Even More Pronatalist Research Showing Local Governments Can Boost Birth Rates (Study: 1,741 Cities in Japan!)
Tokyo and Osaka show identical urban fertility penalties, but Nagoya escaped through aggressive municipal child welfare spending
Japan’s fertility rate sits at 1.38, well below the 2.07 replacement level. Fewer than 800,000 babies were born in 2022, the lowest figure since records began in 1899. By 2070, the elderly dependency ratio will hit 38.7%, making pension systems unworkable and regional economies unviable.
Where you live matters more than national policy. New research on all 1,741 Japanese municipalities calculated local “desired birth rates” (希望出生率, kibō shusseritsu) and found something striking: only 15% can realistically hit the national target of 1.8 under optimistic assumptions. Under conservative assumptions, none can. The variation is enormous, from 0.81 to 1.97, a 143% spread between lowest and highest. Anyone who been noticing a pattern with me, is that I care *a lot* about varation and the causes behind variation (it’s in the name “Governance Cybernetics*). Local governance choices can affect birthrates along side national policy, in some cases may have a bigger impact.
Akashi city (明石市) west of Kobe hit 1.65 fertility against a national 1.3 through aggressive municipal spending. Bloomberg found that Nagareyama (流山市), a Tokyo suburb with innovative childcare logistics, reached 1.50 versus the national 1.26. These aren’t are not one offs (or limited to Japan). They result from specific, replicable policy choices that most Japanese localities simply don’t make.
The research, published in the Journal of the City Planning Institute of Japan by Kaoru Kawasaki of NTT East and Eiji Morimoto of Shinshu University, exposes an uncomfortable reality. The national target of 1.8, adopted uniformly by prefectures and municipalities, bears no relationship to structural conditions in 85% of Japanese territory. When local governments chase unachievable goals, they design interventions for demographic situations that don’t exist. The analysis identifies what distinguishes successful municipalities from failing ones. The answers suggest Japanese demographic policy has been looking in largely the wrong places for a decade.
Understanding the Measurement Problem
Japan introduced “desired birth rate” in 2014 to capture something the standard fertility rate misses: the gap between what people say they want and what they achieve. The calculation combines the proportion of young women (15-34) already married with regional survey data on marriage intentions, ideal family size, and remarriage effects. If unmarried women say they intend to marry, and married women want 2.1 children on average, then Japan’s “real” preference sits around 1.8, far above the actual 1.36 total fertility rate (TFR, 合計特殊出生率).
This improves on TFR alone but doesn’t fully decompose the problem. Stephen J. Shaw’s 2025 analysis in Scientific Reports examined 314 million mothers across 33 countries and found that TFR conflates two statistically independent phenomena. The proportion of women becoming mothers (Total Maternal Rate, or TMR) moves separately from average family size among those who do become mothers (Children Per Mother, or CPM). These respond to different forces and need entirely different policy interventions.
Shaw’s information theory is that the TMR alone contains more diagnostic information (3.64 bits) than combined TFR (3.43 bits). Policymakers using TFR as their primary metric are missing nearly half the information needed to design effective interventions. What’s needed is a basket of complementary metrics, not reliance on one or two headline numbers.
The calculation depends on regional survey data (11 broad regions from Hokkaido to Kyushu), not municipality-specific responses. This means hundreds of cities within each region get assigned identical marriage intention figures. The regression therefore captures variation in who’s actually getting married more than variation in aspirations themselves.
The optimal approach combines multiple diagnostic tools. TFR tracks the outcome. Desired birth rate shows intention-behavior gaps. TMR/CPM decomposition identifies which mechanism is failing.
Kawasaki and Morimoto’s municipality-level calculations extend this diagnostic framework geographically. Their desired birth rates vary from 0.81 to 1.97 across Japanese municipalities, patterns that neither national TFR figures nor uniform policy targets capture. Combined with Shaw’s TMR/CPM lens, this creates a comprehensive diagnostic toolkit showing not just that fertility is low, but where it’s low, why it’s low (entry problem vs family size problem), and what residents say they want but aren’t achieving.
Success Stories
The spatial analysis produces stark patterns. Kyushu and rural regions show elevated desired rates, often reaching 1.7-1.9. Tokyo and Osaka metropolitan areas consistently underperform, clustering around 1.0-1.2. Hokkaido and Tohoku show the lowest rates, while Chugoku and Kyushu show the highest, a west-high, east-low pattern that matches existing social capital research showing similar geographic distributions of community trust and civic engagement.
The variation within regions proves more instructive than the regional patterns themselves. Tokunoshima Town, Kagoshima (徳之島町、鹿児島県) registers Japan’s highest actual fertility rate at 2.25, extraordinary in a developed economy. Municipal surveys in the paper indicate most residents report parents, siblings, friends, and neighbors actively providing childcare support. The social infrastructure for child-rearing exists organically, embedded in daily life rather than delivered through formal programs.
Minamidaito Village, Okinawa (南大東村、沖縄県), another remote island barely accessible and with minimal urban amenities, shows similarly elevated desired rates. The remote island pattern proves statistically significant. Municipalities designated for official remote island promotion programs (離島振興対策実施地域) show desired birth rates 0.12 standard deviations higher than otherwise similar locations, even controlling for income, services, and employment. Designated mountainous areas (特定農山村) show no comparable effect. Both categories face depopulation and receive targeted subsidies, but something about island isolation drives higher desire for children in ways mountainous isolation does not.
Horonobu Town, Hokkaido (幌延町) presents a counterintuitive case. Despite Hokkaido’s generally depressed fertility, this municipality of roughly 2,500 shows relatively elevated desired rates. The distinguishing factor: active recruitment of energy and geological research facilities, creating stable professional employment in a remote location. Young scientists with secure positions adjust family formation plans upward compared to peers in uncertain urban labor markets.
Kasuga Town, Fukuoka (粕屋町) pursued the most aggressive fiscal strategy. Allocating approximately 41% of municipal budget to child welfare spending (児童福祉費), an extreme figure nationally, this bedroom community operates six train stations connecting to Hakata, explicitly positioning itself as family-friendly for metropolitan commuters. This level of spending is roughly triple what most comparable municipalities allocate. Desired birth rates substantially exceed the Fukuoka average.
NPR reporting provides additional detail. Akashi city’s former mayor Fusaho Izumi doubled child welfare spending during his 2011-2022 tenure, funding the increase by cutting public works rather than raising taxes. The fiscal reallocation offended bureaucrats and construction interests but attracted young families at rates that increased tax revenue enough to sustain the programs, a self-reinforcing cycle. Akashi offers free medical care to age 18, free school lunches to 15, free nursery school and kindergarten for families with two or more children, and free diapers for babies under age 1, delivered by midwives who provide professional advice.
Bloomberg found that Nagareyama’s Mayor Yoshiharu Izaki, who took office in 2003, recognized that the town’s baby boomer cohort created an aging crisis the municipality couldn’t financially survive. He implemented station-based childcare pickup and dropoff systems allowing parents to commute to Tokyo while children are transported to daycares, rapid construction eliminating nursery waitlists entirely, and careful cultivation of the town’s identity as “the forest city nearest to Tokyo.” The 30-minute Tsukuba Express (つくばエクスプレス) connection to central Tokyo proved critical. Nagareyama’s data shows 29% of elementary school families have three or more children and only 13% have one child, suggesting the environment enables families to approach their ideal family sizes.
Tokyo’s 23 Wards
Tokyo’s 23 special wards show among the lowest desired rates nationally, averaging around 1.1 despite having the highest incomes in Japan. The wards face binding constraints the research cannot fully measure: housing costs that make family-sized apartments unaffordable, commute patterns that impose severe time poverty, and cultural norms around children in public spaces.
But the comparison with Nagoya is instructive. Nagoya metropolitan area (名古屋都市圏) municipalities show no significant fertility penalty despite similar density and income levels. The difference traces to fiscal priorities. Residual analysis shows Nagoya-area municipalities spend significantly more on social welfare (社会福祉費), elderly welfare (老人福祉費), and child welfare compared to Tokyo or Osaka counterparts, even controlling for tax base. Municipal governments can offset big-city fertility penalties through aggressive family support spending. Tokyo’s wards choose not to, probably because elderly voters who outnumber young families prefer different spending priorities.
What Actually Matters
Child welfare spending dominates everything else. Municipalities that allocate more budget share to child services show dramatically higher desired rates, a 0.47 coefficient that dwarfs all other factors. Kasuga Town pushes this to the extreme at 41% of budget, roughly triple what comparable cities spend. Without municipal income data, though, distinguishing wealth effects from priorities blurs. Richer localities can afford generous programs. Or resource-constrained cities can prioritize differently. The data can’t separate these.
Working close to home shows up strongly too. Places where residents work locally (自市区町村で従業) show desired rates 0.19 standard deviations higher than places where residents commute elsewhere. The statistical effect is larger than anything related to transit access or high-speed rail connections. Tokyo built the Tsukuba Express to connect Nagareyama in 30 minutes, but Nagareyama’s success probably has more to do with Mayor Izaki’s station childcare system than the train itself.
Commute method matters more than commute length. Car commuters show higher desired rates than train commuters (0.30 vs -0.13), not because cars are faster but because they offer flexibility. You can handle an unexpected daycare pickup. You’re not locked into the 5:47 departure. Walking to work also shows loosely positive effects (0.05), though this overlaps with the proximity-to-work finding. If you can walk to work, you’re definitionally working close to home. Walking combines both advantages, proximity and total flexibility, making dual-income family structures operationally feasible.
Kindergarten access (幼稚園) predicts higher desired rates (0.19). So does municipal administrative office accessibility (市区町村機関, 0.18), probably because busy families need to handle tasks without travel. Hospital density (病院) shows modest effects (0.07), though grocery retail (飲食料品小売業) shows none.
Community festivals and events (各種イベントの開催) drive higher desired rates (0.13). General community function (集落機能) shows borderline significance (0.05), but other forms of civic organization show no effects. The mechanism probably runs through social connection and informal childcare support networks. Parents, siblings, friends, neighbors who watch children briefly, provide backup during illness, collectively supervise outdoor play. These networks emerge from repeated neighborhood interaction. Festival culture both reflects and reinforces this.
Agriculture, forestry, fishery employment (第1次産業就業者) correlates with higher desired rates (0.07), though the modest coefficient suggests this represents one factor among many. Remote island designation (離島振興対策) shows positive effects (0.12) even controlling for other factors, suggesting that targeted regional policy can work if designed appropriately.
Living in Tokyo or Osaka metros (東京都市圏・大阪都市圏) lowers desired rates beyond what employment, income, or services explain. Both show identical -0.26 coefficients. Big-city residence itself suppresses desire for children. Nagoya shows no such penalty, suggesting fiscal priorities can offset urban constraints. University presence (高等教育機関) shows negative effects (-0.12), possibly reflecting career timeline delays or age composition effects. Railway commuting drives lower rates, the inflexibility problem noted above.
Budget Priorities
Beyond child welfare spending, the analysis shows how different budget allocations link to desired birth rates. The patterns suggest municipalities face tradeoffs:
Strong positive link:
Child welfare spending: +0.47
Moderate positive link:
Agriculture, forestry, fishery spending (農林水産業費): +0.14
Loose positive link:
Social education spending (社会教育費): +0.04
Moderate negative link:
General social welfare spending, non-child: -0.10
Loose negative links:
Livelihood protection spending (生活保護費): -0.08
Commerce and industry spending (商工費): -0.08
Fire service spending (消防費): -0.08
Labor spending (労働費): -0.07
Urban planning spending (都市計画費): -0.06
Health and physical education spending (保健体育費): -0.06
The negative links for general social welfare and livelihood protection suggest municipalities facing greater poverty struggle to maintain high desired birth rates. The negative links for commerce, urban planning, fire services, and labor spending suggest municipalities forced to allocate substantial resources to economic development, infrastructure, and emergency services face tradeoffs that constrain family support spending.
The evacuation zone designation (帰宅困難区域) for Fukushima-related displacement shows negative effects (-0.04). Nuclear family household share (核家族世帯割合) shows no significant relationship. The widespread assumption that multi-generational households enable higher fertility finds no confirmation in preference data.
For Families Deciding Where to Live
The research translates to concrete considerations for dual-income families weighing location choices.
Proximity to work matters more than transit access. Municipalities where most residents work locally show 0.19 standard deviations higher desired rates than commuter towns. If you must commute, car flexibility matters more than shorter train rides. The ability to handle unexpected pickup or route changes makes dual-income parenting operationally feasible.
Municipal budget priorities predict family-friendliness better than city size. These priorities are measurable through public documents. Localities allocating substantial shares to child welfare, not just elderly care or infrastructure, show dramatically higher desired rates. Kasuga Town’s 41% allocation represents an extreme, but municipalities above 25-30% consistently outperform peers.
Community matters, but in specific ways. Festival culture and local events correlate with higher desired rates. General civic organizations don’t. This reflects informal childcare networks that emerge from repeated neighborhood interaction.
The urban penalty isn’t universal. Tokyo and Osaka show large fertility penalties, but Nagoya doesn’t, because Nagoya-area municipalities spend more aggressively on family support. If considering urban locations, municipal fiscal priorities matter more than density itself.
What the Research Cannot Show
The authors construct two scenarios around marriage intentions. “Good” (89.3% of unmarried women “intend to marry eventually” いずれ結婚するつもり) and “Bad” (only near-term marriage intentions: “within one year” 一年以内に結婚したい or “with ideal partner within one year” 理想の相手なら一年以内に結婚してもよい). The 89.3% figure appears high compared to a 2021 Cabinet Office survey finding only 60% marriage intentions, probably reflecting social desirability bias. The “Bad” scenario captures more genuine intentions but is analytically unhelpful. No municipalities reach 1.8.
The research lacks municipal-level income data, a glaring omission the authors acknowledge. The analysis cannot directly measure economic security. Housing costs receive no measurement despite probably being the binding constraint in metropolitan areas. The cross-sectional design prevents causal claims. The 2015 data predates COVID-19 and remote work normalization. If commuting friction suppresses desire for children, flexible work arrangements should theoretically elevate desired birth rates, but we still have limited studies on this.
National Policy Misses What Matters
The Japanese government’s 2023 “Children’s Future Strategy” emphasizes income support: child allowances, parental leave, subsidized childcare. Former Prime Minister Kishida pledged to double child care spending by the early 2030s, calling it “the last chance to reverse the declining birth rate trend.” An Asahi Shimbun poll found 73% of respondents don’t believe it will work. Three decades of similar initiatives producing nothing justifies their skepticism.
The regression shows why. Time poverty from commuting, spatial constraints from housing, social infrastructure from community networks, and municipal spending priorities matter as much as income support. These require local control: zoning, transit, public space, budgets. National policy barely touches them.
Akashi’s mayor cut public works to fund child welfare. Nagareyama built station childcare systems that make dual-income parenting physically possible. Most municipalities don’t do this. Elderly voters outnumber young families. Construction interests want their budgets. No national policy can force municipal spending shifts when local incentives point elsewhere.
The data has holes. No municipal income data exists. Housing costs get no measurement. The 2015 timing predates remote work. The cross-sectional design can’t prove causation.
What This Shows
The 143% variation between municipalities (0.81 to 1.97) isn’t measurement error. Shaw’s research shows TFR conflates two (and maybe even more) independent processes. Using it as your primary metric means you are flying blind. Policy designed with single or couple metrics reduce the chance for success.
The local pattern repeats. Akashi hit 1.65 versus national 1.3. Nagareyama reached 1.50 versus 1.26. Tokunoshima sustains 2.25 through organic networks. Child welfare spending dominates every other predictor (0.47 coefficient). Nagoya shows no urban fertility penalty despite similar density to Tokyo and Osaka because Nagoya municipalities spend aggressively on family support.
Local governments control every lever here: budgets, commute patterns, housing policy. They alone can’t get us above replacement level, but they are part of the solution to get us there, provincial, state, prefecture, and national level governments need to do their part. The evidence accumulates that they can move the needle. Akashi and Nagareyama aren’t miracles. They’re proof.