In his State of the Union remarks in February, President Joe Biden celebrated the federal government’s recent efforts to boost the semiconductor industry in the United States. The development of semiconductor chips, now used in phones, cars, TVs, and household appliances, is “a story of American genius and possibilities,” he said. The CHIPS and Science Act, signed into law last August, will pour more than $50 billion into the industry, including funding for R&D.
The initiative was motivated partly by the need to control more of the supply chain. But a new study co-authored by Song Ma, an associate professor of finance at Yale SOM, suggests that the move also could also support future innovation in a variety of fields.
Ma and Ernest Liu, now at Princeton University, wanted to determine the best way for countries to allocate R&D funds among scientific fields. The key consideration in their approach is that investing in one area could have ripple effects years or decades later on innovations in other fields. “They are intertwined with each other,” Ma says. For instance, an advance in chemistry might not initially seem useful in other industries. But it could later enable new developments in biotechnology, which in turn could jump-start pharmaceutical improvements.
“You do not necessarily see immediate payoff,” he says. “But down the road, they could be really, really important.”
Ma and Liu created a model formalizing the role of “innovation network,” accounting for both short-term and long-term benefits of R&D, and analyzed data on millions of patents and R&D expense records from around the world. Their findings suggest that the United States could reap substantial benefits from adjusting its distribution of R&D funds—including more investment in fields such as alternative energy, biochemistry, and yes, semiconductors.
It’s useful to think not only about increasing money for R&D, but to “put the right money in the right place,” Ma says.
Ma and Liu started by creating a model to represent the innovation network in mathematical terms. Each advance could yield immediate benefits for consumers, and it could trigger a cascade of future inventions in other sectors. “It keeps exploding,” Ma says.
The model also captured knowledge flow across borders. Countries could piggyback off research in other countries to kickstart their own technological advances.
Next, the researchers obtained two sets of real-world data. First, they gathered information about consumer preferences in 43 countries—essentially, how much people bought products from each sector. This data helped the team estimate the benefit to consumers when innovations in those sectors appeared.
Second, the team retrieved details from Google Patents about more than 36 million patents around the world. For each invention, the researchers checked whether it had been cited by other patents—an indication that the later advance had built upon the earlier one.
The innovation network analysis suggested that medical science, computing, semiconductors, and electric communication technology were among the most central fields in the network. In other words, they tended to have the most impact on other sectors’ innovations. And among countries, the United States was the most central, followed by Japan; other countries frequently cited their patents.
Ma and Liu then calculated the optimal R&D allocation for each country. The goal was to maximize welfare, meaning the value that consumers received from the goods they purchase, combining both consumption in the short run and long run.
How well did these ideal scenarios line up with reality? To figure out each country’s actual allocation, the researchers examined data on R&D expenditures across different sectors in different countries. In an analysis of 10 countries with the most patents from 2010 to 2014, they found that Japan allocated its funds the most efficiently, followed by Germany, South Korea, the U.S., and China. In Russia, France, the UK, Canada, and the Netherlands, allocations were farther off the mark.
Next, the researchers wanted to know how much consumers would benefit if countries improved their R&D allocation strategies. They estimated that if the U.S. had allocated funds as efficiently as Japan did in 2010, the value that American consumers received would increase by 20%, “which is a huge potential improvement,” Ma says. In other words, products would have 20% higher quality, on average, or consumers could buy 20% more products with the same amount of money. The United States could improve its allocation by investing less in phone and wireless communication and more in semiconductors, biochemistry, and green technology such as alternative energy. That doesn’t mean that the former are not important, Ma notes, but that the latter are particularly central to the long-term innovation and growth of the economy.
Finally, Ma and Liu tried to figure out why some countries were better at allocating funds than others. For the industrial research they were analyzing in this study, government agencies weren’t necessarily the ones deciding how money should be distributed; those decisions were often happening at firms, as each company selected projects to pursue.
The team found that countries with efficient R&D allocation, such as Japan, also tended to have more of their research activity concentrated in big firms that operated across multiple sectors. The stronger the presence of these “innovation hub” companies, the less a country tended to misallocate money. The large firms might be more motivated to invest in fundamental research because those advances could later benefit other departments within the same company.
In other words, some countries might allocate R&D funds better because they “have more of those firms that can take the long view,” he says. If R&D funds aren’t being well-allocated at firms, the government might need to step in to boost investment in critical areas—for instance, by offering subsidies or tax breaks.
Using a less-than-ideal allocation strategy doesn’t mean a country isn’t innovative. The United States produces a lot of high-impact patents “even though the country doesn’t allocate particularly efficiently,” Ma says. The situation is analogous to an extremely wealthy family that hasn’t optimized their budget. But if the U.S. did distribute its R&D money better, the country could realize “a lot of gains,” he says.