Replication data and code for: "Corridor, Not Factory: Trade Reorientation and the Missing Investment Response in Kazakhstan, 2022–2025"

Description

A large, sector-specific rise in demand for a tradeable good is expected to induce domestic investment. After February 2022, with Russia's direct trade with its main partners disrupted, part of it was reallocated through neighbouring economies, Kazakhstan among them. Using product-level trade data (HS6, 2018–2025; UN Comtrade), this dataset and code measure the demand shock, calibrate what Kazakhstan retains (a wholesale-and-freight margin of 6–14% propagated through the OECD Inter-Country Input–Output table implies 5–11 cents of domestic value added per rerouted dollar, against about three-quarters for a produced dollar), and test for an investment response using three commercial deal databases (S&P Capital IQ, PitchBook, Preqin). It finds none in the reoriented product lines and reads the null through three gates: market access, irreversibility, and capital-market institutions. Coverage. Geographic: Kazakhstan (primary); Russia, Armenia, the Kyrgyz Republic, Georgia, Türkiye, Belarus, China, EU-27, United Kingdom, United States, Japan, Korea, Switzerland, Norway (trade partners/comparators). Temporal: trade data 2018–2025 annual and 2019m1–2024m2 plus 2025 monthly; deal data 2015–2025; OECD ICIO 2019 reference year; Kazakhstan Bureau of National Statistics input–output 2023 reference year. Unit of observation: HS6 product line × period (trade panels); transaction (deal data); ISIC rev. 4 industry (input–output, sector-priority tables). Universe: 75 candidate HS6 lines (50 from the EU/US/UK/JP List of Common High Priority Items, Feb 2024, plus 25 civilian control lines), from which the 29-line "surge basket" is data-selected; all merger, acquisition, private-equity, and venture transactions with a Kazakhstan-domiciled target/issuer announced 2015–2025 (N ≈ 493 after de-duplication). This dataset regenerates every table, figure, and load-bearing numeric claim in the manuscript from public data plus one academically-licensed deal extract (not itself redistributed). R only.

Steps to reproduce

Restore the R environment from the repo root: R -e 'renv::restore()' (R 4.5.3; full package list in output/renv.lock and output/sessionInfo.txt). Set COMTRADE_PRIMARY=<key> in ~/.Renviron for the authenticated monthly Comtrade pulls. Place the two pointer inputs the licenses do not permit redistributing: OECD ICIO 2019_SML.csv at code/kz_passthrough/_data/icio/icio_ex/, and (license holders only) the Capital IQ/PitchBook/Preqin deal extract at code/kz_valueadd/_data/ (see data/raw/access-restricted-data.md to reconstruct it from the query specification and source deal identifiers if you do not already hold it). Fetch raw trade data: bash fetch_annual.sh, bash fetch_keyed.sh, bash fetch_neighbors.sh in code/kz_passthrough/; bash fetch_kz_imports_hs2.sh in code/kz_valueadd/. Run Rscript 00_run_all.R in code/kz_passthrough/ (trade shock, value capture, DiD, structural breaks), then in code/kz_valueadd/ (investment response, sector priority — requires the deal extract from step 3). Outputs land in each pipeline's _outputs/; the Table/Figure → script:line map in the deposited README.md §6 traces every manuscript exhibit to its producing file. Expected runtime: ~12–18 minutes analysis-only (single-core, <4 GB RAM), plus 15–30 minutes for the live API pulls.

Institutions

• Nazarbayev University

  Astana, Astana

Categories

Licence