Wikipedia defines predictive analytics as a set of statistical techniques, - such as data mining, business analytics, and machine learning, - “that analyze current and historical facts to make predictions about future or otherwise unknown events.” Increasingly powerful and inexpensive computing technologies, new algorithms and models, and huge amounts of data on almost any subject have led to major advances in predictive analytics over the past two decades.
“Worldwide revenue for ‘big data’ and business analytics solutions is forecasted to reach $274.3 billion by 2022,” wrote economists Erik Brynjolfsson, Wang Jin, and Kristina McElheran in The Power of Prediction, a research article published earlier this year. In principle, such widespread use of predictive analytics should have a positive impact of the performance of firms. “However, these investments have yet to yield productivity gains in the aggregate,” said the authors. “At the firm level, managers struggle to close the gap between the promise of predictive analytics and its performance. These concerns have been difficult to tackle empirically due to the rate of technological change and, ironically, a dearth of data.”
To address these concerns, the authors launched a research study in collaboration with the US Census Bureau to collect information on the use of predictive analytics in a representative sample of the US manufacturing industry, - an industry that’s historically been an early adopter of leading-edge technologies.
Every 5 years, the Census Bureau conducts a Management and Organizational Practices Survey (MOPS) to better understand the management and organizational practices at US manufacturing plants and their impact on productivity growth. As part of the study, a new set of questions were added to the 2015 MOPS about the use of predictive analytics and related workplace characteristics in the manufacturing industry, - for example: “How frequently does this establishment typically rely on predictive analytics (statistical models that provide forecasts in areas such as demand, production, or human resources)?”
Since a response to the MOPS survey is required by law, the response rate was 70.9%, yielding data on over 30,000 manufacturing establishments. Overall, the data showed that “adoption of predictive analytics has become widespread in the American manufacturing sector. More than 70 percent of our representative sample adopted some level of predictive analytics as early as 2010, with pervasive penetration across geographies and industries, as well as plant size and age distributions.”
Most firms in the study relied on widely used, pre-AI analytics methods. While most research and business articles on prediction technologies are currently focused on AI methods like machine learning, their use was limited to a relatively small number of leading-edge manufacturing firms in the 2010-2015 period covered by the study. The use of advanced AI methods is expected to increase in the coming years.
Based on a thorough analysis of the data, the study was able to empirically estimate the productivity impact of predictive analytics. The analysis revealed two major findings:
Adoption of predictive analytics is associated with statistically and economically significant productivity gains. The manufacturing plants reporting use of predictive analytics enjoyed 1% to 3% productivity gains on average, which represents roughly $464,000 to $918,000 in increased sales. The evidence indicates a causal relationship: use of predictive analytics precedes performance gains, but not vice-versa; and increased use of predictive analytics leads to higher productivity gains.
Four workplace complements explain why some firms reap large gains from predictive analytics while others see little or no benefit. The study found that productivity gains from predictive analytics have a strong dependency on four key workforce complements: high IT capital investments; a significant share of educated employees; high managerial and organizational capacity; and high-volume, flow-efficient manufacturing processes, that is, processes whose instrumentation and sensors generate significant amounts of production data.
“Our goal is not only to assess the causal impacts of predictive analytics use, but also to make progress on a practical roadmap that managers can follow to better leverage these new tools,” explain the authors. “Awareness of the organizational constraints can help firms allocate scarce analytics resources, targeting areas that are most likely to yield timely returns or funding coordinated, complementary investments with better-understood timelines.”
These results are consistent with those of earlier studies, which similarly found that investments in leading-edge technologies will not lead to significantly higher productivity and growth unless accompanied by complementary factors. Let me conclude by discussing a few of these studies.
Wired for Innovation: How Information Technology is Reshaping the Economy, a 2009 book co-authored by Brynjolfsson, noted that “The companies with the highest returns on their technology investments did more than just buy technology; they invested in organizational capital to become digital organizations. Productivity studies at both the firm level and the establishment (or plant) level during the period 1995-2008 reveal that the firms that saw high returns on their technology investments were the same firms that adopted certain productivity-enhancing business practices.”
More recently, The Productivity J-Curve, a 2020 research paper co-authored by Brynjolfsson discussed the systematic need for complementary investments to reap the benefits of transformative technologies like AI. The paper identified two phases, investment and harvesting, in the life cycle of historically transformative technologies. Such technologies require massive complementary investments for their full benefits to be realized, including new products, processes and business models, and the re-skilling of the workforce. Moreover, the more transformative the technology, the longer it takes to reach the harvesting phase, when it will be widely embraced by companies and industries. Translating technological advances into productivity gains requires major transformations in the strategy, organization and culture of institutions - and these take considerable time.
Finally, over the past few years McKinsey has been conducting surveys on the current state of AI. Their 2017 survey of over 3,000 AI-aware executives found that only 20% of respondents had adopted AI in production in a core part of their business. A common theme throughout the report was that the same players who were leaders in the earlier waves of digitization and analytics were now leading the AI wave. “A successful program requires firms to address many elements of a digital and analytics transformation: identify the business case, set up the right data ecosystem, build or buy appropriate AI tools, and adapt workflow processes, capabilities, and culture.”
McKinsey’s most recent survey, The State of AI in 2020, found that a small share of high performing companies were achieving outsize business results from their AI investments. Three specific value-capturing practices separated these top AI performers from the rest: they have a clearly defined AI strategy which is aligned with their overall business strategy; they have better overall leadership, including engaged and knowledgeable champions in the C-suite that are fully committed to the AI strategy; and they invest more of their digital budgets in AI than their counterparts, and plan to increase these investments in the coming years.
Comments