Productivity and Growth: Least Absolute Deviation estimator and bootstrap techniques to predict aggregate production elasticities in the Palestinian manufacturing industry

We propose the estimation of a log-log Cobb-Douglas aggregate production function for the Palestinian industry. We find that traditional OLS estimates are not reliable they are bad predictors), due to the fact that only limited is available (small samples), and variables are characterised by high variability across time. The oligopolistic structure of the manufacturing sector also contributes. We propose to use bootstrap least deviance technique and find that the estimated elasticities are both significant and robust. For time-saving purposes (being inefficient to set-up the panel dataset), we apply the model to three available cross-sections of 71 manufacturing aggregates: 2000, 2002 and 2006 and find increasing returns to scale, which are supposed to reflect the imperfect competition of the market and/or the existence of high set-up or sunk costs which are mandatory in order to produce at all.