The Decline of U.S. Business Dynamism

While the young-firm economy thinned out, the top of the distribution thickened. Market concentration rose, with the share of industry sales captured by the four largest firms climbing through the 1980s and again after 2000. Average markups, a common proxy for market power, rose from roughly 1.2 to about 1.6 over the same window, and the corporate profit share of GDP rose alongside them.

Labor's share of output moved the other way, drifting down from the early 1980s and falling sharply after 2000. Akcigit and Ates stress that these two trends are linked rather than coincidental. Across U.S. sectors, the industries where concentration rose the most are the industries where labor's share fell the most.

The chart shows that association sector by sector. In finance, services, utilities and transport, manufacturing, retail, and wholesale alike, a rise in concentration lines up with a fall in labor's share. The association is strongest in finance and weakest in wholesale, yet it points the same way in every sector. Read together, the picture is of a small number of firms capturing a larger slice of revenue while paying out a smaller slice as wages.

To test that story, Akcigit and Ates build a structural model of an economy made of many product markets, each contested by a leader and a follower who race through R&D for market leadership. Markups are endogenous, so a firm's price and profit depend on how far ahead of its rival it sits. They calibrate the model to the U.S. economy of around 1980, then subject it to four candidate shocks over thirty years: lower corporate taxes, higher R&D subsidies, higher entry costs, and slower knowledge diffusion.

Because the calibration is disciplined by pre-1980 targets, the model's response is close to an out-of-sample test. The chart compares the actual 1980 to 2010 change in five margins the model was never asked to match against the change the model generates on its own. Labor share, concentration, job reallocation, firm-growth dispersion, and the productivity gap all move in the right direction and roughly the right size.

The match is not exact. The model understates how far sales concentration and the productivity gap widened, partly because concentration in this stylized economy can take only a couple of values. But that a single calibrated transition reproduces the sign of every trend at once is the payoff of studying ten facts jointly rather than one at a time.

The model lets the authors turn each shock off one at a time and measure how much of the total change it accounts for. The verdict is decisive. A calibrated 60 percent decline in the rate of knowledge diffusion accounts for more than 70 percent of the variation in almost every trend, while the other three forces rarely explain more than a tenth.

The chart reports the diffusion channel's contribution to each of the nine model-generated trends. It explains 96 percent of the rise in concentration, 84 percent of the rise in markups, 79 percent of the fall in labor share, and 71 percent of the decline in young firms' employment share. For firm-growth dispersion the channel alone overshoots the observed change, which is why its bar passes 100 percent. Higher entry costs bear mostly on entry itself, and even there account for only about 18 percent of its decline.

The mechanism is a chain. When knowledge stops flowing from leaders to followers, leaders keep their edge, concentration and markups rise, and the profit share climbs while labor's share falls. Followers and entrants, unable to catch up, stop trying, so entry and reallocation dry up. The decline in U.S. dynamism, in this account, is not a mystery of animal spirits but a measurable consequence of slower knowledge diffusion and higher barriers to imitation, and both are things policy can act on.