Dynamics of the Swiss Franc Appreciation Explained by the Dornbusch Model an Approach

Transcription

1 University of Bern Faculty of Business, Economics & Social Sciences Institute of Economics Prof. Dr. Klaus Neusser Bachelor Thesis Explained by the Dornbusch Model an Approach Andreas Kraml Bachelor of Science in Economics; Kappelenstrasse Lyss andreas.kraml@students.unibe.ch 27 April 2012

2 Abstract This paper contains an explanation of the Swiss Franc appreciation in summer Exchange rate movements are analysed by two different versions of Rüdiger Dornbusch s overshooting model from Recent European Central Bank s monetary policy decisions operate as exogenous shock hitting the Swiss economy. Focused on real indices, the subsequent chapters prove the applicability of the Dornbusch model by empirical data and justify the assumption of rational behaviour. The analysis also points out implications for economic policy and explains possible problems such as inflation stabilisation caused by volatile exchange rates. 2

3 Index 1. Introduction The Dornbusch model in general: Setup and visualisation Elements of the model: The asset market Elements of the model: The goods market Dynamic evolution of the economy and the phase diagram European Central Bank s monetary policy as exogenous shock? Interventions and recent liquidity injections Implications on Swiss Franc Euro exchange rate Exchange rate dynamics in the Dornbusch model following an unexpected foreign interest rate shock Adjustments of the Dornbusch model Technical analysis and the path of a foreign interest rate shock Economic intuition and comparison with empirical results An alternative adjustment implying an overshooting Explanatory power of the model and evidence of rationality A critical view Implications for economic policy Conclusion Sources Self-declaration List of illustrations Figure 1: Phase diagram with flexible exchange rates (δ<1)....9 Figure 2: Benchmark interest rate of the Euro area 11 Figure 3: Growth in ECB s balance sheet 11 Figure 4: Proportions of financial guarantees.. 11 Figure 5: Real interest rate spread Switzerland Euro area Figure 6: Real exchange rate indices Swiss Franc Euro area (export-weighted).. 13 Figure 7: Phase diagram with flexible exchange rates (δ>1) Figure 8: Technical analysis (δ>1) Figure 9: Change in GDP (Switzerland) Figure 10: Change in consumer prices...18 Figure 11: Technical analysis (δ<1)

4 1. Introduction When Rüdiger Dornbusch, a German economist working at the Massachusetts Institute of Technology, presented his macroeconomic model of flexible exchange rates in 1976, many researchers were in doubts about his concept which assumed sticky prices. This assumption was under severe attack at that time but today, there is a broad consensus across schools about certain price rigidities in reality. The Dornbusch overshooting model explains exchange rate dynamics following a monetary shock as a predominant source of disturbances. Kenneth Rogoff evaluated this model after 25 years in a paper of the International Monetary Fund (IMF) and attested its importance: [ ] the Dornbusch model defines a high-water mark of theoretical simplicity and elegance in international finance, one which inspired a generation of students, and which still stands today as fundamental. Even today, the model in its original form remains relevant for policy analysis. Dornbusch (1976) is truly an extraordinary paper, one of the handful of most influential papers in macroeconomics [ ] Kenneth Rogoff (2002, p.19). The model experiences a certain renaissance those days because it provides rational explanations of short term exchange rate movements. Various currencies (including USD, GBP, JPY, Euro) exhibit an exceptionally high volatility of prices in the past two years as a result of the financial crisis and the dept problems. Several monetary shocks could be observed in recent years. Central banks were challenged and implemented policy strategies in order to inject liquidity in a system that suffers from a credit crunch. At the same time, the Swiss Franc showed an undesirable development and began to appreciate massively. This paper aims to investigate if the exchange rate development of the Swiss currency in the summer period 2011 can be explained by the Dornbusch overshooting model from Therefore, Dornbusch s basic assumptions are supposed to be mainly consistent with the reality. In addition to a general setup and visualisation of this flexible exchange rate model in the first half, the paper also investigates if the monetary policy of the European Central Bank (ECB) can be considered as exogenous shock affecting the Swiss economy. There were various macroeconomic shocks observable in the last few years. It is therefore necessary to focus on one event which could have caused the development of the exchange rate according to the model. A spread in real Swiss and European interest rates is illustrated and analysed. Following the monetary policy of the ECB, a decline in the Euro area s interest rates during the 4

5 summer 2011 is expected to affect the exchange rates. If this was not the case, investors would benefit from arbitrage. Two adjusted versions of the model are considered in the second half as explanations of the dynamics caused by an unanticipated, long-lasting decline in foreign interest rates. Of course, there are a number of macroeconomic indicators involved in order to explain the movements of prices and exchange rates. The paper considers real indices such as real interest rates, real exchange rates or changes in real output instead of nominal values. The subsequent analysis and figures demonstrate that the second version of the model better reflects past developments than the first one due to a changed price elasticity of net exports. It can be shown that the model explains the temporary overshooting of the Swiss Franc appreciation in summer Moreover, the model s predictions about other macroeconomic terms are also consistent with available empirical data. With certain justified assumptions and useful adjustments, the Dornbusch model can be applied in order to analyse the dynamics of the exchange rate. The comparison of Dornbusch s theoretical approach with empirical data clearly indicates rationality. Despite of some restrictions which weaken the explanatory power of the model, the basic assumptions of the model correspond to real circumstances. Besides, the applicability of Dornbusch s concept of volatile exchange rates has various important impacts on economic policy which are mentioned in the last part. In general, flexible exchange rates influence inflation stabilisation as well as domestic competitiveness. Further consequences arise from the fact that the model is expectation-based and the exchange rate values may reflect incorrect beliefs. 5

6 2. The Dornbusch model in general: Setup and visualisation 2.1 Elements of the model: The asset market The Dornbusch overshooting model explains exchange rate overshooting in a small open economy with flexible exchange rates. Asset markets consist of a bond market, a money market and a market for foreign exchange. Capital is assumed to be perfectly mobile and the domestic interest rate 1 minus an expected rate of depreciation must equal the interest rate of the rest of the world as a consequence (Dornbusch, 1976, p.1162). Domestic and foreign bonds are considered as perfect substitutes and the corresponding rates of return are assumed to even out anticipated exchange rate changes. Equation (1) below states this so called Uncovered Interest Rate Parity (UIP), holding that there should not be any arbitrage across currencies, where r t denotes the domestic interest rate, r f the given world rate of interest and e n describes the logarithmised expected rate of domestic depreciation 2 (Rogoff, 2002, p. 6). r t = r f + e n (1) Since we have no stochastic shocks and people are assumed to be rational, agents have perfect foresight and the formation of expectations occurs ad hoc (Dornbusch, 1976, p.1163). Therefore, an unanticipated shock leads to a discrete jump of the flexible exchange rate and unanticipated windfall gains or losses on assets denominated in foreign currency are generated, meaning that the no-arbitrage condition is temporarily invalid. However, the exchange rate is constant in the long-run ( e n domestic and foreign interest rate will adjust (r t = r f ). = 0) and the Looking at the balanced money market, the demand for real money will equalise real money supply, expressed by the LM equation (2), meaning that money demand depends negatively on interest rates due to opportunity cost of holding money and positively on output according to the transaction motive (Blanchard & Illing, 2006, p. 110). All variables are logarithmised and nominal money supply divided by domestic price level becomes m p t which states real money supply on the left hand side. m p t = y t d ȳ ε(r t r f ) (2) Parameter ε is positive describing the sensitivity of money demand with respect to the interest rate. 1 For simplification, domestic and foreign inflation are assumed to be zero (π = π f = 0). Hence, we do not distinguish between real and nominal interest rates in this setup. 2 E n denotes the nominal exchange rate, i.e. the price of foreign currency denominated in terms of domestic currency. Taking the logarithm, we define e n = lne n such that e n = lne n / t = 1/E n E n / t measuring the (expected) percentage change in the nominal exchange rate (Sørensen, 2003, p.34). 6

7 2.2 Elements of the model: The goods market The Dornbusch model is provided with important Keynesian elements regarding the goods market. We assume the world price of imports to be given exogenously whereas prices of domestic goods depend on the aggregate demand since domestic output is an imperfect substitute for imports (Dornbusch, 1976, p.1162). Equation (3), known as the IS equation, states the aggregate demand for domestic output y d t as a decreasing function of the relative price of domestic goods and the interest rate (Blanchard & Illing, 2006, p.144). Again, all variables are denominated in logarithmic terms where the difference e n p t denotes the real exchange rate with δ > 0 describing the price elasticity of net exports. The interest rate sensitivity of the aggregate demand is denoted by the parameter ϵ > 0 and the exogenous variable u captures shocks to aggregate demand. y d t = ȳ + u + δ(e n p t ) ϵ(r t r f ) (3) The key feature of the Keynesian approach according to the goods market is the assumption of sticky prices of domestic goods in the short run and the lag of time to adjust (Rogoff, 2002, p.14). Hence, aggregate demand y d t can deviate 3 temporarily from the natural output level ȳ. The connection between sticky prices and deviations of aggregate demand from potential output ȳ is given by the price adjustment equation 4 (4) where prices proportionally change with deviations from potential output (Sørensen, 2003, p.9). p t = α(y d t - ȳ ) α > 0 (4) In particular, prices increase due to a positive demand shock and decrease if aggregate demand is below potential output ȳ. The price level is a predetermined variable whose value is tied down by its historical value in the current period (Sørensen, 2003, p.44). 2.3 Dynamic evolution of the economy and the phase diagram Given those four equations, the dynamic evolution of the economy can be derived by solving equations (2) and (3) for y t d and r t in terms of p t and e n which is shown by equations (5) and (6). 3 Equation (3) shows the deviation and output gap: d yt - ȳ = u + δ(e n p t) ϵ(r t r f ) where y d t - ȳ = 0 in the long run (and e n = 0). 4 This relation is known as the Philips curve, assuming that expected inflation is zero. 7

8 y t d - ȳ = ( δε ε+ϵ )en - ( δε+ϵ ε+ϵ )p t + ( εu+ϵm ε+ϵ ) (5) r t r f = ( δ ε+ϵ )en + ( 1-δ ε+ϵ )p t + ( u-m ε+ϵ ) (6) Considering equation (6), an increase in prices has two counteracting effects: A higher p t tends to decrease real money supply which drives up interest rates, but a higher price level decreases net exports and thus output, leading to lower money demand and therefore decreasing domestic interest rates (Tønners, 2011, p.8). The value of parameter δ is crucial for the detection of domination, the first effect dominates if δ<1. In order to find the instantaneous rate of change in p t, equation (5) is inserted in (4) which results in equation (7). Equation (8) is the combination of (6) and (1) and explains the rate of change in e n. p t = ( αδε ε+ϵ )en - α( δε+ϵ ε+ϵ )p t + α( εu+ϵm ε+ϵ ) (7) e n = ( δ ε+ϵ )en + ( 1-δ ε+ϵ )p t + ( u-m ε+ϵ ) (8) In the long run, the Dornbusch model assumes markets to clear and exchange rates to be constant, meaning that (7) and (8) are equal to zero in the steady state (Dornbusch, 1976, p.1165). Equation (1) implies r* = r f in the long run, y* equals ȳ due to relation (4), equation (2) implies p* = m and (e n )* = m - u δ according to the IS equation (3). Solving equation (7) for the long term with stable prices, it can be written as the exchange rate depending on prices, demand shocks and nominal money supply: e n = ( δε+ϵ δε )p t - ( εu+ϵm δε ) (9) Equation (9) describes the upward-sloping Goods Market curve (GM) with a slope larger than one. The intuition of a positive slope is explained by the case of an increase in prices which decreases demand for domestic goods and weakens the competitiveness of the domestic economy (Sørensen, 2003, p.38). According to a δ<1, interest rates increase due to a decline of real money supply. Since we assume y* = ȳ and stable prices, the exchange rate is expected to depreciate (a higher e n ) in order to stimulate the economy via net exports. At all points above the GM curve, the high depreciation increases domestic competitiveness (y t > ȳ) and hence, price level adjusts according to the price adjustment equation (4) and vice versa for all points below the GM curve. The dynamics are indicated in figure 1. 8

9 Also equation (8) is solved for e n as endogenous variable maintaining a constant exchange rate which gives the long run asset market equilibrium (10): e n = ( m-u ) δ - (1-δ δ )p t (10) The Asset Market curve (AM) must be negative 5 assuming δ<1 and implies e n = 0 and r* = r f but output need not be equal to its natural rate. The negative slope can be explained by an increase in prices as example. Assuming δ<1, interest rates increase because of the domination of declining real money supply and might exceed r f. Domestic currency is therefore expected to appreciate in order to reduce aggregate demand and output until the declining transaction demand for money has cancelled the upward pressure on domestic interest rates (Sørensen, 2003, p.39). Below the AM curve, e n is too low and the economy suffers from the strong domestic currency. The low transaction demand for money reduces interest rates and the domestic currency must appreciate ( e n <0) so that domestic bonds are as attractive as foreign bonds 6. Above the AM curve, the economy benefits from a strong competitiveness, but r t > r f and a depreciation is needed for the UIP to hold. According to those dynamic forces, figure 1 shows the direction of motions. It is to mention that path I implies forever increasing prices and infinite depreciation whereas movement II states a negative exchange rate 7 and forever declining prices which means that motions I and II are unstable paths (Dornbusch, 1976, p.1166). III I II Figure 1: Phase diagram with flexible exchange rates (assuming δ<1) (Sørensen, 2003, p.41) Since agents are expected to have perfect foresight, such bubbles can be ruled out and there must be a saddle path between I and II leading to a steady state (path III). It is 5 Note that the AM curve is horizontal in the case of δ=1 meaning that e n is independent of p t and e n = (m-u). 6 See equation (1). 7 The free disposal of a currency contradicts such a scenario in which people have to be paid e.g. a subsidy to accept foreign currency. 9

10 obvious that an economy is expected to be on the saddle path where it converges towards the long run equilibrium point Ē with stable prices and exchange rates. As mentioned above, steady state conditions imply r* = r f, y* = ȳ, p* = m and (e n )* = m - u δ. In any adjustment process, the jump variable e n reacts immediately but prices (as state variables) are predetermined (Dornbusch, 1976, p.1166). 3. European Central Bank s monetary policy as exogenous shock? 3.1 Interventions and recent liquidity injections The effects of the American subprime crisis seriously affected the European economy and tendencies of an economic recession in Europe became apparent. As a result of the liquidity crisis in October 2008, the Eurosystem intervened with massive liquidity injections by open market operations and marginal lending facilities in order to avoid a collapse of the banking system (De Grauwe, 2009, p. 218). After October 2008, the ECB has provided banks with unlimited amounts of money in its liquidity-providing tenders to stimulate lending between banks (Randow, 2009). Main refinancing operations, the bulk of ECB s open market operations, consist of liquidity-providing reverse transactions with duration of one week, executed by the National Central Banks on the basis of standard tenders (European Central Bank [ECB], 2012). At the beginning of 2009, a rapid cut of the ECB s benchmark interest rate 8 (which is applied on the main refinancing operations) was observable since the economy of the European Union was deep in recession and inflation was falling. Traditional monetary economics aims to influence the interest rate by purchasing and selling short-term bonds. A purchase expands the money supply and decreases short-term interest rates, selling causes the opposite (Brunetti, 2011, p.106). Additionally, unconventional monetary policy was implemented such as the purchase of bonds with a longer duration, called Credit Easing. The ECB Monthly Bulletin from June (ECB, 2009) underlines: [ ] the past few months have witnessed a sizeable fall in term money market and loan interest rates, which have declined even faster than the key policy interest rate. Consequently, the ECB s monetary policy decisions and liquidity measures have been effective in averting a dramatic contraction in credit volumes (p.10). Figure 2 illustrates the reduction in the rate on the main refinancing operations of 8 See figure 2. 10

11 325 basis points since October Brunetti (2011, p.110) points out that the combination of traditional and unconventional monetary policy contributed support to the economy but simultaneously, ECB s balance sheet was extended to an unprecedented dimension. Also De Grauwe (2009, p.219) states that an important implication of ECB s liquidity injections was a massive expansion of the balance sheet of the Eurosystem. Illustration 3 shows the evolution of ECB s balance sheet which has more than doubled compared to the situation in 2007, prior to the credit crisis. However, the implementation of such a policy means incurring various risks 9. Figure 2: Benchmark interest rate of the Euro area (Trading Economics, 2012) The evolution of ECB s collaterals and its proportions is presented in figure 4, where a significant increase in Asset-backed securities, bank bonds and other securities over the past few years becomes obvious and confirms the above described strategies. Due to the involved inflation risks, the ECB decided a rise of the interest rate for main Figure 3: Growth in ECB s balance sheet Figure 4: Proportions of financial (Barber, 2011) guarantees (Brendel & Pauly, 2011) 9 De Grauwe (2009, p.220) lists two main problems: 1) National governments are responsible for losses on assets held by the Eurosystem and 2) the massive liquidity injections might have inflationary consequences (in the longrun). 11

12 refinancing operations to 1.5% at the beginning of July An inflation of 2.6% 10 in the same time frame compared to prices in July 2010 clearly indicates a deviation of ECB s long-term inflation target of just below 2%. Moreover, the observation of inflation rates above 2.6% in the antecedent months 11 justifies high expected inflation in the following time. Therefore, a consideration of real interest rates is a more appropriate way to examine certain shocks and impacts for the Swiss economy. Since (ex post) real interest rates according to Blanchard & Illing 12 (2006, p.414) result from the difference between nominal interest rate and inflation, real interest rates are lower the higher the inflation given the nominal interest rates. Decreasing real interest rates in the Eurozone can be observed in the illustration below (figure 5) where rates in Switzerland and in the Real interest rates [%] Switzerland Euro area source: Handelszeitung Figure 5: Real interest rate spread Switzerland Euro area Euro area diverge during the summer months. Switzerland suffers from high interest rates making domestic investments more expensive. The interest rate gap between Eurozone and Switzerland is a long-lasting effect. Handelszeitung (2012) denoted real interest rates in Switzerland of 1.8% in March 2012 and -0.74% in the Euro area at the same time. 10 Harmonized Indices of Consumer Prices (HICPs) illustrate inflation growth rates in the Euro zone relative to the accordant month of the previous year (M/M-12). Further tables and graphs are available on 11 See footnote 10, June: 2.7%, May: 2.7%, April: 2.8%, March: 2.7%. 12 For nominal interest rates and inflation below 20 %, the relation can be described by the approximation r t = i t π t, see Blanchard, O. & Illing, G. (2006). Makroökonomie, 4 th Edition, pp for more information. 12

13 3.2 Implications on Swiss Franc Euro exchange rate A comparison of European and Swiss real interest rates offers valuable clues to the question how and in what dimension Europe s monetary policy appears as an exogenous shock for the Swiss economy. The observable discrepancy is expected to influence the CHF/Euro exchange rate. If this was not the case, the no arbitrage condition expressed by the Uncovered Interest Rate Parity UIP (equation 1) would not apply. After 6 September, the exchange rate was set a minimum value of 1.20 CHF by the Swiss National Bank SNB 13, followed by a diminishing volatility of the exchange rate. An observable appreciation of the Swiss Franc in July and August becomes visible considering figure 6 below January 1999 = 100, source: SNB Figure 6: Real exchange rate indices Swiss Franc Euro area (export-weighted) This appreciation 14, severely affecting the Swiss export industry, could indicate a significant exogenous shock from the Eurosystem. Assuming the decrease in real interest rates in the Eurozone to be this significant impact, the following part explains the recently occurred dynamics of the Swiss Franc Euro exchange rate by an analysis using the Dornbusch overshooting model from chapter 2. The analysis will focus on the decrease in real interest rates as initial, long-lasting shock and will rule out other possible causes. Since capital is highly mobile in European countries and a large number of arbitrage seeking agents are expected to operate in the financial system (i.e. Carry Traders), a quick reaction of agents due to interest and exchange rate movements 13 See press release of the SNB from 6 September 2011: Swiss National Bank sets minimum exchange rate at CHF 1.20 per euro, available on: 14 Observing nominal values, an exchange rate of SFR/EUR in April 2011 is followed by a minimum value of SFR/EUR just less than five months later (see 13

14 seems self-evident. Hence, the definition of ECB s monetary policy as a significant exogenous shock may be justified. 4. Exchange rate dynamics in the Dornbusch model following an unexpected foreign interest rate shock 4.1 Adjustments of the Dornbusch model It is necessary to modify the Dornbusch model from chapter 2 for a better approach with regard to the current economic environment. Switzerland as the domestic country is seen as a small open economy. Thus, trade is supposed to have an important impact and the price elasticity of net exports is assumed to be high. A higher value of parameter δ implies a significant exchange rate or price sensitivity of aggregate demand. With δ>1, the Goods market curve (GM) as well as the Asset Market curve (AM) are positively sloped. A comparison of both slopes 15 shows that the AM curve is flatter than the GM curve but its slope converges to one as δ becomes infinitely high. Due to a positive AM slope, the directions of motions in the phase diagram change as seen in the subsequent figure and accordingly, the stable arm has a positive slope. Note from chapter 2.3 that a higher level of domestic prices decreases domestic interest rates, since the effect of lower money demand following lower net exports and output dominates in the case of δ>1. e GM AM e* p* p Figure 7: Phase diagram with flexible exchange rates (assuming δ>1) (author s construction) 15 The slope of the GM is given by equation (9): ( δε+ϵ δε ) > 1, the AM slope is part of equation (10): - ( 1-δ δ ) = 1-1 δ < 1. 14

15 The exchange rate has to depreciate (an increase in e n ) to prevent r t from falling below foreign rates r f because an increase in aggregate demand and output raises transaction demand for money until the downward pressure on domestic interest rates is eliminated and r t = r f in the long run (Sørensen, 2003, p.39-40). This mechanism explains the upward-sloping AM curve. Furthermore, a modern small opened economy is faced with new payment methods that have an influence on parameters in the equations. The LM curve can be written as m p t = φy t εr t alternatively 16. Parameter φ describes the income elasticity of money demand. In a study for the Oesterreichische Nationalbank, Knell and Stix (2005) analysed this elasticity across different OECD countries with empirical results and state that: The fact that income elasticities for non-us countries are significantly higher suggests, however, that the accuracy of specific money demand theories depends on the particular characteristics of a country. [ ] the dissemination of payment cards is likely to have a significant impact on the demand for narrow money (p.530). According to their study, true income elasticities 17 for 16 OECD countries apart from the US 18 lie in the range from 1.0 to 1.3. Operating with a φ-value of one in the subsequent part seems to be a justified approximation for Switzerland. Additionally, equation (8) from chapter two will be replaced by the following expression (11) because r f cancelled out by inserting (6) in (1). e n = ( δ ε+ϵ )en + ( 1-δ ε+ϵ )p t + ( u-m ε+ϵ ) - rf (11) By setting e n = 0 (in order to describe the long-run equilibrium) and solving for e n as endogenous variable, we get the adjusted AM curve. e n = ( m-u ) δ - (1-δ δ )p t + r f ( ε+ϵ ) δ (12) Since the initial shock refers to real interest rates affecting the exchange rate, it is reasonable to still focus on real indices. 4.2 Technical analysis and the path of a foreign interest rate shock An unanticipated decline in foreign interest rates r f shifts the AM curve (12) downwards whereas the GM curve (9) remains unaffected. Figure 8 illustrates the movement of the AM curve. According to Dornbusch (1976, p.1166), p t is a predetermined state variable 16 In Chapter 2, LM curve was defined as m pt = y t d ȳ ε(r t r f ) which implies a φ = Narrow money consisting of money M0 and M1 according to Knell and Stix (2005, p. 531). 18 True income elasticities for the US show values between 0.4 and 0.5 (narrow money). 15

16 but e n is free to jump 19. Before time t 0, the economy is at point E 0. If the (real) exchange rate (e r ) jumps, it must be at time t 0 when the shock occurs. Sooner or later, the economy must be on the new saddle path. As shown in figure 8, e r jumps to point A and converges to the new equilibrium E 1. In the long-run, the price level has decreased and the exchange rate has appreciated compared to the initial situation. e GM AM 0 e 0 E 0 AM 1 undershooting e 1 S E 1 A S p 1 p 0 Figure 8: Technical analysis (δ>1) (author s construction) p 4.3 Economic intuition and comparison with empirical results The economic intuition of the above mentioned movements is pointed out in the following part. An unexpected as well as an expected decrease in (real) foreign interest rates appreciate the domestic currency in the long run. As written before, the decrease is not announced ex ante and the shock occurs unanticipated. Agents were aware of an increase 20 in nominal interest rates at the beginning of July, but current inflation is hardly observable so that movements of the real interest rates are assumed to be unexpected. Agents have perfect foresight and instantly sell foreign currency (buy domestic currency) in order to enjoy a capital gain. This fact immediately drives the (real) exchange rate down to a lower level. There are two counteracting effects at the same time: A decrease in r f makes foreign bonds less attractive relative to domestic substitutes which tends to further decrease the exchange rate according to the UIP (1) because investors will purchase the domestic currency in order to buy domestic bonds. The second effect results from the decrease in the exchange rate which decreases ceteris paribus real exchange rate. As indicated in IS equation (3), demand for domestic goods and thus real 19 As mentioned earlier, agents immediately form expectations about the exchange rate and are assumed to be fully rational. 20 See press release from European Central Bank, Monetary policy decisions, 7 July 2011, 16

17 output will decline since it becomes more expensive to import from the domestic country. Due to a lower (real) output, demand for money decreases which reduces the domestic interest rate (Blanchard & Illing, 2006, p.113). A decreasing domestic interest rate is expected to have the opposite effect on the exchange rate as the initial shock. Since δ>1, the effect of an appreciation dominates but as seen in figure 8, the exchange rate initially undershoots its new long-run equilibrium. At point A, y t < ȳ because of the lower demand for domestic goods after a domestic appreciation. Implied by equation (4), the price level begins to fall 21 and the competitiveness of domestic goods improves. Consequently, real output begins to increase when the economy tends to converge towards the new equilibrium point E 1. In the case of δ>1, declining prices also tend to increase domestic interest rates (see chapter 2.3, detection of dominance) which make domestic bonds more attractive. Therefore, a further appreciation of the Swiss Franc should be observable according to figure 8, together with a decline in prices. Both effects tend the economy to converge along the saddle path from A to E The appreciation in the short run followed by another appreciation from A to E 1 implies an initial undershooting of the exchange rate as seen in figure 8. Empirical observations are expected to prove and underline the theoretical process. Beginning chronologically, an immediate appreciation of the Swiss Franc should be observable after the decrease in real interest rates occurs. Indeed, between 14 and 21 July 23, a sharp decline in Euro interest rates occurs (from +0.24% to -0.04%, see figure 5) and the real exchange rate appreciates at the same time (see July 2011, figure 6). Change in percent [%] GDP by type of expenditure - real, source: SNB Figure 9: Change in GDP (Switzerland) from the corresponding quarter of the previous year Note from chapter 2.2 that the Dornbusch model assumes sticky prices in the short-run. Therefore, prices are expected to decline with a time lag. 22 The long-run equilibrium holds that y* = ȳ and r* = r f. 23 This week is considered as time t0 when the initial exogenous shock occurs. 24 Quarterly data are not seasonally adjusted, at prices of the previous year. 17

18 Since domestic goods become more expensive for foreign buyers, real GDP of a small opened economy should decline as well. Empirical data shown in figure 9 confirm this effect and a declining real GDP after July is visible. According to a diminishing GDP, money demand decreases which reduces domestic interest rate. Figure 5 underlines that real interest rates in Switzerland moved downwards in the same time (from 1.3% to 1.0%). This impact on the exchange rate is dominated by the appreciation (due to δ>1) but an undershooting occurs. Declining prices as a consequence become apparent in figure 10 where consumer prices converge to a lower level after September Those movements imply a certain time-lag which is consistent with the assumption of sticky prices in the Dornbusch model. Together with decreasing prices, domestic goods become more competitive and real output is expected to increase. Price adjustments as well as an increase of y t occur with a delay but such a trend in GDP is not indicated given the most recent data available in figure 9. However, the higher money demand leading to higher domestic interest rates can be illustrated in figure 5. In September, when prices significantly begin to decrease, a simultaneous increase 25 in 4 Change in percent [%] European Union Switzerland Consumer prices in Switzerland and the European Union, source: SNB Figure 10: Change in consumer prices from the corresponding month of the previous year real interest rates takes place. As indicated in chart 8, the initial undershooting of the exchange rate is followed by a further appreciation of the Swiss Franc. The available data support this process only for a certain period of time. In the second half of August, the real exchange rate shows a turnaround and begins to depreciate which contradicts the model (see figure 6). This development of the real exchange rate rather corresponds to an overshooting. The initial undershooting (as graphically explained in figure 8) is supported by empirical data in the short-run. The model predicts a further appreciation contemporaneous to increasing domestic interest rates and declining prices over time August: 0.36%, 1 September: 0.5%, 8 September: 0.56%, 15 September: 0.83%. 18

19 (after t 0 ). The latter effect is well demonstrated by the chart above whereas the appreciation ends mid August and hence, empirical facts limit the applicability of this version of the Dornbusch model. Nevertheless, the real exchange rate has appreciated compared to the first quarter in 2011 and consumer prices are converging to a lower level exhibiting a time lag. This adjusted model does not generally turn out to be wrong since the main movements and long-run results can be empirically confirmed (provided that data are available yet). It is important to mention that the exchange rate was essentially influenced by the Swiss National Bank after 6 September (see chapter 3.2) when a monetary counteraction was implemented. Apart from that intervention, there are countless influencing factors that determine the value of exchange rates so that the application of theoretical models in reality always remains challenging. The consideration of a Dornbusch model in which an overshooting occurs might show a better match and is analysed in the subsequent chapter. 4.4 An alternative adjustment implying an overshooting The same unanticipated shock is analysed given the primary model from chapter 2.3 with the same assumptions. According to a δ<1, the Asset Market curve is negatively sloped as shown in figure 1. A price elasticity of exports below one implies that aggregate demand and output react less sensitive to exchange rate changes relative to the previous model in 4.1 (see equation (3)). Tressel and Arda (2011) examined amongst others the impact of the exchange rate on export performance in their IMF Country Report about Switzerland. They conclude that Swiss exports to advanced economies such as Euro area countries are less sensitive to real exchange rate movements (Tressel & Arda, 2011, p.11). According to their studies, the real exchange rate (RER) elasticity of exports to the Euro area is estimated at 0.4% whereas the RER elasticity to non-euro area is much higher (70%). Results of those regressions support the assumption of a lower export and output sensitivity with respect to exchange rate movements. Therefore, the case of δ<1 seems to be a more appropriate way of explaining the Swiss Franc dynamics. Again, the economy is assumed to be in the steady state before t 0 and the AM curve is described by equation (12). Directions of motions are obvious in figure 1 and movements of the e n =0 curve in the original model are presented in figure 11. The technical analysis is quite similar to the one in part 4.2: The unexpected decline in r f shifts the AM curve downwards with e n as jump variable and p t as predetermined state 19

20 variable. The real exchange rate jumps at time t 0 when the movement of r f takes place. In the unannounced case of a shock, the economy jumps from E 0 to A on the new saddle path and converges to the new steady state E 1. Again, price level in E 1 is lower and the exchange rate has appreciated in the long-run. The economic intuition begins similar to the first case. Agents realise the real interest rate shock and know the long-run e AM 0 GM AM 1 E 0 e 0 overshooting e 1 S E 1 A S p 1 p 0 p Figure 11: Technical analysis (δ<1) (author s construction) consequences with a lower steady state value of e n. Thus, they sell foreign currency because of a possible capital gain and the exchange rate immediately appreciates. An interest rate spread further decreases e n since Swiss bonds are relatively more attractive for investors. This appreciation has the same effect as described in 4.3 and according to equation (3), demand for domestic goods and output decline. Point A exhibits an y t < ȳ which reduces prices as implied by the price adjustment equation (4). The main difference with respect to the process in chapter 4.3 results from a δ<1: Decreasing prices increase real money supply (m-p t ) and thereby decrease real interest rates as described in chapter 2.3. Falling prices enhance the economy s competitiveness and output as well as net exports begin to increase as the economy converges from A to E 1. Lower real interest rates in the same period of time make foreign bonds more attractive and imply a moderate depreciation of the domestic currency. Since net exports are not so sensitive to exchange rate movements (δ<1), the initial appreciation of the Swiss Franc must be large enough to generate the decline in aggregate demand which decreases prices as consequence (Sørensen, 2003, p.40). In general, this version describes a stronger initial appreciation from E 0 to A followed by a depreciation (A to E 1 ) in order to reach the new steady state. The resulting overshooting contrasts with our first version of the Dornbusch model where the effect of an undershooting was observable. 20

21 Empirical data confirm an initial appreciation mid July following a significant decrease in real interest rates (Euro area) as explained in the previous part. Consequently, real GDP exhibits a turnaround and begins to decrease as shown in figure 9. A decline in consumer prices after July and September becomes obvious in figure 10. Due to a risen money supply, domestic real interest rates should decrease which is the opposite effect compared to the first version. Such movements can be proved by figure 5: The chart denotes increasing real interest rates partly in August and September, but apart from that period, the overall direction is negative 26 after July Changes in consumer prices converge to an annual decline of -1% which seems to be a long-lasting development. Although figure 9 consists of the most recent GDP data available from the Swiss National Bank, it does not illustrate a rise in real GDP in Switzerland. However, the Swiss research institute BAKBasel (2012) forecasts a higher GDP growth for 2012 and recently revised its prediction by +0.3%. They expect an annual (nominal) GDP growth of 0.7%. Parallel to a (constant 27 ) low domestic inflation, we might expect a certain GDP improvement relative to the initial economic prediction in the last quarter. The currency depreciates due to lower interest rates as seen in figure 6. Mid August, the Swiss Franc appreciation reaches its peak and the downturn begins. The real exchange rate converges to a stable value (October 2011) which is still higher than the initial rate as shown by the graph. 4.5 Explanatory power of the model and evidence of rationality A critical view Both versions of the Dornbusch model exhibit equal results: A lower price level and a lower value of e n in the new steady state E 1. This long-run output of the Dornbusch model is supported by empirical data and beyond dispute. The process on the other hand differs and both versions explain opposite movements of certain indices. The value of parameter δ causes the difference between both models. Decreasing prices as explained in part 2.3 might have opposite effects on interest rates corresponding to parameter δ. Movements of the interest rates are thereby crucial for the detection of an undershooting or an overshooting. The decisive empirical facts that support the version of an overshooting are the motions of real interest rates, the development of the real exchange rate and the sensitivity of net exports. In spite of a spread in real interest rates, 26 Real interest rates in Switzerland: 7 July: 1.3%, 27 October: 0.5%. 27 Last reported inflation in Switzerland: -0.9%, see Handelszeitung No. 12, 22 March

22 chart 5 provides evidence for an overall decline in domestic interest rates. Furthermore, a strong appreciation followed by a (smaller) depreciation clearly indicates an overshooting rather than an undershooting. Together with the results from Tressel & Arda (2011) who found a certain inelasticity of net exports to the Euro area, the second version of the Dornbusch model (δ<1) with an overshooting effect may be confirmed as the better approach for an explanation of the recent Swiss Franc exchange rate dynamics. The basic assumption of rationality can be affirmed since empirical data are consistent with the model based on that principle. A strong indication of rational behaviour e.g. is suggested by chart 6 with arbitrage seeking agents who drive down the exchange rate. Despite the approval of the Dornbusch model, it is important to question the explanatory power of this analysis. Many papers investigated for example whether the UIP holds in reality or not. Chinn and Meredith (2005, p.16) concluded that interest rate differentials were bad predictors of short-term exchange rate movements and that even in the long run only a small proportion of exchange rate variances was explained. Rogoff (2002, p.5) mentioned that Rüdiger Dornbusch s explanation for an overshooting had proven of relatively limited value empirically. Moreover, it may be doubted if prices and wages are flexible in the long-run in the economy as implied by the model. Alternative shocks occurring at the same time were ruled out in the previous analysis even though exchange rates are variably influenced by many other factors e.g. beliefs about the Euro currency and the global position of the Swiss Franc as safe haven 28. Finally, it may be questioned if the Swiss economy was in a Steady State before the shock occurred as assumed in the model. After the financial crisis and a significant economic downturn, scepticism concerning the Steady State starting point seems legitimate. 5. Implications for economic policy Independent of the shock considered in the previous chapters, Rüdiger Dornbusch s model has important implications for economic policy. The crucial point is the concept of flexible exchange rates. Dornbusch (1979, pp.22-23) points out that inflation stabilisation becomes more difficult under volatile exchange rates in countries that have a soft currency whereas hard currency countries benefit. Since exchange rates are expected to 28 For further details: Flight to safety, The Economist, 20 July 2011, available on: 22

23 react and adjust immediately, monetary policy exerts instantaneous inflationary pressure in expanding countries and causes inflation-dampening in rather tight countries. Dornbusch (1979, p.23) further states that monetary policy becomes less effective if agents are aware of increasing domestic prices due to the inflationary pressure of the depreciation. A higher domestic price level limits the benefits from the depreciation, namely the competitiveness. Another important implication is the assumption of sticky prices which enables exchange rates to be a tool for monetary policy. Dornbusch (1979) concludes: Monetary policy under flexible rates and high capital mobility works not only by affecting the interest sensitive components of aggregate demand but also by increasing net exports (p.47). Furthermore, if monetary policy influences beliefs about the exchange rate, adjustments of the rate have to be even more pronounced (Dornbusch, 1979, p.49). If flexible exchange rates are seen as a tool for monetary policy, it is necessary to mention the impacts on import prices which are directly affected. A monetary expansion depreciates the domestic currency and export goods become less expensive for foreign importers but increase at the same time prices of imports for the domestic economy. Increasing import prices spill over into consumer, wholesale and producer prices and have an effect on domestic inflation. Dornbusch (1979) points out: The more rapid and the more substantial the spillover of import prices into domestic prices the more inflationary is monetary policy and the less effective it is with respect to aggregate demand (p.49). Additionally, the higher the sensitivity of exchange rates to monetary policy, the steeper the Philips curve when monetary policy serves as a means to influence real output (Dornbusch, 1980, p.177). In this case, a change in interest rates significantly affects exchange rates and hence domestic inflation through changes in import prices. A steeper Philips curve implies that unemployment is less sensitive to monetary policy. Thus, certain level of unemployment may be seen as given, independent of monetary interventions. Flexible exchange rates also entail considerations of income policies. Dornbusch (1982, p.10) mentions the welfare that is influenced by real exchange rate movements. Fiscal policy could serve as a stabilisation and accommodates the changes in the standard of living. Chapter 2.1 described Dornbusch s theory as an expectation based model meaning that agents have beliefs about the exchange rate. This implies that exchange rate movements are driven by people s assumptions and expectations even if they prove false. Indeed, Dornbusch (1982) states: But many of the disturbances in the world economy are not the result of actual changes in fundamentals but rather of changes in 23

24 expectations about the future course of these fundamentals (p.19). Hence, exchange rates may exhibit paths that do not promote macroeconomic stability. The researcher concludes that exchange rates sometimes deviate from their fundamentals (p.20). A related problem is called peso problem, meaning that beliefs about the possibility of regime changes are a source of disequilibrium exchange rates (p.20). It can be concluded that very flexible exchange rates might have severe consequences for monetary policy interventions since their values are governed by expectations. 24

25 6. Conclusion According to recent exchange rate disturbances, the key issue was the question whether the flexible exchange rate model from Rüdiger Dornbusch could explain the Swiss Franc appreciation in summer In fact, the analysis of the model can serve as an explanation for the development of several macroeconomic indices and the implications are consistent with empirical data. The occurred overshooting with regard to the Swiss Franc appreciation became as obvious as impacts on domestic output (which declined), on consumer prices (converging to a lower level) and on real interest rates (decreasing as well). It is necessary to assume a lower price elasticity of net exports since the model otherwise presents an undershooting and predicts movements of real interest rates that are not consistent with real occurrences. The applicability of the Dornbusch model implies rationality in the economy and confirms that the observed real interest rate spread as initial shock has a significant impact on the relationship Swiss Franc - Euro. Nevertheless, there are some questions remaining. Dornbusch s concept gives reasons for short term developments but attaches less weight to the transition from temporary adjustments to long-run states and provides little information about harmonisations over a longer period. Other problems, such as the lack of validity with regard to the Uncovered Interest Rate Parity (UIP) were highlighted in chapter 4.5 and restrict the explanatory power of the model in general. When the Swiss National Bank set a minimum exchange rate of 1.20 CHF/Euro at the beginning of September, the value was significantly influenced and the model s expressiveness may be faced with a further limitation. Despite of various constraints, the analysis proves that Dornbusch s (1976) famous flexible exchange rate model can be applied in order to explain the dynamics of the Swiss Franc last year and helps to predict the directions of movements of main indices. Dornbusch s intention was not the creation of an instrument that perfectly predicts future exchange rates but to develop a tool which helps to understand why the rates could be so volatile in the short-run and how they adjust. Finally, the Dornbusch model serves as a good approximation of a complex, flexible system whose output, the price of a currency, is determined by countless factors and variables. 25