Risk management usually consists of identifying the risk, quantifying the likelihood of its occurrence, and measuring its potential effect on the business. But managing risk in global supply chains is a much more daunting task. The distributed nature of a supply network, where information about potential risks resides across hundreds of external and internal stakeholders, makes it much more difficult to manage these potential hazards to the company. For example, as a company like Gap, Inc. tries to manage risk to its holiday shopping season, a factory manager in China might be the best person equipped to quantify the risk of a machine shutdown to the season's supply of cashmere sweaters.
So how can companies tap into the wealth of information that resides among stakeholders to better manage risk? And how can they do it in a structured way that maintains the integrity of each stakeholder's contribution and point of view, aggregates these insights into a meaningful measure of the likelihood of a problem, and gleans the effect of a certain risk? One viable framework is predictive markets.
Here we define predictive markets, their deployment in business to tap into collective wisdom, and their applicability to manage supply chain risk.
Predictive markets--sometimes referred to as information markets or idea futures--are speculative markets created for the purpose of making predictions of a specific outcome. Cash value is tied to an asset, either a particular event (like the election of a new president) or a parameter (like the total sales for next year). As a free market, prices are set primarily based on the supply and demand for those assets. Subsequently, the current prices of assets should reflect the group's collective prediction of the occurrence of that event or parameter.
Probably the most well-known information market is the Iowa Electronic Political Market, which currently has a nomination market and an election market. Both are real-money futures markets, in which traders buy candidate stocks and trade them in an open market. The share price for each candidate would be an accurate indicator of the collective perception of the likelihood of that candidate winning the election.
Now some global companies are using predictive markets to solicit employees' insights to forecast the likelihood of a future event or future business value of competing projects. The findings of predictive markets are meant to augment traditional decision-making approaches, like statistical forecasting, focus groups, and client input.
Google, for example, has used predictive markets to forecast such things as product launch dates and the impact of different strategic directions on the future of the business. Companies like Hewlett-Packard, Eli Lilly, Microsoft, Pfizer, and Siemens have also used them to augment sales forecasting and product feature prioritization processes, effectively any areas that benefit from another source of accurate outcome or event prediction.
Most notably in the area of supply chain risk management, Intel recently published a study about its pilot of internal prediction markets to assess demand risk. Early results suggest that predictive markets forecasts are meeting or beating traditional sales forecasts in increased accuracy and responsiveness to demand shifts and reduced volatility.
The markets are producing forecasts as much as 20% more accurate than the official forecasts. This is impressive because the official forecasts' error percentage is only in the single digits. At the time of publishing the report, six of eight market forecasts fell within 2.7% of actual sales. AMR Research's benchmarking work has shown that demand forecast accuracy is a vital predictor of supply chain responsiveness.
Interestingly, Intel experienced an unexpected benefit: employees were better able to tap into available data, fueled by the desire to base their trading on the most accurate supply chain information.
Proponents of predictive markets point to Intel's experience and similar projects in which predictive markets forecasted future events more accurately than traditional forecasting techniques. But critics need only to point to New Hampshire's Democratic primary as proof of the inaccuracy of predictive markets. The day before the primary election, the closing price of a Hillary Clinton share was $0.23, compared to a Barack Obama share value of $0.7, largely based on polls the days before the election that showed Senator Obama had surged ahead of Senator Clinton. Mrs. Clinton ended up winning the primary 39% to 36%.
Examining these data points, some strengths and weaknesses of predictive markets emerge. Below is an analysis of the pros and cons of predictive markets, specifically as it relates to their suitability for managing supply chain risk.
Here are a few strengths:
Predictive markets take advantage of individual and collective expertise: Unlike traditional methods of gathering insights, like face-to-face meetings, the markets can capitalize on collective insights of employees and trading partners. Moreover, anonymous predictive markets can ensure that these insights are not skewed by a dominant voice of a high-level supervisor or an outspoken team member. And given that the end game is to accurately forecast the outcome, every trader is motivated to make the best prediction they can, based on their expertise and available information.
Predictive markets have a low initial setup cost: Companies can quickly deploy a basic predictive market. They can build one from scratch, using basic tools like Microsoft Access or Excel, or use platforms from companies like NewsFutures or Inkling. With a small investment, companies can tap into an alternative source of insights to augment traditional approaches like using statistical forecasts or mining partners' data. Initial investment is mostly dedicated to user education and training on how to use available data to make trades most representative of their opinion.
Predictive markets not only forecast but also assess validity of risk responses: Predictive markets can indicate the efficacy of a company's actions to mitigate a risk. For example, the market might predict a high risk of supply disruptions for materials sourced from a specific supplier. To mitigate that risk, the company moves to a dual-sourcing model or sources from another supplier. Yet the market value of that risk remains high. This could imply that the company's actions did not eliminate the risk. It could point to the real culprit behind the disruption risk: not the materials supplier, but possibly the unreliability of inbound transportation resources.
Conversely, here are some of the shortcomings:
Predictive markets assume traders' actions truly reflect their beliefs: Physical proximity of the traders, for example, can sway individual trading decisions and lead to groupthink. Also, traders might shy away from negative predictions in fear of retribution. For example, if traders are predicting the company's sales forecast, they might be reluctant to express personal doubts about the company's future financial performance. Similarly, trading to predict the value of a business initiative that is sponsored by their immediate manager might be too political to solicit objective insights. Trader anonymity can help with that concern.
Predictive markets' outcome is constrained by traders' information sources: If all traders base their buy and sell decisions on information from a single source, the prediction will merely mirror that information. This was arguably exemplified by the Iowa Electronic Market, in which media coverage of flawed polls significantly boosted Barack Obama's standing prior to the primary. Similarly, the presence of traders with limited understanding of a business issue or the lack of representation of the true expert stakeholders could result in a distorted prediction of future outcomes.
Predictive markets may be unsuitable for granular, long-term forecasts: Using the market to predict the demand risk for a particular product may not be possible, given the lack of insights that traders will have on that granular level. Here, statistical modeling will probably result in more accurate predictions because it can take advantage of a wealth of historical and causal data, modeling all the possible outcomes. Similarly, a predictive market might not be able to forecast the effect of a business initiative on the company two years out or more, since traders cannot account for all the possibilities of outcomes over that far a horizon.
With their ability to cost-effectively tap into the wisdom of the crowd and use this wisdom as an augmentation to existing risk management approaches, predictive markets present a promising, innovative approach for managing supply chain risk. Early results at global companies are encouraging, but more work is needed to prove the suitability of the framework in managing risks on larger scales and in different environments.
This is the first of a series of articles that will examine innovations in technology and services in managing supply chain risk. If your company has implemented such practices, please contact us at firstname.lastname@example.org.
Enjoy curated articles directly to your inbox.