Unless we keep raising research inputs, economic growth will continue to slow in advanced nations.

There’s been ongoing dialogue in the past few years about whether tech innovations have plateaued. While some say that we’re still in a golden age of innovation, a 2016 headline in The Wall Street Journal declared, “The Economy’s Hidden Problem: We’re Out of Big Ideas.” The article cited slower gains in science, medicine, and technology that hold back economic growth and posited that U.S. businesses may be too risk averse.

At the MIT symposium in November 2017, director of the MIT Initiative on the Digital Economy Erik Brynjolfsson answered the question, “Are we running out of inventions?” with a definitive “no.” He spoke about improvements in machine learning, from neural networks to voice recognition, and noted that there has been “a flood of research” in artificial intelligence in recent years that will likely lead to new breakthroughs.

What’s going on? Are we so jaded by technological breakthroughs that incremental innovation is discounted? Or are we facing a more serious problem?

Our latest research shows encouraging signs that new concepts have not been depleted. However, unique, original, and untapped ideas are getting more expensive to find — and that’s a problem.

A Research Productivity Gap

A range of evidence from various industries, products, and companies shows that U.S. research efforts are rising substantially while research productivity is sharply declining. Optimists hope for a fourth industrial revolution that will raise the bar again, while pessimists lament that most potential productivity growth has already occurred.

We believe that these differing views revolve around resource allocation. To maintain a given rate of economic growth, resources devoted to research must increase over time — but in many areas that’s not happening fast enough. Aggregate evidence as well as measures of research and development (R&D) productivity in specific industries — especially computers, agriculture, and medicine — illustrate this.

Take Moore’s Law, for example, which Wikipedia defines as “the observation that the number of transistors in a dense integrated circuit doubles approximately every two years.” Although this translates into an impressive increase in technical progress of 35% per year, our research finds that the effective number of semiconductor researchers has increased by around 18 times since 1971, which implies that research productivity on computer chips has in fact declined at an average annual rate of 6.8%. This is significant, not only because of the high expectations that technology has set for fast-paced, awe-inspiring innovation and growth but also because the two are extrinsically linked: Economic growth arises from people creating ideas and innovation. The greater the research investment, the greater the rate of growth.

The High Cost of Mining for Ideas

Geologists have been forecasting “peak oil” for decades, only to be surprised by deep-sea discoveries and shale oil. Likewise, we see a continuing stream of innovations — but, just as newer oil sources are increasingly costly to extract, coming up with new ideas is getting more expensive. The issue is not just how many ideas for productivity growth are left but what it would cost to get them out of the ground — and, crucially, how much we’re prepared to spend to do it.

Our recent study shows that these costs have increased sharply over time and that research productivity, or the innovation bang for the R&D buck, has declined. In an accounting sense, low productivity growth in the economy is a direct consequence of research efforts failing to increase quickly enough to offset declining research productivity. If we want to restore economic growth, we need to pay for it.

Research productivity in the U.S. has fallen 5.3% per year on average, according to our estimates. In order to offset the increased difficulty of finding new ideas, the level of research looking for new ideas must be doubled in the United States every 13 years, just to sustain constant GDP growth per person — and that’s a tall order.

It may be that part of the drop in research productivity is that more R&D is devoted to defending market share rather than expanding the market. One example is “me too” drugs that are only slightly better than existing treatments but can tip the market to a new pharmaceutical company. Another issue may be that basic research is a smaller fraction of overall R&D in part because of government cutbacks.

We find similar trends whether we look at crop yields for corn and soybeans or at medical innovations that reduce mortality from heart disease and breast cancer. Although there have been technological improvements, they require the devotion of ever-growing amounts of resources to the research process to maintain steady rates of improvement. We also find a similar pattern using company-level data. While there was substantial heterogeneity across companies, we found that research productivity declined for more than four-fifths of our sample.

To our minds, all of this points to the conclusion that ideas are becoming more expensive to find. Unless research inputs continue to rise — at the university, government, and individual business levels — economic growth will continue to slow in advanced nations such as the United States. Only top-level commitment and resources will stem the tide.