Thursday, March 29, 2012

Big Data in Supply Chain: Better Economic Forecast from the Cloud


Someday soon, the Federal Reserve may be dead, at least for economic information. Blame it on, or credit it to, cloud computing.


I recently visited the Oakland, Calif., headquarters of GT Nexus, a company that monitors supply chain information — things like ordering, payment, manufacture and logistics — for 100 major companies. Think of GT Nexus as a Facebook for corporate behavior: business partners are friends, and the completion of tasks are status updates.


The clients inside this cloud-based system include Home Depot, Sears Holdings, Adidas, Pfizer, Procter & Gamble, Hewlett-Packard and Fiat Industrial. Each giant company typically has about 1,000 other organizations involved in its business, including suppliers and manufacturers, freight companies, customs officials and banks. GT Nexus keeps straight all their goods, services and the payments involved, in real time, on a cloud-based computing platform. It is possible because so many people, things and computers are now connected to the Internet.


Using a laptop, I saw a $270 pair of Adidas soccer shoes was ordered in the United States on Oct. 25, manufactured in Lianjiang, China, on Dec. 25 at 04:00 Pacific time, and air-freighted to the United States on Jan. 6. I saw 4,941,554 units of the anti-anxiety drug Xanax were sent from Pfizer’s Puerto Rico factory to Narita airport in Japan, at a temperature appropriate to keep the drugs safe, ahead of the tsunami anniversary.


I saw that investors in Mauritius had bought Caterpillar’s largest truck, and shipped it (in pieces, to avoid taxes) to South Africa, where it was assembled for final delivery to a mining project in Congo. I zoomed up a level, and looked at thousands of container cargoes as they moved around the planet.


As thrilling, or geeky, as that may sound, something important is going on. GT Nexus can abstract all the data among all these supply chains into one really big picture, and get a remarkably clear view of all kinds of real-time world activity.


Its customers have total revenues that amount to $500 billion, about 2 percent of the world’s gross domestic product. You can take a number like that and start to extrapolate an overall picture of the economy. In fact, that is exactly what economists at the Federal Reserve and the Commerce Department do, though not quite as slickly, when they crank out those all-important numbers on economic growth and gross domestic product.


Looking at the activity of cargo shipments over several years, we saw flat activity in the recession year of 2009, then an increase (compared with a year earlier) in August 2010. That was from companies ordering goods ahead of the Christmas season, though they waited until the last minute because they were not yet confident that the consumer would shop. In 2011, the activity began as early as May, indicating the big companies felt more confident.


So far this year, things are above last year’s levels, and payments are made relatively quickly, indicating the economy is still doing well.


Take a purchase order, say from J.C. Penney, for clothing by Under Armour; the clothing is manufactured overseas and shipped to the United States. That order, mixed with the orders from many more retailers, provides a snapshot of expectations about how much and what people will buy, 270 days before the products are in stores. Can a government economist do that?


As GT Nexus, and companies like it, build out these kinds of systems, their information becomes more telling about the whole world. “Globalization was instrumental to our growth,” says Aaron Sasson, GT Nexus’s founder and chief executive, who has spent 13 years and $30 million of the money he and his brother made selling another software company. “China was very popular five years ago. Now customers are moving to Vietnam, Bangladesh, Kazakhstan, wherever the costs are lower. We just move the system with them.”


Mr. Sasson has not yet made a product from all this data, but sees the possibility. Customers are not used to sharing their information so readily, and he needs to build trust. Recently 45 companies contributed their data in order to figure out the average time a cargo stayed in port. Should they choose to pool all their data, however, they could have one of the most powerful and reliable economic indicators. If they published it, the world would pay attention, even if the companies themselves got, say, a 30-minute lead time on the release of the info.


It is also possible to merge different databases and model future consumption, a feature that could someday shape corporate and government policy. In one scenario, a shipment from Shenzhen to Chicago was modeled for different pollution outcomes, depending on whether it was sent via the ports of Los Angeles; Tacoma, Wash.; or Norfolk, Va., as well as the total carbon produced by the ship, rail cars and trucks used along the way. In an ideal world, a route through Tacoma, which used a lot of rail in the United States, produced the least carbon. But since many of the rail cars were reserved, another database chimed in, the Tacoma shipment would actually require a lot of dirty trucking. Long Beach, Calif., worked better.


It is not clear that GT Nexus has this market won. As they build out their clouds, companies like SAP and Oracle may be able to deliver similar insights across industries and economies, in something near real time. The greatest challenge may be in teaching their customers to share data, because the computer technology, from sensors at the edge to big servers in the cloud, is already in place.


Often enough, successful technologies change society itself. This kind of cloud-based information is rapidly creating huge pools of data in real time, which can be related to other pools of information like never before.


http://bits.blogs.nytimes.com/2012/03/15/better-forecasts-from-the-cloud/?nl=todaysheadlines&emc=tha26_20120316

No comments: