INTERVENTION AND FORECAST MODELS FOR THE PRICE PAID TO PRODUCER OF BEE (Apis mellifera L.) HONEY IN MEXICO

Abstract

Bee (Apis mellifera L) honey is one of the oldest foods that humans have used. Since ancient times, it has been used as a healthy product due to its sweetening and healing properties. In 2020, Mexico produced 54 121 tons (Mg), which ranked the country as the tenth largest producer in the world. The hypothesis was that current honey prices can be explained by previous prices and that they influence the increase in the population of hives and the production of honey in Mexico. To test this hypothesis, the objective of this research was to develop a forecast model for the annual average prices of honey in Mexico (AAPH). The data comprised the 1966 to 2019 prices and the Box-Jenkins methodology of Autoregressive Integrated Moving Average (ARIMA), with and without intervention, was used. The parameters of the models were estimated with the maximum likelihood method of the SAS® software, while the structural change was calculated with the corresponding library (strucchange) of the R software. A model based on the AAPH series was adapted for the 1966−2019 period and validated with data from 2018 and 2019. The series presents five periods of trend structural changes of AAPH: 1966−1985; 1986−1995; 1996−2003; 2004−2008; and 2009−2019. The best estimated model without intervention was ARIMA (1, 1, 1) and the best model with intervention was ARIMA (1, 1, 0), which indicates that the prices of previous years can explain the AAPH. The predictions had a mean absolute percentage error (MAPE) of 8.16 % for the model without intervention and 4.02 % for the model with intervention. Both estimated models suggested that the AAPH have an upward trend in the medium term. The ARIMA model with intervention provided a more accurate estimation of the AAPH and information to plan and make decisions for the next five years.

Evaluating Time Series Prediction Models: Egg Prices in Mexico

“…In both comparisons, only the ARIMA models with interventions meet the assumptions of white noise and normality in the residuals, in addition to having lower variance. Luis-Rojas et al (2019) found as the best model for egg price prediction the ARIMA (0,1,1)(1,0,1) [12], with an Akaike index of -332.538 and a variance of 0.0104. According to these authors, the model's short-term predictions deviate by 12.38 % from the observed data.…”

“…In Venezuela, a statistical model was used to analyze inflation with a time series covering the period from June 1995 to July 2000; applying the Box-Jenkins methodology, the model that best fit the time series was the ARIMA with a confidence interval of 95 % (Seijas, 2002). Luis-Rojas et al (2019, 2022 used ARIMA models for the prediction of the egg price paid to the producer, taking sorghum and maize production as auxiliary variables. For the case of climatic events, a time series study was conducted in Peru with the monthly rainfall of two different regions using the ARIMA model; depending on the region, the ARIMA structure of the rainfall of the fluviometric stations can be different or similar (Reyes, 2014).…”

Relación del precio de la alta fructosa y el precio de la miel de abeja mexicana

“…La apicultura en México como generadora de divisas se ubicó entre los tres primeros lugares en el sector pecuario. El producto principal de esta actividad es la miel, que constituye uno de los alimentos más antiguos que el hombre ha aprovechado para nutrirse (Luis-Rojas et al 2022). El decrecimiento de las colmenas a nivel mundial se asoció con el cambio climático, la deforestación y el uso de pesticidas en las parcelas agrícolas (Rossi et al 2020).…”