Overview

1 Project motivation

Perennial wheatgrass, commercially known as Kernza, is a deep-rooted grain crop that offers substantial environmental advantages over annual wheat: reduced soil erosion, higher water-use efficiency, and persistent carbon sequestration. An open empirical question is how a regional shift in production from traditional wheat to perennial wheatgrass would ripple through the broader economy, affecting input suppliers, processors, transportation, and labor markets.

This project builds a regional input-output (IO) model using non-survey methods anchored to the 2017 BEA national accounts. The model compares the economic multiplier effects of the two crops and quantifies the net economic and environmental impacts of a supply-side substitution at the regional scale.

2 How the analysis is built

The pipeline proceeds in five stages, each implemented as a module that consumes and produces the same NationalIOTable structure.

Stage Module Purpose
1 national_io.py Build and validate the national I×I table from BEA Use/Make data
2 disaggregation.py Split the national grain sector into Traditional Wheat and Perennial Wheatgrass
3 balancing.py Re-balance an expanded table via RAS when independent targets are supplied
4 regionalization.py Downscale to a sub-national region using Flegg’s Location Quotient
5 scenario.py Run adoption scenarios; compute multipliers, time paths, and satellite accounts

3 Roadmap

The sections are ordered for reading start to finish. The first three cover the method and the national table, the next four calibrate the crop split and add the environmental satellite accounts, and the last two regionalize the model and project it to 2060.

3.1 IO Primer

Introduces the Leontief framework: the transaction table, technical coefficients, the Leontief inverse, and Type I and Type II multipliers, with historical context from Leontief’s 1936 table to the current BEA benchmark accounts. Readers unfamiliar with input-output methodology should begin here; none of it is specific to this project.

3.2 National Input-Output Table

Documents Stage 1: construction and validation of the 397-sector national I×I table from the 2017 BEA Use and Make data, covering the row and column accounting identities, the Industry Technology Assumption (ITA) construction (Z_ITA = C @ Z_use), the commodity-versus-industry cross-gap, PCE normalization for the closed (Type II) model, and the national multiplier algebra. Every downstream stage inherits this table’s structure.

3.3 Sector Disaggregation

Stage 2: a hybrid physical-monetary method for splitting the single BEA grain-farming sector into Traditional Wheat and Perennial Wheatgrass. It derives the parameters (land share α, physical yield ratio φ_physical, price ratio ρ, the operational monetary yield factor φ = φ_physical × ρ, and the input-specific γ intensities for seed, fuel, fertilizer, pesticide, and ag services) and runs a national-level price/yield/intensity sensitivity analysis. The split preserves balance by construction. A closing section lists extensions the current steady-state model does not capture.

3.4 Yield

A survey of reported φ_physical (Kernza-to-wheat physical yield ratio) values from field trials in Kansas, New York, Washington, Wisconsin, and Sweden, with a plot. It provides the evidence base for the φ_physical ≈ 0.087 assumption used in the model.

3.5 Price Premium

A companion survey covering ρ, the Kernza/wheat price ratio. It includes only sources that report both a Kernza price and a wheat reference price in the same document, which makes ρ directly computable; the reported values range from a feed-grade discount (0.64×) to a large specialty premium (14.5×).

3.6 Regionalization

Stage 4, as an end-to-end prototype: national ITA table → grain disaggregation → Flegg’s Location Quotient (FLQ) regional model for a grain and livestock region. It shows why the FLQ size discount matters: it brings the Perennial WG column sum below 1 and makes the regional Leontief inverse well-conditioned. Stage 3 uses real regional and national employment by BEA sector, so the location quotients reflect the region’s actual specialization; the remaining approximations (two Stage 2 input intensities, the scenario land share and price premium, the employment-ratio estimate of regional output) are flagged throughout.

3.7 Scenario Analysis

Stage 5: projections of the regional economic impact of Kernza adoption from the 2017 baseline to 2060 under three pathways (conservative, baseline, optimistic) that differ in adoption speed, breeding-driven yield gains, and price-premium evolution. It covers parameter trajectories, single-parameter sensitivity, time-path value-added projections against a no-adoption counterfactual, break-even ρ* analysis, and multiplier dynamics. The environmental scenario projections moved to Environmental Satellite Accounts.

3.8 Scenario Analysis: Basin Region

Reruns the three pathways on the upper Smoky Hill basin from the Regionalization page instead of the 1% placeholder region, and compares the two. The scenario ranking is unchanged. The regional grain multipliers fall close to 1, because FLQ treats nearly all grain-farm inputs as imports in a small region that specializes in grain.

3.9 Environmental Satellite Accounts

Attaches physical resource intensities (water, nitrogen leaching, energy, greenhouse gas, pesticide) to the disaggregated model. The farm-level tier divides each per-hectare load by revenue per hectare to get a per-dollar intensity and solves for the break-even price premium ρ* at which Kernza matches wheat. The full-economy tier adds EPA USEEIO industry coefficients for every other sector so the burden embodied in purchased fertilizer, fuel, and electricity can be traced through the Leontief inverse (production-based versus consumption-based accounting). Extended agronomic derivations are in two supplementary pages, Water Use Satellite Account and Environmental Externalities.

3.10 References

The consolidated bibliography for every citation used across the site.

3.11 Appendix: multiplier tables

Filterable full-detail tables generated by the pipeline:

4 Key results (2017 BEA, 397-sector detail)

At maximum NAICS detail the national I×I table has 397 sectors. The table below is generated from output/national_multipliers_2017_detail.csv (written by main.py) so it always matches the model, not a transcribed snapshot. The spectral radius and the Leontief-convergence check are on the National IO page.

Table 1
Value
Quantity
Sectors 397
Total output ($X$) $34.3 trillion
Type I output multiplier, mean (range) 2.37 (1.00 to 4.19)
Type II output multiplier, mean (range) 4.43 (1.28 to 8.51)
Type I income multiplier, mean (range) 3.28 (-9.05 to 40.91)
Type II income multiplier, mean (range) 6.25 (-17.23 to 77.90)

Income multipliers span a wide range, including negative values, for the handful of sectors with near-zero or negative direct value added (owner-occupied housing, some government enterprises); see National IO for the normalisation. The complete 397-sector table is output/national_multipliers_2017_detail.csv.