Foreign Exchange Markets - Research Paper Example

Forecasting Currencies in 2014 1. Introduction 1.1. Rationale of the assignment The research paper illustrates and discusses aspects of currency rate forecasting. The following forecasting techniques were discussed and evaluated for profitability: purchasing power parity, the balance of payment, monetary approach, asset market approach and the technical analysis techniques. The data which was used in the study is the latest currency data in the market which encompassed four states: Japan, US, Turkey and Australia. The research also explains the implications of data frequency and linearity in the global arena. The paper also provides insight into the two major methods of analysis which are used to forecast the behavior of the Forex market. It is depicted that the broad categories on which the methods lie are; technical and fundamental analysis. These two are the broad spectrum under which the smaller subdivisions occur. These two classifications have the same objective; tools of forecasting forex trade directions. 1.2. Null hypothesis In this research one null hypothesis was; foreign exchange rate normally exhibit complex patterns which can be termed as non-linearity of the exchange rates, this pattern may be profitably exploited by the concerned parties. 1.3. Challenges There was a huge base of information to collect from, this required thorough analysis. The number of currencies in the world today is overwhelming which brings the comparison problems. Another challenge was the continuous fluctuation of the currencies in the market. The dollars normally fluctuate on different occasions hence uncertainty. 2. Literature review 2.1. Terminology 2.1.1. Linearity According to the oxford dictionary linearity is the property of having one direction. The term is used to express the concept that the model poses properties of addivity and homogeneity. Change in one variable leads to a proportional change in another variable. Thus, linearity is the property of having dimensions of occurrence on more than one framework. 2.1.2. Data set For reliability purposes the data set is based on hourly observations of the US dollar and the Australian dollar, Turkish Lira and the Japanese Yen in 2014. From the hourly observations on these currencies the daily results were recorded. The difference exhibited between the currencies was assessed and the direction of the currency was depicted. The data exhibited a number of relations which also aids decision making on currency forecasting. Some of the data was also collected from the secondary sources which emanated from different national archives. 2.1.3. Technical analysis Method of predicting the movement of prices of currencies in the near future; It is done by the study of charts of past market actions or trends. This aspect is more concerned on what has happened in the market rather than what should happen hence taking into considerations the price instruments and volume of trading; creates charts from the data set it obtains for analysis. 2.1.4. Fundamental analysis This method employs the aspects of economy, political, environmental and statistical to gauge its analysis on the currency prediction. These aspects rely on the basic forces in the market which is the demand and supply. Fundamental instruments convey the information on the instruments instantaneously. 2.2. Research articles The research was carried out on both observable attributes and the statistical analysis of prior work in the field. It means that the information which had previously been recorded in the national research institutes played a major role to bring out the data which was needed in the analysis of this assignment (Shamah, 2003). 2.3. Summary, Comments and Criticisms The fundamental analysis model spawns the impacts of economic and financial relationships between different states to arrive at a forecast. This takes into considerations the short term and long term horizons and their respective models. On the other hand the technical approach relies on the historical data of particular states to arrive at a forecasted exchange rate system. This means that for accuracy of the data to be ascertained by the use of this approach it has to be a short term horizon (Shammah, 2003). There are some other subsidiary methods of currency forecasting, this includes efficient market approach and performance of the forecasters. These two aspects of forecasting are not too often put into practice because of the amount of discrepancies they hold. For example the performance forecasters is immensely prejudiced and skewed to one side of judgement. The forecaster will tend to manipulate the figures to suite their own perspectives. Coming up with a 100% accuracy in prediction may be challenging especially with the varying trends in the currencies in the world. The inconsistent trend depicted by the world currencies makes it cumbersome to come up with accurate data to showcase the global trends. Australian Dollar, Japanese Yen and the Turkish Lira have recorded a significant change in their strengths against the US dollar. For instance in 2014 1 Australian dollar is equivalent to 0.89 US dollar, 1 Japanese Yen is equivalent to 0.0099 US dollars and 1 Turkish Lira is 0.45 US dollars. This is quite different from the same currency factors in the year 2013; this is the year when the dollar raised a record higher against the other currencies in the globe. The impact was catalysed by the effects and actions Japan and the rest of the nations undertook to strengthen their currencies in the long run (Sarno & Taylor, 2005). Inconsistencies in the money market can be attributed to political unrest and general currency weaknesses. Employing the fundamental and analytical approaches in trying to come up with a clearly defined forecast can be a cumbersome task because of these factors. For instance, if forecasting of Japanese Yen against the US dollar is done and yet Japan is in a political crisis, the data obtained will never give the right structure model as anticipated. There has to be a common ground on which well-articulated assumptions have to be made before forecasting is done (Rogoff, 2002). Recognition of the various perspectives is always very important to analyse the requirements of the forecasts. According to Win Thin who came up with a forecast model of the four states (see appendix 1) a forecaster should consider the volatility of the currency in the market. Also the economy of the country will tend to dictate the direction of fluctuation of the currency. If the economy drops the currency will tend to follow in the same direction un-proportionally (Hanson, 2005). 3. Methodology 3.1. Sample data and period The exchange rate information was obtained from a number of sources in the country and other global companies; this information is an accurate and authenticated data that has been collected for the past 2 years on a continuous monthly basis. In case of huge fluctuations experienced for example, during the Euro crisis the data of such circumstances is recorded as an abnormality. The information depicted here is the US dollar relative to other currencies in the global arena. There is an adverse use of the Siegel paradox which to large extent depicts the superiority of the regression and the forecasting of these currencies. This makes it important to generate the log of the daily exchange rate status in the economy. Each information series was accurately tasted for any abnormalities and non-stationary implements through the method of creation of a mean series in the accounted data. The data further is continuously compounded to a certain percentage in each currency market; the main reason for undertaking this analysis is to come up with effective data systems to portray almost the same implications as the real data in the field. To formulate an out sample period, there was a transformation of exchange rates and splitting them into two portions. This is based on the observations which were done in the different periods, thus, 2014 data was selected as the best sample to illustrate or used in forecasting. From the year 2012, the observations carried out in this period were termed as out of sample from which the forecasting of these models was to be done. However, the in sample was depicted by approximately 25% of the data which to high degree was an ideal formula for the assessment to be done (Finke, 2006). The in sample led to the development of the ARCH model which was responsible for the estimation of; R t is the higher order in the period which is categorized as the structural architecture of the period; weeks, months, days. Box-Jenkins approach was also amalgamated in the model to aid the analysis of the autoregressive structures, a number of tests were then carried out which were either applied on R t or the resultants of AR model, this was to detect the ARCH effects in the model. The LM test and LBS1 and LBS2 tests were assigned on the data. The Q20 statistic model played an integral role in trying to harmonize the data and come up with the right model of evaluation (Yu, Wang & Lai, 2007). The currencies which depicted the presence of the ARCH effects then the ARCH and GARCH models were used in analysis and forecasting. According to the econometric theory there is no direct formula of telling which model is the best to the forecaster; there are numerous challenges in model selection as each of the models conveys the same results. However the Box- Jenkins approach suggests otherwise; the lowest model with the largest sample is to be the best choice. Because there are a number of limitations on the ARCH mode, this paper suggests that an alternative approach be absorbed; regression systems (et2 = a + bst2 + nt). Thus, according to regression there should be a selection which maximizes on R t. The ARCH model is selected because it produces a series of results which can be compared favorably to verify the correctness of the information it contains. If a conjunction between the AR and ARCH model is carried out in the data the result may be as depicted in appendix 2. The dependent variable R t is assumed to be a past value enhancing prediction of the present data. In this study data collection encompassed a broad range of sources, sensitivity of the financial reports archived by some of the companies was a challenge to access and retrieve the information (Wang, 2009). 4. Results and discussions Mostly in the international arena transactions are taken or settled as per the future. This is where the importance of the exchange forecasts to enable the international cash flow mechanisms in the transactions. It is through the interest rate forecasting that a given organization will be able to evaluate the magnitude of the risks it is subjected to in the market. Forecasting is all about the expectations in the near future value of given assets. Presently there are two broad procedures on which to forecast exchange rates; the fundamental and technical approaches (Macdonald, 2010). 4.1. Fundamental approach This process encompasses a wide range of data which can be termed as fundamental economic variables which influence the exchange rates in the market; the economic variables in this approach are collected from the economic models in the market. Some of the variables which are included; GNP, Consumption, Inflation rates, Unemployment level,, Interest rates among many other economic attributes in the entire economy. The modification of these models is done to fit the statistical aspects of the data and the forecaster’s experience in the field. Experts in this field make use of the structural models to come up with equilibrium exchange rates. These rates are then used for the projections of the trading signals. A trade signal can be established each moment there is a difference in the forecasted exchange rates and the exchange rates in the market. If a difference lies between these two; actual and forecasted rates then the practitioner will be in a position to determine the real cause of the problem; whether it is the effect of the pricing techniques in the market or the risks which underlie production. If a difference emanates from the concepts of pricing then a buy or sell signal is generated. A fundamental procedure commences with a model which is capable of producing a forecasting equation. This aspect can be based on a number of theories for example, PPP or other combinations of variables which can be beneficial to the practitioners. On the basis of the initial data the forecaster collects the data which is required in the formulation of the equation. If it occurs to the forecaster that the information contained in the data is important then the forecaster proceeds to the next step; generation of forecasts the lasts step is the evaluation of the forecast. As explained above, forecasting is all about the future expectations of values. In this paper I will outline the future value of the exchange rates in the market, S1. The information selected by the forecaster is the one which is used to establish the equation to be constructed. Time t, is the time the information set is availed, thus the notation can be described as S1+ t hence the resulting equation is depicted as E (S1+t). Where E is the expectation at time t. there is an associated forecasting error in each equation, in this case the error can be depicted as e t +1. Hence the full description is; e t +1= S1+1-E (St+1) (David & Stewart, 2010). The use of the forecasting error will be to judge the quality and extent of the forecast which will be made. A common standard used in this scenario is the Mean Square Error (MSE). This can be defined as; MSE=[(et+1)2 + (et+1)2 + (et+3)2 +…+(e t+Q)2]/Q. in this case Q represents the number of forecasts, the higher the magnitude of the MSE the less accurate the model will be in forecasting (see appendix 1). Two kinds of models can be grouped in this approach; the in sample and out sample. The in-sample model works best within the sample which is arrived at while the out-sample works outside the sample (James, Marsh & Sarno, 2012). 4.2. Technical approach This model does emphasize on the smaller subsets of the data. It is generally based on the prices in the market. The model does not rely on the fundamental aspects in the economy, it only chooses on the extrapolations of the price trends in the market. The analysis focuses on the repetition of the prices and the associated patterns which can be collected. The computer models which lie squarely under this model try to point out the major trends and critical turning points which are the main points of generating the trade signals; purchase signals (Finke, 2006). The models in this category are very simple and rely on the moving averages (MA) or the momentum of the indicators. MA models: the main purpose of this model is to align the frantic movement of daily activities in price aspects; this will aid the major signals or trends. The model can be described by the following equation; SMA = (St + St-1 + St-2 + ... + St-(Q-1))/Q. there are two aspects which can be illustrated in this model for example, if the most recent prices are used then a short run (SRMA) equation will be used but if it’s a longer series of price data then a long run (LRMA) is used. Buying signals in the model is showed by the past rates. For instance if a currency is experiencing a downward movement then the SRMA will be below LRMA it is this relationship which triggers the signals of buying, see appendix 1 (Coyle, 2000). Filter models: the most popular of all the models in this category, it is essentially based on the findings that prices depict a small autocorrelations. Price increases will tend to be associated with increases while the price decreases tend to be associated with decreases, the signals which emanate from this information can be used to estimate the impacts on profits. The model relies on showing when the system shows a certain change in the trends as opposed other factors which may not be very important. When an exchange rate rises X% there is a buy signal created however when there’s an X% decrease from its initial point there is a sell signal (Hanson, 2005). 5. Conclusion Exchange rate forecasting is an important framework of money market analysis. Continuous monitoring of the system is very important to the stakeholders who will want to spearhead their productivity in operations. The models described above are very crucial analytic movement in forecasting. However, there lie a number of factors which hamper the efficiency of the process, for example, cost and time aspects (Wang, 2009). A graph of time series is an essential approach to the determination of the facts used for forecasting. At times a hybrid system is used to trace data movement; trends in the market are very essential in that it encompasses almost all models making it one of the most efficient approaches to exchange rate forecasting. A closer look at the four countries; US, Japan, Australia and Turkey will depict that none of this country is a third world nation hence there is a slight economy difference depicted in their GDP. It is their strong correlation which puts exchange rates forecasting between the states to be applicable. If a third world was to be involved in the analysis the final results will obviously be incorrect s the stronger currency will tend to melt the weak currencies in the market. With the continued strengthening of the Lira against the US dollar, interest rates in Turkey are continuously revised to keep in check the inflation rate (Coyle, 2000). References Coyle, B. 2000. Foreign exchange markets. Canterbury, Financial World Pub./BPP. David, P. A., & Stewart, R. D. 2010. International logistics. Mason, Cengage Learning. Finke, S. 2006. SAP foreign currency revaluation FAS 52 and GAAP requirements. Hoboken, N.J., Wiley. Hanson, D. 2005. CE marking, product standards and world trade. Cheltenham [u.a.], Elgar. James, J., Marsh, I. W., & Sarno, L. 2012. Handbook of exchange rates. Hoboken, New Jersey, John Wiley & Sons, Inc. Macdonald, R. 2010. Currency Union and Exchange Rate Issues. Cheltenham, Edward Elgar Pub. http://public.eblib.com/EBLPublic/PublicView.do?ptiID=534835. Rogoff, K. (2002). The Failure of Empirical Exchange Rate Models: No Longer New, McGraw. Sarno, L., & Taylor, M. P. 2005. The economics of exchange rates. Cambridge [u.a.], Cambridge Univ. Press. Shamah, S. B. 2003. A Foreign Exchange Primer. Chichester, John Wiley & Sons. http://public.eblib.com/EBLPublic/PublicView.do?ptiID=155667userid=^u. Wang, P. 2009. The economics of foreign exchange and global finance. [Berlin], Springer- Verlag. http://public.eblib.com/EBLPublic/PublicView.do?ptiID=429163. Yu, L., Wang, S., & Lai, K. K. 2007. Foreign-exchange-rate forecasting with artificial neural networks. New York, Springer. Appendix 1 In US terms Currently 2013 2012 2011 2010 Yen 103 107 109 106 109 Australian Dollar 0.89 0.87 0.85 0.88 0.89 Lira 78 88 100 90 89 Appendix 2 Rt = a 0 + 1Rt-1 + et t = 1,......,T et ½yt-1 ~ N ( 0 , ht ) (1) ht = a0 + a1 et-12 Read More