Nowcasting Commodity Markets Using Real Time Data Stream ...

Available online at www.sciencedirect.com

ScienceDirect Procedia - Social and Behavioral Sciences 213 (2015) 481 – 484

20th International Scientific Conference Economics and Management - 2015 (ICEM-2015)

Nowcasting commodity markets using real time data stream Andrius Guzaviciusa, * a

Kaunas University RI7HFKQRORJ\.'RQHODLþLRJ.DXQDV/7-44249, Lithuania

Abstract Research purpose. Because of the complexity of financial instruments and difficult economic situation today instead of forecasting it is useful to have real-time and reliable data, that might identify current stage in commodity market with holistic approach. Methodology. Different nowcasting models using heterogeneous set of predictors, including different statistics, as also surveys. Main results. The results of the research provide readers with understanding of nowcasting approach regarding regional heterogeneity in commodity market. © 2015 2015Published The Authors. Published by Elsevier Ltd. by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of Kaunas University of Technology, School of Economics and Business. Peer-review under responsibility of Kaunas University of Technology, School of Economics and Business Keywords: Nowcasting; Commodity Marke; Real Time Data Stream.

Introduction There are different scientific opinions, that some financial indicators have power to predict market action providing readers with understanding of nowcasting approach regarding regional heterogeneity in commodity markets. Nowcasting can be defined as the prediction of the present, the very near future and the very recent past. Until recently, nowcasting had received very little attention by the academic literature, although it was routinely conducted in policy institutions on the basis of simple models. Crucial in this process is to use timely information in order to nowcast key commodity markets variables. Because of the complexity financial instruments and difficult economical situation today instead of forecasting it is useful to have real-time and reliable data (Schumacher, Breitung, 2008; Giannone, Reichlin, Small, 2008; Chauvet; Piger, 2008; Das, Ester, Kaczmirek, 2011; Kuzin, Marcellino, Schumacher, 2012; Banbura, Giannone,

* Corresponding author. Tel.: NA. E-mail address: [email protected]

1877-0428 © 2015 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of Kaunas University of Technology, School of Economics and Business doi:10.1016/j.sbspro.2015.11.437

482

Andrius Guzavicius / Procedia - Social and Behavioral Sciences 213 (2015) 481 – 484

Modugno, Reichlin, 2012; Amisano, Geweke, 2013; Askitas, Zimmermann, 2013; Stock, Watson, 2014; Aastveit, Gerdrup, Jore, Thorsrud, 2014). Methodology of nowcasting commodity market by using real time data stream is designed to include essential characteristics of the holistic economic approach creating statistical model which produces a sequence of nowcasts in commodity market. 1. Nowcasting integrating statistical models incorporating real time available data Share of primary commodities in global markets has declined over last time, however, fluctuation in commodity markets still affecting global economy. Price forecast in commodity market remain important part of export activities, especially for developing countries. Nowcasting of price of primary commodity markets is a key factor to macroeconomic policy planning. The basic principle of nowcasting is the exploitation of the real time data stream at higher frequencies in order to obtain early estimate in commodity markets. Additionally considering information containing forward looking financial indicators. It is necessary to use timely information and real time data stream from various sources at wide range of frequencies in a nonsynchronous manner and with different degrees of delay dealing with very large information set. This framework provides a comprehensive approach dealing with nowcasting based on multivariate dynamic models by monitoring many real time data and using. Kalman filter to generate projections for many variables. Methodology of forecasting in commodity market is designed to include essential characteristics of the holistic economic approach. There are three types of price forecast in commodity markets: based on judgments (qualitative analysis with variety of factors), based on historical price data (quantitative analysis) and forecast based on supply and demand analysis see Fig. 1.

Nowcasting based on statistical models relying on historical data systematically incorporating real time available data.

Forecast based on qualitative analysis of variety of factors

P

Gaussian prediction of initial state of the system and Gaussian measuring of state of the system at time t=1 using Kalman filter

y S

D Q Analysis of supply (S) and demand (D)

x Forecast based on quantitative analysis (x,y)

Fig. 1. Modeling nowcasting integrating statistical models relying on historical data systematically incorporating real time available data

Techniques of nowcasting have been based on simplified heuristic approaches and can be formalized in a different statistical models. Models collecting information from a large quantity of data series at different

483

Andrius Guzavicius / Procedia - Social and Behavioral Sciences 213 (2015) 481 – 484

frequencies and with different time lags. Signals about the direction of change in main economic indicators can be extracted from heterogeneous set of information sources, data are used to compute the real time flow of data. Nowcasting can be based on statistical models relying on historical data but also incorporating real time available data. 2. Prediction of initial state of the system fusing the data from prediction and the measurement Techniques of nowcasting have been based on simplified heuristic approaches and can be formalized in a different statistical models. Models collecting information from a large quantity of data series at different frequencies and with different time lags. Signals about the direction of change in main economic indicators can be extracted from heterogeneous set of information sources, data are used to compute the real time flow of data. The methodology allows to process a large amount of information, it provides an explicit link between the news in consecutive data releases and the resulting forecast revisions. Many different courses have come to varying opinions regarding prediction of price in commodities market. Sometimes futures prices in commodities market are not successful by forecasting of future spot price inefficiencies might be observed in commodity markets with limited liquidity (Avsar, Goss, 2001). The commodity market representing sector, that plays a main role in facilitating economic activities between sectors and across regions, and can be useful in monitoring and identifying cyclical turning points the current state of the economy, however, regional heterogeneity and different approaches regarding liquidity in financial markets and data availability are necessary. Y

Y

Gaussian initial state of the system at time t=0

X

Gaussian prediction of state of the system at time t=1

X

Y

Y

Gaussian prediction of initial state of the system and Gaussian measuring of state of the system at time t=1

X

Gaussian prediction of initial state of the system and Gaussian measuring of state of the system at time t=1 by fusing the data from prediction and the measurement

Fig.2 . Gaussian prediction of initial state of the system fusing the data from prediction and the measurement

X

484

Andrius Guzavicius / Procedia - Social and Behavioral Sciences 213 (2015) 481 – 484

Going from initial state of the Gaussian function system at time t=0 to prediction of new system with new variance and mean when t=1. This step is represented by new Gausian function with new variance and mean representing reduced certainty in the accuracy from time t=0 to time t=1. The best estimation of position of the system is provided by combining knowledge from the prediction and the measurement by multiplying two Gaussians' together see Fig. 2. However, the prediction and measurements must be made in the same coordinate frame with the same units. This has to be a particularly concise pair of equations representing prediction and measurements update stages. However, in reality a function is required to map prediction and measurements into same domain. One of the Gaussian function properties is possibility to make new functions by connecting to different Gaussian functions. This important issue ensure endless number of Gaussian functions to be multiplied over time with no effect on complexity. Conclusions Methodology of nowcasting commodity market by using real time data stream might be designed to include essential characteristics of the holistic economic approach in financial markets and create a statistical model which produces a sequence of nowcasts in relation to the real time releases of various economic data regarding commodity market. Methodology of forecasting in commodity market is designed to include essential characteristics of the holistic economic approach. There are three types of price forecast in commodity markets: based on judgments (qualitative analysis with variety of factors), based on historical price data (quantitative analysis) and forecast based on supply and demand analysis. Nowcasting in commodity market is using information and real time data stream from various sources at wide range of frequencies with different degrees of delay dealing with very large information set. This framework provides a comprehensive approach dealing with nowcasting based on multivariate dynamic models by monitoring many real time data and using Kalman filter to generate projections for many variables. Nowcasting price using real time data in commodity markets is required to map prediction and measurements into same domain in order to make new functions by connecting to different Gaussian functions. This important issue ensure endless number of Gaussian functions to be multiplied over time with no effect on complexity. References Aastveit, K., K. Gerdrup, A. Jore, and L. Thorsrud (2014). Nowcasting GDP in real time: Adensity combination approach. Journal of Business and Economic Statistics 32, 48{68. Amisano, G. and J. Geweke (2013). Prediction using several macroeconomic models. Working Paper Series, European Central Bank 1537, European Central Bank. Askitas, N & Zimmermann, KF (2013), ‘Nowcasting Business Cycles Using Toll Data’, Journal of Forecasting, 32, 299-306. DOI:10.1002/for.1262. Avsar, S., A. Goss, (2001). “Forecast Errors and Efficiency in the U.S. Electricity Futures Market,” Australian Economic Papers, 40, 479–499. Banbura, M., D. Giannone, M. Modugno, and L. Reichlin (2012). Now-casting and the real-time data flow. In G. Elliott and A. Timmermann (Eds.), Handbook of Economic Forecasting Volume 2 . Elsevier North Holland. Chauvet, M. and J. Piger (2008). Comparison of the real-time performance of business cycle dating methods. Journal of Business and Economic Statistics 26, 42{49. Das, M, Ester, P & Kaczmirek, L (2011)., Social and behavioral Research and the Internet, Advances in Applied Methods and Research Strategies, Routledge, New York et al. Giannone, D., Reichlin, L., & Small, D. (2008). Nowcasting: The real-time informational content of macroeconomic data. Journal of Monetary Economics, 55, 665–676. Kuzin, V., Marcellino, M. and Schumacher, C. (2012). Pooling versus Model Selection for Nowcasting GDP with Many Predictors: Empirical Evidence for Six Industrialized Countries. Journal of Applied Econometrics. Schumacher, C. & Breitung, J. (2008). Real-time forecasting of German GDP based on a large factor model with monthly and quarterly data. International Journal of Forecasting, 24, 386–398. Stock, J. H. and M. W. Watson (2014). Estimating turning points using large data sets. Journal of Econometrics 178, 368{381.