies and Campbell, Serfaty-de Medeiros and Viceira find that these two currencies are indeed positively related to market returns (2009).
Habib and Stracca test three fundamentals of a country for being a safe haven currency. First of all, the country should be safe and have a low interest rate (2012). The safe status of a country, although this is a broad term, will attract investment in times of increased risk aversion. The low interest rate reflects the absence of risk.
The second explanation by Habib and Stracca is that size and liquidity of financial markets matter (2012). During financial stress and increased risk aversion, the chances of a liquidity crisis to occur increase. Therefore, increased liquidity represents more safety.
The third explanation Habib and Stracca test is the openness of a country towards global investments. They find that financially open countries are more sensitive for global turbulence. Safe haven countries should therefore be isolated from global risk and its corresponding turbulence (Habib & Stracca, 2012). However, perfect capital mobility within countries is necessary.
These three fundamentals coincide with the description of Botman, Carvalho Filho and Lam, who explain that the three conditions of a safe haven currency are low interest rates, a strong net foreign asset position and a highly liquid financial market (2013). A strong net foreign asset position is positively linked to safe haven currencies (Botman, Carvalho Filho & Lam, 2013). During periods of increased financial and/or global risk, countries with safe haven characteristics were the countries with a positive net foreign asset position (Habib & Stracca, 2012). Countries with a strong net external position in times of risk tend to have appreciating currencies.
2.2 Risk
Risk plays an important role in the definition of a safe haven. A safe haven currency is being shaped by market risk. If market risk increases, normal currencies tend to depreciate. In contrast, the Swiss franc as a safe haven currency will appreciate.
7
Risk therefore drives the exchange rates movements. However, there are multiple ways to define market risk. Ronaldo and Söderlind discuss two types of market risk: volatility risk and liquidity risk (2012).
Volatility risk concerns market volatility. Market volatility is defined by the volatility on market returns. The bigger this volatility is, the higher the level of uncertainty on the returns of investments. The VIX is a volatility index measuring this specific type of risk spread (Brunnermeier, Nagel, & Pedersen, 2008).
The second type of market risk is liquidity risk. The term liquidity means the ability to trade on a quick notice without an adjustment of the price (Pastor & Stambaugh, 2003). When an asset is more sensitive to aggregate liquidity, a higher liquidity risk premium is required. Liquidity risk is measured using a TED spread. The TED spread is the difference between the London Interbank Offered Rate (LIBOR), being the rate at which banks themselves can borrow and lend to each other, and the risk-free interest rate (Brunnermeier, Nagel, & Pedersen, 2008). Bandi, Moise and Russell explain that volatility risk and liquidity risk are two systematic risk factors that are priced in in the return of a stock (2008). These two types of risks are negatively correlated to returns. 2.3 Risk hedging and the Uncovered Interest Parity One reason to invest in a foreign currency in times of market turmoil is to hedge risk. For example, if the rate of inflation in a domestic country is uncertain, investing abroad can hedge inflationary risk and therefore hedge currency risk (Glen, & Jorion, 1993). A second reason may be the presence of a correlation between an equity investment and a currency. If the domestic currency is positively related to equity returns, investing in a foreign currency gives the optimal currency position (Campbell, Serfaty-de Medeiros, &Viceira, 2009). If a currency is optimally hedged, it will reduce risks for equity investors (Glen, & Jorion, 1993).
This risk, which leads to the demand for hedging, comes from the disequilibrium of the Uncovered Interest Parity (UIP). This disequilibrium and its associated risk in turn generate investors to require a risk premium (Campbell, Serfaty-de Medeiros, & Viceira, 2009). To show the relation between a risk premium and disequilibrium of UIP, the term carry trade is relevant.
Carry trade describes that if interest rates among countries differ, investors will sell the currency from a low interest rate country and buy the currency of high interest rate countries. (Brunnermeier, Nagel, & Pedersen, 2008). If the UIP would be in equilibrium, the difference in interest rates across countries is eliminated by exchange rates (Menkhoff, Sarno, Schmeling, & Schrimpf, 2012). The country with the high interest rate, say this is the foreign
8
country, will depreciate until the UIP equilibrium is restored (Brunnermeier, Nagel, & Pedersen, 2008).
In reality however, the opposite is observed. There exists a systematic deviation from the UIP, which is often referred to as the “forward premium puzzle” (Menkhoff, Sarno, Schmeling, & Schrimpf, 2012). Due to this deviation from the UIP equilibrium, carry trade is profitable. The Swiss franc exhibits the same deviation, as it appears to appreciate when invested in.
The reason for this systematic deviation is a false assumption of the UIP model, which states that currencies are equally risky. In reality currencies do not equal in riskiness. Therefore a risk premium should be added to compensate for the inequality of currency risks. People become more risk averse in times of financial stress and therefore require a higher risk premium (Verdelhan, 2010) to make up for the extra risk of an investment. Solnik explains that the risk premium of a security on top of the domestic risk free interest rate is proportional to the foreign market risk (1974).
2.4 Currency appreciation
When a country experiences low interest rates and low inflation, which is typically the case for safe haven currency countries, an appreciation of their currency due to the hedging of financial risk could feed a risk of deflation for this country. This happened in Switzerland. Deflation in turn provides a downward pressure on aggregate demand (Botman, Carvalho Filho, & Lam, 2013). This will harm the country’s GDP and economic growth. Besides that, heavy appreciation results in high adjustment costs to the economy (Botman, Carvalho Filho, & Lam, 2013) and it causes a decline in real exports and/or profit margins (Latsos & Schnabl, 2008). The reason is that appreciation makes export more expensive and imports cheaper (Kandil, 2015). For an importing country, this would be beneficial. However, in countries where export significantly exceeds imports like Switzerland and Japan, the monetary authorities try to reduce appreciation (Latsos & Schnabl, 2008). These disadvantages of appreciation were the reason for the announcement by the Swiss National Bank of a lower boundary of Swiss franc against the Euro (Latsos & Schnabl, 2008).
The Swiss National Bank (SNB) is committed to a flexible exchange rate regime with price stability as its main goal (Latsos & Schnabl, 2008). So is the exchange rate policy of the Japanese. This exchange rate mechanism makes monetary policy autonomy possible, meaning that problems with output or inflation can be dealt with.
9
3. Methodology and data
3.1 MethodologyIn this paper, the United Stated is seen as the domestic country, just like in the paper by Ranaldo and Söderlind (2010) and Switzerland as the foreign country. As explained before, a safe haven currency is a currency of which the value increases during more difficult market situations (Ranaldo & Söderlind, 2010). So what should be tested is a value change of the Swiss franc with respect to the U.S. dollar. This will be tested using a risk premium.
The risk premium is defined as a premium required on an investment in the United States (domestic country). This equals the excess return of an investment in Switzerland (foreign country). To find the excess return, the following formula will be used, similar to a formula used by Atanasov and Nitschka (2013):
𝑅!!!! = 𝑟
!∗− 𝑟!− ∆𝑠!!!
In this formula 𝑅!!!! is the excess return of investing in Swiss francs as
compared to U.S. dollars. i (*) is the short-term interest rate of the United States (Switzerland). ∆𝑠!!! is the change in the log of exchange rate. The exchange rate is defined as units of U.S. dollars per unit of Swiss francs. Therefore, an appreciation of the Swiss franc means an increase in exchange rate and a depreciation means a decrease in exchange rate.
The excess return on an investment in Swiss francs will be used as the dependent variable in the regression. This ability to gain excess return is a reason to invest in a foreign currency. The regression is the following, as used by Ranaldo and Söderlind (2013): Regression 1: 𝑅!! = β !𝑆&𝑃!+ 𝛽!𝑇𝑟𝑒𝑎𝑠𝑢𝑟𝑦𝑁𝑜𝑡𝑒!+ β!𝑅𝑖𝑠𝑘!+ β!𝑆&𝑃!!!+ β!𝑇𝑟𝑒𝑎𝑠𝑢𝑟𝑦𝑁𝑜𝑡𝑒!!! + β!𝑅𝑖𝑠𝑘!!!+ β!𝑅!!!! + α + ε ! S&Pt is an explanatory variable modelling the return on Standard & Poor’s 500. This is an index for the market value of the 500 largest U.S. firms trading on the stock exchange market. As it represents a return, the values will be in percentages. TreasuryNote is an explanatory variable for Treasury note. This indicates a safe U.S. debt security with fixed interest rates (Engle, Fleming, Ghysels & Nguyen, 2012). This long-term government bond yield will also be listed in percentages. The explanatory variable Risk consists of both the volatility risk and liquidity risk, measured by the VIX and TED respectively. These two types of risk will be regressed separately. The VIX is listed in percentages. α is a
10 constant and ε! is the error term. The rest of the variables are lagged variables and function as control variables. Additionally, this thesis will carry out a second regression. A safe haven currency is explained as an appreciation of the currency in financially stressed periods. Therefore, the change in exchange rate will be regressed too. The variables that have an impact on a change in the exchange rate are equal to the variables to have an influence on the excess return of an investment in Swiss francs. Therefore, the same regression will be performed using the change in exchange rate as a dependent variable. This leads to the following regression:
Regression 2:
∆𝑆 = β!S&P!+ β!TreasuryNote!+ β!Risk!+ β!S&P!!!+ β!TreasuryNote!!! + β!Risk!!!+ β!∆𝑆!!!+ α + ε!
In favour of the results for both regressions, the values for TED and TreasuryNote are multiplied by 1000. Gelman describes in his paper that rescaling does not have an effect on the t-statistic or the p-value (2008).
Besides these two regressions, the fundamentals of a safe haven currency as described in the literature review will be analysed for the second time period.
3.2 Assumptions
One of the assumptions of an OLS regression is that no perfect multicollinearity exists between two variables. To test this assumption, the correlation between variables plays a role. The correlations of regression 1 for the first time period from 01-01-1992 to 01-02-2009 are listed in table 1. The correlation tables for the other time period and for the second regression are listed in tables 5, 6 and 7 in the appendix.
The tables show that no perfect multicollinearity occurs between any two variables. However, in all the correlation tables, there is a high correlation between TreasuryNote and TreasuryNotet-1; between VIX and VIXt-1 and between TED and TEDt-1. This is no surprisingly result. As this paper concerns a time series regression, using monthly data, it is not remarkable that the values for TreasuryNote, VIX and TED are affected by their own variables of the previous month.
A high correlation may have an effect on the standard errors of the independent variables and with that its variance (Daoud, 2018). Multicollinearity can cause the estimators to be less precise and predictive (Stock & Watson, 2015). Therefore, a Variance Inflation Factor (VIF) test is performed to measure the extent of multicollinearity within the regression.
11 Table 1: correlations regression 1; 01-01-1992 - 01-02-2009 The asterisk (*) indicates a correlation larger than 0.80.
Daoud states that if the VIF exceeds 5, there is a high correlation. A VIF between 1 and 5 indicates a moderate correlation. If VIF is 1, there is no correlation (2018). Table 2 shows the VIF results for the first regression over the first time period. Table 8, 9 and 10 in the appendix show the VIF results for the other regression and time periods. Table 2: Variance Inflation Factor regression 1; 01-01-1992 - 01-02-2009 Both tables for the first time period show that TreasuryNote and VIX have a VIF larger than 5. Both tables for the second time period show that also the TED has a VIF well exceeding 5. However, as we are regressing time series data and lagged variables are included, a high correlation of these lagged variables to its own variable one period later can be expected. After controlling for these high VIF variables, which are TreasuryNotet-1, VIXt-1 and TEDt-1, all the VIF values
RP S&P
500
Treas. Note
VIX TED S&P 500t-1 Treas. Notet-1 VIXt-1 TEDt-1 RPt-1 RP 1.0000 S&P500 -0.0039 1.0000 Treas. Note 0.0696 0.1726 1.0000 VIX -0.0245 -0.3476 -0.3962 1.0000 TED -0.1245 -0.1931 -0.1479 0.5299 1.0000 S&P 500t-1 -0.0740 0.1101 0.1731 -0.4222 -0.2100 1.0000 Treas. Notet-1 0.1070 0.1746 0.9785* -0.3639 -0.1180 0.1483 1.0000 VIXt-1 -0.0517 -0.1145 -0.4057 0.8964* 0.4118 -0.3197 -0.3764 1.0000 TEDt-1 -0.2069 -0.1704 -0.1669 0.5360 0.8523* -0.1845 -0.1399 0.5295 1.0000 RPt-1 0.0418 0.2015 0.0548 -0.1028 -0.1904 -0.0179 0.0595 -0.0090 -0.1203 1.0000 Variable VIF TreasuryNote 24.85 TreasuryNotet-1 24.27 VIX 11.17 VIXt-1 9.45 TED 5.01 TEDt-1 4.95 S&P500 1.62 S&P500t-1 1.28 RPt-1 1.09 Mean VIF 9.30
12
become smaller than 2. This means that there is only a moderate correlation between any two other variables. The only high correlation is between variables with its lagged variable.
Due to the high correlation between TreasuryNote, VIX and TED with their lagged variables, each regression will also be carried out in the absence of these lagged variables. The results of the regressions including and excluding these lagged variables will be compared to each other to see if these lagged variables should indeed be excluded from the regression.
3.3 Hypothesis
Whether a currency will be a safe haven depends on the value of the betas. An increase in 𝑅!! means that investing in the Swiss franc is beneficial and an
increase in S means an appreciation of the Swiss franc, which also makes it beneficial to invest in the Swiss franc. Therefore, the signs for the betas in both regressions are expected to behave similarly. The following hypotheses regard the first period, in which the Swiss franc certainly was a safe haven.
The first hypothesis is that β1 is negative. The return on S&P500 represents the U.S. stock return. The higher this stock return, the lower the excess return from investing in a Swiss monetary instrument (Campbell, Serfaty-de Medeiros & Viceira, 2009). A decrease in the return on S&P500 means that the overall market performance has worsened, which would indicate a higher excess return. The same reasoning holds for direction of the exchange rate. Therefore, β1 is negatively correlated to 𝑅!! and ∆𝑆.
The second hypothesis is that β2 is positive. Treasury notes represent U.S. government bonds, which themselves are very safe and considered to hold its value if uncertainty and risk increase, as worldwide it is the largest and most active financial market (Mensi, Hammoudeh, Reboredo & Nguyen, 2015). Due to the creditworthiness and liquidity of Treasury notes, they are commonly used for a hedging position (Giang, 2012). Therefore, a positive relation to these bonds is expected, as they act in the same way as a safe haven currency.
The third hypothesis is that β3 is positive. A safe haven currency will appreciate with increasing market risk (Botman, Carvalho Filho & Waikei, 2013). Both VIX and the TED spread are volatilities explaining market risk. Therefore, a positive volatility is expected to appreciate the Swiss franc.
If the Swiss franc would have completely renounced its safe haven status in the second period, the beta coefficient for the return on S&P500 (β1) is expected to be positive and the betas for Treasury notes (β2) and risk (β3) are expected to be negative. On the other hand, if the Swiss franc would still be a safe haven currency after the change in monetary policy, the betas are expected to be similar to the first period.
13
3.4 Data
This thesis makes use of monthly data. This data is collected from different databases. The interest rates of the U.S. and Switzerland come from the Federal Reserve Economic Research (https://fred.stlouisfed.org) database and the Sweizerische Nationalbank (https://data.snb.ch) respectively. For the Swiss interest rates, 3-month money market rates are used. For U.S. interest rates, 3-month treasury bills are used. Also the exchange rate, the Treasury notes and the data on the two types of risk are collected from the FRED. The return on S&P 500 is collected from https://finance.yahoo.com. The exchange rates are collected as units of Swiss francs per unit of U.S. dollar. However, the regression makes use of the exchange rate defined as units of U.S. dollars per unit of Swiss franc. Therefore, the inverse of the exchange rate collected from the FRED is used for the regression.
Each regression will be carried out for two different periods in time. Once, the regression concerns the period 01-09-1990 till 01-02-2009. The second time period will be from 01-03-2009 till 01-04-2018. This gives 204 and 205 observations for the first time period regressing R!! and ∆S respectively. The
second time period has 109 observations for both regressions.
The reason that the first time period starts at 09-1990 is because there is many missing data in the prior periods. As of 09-1990, almost all the data needed for this research was available, except for 6 values for the Swiss interest rate. The dates for these missing values were 01/03/1992; 01/06/1992; 01/09/1992; 01/12/1992; 01/09/1993 and 01/12/1993. Therefore, for these dates the average between the preceding and subsequent interest rate is calculated using 𝑟!= (𝑟!!!∗ 𝑟!!!)!,!.
The periods are divided at 01-03-2009. The reason for a division is because this is the month at which the Swiss National Bank decided to announce and change its monetary policy. The purpose of thesis is to see if this change in monetary policy has actually changed the safe status of the Swiss franc and therefore these two different time periods will be compared with each other.
4. Results and empirical data analysis
The results for the two regressions in the first period from 01-01-1992 to 01-02-2009 are listed in table 3. The results for the second period from 01-03-2009 to 01-05-2018 are listed in table 4. These two tables show the results for the full regressions 1 and 2 as given in the methodology and they show the results for the regression excluding the highly correlated lagged variables. The excluded variables are TreasuryNotet-1, VIXt-1 and TEDt-1.14 The VIF test for regressions 1 and 2 turned out to exceed the value of 5 and is therefore too high. However, when comparing both of the regressions, with and without the highly correlated lagged variables, the complete regression as given in the methodology gives better results in terms of significance and R2 than the regression with the omitted variables. The high values for the VIF test are explained by the fact that there are lagged variables in the regression. As monthly data is used, it is no surprise that the variables are highly correlated to its own lagged variable. As no correlation between other variables is present, which can be derived from the VIF tests where the highly correlated variables are omitted, multicollinearity is ruled out. It will not be necessary to omit the lagged variables and therefore the complete regressions 1 and 2 will be used as for the analysis.
As many researchers proved that the Swiss franc was at least a safe haven currency in the first time period (Botman, Carvalho Filho, & Waikei, 2013), table 3 is meant to confirm the usefulness of the variables in explaining the safe haven status of the Swiss franc. The results in table 4 are used to formulate the answer on the present status of the Swiss franc. It follows from the regressions in the first time period that the S&P500 return is negatively correlated to the dependent variables of regressions 1 and 2. This is in accordance to the previously described definition of a safe haven currency, namely that the demand for a safe haven currency increases as the economy is doing worse (Ranaldo and Söderlind, 2010). In the second time period, the sign for the S&P coefficients turn positive for both regressions.
Treasury notes in the first period, being highly significant in regression 2, turn out to be negatively correlated to the dependent variables of both regressions. This is a surprising result. Giang explains that safe haven seeking investors are typically investing in Treasury notes (2012). Therefore a positive coefficient was expected. The lagged variable TreasuryNotet-1 on the other hand does have a positive coefficient, being significant for both regressions. Therefore, the lagged variable TreasuryNotet-1 will be a valuable variable in the analysis for the safe haven status. In the second time period, TreasuryNote is still a negative coefficient and TreasuryNotet-1 is still a positive coefficient for both regressions. The coefficients for VIX and TED result in opposite values in the first period. The VIX has a negative correlation to both the excess return and the change in exchange rate, whereas the TED is positively correlated. In the second period, both the risk factors still result in opposite signs to each other. Additionally, now the signs for both VIX and TED are also opposite for regressions 1 as compared to regression 2. For the continuation of this paper, the TED variable will be the risk factor analysed, as this variable is corresponding to the definition of a safe haven in the first time period.
15
Table 3: Period from 01-01-1992 to 01-02-2009
Listed are the regression coefficients. The standard errors are in parenthesis. * indicates the significance, in the following way: *= p<0.10, **= p<0.05, *** =p<0.01. VIX, TED spread, exchange rate and treasury notes are collected from the FRED. The S&P500 returns come from finance.yahoo.com. R!! is calculated as described earlier. ∆S means the change in the exchange
rate with respect to the previous month. This is calculated using the formula ∆S = (𝑆!− 𝑆!!!)/𝑆!!!. Monthly data is used.
Variable Regression 1 Regression 2 Regression 1 Regression 2
Dependent Variable R!! ∆S R ! ! ∆S Independent Variables S&P500 -.3136496 -.0082865 -.1308508 .0011881 (.4824968) (.0516827) (.4175294) (.0472895) TreasuryNote -.017286 -.0035864 *** .0018218 -.0001712 (.0070063) (.0007542) (.0015866) (.0001782) VIX -.099019 -.0001757 .1276942 -.0154843 (.6447812) (.0692554) (.2804467) (.0317528) TED spread .0108433 .0005509 -.0094747 -.000462 (.008866) (.000949) (.0048344) (.000547) S&P500t-1 -.4614637 -.0737306 -.5793001 -.0940422 (.4368062) (.0467869) (.4394749) (.0497028) TreasuryNote t-1 .0194735 * .0035292 *** (.0069534) (.0007486) VIX t-1 .3413625 -.0154309 (.6075005) (.0651726) TED spread t-1 -.0256692 ** -.0013911 (.0088244) (.0009501) Dependent var t-1 .0218014 .2142332*** .0147135 .2457503*** (.0709834) (.0655374) (.0728523) (.0687601) Constant -.1114541 .0114981 -.0960019 .01653 (.110782) (.011931) (.1127818) (.0127259) R2 0.1097 0.1904 0.0333 0.0856 N 204 205 204 205
16
Table 4: Period from 01-03-2009 to 01-05-2018
Listed are the regression coefficients. The standard errors are in parenthesis. * indicates the significance in the following way: * p<0.10 ** p<0.05 *** p<0.01. See table 1 for information on the data.
Variable Regression 1 Regression 2 Regression 1 Regression 2
Dependent Variable R!! ∆S R!! ∆S Independent Variables S&P500 62.51733 .2199542*** 53.30272 .1490198** (58.79035) (-.0013553) (44.64013) (.0637392) TreasuryNote -1.243464 -.0013553 -.384115 .0005421 (.9777877) (.0013427) (.2997993) (.0004281) VIX -4.15578 .0882267 42.59647 -.0093587 (65.28785) (.0906823) (29.99421) (.0428271) TED spread 1.410403 -.0073365 -2.381311 -.0010272 (3.063177) (.0043343) (1.3601) (.001942) S&P500t-1 138.6951** .0378231r 107.3098** .0642168 (48.12072) (.0670511) (43.97409) (.0119808) TreasuryNote t-1 .9263941 .0020752 (1.013252) (.0013901) VIX t-1 46.62068 -.1452815 (60.12164) (.0833107) TED spread t-1 -3.888242 .0081248 (2.813714) (.0039675) Dependent var t-1 -.0252037 .0547478 -.032646 .0846771 (.0970547) (.0994595) (.0960386) (.1001567) Constant 6.979669 -.0111906 8.72972 .01653 (8.626372) (.0118982) (8.390806) (.0127259) R2 0.1076 0.1788 0.0778 0.0950 N 109 109 109 109 The Swiss National Bank adjusted its monetary policy with the ambition that the Swiss franc ends its continuously appreciating behavior. Figure 2 shows a graph with the Swiss exchange rate collected from the Federal Reserve Economic Database. The grey areas indicate a recession in the U.S. Figure 2 shows that after depreciating in 2011, the exchange rate has been more or less constant. As it is no longer appreciating, the exchange rate movements are in line with the desired effect of the SNB.
17
Figure 2: Exchange rate as units of U.S. dollars per unit of Swiss franc.
The grey areas indicate a U.S. recession. Source: FRED economic research
Previously, three fundamentals for a county with a safe haven currency were listed. These fundamentals are; low interest rates, a strong net foreign asset position and a highly liquid financial market (Botman, Carvalho Filho & Lam, 2013). These fundamentals are analysed for the second period. First of all, the interest rate in Switzerland has decreased up to a negative amount of -0,75%. This is shown in figure 3. The -0,75% is the benchmark level set by the Swiss National Bank (SNB). Figure 3: The interest rate (three-month LIBOR) of Switzerland The data contains the years from 2000 until 2018. The highest value was 3,5 in 2001. Ever since January 2015, the Swiss National Bank (SNB) keeps the interest rate at -0,75%. The interest rate is given in percentages. Source: Trading economics. Secondly, a country with a safe haven currency would have a strong net foreign asset position. Figure 4 shows the Swiss’ net international investment position. Ever since 2009, the value has been decreasing.
18 Figure 4: Swiss Net International Investment Position The data contains the years from 31-12-2000 until 31-12-2017. Source: quandl.
Thirdly, in the period from 15 September 2008 to 21 January 2015, Switzerland experienced excess liquidity (Fuhrer, 2015). The subsequent period is characterized by a period of excess reserves (Fuhrer, 2015). Therefore, Switzerland may be considered a highly liquid financial market.
5. Discussion
From the VIF test followed high results. However, as a high correlation between a variable and its lagged variable is no surprise and there is no high correlation present between any two other variables, the complete regressions as mentioned in the methodology are used for the analysis.
From the analysis regarding the first period, the S&P500 turns out to be indeed an approved variable in explaining a safe haven currency. In the period after the monetary policy adjustments, upon an increase in the return of S&P500, the return on the Swiss franc tends to increase and the exchange rate tends to appreciate. The hypothesis for a safe haven currency on the return of S&P500 is that its relation to excess return and the change in exchange rate is negative. So the results for S&P500 in the second period create the idea that the Swiss franc has lost its safe haven status.
The value of U.S. Treasury notes was hypothesized to commove with the excess return on Swiss franc and the change in exchange rate, as U.S. Treasury notes are considered one of the safest investment options. Surprisingly, this is not the case for the first time period. The lagged variable of Treasury note on the other hand is in accordance to the hypothesized results for a safe haven currency. In the period after the announcement by the SNB, the effect of Treasury notes and the lagged Treasury notes on the value of the Swiss franc remain the same as in the period before the announcement. Because a change in Treasury note still has the same pressure on the Swiss franc, according to Treasury notes the status of the Swiss franc has not changed.
19
When looking at the risk factor, in the first period it becomes clear that if risk increases and thus the TED spread increases, the value of the Swiss franc will increase. In the second time period, this relation is less clear as regressions 1 and 2 give different results. An increase in the TED spread would increase the return on the Swiss franc and at the same time the Swiss franc will depreciate. These results are inconsistent. However, there is no telling which regression is superior to the other, due to a similar R2 and almost equal levels of significance. Therefore, it is questionable on which of the two regressions to base the result regarding the risk factor TED and its corresponding effect on the Swiss franc.
Consecutive to the regression results, the fundamentals of a safe haven currency are analysed. Ever since the announcement by the SNB of a change in monetary policy, Switzerland experiences low interest rates and a highly liquid financial market. These are two fundamentals in favour of a safe haven currency. However, the Swiss Net Foreign Asset Position has been declining, which is opposite to the fundamental of a safe haven. Although these findings are not clearly pointing to an answer about the current status of the Swiss franc, the Swiss franc has stopped appreciating as demanded by the SNB. Therefore, the change in monetary policy has worked out as intended.
6. Conclusion
Due to the fear of deflation, the Swiss National Bank decided to adjust its monetary policy. As the Swiss franc has been seen a safe haven currency up until this point, this raised the following question: can the Swiss franc still be considered a safe haven currency after the announcements by the Swiss National Bank to adjust its monetary policy? To answer this question, this paper carried out two OLS-regressions, one on the excess return of an investment in Swiss francs and one on the change in exchange rate. The data used was divided into two periods, namely the period prior to the announcement being from 01-01-1992 to 01-02-2009 and the period after the announcement from 01-03-2009 to 01-05-2018. Besides the regressions, the fundamentals of a safe haven currency in this second time period were analysed.The overall findings of the paper are ambiguous. If the observed movement of the S&P500 in relation to the Swiss franc were to be considered, it would suggest that the Swiss franc is no longer a safe haven currency. The relation between Treasury notes and both excess return and the change in exchange rate suggest the opposite, namely that the Swiss franc still is a safe haven. The risk factor measured by the TED spread gives different results for both regressions, causing no obvious answer. Also the fundamentals for a safe haven currency lead to ambiguous results. Even though Switzerland experiences
20
low interest rates and has a highly liquid financial market in favour of the safe haven status, its net foreign asset position is decreasing which contradicts this status.
The overall findings of this paper lead to inconclusive results. This suggests further research. One suggestion is a careful search for the factor describing risk. Risk is defined to have a major effect on the behaviour of a safe haven currency. In this paper however, the results for risk were not significant. The effect of fiscal policy on the safe haven status is another question left for further research. Including fiscal variables in the regression might increase the R2 and the overall significance.
Reference list
Atanasov, V., & Nitschka, T. (2014). Currency excess returns and global downside market risk. J. Int. Money Finan. 47, 268–285. Bandi, F. M., Moise, C. E., & Russell, J. R. (2008). The joint pricing of volatility and liquidity, mimeo, University of Chicago. Barrios, s., Iversen. P, Lewandowska, M., & Setzer, R. (2009). Determinants of inta- euro area government bond spreads during the financial crisis. European Commission. Baur, & Mcdermott. (2010). Is gold a safe haven? International evidence. Journal of Banking and Finance, 34(8), 1886-1898. Botman, D., Carvalho Filho, I, & Waikei L. R. (2013). The Curious Case of the Yen as a Safe Haven Currency; A Forensic Analysis. 13/228. Brunnermeier, M., Nagel, S., & Pedersen, L. (2008). Carry Trades and Currency Crashes. NBER Macroeconomics Annual, 23(1), 313-348. doi:10.1086/593088 Campbell, J. Y., Serfaty-de Medeiros, K., & Viceira, L. M. (2009). Global currency hedging, Journal of Finance, forthcoming. Daoud, J. I. (2018). Multicollinearity and Regression Analysis. Journal of Physics: Conference Series, 949(1) Engle, R., Fleming, M., Ghysels, E., & Nguyen, G. (2012). Liquidity, volatility, and flights to safety in the U.S. treasury market: Evidence from a new class of dynamic order book models. 590. Fuhrer, L. M., (2017). Liquidity in the Repo Market. SNB working papers. Gelman, A. (2008). Scaling regression inputs by dividing by two standard deviations. Statistics in Medicine, 27(15), 2865-2873 Glen, J., & Jorion, P. (1993). Currency Hedging for International Portfolios. The Journal of Finance, 48(5), 1865-1886. doi:10.2307/2329070 Habib, & Stracca. (2012). Getting beyond carry trade: What makes a safe21 haven currency? Journal of International Economics, 87(1), 50-64. Kandil, M. (2015). On the benefits of nominal appreciations: Contrasting evidence across developed and developing countries. Borsa Istanbul Review, 15(4), 223-236. Kaul, A., & Sapp, S. (2006). Y2k fears and safe haven trading of the u.s. dollar, Journal of International Money and Finance 25, 760–779. Latsos, S., & Schnabl, G. (2018). Net foreign asset positions and appreciation expectations on the Swiss franc and the Japanese Yen. International Economics and Economic Policy,15(2), 261-280. Maechler, A. M. (2017, November 16). Developments on the Swiss franc capital market and the SNB's monetary policy. Speech presented at Money Market Event, Geneva. Menkhoff, L., Sarno, L., Schmeling, M., & Schrimpf, A., (2012). Carry trades and global foreign exchange volatility. Journal of Finance 67 (2), 681–718. Mensi, Hammoudeh, Reboredo, & Nguyen. (2015). Are Sharia stocks, gold and U.S. Treasury hedges and/or safe havens for the oil-based GCC markets? Emerging Markets Review, 24, 101-121. Nelson, S., & Katzenstein, p. (2014). Uncertainty, Risk and the financial crisis of 2008. International Organization, 68(2), 361-392. Pástor, U., & Stambaugh, R. (2003). Liquidity Risk and Expected Stock Returns. Journal of Political Economy, 111(3), 642-685. Ranaldo, A., & Söderlind. P. (2010). Safe Haven Currencies. Review of Finance 14, 385-407. doi: 10.1093/rof/rfq007 Solnik, B. (1974). An equilibrium model of the international capital market. Journal of Economic Theory, 8(4), 500-524. Stock, J. H., & Watson, M. W. (2015). Introduction to econometrics (Vol. 104). Boston: Addison Wesley. Tachibana, M. (2018). Safe-haven and hedge currencies for the US, UK, and Euro area stock markets: A copula-based approach. Global Finance Journal, 35, 82-96. Verdelhan, A. (2010). A habit-based explanation of the exchange rate risk premium. The Journal of Finance: The Journal of the American Finance Association, 65(1), 123-146. Zurbrügg, F. (2012, November 21). Fiscal and monetary policy: Interdependence and possible sources of tension. Speech presented in University of Lucerne, Lucerne.
22
Appendix
Table 5: correlations regression 1; 01-03-2009 - 01-05-2018 RP S&P 500 Treas. NoteVIX TED S&P 500t-1 Treas. Notet-1 VIXt-1 TEDt-1 RPt-1 RP 1.0000 S&P500 0.0617 1.0000 Treas. Note 0.0017 0.0843 1.0000 VIX 0.0240 -0.0276 0.3003 1.0000 TED -0.0867 0.1583 -0.1996 0.4224 1.0000 S&P 500t-1 0.1951 -0.0360 0.1642 -0.0703 0.0265 1.0000 Treas. Notet-1 0.0192 -0.0297 0.9529* 0.3852 -0.2825 0.0826 1.0000 VIXt-1 0.0608 0.2253 0.2866 0.8526* 0.3804 -0.0276 0.3003 1.0000 TEDt-1 -0.0909 0.1990 -0.1433 0.2932 0.9035* 0.1583 -0.1996 0.4224 1.0000 RPt-1 -0.0027 -0.0362 0.0327 0.0228 -0.1034 0.0618 0.0021 0.0240 -0.0867 1.0000 The asterisk (*) means a correlation >0.80. Table 6: correlations regression 2; 01-01-1992 - 01-02-2009 ∆S S&P 500 Treas. Note
VIX TED S&P 500t-1 Treas. Notet-1 VIXt-1 TEDt-1 ∆𝑆!!! ∆S 1.0000 S&P500 0.0048 1.0000 Treas. Note -0.0715 0.1676 1.0000 VIX 0.0055 -0.3466 -0.3960 1.0000 TED -0.0532 -0.1916 -0.1504 0.5301 1.0000 S&P 500t-1 -0.1233 0.1097 0.1667 -0.4209 -0.2079 1.0000 Treas. Notet-1 -0.0008 0.1747 0.9786* -0.3640 -0.1215 0.1676 1.0000 VIXt-1 -0.0239 -0.1147 -0.4048 0.8964* 0.4121 -0.3466 -0.3960 1.0000 TEDt-1 -0.1044 -0.1706 -0.1697 0.5362 0.8524* -0.1916 -0.1504 0.5301 1.0000 ∆𝑆!!! 0.2535 0.0309 -0.1064 0.0046 -0.0479 0.0048 -0.0715 0.0055 -0.0532 1.0000 The asterisk (*) means a correlation >0.80.
23 Table 7: correlations regression 2; 01-03-2009 - 01-05-2018 ∆S S&P 500 Treas. Note
VIX TED S&P 500t-1 Treas. Notet-1 VIXt-1 TEDt-1 ∆𝑆!!! ∆S 1.0000 S&P500 0.2267 1.0000 Treas. Note 0.1711 0.0843 1.0000 VIX -0.0201 -0.0276 0.3003 1.0000 TED -0.0606 0.1583 -0.1996 0.4224 1.0000 S&P 500t-1 0.1174 -0.0360 0.1642 -0.0703 0.0265 1.0000 Treas. Notet-1 0.1703 -0.0297 0.9529* 0.3852 -0.2825 0.0826 1.0000 VIXt-1 -0.0181 0.2253 0.2866 0.8526* 0.3804 0.0826 0.3003 1.0000 TEDt-1 0.0207 0.1990 -0.1433 0.2932 0.9035* 0.1583 -0.1996 0.4224 1.0000 ∆𝑆!!! 0.1069 -0.1626 0.1810 0.0049 -0.1446 0.2316 0.1827 -0.0201 -0.0606 1.0000 The asterisk (*) means a correlation >0.80. Table 8: Variance Inflation Factor regression 1; 01-03-2009 - 01-05-2018 Variable VIF TreasuryNote 14.94 TreasuryNotet-1 16.03 VIX 6.68 VIXt-1 6.67 TED 6.79 TEDt-1 7.22 S&P500 1.84 S&P500t-1 1.28 RPt-1 1.05 Mean VIF 6.94
24 Table 9: Variance Inflation Factor regression 2; 01-01-1992 - 01-02-2009 Table 10: Variance Inflation Factor regression 2; 01-03-2009 - 01-05-2018 Variable VIF TreasuryNote 25.23 TreasuryNotet-1 24.54 VIX 11.12 VIXt-1 9.38 TED 4.97 TEDt-1 4.96 S&P500 1.61 S&P500t-1 1.27 ∆𝑆!!! 1.03 Mean VIF 9.34 Variable VIF TreasuryNote 14.72 TreasuryNotet-1 15.77 VIX 6.73 VIXt-1 6.69 TED 7.10 TEDt-1 7.51 S&P500 1.92 S&P500t-1 1.30 ∆𝑆!!! 1.18 Mean VIF 6.99
