Alexander Field’s new book gives us new estimates of US private sector productivity in 1941–48 using advanced methods similar to those used by the official statistical agencies. The book focuses especially on the manufacturing sector, which was centrally important to the nation’s effort in World War II. Based on data about investment by the firms in that sector, he has constructed estimates of capital stocks and can construct total factor productivity estimates that take both labor and capital into account.
The overall finding is that productivity growth was slow during this period, although the US did famously produce staggering numbers of aircraft, ships, and tanks. The finding makes sense as Field explains it: manufacturers had to switch massively from producing consumer goods to new and uncommon military equipment, and then in the same decade back to consumer goods. This cost them greatly in resources, time, labor, and equipment.
Furthermore, some of these manufacturers depended on foreign sources for natural rubber, notably for tires, and with the Japanese conquest of Southeast Asia, 90 percent of the usual supply of natural rubber was cut off. This led to a desperate effort to make synthetic rubber. This additional shock to the manufacturers reduced their productivity further. This story is a memorable example of induced innovation.
The book makes estimates of labor hours in manufacturing. Because of the extraordinary pressure to produce, huge numbers of inexperienced people were hired, and they often worked long hours, thus producing more output but lowering output per hour worked.
This is a big subject, and the productivity estimates are new. The estimates make sense. The story is convincing enough to convey why readers should expect to find low or declining productivity growth in wartime generally.
Thousands of data items are embedded in aggregate statistics like these estimates of capital input and labor input. One challenge helpfully discussed here is that standard price indexes do not easily address a situation where there were price controls and limits on production for consumer goods during the war, and where production of goods for the military was so regulated.
The author makes important adjustments to include measures of government capital available to the manufacturing sector. In postwar statistics, government-owned structures and equipment are left out of manufacturing sector statistics because manufacturing is almost always a private-sector activity in the US, but during the war, government capital contributed significantly to producing output. For this reason among others, the time series of capital and productivity estimates here do not directly extend the standard postwar series backward in time. There are some differences in measurement concepts, but the adjustments are appropriate to the war period and data available.
For non-manufacturing sectors, some calculations in the book assume that capital made up a steady 30 percent of the costs of production. Since there were volatile flows of capital and labor across sectors during the war, the assumption appears to be somewhat extreme. It could perhaps be tested in future work.
One subject of special interest to this reader was the author’s use of patent data. The book shows that during the war, there was a decline in the number of patents filed per year in the militarily relevant technical fields relative to the prewar pace. The text suggests that this might be related to weak productivity improvement. However, there is another likely source for the patent finding, which is known from other data but has not been fully studied in economic history. Patents are public documents, and patents for new equipment relevant to the military can be sensitive. During wartime, governments, inventors, and other institutions in the warring countries often don’t want to make militarily relevant inventions visible to the opposing militaries. Special laws in those periods in the warring countries gave the governments the power to keep certain patents secret. Naturally, there is a decline in the number of patent applications from inventors in the opposing countries. At a variety of levels, therefore, patenting is reduced or suppressed, even as R&D activities charge ahead at full speed. My work on aeronautical patents found a strong decline in patents during World War I despite the clear continuation of technological progress. There is a regularity: Patenting is depressed in wartime, at least for militarily relevant technology, for reasons unrelated to short-run productivity.
Three key reasons for the low manufacturing productivity growth in this period are fully and memorably discussed in this book. First, there was a huge shift from production of consumer goods/services to war goods, and then back. Second, there were sharp shocks to goods supply chains, notably the loss of access to natural rubber supplies. Third, inexperienced people worked many of the additional hours.
An underlying fourth cause is left implied. It has to do with the scorekeeping for productivity. The measures of output for productivity incorporate prices and volumes of goods consumers buy. Thus they reflect consumer preferences. They are constructed to match a social welfare function in which consumer preferences matter. If society decides to pursue another objective – mobilization for all-out war – that is a different social objective. When we use GDP as an output measure to cover the mobilization period, productivity concepts are measured in the same way as before; it’s the same “measuring stick.” However, this output measure is not aligned with objectives of the social planners; for better or worse, the planners have decided to value products based on their relevance to mobilization.
In the context of society massively building tanks, ships, airplanes, and an atomic bomb, it can be expected that productivity estimates using the earlier consumer-focused measuring stick would be reduced, even if there were no transition costs. Other changes in social choice might be analogous, such as shutdowns in response to a pandemic. These involve changes in what society has decided to do, away from maximizing the standard measures of output, and could call for different output and productivity concepts and measures.
