Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Cleaner Logistics and Supply Chain 8 (2023) 100115

Contents lists available at ScienceDirect

Cleaner Logistics and Supply Chain


journal homepage: www.journals.elsevier.com/cleaner-logistics-and-supply-chain

An intelligent method for supply chain finance selection using supplier


segmentation: A payment risk portfolio approach
Kedar Shiralkar a, Arunkumar Bongale a, *, Satish Kumar a, c, Anupkumar M. Bongale b
a
Symbiosis Institute of Technology, Symbiosis International (Deemed University), Lavale, Pune 412115, Maharashtra, India
b
Department of Artificial Intelligence and Machine Learning, Symbiosis Institute of Technology, Symbiosis International (Deemed University), Lavale, Pune 412115,
Maharashtra, India
c
Symbiosis Centre for Applied Artificial Intelligence, Symbiosis International (Deemed University), Lavale, Pune 412115, Maharashtra, India

A R T I C L E I N F O A B S T R A C T

Keywords: The COVID-19 pandemic-driven financial crisis grew significant interest among firms to adopt supply chain
Supply chain finance (SCF) finance (SCF) to optimize working capital for the financial stability of the supply chain. However, it is impractical
Supplier segmentation for firms with a diverse and extensive supplier base to strategize the SCF solutions for individual suppliers by
Supplier categorization
assessing their financial risk. Hence, this study conceptualizes an intelligent method to demonstrate how supplier
Risk portfolio model
Supply chain sustainability
segmentation based on suppliers’ payment risk portfolios helps supply chain practitioners to assess suppliers’
Supplier relationship management financial risk and strategize manageable supply chain finance solutions for them. This method employs a sto­
Modern portfolio theory chastic optimization model to compute suppliers’ optimum payment risk portfolios and generate a supplier
Trade credit segmentation matrix to offer supply chain practitioners the cognitive ability to select appropriate SCF solutions
Factoring for their suppliers. The proposed method can be implemented into an AI-driven explainable recommendation
Dynamic discounting system to aid supply chain practitioners in applying smart strategic thinking in supply chain finance decision-
making.

1. Introduction and services caused serious cash liquidity issues for supply chain part­
ners, pushing them for job cuts and postponing capital investments to
In recent times, supply chain finance (SCF) solutions have gained reduce expenditures. These problems led to a significant downgrading of
significant attention to mitigate economic turmoil in the global supply business revenue for firms. Thus, the COVID-19 pandemic-related sup­
chain caused by financial distress during disruptions (Hakovirta and ply chain financial disruption emphasized the need for the evolution of
Denuwara, 2020). SCF solutions equip the supply chain businesses with supply chain finance that holistically considers the financing of all the
the working capital necessary to keep the operations running and thus transactions done in the end-to-end supply chain. The new supply chain
bring financial efficiencies in the supply chain (Pfohl and Gomm, 2009). finance strategies need to focus on long-term funding of the resources in
Perhaps SCF solutions exploit deep knowledge of supply chain relations the supply chain to improve their financial capacities to operate effi­
and dynamics to optimize financial performance and control working ciently even during disruptions. Therefore, the new SCF solutions should
capital, thus improving firms’ competitiveness (Chen et al., 2019). offer suppliers comprehensive financing options to maintain adequate
Hence, many firms use supply chain finance as a risk mitigation tool that working capital to make the supply chain resilient to future financial
provides faster payment options to make their supply chain a true value crises (Hofmann et al., 2021).
chain. Firms prominently use SCF solutions such as Trade Credit and The fundamental principle of supply chain finance is to extend a
Factoring to offer financial access to the supply chain members to firm’s financial strengths to its weak supply chain partners to improve
strengthen them during crises (Xu et al., 2018). working capital in the supply chain (Wuttke et al., 2019). Therefore to
The COVID-19 pandemic caused severe adverse economic implica­ find weak suppliers, it is essential to determine the individual supplier’s
tions to firms and their supply chain partners, leading to significant risk portfolio that reflects its overall risk by assessing its payment per­
distress in managing financial liquidity in the supply chain to facilitate formance pattern over time. Moreover, firms with a diversified supplier
demand and supply challenges. The dramatic fall in demand for goods base have challenges in formulating the supply chain finance strategy

* Corresponding author.
E-mail addresses: arunbongale1980@gmail.com (A. Bongale), ambongale@gmail.com (A.M. Bongale).

https://doi.org/10.1016/j.clscn.2023.100115
Received 2 December 2022; Received in revised form 15 May 2023; Accepted 27 June 2023
Available online 14 July 2023
2772-3909/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

for an individual supplier by determining its payment risk portfolio. the % Paid invoices means low outstanding account receivables and thus
Hence, firms need to categorize their suppliers into manageable cate­ could be a good indicator of how suppliers manage their business
gories by evaluating their payment risk portfolios. This process is called effectively and secure adequate working capital (Basu and Nair, 2012).
supplier segmentation. The firms then can develop the appropriate The “Average payment days” indicator signifies the account re­
supply chain finance strategy for each group of suppliers to make the ceivable period. Offering the extended term for the payment to a capital-
supply chain financially stable and efficient (Ali et al., 2018). In fact, one constrained supplier provides cash liquidity to ease operational expense-
exploratory research study on supply chain finance emphasized that that related challenges to steer the supply chain operations efficiently.
supplier segmentation assists firms in planning appropriate supply chain Moreover, offering trade credit with extended payment terms allows a
finance strategies for their suppliers to minimize supply chain risks firm to obtain profit maximization benefits through higher interest
associated with their suppliers (de Goeij and Steeman, 2016). Over earned on credit due to an extended payment period. Thus, it creates a
recent years, a variety of supply chain finance solutions are available for win–win situation for both firm and suppliers (Yang et al., 2021; Jing
firms to mitigate the financial risks associated with supply chain dis­ and Seidmann, 2014).
ruptions. However, the existing methods of selecting appropriate supply Therefore, building a supplier’s payment risk portfolio based on
chain finance solutions are not adequate to gain optimum benefits. This these two indicators could be helpful to determine the supplier risk
is due to the lack of a method of assessing the supplier risk portfolio aversion attitude to aid in supply chain finance decision-making.
based on suppliers’ payment behavior. Also, with a large supplier base,
firms struggle to choose appropriate supply chain finance solutions for 2.3. Selection of SCF solutions based on Supplier’s risk attitudes
different suppliers. Therefore, this study aims to conceptualize such a
method that aids firms in improving their supply chain finance decision- Over recent years, a variety of SCF solutions have been made avail­
making even with a large supplier base. Also, the suppliers’ behavior is able to firms to mitigate the impact of supply chain disruption on their
inconsistent depending on their ability to face market crisis situations. cash flow. Table 1 illustrates the choice of SCF solutions based on sup­
Therefore, the proposed method should be able to assess the variability pliers’ risk attitudes to improve supply chain performance and
in the suppliers’ behavior more accurately. Hence, to make this method profitability.
reliable, we propose to implement stochastic optimization techniques
for suppliers’ portfolio computation. 2.4. Methods to build Suppliers’ risk portfolio

2. Literature review In supply chain management, the classical newsvendor problem-


solving approach is employed to design the supplier risk portfolios.
In the literature review, we analyzed the literature on the role of However, these approaches do not consider the risk-averse attitude of
supply chain finance for supply chain risk improvement and the factors the supplier and focus on the objective of either maximizing the ex­
affecting supply chain finance decision-making. We also studied the pected profit or minimizing the expected cost (Biswas et al., 2020).
literature on the methods to assess suppliers’ risk portfolios. This is Recent Researches have widely adopted dynamic portfolio optimi­
because suppliers’ risk portfolio represents suppliers’ attitude which is zation models for supplier risk portfolio. Hosseininasab and Ahmadi
crucial for firms to financially deal with the suppliers during the supply proposed a multi-objective portfolio optimization model that de­
chain crisis. termines the supplier risk portfolio by maximizing the expected value
and minimizing their correlated risk (Hosseininasab and Ahmadi, 2015).
2.1. Role of SCF in supply chain risk mitigation Torres-Ruiz and Ravindran developed a Multi-criteria Supplier Sus­
tainability Risk Assessment Framework (SSRAF) that assesses sustain­
The complexity of modern supply chains makes the supply chain ability risk based on economic, environmental, and social risk factors
finance increasingly popular because it can provide untapped promising (Torres-Ruiz and Ravindran, 2018). Mokhtar et al. use a combined
sources of financing within inter-organizational settings to mitigate cash simulation–optimization framework for optimal selection of supplier
liquidity issues effectively. Thus, it reduces the supply chain’s financial portfolio and use indicators such as supplier’s profitability, production
risk resulting in improved firm performance and supply chain effec­ quality, and suppliers’ cost (Mokhtar et al., 2019). The main drawback
tiveness (Hofmann and Johnson, 2016; Ali et al., 2020). With acceler­ of this method is that it is more compute-intensive, and the practitioner
ated funding using SCF solutions, firms can boost production volume to has to determine the objective function precisely; otherwise, it could
meet market demands and take off the trade debt from the balance sheet, lead to an incorrect outcome. One study used the integrated Best-Worst
reflecting their financial position more robust in the balance sheet method (BWM) and K-means clustering approach to obtain different
(Dyckman, 2009). groups of alternative suppliers with similar supply risk levels (Er Kara
The digitalization of SCF solutions brings transparency in financial and Oktay Firat, 2018). The BWM method used in this study to assign
transactions in the supply chain, thus improving trust and commitment weights to the criteria does not identify the optimum risk portfolio and
among supply chain partners leading to profitability throughout the thus may affect the final decision result (Rezaei, 2015).
supply chain (Banerjee et al., 2021; Gelsomino et al., 2016). Researchers have developed the Mean–Variance-based models to
address supply chain risk issues, mainly arising from uncertain market
2.2. Indicators determining Suppliers’ risk attitude demand (Biswas et al., 2020; Liu et al., 2016; Choi and Chiu, 2012; Chiu
and Choi, 2016). However, Mean–Variance models are very sensitive to
Firms increasingly adopt SCF solutions to offer financing alternatives a small change in model inputs, significantly affecting decisions (Huang
to their suppliers but struggle to get expected benefits (Martin and et al., 2021). Furthermore, some researchers used conditional value-at-
Hofmann, 2019). Therefore, some guidance is needed for a firm to risk (CVaR) minimization models to investigate supplier’s risk-averse
choose an appropriate supply chain finance strategy for their suppliers. level based on demand and pricing-related indicators (Li et al., 2014;
Some research studies emphasized that determining the risk aversion Soleimani et al., 2014). The main limitation of the CVaR models is that
attitude of the supplier guides supply chain experts to plan SCF solutions they focus only on risk and ignore the upside potential in the asset. Keren
for their suppliers (Yang et al., 2021; Yang et al., 2021; Chen et al., and Pliskin used Expected-utility (EU) maximization model and
2019). The assessment of the supplier’s risk based on the payment demonstrated that this model provides an insightful and practical
behavior can effectively determine the supplier’s risk-aversion attitude interpretation of suppliers’ risk-averse behavior (Keren and Pliskin,
(Chiu et al., 2011). 2006). However, this method does not consider correlated risk when
The “% Paid invoices” represents account receivable volume. Higher maximizing profit. The modern portfolio theory (MPT) is a widely

2
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

Table 1
Selection of SCF Solutions based on suppliers’ risk attitudes.
SCF Name Benefits When to use? Author(s)

Trade Credit Cash liquidity. profit maximization benefits due to Capital-constrained supplier Yang et al. (2021) (Yang et al., 2021);
interest earned on trade credit with extended with moderate risk-averse Jing and Seidmann, 2014 (Jing and
payment terms. attitude Seidmann, 2014)
Dynamic competitive market pricing due to reduction in cost Cash rich supplier with risk- Gelsomino et al. (2016) (Gelsomino et al.,
Discounting of goods sold (COGS) neutral attitude 2016); Yang et al. (2021) (Yang et al.,
2021)
Partial trade credit Reduction in bad debts Capital-constrained supplier Yan et al. (2021) (Yan et al., 2021)
with bank loan with high-risk seeking attitude
Factoring “Fulfill immediate cash needs More investments in Cash rich supplier with high Yan et al. (2021) (Yan et al., 2021)
supply chain to overcome financial inefficiencies ” risk seeking attitude

adopted method in investment fields for computing investment risk- efficient and manageable to the supply chain experts.
returns portfolios. However, one research study depicts that its risk Kraljic matrix is a strategic method to assist supply chain experts to
analysis and control methods are very meaningful for controlling supply find profit maximization opportunities by segmenting suppliers into four
chain disruptions (Lao and Liu, 2007). The MPT theory attempts to classes in a 2 × 2 matrix based on purchase risk portfolio (Shiralkar
maximize portfolio expected return for a given portfolio risk level or et al., 2022). Fig. 1 illustrates the Kraljic Matrix supplier segmentation
equivalently minimize risk for a given level of expected return. Tech­ method.
nically, the MPT models consider an asset return as a normally distrib­ Kraljic matrix uses momentary state of suppliers’ behavior while
uted function, define risk as the standard deviation of return and assessing suppliers’ purchase risk portfolio. However, in real-world
generate a weighted combination of assets to determine the portfolio practice, suppliers’ attitude is highly inconsistent depending on their
returns (Iyiola et al., 2012). MPT assesses risk and returns for the ability to face changing market conditions (Hudnurkar et al., 2016).
portfolio based on the cumulative interaction between the assets. Moreover, abrupt supply chain disruptions put an additional random
Moreover, when there is a choice between low-risk and high-risk port­ financial burden on the suppliers, affecting their payment performance
folios with the same returns, MPT chooses a portfolio with a low risk and thus bringing inconsistency in their payment behavior. MPT can
(Persson et al., 2007). In the context of Supply chain, different supplier’s detect this variability in supplier behavior while assessing the supplier’s
combination leads to different efficiency. Therefore, to find an effective risk portfolio. Furthermore, using a weighted combination of assets and
supplier’s combination, it is essential for firms to describe the expected time-based compounding of returns ensures quality in optimum risk
return and risks associated with each supplier. Here, the expected return portfolio computation.
is the return that the firm obtains from the supplier and the risk refers to Therefore, we hypothesize that there is a need for two-step model
the intensity that various uncertain accidents make the supplier to which first computes the optimum risk portfolios for suppliers using
deviate from the expected return. Thus, we theoretically infer that if the historical payment performances data extending Modern Portfolio
suppliers chosen are up to a certain risk and the expected return that the Theory. Then bcay applying Kraljic Matrix approach, suppliers can be
firm gets from them is not relevant perfectly or relevant negatively, the segmented into four categories based on their optimum payment risk
risk of the supplier can be reduced, and supply chain crisis can be payment portfolios as illustrated in Fig. 1. The Supply chain managers
controlled effectively. Therefore, the model based MPT which has then can apply appropriate SCF solutions for the suppliers of each
promising capability of computing optimum risk portfolio based on category. Table 2 below summarizes the difference between the prior
expected return within acceptable risk range, can be implemented for literature and proposed model in this study against various criteria.
assessing suppliers’ payment risk portfolio more effectively.
Furthermore, the model can be extended to use the concepts of 3. Research methodology
Kraljic Matrix for categorizing suppliers based on their optimum pay­
ment risk portfolio to make supply chain finance decision making more This experimental study theorizes a supplier segmentation procedure

Fig. 1. Kraljic Matrix.

3
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

Table 2 and then generate the 2 × 2 matrix with four categories for supplier
Comparison between prior literature and proposed Model in this study. segmentation based on their optimum portfolios. Fig. 2 illustrates Flow
Criteria Prior Literature Proposed Model diagram of the proposed supplier segmentation procedure.
The analysis of reviewed literature theoretically supposes that ”%
Suppliers’ risk Partially Yes Yes
averse attitude Paid invoices” and ”Average payment days” indicators could effectively
Consideration assess supplier payment risk portfolio. Hence, we treat these indicators
Financial risk No Yes as assets for portfolio computation using the stochastic optimization
Indicators model. We subsequently discuss the different stages of the supplier
Stochastic Partially Yes Yes
techniques used
segmentation procedure. We incorporated steps 2 and 3 of the supplier
Optimum risk No Yes segmentation procedure in the stochastic optimization model for opti­
portfolio mum payment risk portfolio computation.
Computation
Model sensitivity Highly sensitive to Very less sensitive to
small changes, so small change so 3.1. Pre-processing
model reliability less model reliability
better The pre-processing step is the first step of the proposed supplier
segmentation procedure to fix any data quality issues in the input
that implements the stochastic optimization model to generate an dataset and provide standard semantics to ensure the model’s reusability
optimal payment risk portfolio for an individual supplier. The procedure with different indicators to compute optimum risk portfolio. This pro­
then uses the output of this model to define the supplier segmentation cedure assumes each performance indicator as an asset for the compu­
matrix categorizing the suppliers based on their payment risk portfolio. tation of the risk portfolio. This step also compute the correlation matrix
This experimental study uses quasi-experimental research design in for each supplier in the dataset to measure the correlation coefficients
which multiple treatments (series of optimum risk portfolios) are between two assets to determine the suppliers’ risk-averse attitude. To
applied to only one group whose members are randomly assigned for calculate the correlation, we use logarithmic changes in the returns of
what-if analysis. We discussed the different stages of the proposed assets over interval periods to ensure better accuracy and consistency in
supplier segmentation procedure in detail below. correlation computation.
We obtained the open-source payment practices dataset from Kaggle, The correlation matrix uses the cause-effect relationship and mea­
a digital community platform from Google for data scientists and ma­ sures coefficients on a scale of − 1.0 to 1.0 as shown in Eq. (1):
chine learning practitioners to develop this method. The dataset con­ Cov(rx , ry )
tains information such as supplier names, reporting period, %Paid ρxy = (1)
σx σy
Invoices, Average payment days, Standard payment days and fields
related to payment contractual terms agreed with the suppliers and Where, ρxy is Correlation between returns from the two assets X and
other business policies such as E-Invoicing, Policy coverage etc. Y. Cov(rx , ry ) is Covariance of return from asset X and Covariance of
The portfolio optimization requires optimum weight values that return from asset Y. σx is standard deviation of return from asset X. σ y is
maximize the returns while minimizing the risk. To determine these standard deviation of return from asset Y.
weights, we have enhanced the MPT-based portfolio optimization model
and introduced a new hyperparameter to generate several random risk 3.1.1. Risk portfolios simulation
portfolios. The model then chooses a portfolio that maximizes the This step is the first step of the stochastic portfolio optimization
returns with minimum risk. The procedure iteratively calls this model to model, which computes the spectrum of the supplier’s payment risk
compute optimum payment risk portfolios for all suppliers in the dataset portfolios using random weights. We discuss the numerical approach we
used for this model below.

Fig. 2. Flow diagram of Supplier Segmentation Procedure..

4
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

3.1.2. Portfolio expected returns


The expected returns refers to the profit or loss anticipated by a firm min (w)T .Covportfolio .w
from a particular supplier portfolio. To mathematically compute this, we subject to (w)T .μ⩾μ.Target (6)
(w)T .1 = 1
define Rp as a random variable to represent the return of a portfolio. Let
wi ⩾0
Ri be a random variable that represents the return of the i-th asset in the
portfolio. Let wi be the “weight” of the i-th asset in the portfolio. All The first line in the objective function represents that the objective is
weights are real numbers between 0 and 1, and they all sum up to 1. to minimize the portfolio volatility and therefore minimize the risk. The
Hence, Rp is a weighted sum of all returns Ri, with weights wi and can be constraints imply that the returns have to be greater than a specific
determined by Eq. (2). target return, the sum of all the weights should be 1, and weight values
∑ should not be negative.
Rp = w i Ri (2) To select optimum risk portfolio, we choose Maximum Sharpe Ratio
i
∑ method. Mathematically, the Sharpe Ratio (Franco-Laverde et al., 2012)
Where, wi ∈ [0, 1]∀i , i wi = 1. is defined as per Eq. (7):
Using the linearity of expectation, the expected return for the port­
Rp − Rf
folio is given by Eq. (3): SharpeRatio = (7)
∑ σp
Rp = wi E(Ri ) (3)
i
Where, Rp is return of portfolio, Rf is risk-free rate, and σp is standard
deviation of the portfolio.
Next, we calculate percentage change in both these assets from The procedure iteratively executes Step 3.1.1 and Step 3.2 for each
period i to i +1 and take log of percentage change as the log of the supplier in the input dataset to compute optimum financial risk
returns is time additive and provide reliable and quality results due to portfolio.
continuous compounding of portfolio asset returns.
Thus, the returns from an asset from period i to i +1 can be expressed 3.3. Supplier segmentation matrix definition
as shown in Eq. (4),
( )
Pi+1 − Pi This is the final step in the proposed supplier segmentation proced­
Ri = log × 100 (4) ure. This step generates a four-quadrant matrix for supplier segmenta­
Pi
tion based on suppliers’ optimum payment risk portfolios. The X-axis of
where, P denotes the asset value for ith interval period. the matrix represents Returns, and Y-axis represents the associated Risk.
To estimate E(Ri ) expected total returns for each individual asset Supplier segmentation is a longitudinal process and assigning the
over historical data, we need to calculate mean of logarithmic returns for supplier into the appropriate category could change depending upon the
each interval period and multiply it by the number of continuous in­ evolution of its performance over time. Hence, the supplier segmenta­
terval periods of reporting assets. Here, we assume that returns for each tion matrix should be redefinable to put suppliers into relevant segments
interval period are independent and identically distributed. for effective decision-making. In practice, the decision-making method
using the performance measure follows the normal distribution char­
3.1.3. Portfolio volatility acteristic (Fu et al., 2020). Hence, we assume that the distribution of
Portfolio Volatility is also termed as Portfolio Risk. To determine the suppliers’ risk portfolios follows a normal distribution. Therefore, to
Portfolio Volatility, we calculated the covariance matrix to understand define the coordinates of the matrix’s origin, we use the mean of opti­
how different assets behave with respect to each other. The Covariance mum returns values of all suppliers as X-coordinate and mean of asso­
matrix is then multiplied with weight vectors to compute the portfolio ciated risk values of all suppliers Y-coordinate. The optimum risk
variance. The square root of the variance calculated determines stan­ portfolio of each supplier is represented in the matrix by the return value
dard deviation. This standard deviation is called as Portfolio Volatility. of the optimum portfolio RP as X-Coordinate and the risk value of the
Mathematically, the portfolio variance is computed using Eq. (5), optimum portfolio σ2p as Y-coordinate. Thus, we ensure that the supplier
(Iyiola et al., 2012). segmentation matrix is redefinable with the changing suppliers’ per­
formances and therefore helps to make the supplier segmentation
σ 2portfolio = (w)T .Covportfolio .w (5)
decision-making more effective and reliable.
Where, Covportfolio is the covariance matrix of the portfolio. w is the To mathematically compute the coordinates of supplier segmenta­
vector of weights allocated to each asset in the portfolio. tion matrix origin, consider O(x, y) as origin of Matrix. Thus, the co­
Another purpose of this step in the model is to find the efficient part ordinates x and y of the matrix origin are defined by the Eq. (8).
of the risk-return spectrum that maximizes the returns for a given level ( )
1∑ n
1∑ n
of risk or minimizes risk for a given expected return level. Hence, it is O(x, y) = Rpi , σ2pi (8)
essential to find the optimal weights of assets. The model generates n i=1 n i=1
random weights based on nump ortfolios hyperparameter and simulates
Where, n is Number of suppliers, Rpi is return of optimum portfolio of
risk portfolios using those weights to compute the optimal weights. This
each supplier, and σ2pi is risk of optimum portfolio of each supplier.
step also incorporates the logic to visualize the trend of supplier’s risk
portfolios with different weights.
4. Results and discussion
3.2. Optimum risk portfolio selection
We transformed the proposed supplier segmentation procedure into
a computer-assisted intelligent method using Python. The automated
This is the optimization step of the stochastic optimization model. To
method uses the Python Pandas library to prepare the input dataset for
obtain the optimum risk portfolio for a supplier from a set of risk port­
numerical analysis. Furthermore, this method uses Python’s NumPy li­
folios generated in Section 3.1.1, we use the objective function (6).
brary to carry out the numerical steps of the stochastic portfolio opti­
mization model and supplier segmentation matrix definition. This
method then generates rich graphs and charts to visualize the outputs of
different stages of the supplier segmentation procedure using Python’s
matplotlib and mplcursors library.

5
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

We used historical payment practices data for 20 suppliers to portfolio. Furthermore, we discuss how this supplier segmentation ma­
implement this method. We choose the value of hyperparameter trix and portfolio metrics of each supplier provide cognitive interpre­
num_portfolio as 100 to compute the supplier’s optimum payment risk tation about the supplier behavior and its risk-aversion attitude so that
portfolio. Fig. 3 below illustrates the relationship graph produced by the supply chain managers can apply an effective supply chain financing
method indicating the correlation patterns between two indicators strategy to their suppliers.
observed for suppliers in the dataset. Each indicator is termed as a Low Returns & Low Risk Category: The method terms this category of
feature in this relationship graph. The scatter plot shows the relationship suppliers as Basic. Suppliers belonging to this category give low returns
between the two features, and the diagonal plot showcases the histo­ for a given risk level. Also, the risk associated with these suppliers to
gram to visualize a feature’s probability distribution. achieve the targeted returns is low. Thus, the suppliers from this cate­
The analysis of Fig. 3 implies that for suppliers with positive corre­ gory have a low potential to add value to improve supply chain per­
lation, a strong correlation exists between two indicators. Thus, we formance. The literature review indicated that suppliers from this
deduce that extending the payment terms for these suppliers signifi­ category have limited financial impact on the supply chain. Hence, firms
cantly increases the account receivables. On the other hand, for sup­ should work on reducing the supply chain risk jointly to minimize losses.
pliers with negative correlation, the correlation between two indicators The analysis of the optimum risk portfolio of each supplier of this
was found to be strong for some suppliers and weak for others. Thus, it category indicates that the method has assigned more weightage to the
indicates that for these suppliers, extending the payment terms may not Average Payment Days indicator in their optimum risk portfolio. Thus, it
increase the account receivables. means that suppliers from this category perform well when the firm
The algorithm of the stochastic optimization model generates 100 extends payment terms to them. Therefore, we imply that the suppliers
random risk portfolios for each supplier because the value of the from this category could be capital-constrained but also have a low
num_portfolio hyperparameter is configured to 100. Fig. 4 illustrates the potential to increase the firm’s performance.
risk portfolio spectrum for SUPPLIER20 in the dataset for model High Returns & Low Risk Category: The method terms this category of
outcome analysis purpose. The red star represents the minimum vola­ suppliers as Leverage. Suppliers belonging to this category give high
tility portfolio, and the green star represents the optimum volatility returns for a given risk level. Moreover, the risk associated with these
portfolio for the supplier. suppliers to achieve the targeted returns is low. Thus, the suppliers from
The interpretation of Fig. 4 indicates that the optimum risk portfolio this category are the best value providers. These suppliers have the
of SUPPLIER20 provides a 32% return, and the corresponding risk potential to handle expected disruptions. The literature review analysis
associated with this supplier is 51%. The corresponding weights portrays that firms should emphasize reducing the cost by taking
assigned to obtain the optimum portfolio for SUPPLIER20 would be advantage of the competitiveness of the supplier. Firms may invest in
0.998 for the % Paid invoices indicator and 0.0112 for the Average those suppliers to mitigate their short-term cash needs and exploit their
payment days indicator. Thus, to maximize the returns from SUP­ strengths to reduce supply chain bottlenecks.
PLIER20, the firm should give more attention to find alternatives to The analysis of the optimum risk portfolio of each supplier from this
increase the account receivables instead of extending payment terms. category shows that the method has assigned more weightage to the
The proposed method iteratively executes the stochastic optimiza­ Average Payment Days indicator for certain suppliers. In contrast, for
tion model to compute the optimum payment risk portfolio for each other suppliers, more weightage is given to the % Paid invoices indicator
supplier in the input dataset and generates the supplier segmentation in their optimum risk portfolio. Thus, it implies that suppliers from this
matrix using the logic explained in the Supplier Segmentation Matrix category with higher weightage to the Average Payment Days indicator
definition step of the proposed supplier segmentation procedure. The could be capital-constrained and therefore provide higher returns on the
four quadrants of the supplier segmentation matrix are programmed to extension of the payment term. The suppliers from this category with
confirm with Kraljic Matrix nomenclature, as portrayed in Fig. 1. The higher weightage to the % Paid invoices indicator signifies that these
method also shows the relevant metrics of each supplier’s optimum risk suppliers have adequate cash liquidity and have a high potential to
portfolio using tooltips. improve the firm’s performance.
Fig. 5 below illustrates how the intelligent method categorizes the Low Returns & High Risk Category: The method terms this category of
suppliers into four segments by assessing their optimum payment risk suppliers as Bottleneck. Suppliers in this category give low returns for a

Fig. 3. Relationship graph showing the correlation patterns between two indicators for suppliers.

6
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

Fig. 4. Supplier Risk portfolio spectrum.

Fig. 5. Supplier Segmentation Matrix.

given risk level. Besides, the risk associated with these suppliers to suppliers as Strategic. Suppliers falling under this category give high
achieve the targeted returns is high. Thus, the suppliers from this cate­ returns for a given risk level. Moreover, the risk associated with these
gory have a low potential to add value to improve supply chain per­ suppliers to achieve the targeted returns is also high. The suppliers from
formance. These suppliers usually have quality and service issues and this category ensure long-term profitable growth of the supply chain
hence could adversely impact supply chain performance and resiliency. business and are critical for competitive advantage. The literature re­
As per the literature review, firms need to focus on cost and risk mini­ view emphasized exploiting these suppliers’ strengths on product and
mization and develop new suppliers to replace them. process innovation to eliminate bottlenecks for the long-term sustain­
The analysis of the optimum risk portfolio of each supplier from this ability of the supply chain. Firms should also focus on profit maximi­
category implies that the method assigns higher weightage to the % Paid zation leveraging the competitive advantage of these suppliers.
invoices indicator. Thus, it indicates that these suppliers could have a Analyzing the optimum risk portfolio of each supplier from this
risk-averse attitude. Besides this, because of quality issues with their category implies that the method assigns higher weightage to the % Paid
services, they could potentially be detrimental to the firm’s supply chain invoices indicator. Thus, it indicates that these suppliers have adequate
business in terms of stability and efficiency. . working capital and the potential to be the strategic partner for the firm
High Returns & High Risk Category: The method terms this category of for long-term benefits.

7
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

The analysis of the supplier segmentation matrix further indicates Table 3


the firm’s level of exposure to risk. For example, the analysis of the Recommended SCF Solutions For Different Supplier Segmentation Categories.
supplier segmentation matrix illustrated by Fig. 5 implies that the input Segmentation Recommended SCF Why?
dataset used in this study has a significant number of suppliers falling Category Solution
under the Bottleneck category. On the contrary, the number of suppliers Basic Partial Trade Credit Capital-constrained
in the Strategic category is significantly less. Thus, it infers that the firm with bank loan suppliers with risk-
has a high level of exposure to financial risk. Therefore, supply chain averse attitude. Hence,
experts need to prioritize exploring new suppliers and promoting new minimize bad debts by
insisting suppliers to
ways to develop more strategic partners. opt for partial bank loan
and provide partial
4.1. Validation of results credit to meet its
immediate cash needs.
Leverage Trade credit for Profit maximization
The stochastic optimization model used in the proposed method is
capital - constrained benefit by offering
validated for robustness using parameter sensitivity analysis. For para­ supplier trade credit for capital-
metric sensitivity analysis, we selected one supplier from each category constrained suppliers.
of the supplier segmentation matrix illustrated in Fig. 5 for sensitivity Factoring for Leverage investments in
analysis. The sensitivity analysis is performed by generating the opti­ suppliers with the supply chain to
adequate working maintain adequate cash
mum payment risk portfolios of the four suppliers by using different capital flow with the working
values for the num_portfolio hyperparameter to validate the consistency capital available with
of the model. We used systematic sampling to select values for the the cash-rich suppliers.
num_portfolio hyperparameter to ensure randomization in optimum Bottleneck Dynamic Discounting Minimize the bad debts
by encouraging
payment risk portfolios computation for each supplier. We chose a
suppliers for early
scatter plot to give direct visual indication of sensitivity. Fig. 6 illustrates payment by offering
the seaborn scatter plot portraying the sensitivity of the stochastic discounts.
optimization model. Strategic Factoring Cash-rich suppliers who
The systematic analysis of the scatter plot in Fig. 6 indicates that the are ready to take risk in
the business and hence
model is reasonably robust to provide suppliers’ optimum payment risk encourage them by
portfolios consistently with different values of num_portfolio giving discounts on
hyperparameter. early payments.

5. Managerial implications
to varying operational expenses caused by changing market conditions.
Hence, supply chain managers need revaluation of suppliers’ financial
Recent COVID-19 pandemic-driven supply chain disruption has
risk regularly so that they can act promptly to minimize firm’s losses.
pushed supply chain experts to rethink their supply chain financing
The proposed intelligent method can recompute suppliers’ risk exposure
strategies to secure the financial strength of the supply chain for future
when desired by supply chain managers so that they can rethink their
crises.Table 3 demonstrates how the proposed method provides intelli­
supplier relationship strategy to ensure supply chain efficiency and
gence to supply chain managers in selecting appropriate SCF finance
resiliency.
solutions for different supplier groups.
With the realization of the proposed intelligent method into practice,
In real-world practice, the supplier’s financial health fluctuates due

Fig. 6. Scatter plot portraying sensitivity of the stochastic optimization model.

8
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

the scope of work of supply chain managers is limited to business-centric 6.1. Limitations and future scope of research
activities as below:
The proposed method being a mathematical model has a limitation
• Measure the effectiveness of existing SCF solutions on overall supply in that it may not detect unusual situations which cannot be predicted
chain performance. using historical data. Therefore, it may not be able to produce accurate
• Rethink supply chain finance strategies that holistically consider insights for effective supply chain finance decision-making all the time.
end-to-end financial transactions within the supply chain and rede­ In future studies, this method can be extended to include more cat­
sign the existing SCF solutions based on those considerations. egories into the supplier segmentation matrix and more indicators for
• Define thresholds for weights of payment performance indicators to risk portfolio computation to make it more inclusive. Furthermore, we
assign appropriate SCF solutions to a sub-group of suppliers within observe the randomness in the variation of optimum risk portfolio for
each group of supplier segmentation matrix. some suppliers using this method, which requires more investigation in
• Define criteria for level of engagement for each supplier group and future studies.
work on building appropriate strategies with them to improve
overall supply chain coordination and performance.
Declaration of Competing Interest
• Intermittent quality check to validate the consistency in the perfor­
mance of the recommendation system and retrain the system if
The authors declare that they have no known competing financial
required by reconfiguring the value of num_portfolio
interests or personal relationships that could have appeared to influence
hyperparameter.
the work reported in this paper.
On the other hand, implementing an intelligent method into an AI-
Data availability
driven recommendation system will automate below tasks.

No data was used for the research described in the article.


• Connect to source data systems to gather historical data of suppliers’
payment practices
• Pre-processing step to integrate and validate the payment practices References
data of suppliers for accuracy and completeness.
Ali, Z., Gongbing, B., Mehreen, A., 2018. Predicting supply chain effectiveness through
• Optimum payment risk portfolio computation for each supplier using supply chain finance: evidence from small and medium enterprises. Int. J. Logist.
simulations to compute payment risk portfolio spectrum for each Manage.
supplier and choose optimum payment risk portfolio. Ali, Z., Gongbing, B., Mehreen, A., Ghani, U., 2020. Predicting firm performance through
supply chain finance: a moderated and mediated model link. Int. J. Logist. Res. Appl.
• Aggregate all suppliers’ optimum payment risk portfolios and 23, 121–138.
compute supplier segmentation matrix to group suppliers based on Banerjee, A., Lücker, F., Ries, J.M., 2021. An empirical analysis of suppliers’ trade-off
their portfolios. behaviour in adopting digital supply chain financing solutions. Int. J. Oper. Prod.
Manage. 41, 313–335.
• Provide recommendations with reasoning for SCF solution to each Basu, P., Nair, S.K., 2012. Supply chain finance enabled early pay: Unlocking trapped
supplier group or sub-group of suppliers within a group based on the value in b2b logistics. Int. J. Logist. Syst. Manage. 12, 334–353.
weights assigned to payment performance indicators within the Biswas, I., Adhikari, A., Biswas, B., 2020. Channel coordination of a risk-averse supply
chain: a mean–variance approach. Decision 47, 415–429.
portfolio of each supplier. Chen, X., Liu, C., Li, S., 2019. The role of supply chain finance in improving the
• Redefine supplier segmentation matrix and recommendations when competitive advantage of online retailing enterprises. Electron. Commer. Res. Appl.
new payment practices data is added or existing data is updated to 33, 100821.
Chen, Z., Yuan, K., Zhou, S., 2019. Supply chain coordination with trade credit under the
improve the effectiveness of supply chain financing.
cvar criterion. Int. J. Prod. Res. 57, 3538–3553.
Chiu, C.H., Choi, T.M., 2016. Supply chain risk analysis with mean-variance models: A
Thus, the proposed intelligent method automates more complex and technical review. Ann. Oper. Res. 240, 489–507.
Chiu, C.H., Choi, T.M., Li, X., 2011. Supply chain coordination with risk sensitive retailer
repetitive tasks of supplier chain managers and offers necessary intel­
under target sales rebate. Automatica 47, 1617–1625.
lectual thinking to let them work on business-centric innovations to Choi, T.M., Chiu, C.H., 2012. Mean-downside-risk and mean-variance newsvendor
make the supply chain more efficient and profitable. models: Implications for sustainable fashion retailing. Int. J. Prod. Econ. 135,
552–560.
Dyckman, B., 2009. Integrating supply chain finance into the payables process.
6. Conclusion J. Payments Strategy Syst. 3, 311–319.
Er Kara, M., Oktay Firat, S.Ü., 2018. Supplier risk assessment based on best-worst
This research study contributes to the industry and research litera­ method and k-means clustering: a case study. Sustainability 10, 1066.
Franco-Laverde, J., Littlewood, A., Ellis, C., Schraner, I., Varua, M.E., 2012. Fmcg
ture by conceptualizing the intelligent method for selecting supply chain portfolio budget allocation to price promotions using modern portfolio theory (mpt).
finance solutions using supplier segmentation. Thus, it enables supply Int. Rev. Bus. Res. Papers 8, 16–30.
chain managers to concentrate on more business-centric innovations to Fu, S., Qu, X.L., Xiao, Y.Z., Zhou, H.J., Fan, G.B., 2020. Risky multi-attribute decision-
making method based on the interval number of normal distribution. Symmetry 12,
improve supply chain efficiency and profitability, bringing significant 264.
positive managerial implications. Gelsomino, L.M., Mangiaracina, R., Perego, A., Tumino, A., 2016. Supply chain finance: a
The stochastic optimization and simulation model used in the pro­ literature review. Int. J. Phys. Distrib. Logist. Manage.
de Goeij, C., Steeman, M., 2016. An exploratory study into supply chain finance: the
posed conceptual method detects uncertainties in suppliers’ payment
relevance of supplier segmentation. Tijdschrift voor toegepaste logistiek 1, 123–139.
behavior more efficiently and thus improves quality of supply chain Hakovirta, M., Denuwara, N., 2020. How covid-19 redefines the concept of
finance decision-making. This research study also emphasizes that sustainability.
Hofmann, E., Johnson, M., 2016. Supply chain finance–some conceptual thoughts
supplier segmentation and supply chain financing are logically insepa­
reloaded. Int. J. Phys. Distrib. Logist. Manage. 46, 1–8.
rable. Perhaps supplier segmentation makes the supply chain finance Hofmann, E., Templar, S., Rogers, D., Choi, T.Y., Leuschner, R., Korde, R.Y., 2021.
decision-making more logical, efficient, and manageable for supply Supply chain financing and pandemic: Managing cash flows to keep firms and their
chain managers. Besides this, supplier segmentation empowers supply value networks healthy. Rutgers Bus. Rev. 6, 1–23.
Hosseininasab, A., Ahmadi, A., 2015. Selecting a supplier portfolio with value,
chain managers to identify suppliers who are vital to them so that they development, and risk consideration. Eur. J. Oper. Res. 245, 146–156.
give appropriate attention to improve engagement for superior supply Huang, R., Qu, S., Yang, X., Xu, F., Xu, Z., Zhou, W., 2021. Sparse portfolio selection with
chain performance. uncertain probability distribution. Appl. Intell. 51, 6665–6684.
Hudnurkar, M., Rathod, U., Jakhar, S.K., 2016. Multi-criteria decision framework for
supplier classification in collaborative supply chains: Buyer’s perspective. Int. J.
Product. Performance Manage.

9
K. Shiralkar et al. Cleaner Logistics and Supply Chain 8 (2023) 100115

Iyiola, O., Munirat, Y., Nwufo, C., 2012. The modern portfolio theory as an investment Rezaei, J., 2015. Best-worst multi-criteria decision-making method. Omega 53, 49–57.
decision tool. J. Account. Taxation 4, 19–28. Shiralkar, K., Bongale, A., Kumar, S., 2022. Issues with decision making methods for
Jing, B., Seidmann, A., 2014. Finance sourcing in a supply chain. Decis. Support Syst. 58, supplier segmentation in supplier relationship management: A literature review.
15–20. Mater. Today: Proc. 50, 1786–1792.
Keren, B., Pliskin, J.S., 2006. A benchmark solution for the risk-averse newsvendor Soleimani, H., Seyyed-Esfahani, M., Kannan, G., 2014. Incorporating risk measures in
problem. Eur. J. Oper. Res. 174, 1643–1650. closed-loop supply chain network design. Int. J. Prod. Res. 52, 1843–1867.
Lao, B., Liu, N., 2007. Application of portfolio theory in controlling supply chain crisis. Torres-Ruiz, A., Ravindran, A.R., 2018. Multiple criteria framework for the sustainability
In: Integration and Innovation Orient to E-Society Volume 1: Seventh IFIP risk assessment of a supplier portfolio. J. Clean. Prod. 172, 4478–4493.
International Conference on e-Business, e-Services, and e-Society (I3E2007), October Wuttke, D.A., Rosenzweig, E.D., Heese, H.S., 2019. An empirical analysis of supply chain
10–12. Springer, Wuhan, China, pp. 9–16. finance adoption. J. Oper. Manage. 65, 242–261.
Li, B., Chen, P., Li, Q., Wang, W., 2014. Dual-channel supply chain pricing decisions with Xu, X., Chen, X., Jia, F., Brown, S., Gong, Y., Xu, Y., 2018. Supply chain finance: A
a risk-averse retailer. Int. J. Prod. Res. 52, 7132–7147. systematic literature review and bibliometric analysis. Int. J. Prod. Econ. 204,
Liu, M., Cao, E., Salifou, C.K., 2016. Pricing strategies of a dual-channel supply chain 160–173.
with risk aversion. Transp. Res. Part E: Logist. Transp. Rev. 90, 108–120. Yan, N., Xu, X., Huang, W., 2021. Supplier’s capacity investment strategy with factoring
Martin, J., Hofmann, E., 2019. Towards a framework for supply chain finance for the finance. Int. J. Prod. Econ. 238, 108149.
supply side. J. Purchas. Supply Manage. 25, 157–171. Yang, C., Fang, W., Zhang, B., 2021. Financing a risk-averse manufacturer in a pull
Mokhtar, S., Bahri, P.A., Moayer, S., James, A., 2019. Supplier portfolio selection based contract: early payment versus retailer investment. Int. Trans. Oper. Res. 28,
on the monitoring of supply risk indicators. Simul. Model. Pract. Theory 97, 101955. 2548–2580.
Persson, J., Lejon, C., Kierkegaard, K., 2007. Practical application of modern portfolio Yang, H., Zhuo, W., Shao, L., Talluri, S., 2021. Mean-variance analysis of wholesale price
theory. Unpublished Bachelor’s Degree thesis. contracts with a capital-constrained retailer: Trade credit financing vs. bank credit
Pfohl, H.C., Gomm, M., 2009. Supply chain finance: optimizing financial flows in supply financing. Eur. J. Oper. Res. 294, 525–542.
chains. Logist. Res. 1, 149–161.

10

You might also like