This figure plots the total effect of a nondefense R&D appropriations shock on TFP (blue line) from our previous paper, rescaled to match the peak projected effect of the CHIPS Act on public R&D capital (above), along with 95% confidence intervals (gray bands):

This next figure plots the projected effect of full CHIPS Act funding on cumulative business-sector total factor productivity (TFP) growth from this modeling exercise:

This figure plots the projected effect of full CHIPS Act funding on the cumulative growth of real public R&D capital from this modeling exercise:

Here's the figure from our recent Dallas Fed working paper w/ the historical contributions from government R&D spending on U.S. productivity growth that I referenced in this interview

Quite the striking figure from Stock and Watson at
@brookings.edu #bpea S2025: The U.S. recovery from Covid was the first trend-reverting recession recovery on record

The baseline results (phi = 0.11) are not affected by dropping any year/specific policy events, but when the sample omits 5 years around the late 1950s/early 1960s, there are a few instances when significance becomes more marginal or disappears under weak-IV-robust inference

With the slightly lower production function elasticity, the growth contribution of government nondefense R&D capital to post-war TFP growth now rounds down to at least 20% rather than roughly 25%

The higher end of the old range (300%) came from the old specification excluding NASA appropriations just during the space race; we now exclude all NASA appropriations for that specification, which yields estimates that are much closer to the baseline

Using the raw narrative instruments, the quantitative results have changed slightly but remain very similar to our previous draft. The baseline elasticity estimate is now 0.11 (previously 0.12). The range of returns is now 140 to 210% (previously 150 to 300%)

Karel Mertens and I have an updated version of our paper “The Returns to Government R&D: Evidence from U.S. Appropriations Shocks,” short 🧵on what's changed:

This second figures shows our preferred measure of productivity growth (black line), along with our estimates of the contribution of government R&D capital (blue) and public infrastructure capital (red) to that productivity growth: