Sourcing Economics and Supplier Selection Analytics

Introduction

Supplier selection is an economic decision, not merely a purchasing exercise. A quoted unit price is only one component of the cost incurred by an organisation. Freight, duties, financing, inspection, inventory, quality failures, exchange-rate movements, supplier development, disruption and end-of-life disposal may materially change the economic ranking of suppliers. Consequently, professional sourcing decisions require an integrated analysis of cost, value, risk and strategic fit.

Sourcing economics provides the analytical foundation for comparing supply alternatives. It combines total cost of ownership (TCO), landed-cost modelling, cost-structure analysis, learning curves, price-index analysis and scenario modelling. Supplier selection analytics then extends this economic analysis by incorporating qualitative and strategic criteria through multicriteria decision-making (MCDM).

The central managerial principle is that the preferred supplier is not necessarily the supplier offering the lowest quoted price. It is the supplier that provides the best risk-adjusted value over the relevant decision horizon, subject to technical, ethical, legal and operational constraints.

Learning Objectives

By the end of this chapter, the reader should be able to:

1. Distinguish quoted price, landed cost, purchase cost and total cost of ownership.
2. Construct a landed-cost and TCO model using consistent units and assumptions.
3. Interpret supplier cost breakdowns, learning curves and price indices.
4. Apply weighted-scoring and analytic hierarchy approaches to supplier selection.
5. Conduct scenario and sensitivity analysis and identify decision-critical variables.
6. Evaluate supplier choices using economic, operational, ethical and governance criteria.

1. The Economic Architecture of a Sourcing Decision

A sourcing decision normally has four analytical layers:

The layers should not be added indiscriminately. For example, a cost already included in the supplier's delivered price must not be counted again as freight. Similarly, taxes that are recoverable should generally be excluded from economic cost, whereas non-recoverable taxes should be included.

The analysis should begin by defining:

• the decision horizon;
• the unit of comparison, such as one usable component or one production-ready kilogram;
• the currency and exchange-rate convention;
• demand volume and quality requirements;
• relevant cost categories;
• treatment of taxes, financing and salvage;
• constraints that disqualify a supplier regardless of score.

A supplier offering components at €10 each is not directly comparable with one offering €10.30 if the first has a 5% defect rate and the second has a 0.5% defect rate. The economically relevant comparison may be cost per conforming component rather than cost per purchased component.

2. Landed Cost Modelling

Landed cost measures the cost of acquiring and physically receiving goods at a specified destination. A general model is:

where:

• LC = landed cost per shipment or unit;
• P = supplier invoice price;
• F = freight and transportation cost;
• I = cargo insurance;
• D = customs duty and non-recoverable import taxes;
• H = port, terminal and handling charges;
• C = customs brokerage and documentation costs;
• X = other destination costs;
• R = rebates or credits attributable to the purchase.

All terms must be expressed in the same currency and for the same quantity. If a percentage duty is applied to customs value, the customs-value basis must be specified. For a duty rate d, where 0 ≤ d ≤ 1,

where V_c is the legally defined customs value. It may differ from the invoice price if assists, royalties, packing or related-party adjustments are included under applicable customs rules.

Worked Example 1: Landed Cost

An organisation imports 2,000 units at £18 per unit. Freight is £1,400, insurance is £120, port and handling charges are £600, and brokerage is £280. Customs duty is 6% of invoice value. VAT is recoverable and is excluded. There is a rebate of £300.

Invoice value:

P = 2,000 × £18 = £36,000

Duty:

D = 0.06 × £36,000 = £2,160

Total landed shipment cost:

LC = 36,000 + 1,400 + 120 + 2,160 + 600 + 280 − 300 = £40,260

Landed cost per unit:

£40,260 ÷ 2,000 = £20.13

The supplier's £18 quotation therefore understates the acquisition cost by £2.13 per unit, or 11.8% of the invoice price.

3. Total Cost of Ownership

TCO extends landed cost across the complete economic relationship. A useful general expression is:

where:

• C_acq = acquisition cost;
• C_log = logistics and landed-cost additions;
• C_inv = inventory carrying and pipeline inventory cost;
• C_qual = inspection, defects, rework, returns and warranty costs;
• C_oper = operating, tooling, maintenance or implementation costs;
• C_admin = transaction and supplier-management costs;
• C_risk = expected cost of disruption and other risk exposures;
• C_end = end-of-life, disposal or switching costs;
• V_res = residual value, credits or recoveries.

The model is not a universal checklist. Categories must be tailored to the purchase. For a standard catalogue item, engineering support may be immaterial; for a complex capital component, validation, tooling and technical change management may dominate.

Quality-adjusted cost

If the probability of a defect is p_d and the cost of correcting or replacing a defective unit is C_d, expected defect cost per purchased unit is:

If a conforming unit is required and defective units cannot be economically recovered, the expected purchased quantity needed for Q conforming units is approximately:

This approximation assumes independent, stable defect probabilities and no capacity constraint. If rework, sorting or recovery is possible, those effects should be modelled separately rather than using this formula alone.

Inventory carrying cost

For average inventory value V_avg and annual carrying-rate i,

The carrying rate may include cost of capital, storage, insurance, obsolescence and handling. It should not include costs already separately charged in the model. Pipeline inventory may be included when transit time differs materially among suppliers.

Risk-adjusted cost

A simple expected-loss model is:

where p_j is the probability of risk event j during the chosen horizon and L_j is its financial loss. This is appropriate when probabilities and losses are sufficiently estimable. It is not a prediction of actual annual cost; rare events may produce highly variable outcomes. Correlated risks, such as port closure and regional shortages, must not be treated as independent without justification.

4. Cost Breakdown Analysis

A cost breakdown separates the supplier's price into economically meaningful drivers. A manufacturing cost model may include:

where M is direct material, L direct labour, OH manufacturing overhead, E energy, Q quality cost and P profit or margin.

Cost breakdowns support negotiation, should-cost modelling and supplier development. They are most useful when the buyer understands the physical process, bill of materials, labour content, yield, machine time and relevant commodity exposure.

A breakdown should distinguish:

• fixed costs from variable costs;
• controllable from uncontrollable costs;
• market-driven from supplier-specific costs;
• recurring from one-time costs;
• transparent evidence from managerial estimates.

Cost breakdowns have limitations. Suppliers may legitimately treat process know-how as confidential, and a buyer's engineering estimate may be wrong because it ignores local productivity, financing or compliance costs. The objective is not to demand that every supplier earn the same margin, but to identify whether observed price differences are economically explainable and sustainable.

5. Learning Curves and Volume Effects

Learning curves describe declining labour or process cost as cumulative output increases. The classical unit-time model is:

where T_N is the time for the Nth unit, T₁ is the time for the first unit, and b = log₂(r) when r is the learning rate expressed as a decimal. A learning rate of 80% means that the time for each doubling of cumulative output becomes 80% of the previous time.

This model assumes a reasonably stable process, repeatable product and learning driven by cumulative production. It is inappropriate when automation, product redesign or radically different batch sizes drive the change.

Worked Example 2: Learning Curve

The first unit requires 100 labour hours and the learning rate is 80%. For the 16th unit, cumulative output has doubled four times from one unit. Therefore:

T₁6 = 100 × (0.8)⁴ = 40.96 hours.

If labour costs £25 per hour, the labour cost for the 16th unit is:

40.96 × £25 = £1,024

This is the time for the individual 16th unit under the unit model, not the average time for all 16 units. Confusing unit and cumulative-average learning curves is a common analytical error.

Learning effects should be incorporated into negotiations carefully. A supplier may reduce price because of expected learning, but the buyer should verify that volume commitments, process stability and quality performance justify the assumed curve.

6. Price-Index Analysis and Escalation

Long-term contracts often use indices to adjust prices for changes in input costs. A transparent escalation formula is:

where:

• P₀ and P_t are base and adjusted prices;
• M, L, E, F, FX represent material, labour, energy, freight and exchange-rate indices;
• subscript 0 denotes the base period and subscript t the current period;
• w_i are cost weights;
• w_U is the fixed or unindexed share.

The weights should sum to one:

The formula is valid only if the selected indices reasonably represent the cost drivers and the contract defines publication source, frequency, base year, lag, rounding and extraordinary-event treatment. An index should not be used merely because it is convenient. For example, a general consumer price index may poorly represent a copper-intensive component.

Worked Example 3: Index Adjustment

Suppose P₀ = £100, with weights: materials 0.45, labour 0.25, energy 0.10 and fixed share 0.20. Material, labour and energy indices rise respectively by 12%, 6% and 20%.

The adjustment factor is:

0.45(1.12) + 0.25(1.06) + 0.10(1.20) + 0.20 = 1.101

Thus:

P_t = £100 × 1.101 = £110.10

The calculation does not imply that the supplier's total cost rose by exactly 10.1%; it applies the agreed index methodology to the contract price.

7. Multicriteria Supplier Selection

Cost alone is insufficient when suppliers differ in quality, lead time, resilience, innovation or compliance. MCDM makes trade-offs explicit.

Weighted scoring model

Let supplier s be assessed against criteria i. If w_i is the criterion weight and r_si is the normalized supplier rating, the total score is:

with:

Higher scores must consistently mean better performance. For a cost criterion, a lower raw cost should therefore be converted into a higher benefit score, or cost should be handled separately.

A sound process is:

1. establish mandatory pass/fail requirements;
2. define criteria operationally;
3. assign weights with stakeholder involvement;
4. select evidence and rating scales;
5. score independently before group discussion;
6. test sensitivity to weights and ratings;
7. document conflicts of interest and approval authority.

Analytic hierarchy process

The analytic hierarchy process (AHP) uses pairwise comparisons to derive relative priorities. It is useful when decision-makers find direct percentage weights difficult. However, excessive criteria can create comparison fatigue, and subjective judgements remain subjective even when converted into numerical ratios. AHP should therefore complement, not conceal, managerial judgement.

Benefit, cost and risk criteria

One robust architecture separates:

• feasibility gates: legal, technical, safety and ethical minimums;
• economic score: TCO or risk-adjusted TCO;
• performance score: quality, delivery and service;
• strategic score: innovation, resilience and sustainability.

A supplier failing a safety or forced-labour requirement should not compensate by receiving a high price score. This is a governance principle, not merely a mathematical preference.

8. Scenario and Sensitivity Analysis

A single forecast disguises uncertainty. Scenario analysis evaluates coherent combinations of assumptions, such as base, adverse and favourable conditions. Sensitivity analysis varies one or more inputs to determine which assumptions influence the decision most strongly.

For a decision metric Y = f(x₁, x₂, ..., x_n), a local sensitivity approximation is:

This approximation is most useful for small changes and reasonably smooth relationships. For discontinuities, thresholds or interacting risks, full recalculation or simulation is preferable.

A break-even analysis identifies the value of an uncertain variable at which two suppliers become equal. If supplier A has fixed economic cost F_A, variable cost v_A per unit, and supplier B has corresponding values, the break-even volume is:

provided the denominator is non-zero and the resulting volume is within the feasible operating range.

Illustrative comparison

Supplier A has TCO of £1,000,000 at 100,000 units, including £200,000 fixed cost. Supplier B has £900,000 fixed cost and variable cost £1.10 per unit. A's variable cost is therefore:

v_A = (£1,000,000 − £200,000) ÷ 100,000 = £8.00

B's variable cost is £1.10 only if its total cost structure is stated consistently; such a dramatic difference signals that the data need checking. This illustrates an important professional rule: analytical output must be challenged for economic plausibility. A model can be arithmetically correct and conceptually wrong if cost categories are incomplete or defined differently.

In practice, sensitivity analysis should test exchange rates, demand volume, freight rates, defect rates, lead time, duty, energy prices, working-capital cost and disruption probability. Tornado charts, threshold tables and Monte Carlo simulation can communicate results, but simulation quality depends on valid distributions and correlations.

9. Integrated Supplier Selection Process

An effective process combines economics and judgement:

1. Define the specification and decision boundary. Clarify performance, volume, destination, contract term and mandatory requirements.
2. Collect comparable data. Standardise Incoterms, currency, packaging, delivery point, taxes and payment terms.
3. Build the landed-cost model. Identify all logistics and import components.
4. Build the TCO model. Add quality, inventory, administration, switching and risk costs.
5. Validate cost drivers. Use supplier breakdowns, market indices, engineering estimates and historical data.
6. Assess non-cost criteria. Evaluate capacity, quality systems, delivery, technology, resilience, sustainability and governance.
7. Score and rank suppliers. Apply an explicit MCDM method after eligibility screening.
8. Run scenarios and sensitivity tests. Identify whether the ranking is robust.
9. Negotiate and contract. Translate assumptions into service levels, index clauses, audit rights, remedies and change-control procedures.
10. Monitor realised economics. Compare actual performance with the business case and update the model.

10. Illustrative Case: Selecting a Global Component Supplier

Background

A fictional medical-equipment manufacturer requires 50,000 precision housings annually. Supplier North is local; Supplier East is overseas. Both meet the basic technical specification.

Problem

East quotes £42 per unit compared with North's £46. Management must decide whether the apparent £200,000 annual price saving is real after logistics, quality, inventory and disruption effects.

Analysis

The procurement team estimates the following annual costs:

East is cheaper on invoice price but £245,000 more expensive on estimated TCO. The team then tests assumptions. If East reduces its defect rate and secures a nearer consolidation hub, its TCO falls by £130,000. If North's capacity expansion requires tooling of £180,000, North's first-year advantage falls substantially. A two-supplier allocation may therefore be superior to an exclusive award, despite a small loss of volume efficiency.

Approach and solution

The company adopts a 70:30 allocation, subject to quarterly review. North receives the core volume because of lower risk-adjusted TCO, while East remains a qualified second source. The contract includes quality targets, delivery recovery obligations, index-based material adjustments, audit rights, business-continuity requirements and transparent change notification.

Outcome

The organisation avoids treating the overseas quotation as a saving without considering total exposure. It also preserves competitive tension and resilience rather than selecting a single supplier solely on a static score.

Learning points

1. TCO can reverse a unit-price ranking.
2. Risk estimates should be tested rather than accepted mechanically.
3. Dual sourcing is a portfolio decision involving cost and resilience.
4. Contract design is part of sourcing economics.
5. A model must be updated after award using actual performance.

11. Limitations, Ethics and Governance

TCO models can create false precision. Many inputs are estimates, and rare-event probabilities may be weakly evidenced. Analysts should show ranges, confidence levels or scenario bands rather than reporting an apparently exact answer without qualification.

Ethical sourcing requires that cost pressure does not encourage unsafe work, forced labour, discrimination, environmental abuse or concealment of subcontracting. Supplier selection should include due diligence, traceability, worker-protection requirements and credible remediation processes. Audits alone are insufficient when they are predictable, superficial or disconnected from purchasing practices.

Governance controls should include segregation of duties, documented evaluation criteria, conflict-of-interest declarations, controlled access to bids, audit trails and approval of model changes. Artificial intelligence can assist in extracting invoice data, detecting anomalous prices, forecasting indices and screening risk signals. It should not independently make high-impact supplier decisions without explainability, data-quality controls, human review and protection of confidential supplier information.

Contemporary sourcing increasingly addresses carbon, water, circularity, geopolitical exposure, cybersecurity and human-rights risk. These factors may be monetised where credible data exist, but forcing every impact into a monetary value can obscure moral and regulatory obligations. A hybrid approach—mandatory thresholds plus economic and multicriteria analysis—is generally more defensible.

Best-Practice Checklist

• Define the unit of comparison before collecting prices.
• Use a common currency, time basis and delivery point.
• Separate recoverable taxes from economic costs.
• Avoid double counting freight, inventory or quality costs.
• Document every assumption and data source.
• Reconcile supplier cost breakdowns with market evidence.
• Distinguish unit learning curves from cumulative-average curves.
• Use relevant and contractually defined price indices.
• Apply mandatory compliance gates before weighted scoring.
• Test ranking stability under plausible scenarios.
• Include switching, disruption and end-of-life costs where material.
• Review realised cost and supplier performance after implementation.

Chapter Summary

Sourcing economics evaluates suppliers over the full economic life of a purchase rather than by quoted price alone. Landed cost adds freight, insurance, duty, handling and related acquisition costs. Total cost of ownership extends the analysis to inventory, quality, administration, risk, switching and end-of-life effects. Cost breakdowns reveal the drivers behind a price and support should-cost analysis, while learning curves explain potential productivity improvements as cumulative output grows. Price-index formulas provide disciplined escalation mechanisms when their weights and data sources are appropriate.

Supplier selection analytics combines TCO with multicriteria methods such as weighted scoring and AHP. Eligibility gates are essential because legal, safety and ethical failures should not be offset by a favourable price. Scenario and sensitivity analysis reveal whether a supplier ranking is robust or depends on fragile assumptions. The most effective sourcing decisions are transparent, evidence-based, risk-aware and continuously monitored after contract award.

Key Terms

Total cost of ownership: The complete economic cost of acquiring, using, managing and disposing of a purchased item or service.

Landed cost: Cost of bringing goods to a defined destination, including relevant logistics and import costs.

Cost breakdown: A structured decomposition of a supplier price into material, labour, overhead, quality, profit and other components.

Should-cost model: An analytical estimate of what a product or service ought to cost under specified assumptions.

Learning curve: A relationship showing declining unit effort or cost as cumulative production increases.

Price index: A statistical measure of change in the price of a defined basket or input over time.

Escalation clause: A contractual mechanism that adjusts price according to specified cost indices or triggers.

Risk-adjusted cost: Cost including the expected financial effect of identified uncertain events.

Expected loss: Probability of an event multiplied by its financial consequence.

Multicriteria decision-making: Selection among alternatives using multiple weighted or prioritised criteria.

Weighted scoring: A method that multiplies criterion weights by supplier ratings and sums the results.

Analytic hierarchy process: A pairwise-comparison method for deriving priorities among criteria and alternatives.

Eligibility gate: A mandatory requirement that an alternative must satisfy before comparative scoring.

Sensitivity analysis: Examination of how output changes when one or more assumptions change.

Scenario analysis: Evaluation of outcomes under coherent alternative sets of assumptions.

Break-even volume: The quantity at which competing alternatives have equal economic cost.

Incoterms: International commercial terms defining responsibilities, costs and risks in delivery transactions.

Supplier resilience: The capability of a supplier and its network to withstand and recover from disruption.

TCO leakage: The difference between modelled cost and additional or unexpected cost realised in operation.

Review Questions

Short Answer Questions

1. Distinguish landed cost from total cost of ownership.
2. Why must recoverable taxes generally be excluded from economic cost?
3. What is the purpose of a supplier cost breakdown?
4. State two conditions required for meaningful price-index escalation.
5. Why should eligibility gates precede weighted supplier scoring?

Long Answer Questions

1. Explain how an organisation can construct a TCO model for an imported component.
2. Discuss the assumptions, uses and limitations of learning-curve analysis.
3. Compare weighted scoring with the analytic hierarchy process for supplier selection.
4. Explain how risk-adjusted cost and scenario analysis improve sourcing decisions.
5. Critically evaluate the ethical and governance risks of cost-focused supplier selection.

Analytical / Application Questions

1. A supplier offers a low unit price but has longer lead time, higher defect rates and greater disruption exposure. Design a TCO model and identify the data required to compare it with a local supplier.
2. Develop a weighted supplier-selection framework for a critical medical component. Justify the criteria, weights, mandatory gates and sensitivity tests.
3. A long-term supplier requests a price increase based on a general inflation index. Construct an index-based review process that tests whether the index represents the supplier's actual cost drivers.

Multiple Choice Questions with Answers

Question 1

Which cost is normally included in landed cost but not necessarily in the supplier’s quoted unit price?

A. Freight to the buyer’s destination

B. The supplier’s internal profit margin

C. The buyer’s corporate income tax

D. The buyer’s future product revenue

Correct Answer: A. Freight to the buyer’s destination

Question 2

A shipment has an invoice value of £36,000 and customs duty is 6% of that value. What is the duty?

A. £216

B. £2,160

C. £6,000

D. £21,600

Correct Answer: B. £2,160

Question 3

If the defect probability is 0.04 and the cost of correcting a defect is £50, what is the expected defect cost per purchased unit?

A. £0.20

B. £2.00

C. £20.00

D. £1,250

Correct Answer: B. £2.00

Question 4

In a weighted-scoring model, what must the criterion weights normally sum to?

A. 0

B. 10

C. 50

D. 1 or 100%, depending on the scale

Correct Answer: D. 1 or 100%, depending on the scale

Question 5

An 80% learning curve means that, after each doubling of cumulative output, the relevant unit time becomes:

A. 80% of the previous comparable time

B. 20% greater than the previous time

C. Exactly 80 hours

D. Unchanged

Correct Answer: A. 80% of the previous comparable time

Question 6

Which condition is most important for a meaningful price-index escalation clause?

A. The index is selected solely because it is widely known

B. The index represents the relevant cost drivers

C. The index always increases every year

D. The index eliminates all supplier profit

Correct Answer: B. The index represents the relevant cost drivers

Question 7

Why should a mandatory safety requirement normally be treated as an eligibility gate?

A. Safety performance is irrelevant to cost

B. A supplier should not compensate for failing safety through a high commercial score

C. It guarantees the lowest price

D. It removes the need for supplier audits

Correct Answer: B. A supplier should not compensate for failing safety through a high commercial score

Question 8

Expected disruption cost is most simply estimated as:

A. Probability of disruption multiplied by its financial loss

B. Annual volume divided by lead time

C. Unit price multiplied by supplier profit

D. Freight divided by insurance

Correct Answer: A. Probability of disruption multiplied by its financial loss

Question 9

What is the main purpose of sensitivity analysis in supplier selection?

A. To hide uncertain assumptions

B. To determine how changes in assumptions affect the decision

C. To replace all managerial judgement

D. To guarantee the forecast is correct

Correct Answer: B. To determine how changes in assumptions affect the decision

Question 10

Which statement best describes total cost of ownership?

A. It is always identical to invoice price

B. It includes only customs duty and freight

C. It considers relevant acquisition, operating, quality, risk and lifecycle costs

D. It excludes inventory and switching costs in every case

Correct Answer: C. It considers relevant acquisition, operating, quality, risk and lifecycle costs