Excerpt

## Table of Contents

**1. INTRODUCTION**

**2. METHODOLOGY**

**2.1. Identifying Variables:**

*2.1.1. Dependent Variable:*

*2.1.2. Independent Variables:*

**2.2. Model:**

**2.3. Data and Sample:**

**2.4. Scope of the Study:**

**3. FINDINGS AND ANALYSIS**

**3.1. Regression Analysis:**

**4. FORECASTING EXCHANGE RATE**

**4.1. Regression equation:**

**4.2. Forecasting exchange rate for August 2015:**

**4.3. Comparison between Forecasted exchange rate and Actual exchange rate:**

**5. CONCLUSION**

**REFREENCES**

**Forecasting Exchange Rate: A Panel Data Approach**

**Sajjad Hossine Sharif**

**Independent University, Bangladesh**

** Abstract: ** Exchange rate is a daily basis indispensable factor in the foreign exchange market as well as in the international trade. Many traders make profit based on the pip in the foreign exchange market. Moreover, inflation and deflation of a currency against another currency is the root of making profit in the foreign exchange market. Even in the international trade many individual traders and Multi-national Corporation always carefully observes the fluctuation of the exchange rate in order to determine the exchange rate efficiently and accurately. Because the more accurate the forecasted exchange rate is, the higher the chance to make profit only by investing a little amount of money in the foreign exchange market. Exchange rate has also significant impact over the export, import, foreign direct investment etc. However, this paper is established with an aim to explain how an individual or an organization can formulate future exchange rate of any currency in an efficient and time effective way. To meet this demand this paper is established with the help of panel data and regression model. As a sample, this paper considers USD/BDT to forecast. But it must be noted that with different panel data of different currencies, the method will remain same if anyone wants to forecast exchange rate of different currencies.

**Keyword:** Forecasting exchange rate, Panel data, Regression model.

## 1. INTRODUCTION

Exchange rate is one of the most vital issues in the international trade. When a transaction occurs we need to translate the amount of currency of that transaction into the local currency based the foreign currency. After that in making the payment the local currency needs to be converted into the equivalent amount of the foreign currency so that the payment will be successfully made. Therefore exchange rate involves both the translation and conversion of the currency sometimes by using direct quote or indirect quote or cross exchange rate. In most of the exchange rate the currencies are expressed by 3 letters where first 2 letters is designated for the country and the last letter is designated for the currency. However, in direct quotation of exchange rate the domestic currency fluctuates but per foreign currency remains constant while the local currency is the numerator and the foreign currency is the denominator. But in the indirect quotation the domestic currency becomes denominator and remains constant while the foreign currency will be the numerator and fluctuates against per domestic currency. When the exchange rate between two currencies is unknown we use the cross exchange rate. It is an exchange rate involves two currency derive from their exchange rate against another currency. However, this paper is an attempt to forecast USD/BDT by forming regression model based on the panel data of well chosen independent and dependent variables.

## 2. METHODOLOGY

### 2.1. Identifying Variables:

#### 2.1.1. Dependent Variable:

**USD/BDT:** It is the exchange rate which shows that for 1 BDT or Bangladeshi taka how much money a person can get in the United Sates currency (USD). Based on Bangladesh, it is an indirect quote but based on the United States of America it is a direct quote.

#### 2.1.2. Independent Variables:

**Inflation rate:** Inflation rate is a percentage which indicates the increase in the general level of prices for goods and services. In a very simple word, it means how fast a currency loses its value.

**Bank Interest rate:** A bank interest rate is the rate at which the central bank of a country approves the short term loan towards commercial banks.

**Per Person Income:** Per person income is also called per capita income. Per person income indicates the amount of money earned be per person in a certain or given area.

**GDP growth rate:** GDP is the abbreviation form of gross domestic product. GDP growth rate is a measurement of how fast an economy is growing. Or in another way, it is the percentage changes of gross domestic product from one period to another.

**Table 1: Variables with respective symbols**

Abbildung in dieser Leseprobe nicht enthalten

### 2.2. Model:

Regression analysis is a statistical tool which is used to investigate the relationship. More precisely, regression model analyzes the causal effect and statistical significance of the relationship between two or more variables by assembling the previous valid and reliable data of those variables. As an effort to forecast the exchange rate between USD and BDT, the established regression model consists of 5 variables. Change in USD/BDT (CUSBT) is the dependent variable as well as change in inflation rate (CINF) , change in Bank interest rate (CBIR), change in Per person income (CPPI) and change in GDP growth rate (CGGR) are the independent variables. Based on these variables the OLS regression model is illustrated below:

CUSBTt = b0 + b1(CINFt-1) + b2(CBIR t-1) + b3(CPPI t-1) + b4(CGGR t-1)

### 2.3. Data and Sample:

To conduct this study, considered sample data is the previous 10 years monthly data for each of our selected variables. As a result, the number of sample data is 120 periods for the each of the variables. The last 10 years monthly data with a time series of 2005-2015 for each of the selected variables are collected with convenience sampling method. Microsoft excel 2007 software is implemented for the regression analysis with an assumption that all the variables are linearly related.

### 2.4. Scope of the Study:

1. This paper does not consider any sample data that may exists before June, 2005 to maintain equal number of observations for each of the variable as well as to ensure the reliability of the data for each of the variable.

2. This study considers only a few selected independent variables which is a limitation in term of the scope of the study.

## 3. FINDINGS AND ANALYSIS

### 3.1. Regression Analysis:

Based on the regression model dependent variable change in USD/BDT (CUSBT) is regressed by the independent variables change in inflation rate (CINF), change in interest rate (CBIR), change in per person income (CPPI), change in GDP growth rate (CGGR). Table 2 represents the results of the regression model.

**Table 2: Regression analysis of selected independent variables**

Abbildung in dieser Leseprobe nicht enthalten

The regressed result shows that the change in inflation rate has a negative relationship with the change in exchange rate USD/BDT. But the other three independent variables change in interest rate, change in per person income and change in GDP growth rate, each of them has positive relationship with dependent variables. The T-stat value also shows us that change in interest rate, change in per person income and change in GDP growth rate has a significant relationship with the dependent variables change in exchange rate USD/BDT, as all of their T-stat values are significantly different from zero. But for inflation rate the result shows us that as the T-stat value is very close to zero so the relationship between change in inflation rate and change in exchange rate USD/BDT is occurred simply by chance. On the basis of the value of the R-square, it is postulated that 96.7149677% variation of change in USD/BDT can be explained by the independent variables. Moreover, the adjusted R-square which only considers the independent variables that are significantly associated with dependent variable, amplifying the statement about R-square by providing the value of 96.6016908%. It indicates the selected independent variables can reasonably explain the variation of dependent variables by the value of Adjusted R-square 96.6016908%

## 4. FORECASTING EXCHANGE RATE

To forecast exchange rate, it is needed to calculate the percentage change of each of the independent variables by considering immediate previous two periods. After that, the coefficient value of each of the independent variable will be multiplied with the percentage change rate of each of the independent variables.

### 4.1. Regression equation:

CUSBTt = b0 + b1(CINFt-1) + b2(CBIR t-1) + b3(CPPI t-1) + b4(CGGR t-1)

= -0.002420502 - 7.14705E-06(0.00969) + 0.57653468(0.003384) + 0.06550586(0.002666) + 0.334152987(0)

= -0.0002944

### 4.2. Forecasting exchange rate for August 2015 :

In August, 2015 the exchange rate of USD/BDT will be = $0.0128802199 * (1-.0002944) = $0.012876427 ≈ $0.01288.

It has to be noted that in August, 2015 the actual exchange rate between USD/BDT is $0.012853470437018 or ≈ $ 0.01285

### 4.3. Comparison between Forecasted exchange rate and Actual exchange rate:

our forecasted value is = $0.01288- $0.01285 = $0.00003 more than the actual exchange rate of USD/BDT.

## 5. CONCLUSION

This paper is signified by the method it explains how to forecast the exchange rate very accurately. Historical data is implemented with an aim to illustrate that, with this method how much the forecasted exchange rate differs from the actual exchange rate. This result may differ if more data were considered. It should be noted that in the panel data regression model approach the more data is considered the more accurate result will be gained. Another important thing is there are other considerable significant variables exist which have substantial impact on the exchange rate. By considering them the users whether individual or corporation can forecast the future exchange rate of any currency more accurately. To ensure a sustainable growth and profit any corporation or individual trader can implement a panel data regression model approach which is time effective. However, since inflation and deflation of a currency may determine the profit for many traders so that the described method in this paper is just a little effort to help those who wants to make their own forecast and decision while trading in the foreign exchange market or participating in the international trade. This method will also applicable and time effective for those multi-national corporations which are associated with foreign exchange, export, import, foreign direct investment.

## REFREENCES

1. Copyright, Reserved, A. R., displayed, T. S. L., automatically, served, party, third A., advertisers, the, … s, var. (2016). Thefinancialexpressbd.Com. Retrieved December 30, 2016, from http://www.thefinancialexpressbd.com/old/index.php

2. Exchange rates. Retrieved December 30, 2016, from https://www.bb.org.bd/econdata/exchangerate.php

3. Madura, J., Madura, P. J., & Madura (2010). *International financial management, abridged edition - 10th edition* (10th ed.). Boston, MA, United States: South-Western/Cengage Learning.

4. The daily Star Web edition Vol. 5 Num 393. Retrieved December 30, 2016, from http://archive.thedailystar.net/2005/07/05/d50705050346.htm

5. Unb, D. (2007, December 27). Taka gets stronger against US dollar. . Retrieved from http://archive.thedailystar.net/newDesign/news-details.php?nid=16710

6. Retrieved December 30, 2016, from http://www.thefinancialexpressbd.com/2014/12/26/72806

**[...]**

- Quote paper
- Sajjad Hossine Sharif (Author), 2015, Forecasting the exchange rate of currencies. A panel data approach, Munich, GRIN Verlag, https://www.grin.com/document/351362

Publish now - it's free

Comments