Impacts of Sterling Rate Fluctuations on UK Economy Against the Background of 2008 Financial Crisis

中圖分類號：F13/17 文獻標識碼：A

Abstract： This dissertation aims at investigating the impacts of real bilateral exchange rate fluctuations on UK economy as well as the relationship between Pound Sterling Rate Validity and UK GDP growth.

Chapter One： Introduction

Background Information：

On September 15， 2008， with its 500-year history， Lehman Brothers， as one of the five U.S. largest investment banks in the world， declared officially that a credit crisis has been sparked and transferred to the worst financial crisis in history. To be brief， the expansion of the financial crisis is regarded as an outcome of the finance globalization and deregulation monetary policy， indiscreet mortgage and excessive financial creativity. However， all these factors are closely related to real bilateral exchange rate.

Aim：

The aim of this dissertation is to investigate the economic impacts of pound sterling rate fluctuations on UK economy due to the global economic crisis.

Methodology：

Quantitative methods are mainly adopted to explore the impacts of sterling rate volatility on the UK economy. Descriptive analyses through ten figures present the past and current situation as well as future trends of RGDP， RGDP growth rate， RRER， RHCONS， RGCONS， RTB， RKA and RGFCG. Unit root test is then applied to test for nonstationarity of all variables.

Chapter Two： Literature Review

2.1 Impacts of the Real Bilateral Exchange Rate Fluctuations：

Mendelsohn （2009） observed that “Since August 2008， the European Central Bank and the Bank of England have responded to the financial crisis by sharply cutting interest rate， in the hope of boosting economic growth by encouraging consumers and businesses to spend rather than save.”

The depreciation of Pound Sterling enhances UK’s international competitiveness through exports but meanwhile makes imports more expensive. Therefore with higher exports and lower imports it is possible to create higher real GDP growth as well as inflation.

Graph 1 2 – Impacts on Commodity Markets

For example， as it is shown in Graph1 and 2， UK wheat to French has moved from nearly zero in March 2008 to almost EUR -40/t in August 2008. The spread can partly reflect crop quality and availability despite that the GBP-EUR rate is the main determinant. In August 2008， UK wheat futures were around EUR160/ t， whereas French wheat futures were almost EUR 200/t. As a grain net- exporter， UK wheat price competitiveness has been improved due to the benefit of weaker sterling， therefore UK exports to EU members and third member countries have been supported. To conclude， the weaker Pound plays a generally positive role in UK exports as well as farm incomes during the financial crisis. （Mendelsohn， 2009）

2.2 Government Intervention：

Modern economic policy relies greatly on Keynesian economic thought. During the Great Depression at the beginning of the 20th century， the economist John Maynard Keynes argued for a more active government that will adopt interventionist economic policy to affect the economy in times when the market fails or spirals out of control. （Dombusch， Fisher Startz， 2005） It is the guiding principle， now in use in many countries throughout the world， which leads one to believe that government consumption tends to increase during times of crisis.

As UK Government announced a second rescue package for UK banks in its official announcement that “ These further measures are aimed at lending by banks to the economy， which the initial ？400 billion bail-out package in October 2008 failed to induce”. Though the two times of large-scale and sufficient capital injection to the market， difficulties in liquidity of financial institutions has been relieved to some extent. UK government has also cut interest rate for several times which has been mentioned in details above. In addition， many other methods such as reserves and borrowing， reducing inflation have been intervened together as supportive methods. （Hall， 2008）

Chapter Three —— Methodology

3.1 Research Method：

Galliers listed fourteen kinds of methodologies in 1991， from which it is easy to see that there are plenty of research methods. In the matter of exchange rate and economic growth， quantitative method is mainly adopted. However， each method has its advantages and limitations.

3.1.1 Descriptive Statistics and Estimation：

After collecting the initial data from for example International Financial Statistics， ESDS website or National Statistics， clear graphical outputs such as line graphs are drawn to diagnose whether there will be any possible relationship between any two variables or present the current changes and future trends of the variables. Then if it seems to be a linear relationship between these quantitative variables， correlation coefficient should be figured out. It is believed that “Correlation coefficient can examine the strength as well as direction of potential association between variables.”

3.1.2 Unit Root Test：

The Augmented Dickey Fuller test was carried out on the variables to check for unit roots. The presence of unit roots indicates nonstationarity and makes any regression result spurious. In this test， the 1 hypothesis is nonstationarity while the alternative hypothesis is stationarity. The test statistic in this test is the tau statistic. If the tau statistic is less than the critical value at the 5% level of significance then the 1 hypothesis is not rejected and the variable is nonstationary. In such cases， the first difference of the nonstationary variable is taken and then tested again for unit roots. If the first differenced variable is stationary， then the variable is deemed to be integrated of order 1 or I （1）. Also， stationary variables are also termed integrated of order 0 or I （0） （Harris Sollis， 2003）.

3.1.3 Correlation Analysis：

The correlation coefficients of all variables are summarized in a matrix to show whether the correlation between two pairs of variables are positive， negative or none. The degree of correlation will be clearly presented at the same time.

3.1.4 OLS Regression Model：

Data collected from 1992 to 2009 will be tested through Excel to conserve degrees of freedom. By basing the selection of the variables on an economic formula， it is hoped that conserving degrees of freedom will not compromise the fit of the formula. In this study， different specifications of models are also tested in order to check the robustness of the results. It should also be remembered that all variables have been differenced to ensure stationary. This means that any result must be interpreted as short-run relationships （Harris Sollis， 2003）.

3.1.5 Granger Causality Test：

The results of the regression do not strictly imply that the real bilateral exchange rate causes changes in the real gross domestic product. The existence of a relationship does not say anything about the direction of the influence and therefore does not imply causality. However， in time series analysis， it may be possibly to test for causality because after all， time cannot run backward. If the change in the real bilateral exchange rate happens before the change in the real gross domestic product， then it is possible that the real exchange rate could be causing the change in real gross domestic product.

3.2 Data Collection：

Primary data should be collected and gathered on one’s own and observation is of great importance. With careful observation of failures of other research methods， the research findings will be more accurate and reliable. （Adams， Hafiz， Reside and White， 2007） However， secondary data， referring to those collected through works of previous researchers， is the main source for one’s research. In Adams and Hafiz’s view， in spite of the fact that large number of sources can get access to secondary data， notably government official statistics， educational institutions， companies， libraries and the Internet， there are still limitations on information coverage and accuracy.

3.3 The Sources and Analysis Technique：

The data comes mainly from International Financial Statistics and ESDS website which contains all international economic datasets for UK collected by such as International Monetary Fund （IMF）， OECD national account statistics， national statistics office and The World Bank Group. To add， ESDS covers more than 280 countries and 18 groups with relatively the most current and accurate database.

Besides， SPSS software for Windows and Microsoft Excel are the main instruments to analyze descriptive statistics， conduct unit root test， correlation coefficient calculation and judge regression relationships as well as Granger causality test.

3.4 Major Economic Indicators：

Although many variables can， in theory affect real gross domestic product， this study included only the real trade balance， real capital account， real gross fixed capital formation， real household consumption and real government consumption expenditures. This was done both because of lack of continuous data for some other variables and also to conserve degrees of freedom. Data were transformed into real terms through division by the consumer price index to remove the effects of inflation. The data were also all in billion Pounds Sterling to ensure uniformity of data series.

Chapter Four： Research Findings

4.1 Descriptive Analysis：

The real GDP had a mean value of 2.76 billion pounds during the period 1992-2009. The lowest value of 2.12 billion occurred at the start of the period while the highest value of 3.36 billion occurred towards the end of the period in 2008. The recent data then show GDP declining after 2008. This might be a direct result of the financial crisis， as the economy of the United Kingdom is largely based on the finance industry （Yeandle et al.， 2009）. The Real GDP exhibited a slight upward trend of 0.019. In line with this， the real GDP growth rate for the United Kingdom averaged 0.58%. The growth rate reached to the peak at 1.42% in 1994， part of a strong recovery from recession in 1992 （Moser， Pointner and Reitschuler， 2004）. Recently， Real GDP growth rate was actually reduced to -2.4% in 2009， the UK’s worst performance during this period. Revealed the time trend from 1992 to 2009， the real GDP was declining very slightly by 0.012%.

The real sterling rate against the U.S. dollar averaged out at 0.61 in 2009. This means that on the average， 0.61 pound sterling can be exchanged for one U.S. dollar. The real bilateral exchange rate created the weakest record in 2001 and 2002， when 0.706 pound sterling was needed to buy one U.S. dollar. On the other hand， the real bilateral exchange rate was strongest in early 2008 at 0.48 pound per U.S. dollar. It then sharply depreciated in 2009

and is now recovering.

Figure 1 Real gross domestic product of the United Kingdom

Source： International Financial Statistics（the table figures are the same）

Figure 2 Real gross domestic product growth rate of the United Kingdom

Figure 3 Real bilateral exchange rates with the US dollar against the sterling

Real Household Consumption reflects averaged 1.79 billion Pounds Sterling from 1992-2009. It exhibited a substantial upward trend similar to that of real GDP. Like real GDP， real household consumption peaked recently at 2.16 billion pounds but went downhill from that. This recent downward trend is probably also a result of the recent financial crisis driving food and commodity prices up. The United Kingdom is significantly affected because it imports a lot of its food supplies. The resulting increase in inflation has eroded real household income which in turn has stifled consumption （Tang， 2008）. Meanwhile， the lowest value of 1.35 billion took place in 1992， the end of that period’s recession.

Figure 4 Real household consumption expenditures of the United Kingdom

Real Government Consumption hovered at around 0.45 billion from 1992-1998 before steadily increasing to a peak of 0.75 billion pounds sterling in early 2009. The start of the upward trend in real government consumption coincided with the start of the Premiership of Tony Blair.

Figure 5 Real government consumption expenditures of the United Kingdom

Real Trade Balance for the United Kingdom has been on a downward trend since 1998. While it did achieve a maximum of 26.11 million， this happened way back in 1995. The mean of the real trade balance was a negative -50.77 million Pounds Sterling， and most of its values have been negative. In addition， the real trade balance has been worsening， reaching a deficit of 131.53 million Pounds Sterling. According to the IMF Article IV Consultation Concluding Statement in 2008， the state of the trade balance is due to a substantial deficit in the trade for manufactured goods and commodities， as well as food.

Figure 6 Real trade balance of the United Kingdom

As is appropriate for a highly developed country with a large financial sector such as the United Kingdom， its capital account was mostly in surplus from 1992-2009. There is no clear trend as the observations appear to be highly cyclical. The real capital account surplus averaged 3.4 million pounds during the period， with a peak value of 10 million pounds and a lowest value of -6.2 million during one of its rare deficits in 2006.

Figure 7 Real capital account of the United Kingdom

Real Gross Fixed Capital Formation has generally been on an upward trend from 1992 up to early 2008. When reaching a peak of 600 million pounds in 2008， real gross fixed capital formation has dropped sharply， possibly also a result of the financial crisis. This indicator had a mean value of 464 million pounds from 1992-2009.

Figure 8 Real gross fixed capital formation of the United Kingdom （Unit： Billion Pound Sterling）

Through the graphs below it is clearly to see where the real exchange rate and the real gross domestic product is separately， and the real exchange rate and the real GDP growth rate are plotted against each other， it might be tempting to believe that there is no linear relationship between the variables. Real gross domestic product seems to increase without regard to the level of the real bilateral exchange rate， while real GDP growth rate just exhibits the cyclical behavior assumed to be the result of the business cycle. However， visual observation is an unscientific way to conclude relationships. Even though no clear strong relationship can be readily derived from the data， statistical analyses such as correlation and regression should have the final right to decide whether there is any possible statistical relationship.

Figure 9 Real bilateral exchange rates versus the real GDP of the United Kingdom

Figure 10 Real bilateral exchange rates versus the real GDP growth rate of the UK

4.2 Unit Root Test：

As summarized in Table 4.2， it is clearly shown that all variables were found to be nonstationary. Most were found to be integrated of order 1， which means that their first difference transformations have to be taken to attain stationarity. Real GDP， Real Government Consumption， and Real Gross Fixed Capital Formation were found to be integrated of order 2. The widespread finding of unit roots in the data means that the first difference transformations have to be used for both correlation and regression analysis. On the other hand， the finding of different orders of integration also makes it impossible for the Engle-Granger co integration test to be carried out on the real GDP and the real bilateral exchange rate. Co integration testing requires that two variables be integrated of the same order. Since RGDP was found to be integrated of order 2 while RER was found to be integrated of order 1， a long run equilibrium relationship cannot exist between the two variables （Harris Sollis， 2003）. Further analysis on the relationship of the two variables will have to be limited to short-run correlation and regression analysis of their first differenced， stationary transformations. Here it is important to note that the first difference forms of the variables are their interpreted as their growth in value from the previous year to the next.

Table 4.2 Augmented Dickey-Fuller test

4.3 Correlation Analysis：

As all the variables exhibited nonstationarity， their first differenced transformations were used in the correlation matrix presented in Table 4.3. Using nonstationarised variables might have led to artificially high correlation coefficients. A “d” was added to the variable names to denote the differenced， stationary transformation. Table 4.3 summarizes the correlation coefficients of all variables. This provides a measure of their linear relationship on each other （Cohen， 1988）. As can be seen in the table， the most correlated variables are dRGDP and dRHCONS， and dRGFCF and dRGDP. There is medium positive correlation between the two pairs of variables. The rest of the variables had mostly small to no correlation with one another. None of the variables were found to be so highly correlated with each other to warrant dropping them from the subsequent regression. Multicollinearity is unlikely to be a problem. Also， the variables of interest， dRER and dRGDP are shown here to be negatively correlated with each other， having a correlation coefficient of -0.45. For instance， real exchange rate increasing 1% may generally contribute to 0.45% decrease in real gross domestic product.

Table 4.3 Correlation Coefficients of the First-Differenced Variables

5% critical value （two-tailed） = 0.2387 for n = 68

4.4 OLS Regression Model：

Although there are many variables that can potentially affect real gross domestic product， in this study only the six variables are included. This is done to conserve degrees of freedom.

RGDP=βo + β1 dRER + β2 dRTB + β3 dRHCONS + β4 dRGCONS +β5 dRKA + β6 dRGFCF + R2

Following the most general model （the model that includes all six explanatory variables above）， the result is presented in Table 4.4.

Table 4.4 Regression results of the general model

** = significant at the 5 % level

*** = significant at the 1% level

Using multiple regression analysis， it can be concluded that the short run variations in the real exchange rate are explained by the predictors included in the model. The R2 value shows that 59% of the variations in the real gross domestic product can be explained by the explanatory variables included in the model. However， the R2 exhibits a peculiar behavior in multiple regressions. As stated in Wooldridge （2002）， the R2 always increases with the number of explanatory variables. Therefore， the more relevant statistic is the adjusted R2. Using the adjusted R2， it can still be concluded that the model is an acceptable fit to the data， since 55% of the short-run variations in the real gross domestic product can still be explained by the variables in the model.

The significance of the model as a whole was determined using the F-test. Since the computed F-statistic is 14.87， the model as a whole can be deemed significant at the 1% level. The p-value of the F-statistic， which is 0.0000， confirms this conclusion.

The t-test is applied to each explanatory variable to determine whether the effect each one has on the real gross domestic product is statistically significant. The following conclusions were drawn：

The real bilateral exchange rate significantly affects real gross domestic product at the 10 and 5% level. And the real trade balance significantly affects real gross domestic product at the 10 and 5% level. The real household consumption significantly affects real gross domestic product at the 10， 5， and 1% level. Next， the real government consumption significantly affects real gross domestic product at the 10， 5， and 1% level and the real gross fixed capital formation significantly affects real gross domestic product at the 10， 5， and 1% level. The last one， capital account does not significantly affect real gross domestic product.

Therefore， dRER， dRTB， dRHCONS， dRGCONS， dRGFCF all significantly affect the short-run behavior of dRGDP of the United Kingdom. From these findings， the following five conclusions can be inferred：

First of all， there exists a negative relationship between the real bilateral exchange rate and the real gross domestic product. As the change in real bilateral exchange rate increases by one unit， the growth of the real gross domestic product decreases by 0.13.

Besides， there exists a positive relationship between the real trade balance and the real gross domestic product. As the change in real trade balance increases by one unit， the change in real gross domestic product increases by 0.40.

Thirdly， there exists a positive relationship between the real household consumption and the real gross domestic product. As the change in real trade surplus increases by one unit， the growth of real gross domestic product increases by 0.63.

In addition， there exists a positive relationship between the real government consumption and the real gross domestic product. As the change in real trade surplus increases by one unit， the growth of real gross domestic product increases by 0.56.

Finally， there exists a positive relationship between the real gross fixed capital formation and the real gross domestic product. As the change in real trade surplus increases by one unit， the growth of real gross domestic product increases by 0.73.

Intuitively， the signs can be explained by basic macroeconomics that can be learned in macroeconomics books like Dornbusch， et al. （2005） or Krugman and Obtsfeld （2003）. Findings of previous studies can also be related to the findings of this study.

The positive relationship between the real trade balance and real gross domestic product can be explained by aggregate demand. An increase in the trade surplus means that there is an increased demand for UK exports of goods and services. This spurs economic activity to meet that demand. The United Kingdom， being an exporter of oil and financial services， stands to grow significantly with an increase in the real trade balance.

The finding that the change in real exchange rate has a negative relationship with the change in the real gross domestic product seems to be counterintuitive given that higher exchange rates are commonly associated with an improved trade balance and consequently a higher gross domestic product. However， this sign can be explained within the context of the asset market approach. In an open economy like the United Kingdom， with its large financial sector， real exchange rate fluctuations raise the risk that investors face. This is because a possible depreciation may erode the value of their investment. This higher risk implies a negative impact on gross domestic product （Schnabl， 2007）.

4.5 Granger Causality Test：

The Granger Causality Test provides a way to test for this situation provided that the variables are already stationary. There are four possible outcomes as laid out in Gujarati ：

Firstly， unidirectional causality from dRER to dRGDP can be concluded if the estimated coefficients from the lagged dRER in the first regression are collectively statistically different from zero and the estimated coefficients from the lagged dRGDP in the second regression are collectively not statistically different from zero.

Secondly， unidirectional causality from dRGDP to dRER on the other hand can be concluded if the estimated coefficients from the lagged dRER in the first regression are collectively not statistically different from zero and the estimated coefficients from the lagged dRGDP in the second regression are collectively statistically different from zero.

Thirdly， feedback or Bilateral Causality can be concluded when the sets of coefficients of dRGDP and dRER are statistically different from zero in both the first and second regressions.

Last but not the least， independence is implied when the sets of coefficients of dRGDP and dRER are not statistically different from zero in both the first and second regressions.

The lag length of 2 time periods was chosen based on minimizing certain information criteria. The results of the two regressions are shown in Table 4.5.1 and Table 4.5.2. Based on the individual t-tests and the collective F-tests， it can be seen that there seems to be unidirectional Granger causality from dRER to dRGDP. The lagged coefficients of dRER in the first regression were shown to be significant with an F-stat of 2.58， while the lagged coefficients of the dRGDP in the second regression were shown to be statistically insignificant with an F-stat of 0.47. Therefore， the Granger causality test adds to the finding in the OLS regression that changes in the real bilateral exchange rate Granger cause changes in the real gross domestic product.

Table 4.5.1 Regression result of the first regression Dependent variable： dRGDP

Table 4.5.2 Regression result of the second regression Dependent variable： dRER

The finding that the change in real exchange rate has a negative relationship with the change in the real gross domestic product seems to be counterintuitive given that higher exchange rates are commonly associated with an improved trade balance and consequently a higher gross domestic product.

Chapter Five – Conclusion

5.1 Conclusions and Implications：

As real GDP and real exchange rate have a correlation coefficient of -0.45， it is shown that they are negatively correlated with each other. On the contract， real GDP is positively related to RGFCG and RHCONS with correlation coefficient of 0.55 and .053 respectively. But the rest of variables mostly has very small or none correlation with one another， that is， they are slightly found highly related to one another.

In the regression model， 55% of the fluctuations in real GDP in a short run can be explained by variables in the model. The real GDP is significantly affected by real bilateral exchange rate and trade balance at the significant level of 5%， and affected by household government consumption and gross fixed capital formation at a 1% significance level.

5.2 Recommendations for Future Work：

Because of the limited time， resources and length of paper， the impacts were narrowed so I would like to suggest investigating more specific industries or companies such as oil industry or financial services sector in future works in spite of the fact that this paper has covered major economic indicators related to exchange rate.

Finally， as mentioned in ‘Limitation’ sector， methods applied here is relatively simple which only investigate the relationship mainly between two variables， consequently I would propose that the results of this study should be tested by Ramsey’s RESET test for specification errors to ensure accuracy and at the meantime more complicated methods such as dynamic panel procedure should be explored.