A non-conventional model to explain and forecast prices and trade flows of food staples across developing nations in Southeast Asia

2017 Fiscal Year Annual Research Report

• Project/Area Number: 17H07180
• Research Institution: Waseda University
• Principal Investigator: Quek Olivia 早稲田大学, 国際学術院(アジア太平洋研究センター), 助教 (60803517)
• Project Period (FY): 2017-08-25 – 2019-03-31
• Keywords: Spatial economics / Trade / Arbitrage / Food security

Outline of Annual Research Achievements

In order to understand the factors which determine the comovement of rice prices within and across countries, we studied in great detail the political economy that governs the trade of rice in each individual nation. Next, we downloaded data on the prices and trade flows of rice within and across nations, and visualized the data to understand the overall patterns that are present. However, there is a lot of noise in the data that prevents us from finding any statistically significant relationship between the magnitude of price gaps and that of trade flows. When analyzing within country data, we also discovered many pairs of net trade flows that have a sign opposite to what we would expect. We are currently in the process of solving this problem using deep learning algorithms.

Current Status of Research Progress

3: Progress in research has been slightly delayed.

Reason

First, we found that there is a lot of noise in the data and that this prevents us from finding any statistically significant relationship between the magnitude of price gaps and that of trade flows. For example, when analyzing within country data, we discovered many pairs of trade flows that have a sign opposite to what we would expect. We are currently in the process of solving this problem using deep learning algorithms.
Second, I delivered a baby in December 2017 and was not able to proceed with my research as quickly as I envisioned during my maternity leave.

Strategy for Future Research Activity

We are currently in the process of capturing the broad patterns in the data. Due to a high degree of statistical noise (such as misreporting and a lack of updates), statistical techniques have to be employed to make better sense of the data. Once we have achieved this purpose, we shall employ statistical methodologies such as Principal Component Analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE) techniques, to better understand the comovement of prices between trading partners.