Rice import forecasting with ARIMA and Halt Winters methods

Thesis for obtaining a master's degree in business management, marketing orientation

Abstract

In this research, the value of rice imports for the next 5 years has been predicted using the import value of these products during the period 1360-1391. To make the prediction, first, the predictability of the series was checked by the Wald-Wolfwitz ??randomness test. Then non-causal regression methods including ARMA and ARIMA as well as Holt-Winters non-regression methods were compared based on the minimum mean squared error, finally the best prediction model was selected for each series of data and based on that prediction was made for the next 5 years. The results of the research show that the value of rice import based on the Wald-Wolfwitz ??test is non-random and predictable. Then, to predict the value of rice import stationarity, the data was checked with the generalized Dickey-Fuller test, and the value of rice import was stationary at the level and width from the origin and without differentiation at the 5% error level. After that, the forecasting methods were compared with the criterion of minimum mean squared error, and the ARMA method was chosen as the best model for the rice import value series. In the end, by using the selected method, it was predicted for the next 5 years that the value of rice imports for the coming years is on the rise. Introduction

So far, many researches have been done on the methods of predicting economic variables. Time series models are often used for short-term forecasts and try to explain the behavior of a variable based on the past values ??of that variable (and possibly the past values ??of other variables that we wish to predict). These models are able to provide an accurate prediction of the desired variable even in cases where the economic model has an uncertain infrastructure. Unlike econometric models, time series models only use statistical data for forecasting and do not pay attention to the theoretical foundations of economic theories.  Univariate time series models only relate the current values ??of a variable to its past values ??and current and past error values. Multivariate time series models try to explain the behavior of a variable based on its past values ??and a number of different variables simultaneously. In this research, univariate time series have been used to predict the amount of rice imports in the next 5 years. The necessary data of this product during the period of 1360 to 1391 has been collected annually in the form of Fob price from the annuals of foreign trade statistics of Iran in different years.

In the present study, according to the amount of rice imports in the past years and by using two ARMA or ARIMA and sub-seasonal Halt-Winter methods, the amount of rice imports for the next 5 years has been predicted and estimated.

Introduction

Considering the growing population of countries and the limited production resources, providing the consumables needed by society is one of the most essential factors in gaining economic independence. Foreign trade is one of the most important economic sectors in developing countries, and in this sector, import is very important, on the one hand, as one of the factors of growth and gross domestic product, and on the other hand, it is considered as one of the important items in the balance of payments of each country. In the foreign trade of Iran's economy, import is one of the categories whose importance and position has always expanded, therefore, any change and development that occurs in the country's import will have a significant impact on the process of production, growth and development. Considering this and considering the structural dependence of various sectors of Iran's economy on imports, it is very important to know the dimensions of this issue. It is compiled and using the two methods of Box Jenkins and Halt Winters, it is non-seasonal for the next 5 years.

1-2 Statement of the problem (definition of the research topic)

Foreign trade is one of the most important economic sectors in any country, especially in developing countries, and in this sector, import is very important. On the one hand, it is considered as one of the factors of growth and gross domestic product, and on the other hand, it is considered as one of the important items in the balance of payments of each country.. In the foreign trade of Iran's economy, import is one of the categories whose importance and position has constantly expanded, therefore, any change and development that occurs in the country's imports will have a significant impact on the production, growth and development process. Considering this and considering the structural dependence of various sectors of Iran's economy on imports, it is very important to understand the dimensions of this issue. 

For proper planning regarding the amount of allocated currency, the investment process in the construction of new industrial units, the amount of consumption of these products and many industrial and non-industrial policies, forecasting the amount of rice imports in the future is very important. Because in any policy-making, the basis of action is not only the existing situation, but short-term and long-term forecasts of the studied variables are also among the influential indicators. The challenges in predicting time series variables have been mainly affected by the evolution of the methods and tools provided for forecasting. As the importance of predicting time series variables has caused the diversity and breadth of tools. Of course, it should be noted that depending on the nature of the available data, the appropriateness and predictive power of these tools are different from each other. But what can be deduced from the overview of the studies is the comparison of the prediction power of different methods, based on some criteria. Forecasting methods are divided into quantitative and qualitative methods depending on how much mathematical and statistical methods are used in them. In quantitative methods, data related to the past are analyzed with the aim of predicting the future value of the desired variable using statistical and mathematical methods, on the other hand, qualitative methods involve subjective estimation through the opinions of experts.

In this research, due to the high ability of quantitative methods in predicting economic variables, a regression method (Box Jenkins) and a non-regression method (non-seasonal Halt Winters) have been used to predict rice imports. And we are looking for an answer to this question, is the import of rice predictable? And what is the best method to predict the amount of rice import? 1-3 The importance and necessity of the research topic According to the latest statistics of the Iranian customs, the import of rice to the country decreased by 14.96% in terms of weight, but it faced a growth of 4.21% in terms of value. Last year, rice accounted for 2.03% of the country's total imports. allocated India was the main source of Iranian rice import in 2013. So that this country's share of Iran's rice imports in terms of weight and value reached 67.46 and 72.06 percent, respectively. Last year, about 870,000 tons of rice with an approximate value of 943 million dollars were imported from India. 8.94% of Iran's rice imports last year were done through other countries[1]. Since predicting future events plays a major role in the decision-making process, forecasting is important for many organizations and institutions. In addition, the forecasting of economic variables has an effective role in the policies of producers, because producers formulate and implement their policies not only based on the current situation, but also based on short and long-term forecasts of key variables. Therefore, the accuracy of predicting these variables, regardless of the correctness and appropriateness of the policies with the existing situation, are among the keys to the success of these policies (Tarazkar, 2014: 57).

After wheat, rice is the most important source of food supply in our country. The country's annual rice production is about 3.16 million tons with a conversion factor of 64% (turning rice into white rice), about 2.2 million tons of rice, and the average per capita consumption is 37-40 kg per year. In Iran, about 2 million tons of white rice are produced annually, more than 22% of the produced white rice is broken, causing great damage to the economy of the farmer and the country. The loss caused by small rice in the country is estimated to be more than 1497 billion Rials annually.

Due to the high consumption of rice in the country and the limitation of the production of this main food source in the Iranian consumption basket, there is a shortage of domestic rice annually through imports from countries such as India, Pakistan, Thailand, etc. is provided Therefore, forecasting the amount of import of this product can help the Ministry of Commerce, Customs, and other related bodies in making policies for foreign exchange allocation and other planning.