Working Papers

Unfamiliar ground: shared services and the performance of planning authorities in England

with Thomas Elston · Draft, September 2026

Abstract Inter-municipal cooperation is widely promoted as a route to efficiency in local public services, yet evidence on its operational consequences is thin. We study shared planning services in English local government, adopted between 2007 and 2022, with a difference-in-differences design across the border segments between partner councils and their neighbours on around one million planning applications. Shared services raised the probability of a late decision by roughly five percentage points and amplified the long tail of delay by around a third, with no offsetting gain in decision quality or cost and no change in staffing, workload or turnover. The delay has two sources. Officers who work across the boundary receive cases on ground they have never worked, and those are the cases that run late; repetition erodes the penalty. Where partnerships also consolidate their case-management systems, a second and larger slowdown arrives with the new system and falls on every officer in the council. Sharing is costly where it moves work away from what officers know.

Trade war and state intervention: evidence from public procurement in China

with Andrés Rodríguez-Pose · Draft, June 2026 · prepared for the Journal of Economic Geography special issue on the state and the remaking of economic geographies

Abstract Accounts of state capitalism have concentrated on the supply side: the state as owner, investor, and regulator. We study an under-examined demand-side modality, the state as a buyer, by asking how public procurement responds to an external shock. Using the US–China trade war as a large and geographically uneven shock to Chinese localities, the universe of public procurement contracts from 2016 to 2022, and a stacked difference-in-differences design, we find that trade-war exposure raises the value of public procurement by about 35 per cent. The response runs on the extensive margin, drawing more firms and buyers into denser state–firm networks rather than deepening existing ties; it is multi-scalar, with central government responding earliest and local government carrying the larger volumes; and it concentrates in critical and strategic sectors. Rather than retreating into local protectionism, exposed localities source more from suppliers beyond their own jurisdiction. The state as a buyer thus emerges as a consequential, spatially active modality of intervention, a demand-side counterpart to the more familiar state as owner and regulator.

Mapping the policy-instrument space: the making of local AI strategy in China

Draft, 2026 · prepared for the Journal of Economic Geography special issue on advancing economic geography with AI

Abstract Just as regions branch into industries related to what they already do, what a government can feasibly do next is shaped by the policies it has already put in place. Policy has rarely been studied as a relational, evolving object of this kind, because it is locked in unstructured administrative text and hard to observe at scale. This paper addresses that with a large language model, reading 2,983 artificial-intelligence policy documents issued by Chinese governments between 2017 and 2025 and coding, for each, the instruments it deploys, its regulatory stance, and whether it originates in local initiative or central mandate. Treating instruments that recur together within policies as related, it builds a policy-instrument space and, for every city, a relatedness density measuring how close an as-yet-unused instrument is to the city’s existing mix. Three results follow. Bottom-up and top-down policy form sharply different geographies and instrument mixes. The instrument space has a clear core of widely shared instruments, with rarer and more specialised ones at its periphery. And the space evolves path-dependently, but only where cities write their own policy: relatedness predicts which instrument a city adopts next where policy is locally initiated, but not where it is centrally mandated.

Evaluation of sector-led improvement among local planning authorities: does peer review improve productivity?

with Thomas Elston · Draft, March 2026

Abstract Peer review is a form of sector-led improvement for local government, often regarded as less intimidating than inspection and more cost effective than private consultancy. Yet in planning as in other local services, there is no conclusive evidence of its effectiveness at improving organisational performance based on objective comparison of participating and non-participating authorities. Using data on approximately 3.28 million planning applications decided in England between 2004 and 2025, this evaluation shows that peer review substantially improves decision speeds without sacrificing quality. Against a baseline in which 28.1 per cent of applications are determined later than their statutory target date, peer review reduces the probability of delay by 4.3 percentage points, a 15 per cent reduction relative to the pre-treatment mean. The effect is immediate, strengthens over the first twelve months, and sustains for at least a further two years, while refusal rates, appeal rates and application volumes remain unchanged. Benefits vary by fiscal context, decision-making route and application type: reviews conducted before April 2011 achieved reductions more than twice as large as later ones, suggesting that austerity weakened organisational learning. Review team composition and the detail of post-review action plans also shape how much authorities gain.