Measuring the Printing Revolution: What the Economic Evidence Shows

The first two lessons in this course showed that print made books far cheaper and reshaped religious and scholarly life. Yet for a long time economists could find no sign of it in the aggregate numbers. The economic historian Gregory Clark argued the macroeconomic impact was "unmeasurably small", because printing was a small sector with modest price effects on the wider economy.

Jeremiah Dittmar, opening his study in the Quarterly Journal of Economics (2011), points to a modern parallel: until the mid-1990s, economists also found no productivity gains from computers.

Key idea: A technology can transform how information moves without showing up in national income for a long time. If its effects are local, uneven or slow, averaging over a whole economy hides them. The way forward is to find a finer unit of comparison, and for printing that unit is the city.

The obvious test is to compare cities that got a press by 1500 with cities that did not, and see which grew faster afterwards. The problem is selection. Printers were entrepreneurs; they chose where to set up. If they picked cities that were already rich, well connected or about to boom, then press cities would grow faster even if printing did nothing.

Dittmar checks the second row directly: cities that adopted printing in the 1400s had no prior growth advantage. That removes the crudest version of selection, but not a subtler one, where printers foresaw future growth. For that he needed a source of variation in adoption that printers did not choose.

The first lesson noted that printing spread in "concentric circles" from Mainz, because the know-how travelled with people trained there. So a city's distance from Mainz predicts whether it got a press early, for reasons unrelated to the city's own prospects. That makes distance a candidate instrumental variable.

The method runs in two stages:

Stage 1 predicts adoption from distance and controls $X_i$ (such as city size, universities, ports and country). Stage 2 uses only the predicted adoption, the part driven by distance, to estimate the effect $\theta$ on growth from 1500 to 1600.

Definition: An instrument must satisfy two conditions. Relevance: it strongly predicts the treatment (distance did predict adoption). Exclusion: it affects the outcome only through the treatment. Exclusion can never be proven from the data, only argued and tested indirectly.

Dittmar's headline finding: between 1500 and 1600, European cities where printing presses were established in the 1400s grew 60% faster than otherwise similar cities. The instrumental-variable estimates support the ordinary comparison rather than overturning it.

He also runs a placebo test for the exclusion condition. If distance from Mainz merely stood in for distance from Europe's economic centre, distance from other major cities should work just as well. It does not:

Source: Dittmar (2011), Table VIII; the outcome is log population growth 1500 to 1600.

Key idea: Only the true origin of the technology produces a significant estimate. Wittenberg is included as a placebo because Protestant ideas spread from there, so this also checks that the result is not simply the Reformation in disguise.

A growth effect needs a channel. Dittmar's answer is localised spillovers in human capital, especially skills that paid in commerce.

• Commercial arithmetics. From the 1480s, European presses produced a stream of "commercial arithmetics", the first printed mathematics textbooks, teaching merchants to calculate interest, profit shares and exchange rates.
• Bookkeeping and business education. Printed manuals spread bookkeeping, accounting and business practice to a much wider audience than private tutoring could.
• Local benefit. Because books and the craft travelled badly (lesson one: a 360 km journey could add 20% to a book's price), the gains accrued mostly where the press was.

The effect was concentrated in ports. Dittmar finds the growth advantage was largely driven by ports that had presses, even though ports were less likely to get one: in his balanced sample, presses were established by 1500 in 43% of non-port cities but only 36% of sea ports.

In practice: This is the pattern economists expect from a general-purpose technology. The payoff comes where it combines with an activity that uses information intensively, here long-distance trade.

The second lesson presented the Reformation as print's strongest qualitative case. Jared Rubin tested it statistically in "Printing and Protestants" (Review of Economics and Statistics, 2014), with the same instrument.

His finding: cities with at least one printing press by 1500 were at minimum 29 percentage points more likely to be Protestant by 1600.

The design mirrors Dittmar's:

Timing is what makes the instrument plausible here. The presses that count were established by 1500, seventeen years before Luther's theses, so printers could not have been choosing cities for their future Protestantism.

Gotcha: "At least 29 percentage points" is a lower bound on a probability, not a statement that 29% of press cities became Protestant. It compares otherwise similar cities with and without an early press.

Having a press was one thing; how many competed was another. In later work, Dittmar and Skipper Seabold built city-level data on printing output and competition from 1454 to 1600 ("New Media and Competition", LSE Centre for Economic Performance Discussion Paper 1600). Summarising it for the LSE Impact Blog, the authors report:

• Entry lowered prices. When an additional printing firm entered a city market, book prices there fell by 25%.
• A long decline. The raw price of books fell by 2.4% a year for over a hundred years after Gutenberg; adjusting for content and physical features such as format, illustration and colour, the fall was 1.7% a year.
• Competition shaped content. More competitive cities printed more business-education content and spread more Protestant material.

To separate competition from everything else about a city, they use printer deaths: when a printer died, a city's market became less competitive for reasons that had nothing to do with local demand.

Each arrow is a link that at least one of the studies in this lesson tries to measure.

These studies are careful, and they are still estimates. Reading them well means knowing the weak points.

Key idea: The combination is stronger than any single study. Price series, bibliographic counts, city growth, religious adoption and printer competition point the same way, using different data and different assumptions. That convergence, more than any one coefficient, is the case that the printing press changed Europe's economy as well as its ideas.