The Palm Oil Industry: Visualized Through Big Data

Includes: JMHLY, WLMIY
by: Gabriel Thoumi, CFA

Investor herding behavior can be visualized using big data.

Wilmar and Jardine Matheson have less investor herding than competitors.

Herding behavior can be a relevant risk measure.

Image credit: ©Getty Images/slpu9945

Published by the CFA Institute: Enterprising Investor

By Kalev Leetaru and Gabriel Thoumi, CFA, FRM

Palm oil is an inexpensive and highly versatile vegetable oil derived from the fruit of the oil palm tree. It can be found in half of all consumer goods in Western grocery stores, from chocolate, ice cream, and baked goods to soaps, lotions, and detergents.

Palm oil is the highest-yielding vegetable oil crop as well as the most actively traded vegetable oil in the world. As a petroleum alternative, it can power vehicles, heat homes, and manufacture plastic.

And with annual sales of around $50 billion, palm oil is also a big business, albeit one with oversized risks.

Over the past 15 years, palm oil plantations in Indonesia and Malaysia have expanded, tripling their production. The two countries now produce around 85% of the global supply. From one million hectares (ha) in 1990, Indonesia’s oil palm estate grew to 21 million ha in 2015. Now oil palm concessions lease more than 10% of the nation’s total land area.

After falling 4.8% in 2015–2016, the first decline in 18 years, due to El Niño and related climate change effects, global palm oil production is expected to increase by 7.3%, to 63 million tons, in 2016–2017, according to FitchGroup’s BMI Research forecasts.

Risks Grow Too

The expansion in palm oil production has come with considerable costs, and analysts need to consider the environmental, social, and governance [ESG] impacts. Most analysts have no background in natural resource science, so are unfamiliar with ecology functions or how the holistic ecological strength of plantation assets are key revenue drivers. For example, analysts may use the same growth and yield calculations for an oil palm plantation in Sumatra as for valuations of investments in West Africa, despite very different solar radiation and rainfall patterns.

Slave and child labor are often part of the palm oil supply chain. To make way for plantations, companies frequently come into conflict with local and indigenous communities. In 2012, 59% of Indonesia’s roughly 1,000 palm oil firms were involved in land disputes.

The 2015 fires in Southeast Asia are blamed in part on oil palm cultivation. These fires and the resulting haze may have contributed to 100,000 fatalities in Indonesia, Malaysia, and Singapore, according to estimates from Harvard and Columbia universities. The economic toll of the fires exceeded $16 billion - more than double the recovery costs of the 2004 Indian Ocean tsunami and equivalent to 1.8% of Indonesia’s gross domestic product [GDP] - according to World Bank calculations.

Palm oil is the leading cause of deforestation in Indonesia and a major reason why the country is the sixth-worst greenhouse gas emitter in the world. Clearing forests for palm oil production has also pushed Bornean orangutans and Sumatran elephants and tigers, among other species, to the brink of extinction.

Visualizing the Risks

Given the critical human and environmental risks posed by palm oil, how can we better understand the macro-level structure and governance of the palm oil industry?

In particular, is the sector a collection of independent companies with little connection to one another? Or is it a highly interdependent and interconnected web of firms that share officers, board members, and investors?

To answer these questions, we gathered information from public sources as well as from Bloomberg and other financial databases. Using this data, we wrote a set of PERL scripts to construct network diagrams of officers, board members, and shareholders. We then visualized these networks with open-source Gephi software to create a set of interactive charts.

From a simple spreadsheet listing each Southeast Asian palm oil and agriculture company and their senior executives, we constructed a graphic that visualized overlapping executive roles, related governance risks associated with overboarding, and investor positions.

The most common relationships we found among the companies are board interlocks. The figure below shows the board members of each company. You can click on the image to open an interactive visualization in your browser and zoom into any portion of the network to see more detailed information. Clicking on any node reveals the other companies and board members it connects to. All information is current as of Q2 2016.

To map the macro-level structure of the network, we applied a technique called “community finding.” Board members, executives, and companies that are more closely connected than the rest of the network share the same randomly assigned color. In addition, we used Google’s “PageRank” algorithm to determine the relative “importance” of each node to the specific network. The larger the node, the more “important” it is.

Southeast Asia Palm Oil Industry Board Interlocks Networks

outheast Asia Palm Oil Board Interlocks Network

Data derived from publicly available sources as of Q2 2016 and may contain errors.

We employed the same approach to map the major shareholder connections (view interactive version) in the palm oil sector.

As with the board interlocks visualization, groups of investors and companies that are more connected than the rest of the network have the same randomly assigned color, and node size is based on the node’s relative “importance” as determined by Google PageRank. The thickness of the line between a shareholder and a company is a function of the percentage of the company’s outstanding shares owned by that shareholder. The thicker the lines, the greater the ownership.

Southeast Asia Palm Oil Industry Shareholders Network

alm Oil Industry Shareholder Networks

Data derived from publicly available sources as of Q2 2016 and may contain errors.

A key finding demonstrating shareholder clustering, from the upper right quadrant of the visualization: Wilmar International (OTCPK:WLMIY) (green) and Jardine Matheson (OTCPK:JMHLY) (blue) each has a cluster of investors who only own shares in one of the companies, a cluster of shareholders who invest in both but not others, and then an array of investors who invest in one or both as well as other companies.

This shareholder clustering is common among many of the companies surveyed.

Using the interactive visualization, you can click on any company to see its investors as of Q2 2016. You can also search for a specific shareholder or company by name to filter the network since such macro-level trends as overboarded executives or interlocking boards are not so easily discerned by a manual, company-by-company analysis.

This is a pilot project based on publicly available information, so some data may not be accurate.

By demonstrating what’s possible with public information and advanced visualization tools, we hope to encourage a new way of thinking about how industries with these sorts of challenges can be understood and investigated.

Our work demonstrates that using rudimentary data visualization tools, investors can begin to assess other investors' herding behavior and to judge insider behavior from a sector point of view. Furthermore, with this herding behavior described visually, it is now possible to understand how the ability outside investors have to influence an insider-led market.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. Business relationship disclosure: Kalev Leetaru, PhD, and Gabriel Thoumi, CFA, FRM wrote this article for the CFA Institute: Enterprising Investor.

Editor's Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.