Two identical trains can be worth wildly different amounts, depending on which cities they join.
China’s early high-speed rail, above all the line joining Beijing and Shanghai, paid off far beyond what its planners modeled. Later lines to secondary cities paid off much less, though the trains were no worse and the workers no less skilled. Where did the difference come from?
Economists file the answer under total factor productivity, their name for “we cannot explain this part”: the gap between the output that labor, capital and materials predicted and the output that actually appeared. TFP usually gets read as technology or efficiency. The rail lines suggest a better reading: TFP is a synergy measure disguised as a technology measure.
What the residual catches
Connecting Beijing to Shanghai did not just move people faster. It combined two economies that had been semi-isolated. The planners modeled travel time saved and jobs created. They missed the higher-order effects: new patterns of urbanization, logistics industries growing up around the line, supply chains joining into networks. The residual was large because the synergies were large. A line between two mid-tier cities has similar inputs but far less to combine, so the residual shrinks.
TFP, in other words, is an error term. A big positive residual means the model missed how well the inputs would work together; a small or negative one means the model was too optimistic.
When the residual goes negative
The same logic tracks China’s economic trajectory. While the infrastructure gap was enormous, almost any connection between isolated places created multiplicative value. As those synergies were harvested, each new unit of investment found less to combine, but capital kept flowing at the same rate. The bridges to nowhere are investments with negative TFP — the inputs go in, but the synergies do not exist, and the output is less than the sum of the parts.
That is why targeting “productivity growth” backfires, a version of Goodhart’s Law: governments can push the inputs they control, but TFP is precisely the part that is not inputs. More concrete in a city that already has enough roads creates maintenance costs, not network effects.
The sharpest objection is that TFP is just a residual, a measure of our ignorance. That is technically correct: a perfect model would leave a residual of zero. But nobody has that model, and the size of the gap still says how much value the system produces beyond what linear planning predicts.
Read that way, TFP is a legibility gap, the distance between what planners can see and what the system actually does. High TFP means value is arriving through channels nobody designed. Low TFP means the models have become accurate because the system has stopped surprising anyone. The useful question was never how to add more inputs but where the unexploited connections are, and a system that no longer surprises its planners may have stopped finding them.