Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

Product-level value chains from firm data: mapping trophic levels into economic growth

T0 review · 4 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a product-level value chain reconstructed from Italian firms' import-export records carries an intrinsic trophic ordering, and that a country's position in that ordering—measured by the average trophic level of its…

desk verdict The product-level network construction is solid and new; the growth-prediction claim is oversold and has a sign error that needs fixing. read the letter →

arxiv 2505.01133 v1 pith:KWTPJOXQ submitted 2025-05-02 physics.soc-ph econ.GNq-fin.EC

Massimiliano Fessina , Andrea Tacchella , Andrea Zaccaria This is my paper · ORCID
classification physics.soc-phecon.GNq-fin.EC
keywords product-levelvaluechainstrophiclevelsfirm-leveltradedataupstreamnessdownstreamnesseconomicgrowthpredictionbipartiteconfigurationmodelrevealedcomparativeadvantage
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that the fine-grained structure of real value chains can be recovered from firm-level trade data and that this structure carries economic information that sector-level input-output tables miss. Using import and export records of about 59,000 Italian firms, it builds a directed network of roughly 5,000 products, linking an input product to an output product when a statistically validated number of firms import the former and export the latter. Assigning each product a trophic level—its distance from basic inputs—reveals a hierarchy from raw materials to final goods whose coherence is significant only at the most detailed product classification. Country-level indicators built from these product positions, using country trade data, are strongly associated with industrial diversification and add predictive power for ten-year GDP per capita growth beyond standard country controls. If the claim is right, product-level trophic metrics offer a data-driven way to measure where countries stand in global production chains and where growth is likely to come from.

What carries the argument

The load-bearing object is the product-level directed network built by linking each imported product to each exported product of the same firm, weighted by the number of firms doing both, then filtered with the Bipartite Configuration Model so that only links with significant co-occurrence survive. On this network the authors solve the directed-graph trophic-level equation $\Lambda\vec h = \vec v$ (with the symmetrized Laplacian and imbalance vectors) to assign every product a trophic level, and measure the network's trophic coherence to test how tree-like the hierarchy is. The same trophic levels are then applied to country import/export baskets to compute normalized indicators (average import trophic level and trophic jump, each divided by the logarithm of export diversification), which enter the growth regressions as predictors.

What would settle it

One concrete test: compute the product network and trophic levels from firm-level import-export data of a second major economy (e.g., Germany or China) using the identical pipeline, and compare the product trophic-level rankings; if the rankings diverge substantially, or if country-level growth regressions built on the alternative rankings lose significance, the claim that Italian-derived product positions capture global value-chain structure is falsified. A cheaper observational check: see whether the reported 2023 pattern that no poorly diversified country imports upstream products persists when the network is rebuilt from non-Italian firm data.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a product-level input-output network constructed from firm-level import/export data has a statistically significant trophic structure that survives validation against null models only at the 6-digit Harmonized System level. Trophic levels order products from basic inputs (e.g., coarse animal hair) to final goods (e.g., traveling circuses), and the trophic jump—average trophic level of a firm's exports minus that of its imports—is positive for most firms, largest in Arms and Vehicles, and negative only for Works of art. At the country level, more diversified economies import more upstream products, less diversified economies are constrained to import downstream products (the "trophic barrier"), and a country's normalized average import trophic level and normalized trophic jump are statistically significant predictors of future GDP per capita growth in linear regressions, with the combined model reaching the highest adjusted $R^2$. The authors also find that this trophic organization is lost when products are aggregated to 4-digit or 2-digit classifications, which they read as evidence that coarse input-output tables miss the directional structure of value chains.

Load-bearing premise

The entire country-level analysis assumes that the product hierarchy inferred from Italian firms' 2007 trade records generalizes worldwide, despite the authors' own note that Italian-only import/export data may bias product positions.

Editorial extensions

If this is right

  • The combined trophic model raises adjusted $R^2$ from 0.075 (baseline) to 0.218 for ten-year GDP per capita growth, so if the claim is right, product-level trophic indicators are a practically useful complement to standard growth regressions.
  • Trophic coherence being significant only at the 6-digit level implies sector-level input-output analyses likely underestimate the directional structure of production, and firm-level product data should be used where available.
  • The triangular relationship between diversification and import trophic level implies a "trophic barrier" that constrains poorly diversified economies to importing final goods, which would shape how industrial policy and diversification strategies are evaluated.
  • Sector-level trophic jumps identify where value is added in the chain (e.g., Arms and Vehicles transform inputs the most), offering a systematic ranking of industries by their distance between inputs and outputs.
  • Because the regressions hold for 5-year and 10-year horizons and for 2007-2011 averaged trophic levels, the predictive relationship appears temporally stable across the period studied.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the Italian-derived hierarchy is truly global, the same pipeline could be run on firm-level data from several countries and the product rankings compared; a multi-country consensus ranking would remove the single-country bias the authors flag.
  • The negative sign on normalized trophic jump, conditional on import trophic level, suggests that high-jump economies such as Bangladesh and Mexico are positioned as assembly platforms; one testable extension is to check whether high NTJ countries with low NITL systematically underperform in upgrading to more upstream production.
  • Because the data exclude services, the reported indicators likely understate the value added by service-oriented economies such as the US; adding services trade or firm-level service inputs would probably shift their measured positions and could change the growth-regression coefficients.
  • The paper's trophic-level framework could be coupled with product-complexity or fitness metrics to separate "how far from raw materials" from "how hard to produce", which might sharpen the growth predictions further.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper reconstructs a directed, product-level input-output network from Italian firm-level import/export data (2007, about 59,000 firms and 5,000 products), filters links using a bipartite configuration model with Benjamini-Hochberg correction, and computes trophic levels for each product. It shows that the network has weak but statistically significant trophic coherence at the 6-digit product level, which disappears at coarser aggregation levels. The authors then define firm- and sector-level trophic jumps, extend the product-level hierarchy to countries using UN-COMTRADE data, and construct two normalized country-level indicators (NITL and NTJ). These indicators are used in linear regressions of future GDP per capita growth, and the model including both indicators (TL3) achieves the highest adjusted R-squared. The paper claims that these Italian-derived metrics are predictive at the global country level and capture significant input-output structure of global value chains.

Significance. If the central claim holds, the paper makes a valuable methodological contribution by moving value-chain analysis from coarse sector-level input-output tables to a statistically validated product-level network. The network construction is careful: it uses a maximum-entropy null model, explicit sampling, and multiple-testing correction, and the descriptive results on trophic coherence and sector-level jumps are interesting in their own right. However, the country-level generalization is not yet convincingly established. The manuscript itself flags the Italian-data limitation, the fixed-origin robustness check is only partially supportive, and the growth regression is an in-sample test with small relative gains in adjusted R-squared and an unstable sign for one key coefficient. A sign inconsistency in the definition of trophic jump further undermines the current presentation of results.

major comments (4)
  1. [Methods, Eq. (20); Firms' trophic direction] Eq. (20) defines TJ_f = ITL_f − ETL_f, but the text in 'Firms' trophic direction' defines a positive trophic jump as occurring when firms export products with a higher trophic level than their imports. With the displayed formula, a firm exporting more downstream products would have a negative trophic jump. This makes the reported distribution centered around +0.1, the sector rankings in Fig. 4, and the country-level NTJ values in Fig. 6 incompatible with the equation as written. This is not cosmetic: NTJ and its regression coefficients in Table 2 inherit the sign convention. The authors must either correct Eq. (20) to TJ_f = ETL_f − ITL_f or explicitly re-derive every subsequent interpretation and regression sign under the stated convention.
  2. [Discussion; Supplementary S4 (Fig. S11)] The claim that Italian-derived product trophic levels generalize to the global value chain is the load-bearing premise for the entire country-level analysis, yet the manuscript offers no independent validation of that premise. The Discussion itself concedes that the methodology 'can potentially be biased by two factors: being based entirely on Italian data, and considering only imports and exports of Italian firms.' The fixed-origin robustness check in Fig. S11 is only partially supportive: the negative diversification-import-trophic-level correlation is significant for Italy and China but is lost for Malaysia and Saudi Arabia, with only the triangular envelope persisting. An in-sample growth regression that uses the same hierarchy as its main covariate cannot serve as external validation. I would like to see a genuine out-of-sample test, for example deriving trophic levels from a subset of firms or from another country's firm-level data and then testing country-level predictions.
  3. [Trophic levels predict economic growth; Table 2] The growth-regression evidence is presented as the proof of predictive power, but it is an in-sample regression with model selection across TL1, TL2, TL3, EXP, and ETL2. The best model (TL3) improves adjusted R-squared by only 0.016 over EXP (0.218 vs. 0.202), and the NTJ coefficient changes sign from +0.035** in TL2 to −0.045*** in TL3. This instability, even with the residualization reported in Table S7, indicates strong collinearity and does not by itself establish that the trophic metrics add predictive content beyond export diversification. A time-split or out-of-sample evaluation is needed before the abstract's 'prove' claim can be supported.
  4. [Trophic levels predict economic growth; Methods] The regression setup does not state explicitly whether NITL and NTJ are computed from country trade data of the same year y used for the baseline controls, or from the 2023 UN-COMTRADE data introduced in the preceding section. If the latter, the covariates contain information from the end of the prediction window and the growth regression would be subject to look-ahead bias. Supplementary S5 shows temporal robustness of the scatter plots, but the regression tables should clarify the construction year of the country-level indicators for each observation.
minor comments (4)
  1. [Methods, Eqs. (17)-(20)] Equation (19) is an empty equation number between (18) and (20); this is likely a typographical artifact and should be removed or renumbered.
  2. [Figure S12 caption] The caption refers to 'panels a and b' for all six panels; it should refer to panels c-f in the appropriate places.
  3. [Throughout] The text uses comma decimal separators in tables and equations (e.g., '0,075' in Table 2), which is inconsistent with the English-language text and may confuse readers; use periods instead.
  4. [Firms' trophic direction] There is a missing space in 'positivetrophic jump' and a few other minor spacing issues that should be corrected during copyediting.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: product-level trophic levels come from Italian firm data, while country-level growth validation uses external UN-Comtrade and World Bank data.

full rationale

The derivation chain is self-contained: the product network and trophic levels h are computed exclusively from the 2007 ISTAT firm import/export matrices (Methods, eqs. (2)-(15)); the country indicators NITL and NTJ are then built by applying these h to UN-Comtrade country import/export matrices (eqs. (17)-(20) and main text); and the growth regressions use World Bank GDPpc, population, and trade data as outcomes (eq. (1)). No parameter is fitted to GDP growth before the indicators are constructed, so the growth test is an external benchmark rather than a reinsertion of the input data. The firm- and sector-level trophic jumps are descriptive uses of the same firm data, not independent predictions; they are not presented as a validation of the hierarchy. The authors explicitly concede the single-country limitation in the Discussion ('our methodology can potentially be biased by two factors: being based entirely on Italian data, and considering only imports and exports of Italian firms'), and the fixed-country robustness check in S11 shows the correlation is only partially reproduced for Malaysia and Saudi Arabia; these are empirical limitations, not circular reductions. The collinearity between NITL, NTJ, and export diversification is acknowledged and addressed by residualization (S7); whatever the statistical merits of that procedure, NITL is not by construction equal to export diversification or to a standard complexity index. Self-citations (refs. 20, 24, 29, 35, 39) are methodological or supportive and are not invoked to define the target result or to forbid alternatives. No equation-level reduction from output to input can be exhibited.

Assumptions & free parameters 6 free parameters · 5 assumptions · 1 invented entities

The paper introduces no new physical entities. It relies on six free parameters or normalization choices, the most consequential being the RCA threshold, the p-value threshold, and the normalization by log(export diversification). The axioms are dominated by the domain assumption that co-import-export links are production links and that an Italian-derived product hierarchy transfers globally. These are reasonable but untested premises, and they carry much of the paper's interpretive weight.

free parameters (6)
  • RCA threshold = 1
    Imported and exported products are binarized as competitive if Revealed Comparative Advantage >= 1, a standard but arbitrary cutoff that changes which firm-product links enter the network.
  • p-value threshold alpha in Benjamini-Hochberg procedure = 0.01
    Chosen by the authors for validating network links; it directly controls the density and composition of the filtered product network.
  • Number of null model samples = 1000
    Poisson-Binomial p-values are estimated from 1000 explicit samples of the null ensemble; this choice affects the resolution of low p-values.
  • Network construction year = 2007
    Selected as the first year of a five-year span with coherent HS product codes; the choice of base year is data-driven but affects all derived trophic levels.
  • Additive constant of trophic levels = minimum set to 1
    The Laplacian equation defines trophic levels up to an additive constant per connected component; the paper fixes the global minimum to 1, which sets the absolute scale of all country- and firm-level trophic metrics.
  • Normalization divisors NITL and NTJ = log(export diversification)
    The authors divide average import trophic level and trophic jump by the logarithm of export diversification to account for size effects; this transformation introduces collinearity that the paper then has to manage.
assumptions (5)
  • domain assumption A firm that imports product p and competitively exports product p' reveals an input-output relation from p to p'.
    This is the core inference rule behind the network. It fails for trading companies, multi-product assemblers that import for internal use, and firms whose imports and exports belong to separate production lines.
  • domain assumption The Bipartite Configuration Model, preserving degree sequences, captures the relevant null distribution for spurious product links.
    The filtering step removes only co-occurrences explained by firm and product degrees. If firm size, industry, or other confounders generate spurious associations, the retained links still contain bias.
  • domain assumption The product hierarchy inferred from Italian firms generalizes to the global value chain.
    All country-level metrics use trophic levels derived from the Italian firm network in 2007. The authors acknowledge this in the Discussion as a potential bias; if the hierarchy is country-specific, the cross-country predictions inherit that bias.
  • domain assumption Trophic levels on this network correspond to an economic hierarchy from basic goods to final products.
    The mathematical definition of trophic level is purely graph-theoretic; the interpretation as production stage is an economic assumption that the paper motivates but does not prove.
  • standard math The linear growth regression with GDP per capita, population, trade, and year fixed effects isolates a predictive effect of trophic position.
    Standard econometric identification assumptions are required for the coefficient estimates to have predictive meaning; the paper does not perform out-of-sample or causality tests.
invented entities (1)
  • Trophic barrier
    purpose: Describes the empirical pattern that poorly diversified countries do not import upstream products.
    The 'trophic barrier' is a descriptive label for a scatterplot pattern, not a mechanism or entity with independent falsifiable consequences.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Product-level value chains from firm data: mapping trophic levels into economic growth." pith.science (2026). https://pith.science/paper/KWTPJOXQ

@misc{pith2026250501133,
  author       = {Pith},
  title        = {Pith review of: Product-level value chains from firm data: mapping trophic levels into economic growth},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KWTPJOXQ}},
  note         = {Machine review of arXiv:2505.01133}
}
read the original abstract

We reconstruct a product-level input-output network based on firm-level import-export data of Italian firms. We show that the network has a statistically significant, yet nuanced trophic structure, which is evident at the product level but is lost when the classification is coarse-grained. This detailed value chain allows us to characterize the trophic distance between inputs and outputs of single firms, and to derive a coherent picture at the sector level, finding that sectors such as weapons and vehicles are the ones with the largest increase in downstreamness between their inputs and their outputs. Our measure of downstreamness at the product level can be used to derive country-level indicators that characterize industrial strategies and capabilities and act as predictors of economic growth. With respect to the standard input/output analysis, we show that the fine-grained structure is qualitatively different from what can be observed using sector-level data. We finally prove that, even if we leverage exclusively data from Italian firms, the metrics that we derive are predictive at the country level and capture a significant description of the input-output relations of global value chains.

Figures

Figures reproduced from arXiv: 2505.01133 by the authors.

Figure 1
Figure 1. Small subgraph of the directed network of products we reconstruct. Here, a link is present if a statistically high number of firms import the source node and export the target node. In this layout, we order the nodes vertically using the trophic level. This induced hierarchy organized the products from basic inputs (at the bottom), to increasingly complex components up to products for final consumption (at the top) … view at source ↗
Figure 2
Figure 2. Products’ network aggregated at the section level, filtered with the PMFG method32: the size of nodes is proportional to the average trophic level of the products they encompass (the bigger the dot, the more downstream the sector), the width of the arrows to the intensity of links. 3/20 [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Empirical value (vertical red line) and null distributions of the trophic coherence of the input-output networks built at 6 digit (a), 4 digit (b) and 2 digit (c) aggregation level. The trophic coherence is significant with respect to both the Erdòs-Renyi model and the Directed Configuration Model only for products at the 6-digit level (a), signaling that the trophic organization of the product’s network emerges onl… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Average export vs import trophic level of firms grouped into industrial sectors. The distance from the identity line (trophic jump) measures the transformation operated by industries on their inputs, while their position along the identity line indicates at which stage…
Figure 5
Figure 5. Figure 5: Export diversification of countries vs average trophic level of their imports. There is a strong and significant negative correlation between the two quantities, both in terms of values (Pearson) and rankings (Spearman). Asterisks denote the significance level of the c…
Figure 6
Figure 6. Figure 6: Left: export diversification of countries vs average trophic jump. There is a strong and significant positive correlation between the two quantities, both in terms of values (Pearson) and rankings (Spearman). Asterisks denote the significance level of the correlation c…
Figure 7
Figure 7. Figure 7: Construction of the products’ input-output network. Given the tripartite network of imports and exports by Italian firms for a fixed yeary y, a directed monopartite network encoding input-output relationships between products is built by putting a link from product p1 …

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Synthetic supply networks

    econ.GN 2026-08 conditional novelty 6.0 of 10

    A new pipeline generates synthetic firm-level supply networks that match selected empirical network statistics and aggregate exactly to national input-output tables using only public data.

Reference graph

Works this paper leans on

41 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [1]

    & Hillberry, R

    Antràs, P., Chor, D., Fally, T. & Hillberry, R. Measuring the upstreamness of production and trade flows.Am. Econ. Rev. 102, 412–416, DOI: https://doi.org/10.1257/aer.102.3.412 (2012)

    doi:10.1257/aer.102.3.412 2012

  2. [2]

    Production staging: measurement and facts

    Fally, T. Production staging: measurement and facts. Boulder, Color. Univ. Colo. Boulder, May155–168 (2012)

    work page 2012

  3. [3]

    Miller, R. E. & Temurshoev, U. Output upstreamness and input downstreamness of industries/countries in world production. Int. regional science review40, 443–475, DOI: https://doi.org/10.1177/0160017615608095 (2017)

    doi:10.1177/0160017615608095 2017

  4. [4]

    McNerney, J., Savoie, C., Caravelli, F., Carvalho, V . M. & Farmer, J. D. How production networks amplify economic growth. Proc. Natl. Acad. Sci. 119, e2106031118, DOI: https://doi.org/10.1073/pnas.2106031118 (2022)

    doi:10.1073/pnas.2106031118 2022

  5. [5]

    Correlation between upstreamness and downstreamness in random global value chains

    Bartolucci, S., Caccioli, F., Caravelli, F. & Vivo, P. Correlation between upstreamness and downstreamness in random global value chains. arXiv preprint arXiv:2303.06603 DOI: https://doi.org/10.48550/arXiv.2303.06603 (2023)

    work page Pith review arXiv doi:10.48550/arxiv.2303.06603 2023

  6. [6]

    & Vivo, P

    Bartolucci, S., Caccioli, F., Caravelli, F. & Vivo, P. Upstreamness and downstreamness in input–output analysis from local and aggregate information. Sci. Reports 15, 2727, DOI: https://doi.org/10.1038/s41598-025-86380-6 (2025)

    work page doi:10.1038/s41598-025-86380-6 2025

  7. [7]

    Input-output economics (Oxford University Press, 1986)

    Leontief, W. Input-output economics (Oxford University Press, 1986)

    1986

  8. [8]

    M., Ozdaglar, A

    Acemoglu, D., Carvalho, V . M., Ozdaglar, A. & Tahbaz-Salehi, A. The network origins of aggregate fluctuations. Econometrica 80, 1977–2016, DOI: https://doi.org/10.3982/ECTA9623 (2012)

    doi:10.3982/ecta9623 2012

Show all 41 references
  1. [9]

    & Todo, Y

    Inoue, H. & Todo, Y . Firm-level propagation of shocks through supply-chain networks.Nat. Sustain. 2, 841–847, DOI: https://doi.org/10.1038/s41893-019-0351-x (2019)

    2019 doi

  2. [10]

    & Thurner, S

    Diem, C., Borsos, A., Reisch, T., Kertész, J. & Thurner, S. Quantifying firm-level economic systemic risk from nation-wide supply networks. Sci. reports 12, 1–13, DOI: https://doi.org/10.1038/s41598-022-11522-z (2022)

    2022 doi

  3. [11]

    Fessina, M. et al. Inferring firm-level supply chain networks with realistic systemic risk from industry sector-level data. arXiv preprint arXiv:2408.02467 DOI: https://doi.org/10.48550/arXiv.2408.02467 (2024). 12/20

    2024 doi arXiv

  4. [12]

    & Todo, Y

    Inoue, H., Okumura, Y ., Torayashiki, T. & Todo, Y . Simulation of supply chain disruptions considering establishments and power outages. PloS one 18, e0288062, DOI: https://doi.org/10.1371/journal.pone.0288062 (2023)

    2023 doi

  5. [13]

    M., Lafond, F

    Pichler, A., Pangallo, M., del Rio-Chanona, R. M., Lafond, F. & Farmer, J. D. Forecasting the propagation of pandemic shocks with a dynamic input-output model. J. Econ. Dyn. Control. 144, 104527 (2022)

    2022

  6. [14]

    Ialongo, L. N. et al. Reconstructing firm-level interactions in the dutch input–output network from production constraints. Sci. Reports 12, 1–12, DOI: https://doi.org/10.1038/s41598-022-13996-3 (2022)

    2022 doi

  7. [15]

    & Lafond, F

    Mungo, L., Brintrup, A., Garlaschelli, D. & Lafond, F. Reconstructing supply networks. J. Physics: Complex. 5, 012001, DOI: https://doi.org/10.1088/2632-072X/ad30bf (2024)

    2024 doi

  8. [16]

    Kosasih, E. E. & Brintrup, A. A machine learning approach for predicting hidden links in supply chain with graph neural networks. Int. J. Prod. Res. 1–14, DOI: https://doi.org/10.1080/00207543.2021.1956697 (2021)

    2021

  9. [17]

    & McFarlane, D

    Wichmann, P., Brintrup, A., Baker, S., Woodall, P. & McFarlane, D. Extracting supply chain maps from news articles using deep neural networks. Int. J. Prod. Res. 58, 5320–5336, DOI: https://doi.org/10.1080/00207543.2020.1720925 (2020)

    2020

  10. [18]

    & Thurner, S

    Chakraborty, A., Reisch, T., Diem, C., Astudillo-Estévez, P. & Thurner, S. Inequality in economic shock exposures across the global firm-level supply network. Nat. Commun. 15, 3348, DOI: https://doi.org/10.1038/s41467-024-46126-w (2024)

    2024 doi

  11. [19]

    & Lafond, F

    Bacilieri, A., Borsos, A., Astudillo-Estevez, P. & Lafond, F. Firm-level production networks: what do we (really) know? Tech. Rep. 2023-08, INET Oxford Working Paper (2023)

    2023

  12. [20]

    & Squartini, T

    Fessina, M., Zaccaria, A., Cimini, G. & Squartini, T. Pattern-detection in the global automotive industry: a manufacturer- supplier-product network analysis. Chaos, Solitons & Fractals 181, 114630, DOI: https://doi.org/10.1016/j.chaos.2024. 114630 (2024)

    2024 doi

  13. [21]

    & Tiwari, A

    Brintrup, A., Barros, J. & Tiwari, A. The nested structure of emergent supply networks. IEEE Syst. J. 12, 1803–1812, DOI: https://doi.org/10.1109/JSYST.2015.2493345 (2018)

    2018

  14. [22]

    & Thurner, S

    Diem, C., Borsos, A., Reisch, T., Kertész, J. & Thurner, S. Estimating the loss of economic predictability from aggregating firm-level production networks. PNAS nexus 3, pgae064, DOI: https://doi.org/10.1093/pnasnexus/pgae064 (2024)

    2024 doi

  15. [23]

    & Hidalgo, C

    Karbevska, L. & Hidalgo, C. A. Mapping global value chains at the product level. EPJ Data Sci. 14, 21, DOI: https://doi.org/10.1140/epjds/s13688-025-00521-5 (2025)

    2025 doi

  16. [24]

    & Zaccaria, A

    Albora, G. & Zaccaria, A. Machine learning to assess relatedness: the advantage of using firm-level data. Complexity 2022 (2022)

    2022

  17. [25]

    & Garlaschelli, D

    Squartini, T. & Garlaschelli, D. Analytical maximum-likelihood method to detect patterns in real networks. New J. Phys. 13, 083001, DOI: https//doi.org/10.1088/1367-2630/13/8/083001 (2011)

    2011 doi

  18. [26]

    Cimini, G. et al. The statistical physics of real-world networks. Nat. Rev. Phys. 1, 58–71, DOI: https://doi.org/10.1038/ s42254-018-0002-6 (2019)

    2019

  19. [27]

    & Squartini, T

    Saracco, F., Di Clemente, R., Gabrielli, A. & Squartini, T. Randomizing bipartite networks: the case of the world trade web. Sci. Reports 5, 10595, DOI: https://doi.org/10.1038/srep10595 (2015)

    2015 doi

  20. [28]

    Saracco, F. et al. Inferring monopartite projections of bipartite networks: an entropy-based approach. New J. Phys. 19, 053022, DOI: https://doi.org/10.1088/1367-2630/aa6b38 (2017)

    2017 doi

  21. [29]

    & Zaccaria, A

    Cimini, G., Carra, A., Didomenicantonio, L. & Zaccaria, A. Meta-validation of bipartite network projections. Commun. Phys. 5, 76 (2022)

    2022

  22. [30]

    Several measures of trophic structure applicable to complex food webs

    Levine, S. Several measures of trophic structure applicable to complex food webs. J. theoretical Biol. 83, 195–207, DOI: https://doi.org/10.1016/0022-5193(80)90288-X (1980)

    1980 doi

  23. [31]

    S., Johnson, S

    MacKay, R. S., Johnson, S. & Sansom, B. How directed is a directed network? Royal Soc. open science 7, 201138, DOI: https://doi.org/10.1098/rsos.201138 (2020)

    2020 doi

  24. [32]

    & Mantegna, R

    Tumminello, M., Aste, T., Di Matteo, T. & Mantegna, R. N. A tool for filtering information in complex systems. Proc. Natl. Acad. Sci. 102, 10421–10426, DOI: https://doi.org/10.1073/pnas.0500298102 (2005)

    2005 doi

  25. [33]

    & Rényi, A

    Erd˝os, P. & Rényi, A. On random graphs I. Publ. math. debrecen 6, 18, DOI: https://doi.org/10.5486/PMD.1959.6.3-4.12 (1959)

    1959 doi

  26. [34]

    M., Beyer, R

    Wacker, K. M., Beyer, R. C. & Moller, L. C. Leveraging growth regressions for country analysis. Tech. Rep., The World Bank (2024). DOI: https://openknowledge.worldbank.org/entities/publication/c88bdfc7-d358-4fe4-8c79-43b0229d6747. 13/20

    2024

  27. [35]

    & Pietronero, L

    Tacchella, A., Cristelli, M., Caldarelli, G., Gabrielli, A. & Pietronero, L. A new metrics for countries’ fitness and products’ complexity. Sci. reports 2, 723, DOI: https://doi.org/10.1038/srep00723 (2012)

    2012 doi

  28. [36]

    Angrist, J. D. & Pischke, J.-S. Mostly harmless econometrics: An empiricist’s companion (Princeton university press, 2009)

    2009

  29. [37]

    revealed

    Balassa, B. Trade liberalisation and “revealed” comparative advantage 1. The manchester school 33, 99–123, DOI: https://doi.org/10.1111/j.1467-9957.1965.tb00050.x (1965)

    1965

  30. [38]

    & Kuang, D

    Thissen, D., Steinberg, L. & Kuang, D. Quick and easy implementation of the benjamini-hochberg procedure for controlling the false positive rate in multiple comparisons. J. Educ. Behav. Stat. 27, 77–83, DOI: https://doi.org/10.3102/ 10769986027001077 (2002)

    2002

  31. [39]

    WECARE”, CUP B53D23003880006, financed by the Italian Ministry of University and Research (MUR), Piano Nazionale Di Ripresa e Resilienza (PNRR), Missione 4 “Istruzione e Ricerca

    Tacchella, A., Mazzilli, D. & Pietronero, L. A dynamical systems approach to gross domestic product forecasting. Nat. Phys. 14, 861–865, DOI: https://doi.org/10.1038/s41567-018-0204-y (2018). Acknowledgements The authors acknowledge the PRIN project No. 20223W2JKJ “WECARE”, CU...

    2018 doi

  32. [40]

    we erased transactions relative to HS codes not belonging to the official classification: these extra codes were introduced by ISTAT to classify unidentified products

  33. [41]

    not explained

    we retained only those firms appearing as both importers and exporters. Following these steps, we are left with 59.341 firms and 5.008 products, covering more than 90% of the exchanged monetary volume. S2 Trophic level distribution In figure S8 we report the empirical distribu...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.