Neural Network Approach to Estimating Conditional Quantile Polynomial Distributed Lag (QPDL) Model with an Application to Rubber Price Returns
[1]
Kwadwo Agyei Nyantakyi, Ghana Institute of Management and Public Administration (GIMPA), Greenhill College, Business School, Achimota-Accra, Ghana.
In this paper we consider the estimation of the conditional quantile polynomial distributed lag (QPDL) using neural network to investigate the influence of the conditioning variables on the location, scale and shape parameters of the QPDL model developed. This method avoids the need for a distributional assumption and applies conditional quantiles approach which allows the investigator to employ a range of conditional functions which exposes a variety of forms of conditional heterogeneity to give a more comprehensive picture of the effects of the independent variable on the dependent variable. The models fitted were adequate with very high R-square and low AIC values across the quantiles. We observe the effects of the quantiles on the dependent variables through the various GAM-style plots. Also from the actual and the predicted plots we observed that there was no difference between them. The results suggest that neural network used in estimating the QPDL model offers a useful alternative for estimating the conditional density, as artificial neural networks have proven to produce good prediction results in regression problems.
Backpropagation, Effects, Feedforward, Hidden Neurons, Polynomial Transformation
[1]
Almon S., 1965, “The Distributed Lag Between Capital Appropriations and Expenditure” Econometrica 33: pp.178-196
[2]
Barron, A. R. and Barron, R. L., Statistical learning networks: A unifying view, Wegman, E., editor, Computing Science and Statistics: Proc. 20th Symp. Interface, (American Statistical Association, Washington, DC, 1988) 192-203.
[3]
Basset, G.W., Jr. and Koenker, R. 1978, ‘The asymptotic theory of the least absolute error estimator’, Journal of the American Statistical Association, vol. 73, pp. 618–622.
[4]
Cybenko, G., (1989), Approximations by superpositions of a sigmoidal function. Mathematics of Control, Signals and Systems, 2 , 303-314.
[5]
FAOSTAT, Sri Lanka Annual Data (1961-2013), [on line].[Accessed on 10.02.2014]. Available at http://faostat.fao.org
[6]
Galvao Jr, A. F., Montes-Rojas, G. & Park, S. Y. (2009). Quantile autoregressive distributed lag model with an application to house price returns (Report No. 09/04). London, UK: Department of Economics, City University London.
[7]
Geman, S., Bienenstock, E., and Doursat, R., (1992), Neural networks and the bias-variance dilemma. Neural Computation, 4, 1-58.
[8]
Hornik, K., Stinchcombe, M., and White, H., (1990), Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks. Neural Networks, 3, 551 - 560.
[9]
Hornik, K., Stinchcombe, M., White, H., and Auer, P., (1993), Degree of approximation results for feedforward networks approximating unknown mappings and their derivatives. (Discussion paper 93-15, Department of Economics, UCSD)
[10]
Koenker, R. W. and Bassett, G. W. (1982). Robust Tests for Heteroscedasticity based on Regression Quantiles, Econometrica, 50, 43–61.
[11]
Koenker, R. W. (2005). Quantile Regression, Cambridge U. Press.
[12]
Koenker, Roger and Zhijie Xiao (2002), “Inference on the Quantile Regression Process”, Econometrica, 81, 1583–1612
[13]
Scott, G.M. (1993), Knowledge - Based Artificial Neural Networks for Process Modeling and Control. Ph.D. thesis, University of Wisconsin
[14]
Ripley, B. D., Statistical aspects of neural networks, in: Barndor-Nielsen, O., Jensen, J., and Kendall, W., (1993), editors, Networks and Chaos { Statistical and Probabilistic Aspects}. (Chapman and Hall, 1993)
[15]
Ripley, B. D.,(1996), Pattern Recognition and Neural Networks. Oxford Press
[16]
Thrun, S.B. (1994), Extracting Symbolic Knowledge from Artificial Neural Networks. Revised Version of Technical Research Report TR-IAI-93-5, Institut für Informatik III – Universität Bonn
[17]
Towell, G., Shavlik, J. W. (1993), The Extraction of Refined Rules from Knowledge – Based Neural Networks. Machine-Learning, vol.13
[18]
World Bank Pink Sheet Annual Data (1961-2013), [on line]. [Accessed on 08.08.2014]. Available at http://econ.worldbank.org
