Task 1 exchange rate forecasting models

Byapplying regression analysis for Purchasing Power Parity model, Uncovered Interest Rate Parity, andMacro-based model, we come up with the forecasted exchange rate EUR/SEK based on each

Table of Contents

Executive Summary 0

Introduction 1

Task 1: Exchange rate forecasting models 1

Data Source 1

Purchasing power parity 2

PPP Model construction 2

PPP Regression result & analysis: 3

PPP Residual Plot: 4

PPP Forecasting for Q1/2021: 4

Uncover Interest rate Parity (UIP) 5

UIP Model Construction: 5

UIP Regression result & analysis: 5

UIP Residual Plot: 6

UIP Forecasting for Q1202: 6

Macro-based Model 7

Macro-based Model construction 9

Macro-based regression result & analysis: 9

Macro-based Model forecasting for Q1/2021: 10

Task 2: Composite Forecasting 11

Task 3: Qualitative Factors 12

Sweden policy in terms of QE program 12

Sweden’s labour market 12

Economic Growth in Covid-19 Pandemic 13

Sweden import and export policy 13

References: 14

Executive Summary

The purpose of this report is to calculate and forecast the exchange rate of EUR/SEK for Quarter 1 in

2021 The data set used in this report is collected from Quarter 2 of 2003 to Quarter 3 of 2020 Byapplying regression analysis for Purchasing Power Parity model, Uncovered Interest Rate Parity, andMacro-based model, we come up with the forecasted exchange rate EUR/SEK based on each model, with10.6964, 10.388 and 10.5353, respectively The results derived from 3 models indicates that the exchangerate of EUR/SEK will be depreciating To generate a more trustful forecasted exchange rate EUR/SEK,

we decided to construct Composite Forecasting Model, which is the combination of 3 models above, andproduce the final exchange rate EUR/SEK in Q1/2021, with the result of 10.5415

To support the forecasting results from the models, we also discuss the qualitative factors and events affecting the depreciation of EUR/SEK, which are Sweden policy in terms of QE program, Sweden’s

labour market, Economic Growth in Covid-19 Pandemic, Sweden import and export policy.

According to Reinhart and Reinhart (2001), the exchange rate is determined as the value of acurrency of a nation in terms of the other country’s currency Having a deep understanding about thecauses affected by the exchange rate, investors could seize opportunities and receive profit bypredicting how it will change in future In this report, the exchange rate of SEK/EUR in Q1 2020 isforecasted by applying quantitative and qualitative methods

In quantitative methods, there are four models used in order to anticipate the exchange rate The firstmodel, we used the inflation rate differential between Sweden and Euro area, known as PPP theory.The next model is interest rate parity that uses the differential of interest rate of two countries Thethird model attached six macro factors including inflation, interest rate, oil price, GDP growth rate,money supply and current account balance Thus, in the last model, the most significant variableswill be chosen by using the backward elimination

In terms of qualitative, the movement Sweden and Euro area exchange would be forecasted byanalyzing the relevant economic events

Task 1: Exchange rate forecasting models

Data Source

Table 1: Data sources used in our report

1 | P a g e

Purchasing power parity

Purchasing power parity (PPP) is an economic theory which uses the “basket of goods” approach inorder to compare different countries Purchasing Power Parity (PPP) tends to center around theaggregate price levels whereas the law of one price (LOP) focuses on an individual level (basing onSorkin 2012) According to International Monetary Fund Research (1976), PPP among two countriesincludes two versions such as absolute PPP and relative PPP Absolute PPP is considered as the ratio

of the countries' price levels or the product of the exchange rate in a base period Relative PPP is theratio of the countries’ price indices According to Flood and Taylor (2005), absolute PPP supposesthat the exchange rate is the same with the ratio of 2 national price levels while relative PPP showsthat the shift in the exchange rate accompanies the shift in relative national prices Therefore, acountry suffering higher inflation will hope to decrease its currency

The formula for absolute PPP: st = (pt - pt*) + t

The formula for relative PPP: st = (pt - pt*) + t

In which:

St : The spot exchange rate (Domestic price of foreign currency) at time t

pt : The domestic price level at time t

pt*: The foreign price level at time t

% S: The percentage change in the spot exchange rate

% P: The percentage change in the domestic price level (Domestic Inflation)

% P*: The percentage change in the foreign price level (Foreign Inflation)

0: The arbitrary constant term (the intercept)

1: The coefficient of the regressors

: The error (residual) term

Apply to this case, considering the domestic currency is Euro (EUR) and the foreigncurrency is Sweden (SEK), the regression model would be indicated in terms of difference inflationsbetween the Euro Area (Home country) and Sweden (Foreign country) (basing on Hauner, Lee andTakizawa, 2010):

% EUR/SEK = 0 + 1 * (Inf Euro Area + Inf Sweden ) +

In which:

Inf Eu: Inflation rates in the Euro Area

Inf Sweden: Inflation rates in Sweden

Figure 1: Regression result of PPP

PPP Regression result & analysis:

According to figure 1, the relationship between percentage change of exchange rate and inflation rate

of Euro area and Sweden is:

- The intercept coefficients �0 is 0.0045 that is expected that the value of exchange rate between two countries will be 0.0045 when there is no different between their inflation

- The slope of coefficients is - 0.00537 has showed the negative relationship between theexchange rate of EUR/SEK and inflation rate differentials It will be explained that a 1%increase in inflation rate will lead to an increase in % change of spot rate by 0.00537%

- The R-square is 0.031 which means there are only 3.1% of percentage change of exchange rate variation is explained by inflation rate differentials between Sweden and Euro area

- The P-value is extremely high (14.5%), much higher than 0.05 implies that the nullhypothesis is true The results were not statistically significant Hence, there is norelationship between two variables

3 | P a g e

PPP Residual Plot:

Figure 2: Inflation differenntial residual plot

The residual plot diagram displays the most of errors are randomly dispersed around the horizontal axis and fluctuate around -0.08 and 0.006 which is relatively low, indicating a good fit model

PPP Forecasting for Q1/2021:

Predicted

differential exchange rate

Uncover Interest rate Parity (UIP)

The general concept of the Interest Rate Parity (IRP) explains the changes of the exchange rate related tothe nominal interest rate differential between two countries Furthermore, we have two different

covered and uncovered interest arbitrage With the uncovered interest arbitrage, we will predict the

future spot rate and also use that future sport market to exchange currency Hence, due to we are not

hedged against possible risk, our profit gain or loss will be affected by market exchange rate

Otherwise, with covered interest arbitrage, after having the prediction for future exchange rate, we

will have a forward contract at which we will sell currency in the future Nevertheless, using a

covered interest arbitrage can limit the potential risk and make profits from the market

UIP Model Construction:

The equation of UIP model to be examine will be:

UIP Regression result & analysis:

Figure3 UIP Regression Model

The Equation for UIP Regression is that:

EUR

It can be analysed from the UIP regression result that:

The intercept β0 is −0.006617827interest rates of both regions valued at 0 at time t will lead to the fall in the change ofexchange rate EUR/SEK by −0.66 %

5 | P a g eThe coefficient β1 is

of the both interest rates increases by 1% at time t, the change of the exchange rate will alsoincrease by 0.82% in the same period.

The R-squared is calculated at 0.039796375 , showing that about 3.97% of the variations

of the change of the Exchange rate between EUR and SEK can be explained by the change ofthe both interest rates This can be inferred that there is an insignificant relationship betweenchange in exchange rate and interest rate differential between EUR and SEK

The p-value is recorded atsignificant level 5%) illustrate that there is no specific relationship between the change of theexchange rate and interest rate differential between EUR and SEK

UIP Residual Plot:

Figure 4 IR differentials Residual Plot.

From the UIP regression model, the IR differential Residual Plot is provided for testing whether or

not the model is good fitted According to the data from the figure, the residuals value ranging from

-0.05 to 0.08, which is considered relatively low rate, represents the UIP model is highly accurate

Time (Quarterly)

2020 Q4

2021 Q1

Table 3 The UIP model exchange rate forecast.

Due to the booming of the COVID-19 around the world, which leads most of the European region

suffering from one the worst recession in the economy This urged the European Central Bank decisively

reduce the interest rate down to 0% or lower in order to minimize the loss from the sudden

pandemic Also, from the data provided by Trading Economics, they predicted that the interest ratewill remain 0% until 2021, causing interest rate differential of EUR and SEK will remain 0.Therefore, the change of EUR/SEK will maintain at decreasing rate at –0.66% for each quarter untilthe end of 2021 From the calculating, EUR/SEK is predicted to fall at the value of 10.5269, as theEUR is expected to depreciate against SEK.

Macro-based Model

Apart from the PPP and UIP models for forecasting the exchange rate (ER), there have been severalstudies on forecasting the ER using monetary fundamentals According to Mark and Sul (2001),there is an increasing magnitude of ER return regression slope coefficients and R squared as theprediction horizon increases Simultaneously, there is also an improvement in forecast accuracyusing out-of-sample monetary fundamentals relative to the random walk under the same condition(Faust, J et al., 2002, p 36) Based on the two statements, we have constructed a multifactorregression model using the following variables:

1) Short term Interest rate:

The effect of interest rate on ER can be expressed through either Chicago view (asset marketapproach) and Keynesian approach (international goods market approach) (Hüseyin Şen et al., 2019).Regarding the Chicago view, under perfectly flexible prices, a change in market interest rate isprimarily affected by the expected rate of inflation If domestic interest rates were higher than therest of the world, the increases in expected rate of inflation would also be expected by the marketparticipants, putting pressure on the exchange rate to increase (due to the depreciation effect ofinflation) On the other hand, the Keynesian view suggests that a higher interest rate would create anattractive market for capital inflows into the host country (as the returns for investment of foreignersincreases), leading to an appreciation of the domestic currency through a rise in demand of thedomestic currency

2) Inflation rates:

Several studies conducted by Giavazzi and Giovannini (1989), Velasco (1996), and Dornbusch(2001) have proved that in a stable exchange rate regime, not only the price is stable, but theefficiency of monetary policy is also increased Under that condition, a country with a relativelyhigher inflation rate against the rest of the world is likely to suffer from external deficits, leading to afall in its foreign exchange reserves (Hüseyin Şen et al., 2019)

3) Percentage change in oil prices:

7 | P a g e

The connection between the oil market and currency market has long been established Earlyresearch on the role of oil prices by Golub (1983) and Krugman (1983) revealed that when oil pricesrise (fall), an oil-exporting (oil-importing) country may witness an appreciation (depreciation) in itscurrency (Reboredo, J C 2011, p 420) Furthermore, study on oil price movements’ impacts onexchange rate by Bloomberg and Harris (1995) based on the law of one price highlighted that sinceoil is homogenous and internationally traded in USD, a depreciation in USD reduces the oil price inforeigners’ perspective, hence increasing their purchasi ng power and oil demand, pushing up thecrude oil price in USD (Reboredo, J C 2011, p 420).

4) Relative change in real GDP:

The effect of real GDP has a different effect on the exchange rate, depending on which channel isconsidered (trade balance or capital flows) In terms of trade balance, if the growth rate of domesticincome were faster than foreign income, the imports would also grow with a faster pace, leading to ahigher demand for foreign currency (in order to make imports payment) As a result, the domesticcurrency will witness a depreciation

On the other hand, a faster economic growth is associated with a higher return in domestic assets.Therefore, foreign investors will seek an opportunity to invest in domestic assets, pushing up thedemand for domestic currency, hence decrease the exchange rate (appreciation) (Moosa, I, A 2009, p.106)

5) Change in money supply:

The justification for the impact of change in money supply to ER can be expressed through inflation.Theoretically, when a substantial amount of money is injected to the market, it would lead to anincrease in inflation, depreciating the value of domestic currency against other foreign currencies(Sean, M et al, 2018, p 64)

6) Current account (CA) to GDP ratio:

There have been many developments since the original study about exchange rate adjustment to CAimbalances with the price-specific-flow adjustment mechanism under the gold standard by DavidHume in the 18th century In all of these models, regardless of the exchange rate regime, the realexchange rate (RER) movement is found to accompany the adjustment to current account imbalance.For example, a CA surplus following a demand shock will enable the RER to appreciate, eitherthrough nominal exchange rate appreciation or through a rise in domestic inflation, reducingcompetitiveness and exports while favouring imports (Gervais et al., 2015, p 87)

Macro-based Model construction

The model is constructed based on the data in Q1/2003 to Q3/2020 (all variables are at time t-1,except percentage change in EUR/SEK exchange rate), the regression equation of change inEUR/SEK exchange rate is written as below:

(∆ % i EUR −∆ %i SEK )t −1 :relative change in short term interest rate at time t-1 (∆ %π EUR

−∆ %π SEK :relative change in inflation rate at time t-1

∆%oilprice t −1 : percentage change in oil price at time t-1

(∆ %GDPEUR−∆ %GDP)t−1 :relative change in GDP at time t-1

(∆ %M EUR s−∆%M SEK s)t −1 : relative change in money supply at time t-1

(CA/GDP EUR −CA /GDP SEK)t −1 : relative change in current account ratio to GDP at time t-1

ε t : error terms at time t

Macro-based regression result & analysis:

The following figure represents the summary output of the regression model (1):

Figure 5 First Macro-based model regression results

9 | P a g e

)

t−1