The Weak Yen Is Not a Reshoring Policy

This series has followed Japan’s 2021–22 yen depreciation through one subject: the relationship between Japanese multinational parents and their foreign affiliates. The final question is whether a weaker yen should be expected to bring production back to Japan. The most defensible answer is conditional. A depreciation can raise foreign-affiliate activity, income received by Japanese parents, and selected support work in Japan without relocating the production that generated those gains. To understand why, we need to hold the firm’s network, its scarce capabilities, and the national-account measure being discussed in the same frame.

The complete chain runs through the firm’s network

The starting point is not an empty map on which a currency movement decides where the next factory will be built. Japanese firms entered the 2021–22 episode with foreign affiliates already operating across many markets. Their exposure depended on the geography of those networks and on how affiliates earned and spent money. Two parents could both own an affiliate in East Asia, yet face different revenue-side exposure because the currencies associated with their markets moved differently against the yen. The Taiwan–Korea comparison illustrates the point: affiliates with similar local-market revenue shares experienced approximately 22 versus 11 log points of cumulative yen depreciation against the relevant currencies by the end of 2022.

That inherited network creates the first link in the chain. A parent with more revenue-side exposure receives a larger home-currency gain when affiliate activity is translated into yen. This is not yet evidence that the affiliate produced more. A fixed foreign-currency surplus can be worth more in yen even if output and employment do not change. The purchase side can move differently, which is why the study separates revenue exposure from cost exposure rather than treating every foreign-currency connection as one effect.

The Japanese firm evidence shows that the response went beyond a passive translation in important respects. More-exposed affiliates recorded higher value added, current profit, retained earnings, and payments to Japanese parents. Moving from the 10th to the 90th percentile of revenue exposure was associated with approximately a 19 percent increase in reported current profit and about a 15 percent increase in total payments. Those financial outcomes are yen-denominated and can contain valuation, but affiliate employment also rose. The p90–p10 effects were about 4 percent for total affiliate employment and 4.8 percent for workers dispatched from the parent. More exposed affiliates therefore used more labor, even though the data cannot decompose every yen of profit into translation and new production.

The foreign expansion had a corresponding parent-side pattern. Average wages rose at more exposed Japanese parents. Non-production support employment increased by about 9 percent from the 10th to the 90th percentile of exposure, while other parent employment declined by about 7 percent. Total parent employment remained broadly flat, head-office employment did not respond, and domestic fixed assets declined by about 5 percent. The response was stronger among parents with larger foreign affiliate workforces. The important relationship is not “foreign profits replaced Japanese jobs” in a simple one-for-one sense. It is that foreign activity and the parent capabilities supporting it expanded while the broader domestic employment base did not.

This is where the firm’s internal constraint becomes decisive. Coordination, procurement, sales support, logistics, engineering, and operational management are productive inputs. A Japanese parent can use them for domestic production or for an affiliate abroad. It can hire more support, but the model assumes that the cost of expanding this capability rises as the pool grows. The parent therefore faces a choice: allocate more of an existing capability to the affiliate, expand the capability at a higher marginal cost, or leave domestic operations with less support.

In the model, total support is divided between domestic and affiliate operations: N equals domestic support plus affiliate support. A depreciation increases the yen value of the affiliate’s operating return. The marginal value of support at the affiliate consequently rises relative to its domestic use. The parent shifts support toward the affiliate and may expand total support employment and the support wage. If production labor becomes more productive when paired with support, affiliate output and employment rise while domestic multinational production faces downward pressure.

The chain is therefore not a mysterious failure of the exchange rate to work.