“Are Ideas Being Fished Out?”, 2022-11-24 ():
Declining research productivity in US firms cannot be explained by fishing out.
Knowledge production is increasing in the knowledge stock.
Maximum research productivity has been increasing over time.
Declining productivity tracks the federal R&D shift from Development to Research.
Declining research productivity tracks firm adoption of inferior R&D practices.
This paper examines whether declining research productivity can be explained by fishing out—is the production of new knowledge decreasing in the level of existing knowledge?
We estimate the knowledge production function for US firms and find instead that:
knowledge production is increasing in the knowledge stock. This is reinforced by the observations that maximum research productivity across firms is increasing over time, and that research productivity year effects continue to exhibit decline after modeling contributions from knowledge and research labor.
Given that fishing out appears unable to explain the decline in research productivity, we offer preliminary evidence of contingent factors that might contribute to the decline.
[Keywords: endogenous growth, research productivity, fishing out]
… In summary, our results failed to find support for fishing out. Rather, we found positive spillovers of the knowledge stock in the production of new knowledge. Thus, we do not need to be concerned that R&D is getting harder, or accordingly that growth from R&D will decline to zero. Rather, these results support Romer’s original form of the knowledge production function, as well as its expectation that R&D investment can generate growth in perpetuity.
However, and this is critically important, both the government and firms need to recognize that US research productivity has declined dramatically. It had declined 70% at the time of Jones’s first documentation (1995). It has declined again by the same amount in the period since then, as Figure 5 shows. Therefore, to avoid expending increasing amounts of R&D to maintain even the current levels of growth, we need to identify and ameliorate contingent factors contributing to the decline. We provide evidence of such factors at the macro and micro levels.
While our analysis pertains to firms, the findings have important implications for public policy. First, the vast majority of R&D is performed by industry. This likely warrants greater attention to firms in federal innovation policy. The main existing policy instrument for firms is the R&D tax credit. Because of its structure, the tax credit rewards firms for increasing research, but not for increasing development. Thus, the tax credit is likely contributing to the problem of excess research associated with the federal shift from ‘D’ to ‘R’.
Even ignoring an issue of excess research, aggregate data suggest that R&D spending has increased on average 0.014% per year since 1978, when expressed as share of GDP. In absolute dollars, the increase is even more pronounced. Thus, it does not appear that firms need incentives to increase R&D investment. Rather, they need incentives to improve R&D productivity. One policy approach to accomplish this is tying the R&D tax credit to improvements in R&D productivity rather than increases in research spending.
The second implication of our firm-level analysis for public policy is that development, which is required to commercialize or diffuse inventions, is almost exclusively performed by industry. Therefore, if industry R&D is unproductive, then research done by universities and labs becomes de facto unproductive. It may continue to appear productive when measured by patents and publications, but if there is no capacity to develop those inventions, they cannot contribute to economic growth. To remedy this, policymakers could move toward restoring prior ratios of federal funding for development relative to research.
View PDF: