Models to forecast currency value are abundant. In order to gain a greater grasp of the way currencies are forecasted a literature review of the prevalent literature must be undertaken to understand the way currencies should be forecasted. The aim of this literature review is to understand the variables, the statistical models used and statistical models are used. This paper aims to delineate the key variables used for forecasting the Euro and the statistical models used and the significance of the results found by previous researches on currency forecasting. The paper is divided into three sections – variables, statistical methods and the significance of their findings. For this study, we have reviewed six articles on currency forecasting and report the variables that have been used by them for their research.
Data Collection
Hauner, Lee and Takizawa devised a model to forecast currency (3). They conducted a survey of the market expectations and using the data from the survey they undertook. A comparison of the existing forecasting models existing in currency literature and the expectancy data they collected were used for the research. They used data from 55 advanced economic countries (Hauner, Lee and Takizawa 5).
The second paper reviewed is by Neely and Weller who studied the aspects of technical analysis used in the foreign exchange market (Technical Analysis in the Foreign Exchange Market 23). The importance of the paper is the study of market efficiency of the currency. The paper provides a considerably detailed study of the literature on technical analysis and used dollar exchange rate data from the 1970s and 80s for the analysis.
A study conducted by Charles Engel showed how the exchange rate models have changed over the years. The study tried to evaluate the various exchange rate models available in the literature. The review provides various methods used in literature for studying exchange rate forecasting.
Marcellino, Stock and Watson uses time series data from two levels – country level data, and aggregate area data to demonstrate the changes in exchange rate. Rimea, Sarnoc and Sojlie (72) aimed at solving the difference between the macro level and micro level studies of exchange rate prediction. Benavidesa and Capistrán (628) studied the volatility in the change in the exchange rates of the currency under study.
Variables
The variables Hauner, Lee and Takizawa (6-7) have used for the study are as follows:
- Appreciation in percent of the currency vis-à-vis USD
- The short term interest rate as a percentage which is the money market rate
- Inflation – using purchasing power parity data they assumed that
- Real GP
- Current account balance as a percentage of GDP
- Budget balance as a percent of GDP
The study by Neely and Weller did not point out any particular variable but looked at the price of the dollar i.e. the exchange rate over a long and short term period. The first variables used are the price and the volume of transaction of the currency. Second, variable used for the study is asset prices. The third variable used is an axiom that traders will tend to act in similar manner when the environment becomes similar.
The variables that Engels found to be robust through his meta-analysis of the exchange rate model literature are the changes in prices of asset and the expectations of the future fundamentals (Engel 5).
Marcellino, Stock and Watson (5) uses the following variables for the study – real GDP, inflation, industrial production index, and unemployment rate.
Rimea, Sarnoc and Sojlie uses the different variables like exchange rate movement, order flow, and the expectations of the economy of the different macroeconomic variables. The fundamental of the economy play a vital role as a variable in this study. These reflect the heterogeneous expectations of the macro fundamentals of the market. The paper tries to associate the currency and the macro markets.
Benavidesa and Capistrán (629) use the variables such as the individual models, unconditional models, and hybrid forecast model. They used their relation between Mexican peso and USD.
Statistical Method
The study conducted by Hauner, Lee and Takizawa (6) used the exchange rate measure developed by Reinhart-Rogoff. The statistical measure used for analyzing the data is regression. This was used to see the correlation between the independent variables mentioned in variables section with the exchange rate expectations.
The method employed by Neely and Weller was to follow the principles of technical analysis which are mentioned above and tried to identify “trend and reversal trends” (Neely and Weller 24). They conducted an exploratory study of the recent changes in exchange rate to predict the trend of the future change in exchange rate.
Engel’s study pointed out that the statistical methods used in exchange rate literature have been varied. Their study used no statistical method. However, they concluded in their study that most of the studies are based on statistics, but these studies did not pertain to understanding that the change in exchange rate is mostly due to the change in expectations of the change in fundamentals of the economy. That is why they suggested the usage of Keynesian macroeconomic theory to understand the change in exchange rate (Engel 5).
Marcellino, Stock, and Watson used a forecasting model to demonstrate the exchange rate change. The dependent variable for the study is logarithm of GDP. They used autoregressive benchmark forecast, VAR forecast, AR with US variables, and principle component forecast (Marcellino, Stock and Watson 11-12). They used F-test for comparison of the sample data.
Rimea, Sarnoc and Sojlie used forecasting model and the explanatory model to depict the changes in exchange rate. They used high exchange rates of one year of three main exchange rates (79).
Findings
The findings of the study by Hauner, Lee, and Takizawa (10) showed that exchange rate models are inflation and growth rates. The other variable used in the study did not show statistically significant role in exchange rate formation.
The study by Neely and Weller show that the intervention by central banks and the changes in the trading rules provide a robust explanation as to the changes in exchange rates based on the trading rules used in economics. Further, psychological factors found, have a strong influence on the forecasting of currency rates. The study conducted by Engel therefore suggests that the monetary policy targets at changing the exchange rate as well as target inflation change (Engel 5).
The findings of research by Marcellino, Stock and Watson (20) using univariate model shows that the more accurate method to predict exchange rate is through inclusion of the dynamic factors.
Rimea, Sarnoc and Sojlie’s study shows that the findings of the research are as follows – (1) flow of the order is correlated to the macro fundamentals of the economy, and (2) daily movement of the exchange rates are shown in the current movement of the exchange rate.
Benavidesa and Capistrán (632) show that the ARCH-type forecasting model is found to be the best indicator of exchange rate.
Bibliography
Benavidesa, Guillermo and Carlos Capistrán. “Forecasting exchange rate volatility: The superior performance of conditional combinations of time series and option implied forecasts.” Journal of Empirical Finance, 19(5) (2012): 627–639. Web.
Engel, Charles. “Exchange-Rate Models.” Fall 2006. The National Bureau of Economic Research. Web. 9 August 2013.
Hauner, David, Jaewoo Lee and H. Takizawa. “In Which Exchange Rate Models Do Forecasters Trust?” IMF Working Paper No. 11/116 (2011): 3-14. Print.
Marcellino, Massimiliano, James H. Stock and Mark W. Watson. “Macroeconomic Forecasting in the Euro Area: Country Specific versus Area-Wide Information.” European Economic Review, Elsevier, 47(1) (2003): 1-18. Web.
Neely, Christopher J. and Paul A. Weller. “Technical Analysis in the Foreign Exchange Market.” Working Paper 2011 Federal Reserve Bank of St. Luise (2007): 23-38. Print.
Rimea, Dagfinn, Lucio Sarnoc and Elvira Sojlie. “Exchange rate forecasting, order flow and macroeconomic information.” Journal of International Economics, 80(1) (2010): 72–88. Web.