Research of "forecasting foreign exchange rates"

1

Research Paper (Forecasting Foreign Exchange Rates)

Student’s Name

Institutional Affiliation

Corse Code & Title

Instructor’s Name

Assignment Due Date

Research Paper (Forecasting Foreign Exchange Rates)

Background

The calculation of exchange rates used to be fairly straightforward. There had been long stretches of exchange rate stability, mostly during the Bretton Woods Regime (a system placed in place in 1944 once the representatives of allied countries gathered at Bretton Woods to create a stable economic model out of the turmoil of World War II). Despite their flaws, pegged systems have a level of certainty in currency price forecasts (Huang & Nakamori, 2004). There were big system modifications if a currency deviated too far from economics, but there have been few shocks so because changes could be predicted well in progress. This is not the case anymore, however, exchange rate projections are used as the foundation in many, if not all, global business and financial decisions today. Speculations interest rate and exchange rate projections provide companies and banks with the ability to make significant profits. The slow changing of currency fluctuations, coupled with the increased globalization of industry, has created a need for forecasting methods.

Forecasting, in particular, necessitates the assumption of a collection of relationships between the variables. To put it another way, economic forecasting requires the use of models. Forecasting methods of exchange rate are based on structured models and may depend on a presumed chain of causal relationships (e.g., simulation models) or on the data-driven construction of measuring the relationship between both the variable of interest and cumulative sum of the same set (intrinsic models) and the measurements of different exogenous variables in the past (extrinsic models) (Nag & Mitra, 2002). One aspect unique to both experimental and causal and extrinsically statistical analyses would be that their forecasting performance depends on the expectation that relationships formed in the past will continue essentially stable into the future. It makes very little difference if this partnership's essence is defined in terms of even a logical or theoretical context or a statistical dependency. The forecasting method it is not the only requirement for effective forecasting (Huang & Nakamori, 2004). A far more basic need would be that other analysts' behavior does not negate any potential benefit from accurate prediction. Recent advances in statistical analysis theory have resulted in the widespread use of forecasting approaches that include fitting a functional relationship to the sequence's true performance and trying to extrapolate it into the future.

Statement of the Problem and Thesis Statement

World currency is being exchanged daily against each other to the number of billions a day. Through such trading, each currency is weighted and weighed against another by a currency exchange rate. The price of money measured by another currency is known as an exchange rate. The question then becomes: what allows currency fluctuations to fluctuate, and how can future value be predicted? Model detection and model validation are key parameters involved in time series analysis. According to most studies, the identification problem is more complicated (Nag & Mitra, 2002). Estimating the input variables is typically simple once the suitable type of the model is known. (need to be rewritten). It is important to define a design that accurately describes a time series to find one. Make sure you understand the model's intent. (need to be rewritten). Is the primary goal of the model to clarify the essence of the mechanism that generates the sequence? Or whether the model to be evaluated on its ability to forecast future values of the time series(need to be rewritten).; consequently, to come at a framework that highlights the current elements of the series, a choice qualification, which combines model, fit, and It, is necessary to use model complexity (Sermpins et al., 2013). (need to be rewritten). This study aims to find answers to various questions about foreign exchange rate prediction and the best forecasting methods.

Research Objectives

i. To find out the different ways used in forecasting the foreign exchange rates. (show me where in the paper)

ii.

iii. To understand the different models used in forecasting the foreign exchange rates. (show me where in the paper)

iv.

v. To find out the reasons behind currency fluctuation and how this affects the foreign exchange forecasts. (show me where in the paper)

vi.

vii. Understand the different macroeconomic variables and parameters that are used in exchange rate forecasting computation. (show me where in the paper)

Literature Review

Exchange rate fluctuations heavily influence trade and deep economic policies. Economic output is heavily impacted and influenced by these fluctuations. The impact of exchange rate fluctuations on a country's economic effectiveness is primarily due to its reliance on domestic production and the comparative advantage in exports. The state can significantly impact lending activities to improve the economy during these fluctuations, worsening or improving the currency exchange price (Huang & Nakamori, 2004). Financial indicators such as interest rates, gross domestic product, money supplies, commodity prices, and trade account balances have long been considered indicators of the real effective exchange rate in basic exchange rate models. The exchange rate is defined as a highly stable currency in which all information is public, and traders have the same perceptions with no advanced understanding over one another.

Exchange Rate Systems

The main determining factor of an exchange rate is trust in the stability of that specific currency. Choices based on anticipated future events can affect the currencies (Sermpinis et al., 2013). In many situations, choices made today are dictated by the current exchange rates and the forecast rates. There are four exchange rate systems: fixed exchange rates managed floating exchange rates, Free-floating exchange rates, and the Semi-fixed exchange rate. The state (or the banking system functioning on its behalf) conspires in the stock market to maintain the exchange rate stable to a fixed target in a completely fixed exchange rate regime (Nag & Mitra,2002). It adheres to a single fixed exchange rate but does not allow significant deviations from that rate. These are the criteria behind fully fixed exchange rates.

On the other hand, Currencies can shift within allowed fluctuation levels in a semi-fixed exchange rate system. The most important goal is the exchange rate. Making monetary system Interest rate charged to fulfill the expectations, and the exchange rate has a set target. (need to be rewritten).

A floating exchange rate mechanism is one where exchange rates can fluctuate based on the current market factors rather than national governments intervening. Changes in international demand and supply trigger the value of the currency to fluctuate through floating exchange rates. On the other hand, a free-floating exchange rate is uncommon. Most governments attempt to manipulate the exchange rate by interest rate adjustments at some stage (Sermpinis et al., 2013). The exchange rate in a free-floating exchange rate system is calculated primarily by market supply and demand forces in the foreign exchange market. The key influences on currency fluctuations and other measures are competition and foreign investment.

Exchange Rate Variables

Exchange rate fluctuations are relevant because of their impact on the amount we spend on imported goods and the prices we earn for manufactured goods, our exports, and the sum of money that flows into or out of the market. For instance, when the US Dollar rises (value increases), Exports become more appealing, and foreign buyers must seek more US dollars to purchase the same amount of exports. When the US dollar price goes down (loses value), exports are less expensive, and imports become appealing. Because of the weakening of the US currency, US exports are now much more competitive globally (Shen et al., 2015). Currently, foreign buyers of American items must find fewer dollars to purchase the same amount of exports. Import prices can be reduced in any domestic economy, lowering output costs and inflation rates. Other exchange rate variables include the capital inflow and outflow of a particular currency, the currency demand and supply, international competition, and macroeconomic effects.

Messe and Rogof Model

In international economics, the difficulty of forecasting future exchange rates has long been a concern. Exchange rate models are evaluated using the prediction experiment suggested by Richard Meese and Kenneth Rogoff. Meese and Rogoff analyzed the relationship between real exchange rates and lending rates over the current flexible rate era. They concluded that fundamentals such as placed money supplies, real wages, short-term interest rates, predicted inflation variations, and cumulative trade balances all influence the exchange rate (Shen et al., 2015). The basic premise is that commodities market rates change slowly regarding expected disruptions and excess demand. As a result, economic disruptions that are not ideally expected will trigger significant variations in the real effective exchange rate itself from the long-run equilibrium position. They discovered that even a random walk model performed just as well as a linear regression model. Models are based on factors that can be used to forecast exchange rates. In summary, their results indicate that economic fundamentals such as the balance of trade, supply of money, economic output, and other main variables are ineffective in predicting exchange rates among states with comparable inflation rates.

Random Walk Model

Many proponents of structural and economic analysts believe there is no systematic connection between stock price directions from one day to another. They think that prices should find a level that balances market forces. This amount will be achieved unexpectedly, as prices fluctuate in response to the most recent available data or news release. Many well-defined trading techniques, usually on statistics and signal processing, would fail if the random walk theory is right. Price expectation is the best argument against proponents of the random movement. One might claim that all market players understand precisely where prices should go. The perceived spontaneous variation of prices, except uncertainty, is dependent on both the period and the rate of data used (Chao et al., 2011). When using a long period, such as one to two decades, and averaging the information to monitor the synthesis protocol, the prevailing features, as well as temporal and periodic fluctuations, would change. Moving averages and other technical approaches are used to separate these price features. The accumulation of data into annual prices would smooth out erratic daily movements and significantly strengthen the relationship between various prices. Using daily data over a long period creates distortion and obstructs consistent patterns.

Most price volatility reaches a degree of equilibrium in the long run, and exhibits mean reversion over an amount of time. However, the short-term price change can be very different from a random set of numbers. It often exhibits two distinct characteristics: unusually long price runs in a horizontal path and asymmetry, or the unequal scale moves in opposite directions. (this section is not related to the topic or need more information to be related to the topic which is Forecasting Foreign Exchange Rates)

Research Methodology

Most of the time series probability theory suggests that data series have a pattern. A Stationary is defined as having a consistent variability over time. In certain cases, non - stationary time series elements may be omitted to make the series more stable (Chao et al., 2011). For instance, a fixed position to eliminate patterns or to remove variations in a time series. Differences in the seasons' Predictions are made using forecast equations that follow the method. Models parameters are estimated using this method. (need to be rewritten). As a result, the estimation method defined regulates how the forecast generates outcomes. The forecast method allows the user predict one or more time series. It does not allow you to classify models or assess their suitability (Huang et al., 2004). That's why it's crucial to analyze and choose the best forecasting model. Once it is known that there is a quantitative distinction between data, a formula can be established that describes one price movement in terms of the other prices and data, based on the measurement of dependency and correlation properties. These methods perform better when applied, which has previously been used, such as rates are within the variety of relevant data. When variables are extrapolated outside of previous events, forecasting accuracy drops dramatically. Other common approaches would experience the same phenomenon.

Estimation and Forecasting Methods

Methods for estimating time series are focused on observational data analysis. These approaches back up the idea that past data trends could be used to forecasting potential data points. The major statistical methods include Moving averages, Mathematical models, Exponential smoothing, and the Box-Jenkins approach.

Least-Squares Regression

The least-squares regression analysis could be applied to establish the correlation among two dependent variables or to discover how markets are rising when driven by recognized related factors. A simple evaluation process could be used to assess the model qualities of this method. Assume that there is still an extensive price sequence for currency fluctuations, and we would want to know how often prior quarters are optimal for forecasting the next quarter’s price (Shen et al., 2015). The average error in the predictions is used to find the response. The best predictive model could be determined using error analysis and applied with forecasts. This is especially evident when analyzing the error of two separate prediction models for about the same time since it eliminates the bias introduced by different length intervals. It's also a good idea to take the analysis a step further and include forecast error data. This provides a measure of out-of-sample prediction precision and boosts confidence in the method's predictive abilities.

Regression

Linear regression is a correlation between two sets of data that follows a straight line. It is most commonly found using the best match method, which chooses the normal distribution close to much of the data points. Provided a data series and a time on which the line will be constructed, the regression analysis slope returns a straight line slope (Sermpinis et al., 2013. The regression line slope is calculated, and the line is then projected into the future, returning the valuation of the future stage. The consumer must determine the data sequence, the time in which the path will be measured, and the number of future periods. Projected price = starting price + slope x (calculated period + projection period). (this section is not related to the topic or need more information to be related to the topic which is Forecasting Foreign Exchange Rates)

Box Jenkins

The intervention of Box and Jenkins would be to emphasize the consistency of the solution. They specified that the moving average and auto-regression steps could be confined to first- or second-order analysis. To accomplish this, the data had to be de-trended and thereby rendered stationary. De-trending was most easily achieved by differencing the data, generating a new series by deducting each previous term, t-1, from the subsequent time, t-2. By applying all of these procedures in sequence, the ARIMA system must note all of these modifications, or transformations, to restore the final forecast to the correct price notation (Huang et al., 2004). If the Box-Jenkins method fails to find a stationary solution, the details are still not static, and further differencing is required. Moving average exponential smoothing, random walk model, a multivariate analysis, and a Box-Jenkins transfer function model to evaluate the forecasts. Finally, they compare the forecasting approaches based on the actual current price during that period.

Findings

Even though it is difficult to predict the unknown, our research shows that various forecasting approaches have proven to be more accurate for particular currencies. Forecasting exchange rates can be done using a variety of variables. The type of forecasting approach used can be influenced significantly by the exchange market (Chao et al., 2011). If a currency uses a fixed exchange rate, where local currency is linked to a single currency, a specific measure of value, or some other particular standard of value, the prediction would be substantially different than if the exchange rate uses a floating exchange rate mechanism. The market determines the value of a currency in a floating exchange rate system. Since there are so many factors and factors that can influence foreign exchange rates, it's best to predict using a combination of techniques to find the best match for that particular local currency.

Forecasting is not only possible but also required in certain situations. Suppose it's a profit-driven foreign exchange transaction or a business loan from a foreign bank. In that case, various local variables such as government controls, inflation, and weather disasters must be considered. Forecasting is not only possible but also required in certain situations. Suppose it's a profit-driven foreign exchange transaction or a business loan from a foreign bank. In that case, various local variables such as government controls, inflation, and weather disasters must be considered (these two sentences are repeated twice). The price of a currency tends to shift in the same line as the price of a GDP unit (Chao et al., 2011). Inflation and money supply are many other variables that are related (need to be rewritten). The purchasing power of a currency unit depreciates as a result of inflation. Since a large amount of money tends to devalue demand for money, money supply and inflation are related. The different methods highlighted above can be used to estimate and predict the exchange rates between other currencies.

Conclusion

The forex market often does not shift in a predictable manner. Emotions, judgments, and politics may all have an effect on rates. Price forecasts that are purely dependent on data are insufficient (need to be rewritten). You must also be able to predict potential demand information in advance of the release of new statistics. The majority of foreign exchange transactions are derivative trades that trigger rate movement. Until a method for forecasting speculation is developed, it appears that the best predictions of currency exchange rates are based on current and historical prices.

References

Chao, J., Shen, F., & Zhao, J. (2011, July). The forecasting exchange rate with deep belief networks. In The 2011 International Joint Conference on Neural Networks (pp. 1259-1266). IEEE.

Huang, W., Lai, K. K., Nakamori, Y., & Wang, S. (2004). Forecasting foreign exchange rates with artificial neural networks: A review. International Journal of Information Technology & Decision Making, 3(01), 145-165.

Nag, A. K., & Mitra, A. (2002). Forecasting daily foreign exchange rates using genetically optimized neural networks. Journal of Forecasting, 21(7), 501-511.

Sermpinis, G., Theofilatos, K., Karathanasopoulos, A., Georgopoulos, E. F., & Dunis, C. (2013). Forecasting foreign exchange rates with adaptive neural networks using radial-basis functions and particle swarm optimization. European Journal of Operational Research, 225(3),

Shen, F., Chao, J., & Zhao, J. (2015). Forecasting exchange rate using deep belief networks and conjugate gradient method. Neurocomputing, 167, 243-253. 528-540.