Reviewing a Textbook Sidebar on 1930s Colonial Industrialisation
A textbook publisher's review panel is debating a sidebar for a high-school volume. The draft reads: 'During the 1930s, colonial rule brought rapid industrial growth, and living standards rose as factory employment expanded.' The panel has a data appendix: the number of factories in Korea rose from roughly 2,000 in 1919 to more than 6,000 by the late 1930s; Korean-owned firms held a small minority of paid-in capital throughout the period; Korean workers' wages in the same industries averaged roughly half to two-thirds of Japanese workers' wages; rice shipments to Japan rose sharply in the 1920s and 1930s while per-capita rice consumption in Korea fell; and the fastest-growing sectors were chemicals and metals in the northern provinces, financed by zaibatsu capital and tied to military demand after 1931 and again after 1937. One panel member says the rise in factory numbers proves development; another replies that all colonial statistics are unreliable and should be dropped.
1. Rewrite the sidebar into two sentences that are defensible on the appendix data, and explain what you changed and why.
Read the worked answer
Suggested replacement: 'From the early 1930s, colonial policy and Japanese military demand produced rapid growth in mining and heavy industry in the northern provinces, financed largely by zaibatsu capital, while Korean-owned firms remained a small share of invested capital. Output and factory employment expanded, but Korean wages stayed well below those of Japanese workers in the same industries, and per-capita rice consumption in Korea fell as shipments to Japan rose.'
What changed, and the reasoning:
- 'Colonial rule brought' became a specified mechanism. Growth had identifiable drivers - the Manchurian crisis of 1931, the wider war from 1937, subsidised zaibatsu investment, and tariff integration with the metropole. Naming them prevents the sentence from implying that colonial administration as such causes growth.
- 'Living standards rose' was replaced with distributional facts. Higher output and more factory jobs are compatible with falling food consumption; the appendix shows exactly that combination. The sidebar can state both without resolving them into a verdict.
- Ownership and wages were added, because a factory count without capital share and relative pay invites the reader to assume that growth accrued locally.
The two sentences are longer, but each clause is checkable against the appendix.
2. Evaluate the two panel members' arguments: the factory-count inference and the blanket rejection of colonial statistics.
Read the worked answer
The factory-count argument fails on measurement. A count of establishments is sensitive to how a factory is defined, to the registration threshold, and to the growth of small workshops that may be undercounted. It says nothing about capital intensity, output value, ownership, or who received the returns. The appendix itself shows that Korean-owned capital remained a small minority, so a rising count is compatible with growth that was largely externally owned. The inference from 'more factories' to 'development' also requires a distributional judgement that the count cannot supply.
The blanket-rejection argument fails on method. Colonial statistics were compiled by an administration with interests, and they have known biases - under-registration of Korean small enterprise, category choices that flatter mining and heavy industry, wage series built on larger firms. But bias is a direction, not a prohibition. Series can still constrain claims: if wages in the recorded firms were half to two-thirds of the comparator, the true gap is unlikely to have been smaller. Discarding the data leaves the field to whoever is willing to use it carelessly.
The correct posture is source criticism with explicit direction of bias, and triangulation across independent series - police records, missionary accounts, company reports, and later survey data.
3. Which single additional dataset would most change the sidebar, and why does it have more discriminating power than the others?
Read the worked answer
The strongest candidate is a real-wage series for Korean industrial and agricultural workers, deflated by a cost-of-living index that includes food, because it directly tests the claim the sidebar originally made.
The reasoning about discriminating power:
- It targets the actual assertion. 'Living standards rose' is a statement about real consumption per person. Nominal wages, factory counts, and output value all bear on it indirectly; a deflated wage series bears on it directly.
- It interacts with the rice data. The appendix already shows falling per-capita rice consumption alongside rising shipments to Japan. If nominal wages rose while food prices rose faster, the two facts combine into a coherent picture of immiserising specialisation; if real wages rose for industrial workers while rural consumption fell, the story becomes one of sectoral divergence rather than uniform decline.
- It is attainable and bounded. Wage and price series exist for the 1930s, with known coverage limits. That makes it more useful than a wish-list item such as a full income distribution, which nobody can reconstruct.
The runner-up would be land tenure and tenancy data, which would tell the panel what was happening to the rural majority who never entered a factory. A review panel should note both, and choose the wage series because it addresses the sentence the publisher actually printed.