论文信息

Yinghui Wu / Peihong Xie / Yuxue Yang

Frontiers in Sustainable Food Systems

论文解读

AI 辅助整理
方法
26家中国上市农业企业2010-2022年306个观测,PCA与模糊集定性比较分
发现
静态识别4条、年度识别6条路径,仅1条共享;年度解一致性更高,静态解覆盖更广。
机会点
可将时间聚合敏感性纳入农业企业ESG与融资约束组态评估,发展动态QCA方法。

论文摘要

为比较企业在均值静态与年度面板模糊集定性比较分析下的高质量发展组态,本研究基于2010—2022年中国26家上市农业企业的306个企业—年度观测值,通过主成分分析(PCA)构建了包含15个指标的高质量发展指数(HD)。多年均值用于概括长期水平,而年度记录则保留同一年度内的组合。基线识别出四条静态路径和六条年度路径,其中包括一条共享组态,即高环境、社会与治理(ESG)表现、非强融资约束、国有产权和非污染密集型行业地位。相反的ESG与融资约束(SA)状态表明组态层面存在互补性。年度D2由非高ESG、高SA、非国有产权和污染密集型行业地位构成,在等权与熵权法下均被精确保留,并在将不一致性比例降低(PRI)阈值提高至0.65的单独PCA检验中依然成立。其识别在这些设定下均稳健;结果方向检验和案例证据进一步支持其与高HD的对应关系。在基线规则下,年度解在共同年度数据上具有更高一致性,而静态解覆盖范围略广。相对表现随筛选规则而变化,完整的基线共享组态对测量敏感。研究结果揭示了时间聚合下的组态差异,并为结合长期类型与年度状态的企业评估提供了启示。

全文

Frontiers in Sustainable Food Systems

Frontiers in Sustainable Food Systems 01 frontiersin.org
Temporal aggregation and the
identification and stability of
high-quality development
configurations in agricultural
enterprises
Yinghui Wu 1,2, Peihong Xie 3* and Yuxue Yang 4
1International Business School, Shanghai University of International Business and Economics,
Shanghai, China, 2Nanjing Open University, Nanjing, China, 3School of Management, Shanghai
University of International Business and Economics, Shanghai, China, 4School of Economics and
Management, Jiangsu University of Science and Technology, Zhenjiang, China
To compare high-quality development configurations in agricultural enterprises
under firm-mean static and annual panel fuzzy-set qualitative comparative analy-
sis, this study constructs a 15-indicator high-quality development index (HD)
through principal component analysis (PCA), using 306 firm-year observations
from 26 Chinese listed agricultural enterprises during 2010-2022. Multiyear
means summarize long-term levels, whereas annual records preserve combina-
tions within the same year. The baseline identifies four static and six annual paths,
including one shared configuration of high environmental, social, and governance
(ESG) performance, non-strong financing constraints, state ownership, and non-
pollution-intensive industry status. Opposite ESG and financing-constraint (SA)
states suggest complementarity at the configurational level. Annual D2, compris-
ing non-high ESG, high SA, non-state ownership, and pollution-intensive industry
status, is retained exactly under equal and entropy weighting and a separate PCA
check raising the proportional reduction in inconsistency (PRI) cutoff to 0.65.
Its identification is robust across these settings; outcome-direction checks and
case evidence further support its correspondence with high HD. Under baseline
rules, the annual solution has higher consistency on common annual data, while
the static solution has slightly broader coverage. Relative performance changes
with screening rules, and the complete baseline shared configuration is mea -
surement-sensitive. The findings reveal configuration differences under temporal
aggregation and inform enterprise assessment combining long-term types with
annual states.
KEYWORDS
agricultural enterprises, annual panel, economic sustainability, fuzzy-set qualitative
comparative analysis, high-quality development, temporal aggregation
1 Introduction
The economic sustainability of agricultural and food value chains depends partly on enter-
prise development quality: value creation, supply-chain cooperation, and the ability to sustain
operations. Raw-material procurement, processing, and cash collection span different business
cycles. Environmental, social, and governance (ESG) performance, financing constraints, and
regional green finance conditions may therefore combine differently from year to year.
TYPE Original Research
PUBLISHED 17 November 2026
DOI 10.3389/fsufs.2026.2017893
OPEN ACCESS
EDITED BY
Justice Gameli Djokoto,
Dominion University College, Ghana
REVIEWED BY
Qijai Liu,
Shinhan University, Republic of Korea
Shizheng Tan,
Taiyuan University of Technology, China
*CORRESPONDENCE
Peihong Xie
phxie68@163.com
RECEIVED 03 October 2026
REVISED 25 October 2026
ACCEPTED 27 October 2026
PUBLISHED 17 November 2026
CITATION
Wu Y, Xie P, and Yang Y (2026) Temporal
aggregation and the identification and
stability of high-quality development
configurations in agricultural enterprises.
Front. Sustain. Food Syst. 10:2017893.
doi: 10.3389/fsufs.2026.2017893
COPYRIGHT
© 2026 Wu, Xie and Yang. This is an
open-access article distributed under
the terms of the Creative Commons
Attribution License (CC BY). The use,
distribution or reproduction in other
forums is permitted, provided the
original author(s) and the copyright
owner(s) are credited and that the
original publication in this journal is
cited, in accordance with accepted
academic practice. No use, distribution
or reproduction is permitted which does
not comply with these terms.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 02 frontiersin.org
Multi-year averages summarize long-term levels but may group condi-
tions observed in different years into a single enterprise type.
Identifying development paths consequently requires attention not
only to which conditions are present, but also to how those conditions
are represented over time: do the types identified from firm means
match the combinations observed in annual records?
Research on development quality provides a multidimensional
basis for this question. Y ang and Zhang (2021) discuss measurement
through value creation and sustained development. Tang et al. (2026)
incorporate operating capabilities, managerial control, and industrial-
chain coordination into enterprise competitiveness, while Liu and
Zhou (2022) document regional differences and temporal change in
rural development quality. At the enterprise level, Zhang and Xu
(2025) include the ability to combine resources in their measurement
of new quality productivity and identify differences across industries
and ownership types. These studies help define development quality
and enterprise heterogeneity. Research on the European food industry
links ESG performance to financial outcomes (Sandberg et al., 2023),
and research on rural industries examines spatial differences in finan-
cial support (Qin and Pan, 2022). Bringing enterprise and regional
conditions into one framework raises a further question: do long-term
average levels and annual states reveal the same combinations of
development conditions? Recent research on Chinese agribusiness
also finds an association between ESG and investment efficiency (Ma
et al., 2025). Provincial studies examine green finance in relation to
agricultural non-point source pollution (Shen and Zhang, 2024) and
carbon emission intensity ( Tong et al., 2026 ). These studies draw
attention to sustainable enterprise operations and regional financial
environments, but regional financial indicators do not directly mea-
sure the financing received by individual firms. This study examines
how temporal representation changes joint configurations of enter-
prise and regional conditions.
Huang and Tang emphasize the coordination of technology,
institutions, markets, and actors in rural governance ( Huang and
Tang, 2025). Configurational research similarly shifts explanation
from individual relationships to combinations of conditions. Fiss
uses set relations to characterize organizational configurations and
multiple paths to the same outcome ( Fiss, 2007 ; Fiss, 2011 ).
Studies of environmental regulation in agribusiness ( Xia et al.,
2024) and agricultural digitalization ( Wang et al., 2025 ) also iden -
tify diverse combinations associated with development. This diver -
sity raises the issue of temporal representation. Garcia-Castro and
Ariño distinguish pooled, annual cross-sectional, and within-firm
relationships ( Garcia-Castro and Ariño, 2016 ). Bhattacharya dem -
onstrates the construction and interpretation of panel qualitative
comparative analysis (QCA) ( Bhattacharya, 2023 ). Du et al. (2025)
examine complex growth patterns, and Finn (2022) distinguishes
condition identification from temporal process interpretation.
Building on this literature, this study examines averaging itself:
when the same enterprise data are reduced to multi-year means,
which configurations remain visible and which annual combina -
tions disappear?
Using the same firms and conditions, this study compares firm-
mean static and annual panel configurations, evaluates both solu -
tions on common data, and contrasts calibration anchors and the
order of operations. Its contribution is to distinguish shared con -
figurations retained across temporal representations from configu -
ration differences under common condition backgrounds, thereby
separating long-term enterprise types from annual operating
states. Under baseline PCA and screening rules, high ESG, non-
strong financing constraints, state ownership, and non-pollution-
intensive industry status form a shared combination; ESG and
financing combinations differ among non-state-owned, pollution-
intensive enterprises. Comparing ESG and financing states further
suggests complementarity at the configurational level. Combining
weighting and PRI sensitivity checks, outcome-direction assess -
ment, and case support, the study finds robust identification of
annual D2, comprising non-high ESG, high SA, non-state owner -
ship, and pollution-intensive industry status, under the examined
weighting and PRI adjustments. Distinguishing this finding from
measurement-dependent shared combinations informs assessment
combining long-term classification with annual operating
diagnosis.
2 Materials and methods
2.1 Analytical framework and sample
The analysis addresses three questions: which condition com -
binations correspond to high-quality development in agricultural
enterprises, how the combinations identified from firm means and
annual records differ, and how consistently these combinations cor -
respond to the outcome across years and firms. Both models use
fuzzy-set qualitative comparative analysis (fsQCA), implemented
in R 4.4.3 with the QCA package, version 3.24 ( Duşa, 2019 ). Paths
are first identified using the same conditions. Contrasts in aggrega -
tion, calibration order, and calibration reference are then evaluated
on common data, linking path interpretation to stability
assessment.
The analytical unit is the agricultural value-chain enterprise.
Regional green finance is treated as an external condition, combined
with firm-level ESG performance, financing constraints, ownership,
and industry status. Figure 1 connects condition selection, path iden-
tification, and stability assessment. The analysis examines how internal
enterprise conditions combine with regional resources, which combi-
nations survive averaging, which appear in annual records, and how
consistently the identified solutions correspond to development qual-
ity over time.
Condition selection covers enterprise operating conditions, the
external financial environment, and structural attributes. The food-
industry ESG literature informs the relationship between sustainabil-
ity characteristics and operating performance (Sandberg et al., 2023),
while research on rural industries identifies regional differences in
financial conditions (Qin and Pan, 2022). Five conditions are therefore
included at the enterprise and regional levels: ESG performance,
financing constraints, the green finance environment, ownership, and
pollution-intensive industry status. The two models use the same
conditions.
Enterprise operating conditions comprise ESG performance and
financing constraints. ESG summarizes environmental, social respon-
sibility, and governance characteristics; financing constraints are mea-
sured by the SA index. Retaining annual variation in these variables
allows long-term characteristics to be compared with their combina-
tions in individual years.
The green finance environment is measured by the green fund
share in the firm’s city: the proportion of local fund assets allocated to

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 03 frontiersin.org
green funds. Firm-level financing constraints are measured by the SA
index. Including both levels distinguishes regional fund allocation
from firms’ own financing constraints and identifies their different
combinations.
Ownership and pollution-intensive industry status capture
structural differences. They represent resource and governance
backgrounds associated with ownership and the environmental
compliance context associated with industry membership.
Combining these attributes with annual operating conditions
identifies different development paths within broadly similar
enterprise types.
The sample comprises Chinese agricultural value-chain firms
listed on the Shanghai and Shenzhen A-share markets during 2010–
2022. Under the China Securities Regulatory Commission industry
classification (China Securities Regulatory Commission, 2012 ), it
covers agriculture, agricultural and sideline food processing, food
manufacturing, and the manufacture of alcoholic beverages and other
beverages: 7, 15, 2, and 2 firms, respectively. Table 1 lists the 26 firms,
stock codes, and industry categories. The sample spans agricultural
production and downstream processing, with manufacturing firms
accounting for the majority.
Financial and industry data are matched by firm and year.
Condition-source records designated ST, *ST, or PT in a given
year are excluded for that year only; other eligible years of the
same firm are retained. The 15 outcome indicators provide 313
observations. Requiring complete data for the five conditions
leaves 306 observations covering 26 firms and 13 years, with 6–13
usable years per firm. Both models are constructed from these
same 306 observations. ESG data come from the Huazheng
SMART ESG platform. Financial, ownership, market, and indus -
try data come from listed-company disclosures and the CSMAR,
Wind, and CNRDS databases. Green fund data are drawn from
prefecture-level city green finance indicators.
2.2 Outcome index and conditions
The high-quality development index (HD) draws on Y ang and
Zhang’s multidimensional approach (Y ang and Zhang, 2021), Tang
et al. ’s framework of operating, control, and industrial-chain coordi-
nation capabilities (Tang et al., 2026), and the State Council of the
People’s Republic of China’s guidance on rural industrial revitaliza-
tion, which emphasizes value-chain upgrading, benefit-sharing link-
ages, and sustained development ( State Council of the People’s
Republic of China, 2019 ). Fifteen indicators cover value creation,
stakeholder coordination, and operating sustainability. Green man-
agement expenditure, its ratio to revenue, and social donations are
excluded to reduce direct content overlap between HD and the ESG
condition. Seven indicators represent value creation: basic and
diluted earnings per share, stock returns with and without cash-div-
idend reinvestment, sustainable financial growth, employment scale,
and Tobin’s Q. Eight represent stakeholder coordination and operat-
ing sustainability: two comprehensive tax ratios, the income tax ratio,
an employee-compensation proxy, supplier and customer stability,
and the current and quick ratios. Employment, tax ratios, and liquid-
ity describe labor absorption, tax contributions, and short-term sol-
vency backgrounds within the composite evaluation. Data from
CSMAR, Wind, and company disclosures are Z-standardized across
313 observations before PCA. PCA summarizes the correlation
structure among indicators. The Kaiser–Meyer–Olkin (KMO) statis-
tic is 0.581; Bartlett’s test statistic is 6610.422 (105 degrees of free -
dom, p < 0.001). Six components have eigenvalues above one and
explain 80.081% of total variance. All 15 components are weighted
by their variance contributions to retain information; the six-compo-
nent cumulative share describes information concentration rather
than serving as a truncation criterion. Each component is oriented
to correlate positively with the mean of the standardized indicators.
The weighted score is standardized again. In response to the KMO
FIGURE 1
Framework for comparing static and annual panel configurations of agricultural enterprise development. Dashed boundaries distinguish enterprise
operating conditions, structural attributes, and the regional green finance environment. Arrows indicate the analytical sequence of identification and
comparison. ESG = environmental, social, and governance performance; HD = high-quality development; SA = size-age index; fsQCA = fuzzy-set
qualitative comparative analysis.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 04 frontiersin.org
result, Section 3.5.2 examines weighting sensitivity (OECD/European
Union/EC-JRC, 2008). Let F denote component scores, λ eigenvalues,
Q the weighted score, and H the final index:

λ
λ= =
−= = =∑ ∑15 151 1,, k itit k kit k itk Qjj QQQ wF w H s (1)
In Equation 1, i and t denote firm and year; k and j index princi-
pal components; and w is the variance-contribution weight. The
mean of Q and its standard deviation sQ are calculated across the 313
observations, and H is HD. Average employee compensation and
Tobin’s Q have coefficients of −0.002110 and −0.001902 in Q; the
remaining coefficients are positive. These are statistical composite
weights resulting from the correlation structure and component
weighting. Supplementary Tables S1, S2 report the first six compo-
nent loadings. The final column of Supplementary Table S2 gives the
standardized composite coefficients from all 15 components, and
multiplying them by the corresponding indicator Z-scores and sum-
ming reproduces HD.
Supplementary Table S3 lists the indicators, operational defini -
tions, data sources, and scoring coefficients. Policy documents on
leading agricultural industrialization enterprises and agricultural
industrialization consortia provide context for value-chain
upgrading and stable benefit-sharing linkages ( Ministry of
Agriculture and Rural Affairs of the People's Republic of China,
2021; Ministry of Agriculture et al., 2017 ). The employee-compen -
sation proxy is year-end employee compensation payable per
employee.
The ESG condition is the continuous Huazheng overall score
divided by 10. Since rating scope, measurement, and weighting can
differ across providers (Berg et al., 2022), all scenarios use the same
source. Financing constraints are represented by the size–age index
(SA), using the functional form proposed by Hadlock and
Pierce (2010):
= −+− 20.737 0.043 0.040it it it itSA A A L (2)
Here, A is the natural logarithm of total assets measured in mil-
lions of renminbi (RMB), and L is years since establishment. Size and
age are retained as provided in the source data without additional
caps. Signed SA values are preserved, and larger values are calibrated
into the strong-financing-constraints set. “Strong” and “non-strong”
denote membership in the high-SA set and its negation, respectively.
Path expressions retain FC for this set and ~FC for its negation.
The green finance environment is measured by the one-year lag of
the city-level green fund share, as shown in Equation 3:
( ) −= = ,1, Lctct it ci tctBG GG F (3)
G is the city’s green fund share, B is the total market value of green
funds, F is total fund assets, c identifies the city, c(i) is firm i’s city, and
superscript L indicates the one-year lag. The input is already matched
to the preceding year in the firm’s city. It measures allocation structure
rather than the absolute amount of green finance. Ownership and
pollution-intensive industry status are crisp sets: state-owned and
pollution-intensive firms receive one, and other firms receive zero.
Pollution-intensive status is matched to annual industry codes using
the 2008 industry list for environmental verification of listed compa-
nies (General Office of the Ministry of Environmental Protection,
2008). S and D below denote firm-mean static and annual panel rep-
resentations. Table 2 summarizes definitions and calibration anchors.
2.3 Firm-mean static fsQCA
QCA examines set relations between condition combinations and
an outcome. Fiss (2007, 2011) conceptualizes organizational configu-
rations as set relations and uses fuzzy sets to represent combinations
and multiple paths to the same outcome. Calibration and analysis
follow Ragin (2008) and Schneider and Wagemann (2012). A condi-
tion’s membership m ranges from zero to one, with higher values
indicating greater membership. Negation is 1 − m; logical AND takes
the minimum; logical OR takes the maximum. Consistency assesses
TABLE 1 Sample firms.
Stock code Company Industry
000702 Zhenghong Sci-Tech C13
000860 Shunxin Agriculture C15
000876 New Hope C13
000895 Shuanghui Development C13
000998 Longping High-Tech A01
002041 Denghai Seeds A01
002286 Baolingbao C13
002311 Haid Group C13
002385 Dabeinong C13
002548 Jinxinnong C13
002557 Chacha Food C13
002567 Tangrenshen C13
002582 Haoxiangni C13
300511 Xuerong Bio-Tech A01
600108 Yasheng Group A01
600251 Guannong C13
600298 Angel Y east C14
600371 Wanxiang Denong A01
600438 Tongwei C13
600597 Bright Dairy C14
600598 Beidahuang A01
600811 Orient Group C13
600962 SDIC Zhonglu C15
601118 Hainan Rubber A01
603345 Anjoy Foods C13
603517 Juewei Food C13
A01 = agriculture; C13 = agricultural and sideline food processing; C14 = food
manufacturing; C15 = manufacture of alcoholic beverages, other beverages, and refined tea.
Company names are abbreviated English names for identification; stock codes provide the
unique reference.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 05 frontiersin.org
how closely cases belonging to a combination also belong to the high-
quality development set. Coverage assesses the share of the outcome
captured by that combination (Ragin, 2006). They therefore describe
the closeness and breadth of the set relation, respectively.
Let i denote firm, t an eligible year, k a continuous condition, and
T the number of eligible years for the firm. The firm-mean approach
first averages the raw conditions x and outcome y over each firm’s
eligible years using Equation 4 :

= =∑∑11 ,ik itk i it ttiix xy yTT (4)
Static anchors a are obtained from the cross-sectional distribution
of firm means. The direct calibration function g in Equation 5 trans-
forms these means into condition and outcome memberships.
Superscript S denotes the static representation, and subscripts k and y
identify the condition and outcome anchors:
( ) ( )= =g ;, g;S SS S ik i i yik km x a Y ya (5)
For path p, let Ap contain the conditions required to be present
and Bp those required to be absent. Membership in the path is the
minimum of these condition memberships, as shown in Equation 6 :
{ } { }( )= ∈ ∪− ∈min : 1 :SS Sip p pik ikX mk A mk B (6)
If minimization identifies P paths, membership in the whole solu-
tion is the maximum of the path memberships, as shown in
Equation 7 :
= maxSSi p ipXX (7)
Let X denote membership in a path or the whole solution and
Y membership in the outcome. Consistency C and coverage V are
calculated using Equation 8 ; ⇒ denotes the sufficiency relation
being evaluated:

( ) ( ) ( ) ( )⇒= ⇒=∑∑∑∑min , min ,C ,Vii iiii iiii XY XYXY XYXY (8)
The firm-mean approach represents each enterprise as a long-
term average case, preserving relative levels while reducing annual
variation. Averaging conditions and outcomes separately, however,
removes their annual pairing: financing constraints meeting a path
requirement in one year and favorable operating performance in
another may become characteristics of the same average case. Means
summarize long-term types, whereas annual records retain contem-
poraneous combinations.
2.4 Annual panel fsQCA and stability
decomposition
Annual panel fsQCA follows the panel set-theoretic framework
(Garcia-Castro and Ariño, 2016 ; Bhattacharya, 2023 ). In this study,
“dynamic” refers to annual condition combinations and their sta -
bility across years, not to identified causal transitions between
enterprise paths. Path sequences and transitions require trajectory-
based QCA or longitudinal configuration tracking ( Pagliarin and
Gerrits, 2020 ; Verweij and Vis, 2021 ). Firm-year observations are
retained and individually calibrated using common anchors from
the pooled panel distribution, as shown in Equation 9 . Superscript
D denotes the panel representation:
( ) ( )= =g ;, g;D DD D itk it it yitk km x a Y ya (9)
TABLE 2 Variable definitions, measurements, and calibration anchors.
Variable Definition and
measurement
Full non-membership Crossover Full membership
Development quality (HD)
Standardized composite of 15
operating-development indicators
S −0.824
D −1.666
S −0.184
D −0.035
S 1.521
D 1.551
ESG performance
Environmental, social, and
governance performance;
Huazheng score/10
S 6.754
D 6.548
S 7.329
D 7.298
S 7.867
D 8.107
Financing constraints (SA)
Signed size–age index; larger
values calibrated as high FC
S −4.022
D −4.117
S −3.825
D −3.832
S −3.492
D −3.472
Green finance
environment
City green fund share, lagged one
year
S 0.034
D 0.025
S 0.059
D 0.057
S 0.081
D 0.092
Ownership State-owned = 1; otherwise = 0 0 – 1
Pollution-intensive status
Pollution-intensive = 1;
otherwise = 0
0 –
1
Continuous variables are calibrated into sets of high HD, high ESG, strong financing constraints, and a strong green finance environment. S and D use the 5th, 50th, and 95th percentiles of the
firm-mean and annual panel distributions, respectively. HD is denoted H in the equations. SA retains negative values; larger values indicate greater membership in the defined high-SA
financing-constraints set, as explained after Equation 2. State ownership and pollution-intensive industry status are crisp sets and are not continuously calibrated.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 06 frontiersin.org
Annual path memberships and whole-solution memberships are
again calculated using the minimum and maximum operators in
Equation 10:
{ } { }( )= ∈ ∪− ∈min : 1 :DD Ditp p pitk itkX mk A mk B
= maxDDit p itpXX (10)
Let Ω denote all eligible firm-year observations. Pooled measures
use the full panel to assess the overall correspondence between the
solution and the outcome, as shown in Equation 11. Superscript P
denotes pooled evaluation:

( ) ( )
( )
( ) ( )
( )
,,
,,
min , min ,
,
it it it itit itPP
it itit it
XY XY
CV XY
∈Ω ∈Ω
∈Ω ∈Ω
= =
∑∑
∑∑

(11)
Between measures are calculated separately for each year using
Equation 12. They assess the extent to which the same solution cor-
responds to high-quality development in that year. Superscript B
denotes annual evaluation:

( ) ( )
= =∑∑∑∑min , min ,,it it it itBB iitt it itii XY XYCV XY (12)
Within measures are calculated over each firm’s eligible years
using Equation 13. They assess whether the same solution consistently
corresponds to the firm’s high-quality development across years.
Superscript W denotes within-firm evaluation:

( ) ( )
= =∑∑∑∑min , min ,,it it it itWW ttii it ittt XY XYCV XY (13)
Pooled consistency can be expressed using Equation 14 as a
weighted average of annual or within-firm consistency (Garcia-Castro
and Ariño, 2016):
= =∑∑P BB WW tt i itiC wC w C (14)
Superscripts B and W on w denote year and firm weights, respec-
tively. Each weight is the group’s sum of solution memberships divided
by the corresponding full-sample sum. Consistency is not calculated
for a group with a zero denominator.
2.5 Channels of temporal aggregation and
comparison scenarios
Temporal aggregation can change path identification through
three operations. First, firm means and annual observations have dif-
ferent distributions, changing calibration anchors and potentially
assigning different memberships to the same raw value. Second, aver-
aging and nonlinear calibration generally do not commute. With E
denoting the within-firm mean, common panel anchors generally
yield the relation in Equation 15:
( ) ( )≠ g ; g; DDt itk t itk kkEx a E x a (15)
Third, averaging before or after constructing a condition intersec-
tion changes the assessed degree of joint membership. Let z denote the
membership required by a path: m for a present condition and 1 − m
for an absent condition. As the minimum in each period is no greater
than any individual condition membership in that period, Equation 16
follows:
≥    min mink t itk t k itkEz E z (16)
The inequality is strict when the least-satisfied condition changes
across years. If financing constraints and regional green finance meet
a path’ s requirements in different years, averaging before taking the
intersection may increase the assessed degree of joint membership.
Averaging thus changes annual pairing information and the influence
of low-membership conditions on configuration assessment. In an
unbalanced panel, unequal numbers of eligible years also change firm
weights in pooled assessment.
Table 3 specifies five scenarios. S is the firm-mean static model,
and D is the annual panel model; these are the main comparison. A
averages raw values before calibration with panel anchors, providing
an aggregation contrast with D. B calibrates each observation with
panel anchors before averaging memberships, providing an order con-
trast with A. C retains D’s 306 annual observations but calibrates con-
tinuous conditions and the outcome with static anchors, providing an
anchor contrast with D. Figure 2 illustrates the workflow.
The D–A contrast tests whether aggregation still changes configu-
rations under common anchors. A–B examines the order of averaging
and nonlinear calibration, and D–C examines the calibration refer-
ence while annual records are retained. T ogether, these contrasts
clarify different sources of discrepancy. Common data evaluation then
assesses which operating states the identified solutions cover and
TABLE 3 Scenarios for comparing temporal aggregation.
Scenario Case
unit
Temporal
processing
Anchor
reference
Firm-mean static (S) 26 firms Average raw values,
then calibrate
Firm means
Annual panel (D) 306 firm-
years
Calibrate
observations without
aggregation
Annual panel
Average then
calibrate (A)
26 firms Average raw values,
then calibrate
Annual panel
Calibrate then
average (B)
26 firms Calibrate
observations, then
average memberships
Annual panel
Panel with static
anchors (C)
306 firm-
years
Calibrate
observations without
aggregation
Firm means

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 07 frontiersin.org
whether they retain close correspondence with the outcome
across years.
2.6 Truth-table rules and common-data
evaluation
The 5th, 50th, and 95th percentiles serve as anchors for full non-
membership, the crossover, and full membership, followed by direct
logistic calibration. Quantiles use R’s default linear-interpolation algo-
rithm. Memberships exactly equal to 0.5 are adjusted to 0.501 to assign
a truth-table row. In aggregated scenarios, ownership and pollution-
intensive status equal one if their within-firm share of ones is at least
0.5, including ties. Panel scenarios retain annual binary values; major-
ity assignment may therefore differ from an individual year’s struc-
tural attributes.
Proportional reduction in inconsistency (PRI) filters sufficiency
combinations that are simultaneously close to the outcome and its
negation. The baseline uses consistency 0.85, PRI 0.50, and frequency

  1. Consistency thresholds of 0.80 and 0.90 test looser and tighter
    screening; frequency is also increased to 3 at consistency 0.85. Static
    frequency counts firms, while panel frequency counts firm-years.
    Independent firm counts supporting positive rows are additionally
    checked: equal frequency cutoffs do not imply equal support from
    independent firms. Sensitivity analysis follows Skaaning (2011) and
    Oana and Schneider (2024). These thresholds screen truth-table rows;
    whole-solution consistency is calculated after path identification.
    Truth-table analysis and reporting follow configurational research
    guidelines (Greckhamer et al., 2018). Complex solutions exclude logi-
    cal remainders and minimize rows supported by observed cases.
    When equivalent minimal solutions exist, selection first ranks consis-
    tency on each solution’s own identification data, then coverage, and
    finally the software candidate number; all candidate expressions and
    evaluations are retained. Both models use the same selection rule.
    Common-data evaluation is an in-sample comparison, and D is iden-
    tified on the common annual panel itself.
    A consistency threshold of 0.90 is used for single-condition neces-
    sity (Schneider and Wagemann, 2012; Ragin, 2006). The frequency
    with which a condition appears in paths is not a criterion of necessity
    or importance. Individual paths are assessed by consistency, PRI, raw
    coverage, and unique coverage. Method comparisons use consistency,
    coverage, and the dispersion of annual and within-firm correspon-
    dence on common data.
    Specifically, the solutions from S, D, A, B, and C are evaluated
    using the same calibrated memberships for the 306 annual observa-
    tions, yielding Pooled, Between, and Within measures. They are also
    evaluated using the same memberships for the 26 firm means. S and
    D form the main comparison; A, B, and C help explain differences.
    Within means and distances for S and D are compared only among
    firms for which both solutions yield defined consistency.
    For a sequence v of annual or firm-level consistency values, the
    panel set-theoretic adjusted distance D(v; J) is calculated using
    Equation 17 (Garcia-Castro and Ariño, 2016):

( )

=

 −
=
++
∑ ∑
2
1
1
2
1
D;
32
J j
Jj
ll
v
Jv
vJ
J
JJ
(17)
J is the number of eligible years or common firms, j and l index
groups, and the elements of v are the corresponding consistency
values. D here denotes the distance function. Smaller values indicate
more similar consistency across years or firms. Stability is assessed as
FIGURE 2
Temporal aggregation, calibration order, and common-data evaluation. Static anchors come from the distribution of 26 firm means; panel anchors
come from the pooled distribution of 306 complete annual observations and are shared across years. Only continuous variables are averaged.
Aggregated scenarios assign ownership and pollution-intensive status by within-firm majority, while panel scenarios retain annual values.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 08 frontiersin.org
dispersion in whole-solution consistency within the common sample
and interpreted alongside its level, distinguishing closely aligned high
consistency from uniformly low consistency. Annual consistency plots
use 0.80 as a comparison reference.
To assess robustness of configuration identification, the study
retains PCA as the main analysis and separately examines changes in
HD weighting and the PRI screening cutoff. Equal and entropy
weighting use the same 15 baseline indicators. Equal weighting aver-
ages their Z-scores and then standardizes the composite. Entropy
weighting min-max scales each indicator across the 313 scoring
records, adds 10 −12 when calculating proportions, normalizes the
divergence coefficients of standardized entropy into weights, and stan-
dardizes the weighted score. Both checks retain each temporal repre-
sentation’s original 5th/50th/95th-percentile calibration rule,
consistency 0.85, PRI 0.50, and frequency 1. A separate parameter
check changes only the PCA PRI cutoff to 0.65; Table 4 continues to
examine consistency and case frequency. Supplementary Tables S3, S4
provide weights, all selected expressions, and comparisons.
Independent non-high-HD analysis retains baseline PCA and
defines the outcome as one minus calibrated high-HD membership
before rebuilding the truth tables. Separately, each measure’s calibrated
data are used to score the complete annual D2 expression and the
alternative-weight shared combination against high and non-high
HD. Truth tables and paths for independent non-high-HD analysis
appear in Supplementary Tables S5, S7; fixed-expression scores for
both outcome directions appear in Supplementary Table S4. Retention
under weighting and PRI checks, outcome correspondence, and case
support are considered jointly to identify annual configurations sup-
ported by these complementary checks.
2.7 Reproducibility and manuscript
preparation
The replication scripts use the same merged dataset to reproduce
HD construction, calibration, truth tables, all candidate solutions, and
common-data evaluation, including equal weighting, entropy weight-
ing, PRI 0.65, and non-high-HD analysis.
During revision of the authors’ existing manuscript, OpenAI
Codex (GPT-6) assisted with English translation, document format-
ting, and methodological checks. The authors reviewed and approved
the manuscript and take responsibility for its content.
3 Results
3.1 Calibration, empirical diversity, and
necessary conditions
Table 2 shows that firm means and annual observations imply
different calibration references. HD’s full non-membership anchor is
−0.824 for the static model and −1.666 for the panel model; full mem-
bership anchors are 1.521 and 1.551. Static endpoint anchors for ESG
and green finance are also closer to their crossover points. Long-term
means compress some annual extremes, changing relative positions in
the high and non-high sets.
Five conditions imply 32 logical combinations. S observes 17
truth-table rows and D observes all 32, indicating that annual
states of the same firms increase observable configurational diver -
sity. In Table 5, the necessity consistency of every condition and its
negation is below 0.90. The subsequent analysis therefore focuses
on how conditions combine in relation to high-quality
development.
3.2 Static and annual panel configurations
At the common consistency threshold of 0.85, the static model
identifies four paths, S1–S4 (Table 6), and the panel model identifies
six, D1–D6 (Table 7). Their combinations and economic meanings are
examined first, followed by a comparison of shared and contrasting
configurations.
S2 comprises high ESG, non-strong financing constraints, state
ownership, and non-pollution-intensive industry status. Its consis-
tency of 0.876 is the highest among the static paths. This long-term
type combines responsible governance with non-high SA. For these
enterprises, the relationship between resources and development qual-
ity should be interpreted through governance, stakeholder coordina-
tion, and sustained operations together.
S1, S3, and S4 all include non-state ownership and pollution-
intensive industry status but differ in enterprise and regional condi-
tions. S1 combines non-high ESG with a strong green finance
environment, indicating that higher composite operating quality can
coexist with a lower ESG rating. S3 combines high ESG with strong
financing constraints; its raw coverage of 0.347 is the highest among
the static paths, describing a type in which high SA, higher ESG, and
long-term operating quality coexist. Firms sharing ownership and
TABLE 4 Sensitivity to consistency and case-frequency thresholds.
Consistency
cutoff
Frequency Model Paths Pooled
consistency
Coverage Mean
within
Within
distance
0.80 1 Static 4 0.788 0.581 0.811 0.278
0.80 1 Panel 4 0.775 0.700 0.789 0.234
0.85 1 Static 4 0.788 0.581 0.811 0.278
0.85 1 Panel 6 0.858 0.572 0.849 0.185
0.90 1 Static 3 0.822 0.321 0.870 0.255
0.90 1 Panel 3 0.889 0.331 0.927 0.111
0.85 3 Static 1 0.828 0.255 0.891 0.170
0.85 3 Panel 6 0.858 0.572 0.823 0.214
PRI remains 0.50. S frequency counts firms; D frequency counts firm-years. Within-firm comparisons use 21 common firms at frequency 1 and 13 at frequency 3. Coverage is evaluated on the
common annual panel. Bold rows indicate baseline settings; the other rows are sensitivity settings.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 09 frontiersin.org
industry backgrounds therefore need not follow a single ESG or
financing pattern.
S4 combines non-strong financing constraints with a non-strong
green finance environment, with consistency 0.852 and raw coverage
0.282. A lower city green fund share and non-high SA represent
regional allocation structure and firm-level financing constraints,
respectively. Considering both distinguishes local green fund alloca-
tion from the enterprise’s own financing constraints.
Pollution-intensive status appears in three static paths and in D1,
D2, D4, and D6 below. This prevalence relates to sample composition:
16 firms and 185 annual observations belong to this category, repre-
senting a majority of the 26 firms and 306 observations. The variable
is assigned by an industry list and describes the environmental com-
pliance context, not measured emissions. Its necessity consistency is
0.716 in S and 0.673 in D. Among non-state-owned,
pollution-intensive firms, different ESG, financing, and regional com-
binations correspond to relatively high HD, revealing diverse develop-
ment types within a shared industry background.
All four static paths have case support. For S1–S4, the numbers of
firms with both path and high-HD membership above 0.5 are 3, 2, 4,
and 2, respectively. Raw coverage includes coverage shared with other
paths, whereas unique coverage captures incremental coverage relative
to the other paths; the two describe coverage breadth and unique con-
tribution. Supplementary Table S6 provides path-specific support
counts and typical cases.
D1 and D2 both include non-high ESG, strong financing con -
straints, and pollution-intensive industry status, each with consistency
0.914. D1 additionally requires a strong green finance environment,
whereas D2 requires non-state ownership. Both connect a lower rating
TABLE 5 Necessity analysis for individual conditions and their negations.
Condition S consistency S coverage D consistency D coverage
High ESG 0.695 0.675 0.732 0.685
Non-high ESG 0.582 0.592 0.605 0.659
Strong financing constraints 0.690 0.719 0.685 0.678
Non-strong financing
constraints
0.646 0.614 0.622 0.638
Strong green finance
environment
0.641 0.605 0.647 0.653
Non-strong green finance
environment
0.663 0.696 0.673 0.676
State-owned 0.317 0.409 0.357 0.455
Non-state-owned 0.683 0.552 0.643 0.535
Pollution-intensive 0.716 0.578 0.673 0.561
Non-pollution-intensive 0.284 0.367 0.327 0.416
The reference threshold for necessity consistency is 0.90.
TABLE 6 Firm-mean static configurations under baseline PCA measurement.
Condition /
fit measure
S1 S2 S3 S4
High ESG ⊗ ● ●
Strong financing
constraints
⊗ ● ⊗
Strong green
finance
environment
● ⊗
State-owned ⊗ ● ⊗ ⊗
Pollution-intensive ● ⊗ ● ●
Consistency 0.846 0.876 0.873 0.852
PRI 0.662 0.658 0.760 0.738
Raw coverage 0.222 0.124 0.347 0.282
Unique coverage 0.062 0.124 0.107 0.069
● indicates presence, ⊗ negation, and a blank an unrestricted condition. All are complex
solutions without core/peripheral coding. PRI denotes proportional reduction in
inconsistency. The baseline uses 15-indicator PCA, truth-table row consistency 0.85, PRI
0.50, and frequency 1 (firm). Supplementary Table S4 compares each path under equal
weighting, entropy weighting, and PCA PRI 0.65. Shading identifies the complete
configuration shared by static S2 and annual D5 in the baseline analysis.
TABLE 7 Annual panel configurations under baseline PCA measurement.
Condition/
fit measure
D1 D2 D3 D4 D5 D6
High ESG ⊗ ⊗ ● ● ● ●
Strong financing
constraints
● ● ⊗ ⊗ ⊗
Strong green
finance
environment
● ● ●
State-owned ⊗ ● ⊗ ● ⊗
Pollution-
intensive
● ● ● ⊗ ●
Consistency 0.914 0.914 0.885 0.847 0.902 0.860
PRI 0.797 0.804 0.612 0.678 0.645 0.688
Raw coverage 0.226 0.190 0.186 0.223 0.137 0.205
Unique coverage 0.029 0.017 0.012 0.072 0.011 0.035
● indicates presence, ⊗ negation, and a blank an unrestricted condition. All are complex
solutions without core/peripheral coding. PRI denotes proportional reduction in
inconsistency. The baseline uses 15-indicator PCA, truth-table row consistency 0.85, PRI
0.50, and frequency 1 (firm-year). Supplementary Table S4 compares each path under equal
weighting, entropy weighting, and PCA PRI 0.65. Shading identifies the complete
configuration shared by static S2 and annual D5 in the baseline analysis.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 10 frontiersin.org
and high SA with favorable operating quality in the same year.
Compared with static S3, higher annual operating quality does not
always coincide with high ESG; regional resources and ownership also
enter the combinations.
D3 and D5 both combine high ESG, non-strong financing con-
straints, and state ownership, together with a strong green finance
environment and non-pollution-intensive industry status, respec -
tively. Their consistencies are 0.885 and 0.902. These paths describe
contemporaneous combinations of responsible governance and
financing constraints. D5 is identical to S2, while D3 adds a state-
owned enterprise combination in a strong regional green finance
environment.
D4 and D6 both include high ESG, non-state ownership, and
pollution-intensive industry status, combined with non-strong financ-
ing constraints and a strong green finance environment, respectively.
D4 has the highest annual unique coverage, 0.072, while D6 has raw
coverage of 0.205. D4 emphasizes firm-level financing constraints,
whereas D6 emphasizes regional green fund allocation. Similar pro-
cessing enterprises can therefore correspond to higher annual devel-
opment quality through different condition combinations.
The difference between six panel and four static paths arises from
case representation and Boolean minimization. Annual observations
and firm means produce 32 and 17 empirically observed combina-
tions, respectively. The same thresholds retain nine and eight positive
truth-table rows, which reduce to six and four paths. Multi-year aver-
aging compresses observable combinations; annual records preserve
more contemporaneous combinations. More paths indicate greater
empirical diversity, not automatically better fit or broader coverage.
On their own identification data, static solution consistency and
coverage are 0.822 and 0.636, compared with 0.858 and 0.572 for the
annual solution. These fits use different data representations; the
methodological comparisons below use common evaluation data.
Each annual path has positive unique coverage, ranging from
0.0107 to 0.0720, with D4 being the highest. Requiring path and high-
HD membership above 0.5 and membership in each of the other five
paths no higher than 0.5, exclusive support for D1-D6 is 5/2, 4/2, 2/1,
17/6, 3/1, and 11/5 (records/firms), respectively. All six paths have
exclusive annual support; D4 has the most exclusive records, while
exclusive support for D3 and D5 is concentrated in one firm each.
Unique coverage sums to 0.1752, whereas whole-solution coverage is
0.5719, indicating shared coverage. Joint D1, D2, and high-HD mem-
bership above 0.5 involves 10 annual records from four firms.
Supplementary Table S6 reports path support, typical cases, and pair-
wise overlap.
Figure 3 compares the shared combination and financing and
ESG differences in the displayed solutions. Under baseline PCA and
screening rules, S2 and D5 share high ESG, non-strong financing con-
straints, state ownership, and non-pollution-intensive industry status.
Firm-mean and annual representations identify the same expression,
neither restricting regional green finance. This configuration
describes state-owned, non-pollution-intensive enterprises combin-
ing higher ESG and non-high SA; their assessment should jointly
consider governance performance and firm-level financing
constraints.
The financing contrast concerns S3 and D4. Both include high
ESG, non-state ownership, and pollution-intensive status, but S3
specifies strong and D4 non-strong financing constraints. They
describe high-SA and non-high-SA backgrounds in long-term and
annual combinations, respectively. D1 still includes strong financing
constraints, and D6 leaves them unspecified. The paired contrast is
therefore a classification difference within a particular combination,
not a reversal across the entire solution.
The direction of ESG differences depends on the accompany -
ing conditions. Under strong regional green finance, non-state
ownership, and pollution-intensive industry status, S1 requires
non-high ESG and D6 high ESG. Under strong financing con -
straints, non-state ownership, and pollution-intensive industry
status, S3 requires high ESG and D2 non-high ESG. These oppo -
site contrasts show that averaging changes how ESG combines
with other conditions. Annual changes in ratings, financing con -
straints, and operating performance need not coincide, so long-
term types and annual configurations can show different
ESG states.
Haid Group’s firm mean corresponds to S1 and its 2022 observa-
tion to D6, specifying non-high and high ESG, respectively. Its annual
report describes centralized procurement, production close to mar-
kets, and technical services for aquaculture farmers. In 2022, revenue
was RMB 104.715 billion, profit attributable to shareholders was RMB
2.954 billion, and net operating cash flow was RMB 6.818 billion
(Guangdong Haid Group Co., Ltd., 2023). Procurement, production,
and technical services provide operating context for supply-chain
coordination and development quality. The S1/D6 contrast suggests
identifying long-term ESG types separately from annual ratings before
examining their combinations with regional resources and operating
outcomes.
Chacha Food’s firm mean corresponds to S3 and its 2022 obser-
vation to D4: financing-set membership differs, while both have high
HD. This reflects differences in the classification of the same firm’s
financing constraints under firm-mean and annual references.
Sunflower-seed procurement includes contracts with growers, and
distributors generally pay before shipment. In 2022, revenue was
RMB 6.883 billion, profit attributable to shareholders RMB 0.976 bil-
lion, and net operating cash flow RMB 1.532 billion ( Chacha Food
Co., Ltd., 2023). Contracts and advance payments provide context for
stable operating relationships and cash collection. Long-term assess-
ment can examine accumulated governance and supply-chain rela-
tionships, while annual evaluation checks financing constraints
alongside cash collection and investment arrangements.
3.3 Calibration anchors and operation order
Table 8 examines whether calibration anchors alone explain
path differences. D and A share panel anchors but retain annual
observations and firm means, respectively. They identify six and
three paths, with no identical expressions. Common anchors there -
fore do not remove the discrepancy. This contrast also changes
annual pairing, firm weights, and the unit of empirical support.
Firm means preserve long-term relative levels, while annual
records preserve contemporaneous conditions and outcomes.
Classification based only on means may group firms with similar
long-term types but different annual conditions. Combining long-
term classification with annual checks can distinguish an estab -
lished operating foundation from conditions requiring attention in
a particular year.
At consistency 0.85, A and B identify the same three paths. On
their own identification data, consistency is 0.839 and 0.875 and cov-
erage 0.485 and 0.477, respectively. Here, the order of operations does

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 11 frontiersin.org
not change the expressions but changes firm membership in paths and
the outcome, so identical expressions can characterize enterprise per-
formance to different degrees.
C retains D’s annual records but uses static anchors and iden -
tifies three paths, showing that calibration reference also changes
path identification. Evaluating both expressions on the common
annual panel gives consistency and coverage of 0.889 and 0.331
for C, compared with 0.858 and 0.572 for D. C corresponds more
closely to the high-quality development set but covers less of it.
Changing the calibration reference therefore changes the range of
annual operating states captured by the identified
configurations.
FIGURE 3
Baseline contrasts and sensitivity evidence. (A) Compares the shared combination and financing and ESG differences; S3 appears in two contrasts.
(B) Summarizes selected complete expressions, narrower subsets, and supplementary fixed-expression fit. Supplementary Table S4A compares all ten
baseline paths, S4C reports all 23 alternative paths, and S4D reports high/non-high-HD scores and case support.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 12 frontiersin.org
3.4 Consistency and stability on common
data
Table 9 compares the two baseline solutions on common data. On
the annual panel, consistency is 0.788 for S and 0.858 for D, and cover-
age is 0.581 and 0.572. On common firm means, consistency is 0.822
and 0.873, and coverage is 0.636 and 0.612. Under baseline rules, the
annual solution corresponds more closely to the outcome, whereas the
static solution has slightly broader coverage. The models emphasize
the breadth of long-term types and the closer fit of annual combina-
tions to the outcome, respectively.
Figure 4 compares annual consistency. Median consistency is
0.802 for S and 0.874 for D. Seven and ten years, respectively, reach the
0.80 reference, and adjusted annual distances are 0.094 and 0.077.
Under baseline rules, the panel solution retains closer correspondence
in more years and shows less annual dispersion. Using the static solu-
tion to assess annual performance calls for further attention to the
conditions in lower-consistency years.
Within-consistency is defined for 21 firms under S and 23 under
D. Figure 5 compares the 21 common firms. The two points for each
firm indicate the solutions’ consistency across that firm’s annual
records, and the connecting line shows their difference. Mean
consistency is 0.811 for S and 0.849 for D; 12 and 15 firms reach 0.80;
adjusted distances are 0.278 and 0.185. Under baseline rules, the panel
solution has closer within-firm correspondence and lower dispersion
across firms.
3.5 Sensitivity analysis
3.5.1 Consistency and frequency comparisons
Table 4 compares consistency and frequency cutoffs. These cutoffs
change the truth-table rows entering minimization. Static S counts
firms, whereas annual D counts firm-years; observed combinations
and support concentration differ. The same numeric cutoff may there-
fore retain different empirical content and alter relative performance.
The PRI cutoff remains 0.50 throughout Table 4.
Table 4 tests whether comparisons persist under different screening
rules. At consistency 0.80, both models identify four paths. Panel pooled
consistency is 0.775, below static consistency of 0.788, but panel coverage
is higher; its mean within-consistency among common firms is also
lower. At 0.85 and 0.90, panel consistency is higher. At 0.90, both retain
three paths, with coverage falling to 0.321 and 0.331. A higher cutoff
requires closer correspondence between retained truth-table rows and
the outcome but reduces the operating states covered.
The operation-order comparison also depends on screening rules.
At 0.80, A and B identify three and four paths; at 0.85, the same three;
at 0.90, three each, but only two are identical. C has no positive truth-
table rows passing the 0.90 threshold, so no sufficiency solution is
reported. Expressions and coverage must therefore be considered
together when interpreting operation-order differences.
Raising truth-table row frequency to three at consistency 0.85
reduces the static model from four paths to one. The retained com -
bination comprises high ESG, high SA, non-high regional green
finance, non-state ownership, and pollution-intensive industry
status, a narrower context of baseline S3. Four firms have path
membership above 0.5, three of which also belong to high
HD. Evaluated on the common annual panel, coverage falls from
0.581 to 0.255. The four baseline static paths are sensitive to the
frequency cutoff; the higher cutoff retains a context of S3 with
more concentrated case support while reducing the operating
states covered by baseline classification. The annual solution
retains six paths and coverage 0.572. Its nine positive truth-table
rows each involve 1–6 firms; one row’s three records come from the
same firm. Under this setting, the 13 common firms have static/
annual mean within-firm consistency of 0.891/0.823 and adjusted
distance of 0.170/0.214. Thus, identical frequency values do not
imply identical independent-firm support across temporal
representations.
TABLE 8 Identification results across scenarios at consistency 0.85.
Scenario Cases Observed
rows
Positive
rows
Paths Consistency PRI Coverage
Firm-mean static (S) 26 17 8 4 0.822 0.695 0.636
Annual panel (D) 306 32 9 6 0.858 0.706 0.572
Average then calibrate (A) 26 16 6 3 0.839 0.713 0.485
Calibrate then average (B) 26 16 6 3 0.875 0.737 0.477
Panel with static anchors (C) 306 32 4 3 0.838 0.705 0.279
Consistency, PRI, and coverage are calculated on each scenario’s own identification data. Positive rows jointly satisfy consistency, PRI, and frequency rules.
TABLE 9 Evaluation of the baseline solutions on common annual and firm-mean data.
Measure Static S Panel D
Pooled consistency 0.788 0.858
Common-panel coverage 0.581 0.572
Firm-mean consistency 0.822 0.873
Firm-mean coverage 0.636 0.612
Median annual
consistency
0.802 0.874
Adjusted annual distance 0.094 0.077
Y ears with consistency ≥
0.80
7/13 10/13
Mean Within consistency
(common firms)
0.811 0.849
Adjusted Within distance
(common firms)
0.278 0.185
Common firms with
Within consistency ≥ 0.80
12/21
15/21
Annual evaluation uses 306 observations and firm-mean evaluation 26 firms. Within means
and distances use the 21 firms with defined consistency under both solutions.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 13 frontiersin.org
The baseline S and D models have two and four minimal can -
didate solutions, respectively. Across candidate combinations, D
retains higher pooled consistency and mean within-consistency,
and a lower within-distance among common firms. Candidate
selection does not reverse the consistency comparison at 0.80
either. Tables 4, 9 therefore indicate that comparative performance
changes mainly with screening rules. Application requires atten -
tion to both the evaluation scale and desired coverage.
Table 4 shows that different screening rules involve trade-offs
among outcome correspondence, coverage breadth, and case sup -
port. Under baseline rules, annual combinations correspond more
consistently to the outcome; a higher frequency cutoff concen -
trates static classification on fewer long-term types. Enterprise
assessment can combine long-term classification with annual
checks rather than judge either temporal representation by its
number of paths.
3.5.2 Sensitivity to HD weighting
Supplementary Table S4 compares equal weighting, entropy
weighting, and PRI 0.65 with every baseline path, S1–S4 in Table 6
and D1–D6 in Table 7. In compact notation, E denotes high ESG,
FC the financing-constraints set calibrated from SA, G high city
green fund share, O state ownership, P pollution-intensive indus -
try status, and ~ set negation. Equal weighting identifies 4/3
static/annual paths and entropy weighting 3/3. The three annual
expressions are identical under both alternatives and retain base -
line D2 and D5 exactly. Static solutions differ: equal weighting
adds non-high regional green finance to S2, while entropy weight -
ing does not retain that expression. Thus, the complete original
S2/D5 shared relationship is measurement-sensitive; both alterna -
tive weighting methods identify FC · G · ~ O · P as a different
shared expression.
FIGURE 4
Annual consistency of the static and annual solutions. The truth-table consistency screening cutoff is 0.85; the dashed line at 0.80 is a comparison
reference for annual whole-solution consistency.
FIGURE 5
Within-firm consistency for firms common to both evaluations. The 21 common firms are numbered in ascending stock-code order: 1 Shunxin
Agriculture; 2 New Hope; 3 Shuanghui Development; 4 Longping High-Tech; 5 Baolingbao; 6 Haid Group; 7 Dabeinong; 8 Jinxinnong; 9 Chacha Food;
10 Tangrenshen; 11 Haoxiangni; 12 Yasheng Group; 13 Angel Yeast; 14 Tongwei; 15 Bright Dairy; 16 Beidahuang; 17 Orient Group; 18 SDIC Zhonglu; 19
Hainan Rubber; 20 Anjoy Foods; 21 Juewei Food. Lines connect the two solution consistencies for each firm.

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 14 frontiersin.org
Common-annual consistency for S/D is 0.8324/0.8626 under
equal weighting and 0.8345/0.8371 under entropy weighting. Both
checks retain higher common-annual consistency and lower adjusted
distances across years and common firms for D, although the entropy-
weight consistency gap is 0.0027. On common firm means, consis-
tency is 0.8903/0.8856 and 0.9386/0.8891, respectively, favoring S on
this evaluation scale. Each comparison uses common data within the
same measure. Both weighting checks retain the annual solution’s rela-
tive advantage in common-annual consistency, while model rankings
change with the evaluation scale.
3.5.3 Separate PRI parameter check
Changing only the baseline PCA PRI cutoff from 0.50 to 0.65 still
identifies four static and six annual paths. Of the ten baseline expres-
sions, five are retained exactly, four have narrower counterparts, and
one is not retained exactly. S2 and D5 both add non-high regional
green finance, forming the shared subset E · ~ FC · ~ G · O · ~ P . Its
static/annual consistency is 0.9881/0.9746 and coverage 0.0742/0.0768;
support above 0.5 involves one firm and three annual records. Stricter
PRI screening therefore concentrates the original shared relationship
in a narrower empirical context (Supplementary Table S4; Oana and
Schneider, 2024).
3.6 Non-high HD and joint assessment of
paths
Under baseline rules, independent non-high-HD analysis identi-
fies three static and three annual paths. Static whole-solution consis-
tency, PRI, and coverage are 0.8822, 0.8055, and 0.4634, respectively;
annual values are 0.7002, 0.5231, and 0.4499. Static path consistency
ranges from 0.8652 to 0.9288 and annual path consistency from
0.6542 to 0.8120. None of the selected complete non-high expressions
is identical to a baseline high-HD path. The annual non-high solution
has weaker sufficiency support and serves as a supplementary out-
come-direction comparison. Non-high HD denotes the complement
of the relative high-HD fuzzy set (Supplementary Tables S5, S7).
Comparing baseline paths with weighting and PRI sensitivity, out-
come direction, and case support reveals robust identification of
annual D2 (~E · FC · ~ O · P) across the examined measures and PRI
settings. This combination comprises non-high ESG, high SA, non-
state ownership, and pollution-intensive industry status, without
restricting regional green finance. The expression is selected exactly
under baseline PCA, equal weighting, entropy weighting, and PCA
PRI 0.65. Figure 3B displays identification differences across settings
for D2, the baseline shared configuration, and the alternative-weight
shared configuration. High-HD consistency/PRI is 0.9136/0.8042,
0.8944/0.7535, and 0.8578/0.6566 under PCA, equal weighting, and
entropy weighting, respectively; scoring the same expression against
non-high HD yields PRI 0.1524, 0.1659, and 0.2716. Baseline D2 and
high-HD membership jointly exceed 0.5 for 14 annual records from
five firms. Exact retention across settings, stronger correspondence
with high HD, and observed case support jointly inform annual oper-
ating diagnosis for non-state-owned, pollution-intensive enterprises
(Supplementary Tables S4A, S4D1, S6A).
The alternative-weight shared expression FC · G · ~ O · P com-
prises high SA, high city green fund share, non-state ownership, and
pollution-intensive industry status. Across three measures and two
temporal representations, high-HD consistency ranges from 0.8618 to
0.9470 and PRI from 0.7187 to 0.8845; corresponding non-high PRI
ranges from 0.0349 to 0.1950. Equal and entropy weighting select this
shared expression, whereas PCA supplies supplementary fixed-expres-
sion fit rather than joint selection of the same complete path. Joint
path and high-HD membership above 0.5 involves two to three firms
in static data and five in annual data. Supplementary Table S4 reports
all scores and adverse records.
4 Discussion
Development-quality assessment should distinguish long-term
operating types from annual condition states. Panel set-theoretic
research separates pooled, annual, and within-firm relationships
(Garcia-Castro and Ariño, 2016; Bhattacharya, 2023); this study fur-
ther examines which condition information averaging preserves.
Haid’s and Chacha’s configurations and annual-report evidence pro-
vide operating context: long-term ESG or financing classifications can
differ from annual states, and supply-chain cooperation, cash collec-
tion, and investment arrangements help interpret combinations of
conditions and operating quality. Assessment can first identify owner-
ship and industry types, then jointly examine ESG, firm-level financ-
ing constraints, and regional fund allocation.
Baseline S2 and D5 contain exactly the same four conditions,
forming a shared configuration retained across temporal representa-
tions. Other pairs differ under three common conditions: S3/D2 con-
tain high and non-high ESG, respectively, under high SA, non-state
ownership, and pollution-intensive industry status; S1/D6 contain
non-high and high ESG under strong regional green finance, non-
state ownership, and pollution-intensive industry status; S3/D4 con-
tain high and non-high SA under high ESG, non-state ownership, and
pollution-intensive industry status. Of the ten baseline paths, seven
specify both ESG and SA: six pair opposite states, and S3 pairs the
same state (Tables 6, 7; Figure 3A). These opposite pairings suggest
complementarity between ESG and financing conditions at the con-
figurational level: high ESG with non-high SA and non-high ESG with
high SA correspond to distinct high-HD combinations. Comparing
long-term types and annual states further reveals the contextual dif-
ferences in these configurations.
Joint assessment further identifies annual D2, comprising non-
high ESG, high SA, non-state ownership, and pollution-intensive
industry status, as robustly identified across the examined measures
and PRI settings. The complete expression is retained under equal
weighting, entropy weighting, and PCA PRI 0.65; outcome-direction
scores and case records further support its correspondence with high
HD ( Supplementary Tables S4A,S4D1, S6A ). Equal and entropy
weighting also identify FC · G · ~ O · P as a shared combination, while
PCA data provide supplementary fixed-expression fit rather than joint
selection (Supplementary Tables S4C, S4D2). Distinguishing annual
combinations retained across settings from alternative combinations
that change with measurement provides additional evidence for
annual operating assessment beyond long-term types.
4.1 Implications for enterprise assessment
and regional support
Enterprises can first identify development contexts by ownership
and industry, use multi-year information to assess long-term types,

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 15 frontiersin.org
and check annual records to determine whether conditions hold in the
same year. For state-owned, non-pollution-intensive enterprises rep-
resented by baseline S2/D5, assessment can jointly examine gover-
nance, supply-chain coordination, and financing constraints. For
non-state-owned, pollution-intensive enterprises represented by S3/
D4, it should distinguish long-term and annual SA classifications and
consider contractual cooperation, cash collection, and upgrading
investment when assessing current operations (Tables 6, 7). Annual
D2 comprises non-high ESG, high SA, non-state ownership, and
pollution-intensive industry status and is supported jointly by weight-
ing, PRI, outcome-direction, and case evidence. For these enterprises,
annual assessment should combine ESG status with financing con-
straints and current orders, cash collection, and upgrading investment
to evaluate operating quality and identify funding needs. Enterprises
should continue improving environmental compliance and ESG while
assessing operating performance comprehensively.
Regional authorities can use long-term types to classify enterprise
needs and annual condition checks to identify current support priori-
ties. The alternative-weight shared combination FC · G · ~ O · P sug-
gests jointly considering regional green finance allocation, firm-level
financing constraints, and upgrading needs among non-state-owned,
pollution-intensive enterprises. Baseline S4 also draws attention to
operating foundations in cities with lower green fund shares. Support
arrangements can further examine orders, feasible upgrading plans,
cash collection, and financing records to align regional resource allo-
cation with specific enterprise needs.
4.2 Scope and limitations
The direct empirical scope is 26 Chinese listed agricultural value-
chain enterprises during 2010–2022, predominantly downstream
manufacturing firms. Their disclosure, governance, and financing
conditions differ from those of smaller non-listed enterprises and
firms in other institutional environments. Distinguishing long-term
types from annual states offers a transferable assessment perspective;
applying specific paths and numeric anchors to other populations
requires indicators and calibration rules suited to the target context.
5 Conclusion
Agricultural enterprise development quality corresponds to mul-
tiple governance and resource combinations. Under baseline PCA and
screening rules, S2/D5 share high ESG, non-strong financing con -
straints, state ownership, and non-pollution-intensive industry status;
non-state-owned, pollution-intensive enterprises display several ESG,
financing, and regional combinations. Annual D2, comprising non-
high ESG, high SA, non-state ownership, and pollution-intensive
industry status, is retained exactly under equal weighting, entropy
weighting, and PCA PRI 0.65, demonstrating robust identification
under the examined measurement and parameter adjustments.
Outcome-direction checks and case evidence further support its cor-
respondence with high HD. This finding informs annual diagnosis for
non-state-owned, pollution-intensive enterprises: assessment should
jointly examine current ESG, financing constraints, and operating
performance. Annual D5 is also retained exactly under all three mea-
sures. The complete S2/D5 shared expression is measurement-sensi-
tive; equal and entropy weighting instead identify FC · G · ~ O · P ,
comprising high SA, high city green fund share, non-state ownership,
and pollution-intensive industry status, as a different shared
combination.
Temporal aggregation changes some condition combinations.
Multi-year means preserve long-term relative levels but compress
annual variation and change joint membership; annual records retain
contemporaneous combinations. Baseline S3/D2 and S1/D6 show
ESG contrasts in opposite directions, while S3/D4 shows different
SA-set states. These contrasts and the opposite ESG-SA pairings
among baseline paths jointly suggest complementarity at the configu-
rational level. Differences persist under common calibration anchors,
showing that the effects of temporal aggregation do not arise solely
from anchor selection.
The two temporal representations serve complementary assess-
ment tasks. Under baseline rules, D has higher consistency on
common data and less dispersion across years and common firms,
whereas S has slightly broader coverage. Table 4 further shows how
screening rules alter trade-offs among correspondence, coverage, and
case support. Enterprise assessment can therefore use multi-year data
to characterize long-term types and annual records to examine con-
temporaneous combinations. Annual D2, supported by joint assess-
ment, and other more context-specific combinations provide
differentiated evidence for matching enterprise needs with regional
support.
Data availability statement
The data analyzed in this study is subject to the following
licenses/restrictions: the assembled file contains commercially
licensed Huazheng, CSMAR, Wind, and CNRDS data that are not
authorized for public redistribution. Source records also include
listed-company disclosures and city green finance indicators. Any
sharing of licensed records is subject to the data providers' permis -
sion and access conditions. The public Supplementary material con -
tains Supplementary Tables S1–S7 . The separately supplied
replication data and code are for confidential peer review only and
are not intended for publication as Supplementary material .
Requests to access these datasets should be directed to YW ,
25402003@suibe.edu.cn .
Author contributions
YW: Formal analysis, Writing  – original draft, Data curation,
Methodology, Software, Writing – review & editing, Conceptualization.
PX: Methodology, Supervision, Writing  – review & editing,
Conceptualization, Writing – original draft. YY: Writing – review &
editing, Funding acquisition, Validation.
Funding
The author(s) declared that financial support was received for this
work and/or its publication. This research was funded by the Jiangsu
Provincial Social Science Fund, grant number 25EYD004, and the

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 16 frontiersin.org
National Natural Science Foundation of China, Y oung Scientists Fund,
grant number 72302104.
Conflict of interest
The author(s) declared that this work was conducted in the
absence of any commercial or financial relationships that could be
construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was used in the creation
of this manuscript. During revision of the authors’ existing manu-
script, OpenAI Codex (GPT-6) assisted with English translation,
document formatting, and methodological checks. The authors
reviewed and approved the manuscript and take responsibility for its
content.
Any alternative text (alt text) provided alongside figures in this
article has been generated by Frontiers with the support of artificial
intelligence and reasonable efforts have been made to ensure accuracy,
including review by the authors wherever possible. If you identify any
issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors
and do not necessarily represent those of their affiliated organizations,
or those of the publisher, the editors and the reviewers. Any product
that may be evaluated in this article, or claim that may be made by its
manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online
at: https://www.frontiersin.org/articles/10.3389/fsufs.2026.2017893/
full#supplementary-material
References
Berg, F ., Kölbel, J. F ., and Rigobon, R. (2022). Aggregate confusion: the divergence of
ESG ratings. Rev. Financ. 26, 1315–1344. doi: 10.1093/rof/rfac033
Bhattacharya, P . (2023). How to build and analyze a panel data QCA model? A meth-
odological demonstration of Garcia-Castro and Ariño’s panel data QCA model.
Methodol. Innov. 16, 265–275. doi: 10.1177/20597991231179389
Chacha Food Co., Ltd. (2023). Annual Report 2022. Available online at: https://money.
finance.sina.com.cn/corp/view/vCB_AllBulletinDetail.
php?id=9031130&stockid=002557 (accessed October 3, 2026)
China Securities Regulatory Commission. (2012). Guidelines for the Industry
Classification of Listed Companies (2012 Revision). Announcement No. 31. Available
online at: https://www.csrc.gov.cn/csrc/c101864/c1024632/content.shtml (accessed
October 3, 2026).
Du, Y ., Liu, Q., Kim, P . H., and Li, J. (2025). Riding the waves of change: using qualitative
comparative analysis to analyze complex growth patterns in entrepreneurship.
Entrepreneurship Theory Pract. 49, 312–353. doi: 10.1177/10422587241249330
Duşa, A. (2019). QCA with R: A Comprehensive Resource. Cham: Springer.
Finn, V . (2022). A qualitative assessment of QCA: method stretching in large-N studies
and temporality. Qual. Quant. 56, 3815–3830. doi: 10.1007/s11135-021-01278-5
Fiss, P . C. (2007). A set-theoretic approach to organizational configurations. Acad.
Manag. Rev. 32, 1180–1198. doi: 10.5465/amr.2007.26586092
Fiss, P . C. (2011). Building better causal theories: a fuzzy set approach to typologies in
organization research. Acad. Manag. J. 54, 393–420. doi: 10.5465/amj.2011.60263120
Garcia-Castro, R., and Ariño, M. A. (2016). A general approach to panel data set-theo-
retic research. J. Adv. Manag. Sci. Inform. Syst. 2, 63–76. doi: 10.6000/2371-1647.2016.02.06
General Office of the Ministry of Environmental Protection. (2008). Notice on the
Industry Classification Catalogue for Environmental Verification of Listed Companies.
Available online at: https://www.mee.gov.cn/gkml/hbb/bgth/200910/t20091022_174891.
htm (accessed October 3, 2026)
Greckhamer, T., Furnari, S., Fiss, P . C., and Aguilera, R. V . (2018). Studying configura-
tions with qualitative comparative analysis: best practices in strategy and organization
research. Strateg. Organ. 16, 482–495. doi: 10.1177/1476127018786487
Guangdong Haid Group Co., Ltd. (2023). Annual Report 2022. Available online at:
https://vip.stock.finance.sina.com.cn/corp/view/vCB_AllBulletinDetail.
php?id=9064917 (accessed October 3, 2026)
Hadlock, C. J., and Pierce, J. R. (2010). New evidence on measuring financial con -
straints: moving beyond the KZ index. Rev. Financ. Stud. 23, 1909–1940. doi:
10.1093/rfs/hhq009
Huang, H., and Tang, K. (2025). Empowering rural governance through agricultural new
quality productive forces: practical issues, theoretical logic, and implementation mecha-
nisms. J. Henan Norm. Univ. 52, 17–23. doi: 10.16366/j.cnki.1000-2359.2025.06.03
Liu, T., and Zhou, H. (2022). Regional differences and dynamic evolution of high-quality
rural development in China. J. South China Agric. Univ. 21, 1–11. doi: 10.7671/j.
issn.1672-0202.2022.06.001
Ma, A., Gao, Y ., and Xing, L. (2025). The impact of ESG performance on corporate
investment efficiency: evidence from Chinese agribusiness companies. Sustainability
17:7362. doi: 10.3390/su17167362
Ministry of Agriculture, National Development and Reform Commission, Ministry of
Finance, Ministry of Land and Resources, People’s Bank of China, and State
Administration of Taxation. (2017). Guiding Opinions on Promoting Agricultural
Industrialization Consortia. Nong Jing Fa [2017] No. 9. Available online at: https://app.
www.gov.cn/govdata/gov/201710/25/413908/article.html (accessed October 23, 2026)
Ministry of Agriculture and Rural Affairs of the People's Republic of China. (2021).
Opinions on Promoting the Growth and Strengthening of Leading Agricultural
Industrialization Enterprises. Available online at: https://fgs.moa.gov.cn/flfg/202110/
t20211026_6380529.htm (accessed October 23, 2026).
Oana, I. E., and Schneider, C. Q. (2024). A robustness test protocol for applied QCA: theory
and R software application. Sociol. Methods Res. 53, 57–88. doi: 10.1177/00491241211036158
OECD/European Union/EC-JRC (2008). Handbook on Constructing Composite
Indicators: Methodology and User Guide. Paris: OECD Publishing.
Pagliarin, S., and Gerrits, L. (2020). Trajectory-based qualitative comparative analysis:
accounting for case-based time dynamics. Methodol. Innov. 13, 1–11. doi:
10.1177/2059799120959170
Qin, Z., and Pan, Y . (2022). Impact of digital inclusive finance on promoting high-
quality development of rural industries. J. South China Agric. Univ. 21, 23–33. doi:
10.7671/j.issn.1672-0202.2022.05.003
Ragin, C. C. (2006). Set relations in social research: evaluating their consistency and
coverage. Polit. Anal. 14, 291–310. doi: 10.1093/pan/mpj019
Ragin, C. C. (2008). Redesigning Social Inquiry: Fuzzy Sets and beyond . Chicago, IL:
University of Chicago Press.
Sandberg, H., Alnoor, A., and Tiberius, V . (2023). Environmental, social, and gover-
nance ratings and financial performance: evidence from the European food industry.
Bus. Strateg. Environ. 32, 2471–2489. doi: 10.1002/bse.3259
Schneider, C. Q., and Wagemann, C. (2012). Set-Theoretic Methods for the Social Sciences:
A Guide to Qualitative Comparative Analysis. Cambridge: Cambridge University Press.
Shen, Y ., and Zhang, X. (2024). Finance-driven sustainable development: the impact of
green finance on agricultural non-point source pollution and its pathways. Front.
Sustain. Food Syst. 8:1430670. doi: 10.3389/fsufs.2024.1430670
Skaaning, S. E. (2011). Assessing the robustness of crisp-set and fuzzy-set QCA results.
Sociol. Methods Res. 40, 391–408. doi: 10.1177/0049124111404818
State Council of the People’s Republic of China. (2019). Guiding Opinions on Promoting
Rural Industrial Revitalization. Available online at: https://app.www.gov.cn/govdata/
gov/201906/28/444909/article.html (accessed November 8, 2026)
Tang, H., Wu, S., and Zhang, T. (2026). Core competitiveness of Chinese listed compa-
nies: new measurement and validity tests. Syst. Eng. Theory Pract. 46, 2249–2266. doi:
10.12011/SETP2024-2260

Wu et al. 10.3389/fsufs.2026.2017893
Frontiers in Sustainable Food Systems 17 frontiersin.org
Tong, A. H., Niu, H. W ., Jiang, L., Xia, Y ., and Wang, Y . (2026). Does agricultural green
finance help reduce agricultural carbon emission intensity: an empirical analysis based
on 30 provinces in China. Sci. Rep. 16:24469. doi: 10.1038/s41598-026-52382-1
Verweij, S., and Vis, B. (2021). Three strategies to track configurations over time with
qualitative comparative analysis. Eur. Polit. Sci. Rev. 13, 95–111. doi: 10.1017/
s1755773920000375
Wang, Y ., Y an, H., and Dong, F . (2025). Research on multiple paths of agricultural digitaliza-
tion driving cultivation of new quality agricultural productivity: a dynamic QCA configura-
tion analysis. Res. Agric. Mod. 46, 623–636. doi: 10.13872/j.1000-0275.2025.0742
Xia, J., Zhang, L., and Song, Y . (2024). The impact of environmental regulatory instru-
ments on agribusiness technology innovation: a study of configuration effects based on
fsQCA. PLoS One 19:e0294662. doi: 10.1371/journal.pone.0294662
Y ang, Y ., and Zhang, P . (2021). The logic, measurement and governance of China’s high-
quality economic development. Econ. Res. J. 56, 26–42.
Zhang, X., and Xu, H. (2025). Development of new quality productivity in agricultural
enterprises: theoretical analysis and performance measurement—evidence from A-share
agricultural listed companies in China. J. Henan Normal Univ. 52, 64–71. doi: 10.16366/j.
cnki.1000-2359.2025.03.09

论文原文与出版信息

Frontiers in Sustainable Food Systems
查看论文原文

继续探索