How an Economy Shrinks in Space: Concavity-on-Jobs and Upward Consolidation under Demographic Decline

arXiv:2609.09859 · econ.GN, q-fin.EC · Submitted 2026-09-09 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.

Ruth: We're looking at the paper: "How an Economy Shrinks in Space".

Warren: Here's the gist of the paper. When a country's population declines, its aggregate economy contracts on an intensive margin—meaning industrial diversity remains, but every industry shrinks slightly.

Ruth: At the regional level, this contraction manifests extensively: entire industries disappear sequentially.

Warren: The relevant unit is the city, where industries are nested by size according to the hierarchy property of industrial location; each viable only above a minimum population threshold.

Ruth: Necessity industries' thresholds cluster at the low end, making a city's industry count and jobs sharply concave in size (concavity on jobs).

Title and authors: Ruth: Now we're diving into the core summary of this paper, "How an Economy Shrinks in Space: Concavity-on-Jobs and Upward Consolidation under Demographic Decline". It’s about how shrinking populations change the spatial organization of economic activity within a country > <ref:2609.09859#pg2>.

Warren: Basically, they argue that while national level changes are subtle, the mechanism happening at the city level is extensive margin reorganization—as population contracts, the set of industries each city can locally sustain shifts > <ref:2609.09859#pg3>.

Iris: They focus on Japan as an example because Tokyo is at the apex and they use worker-level panel data to test this migration incentive that varies by origin > <ref:2609.09859#pg1>.

Ruth: The mechanism they propose is that when an industry consolidates away, its jobs leave with it, and a worker who was attached to it can no longer do that work at home, forcing them down to a lower-tier industry in their origin area at a reduced wage > <ref:2609.09859#pg1>.

Warren: Moving to Tokyo then restores the original industry and thus the stepped-down income for that worker > <ref:2609.09859#pg1>.

Iris: The paper also identifies four origin cells based on size and decline rate, like cell I, II, III, and IV > <ref:2609.09859#pg3>. They predict that cell IV carries a concavity on jobs because the further an origin is from the means—smaller and faster-declining—the larger the premium for moving to Tokyo is > <ref:2609.09859#pg3>.

Ruth: That's a big implication for understanding migration patterns; it’s not just about moving toward a big city, but how much better you actually get if you move based on where you live now > <ref:2609.09859#pg1>.

Warren: But there are improvements they suggest too, like using triple-IV over-identification strategies with instruments dated years or decades before the panel to test cell-level interactions that single source variation can't catch > one <ref:2609.09859#pg2>.

The paper's summary: Ruth: So we’ve seen how the paper sets up this framework, and now let's look at what they suggest to make it stronger, because testing these things is always tricky > <ref:2609.09859#pg1>.

Warren: They propose using a triple-IV over-identification strategy with instruments dated years to decades before the panel, like demographic Bartik constructed from prefecture level birth records and national age specific survival rates > one <ref:2609.09859#pg2>.

Iris: That approach identifies cell level interactions where single source variation just can't work, and they test those predictions using a Hansen J over-identification test > one <ref:2609.09859#pg2>.

Ruth: And another way to improve the analysis is for AI systems to be designed to predict which margin of contraction—extensive, like industry exits and entries, or intensive, like the contraction of surviving industries—is actually driven by demographic decline > two.

Warren: The paper finds that only the exits respond to demographic decline because that’s where the real consolidation happens > two. That establishes a clear link between industry consolidation and migration patterns.

Iris: They also suggest using AI models to predict the sign and magnitude of the wage premium based on an origin's size and its rate of decline by analyzing cell level interactions in a four way piecewise specification > three.

Ruth: That should help distinguish between large origins where the premium rises with size, versus small fast declining origins where it grows toward that small fast declining corner > three.

The paper's improvements: Warren: So to wrap up, the paper "How an Economy Shrinks in Space: Concavity-on-Jobs and Upward Consolidation under Demographic Decline" shows that contraction isn't just a national average; it’s about how industries are organized spatially > <ref:2609.09859#pg1>.

Ruth: The main finding is that the migration premium for moving to Tokyo varies depending on whether you're in a small or large origin, and specifically, it gets larger for the smaller and faster declining origins like cell IV > three.

Iris: They confirm that these four cells carry the framework’s predicted signs, especially showing a significant premium response to deep decline in cell IV > three.

Warren: The empirical strategy they use measures this premium as a differential across origins using worker level wage growth in real housing inclusive terms > <ref:2609.09859#pg1>.

Ruth: The overall implication is that demographic decline causes industries and jobs to consolidate upward through the hierarchy, and workers move up to the apex city for recovery > <ref:2609.09859#pg1>.

Iris: They also mention that on the small side, they predict the premium grows as origin size falls because smaller origins cross more ubiquitous industry thresholds simultaneously > one <ref:2609.09859#pg2>.

Warren: And they suggest policy could shift from just subsidies to deciding which locations to keep above their thresholds and which ones to consolidate into them based on predictable losses > <ref:2609.09859#pg3>.

Ruth: It's a complex picture showing how spatial organization matters when we look at long term demographic changes in an economy, and it’s a lot for us to digest > <ref:2609.09859#pg1>.

Conclusion: Ruth: So we’ve looked at how population decline causes industries to consolidate upward through the hierarchy, and now we’re wrapping up this paper, "How an Economy Shrinks in Space: Concavity-on-Jobs and Upward Consolidation under Demographic Decline".

Warren: It really highlights how localized economic changes happen on a scale that's hard to see from just looking at national numbers.

Iris: I think the identification is pretty solid because they’re using that triple IV over identification strategy, which gets them past some of those tricky single source variation issues > one.

Ruth: Exactly, and their finding about cell IV—the small and fast declining origins—having the largest premium for moving to Tokyo makes it a really tangible incentive for migration patterns.

Warren: That means if you're in one of those smaller, faster shrinking places, the gain from moving is much bigger than somewhere else > <ref:2609.09859#pg3>.

Iris: And they also found that on the small side, the premium actually grows as origin size falls because those small origins cross more ubiquitous industry thresholds all at once > one.

Ruth: It’s a neat mechanism where demographic decline actually pushes jobs into higher tiers, and workers are forced to move up to maintain their income level > <ref:2609.09859#pg1>.

Warren: The policy question they pose is whether we should focus on subsidies or on deciding which locations we keep above their thresholds and which ones we consolidate into them based on those predictable losses > <ref:2609.09859#pg3>.

Iris: I just want to mention that the paper notes the limitation that it’s focused on the exits—the industries leaving—and doesn't really look at whether new, different kinds of industries are entering in those same locations > two.

Ruth: That’s a fair point, and it opens up some interesting future work for understanding how regional economies actually adapt to these shifts > <ref:2609.09859#pg1>.

Warren: Yeah, I’m curious if we can build an AI model that can predict which margin of contraction—extensive or intensive—is driven by demographic decline, maybe by analyzing the response of those margins > two.

Iris: That seems like a good path because mapping out which parts of the economy are consolidating is crucial for any kind of modeling > three.

Ruth: Well, that’s it for this one on spatial organization and migration, but next time we’ll be looking at how AI affects innovation incentives in R andD with "Bridging Distant Ideas".

Research Institute of Economy, Trade and Industry · Institute of Economic Research, Kyoto University

econ.GN, q-fin.EC

Submitted: 2026-09-09

Updated: 2026-11-16

Comments: 55 pages, 15 figures, 8 tables. The 95-page Online Appendix is included as an ancillary file (anc/online_appendix.pdf)

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 91/100

The gist: When a country's population declines, its aggregate economy contracts on an intensive margin—meaning industrial diversity remains, but every industry shrinks slightly.

Key concepts

Intensive Margin Contraction
This describes how an economy shrinks when the population declines while maintaining industrial diversity. Every existing industry loses a small amount of output or jobs, rather than one industry completely vanishing.
Extensive Margin Contraction
This refers to the broader contraction where entire industries disappear sequentially. At the regional level, this means specific sectors vanish as cities fall below minimum population thresholds.
Concavity-on-Jobs
This concept describes how a city's job count is not linearly related to its size. Small cities have a disproportionately large number of jobs relative to their size because they host many low-threshold necessity industries, making the job count sharply concave.
Upward Consolidation
When an industry leaves a declining city, its jobs move to the next larger city capable of hosting it. This process pushes workers toward higher-tier industries in larger urban centers, which is why moving to an apex city restores lost income.

Terminology

Summary

When a country's population declines, its aggregate economy contracts on an intensive margin—meaning industrial diversity remains, but every industry shrinks slightly. At the regional level, this contraction manifests extensively: entire industries disappear sequentially. p. 0 The relevant unit is the city, where industries are nested by size according to the hierarchy property of industrial location; each viable only above a minimum population threshold. p. 0 Necessity industries' thresholds cluster at the low end, making a city's industry count and jobs sharply concave in size (concavity on jobs). A modest loss pushes small cities below many thresholds simultaneously, whereas a large core sheds specialized industries one by one. p. 0 Lost industries consolidate upward to the next larger city capable of hosting them; for the worker, this means moving to a lower-paid local job. p. 0 To recover income, workers move up to the apex city that hosts the full industry range. p. 0

Studying Japan—with Tokyo at its apex—using worker-level panel data on young workers who carry migration reveals a Tokyo-bound migration incentive that varies by origin, following concavity on jobs. p. 0 The framework posits that this migration wage premium loads on the extensive margin (whether industries are locally present at all), rather than the intensive-margin object of within-industry pooling. p. 5 The mechanism operates through a common principle: when an industry consolidates away, its jobs leave with it, and a worker attached to it can no longer perform that work at home, forcing them down to a lower-tier industry in their origin area at a reduced wage. p. 4 Moving to Tokyo—the apex hosting the full range of industries—restores the original industry and thus the stepped-down income. p. 0

The nature of consolidation and recovery differs sharply between small and large origins. p. 4 At a declining small origin, industries that disappear are ubiquitous necessities, crossing many thresholds at once. p. 4 The smaller and faster-declining the origin, the more these necessities leave, collapsing the worker's local options; consequently, the larger her gain from moving is: the premium is largest at the small-origin corner — the smallest, fastest-declining origins. Conversely, in a large declining core, thresholds are widely spaced; deep decline only pushes out one or two industries. Here, what matters is which industries go—determined by size—and that set of departing industries is specialized and at the highest-order end of the hierarchy where wage gaps are largest. p. 4 Thus, the premium is largest at large-origin corner — one steep step.

The analysis identifies four origin cells based on size and decline rate: Cell I (large times slow), Cell II (large times fast), Cell III (small times slow), and Cell IV (small times fast). The framework predicts that "On the small-origin side, cell IV carries concavity-on-jobs [beta Sg > 0]: the further an origin lies from the means — smaller and faster-declining — the larger the premium. Conversely, On the large side, ME(S) is positive at every decline depth at which it is evaluated. The findings confirm that the four cells carry the framework’s predicted signs," with Cell IV showing a significant premium response to deep decline.

The empirical strategy measures this premium as a differential across origins using worker-level wage growth in real, housing-inclusive terms. p. 5 This differential reveals how the premium changes across origins—its gradient in origin size and decline rate. p. 5 The results show that the premium grows toward the small-origin corner — the smaller and faster-declining origins — where the larger her gain from moving [is] a Tokyo move can carry (many shallow steps aggregated). The framework's content lies in how this premium varies across origins, with equation (6) placing this variation in piecewise interaction coefficients. The four cell estimates are interpreted as quadrant-specific slopes, not the average mover’s premium. p. 7

The dynamic component of the framework activates only as working-age contraction accumulates enough to push origin industry sets across their hosting thresholds at scale. p. 7 The policy question shifts from subsidies to deciding which locations to keep above their thresholds and which to consolidate into them, based on predictable losses. p. 7 A complementary lever involves investments in remote services (e. p. 3g., mobile retail) and digitalization of public services to lower minimum population requirements for local necessity supply, allowing regional cores to retain self-contained bundles of necessities rather than ceding them entirely to Tokyo's catchment area. The analysis confirms that "On the small side it predicts ME(S) < 0 — the premium grows as origin size falls — because smaller origins cross more ubiquitous-industry thresholds simultaneously, so the total wage gain summed across the multiple exiting industries grows." This supports a mechanism where demographic decline consolidates industries and jobs upward through the hierarchy. The framework's content is in how this premium varies across origins.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

1

Robust causal identification for migration effects can be achieved by employing a triple-IV over-identification strategy using instruments dated years to decades before the panel, such as demographic Bartik constructed from prefecture-level birth records and national age-specific survival rates. This approach identifies cell-level interactions where single-source variation cannot, testing the framework's predictions with a Hansen J over-identification test.

2

AI systems can be designed to predict which margin of contraction—extensive (industry exits and entries) or intensive (the contraction of surviving industries)—is driven by demographic decline by analyzing the response of these margins to young-population decline. The paper finds that only the exits respond to demographic decline, establishing a clear link between industry consolidation and migration patterns.

3

AI models can be used to predict the sign and magnitude of the wage premium based on an origin's size and its rate of decline by analyzing cell-level interactions in a four-way piecewise specification. This allows systems to distinguish between large origins where the premium rises with size and small, fast-declining origins where the premium grows toward the small, fast-declining corner.

Sources

Related papers