Indian Rupee: Is it Really Unpredictable? FIN 3560: Financial Markets and Instruments Stephanie Boenawan Connor Boyen Aydarbek Kurbansho Mirela Tadic (Section 04) December 4, 2013

“I pledge my honor that I neither received nor provided any unauthorized assistance during the completion of this work.” “The authors of this paper hereby give permission to Professor Michael Goldstein to distribute this paper by hard copy, to put it on reserve at Horn Library at Babson College, or to post a PDF version of this paper on the Internet.”

Table of Contents

Executive Summary ……………………………………………………………………………. 2 Background ………………………………………………………………………………..…… 3 Regression Analysis ……………………………………………………………………………. 4 Variable Analysis ………………………………………………………………………………. 8 Exchange Rate Calculations …………………………………………………………………... 15 Conclusion ………………………………………………………………….…………………... 16 References ………………………………………………………………………………...……. 17 Exhibits ………………………………………………………………………...……………….. 19

1

Executive Summary

The paper opens with a history on the volatility of the Indian Rupee that was majorly driven by inflation in 1966. Since 1966, the Indian Rupee has gone through some major policy changes that affect the intrinsic value of the currency. The Indian Rupee was fixed to the US Dollar from 1966 until the crisis of 1991, and India was implementing strict protectionist policies with a huge current account deficit1. These factors causes the Indian Rupee to be highly overvalued as the spread between nominal exchange rate and real exchange rate is around 10.5 Rupee to a Dollar. The Indian Rupee devalued against the US Dollar from 4.79 Rupee to a Dollar in 1958 to 18.52 Rupee to a Dollar in 1991 and further depreciated to 62.39 Rupee to a Dollar today2. After 1991, India changed into a floating exchange rate regime. This change has caused a volatility problem to the exchange rate of the Indian Rupee. The source of the volatility problem is deeply analyzed through a regression model that encompasses these variables from 1992 to 2012; total Gross Domestic Product or GDP (in US$ bln.), public debt to GDP, inflation, nominal interest rate, current account, and terms of trade. Our regression model examined the correlation between the exchange rate of the Indian Rupee and each of these six independent variables. The purpose of the model is to see if each variable has a statistically significant effect towards the Indian Rupee. Also it eliminates such variables that might intuitively seem contradictive such as inflation rate, which were perceived as the main problem to the fluctuation of the Indian Rupee. In addition, the regression analysis provides a best subset of variables that gives the highest R2 or coefficient of determination value out of a 100% for predicting the changes in the Indian Rupee to the Dollar. The paper concludes with a result that shows a strong relationship exists between the exchange rate of the Indian Rupee and GDP, public debt to GDP, nominal interest rate, current account deficit, and terms of trade. Since changes in these five variables are statistically proven to be highly correlated to the changes in the exchange rate of the Indian Rupee, it is recommended if policy makers in India would carefully consider these five variables to maintain a more stable Rupee.

1

Jyoti P. "Indian Rupee, seen as Overvalued Against U.S. Dollar, Likely to Fall." Asian Wall Street Journal: 24. May 09 1997. ProQuest. Web. 4 Dec. 2013 . 2 Merchant, Minhaz. "Rupee Is Undervalued by 25%; Fair Value Would Be Rs 40/dollar." The Economic Times. N.p., n.d. Web. 28 Nov. 2013. .

2

Background History The Rupee is one of the oldest forms of currency still used in the modern day world. However, it was not until 1947 when India broke free of British control and began introducing their monetary policy. It took ten years after the induction of the Rupee for a decimalization system to be put into place. After 1957, the paisa was inducted into the system as one hundredth of a Rupee. Over the past 65 years the Rupee has significantly depreciated in value compared to the US dollar. Originally, the ratio was 1:1 for the Indian Rupee to the US dollar. After less than one year the ratio quickly changed to 1 US dollar to every 4.79 Rupees. There have been two major economic crises in India’s recent past that have helped create the downward spiral of the Rupee’s value. In 1966, India’s inflation caused its own goods to become more expensive than foreign goods and therefore the amount of imports increased while deports decreased. The depletion of India’s foreign currency reserves finally blocked foreign aid and subsequently the Rupee was devalued. The 1991 devaluation of the Rupee was primarily because of India’s economic reform. During the eighties, made worse by the Gulf War, India’s oil import bill grew, thus causing the country to have a balance of payments problem.

The

government’s deficit rose in 1981 from 9 percent of the country’s GDP to 12.7 percent in early 1991. At this point, the value of the Rupee declines drastically; the government began expanding the international reserves and by the end of 1991, the Indian government depleted its foreign reserves and had to allow the currency to sharply decrease in value.3 The 1991 devaluation was different than 1966 because the Indian government was trying to increase trade with foreign powers which effectively devalued the Rupee. Similar to 1966, high inflation lowered the amount of imports which caused trade deficits. India’s primary focus was to stabilize the value of the Rupee. Indian currency is susceptible to economic altercations in other areas such as Bangladesh, Pakistan and Nepal. One of the major reasons is because these countries have adopted the Rupee as currency for their nations. Currently, one American dollar is worth roughly 62 Rupees. 4 Gold India is known to be one of the biggest gold importers in the world. There are two major reasons for this huge amount of gold import. The first one is the jewelry industry that is extremely profitable in India, allowing for an existence of a $1.2 billion industry.5 The second cause is the unpredictable performance of

3

World Bank, 23 Aug. 1991. Web. "Money Studies in India." N.p., n.d. Web. 1 Dec. 2013.. 5 Shivom, Seth. "Gold Exports in June Slump 70% in India." Gold News. Mineweb, 19 July 2013. Web. 3 Dec. 2013. . 4

3

the Indian currency. Gold is known to be a safe haven for investors and is one of the best choices to look into during a currency unstable run. However, increased imports hurt India’s current account deficit, putting more pressure on the 6

Rupee . Even though it is great that the ability to possess gold will allow numerous investors to keep their money safe away from the corrupted government issues, sudden spikes in the currency movement and other “cataclysmic” factors, the high amount of gold imports keeps the Rupee weaker and the economic development of India is slowing down. Regression Analysis The type of data collected for our regression was time-series. Please see Exhibit 1 for the regression output from Minitab. We chose six factors which we believe affect the Rupee Exchange Rate and explain the recent currency troubles India has been facing. All observations are collected over a time period from 19922012. Time is an important dimension in a time series data, and our regressions on the six factors which affect the Rupee are dependent across time. Number of observations in our regression is 20, and the significance level is 5% (1.96 = critical value). The six factors chosen and the data indicators used are as follows: 1. Political Stability & Economic Performance : Gross Domestic Product (GDP) 2. Public Debt: Public Debt/GDP 3. Inflation: Consumer Price Index 4. Interest Rates: Nominal Interest Rates 5. Current Account Deficit: Trade of Imports and Exports 6. Terms of Trade: Exports Prices/Import Prices Our hypothesis for the regression is as follows: Ho: Political Stability & Economic Performance, Public Debt, Inflation, Interest Rates, Current Account Deficit, and Terms of Trade have no effect on the exchange rate of the Indian Rupee H1: Political Stability & Economic Performance, Public Debt, Inflation, Interest Rates, Current Account Deficit, and Terms of Trade do have an effect on the exchange rate of the Indian Rupee. In determining whether or not the six factors affect the Indian Rupee Exchange Rate value, we will analyze the critical value test and p-value test to test our assumptions and hypothesis. We will also analyze the collinearity test and best subset test.

6

Canavan, Greg. "Why India Is Buying Gold." The Daily Reckoning Australia. Port Phillip Publishing, 28 June 2012. Web. 03 Dec. 2013. .

4

R2 & Adjusted R2 The coefficient of determination (R2) presents how the data values of the model are organized. A high R2 value shows that all the variables presented in the regression model fit a single straight line at a higher succession rate7. Our model’s R2 is equal to 92.9%. The high percentage means that the majority of movements of a security are explained by movements in the index. The adjusted coefficient of determination accounts all the variables added to the model. Otherwise, it serves the same role as the regular coefficient of determination. Our model’s adjusted R2 is 89.6%, making our model fairly precise. We also looked at the number of variables that is appropriate to give us a regression model that has a justifiable high R2 value by looking at the difference between R2 and R2 adjusted. The difference between R2 and R2 adjusted is 3.3%, which is larger than the normal 2% difference according to goodness of the fit hypothesis8 The difference of 3.3% tells us that we should not add more variables to the model. T-Statistic The T- statistic is a reference to the relationship between a single variable and a single predictor. It is a statistical examination of two population means.9 The absolute value of the T-statistic is used to reject the null hypothesis for the regression model if the absolute value of the T- statistic is higher than the critical value. The absolute value is used due to the two-tailed nature of the statistical analysis. |t| > 1.96*

significant

*The T-Statistic will be analyzed for all six chosen factors following the same statistic rules as above

P-Value The P value serves as an estimated probability to reject the null-hypothesis for the regression model. The lower the P value, the lower the relevance of the null hypothesis of the variable in the regression model10. With a 5% significance level, we determined whether or not the p-value for the regression is significant by using the following statistics rule: P-value < Significance Level 0.000 < 0.05* significant

7

"R-Squared." Investopedia. Investopedia US, n.d. Web. 30 Nov. 2013. . 8 Berenson, Mark L., David M. Levine, and Timothy C. Khrebiel. Basic Business Statistics: Concepts and Applications. Upper Saddle River, NJ: Prentice-Hall, 1999. Print. 9 "T-Test." Investopedia. Investopedia US, n.d. Web. 30 Nov. 2013. . 10 "P Values." Statistical Help. Statsdirect.com, n.d. Web. 30 Nov. 2013. .

5

The regression’s p-value of 0.000 indicates that the overall regression we ran with the six factors is significant. The majority of our P-values for the chosen variables are close enough to zero to allow us to consider these variables to be more statistically relevant. The inflation variable, that doesn’t look statistically variable, might be affected by the nature of this particular variable. (This will be analyzed further) *The P-Value will be analyzed for all six chosen factors in the same format as above

Collinearity Test It is very important to test collinearity problem in our regression model, because an independent variable that is highly correlated to other independent variables in the model would cloud the result in exchange rate of Indian Rupee or our dependent variable. To test this problem, we looked at the VIF value of each independent variable in our model that is higher than the appropriate 5. A VIF higher than 5 means that there is a collinearity problem according to VIF test11.

The result above shows that GDP ($bln.) variable and Current Account Deficit variable have the highest VIF in the model. This means that there is a high correlation between GDP ($bln.) and Public Debt/GDP, Inflation, Nominal Interest Rate, and Current Account Deficit. Also it means that Current Account Deficit is highly correlated with GDP ($bln.), Public Debt/GDP, Inflation, and Nominal Interest Rate. Best Subset (Please see exhibit 1b) Best subset regression tells us the exact variables and number of variables that would give the best R2, R2 adjusted value, and standard error of the estimate (S) value. The best amount of variables that gives the highest R2 and R2 adjusted value is five variables, which are GDP ($bln.), Public Debt/GDP, Nominal Interest Rate, Current Account Deficit, and Terms of Trade. These variables also give the lowest S value of 2.2548, which means that the variability of the data around the regression line is 2.2548 points away. In addition, the rule for coefficient of variation says that the closer the formula

to 0 means the less volatility

or variation in the data to the regression line of the exchange rate of Indian Rupee12.

11

Berenson, Mark L., David M. Levine, and Timothy C. Khrebiel. Basic Business Statistics: Concepts and Applications. Upper Saddle River, NJ: Prentice-Hall, 1999. Print. 12 "Coefficient Of Variation." Investopedia. N.p., n.d. Web. 22 Nov. 2013. .

6

Changes to the Model After considering the collinearity problem in the regression model, we decided to run a regression without the GDP ($bln.) and current account deficit variable. The result shows that there is very little collinearity in the model, which means that each independent variable is not too correlated that it clouds the data. Predictor Constant Public Debt/GDP Inflation Nominal Interest Rate ToT S = 3.27203

Coef 104.82 -0.3127 -0.1511 -2.7897 -0.00000000

R-Sq = 83.9%

SE Coef 14.40 0.1657 0.2934 0.3528 0.00000000

T 7.28 -1.89 -0.52 -7.91 -1.90

P 0.000 0.079 0.614 0.000 0.076

VIF 1.227 1.428 1.331 1.597

R-Sq(adj) = 79.6%

However, the result shows a poorer R2 and R2 adjacent difference of 4.3% as well as a poorer R2 value of 83.9% compared to our original 92.9%. This led us to do another change to the model according to our t-test and p-value result by eliminating insignificant independent variables such as inflation from the model. The result shows an improvement in our R2 adjusted with a value of 90.3%, which makes the difference between R2 and R2 adjusted smaller with a value of 2.6%. This means that the model is a better predictor of changes in the exchange rate of Indian Rupee given the variables listed below. Predictor Constant GDP ($bln.) Public Debt/GDP Nominal Interest Rate Current Acc.Deficit ToT S = 2.25477

Coef 107.88 0.019522 -0.6309 -1.9537 0.31275 -0.00000000

R-Sq = 92.9%

SE Coef 11.35 0.004583 0.1451 0.3341 0.07745 0.00000000

T 9.51 4.26 -4.35 -5.85 4.04 -3.71

P 0.000 0.001 0.001 0.000 0.001 0.002

VIF 19.207 1.983 2.514 16.265 1.325

R-Sq(adj) = 90.3%

Residual Plots Assumptions The normal probability plot graph further shows a supporting point for the high R2 value that the data points of the exchange rate of Indian Rupee lie closely to the regression line assuming a straight line relationship as seen on exhibit 1c. Since, we ran a regression model instead of a time series, the Versus Fits graph show no pattern for the data points to show that the data is not cyclical. The Versus Fits graph indicates that the residuals in the models are independent of each other, which means that the assumption of independence of errors is valid13. In addition, the histogram on exhibit 1c shows a peak in the middle at 0 on 13

Berenson, Mark L., David M. Levine, and Timothy C. Khrebiel. Basic Business Statistics: Concepts and Applications. Upper Saddle River, NJ: Prentice-Hall, 1999. Print.

7

the x-axis and two tails at each end of the graph. The histogram does not show a perfectly shaped normal distribution pattern, but it does portray a closer picture to a bell-curved shape, which justify our reasoning to do a two-tailed test type for our t-test. Lastly, the Versus Order graph assumes an equal variance14 for the regression model by looking at the data points that lie inside the -2 and 2 bands on the y-axis. The graph shows that the equal variance assumption is valid due to all data points lie inside the -2 and 2 bands on the yaxis.

Variable Analysis Political and Economic Stability In the last 20 years, the Indian economy has been one of the most cherishing economies in the world. Thanks to their service industry that is mostly based off IT-support for a lot of western companies, the development rates of the Indian economy have been somewhat stable. A slow decline in the economic development as well as the Rupee exchange rate was witnessed in 2012 due to global economic issues, mostly related to the economical downfall in several countries of the European Union, as well as the slow decision making from the Indian parliament when it comes down to laws15. To determine the political and economic situation in India, we decided to use the main factor that reflects changes in both of the factors – Gross Domestic Product. The stability of GDP without any radical changes shows that the economic performance (which is based off the government actions). Our hypothesis for GDP is as follows: Ho: GDP has no effect on the exchange rate of the Indian Rupee H1: A higher GDP increases the value of the Indian Rupee.

Our regression indicated a T-statistic of 4.03 for the GDP variable |4.03| > 1.96 The absolute value of 4.03 is 4.03, which is larger than the critical value of 1.96. Therefore, we are rejecting the null hypothesis and determine that according to the t-statistic, a higher GDP value increases the value of the Indian Rupee.

14

Equal variance means that there is no major differences in the variability of the residuals for different X i values (Berenson 541). 15 Potia, Zeenat, and Tarun Khanna. "Behind India’s Economic and Political Woes." HBS Working Knowledge. Harvard Business School, n.d. Web. 22 Nov. 2013. .

8

Our regression indicated a P-value of 0.001 for the GDP variable 0.001 < 0.5 The p-value of the GDP variable is 0.001, which is smaller than the significance level of 0.5. Therefore, we are rejecting the null hypothesis and determine that according to the p-value test, a higher GDP increases the value of the Indian Rupee. After examining the GDP variable for India, we have determined that economic and political stability is significant, and does affect the Indian Rupee Exchange Rate. The higher the GDP in India, the higher the exchange rate value will be. However it is important to note an usual observation on year 2007 in GDP that caused the value of Indian Rupee to be 4.096 points away given its actual value of 41.350 Rupee per Dollar and its predicted value of 45.446 Rupee per Dollar according to our regression model 16. Since the GDP changes relate to the currency rate, we can make a connection with the politic and economic stability.

Public Debt Public or government debt is important factor, when it comes down to the foreign exchange rates fluctuation due to its ability to predict the stability of country’s economic performance for the foreign investors. Depending on how much the government borrowed and how capable it is to pay the debts back, the investors make the final decision to invest in the currency. Therefore, the strength of the currency is dependent on the demand from the investors. 1718 Our hypothesis for Public Debt is as follows: Ho: Public Debt has no effect on the exchange rate of the Indian Rupee H1: A higher Public Debt increases the value of the Indian Rupee.

Our regression indicated a T-statistic of -4.08 for the Public Debt variable |-4.08| > 1.96 The absolute value of -4.08 is 4.08, which is larger than the critical value of 1.96. Therefore, we are rejecting the null hypothesis and determine that according to the t-statistic, a higher Public Debt/GDP value decreases the value of the Indian Rupee.

16

Another important note is that the unusual observation for our regression model is only 4.76% of our total data points. Bergen, Jason Van. "6 Factors That Influence Exchange Rates." Investopedia. Investopedia US, n.d. Web. 22 Nov. 2013. . 18 "A Walk on the Wild Side." The Economist. The Economist Newspaper Ltd., 23 Feb. 2013. Web. 23 Nov. 2013. . 17

9

Our regression indicated a P-value of 0.001 for the Public Debt variable 0.001 < 0.5 The p-value of the Public Debt/GDP variable is 0.001, which is smaller than the significance level of 0.5. Therefore, we are rejecting the null hypothesis and determine that according to the p-value test, a higher Public Debt/GDP decreases the value of the Indian Rupee. After examining the Public Debt variable for India, we have determined that Public Debt is significant, and does affect the Indian Rupee Exchange Rate. The higher the Public Debt in India, the lower the exchange rate value will be. Inflation WPI (Wholesale Price Index) is the most common inflationary measured used by policy makers in India. WPI ‘represents the price of goods at a wholesale stage’19. On the other hand, CPI (Consumer Price Index) measures ‘the weighted average of prices of a basket of consumer goods and services’ 20. However, India’s economy is moving in reaction towards changes in consumer-price inflation. This is because more than 800 million people in India are living on less than $2 per day21. This is a major reason that caused us to use CPI as inflation measure rather than WPI, because CPI will provide a better grasp of the volatility in price change and its impact towards the Indian Rupee. Inflation rate or changes in price of goods and services will impact the exchange rate of the Indian Rupee. An increase in inflation rate means price of goods and services have become higher or more expensive due to lower purchasing power of Indian Rupee. This means that consumers in India will be less willing and able to purchase goods and services. On the other hand, domestic goods and services or India’s export will be demanded less as price becomes more expensive. Our hypothesis for Inflation (CPI) is as follows: Ho: Inflation rate has no effect on the exchange rate of the Indian Rupee H1: An increase in the inflation rate will cause a decrease in the value of the Rupee

19

"Wholesale Price Index." The Economic Times. N.p., n.d. Web. 28 Nov. 2013. . 20 "Consumer Price Index CPI." Investopedia. N.p., n.d. Web. 22 Nov. 2013. . 21 Goyal, Kartik. "Rajan Spurs Surge in India's Reserves to Support Rupee: Economy." Bloomberg.com. Bloomberg, 12 Nov. 2013. Web. 22 Nov. 2013. .

10

Our regression indicated a T-statistic of -0.34 for the Inflation variable |-0.34| > 1.96 The absolute value of -0.34 is 0.34, which is smaller than the critical value of 1.96. Therefore, we accept the null hypothesis and determined that inflation rate has no effect on the exchange rate of the Indian Rupee. Our regression indicated a P-value of 0.740 for the Inflation variable 0.740 < 0.05 The p-value of the inflation variable is 0.740, which is bigger than the significance level of 0.05. Therefore, we accept the null hypothesis and determine that inflation rate has no effect on the exchange rate of the Indian Rupee. Inflation rate in India is not a very significant driver in the exchange rate of Indian Rupee according to the t-test performed on the regression model. This is largely due to the volatility of the inflation variable that causes too much movement on data points for the model that it becomes statistically insignificant. However, this does not mean that inflation rate is not a key indicator for policy makers in India that would inevitably affect the Indian Rupee indirectly. Interest Rate Nominal interest rate that is used in the regression model is the lending interest rate for ‘short and medium-term financing needs of the private sector’22. Nominal Interest rates signals borrowers and lenders on the rate of borrowing and lending money, which will affect spending and investment by firms and the public. As interest rates rises for lenders, firms and public will be less willing to borrow money as borrowing cost becomes more expensive. This will slow down India’s growth, but it will also mean that foreign investors will be more willing to invest in India as rate rises. As demand for Rupee rises, so will its value compared to other currencies. Our hypothesis for Interest Rates is as follows: Ho: Nominal interest rates have no effect on the exchange rate of the Indian Rupee H1: An increase in the nominal interest rate will cause an increase in the value of the Indian Rupee Our regression indicated a T-statistic of -4.82 for the Interest Rate variable |-4.82| > 1.96

22

"Deposit Interest Rate (%)." Data. .

11

N.p.,

n.d.

Web.

22

Nov.

2013.

The absolute value of -4.82 is 4.82, which is greater than the critical value of 1.96. Therefore, we reject the null hypothesis and determine that an increase in nominal interest rate will cause an increase in the value of Indian Rupee. Our regression indicated a P-value of 0.000 for the Interest Rate variable 0.000 < 0.05 The p-value for nominal interest rate variable is smaller than the significance level of 0.05. Therefore, we reject the null hypothesis and determined that an increase in nominal interest rate will cause an increase in the value of Indian Rupee. According to our t-test, nominal interest rate is significant as a key driver to the exchange rate of Indian Rupee. Since the relationship between nominal interest rate and exchange rate of Indian Rupee is negatively correlated, an increase in nominal interest rate will caused an increase in the value of Indian Rupee. Therefore when making a policy that will affect the Rupee, policy makers should be aware of the change in nominal interest rate. Current Account Deficit The current account deals with the trade of goods and services between two countries. The monetary value of exports from a country and imports into a country are measured in the current account. If the value of a country’s exports exceeds the values of the goods and services it imports, then that country has a trade surplus.23 Our hypothesis for the Current Account Deficit is as follows: Ho: The current account deficit has no effect on the exchange rate of the Indian Rupee H1: A higher current account deficit will cause a decrease in the value of the Indian Rupee Our regression indicated a T-statistic of 3.56 for the Current Account Deficit variable |3.56| > 1.96 The absolute value of 3.56 is 3.56, which is larger than the critical value of 1.96. Therefore, we are rejecting the null hypothesis and determine that according to the t-statistic, a higher current account deficit value decreases the value of the Indian Rupee.

23

"Current Account Deficit." Investopedia. N.p., n.d. Web. 04 Dec. 2013.

12

Our regression indicated a P-value of 0.004 for the Current Account Deficit variable 0.004 < 0.05 The p-value of the Terms of Trade variable is 0.004, which is smaller than the significance level of 0.5. Therefore, we are rejecting the null hypothesis and determine that according to the p-value test, a higher current account deficit decreases the value of the Indian Rupee. After examining the Current Account Deficit Variable for India, we have determined that the Current Account is significant, and does affect the Indian Rupee Exchange Rate. The higher the deficit in India, the lower the exchange rate value will be. In recent months, India’s imports have fallen, while its exports climbed 13%, causing the trade deficit no fall from 20.1 billion in May, to $10.9 billion in August.24 The Reserve Bank of India recently came out with a statement saying that “the current account deficit in 2013-14 will be USD 56 billion” and that due to this number being lower than projected earlier; there is no reason for the Rupee, India’s currency, to depreciate. This figure is less than three percent of India’s GDP and 32 Billion dollars less than last year’s figure, which is a positive movement. Governor Raghuram Rajan said that although this is a good movement some of it can be explained by “our strong measures to curb gold import”. It is fortunate that this figure fell because the CAD reached an all-time high in India from 2012-2013 at 88.2 Billion dollars and 4.8 percent of GDP. Since 2008, India’s CAD was steady between 2008 and 2010, but then grew significantly and has been fluctuating ever since. 25 Terms of Trade In India, the terms of trade effect corresponds to the ratio of price of exportable goods to the price of importable goods. 26 Our hypothesis for Terms of Trade is as follows: Ho: Terms of Trade has no effect on the exchange rate of the Indian Rupee H1: A higher terms of trade will cause an increase in the value of the Indian Rupee Our regression indicated a T-statistic of -2.95 for the Terms of Trade variable |-2.95| > 1.96

24

Raza, Syed. "Effects of Terms and Trade." N.p., 3 Apr. 2012. Web. 3 Dec. 2013. . 25 "India Terms of Trade." TRADING ECONOMICS. N.p., n.d. Web. 04 Dec. 2013. . 26 Raza, Syed. "Effects of Terms and Trade." N.p., 3 Apr. 2012. Web. 3 Dec. 2013. .

13

The absolute value of -2.95 is 2.95, which is larger than the critical value of 1.96. Therefore, we are rejecting the null hypothesis and determine that according to the t-statistic, a higher Terms of Trade value increases the value of the Indian Rupee. Our regression indicated a P-value of 0.011 for the Terms of Trade variable 0.011 < 0.05 The p-value of the Terms of Trade variable is 0.011, which is smaller than the significance level of 0.5. Therefore, we are rejecting the null hypothesis and determine that according to the p-value test, a higher Terms of Trade increases the value of the Indian Rupee. After examining the Terms of Trade Variable for India, we have determined that Terms of Trade is significant, and does affect the Indian Rupee Exchange Rate. The higher the Terms of Trade in India, the higher the exchange rate value will be. When a nation’s Terms of Trade improves, thus making the Rupee exchange rate higher, the country can buy more imports for any given level of exports. A higher value in the currency lowers the prices of its imports. The terms of trade in India are reported by the Reserve Bank of India. In the last few decades the Terms of Trade situation in India has been improving. “In the 1980’s the average terms of trade was 84, in 1990’s it increased to 105 and in the decade of 2000 the average terms of trade marginally improved and became 107.” Similarly, during the period of time, as the Terms of Trade was improving India’s GDP was also improving, and it has been shown that there is a connection between the two factors. Between the years 2000 and 2011, India hit its lowest point in term of trade of 77 Index Points in the year 2007. It is forecasted to continue improving just slightly in the coming years and there has been an upward trend since its lowest point almost seven years ago. This means that India is continuously exporting more that it is importing at an increasing rate, which causes capital to flow into the country, which is a very positive indicator for India’s continued growth. 27

27

"Ideas for India." Exchange Rate Movements and Indian Firms' Exports. N.p., n.d. Web. 04 Dec. 2013. .

14

Exchange Rate Calculations28 Purchasing Power Parity A difference in inflation rates between countries such as India and the US can affect the exchange rate. The purchasing power parity takes into consideration the differences in inflation rates between the US and India. It is important to note the inaccuracy of the purchasing power parity in this exchange. The inaccuracy comes from the assumption that there are other factors affecting the exchange rate such as interest rates and GDP. The original equation is derived from the assumption that each country’s real interest rates are the same and therefore can be set equal to one. Eliminating this variable from the equation allows for the inflation rate of the US minus the inflation rate of India to equal the change in the spot price divided by the current spot price. In theory, a higher inflation rate will create a lower value for the currency.

US Inflation Rate - Indian Inflation Rate = Change in the spot exchange rate/ Current Spot Exchange Rate IPus-IPI= (ΔSus/I)/(Sus/I) The purchasing power theory describes that in the long run exchange rates will theoretically move towards rates that would equalize the price of an identical basket of goods in two different countries. Essentially, the purchasing power theory states that a good such as a cheeseburger should cost the same in two separate countries once the currency is converted using the exchange rate. In Exhibit 2c, the US inflation rate is compared to the Indian inflation rate on a monthly basis since the middle of 1994. Finding the difference in the inflation rates (per month) allows for the change in the spot price to be calculated. It is important to note that the inflation rates pertain to the entire month while the spot rates are from the beginning of the month. Using these variables allows for the calculation of the next months predicted spot rate. Comparing the predicted spot rate to the actual spot rate of the month allows for the percentage difference to be calculated. Exhibit 2a shows the graph of the difference between the predicted and the actual. There are several time periods where sharp spikes occur that prove the purchasing power parity inaccuracy. Exhibit 2b is a graph of the 10 year time period in the middle where the difference in predicted and actual is relatively evenly balanced. The spike to negative .07 at the 18 th observation in Exhibit 2b happens to fall on September 2001. Clearly, the September 11th attack on the United States had a quick and dramatic effect on the spot rate. This proves the assumption that there are more factors than just inflation that 28

24 US Department of the Treasury, n.d. Web. . 25 "India Interest Rate." TRADING ECONOMICS. N.p., n.d. Web. 04 Dec. 2013. . 26 "United States | Economic Indicators." United States | Economic Indicators. N.p., n.d. Web. 04 Dec. 2013. . 27 Reserve Bank of India, n.d. Web. . 28 "TRADING ECONOMICS | 300.00 INDICATORS | 196 COUNTRIES." TRADING ECONOMICS | 300.00 INDICATORS | 196 COUNTRIES. N.p., n.d. Web. 04 Dec. 2013. .

15

affect the exchange rate. If the theory were exactly correct then the difference should be zero and therefore the graph would not fluctuate as it does. Exhibit 2c shows the regression analysis of the difference between the actual and predicted spot rates. The sample size is 234 observations which include each month over the past 19 years. The t stat of .91701433974977 is lower than the significance level which leads to the conclusion that inflation is not directly correlated to the change in spot price. There are other factors that can affect the rate. Exhibit 2b proves that there are other factors The value of the rupee has dropped lower than expectations based on the higher rate of inflation in India. Uncovered Interest Rate Parity The expected spot price, E(S), for the Indian Rupee is the rate at which a bank believes the value of the foreign exchange will be. The price of the expected spot rate is based off of the current spot rate. The price is also adjusted with the cost of carrying the currency. The spot exchange rate and the difference in the two countries interest rates are the key variable to using the interest rate parity. Under the uncovered interest rate parity, the expected spot rate is going to equal the US interest rate divided by the current spot rate multiplied by the interest rate of India. The interest rates used to calculate the E(S) are 3 month Treasury bill yields. Using 3 month T-bills will theoretically predict the spot rate for 3 months in advance using the current month’s interest rates as shown in Exhibit 3b. The interest rates for India are substantially higher than that of the US and therefore the value of the rupee will be decrease. Similar to the purchasing power parity, the uncovered interest parity proves that interest rates are not the only determinant of the exchange rate.

Conclusion The purpose of this paper is to understand the underlying variables behind the volatility of the exchange rate of the Indian Rupee through a multivariate regression analysis. The analysis concludes with a result that says the GDP, public debt to GDP, nominal interest rate, current account deficit, and terms of trade have a statistically significant effect towards the Indian Rupee. This means that as each of these variables changes, so does the Indian Rupee according to each positive or negative relationship between the Rupee and each variable. The most statistically significant variable in the model after removing the inflation rate variable is nominal interest rate. Nominal interest rate of the United States and India are also a factor in determining the forward exchange rate of the Indian Rupee. From our analysis based on the interest rate parity, we concluded that the difference between the 3-month Treasury bills in the US and the 3-month Treasury bills in India is statistically significant with the difference between the forward exchange rate of the Indian Rupee and the spot exchange rate of the Indian Rupee. This result allows us to predict for the forward exchange rate of the Indian Rupee using the interest rate parity theory more accurately on a monthly basis for the next two years. 16

References "Money Studies in India." N.p., n.d. Web. 1 Dec. 2013. . World Bank, 23 Aug. 1991. Web. http://www-wds.worldbank.org/external/default/WDSContentServer/ WDSP/IB/1991/08/23/000009265_3960930195417/Rendered/PDF/multi0page.pdf Shivom, Seth. "Gold Exports in June Slump 70% in India." Gold News. Mineweb, 19 July 2013. Web. 3 Dec. 2013. . Canavan, Greg. "Why India Is Buying Gold." The Daily Reckoning Australia. Port Phillip Publishing, 28 June 2012. Web. 03 Dec. 2013. . "R-Squared." Investopedia. Investopedia US, n.d. Web. 30 Nov. 2013. . Berenson, Mark L., David M. Levine, and Timothy C. Khrebiel. Basic Business Statistics: Concepts and Applications. Upper Saddle River, NJ: Prentice-Hall, 1999. Print. "T-Test." Investopedia. Investopedia US, n.d. Web. 30 Nov. 2013. . "P Values." Statistical Help. Statsdirect.com, n.d. Web. 30 Nov. 2013. . Berenson, Mark L., David M. Levine, and Timothy C. Khrebiel. Basic Business Statistics: Concepts and Applications. Upper Saddle River, NJ: Prentice-Hall, 1999. Print. "Coefficient Of Variation." Investopedia. N.p., n.d. Web. 22 Nov. 2013. . Potia, Zeenat, and Tarun Khanna. "Behind India’s Economic and Political Woes." HBS Working Knowledge. Harvard Business School, n.d. Web. 22 Nov. 2013. . Bergen, Jason Van. "6 Factors That Influence Exchange Rates." Investopedia. Investopedia US, n.d. Web. 22 Nov. 2013. . "A Walk on the Wild Side." The Economist. The Economist Newspaper Ltd., 23 Feb. 2013. Web. 23 Nov. 2013. . "Wholesale Price Index." The Economic Times. N.p., n.d. Web. 28 Nov. 2013. . 17

"Consumer Price Index - CPI." Investopedia. N.p., n.d. Web. 22 Nov. 2013. . Goyal, Kartik. "Rajan Spurs Surge in India's Reserves to Support Rupee: Economy." Bloomberg.com. Bloomberg, 12 Nov. 2013. Web. 22 Nov. 2013. . "Deposit Interest Rate (%)." Data. N.p., n.d. Web. 22 Nov. 2013. . "Current Account Deficit." Investopedia. N.p., n.d. Web. 04 Dec. 2013. 1 Raza, Syed. "Effects of Terms and Trade." N.p., 3 Apr. 2012. Web. 3 Dec. 2013. . "India Terms of Trade." TRADING ECONOMICS. N.p., n.d. Web. 04 Dec. 2013. "Ideas for India." Exchange Rate Movements and Indian Firms' Exports. N.p., n.d. Web. 04 Dec. 2013. . US Department of the Treasury, n.d. Web. . "India Interest Rate." TRADING ECONOMICS. N.p., n.d. Web. 04 Dec. 2013. .1 Reserve Bank of India, n.d. Web. . "TRADING ECONOMICS | 300.00 INDICATORS | 196 COUNTRIES." TRADING ECONOMICS | 300.00 INDICATORS | 196 COUNTRIES. N.p., n.d. Web. 04 Dec. 2013. . "India Inflation Rate." TRADING ECONOMICS. N.p., n.d. Web. 04 Dec. 2013. . http://www.federalreserve.gov/releases/h15/data.htm "India Treasury Bill Yield." TRADING ECONOMICS. N.p., n.d. Web. 04 Dec. 2013. ."Selected Interest Rates (Daily) H.15." FRB: H.15 Release--Selected Interest Rates--Historical Data. Federal Reserve Bank, n.d. Web. 02 Dec. 2013. .

18

Exhibits

Exhibit 1: Regression Output from Minitab Regression Analysis: Exch.rate versus GDP ($bln.), GDP/Capita PPP, ... The regression equation is Exch.rate = 118 + 0.0318 GDP ($bln.) - 0.0078 GDP/Capita PPP - 0.638 Public Debt/GDP - 0.069 Inflation - 2.12 Nominal Interest Rate + 0.366 Current Acc.Deficit - 0.000000 ToT 20 cases used, 1 cases contain missing values Predictor Constant GDP ($bln.) GDP/Capita PPP Public Debt/GDP Inflation Nominal Interest Rate Current Acc.Deficit ToT S = 2.39903

Coef 118.35 0.03183 -0.00778 -0.6379 -0.0687 -2.1211 0.3660 -0.00000000

R-Sq = 93.1%

SE Coef 25.70 0.02515 0.01530 0.1600 0.2711 0.6068 0.1519 0.00000000

T 4.60 1.27 -0.51 -3.99 -0.25 -3.50 2.41 -2.82

P 0.001 0.230 0.620 0.002 0.804 0.004 0.033 0.015

R-Sq(adj) = 89.0%

Analysis of Variance Source Regression Residual Error Total

DF 7 12 19

SS 926.66 69.06 995.73

Source GDP ($bln.) GDP/Capita PPP Public Debt/GDP Inflation Nominal Interest Rate Current Acc.Deficit ToT

DF 1 1 1 1 1 1 1

MS 132.38 5.76

F 23.00

P 0.000

Seq SS 428.01 309.92 73.70 36.08 23.86 9.31 45.78

Unusual Observations Obs 16 18

GDP ($bln.) 949 1224

Exch.rate 41.350 48.410

Fit 44.860 43.931

SE Fit 1.644 1.453

Residual -3.510 4.479

St Resid -2.01R 2.35R

R denotes an observation with a large standardized residual.

19

Exhibit 1a continued: Regression Output from Minitab Regression Analysis: Exch.rate versus GDP ($bln.), Public Debt/GDP, ... The regression equation is Exch.rate = 107 + 0.0193 GDP ($bln.) - 0.622 Public Debt/GDP - 1.89 Nominal Interest Rate + 0.303 Current Acc.Deficit - 0.000000 ToT - 0.088 Inflation 20 cases used, 1 cases contain missing values Predictor Constant GDP ($bln.) Public Debt/GDP Nominal Interest Rate Current Acc.Deficit ToT Inflation S = 2.32961

Coef 106.91 0.019293 -0.6219 -1.8907 0.30291 -0.00000000 -0.0883

R-Sq = 92.9%

SE Coef 12.07 0.004783 0.1523 0.3920 0.08513 0.00000000 0.2606

T 8.86 4.03 -4.08 -4.82 3.56 -2.95 -0.34

P 0.000 0.001 0.001 0.000 0.004 0.011 0.740

R-Sq(adj) = 89.6%

Analysis of Variance Source Regression Residual Error Total

DF 6 13 19

SS 925.18 70.55 995.73

MS 154.20 5.43

F 28.41

P 0.000

There are no replicates. Minitab cannot do the lack of fit test based on pure error. Source GDP ($bln.) Public Debt/GDP Nominal Interest Rate Current Acc.Deficit ToT Inflation

DF 1 1 1 1 1 1

Seq SS 428.01 0.85 352.51 73.38 69.81 0.62

Unusual Observations Obs 16

GDP ($bln.) 949

Exch.rate 41.350

Fit 45.446

SE Fit 1.139

Residual -4.096

St Resid -2.02R

R denotes an observation with a large standardized residual.

20

VIF 19.599 2.045 3.242 18.409 1.755 2.223

Exhibit 1b: Best Subset Best Subsets Regression: Exch.rate versus GDP ($bln.), Public Debt/, ... Response is Exch.rate 20 cases used, 1 cases contain missing values

Vars 1 1 2 2 3 3 4 4 5 5 6

R-Sq 76.7 43.0 80.0 78.0 83.6 81.7 85.8 84.5 92.9 88.2 92.9

R-Sq(adj) 75.4 39.8 77.6 75.4 80.5 78.2 82.1 80.4 90.3 84.0 89.6

Mallows Cp 26.7 88.6 22.7 26.4 18.1 21.6 16.0 18.4 5.1 13.7 7.0

21

S 3.5880 5.6160 3.4250 3.5922 3.1960 3.3777 3.0657 3.2049 2.2548 2.8999 2.3296

P u b l G i D c P D ( e $ b b t l / n G . D ) P

N o m i n a l

I n f l a t i o n

C u r r e n I t n t A e c r c e . s D t e f R i a c T t i o e t T X

X X X X X X X X

X X X X X

X X X X X X X X X X X X X

X X X X X X X X X

Exhibit 1c: Residual Plots Graph from Minitab

Residual Plots for Exch.rate Normal Probability Plot

Versus Fits Standardized Residual

99

Percent

90 50 10 1

-2

-1 0 1 Standardized Residual

2 1 0 -1 -2

2

30

36 42 Fitted Value

Histogram Standardized Residual

Frequency

6 4 2 -2

-1 0 1 Standardized Residual

54

Versus Order

8

0

48

2

2 1 0 -1 -2 2

22

4

6

8 10 12 14 16 Observation Order

18

20

Exhibit 2: Purchasing Power Parity29 US Inflation Rate – Indian Inflation Rate = Change in the Spot Exchange Rate/Current Spot Exchange Rate IPus-IPI= (ΔSus/I)/(Sus/I)

Exhibit 2a 0.15

0.1

0.05

Series1 1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226

0

-0.05

-0.1

-0.15

Exhibit 2b 0.06 0.04 0.02 0 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 -0.02 -0.04 -0.06 -0.08 Series1

29

"India Inflation Rate." TRADING ECONOMICS. N.p., n.d. Web. 04 Dec. 2013. .

23

Exhibit 2c

Regression Analysis T-Stat Average St. Deviation Standard Error Square Root (234)

0.91701433974977 0.00160559483121 0.02678352677939 0.00175089391911 15.29705854077840

Exhibit 2d New Spot Rate

Actual

Old Spot Rate - New Spot Rate

Jun-94 Jul-94 Aug-94 Sep-94 Oct-94 Nov-94

31.77367739 31.62930831 31.7585756 31.41690925 31.50629434

31.3675 31.39 31.375 31.36 31.405

0.012948988 0.007623712 0.012225517 0.001814708 0.003225421

May-13 Jun-13 Jul-13 Aug-13 Sep-13 Oct-13

54.72578356 54.19943143 57.21724689 60.46102891 61.03792062 66.82253294

53.806 56.581 59.533 60.635 66.595 62.585

0.017094442 -0.042091313 -0.038898646 -0.002869153 -0.083445895 0.067708444

24

Exhibit 3: Interest Rate Parity3031 The India interest rates are the 3 month treasury bill yield The US interest rates are 3 month treasury bills on a monthly frequency Exhibit 3a

Uncovered Interest Rate Parity (1+ius)= S rupees/$ *(1+iI)*(1$/E(S) rupees) E(S)= S*((1+id)/(1+if)) F= forward exchange rate S= current spot price id= interest rate of domestic (US) if= interest rate foreign (India) Exhibit 3b E(Spot Price)

(3 month prediction)

(Predicted SP-Actual SP)/ Actual SP

46.2000237 45.93405258 45.70263088 45.69640445

Spot Rate 45.82 45.55 45.32 45.32 43.4 44.53 45.42

Dec-03 Jan-04 Feb-04 Mar-04 Apr-04 May-04 Jun-04

0.021645011 0.021770341 0.021880578 0.02188356

Mar-13

0.018056173

55.38272202

54.145

0.022859397

Apr-13

0.017906528

55.84555453

54.285

0.028747435

May-13

0.018405297

54.33218375

53.806

0.009779277

Jun-13

0.018104883

55.23371928

56.581

-0.02381154

Jul-13

0.018061744

55.36563868

59.533

-0.070000862

Aug-13

0.018242178

54.81801656

60.635

-0.095934418

Sep-13

0.017358679

57.60807138

66.595

-0.134948999

Oct-13

0.016490321

60.64163173

62.585

-0.031051662

30

0.019417999 0.058388308 0.026333503 0.006085523

"Selected Interest Rates (Daily) - H.15." FRB: H.15 Release--Selected Interest Rates--Historical Data. Federal Reserve Bank, n.d. Web. 02 Dec. 2013. . 31 "India Treasury Bill Yield." TRADING ECONOMICS. N.p., n.d. Web. 04 Dec. 2013. .

25

Exhibit 4: Data used in Regression

Exchange Rate: Rupee/US Dollar

Inflation (CPI)

Nominal Interest Rate

Current Account Deficit (USD in Bln)

Public Debt (Public Debt/GDP)

Terms of Trade (in constant Rupee)

1992

25.92

11.8

18.9

-3.526

76.351

125,178,321,800.22

1205.28

1993

30.49

6.4

16.3

-1.158

76.787

345,418,259,528.26

1246.87

1994

31.37

10.2

14.8

-3.369

76.939

485,764,733,727.22

1281.5

1995

32.43

10.2

15.5

-5.911

74.109

394,725,008,164.72

1341.57

1996

35.43

9

16

-4.619

70.365

207,998,107,331.00

1416.99

1997

36.31

7.2

13.8

-5.499

68.711

549,450,552,310.18

1496.8

1998

41.26

13.2

13.5

-4.038

67.623

710,462,678,301.72

1530.2

1999

43.06

4.7

12.5

-4.698

67.818

460,404,304,403.43

1596.96

2000

44.94

4

12.3

-2.666

70.122

414,090,285,569.64

1702.93

2001

47.19

3.7

12.1

3.4

72.731

381,962,182,034.26

1741.32

2002

48.61

4.4

11.9

6.345

77.849

125,634,787,737.82

1797.68

2003

46.58

3.8

11.5

14.083

82.199

419,907,033,367.77

1838.08

2004

45.32

3.8

10.9

-2.47

84.3

-

1953.11

2005

44.10

4.2

10.8

-9.902

84.063

90,716,563,175.35

2074.47

2006

45.31

6.1

11.2

-9.565

81.764

132,920,373,567.44

2233.86

2007

41.35

6.4

13

-15.736

78.49

138,927,975,725.48

2406.34

2008

43.51

8.4

13.3

-27.913

75.44

742,039,208,631.76

2606.16

2009

48.41

10.9

12.2

-38.182

74.724

529,797,577,678.63

2671.68

2010

45.73

12

8.3

-45.946

74.973

931,591,036,361.65

2860.55

2011

46.67

8.9

10.2

-78.154

69.427

988,707,611,989.04

3121.62

2012

53.44

9.3

10.6

-88.163

68.053

41,359,071,322.05

3277.01

Year

26

Political Stability & Economic Performance (GDP)