How a Chinese Logistics Boss Applied Principal‑Agent Theory to Clear 1.05 M Debt

Traditional Model Dilemma

In the standard principal‑agent model, the boss’s profit and the employee’s profit both depend on the unobservable effort level. Information asymmetry leads to a sub‑optimal equilibrium: employees choose minimal effort, while the boss lowers wages and tightens supervision.

Real‑world evidence from Yonghui Supermarket’s failed redesign illustrates this: despite investing 8 million yuan in store layout, sales fell from 200 million to 60 million yuan per day because the incentive function remained unchanged, violating incentive‑compatibility constraints.

Data from "Pang Donglai" (2024)

Revenue: 170 billion yuan, profit: 8 billion yuan (2023 profit was only 1.4 billion)

13 stores, about 10,000 employees

Average monthly salary: 9,000 yuan, turnover rate: 2.01%

per‑capita revenue = 170 billion ÷ 10 k = 170 k
per‑capita annual salary = 9 k × 12 = 108 k
efficiency ratio = 170 ÷ 108 = 15.74
net profit margin = 8 ÷ 170 = 4.71%
profit growth = 471%
annual profit per store = 61.54 million (exceeds 2,198 A‑share companies)

Incentive Function Reconstruction

Traditional model: fixed individual performance bonus.

Pang Donglai model: a large share of profit (β) is allocated to employees, with the boss retaining only a small portion.

Marginal Incentives

Base salary is already twice the market level, reducing the need for additional cash incentives.

Social pressure raises the cost of shirking.

Cultural identification lowers the disutility of effort.

These factors make the marginal incentive appear small but produce a far larger actual effect than the traditional model.

Team Incentive Mechanism

Holmstrom (1982) proved that under a budget‑balanced constraint, the Nash‑equilibrium effort level is strictly below the Pareto‑optimal level because of “free‑riding”.

Solution: introduce a residual‑claimant principal to break the budget balance. In the Pang Donglai model the boss keeps only 5% of profit, creating a team incentive where individual laziness is detected by peers, and helping colleagues increases one’s own payoff.

Efficiency Analysis

Per‑capita profit: 80 k yuan; per‑capita salary: 108 k yuan, appearing as a 28 k yuan loss. Hidden benefits are substantial:

Turnover Reduction

Reduced turnover = (0.25 – 0.0201) × 10 k = 2,299 employees per year

Cost saved = 2,299 × 50 k = 115 million yuan

Loss Reduction

Fresh‑produce loss: 2.8% vs industry average 15%

Saving = 85 billion × (0.15 – 0.028) = 1.037 billion yuan

Combined ROI = (115 m + 1.037 b) ÷ 28 k ≈ 3.11, i.e., every 1 yuan of high salary yields 3.11 yuan of total return.

Why the Zhengzhou Boss Succeeded

The situation is a repeated‑game problem. When heavily indebted, the boss’s only rational choice is to honor commitments, because there is little to gain from default. This “death‑by‑a‑thousand‑cuts” strategy creates a credible promise.

Boundary Conditions for Replication

Constraint 1: Scale Limit – Management capacity caps at about 13 stores (≈770 employees per store).

Constraint 2: Cash Flow – Paying 9 k versus a 4 k market salary requires 2.25× labor cost; a 50 million‑revenue firm would need an extra 18.75 million yuan cash reserve.

Constraint 3: Trust – Full financial transparency and proven case studies are mandatory.

All three constraints must be satisfied for the model to be replicable.

Revision of Classic Theory

The traditional Holmstrom‑Milgrom model assumes an optimal upper bound on effort. The Pang Donglai approach breaks this bound by allowing a larger profit share for workers.

New Conclusion : Under specific conditions (moderate scale, guaranteed base pay, full transparency, strong cultural alignment) the revised incentive scheme can be Pareto‑optimal.

Properties :

There exists a critical point where marginal benefit equals marginal cost.

As scale grows, the optimal effort level declines.

Even with a small profit share for the boss, total profit growth keeps absolute returns non‑decreasing.

Practical Recommendations

Three Phases :

Pilot (3 months): β = 0.3, salary +30%, 50‑person team.

Scale‑up (6 months): β = 0.5, salary +50‑100%, eliminate clock‑in.

Institutionalize (ongoing): β = 0.7‑0.9, embed in company charter.

Avoid Common Mistakes :

Only impose requirements without salary increase.

Focus on formality without substance.

Fail to deliver promised incentives.

Lack of financial transparency.

Three‑Month Self‑Check : monitor turnover, performance, and culture; if four out of five indicators deteriorate, stop the experiment immediately.

Core Inequality

When the inequality defining the profit‑share β is satisfied, employee‑initiated effort becomes a Nash equilibrium without the need for supervision.

Empirical evidence shows profit rising from 1.4 billion to 8 billion yuan (471% growth), confirming the model’s scalability. Four indispensable conditions are required: genuine profit sharing (β ≥ 0.5), truly high wages (≥2× market), full transparency (T = 1), and controlled scale (n ≤ n_max).

When incentive‑compatibility holds, supervision becomes unnecessary and the Nash equilibrium converges to the Pareto‑optimal outcome.

Data Sources : Public disclosures of Pang Donglai (2025‑01‑08), Xinhua News Agency reports, Yonghui financial statements.

References : Holmstrom (1982), Zhang Weiying (1996).