Bwog’s resident carbon confidant, Zach Kagan, brings us sordid tales of sustainability, statistics, and snacks.
This midterm season, an incalculable number of sodas and assorted caffeinated beverages will be drunk. Scores of candy bar pick-me-ups will be scoffed. Bag after bag of vending machine purchased chips will be opened and consumed. Exams and snacking go hand and hand this time of the semester, but as you cram for your sustainable development midterm, you may pause and wonder about the sustainability of your own increasingly junk food fueled diet. As it happens, PepsiCo wondered the exact same thing and, thanks to engineering professor Christoph Meinrenken, they know quite a bit more about it.
But PepsiCo didn’t just want to analyze one product. It wanted to determine the carbon footprint of each of its 10,000+ products, not an easy task. Determining the footprint of something requires so-called “life-cycle assessment,” following the product on each stage of its development from bare materials to consumption. Each step of production has its own set of emission factors that needs to be accounted for. Each requires a different team of experts to analyze and compute the footprint. That data needs to then be used to create a meaningful description of the carbon footprint of an individual product. To repeat that process for each of PepsiCo’s products would take an exorbitant amount of time, money, and patience. Prof. Meinrenken sensed that this problem could be solved much more efficiently, but to do so he would have to think outside the box.
“That’s when my competitive juices started flowing,” said Dr. Meinrenken when I sat down with him this past week. The result was an algorithm that can accurately calculate the carbon footprint of almost any consumer good (given the right information, of course). But, most importantly, it doesn’t need all of a product’s most intimate details to get the job done. Instead, it looks at each product’s bill of materials (a list of all the stuff that goes into making the product) and makes educated estimations about the emission factors associated with each part of the product’s life-cycle. This is important because most producers really have no idea of the emissions they create by transporting their goods or refrigerating them, but the model can estimate those numbers fairly well. Meinrenken compares it to Netflix’s film recommendation system, but instead of suggesting you watch The Shawshank Redemption because you’re an 18-24 male who enjoyed Morgan Freeman in the Dark Knight, it’s estimating that the emissions of farming corn for the production of Tostitos chips is probably similar to the emissions of farming corn for Fritos.
But Meinrenken’s model can do more than snack-based social media data crunching. It also has a sophisticated way of fact checking itself. In statistician lingo, PepsiCo’s algorithm avoids the problem of “garbage in, garbage out” by recognizing when data may in fact be garbage. Big companies like PepsiCo are usually good at taking data about their day to day activities, but sometimes there are mix-ups. It’s not unheard of for a Lay’s factory to produce thousands of bags of chips without reporting a single potato used. Meinrenken’s model can spot these inaccuracies and bring them to attention before they screw up the model’s results. Similarly, the model can identify what step in a product’s life cycle contributes the largest emission factor so that the user can determine which data is most important to measure accurately.
The result is a model that can turn a massive and somewhat unorganized data-set into tangible carbon footprints for each product listed. That list can then be used by PepsiCo to identify which stages of production contribute most to carbon emissions. For example, the Columbia team first project with PepsiCo was to calculate the carbon footprint of Tropicana Orange Juice. They were surprised to find that a sizable portion of emissions didn’t come from transportation or packaging, but instead from the petroleum based fertilizers used in the farming of the oranges themselves. Thus by just switching to natural, compost based fertilizers, PepsiCo could dramatically reduce Tropicana’s carbon footprint.
But the cynical truth behind Meinrenken’s analysis is that most companies, including PepsiCo, already know where they can stand to improve carbon emissions, even without models and life-cycle assessments. Companies use these tools to make business decisions about sustainability and it becomes an issue of branding more than an issue of environmental well-being. However, Meinrenken’s model does show a sizable (50%) correspondence between emission factors and production cost. That means the more carbon intensive a stage of production is, the more likely it is to be expensive as well. While the sustainable alternative may be more expensive still, it goes to show that there are market forces in action when it comes to emissions.
Sadly, it may take economic incentives to get people to take sustainability seriously in their everyday lives. There’s a lot of little things that each of us can do, like cutting out meat and animal products from our diets and unplugging unused electronics. Actually making these commitments, however, is easier said than done. “It’s frustrating that there are totally known methods to promote sustainability,” Dr. Meinrenken said, “but we have a hard time convincing people to give a fuck.”
This article has been updated to correct a few factual inaccuracies.