Munich Personal RePEc Archive

Seasonal Unit Roots and Structural Breaks in agricultural time series: Monthly exports and domestic supply in Argentina

Mendez Parra, Maximiliano (2015): Seasonal Unit Roots and Structural Breaks in agricultural time series: Monthly exports and domestic supply in Argentina.

[img]
Preview
PDF
MPRA_paper_63831.pdf

Download (503kB) | Preview

Abstract

Monthly time-series data based on agricultural commodities tend to present strong and particular patterns of seasonality. The presence of zero values in some of the seasons is not explained by the absence of reporting but is the result of actual features of agricultural processes. Seasonal unit root tests have never been applied to data that exhibit these characteristics, with a consequent lack of critical values to be used in the inference. Monte Carlo simulations are performed to obtain critical values that can be used for this type of data. In addition, seasonal unit roots under the presence of unknown structural breaks have never been applied to any kind of monthly time series, with the associated absence of critical values to be used in the testing procedure. Monte Carlo simulations are also performed to tabulate these critical values. It is observed that the presence of zero values does not invalidate the critical values available, with or without unknown structural breaks; the values obtained here for the monthly seasonal unit root tests under unknown structural breaks can be used in any other kinds of exercise. A seasonal unit root test with more power is also considered and critical values are obtained to perform the inference. The capability of the seasonal unit root tests to select the right break date is analysed, with some divergent results with respect to previous findings. An application of these techniques on the monthly quantities of exports and domestic supply of three agricultural commodities in Argentina between 1994 and 2008, which observe the patterns of seasonality described, is presented. Although, some evidence of stochastic seasonality has been found in some of these series, in general a deterministic approach can adequately describe their seasonality

UB_LMU-Logo
MPRA is a RePEc service hosted by
the Munich University Library in Germany.