IEB ยท Session 08 ยท Exam Preparation

Transaction Costs &
Digital Transactions

Analyzing why firms choose between hierarchies and markets, and how digitalization reshapes the boundaries of the firm.

Transaction Cost Theory (TCT) โ€” Williamson (1979, 1986)

Transaction Cost Theory provides a fundamental explanation for why firms exist and how they choose between hierarchical integration and market transactions. At its core, TCT argues that the firm is a governance structure designed to minimize the combined costs of production and coordination.

The Core Insight

Organizations choose between Make (Hierarchy), Buy (Market), and Partner (Hybrid) based on which structure minimizes total costs:

  • Production Costs: The costs of actually manufacturing or delivering a good/service
  • Transaction Costs: The costs of coordinating, monitoring, and governing exchanges

The firm exists because markets are imperfect. If all information were free and contracts were costless to enforce, everything would be outsourced to the cheapest provider.

Williamson's Behavioral Assumptions

Williamson identified two key human behaviors that make transaction costs non-trivial:

๐Ÿง 

Bounded Rationality

People are not perfectly rational. They have limited cognitive capacity and cannot anticipate all contingencies when writing contracts or making decisions.

๐Ÿ˜ˆ

Opportunism

People act in their own self-interest, sometimes with guile. Without proper monitoring and incentives, partners will exploit information asymmetries or breach agreements when beneficial.

Williamson's Three Critical Dimensions

These factors determine which governance form (Hierarchy, Market, or Hybrid) is most efficient:

Dimension Definition Implication for Structure
Asset Specificity The degree to which an investment is specialized to a particular relationship (e.g., custom software, proprietary data). Low specificity = easily redeployed; high specificity = "locked in" to one partner. High specificity โ†’ Hierarchy: Firms internalize to protect their investments from opportunism. Low specificity โ†’ Market: Can easily switch suppliers.
Uncertainty The degree to which the future environment is unpredictable (market volatility, technological change, behavioral ambiguity). Hard to write complete contracts when outcomes are uncertain. High uncertainty โ†’ Hierarchy: Easier to adapt by internal reorganization. Low uncertainty โ†’ Market: Can write detailed fixed-price contracts.
Frequency How often the transaction occurs. One-time transactions vs. ongoing relationships. High frequency โ†’ Hierarchy or stable Hybrid: Justifies investment in monitoring systems. Low frequency โ†’ Spot market: Not worth investing in governance.

Types of Transaction Costs

Transaction costs arise at different stages of an exchange. Medaglia (Session 8) identifies four main categories:

๐Ÿ”Ž

Search Costs

The cost of finding a suitable supplier or partner in the market. Includes advertising, travel, information gathering, and due diligence. Digitalization dramatically reduces these.

๐Ÿ“‹

Information Costs

The cost of measuring quality and verifying product/service specifications before the transaction. Includes testing, inspections, third-party certifications, and assessments.

๐Ÿค

Bargaining & Contracting Costs

The cost of negotiating terms, drafting contracts, and reaching an agreement. Includes legal fees, time spent in negotiations, and insurance requirements.

๐Ÿ”’

Policing & Monitoring Costs

The cost of ensuring the partner performs as agreed, preventing shirking or deviation from contract terms. Includes audits, performance tracking, and enforcement actions.

Post-Transaction vs. Pre-Transaction Costs

Medaglia emphasizes: Digitalization primarily reduces ex-ante (pre-transaction) costs like search and information gathering. However, it may increase ex-post (post-transaction) policing costs in complex relationships due to difficulty in measuring digital outputs.

The Total Cost Approach

The fundamental insight of TCT is that the optimal organizational form minimizes the sum of production costs and transaction costs. This creates a trade-off:

The Make-or-Buy Trade-off

  • MAKE (Hierarchy): Lower transaction costs (direct control, no opportunism risk), but potentially higher production costs (in-house often less efficient than specialists).
  • BUY (Market): Lower production costs (specialists are more efficient), but higher transaction costs (search, monitoring, risk of opportunism).
  • PARTNER (Hybrid/Networks): A middle ground attempting to capture benefits of both while minimizing downsides.
Integration Level โ†’ Cost MARKET HIERARCHY Production Cost Transaction Cost Total Cost Optimal
Form
Fig. 1 โ€” The Total Cost Minimization Framework. The optimal organizational form sits at the intersection of rising transaction costs and declining production costs. It's rarely at pure extremes.

Why Organizations Rarely Exist at Pure Extremes

Perfect hierarchies (complete integration) lead to bureaucratic inefficiency and rising production costs. Perfect markets leave firms vulnerable to opportunism and high monitoring costs. Most firms are positioned somewhere in the middle, using hybrid forms like joint ventures, strategic alliances, franchising, and platform ecosystems to balance the trade-off.

Move to the Middle

A critical prediction of TCT is that most organizations will gravitate toward hybrid or intermediate forms rather than pure hierarchies or pure markets. This is because the total cost curve (shown above) has its minimum in the middle, not at the extremes.

What Does "Middle" Mean?

The "middle" encompasses a spectrum of hybrid organizational forms that blend market and hierarchy characteristics:

  • Strategic Alliances & Joint Ventures: Shared control between independent firms, with contractual governance.
  • Networks & Ecosystems: Multiple interconnected firms with relational (trust-based) rather than purely contractual governance.
  • Franchising & Licensing: Partial integration where the franchisor retains control over brand and standards while franchisees retain operational autonomy.
  • Platform Ecosystems: Firms like Uber or Airbnb sit between markets (they don't employ drivers/hosts) and hierarchies (they exert significant control).
  • Supply Chain Partnerships: Long-term relationships with preferred suppliers rather than spot-market transactions.

Evidence of Moving to the Middle

Historically: Historically, firms were either heavily vertically integrated (GM made its own steel) or used pure spot markets. Today:

๐Ÿญ

De-Integration of Large Firms

Many conglomerates broke up in the 1980s-2000s, moving away from full hierarchy toward hybrid forms and partnerships (e.g., Toyota's supplier networks).

๐ŸŒ

Rise of Platform Economies

Digital platforms (Uber, Airbnb, Spotify) are quintessentially "middle" โ€” neither pure hierarchies nor pure markets, but hybrid governance.

Why Digitalization Reinforces Move to the Middle

Digital tools make hybrid forms feasible because they enable:

  • Real-time monitoring: Firms can now monitor performance of partners without owning them (e.g., Uber tracking driver locations and ratings).
  • Rapid communication: Coordination across firm boundaries becomes nearly as fast as internal hierarchical coordination.
  • Automated contracting: Smart contracts and APIs can encode many terms automatically, reducing bargaining costs.
  • Transparency: Data sharing between firms becomes easier (supply chain visibility, API integrations).

The Electronic Market Hypothesis (EMH)

Malone et al. (1987) proposed that information technology would fundamentally reshape organizational boundaries by reducing coordination costs more than production costs.

The EMH Prediction

Because markets are generally more efficient at resource allocation than hierarchies (firms have to support idle capacity; hierarchies create bureaucratic overhead), a decline in coordination costs would shift the optimal trade-off curve downward, favoring markets over hierarchies.

Expected outcome: More outsourcing, shorter supply chains, smaller firm size, and greater use of spot markets. This would lead to "electronic markets" replacing internal departments.

What Actually Happened?

The EMH predicted a wholesale shift toward markets and de-integration. However, reality has been more nuanced:

Prediction (EMH) Actual Outcome Why the Difference?
Firms shrink; outsource everything Firms grew in some industries (tech), outsourced in others (manufacturing) IT creates winner-take-all dynamics and platform economies, increasing firm size in software/cloud sectors.
Spot markets dominate Strategic alliances and platform ecosystems emerged instead Hidden action (agency) and asset specificity remain high in digital partnerships, requiring hybrid governance.
Supplier switching is frictionless Lock-in increased due to data and API switching costs Digitalization created new forms of asset specificity (proprietary data, custom integrations).

The Irony of Digital Platforms

Malone predicted electronic markets would dominate. Instead, we got digital platforms (e.g., Uber, Amazon, Meta) โ€” essentially new forms of hierarchy dressed up as markets. These platforms are intermediaries that exert substantial control over third-party suppliers, contradicting EMH's prediction of decentralized spot markets.

Transaction Risks in Digital Economies

While digitalization reduces many transaction costs, it introduces or amplifies new risks. Organizations must balance efficiency gains against these emerging vulnerabilities.

๐Ÿ˜ˆ

Opportunism & Holdup

A partner can threaten to terminate or degrade service once you've invested heavily (asset specificity). In digital contracts, this manifests as API deprecation or pricing increases post-lock-in.

๐Ÿ‘๏ธ

Hidden Action (Shirking)

The partner under-delivers because outputs are hard to measure digitally (e.g., "Is the AI model truly optimized for our use case?"). Black-box models exacerbate this.

๐Ÿ”

Data Asymmetry

In digital partnerships, one party may have vastly more information about usage patterns, performance, or competitive activity. Difficult to verify partner honesty.

๐Ÿ”’

Lock-in via Switching Costs

Proprietary data formats, custom APIs, and trained models create massive switching costs. Firms become dependent on partners, increasing vulnerability to renegotiation.

Novo Nordisk & OpenAI: A TCT Analysis

How can we apply Transaction Cost Theory to Novo Nordisk's decision to partner with OpenAI for drug discovery?

The Make-or-Buy Decision

Option 1: MAKE (Build in-house LLM)

  • Production Costs: $50M+ investment in AI research, talent acquisition, ongoing R&D. Massive.
  • Transaction Costs: Minimal (internal control, no external coordination).
  • Total: Prohibitively expensive. Novo lacks expertise in LLM training; opportunity cost is high.

Option 2: BUY (Use OpenAI's API)

  • Production Costs: Low. Pay-per-use or subscription to OpenAI's model. Leverage their R&D.
  • Transaction Costs: Search costs near-zero (OpenAI is famous); information costs low (benchmarked models); bargaining straightforward (standard contracts); but policing costs are high โ€” hard to verify if GPT is truly optimized for drug discovery.
  • Total: Lower than MAKE, but exposed to risks below.

The Transaction Cost Challenges

  • Asset Specificity (HIGH): Novo's investment in OpenAI integration, prompt engineering, and data pipelines becomes specific to OpenAI's API. Switching to Google Gemini or Claude would require rework.
  • Uncertainty (VERY HIGH): The AI landscape is rapidly evolving. OpenAI could change pricing, features, or terms. Drug discovery outcomes are unpredictable.
  • Frequency (HIGH): Novo will use the API repeatedly for continuous drug discovery, justifying investment in governance.
  • Opportunism Risk (CRITICAL): OpenAI could raise prices once Novo is locked in, knowing switching costs are high. This is "Small Numbers Bargaining" โ€” only a few LLM providers can offer what OpenAI does.
  • Hidden Action (CRITICAL): Novo cannot easily verify that OpenAI's model is performing optimally for drug discovery, or that OpenAI isn't allocating compute resources away from Novo's requests during peak usage.

Why a Hybrid Form Might Emerge

Given these risks, we might expect Novo Nordisk to:

  • Negotiate a long-term contract with price caps and service-level agreements (SLAs) to mitigate opportunism risk.
  • Maintain internal AI capabilities as a "shadow option" to reduce lock-in and verify OpenAI's performance.
  • Diversify โ€” use OpenAI for some tasks but also explore Google, Anthropic, or open-source models.
  • Invest in integration โ€” create proprietary layers (fine-tuning, domain-specific adaptations) that increase switching costs in Novo's favor.

Likely Oral Exam Questions