5D Library › Cross-Cultural Contracts, Real Options and Game Theory of Southeast Asian Market Entry · 中文
Chapter 3 Incomplete Contracts and Relational Contracts
The contract that cannot be finished
If a contract could set down every future situation and what each party should do in it, cooperation would have no disputes: when circumstances change, you turn to the matching page and follow it. Real contracts cannot do this. There are too many future situations, and some cannot be foreseen; some can be foreseen but cost too much to write down; and some can be written down but a court could not verify whether they occurred. Economists call such contracts incomplete.
Business practice adapted to this long ago. Macaulay’s interviews with firms in Wisconsin found that business people rarely relied on contract terms or went to law in everyday dealings; disagreements were usually settled by a phone call, and many felt that overly detailed terms signalled distrust (Macaulay 1963).
Who holds the residual control rights?
Grossman and Hart, and Hart and Moore, proposed a framework for the problem of incomplete contracts (Grossman and Hart 1986; Hart and Moore 1990). Since the contract cannot specify everything, it matters greatly who decides in the situations it does not cover. They called this right residual control, and tied it to the ownership of assets: whoever owns the key asset decides in the gaps of the contract, and so has a better fallback in later bargaining.
This affects incentives to invest beforehand. Suppose both sides must make specific investments in the venture: headquarters contributes product, brand and technology; the local partner contributes channels, customer relationships and the ability to deal with regulators. A specific investment is worth less elsewhere if the relationship breaks down, so the investing party is easily “held up” by the other in later bargaining (Klein et al. 1978). Anticipating this, both invest less. Owning the asset reduces one’s own exposure to hold-up and so encourages investment; but the asset can belong to only one party, and giving it to one weakens the other’s incentive.
Calculation
A numerical version of the Grossman–Hart–Moore model. The two parties invest and then bargain, splitting equally (Nash bargaining) whatever value exceeds their outside options. If bargaining fails, the owner of the key asset can still use half the value of its own investment, a non-owner only a fifth. Under joint ownership either party can block the other’s use of the asset, so neither has an outside option.
# Who should own the asset? The Grossman-Hart-Moore property-rights model, numerically.
# Two parties invest before they bargain: H (the foreign headquarters: product, brand, technology,
# investment h) and L (the local partner: distribution, customer and regulator relationships, l).
# Investment costs h^2/2 and l^2/2. Inside the relationship the venture is worth
# V = a*h + b*l.
# If bargaining fails, each party has an outside option that depends on who owns the key asset
# (for example the local operating company holding licences and the customer base). The owner can
# still use part of its own investment with the asset (share 0.5); a non-owner keeps only 0.2 of
# its investment's value. Ex post they split the gain over outside options 50/50 (Nash bargaining),
# so each invests where marginal cost = 1/2 (marginal value inside + marginal outside option).
# Under joint ownership either party can block use of the asset, so neither has an outside option.
def outcome(owner, a, b):
oh = 0.5 if owner == "H" else 0.2 # share of own investment usable outside the relationship
ol = 0.5 if owner == "L" else 0.2
if owner == "joint": # joint ownership: either party can veto use of the asset
oh = ol = 0.0
h = 0.5 * (a + oh * a) # first-order conditions: h = (a + oh*a)/2
l = 0.5 * (b + ol * b)
surplus = a * h + b * l - h * h / 2 - l * l / 2
return h, l, surplus
for a, b, story in ((2.0, 1.0, "HQ's investment matters more (technology-led product)"),
(1.0, 2.0, "local investment matters more (distribution- and licence-led market)")):
first_best = (a * a + b * b) / 2
print(f"{story}: first-best surplus {first_best:.2f}")
for owner in ("H", "L", "joint"):
h, l, s = outcome(owner, a, b)
print(f" owner {owner:5s}: investments h = {h:.2f}, l = {l:.2f}, surplus {s:.2f} ({s / first_best:.0%} of first best)")
# HQ's investment matters more (technology-led product): first-best surplus 2.50
# owner H : investments h = 1.50, l = 0.60, surplus 2.29 (92% of first best)
# owner L : investments h = 1.20, l = 0.75, surplus 2.15 (86% of first best)
# owner joint: investments h = 1.00, l = 0.50, surplus 1.88 (75% of first best)
# local investment matters more (distribution- and licence-led market): first-best surplus 2.50
# owner H : investments h = 0.75, l = 1.20, surplus 2.15 (86% of first best)
# owner L : investments h = 0.60, l = 1.50, surplus 2.29 (92% of first best)
# owner joint: investments h = 0.50, l = 1.00, surplus 1.88 (75% of first best)The conclusion agrees with the theory: the asset should be owned by the party whose investment matters more. In a business led by product and technology, ownership by headquarters brings the surplus to 92 % of the first-best level; in a business led by distribution and licences, the local partner should own it. Joint ownership is the worst in both cases, at 75 % of the first best: each side can veto the other, and as a result neither has enough reason to invest.
Note
A real constraint in Southeast Asia. In Southeast Asia, who owns the key asset is often not entirely the parties’ free choice. Thailand’s Foreign Business Act (1999) restricts majority foreign ownership in many lines of business; Indonesia sets the conditions for foreign shareholding by sector in a positive investment list; Malaysia and other countries have local ownership requirements in some sectors. These rules change, and the official text in force should be checked. Their economic implication is that in some sectors the local partner necessarily holds legal control of the operating company, and what headquarters can design is the ownership of other assets (brand, technology, data, supply chain) and the allocation of decision rights in the contract. The checklist in Chapter 8 therefore makes the residual-control matrix its second step.
Self-enforcing promises: relational contracts
Residual control answers “who decides in the gaps”. Another approach is to skip the courts altogether and let the relationship enforce itself. Baker, Gibbons and Murphy call such informal agreements, enforced by both sides’ expectations of the value of future cooperation, relational contracts (Baker et al. 2002).
The logic comes from repeated games. As long as future cooperation is valuable enough, neither side will give it up for the one-off gain from reneging. If each period’s cooperation creates surplus , a party gets per period after the relationship breaks, and the discount factor is , the largest temptation to renege that an informal promise can withstand is:
When an unexpected shock in some period makes reneging worth more than this cap, the promise breaks.
An important finding of Baker and colleagues is that formal and relational contracts can be either complements or substitutes (Baker et al. 1994, 2002). A formal contract can cover some situations and lighten the relationship’s load; but the better the formal contract, the better the fallback after the relationship breaks, the smaller the cost of reneging, and the more fragile the relational promise may become.
Evidence
Poppo and Zenger, using survey data on information-services outsourcing by US firms, found that contractual complexity and relational governance were positively correlated, and that together they were associated with better exchange performance; they concluded that in practice the two function more as complements than as substitutes (Poppo and Zenger 2002). Hart and Moore later proposed that contracts act as reference points: a contract is not only the basis of enforcement but also shapes what both sides consider fair, and adjustments outside the contract that depart from this reference point provoke resentment and retaliatory shading (Hart and Moore 2008).
Three forms of governance compared
Now bring in the decoding problem of Chapter 1. Under a relational contract, even when no one reneges, one side may misread a fair adjustment by the other as a breach and end the relationship. Between partners from different cultures such misreadings are more likely. Explicit state-contingent milestones (writing in the contract what adjustment will be made under what observable conditions) are exactly what reduce misreading: an adjustment that is written down needs no interpretation.
Calculation
Compare three forms of governance: a formal contract alone; a relational promise alone; and a hybrid of formal milestones plus relational governance. The speed of environmental change is the mean size of unexpected shocks. Drafting formal clauses costs more the more they cover; an uncovered shock is either renegotiated at arm’s length, losing part of the surplus, or left to the relationship. In the hybrid, the formal contract remains after a breach, so the cap of the relational promise is smaller (the substitution effect of Baker and colleagues).
# Formal, relational or hybrid governance when the environment moves faster than contracts can be
# written. A stylised model in the spirit of Baker, Gibbons & Murphy (2002); numbers illustrative.
# Each period the venture creates surplus S = 10, and a shock of size z arrives, z ~ exponential with
# mean sigma (the speed of environmental change). Discount factor delta = 0.8.
# Formal only: written clauses cover shocks up to Zw (drafting costs 0.15 * Zw per period). A larger
# shock is a gap: the parties renegotiate at arm's length and lose 6 of that period's surplus.
# Relational only: every shock is handled by an informal promise to adjust fairly. The promise holds
# while z <= cap = delta / (1 - delta) * (S - fallback); a larger z is worth reneging over, and the
# relationship ends. Even when nobody reneges, a partner may MISREAD a fair adjustment as a breach
# (probability m per period); that also ends the relationship. m is higher across cultures.
# Hybrid: written state-contingent milestones handle shocks up to Zw with no room for misreading;
# the promise handles the rest. Breaking the promise now leaves the formal contract (worth 2 per
# period), so the fallback rises and the cap shrinks: the substitution effect in Baker et al. (1994).
# Values are per-period equivalents, (1 - delta) times the discounted value of the relationship.
import math
S, LOSS, DRAFT, DELTA = 10.0, 6.0, 0.15, 0.8
def tail(a, sigma):
return math.exp(-a / sigma)
def per_period(flow, p_end):
return flow * (1 - DELTA) / (1 - DELTA * (1 - p_end))
def formal(sigma, m):
return max((S - DRAFT * zw - LOSS * tail(zw, sigma), tail(zw, sigma))
for zw in (k * 0.25 for k in range(401)))
def relational(sigma, m):
cap = DELTA / (1 - DELTA) * S
p_end = tail(cap, sigma) + (1 - tail(cap, sigma)) * m
return per_period(S, p_end), p_end
def hybrid(sigma, m):
cap = DELTA / (1 - DELTA) * (S - 2.0)
best = None
for zw in (k * 0.25 for k in range(401)):
covered = 1 - tail(zw, sigma)
reneg = tail(max(zw, cap), sigma)
p_end = reneg + (1 - covered - reneg) * m # misreading only where nothing is written
v = per_period(S - DRAFT * zw, p_end)
if best is None or v > best[0]: best = (v, p_end)
return best
for m, who in ((0.005, "same-culture pair, misreading 0.5 %"), (0.04, "cross-cultural pair, misreading 4 %")):
print(f"{who}:")
print(" speed of change | formal: value (gaps) | relational: value (ends) | hybrid: value (ends) | best")
for sigma in (1, 3, 6, 12, 24):
fv, fp = formal(sigma, m); rv, rp = relational(sigma, m); hv, hp = hybrid(sigma, m)
best = max((("formal", fv), ("relational", rv), ("hybrid", hv)), key=lambda x: x[1])[0]
print(f" {sigma:15d} | {fv:6.2f} ({fp:5.1%}) | {rv:6.2f} ({rp:5.1%}) |"
f" {hv:6.2f} ({hp:5.1%}) | {best}")
# same-culture pair, misreading 0.5 %:
# speed of change | formal: value (gaps) | relational: value (ends) | hybrid: value (ends) | best
# 1 | 9.30 ( 2.4%) | 9.80 ( 0.5%) | 9.81 ( 0.4%) | hybrid
# 3 | 8.38 ( 7.6%) | 9.80 ( 0.5%) | 9.80 ( 0.5%) | relational
# 6 | 7.39 (14.7%) | 9.76 ( 0.6%) | 9.62 ( 1.0%) | relational
# 12 | 6.03 (29.9%) | 8.61 ( 4.0%) | 7.71 ( 7.4%) | relational
# 24 | 4.56 (60.0%) | 5.64 (19.3%) | 4.83 (26.7%) | relational
# cross-cultural pair, misreading 4 %:
# speed of change | formal: value (gaps) | relational: value (ends) | hybrid: value (ends) | best
# 1 | 9.30 ( 2.4%) | 8.62 ( 4.0%) | 9.50 ( 0.4%) | hybrid
# 3 | 8.38 ( 7.6%) | 8.62 ( 4.0%) | 9.02 ( 1.2%) | hybrid
# 6 | 7.39 (14.7%) | 8.58 ( 4.1%) | 8.56 ( 3.1%) | relational
# 12 | 6.03 (29.9%) | 7.71 ( 7.4%) | 7.01 (10.7%) | relational
# 24 | 4.56 (60.0%) | 5.30 (22.1%) | 4.60 (29.3%) | relational
The model gives a more careful answer than “the hybrid contract is best”:
- A formal contract alone is never best, and the faster the environment changes, the more gaps it has.
- Between same-culture partners, misreading is rare, and relational governance alone is best or level with the hybrid in almost every environment.
- Between cross-cultural partners, misreading drags down the value of relational governance alone; when change is not too fast (mean shock 1–3), the hybrid is best, its written milestones removing most misreadings.
- When change is very fast, relational governance alone catches up with and overtakes the hybrid in the model, even for cross-cultural partners: at a mean shock of 6 the two are about level, and from 12 upwards relational governance alone is clearly better. The reason is the substitution effect pointed out by Baker and colleagues: in the hybrid, the fallback provided by the formal contract weakens the shocks the relational promise can absorb.
So the advantage of an “explicit state-machine hybrid contract” is conditional: it is largest when the risk of misreading is high and environmental change is moderate. That is exactly the situation of many Chinese firms entering Southeast Asian markets for the first time, but not every situation.
Hypothesis · not established
The milestones-reduce-misreading hypothesis (not yet established). In cross-cultural cooperation, writing adjustment conditions as observable milestones mainly reduces relationship break-ups caused by misreading, not genuine reneging. A testable implication: among ventures using written milestones, the share terminated early for reasons such as “communication misunderstanding” or “the other side did not keep its word” should be lower than among ventures without them, while the share terminated for “market reasons” should not differ much. If both kinds of termination fall in the same proportion, or neither falls, the hypothesis is weakened.
Summary
Contracts cannot be finished, so two things must be decided: who decides in the gaps (residual control rights), and what sustains cooperation beyond the contract (the relationship). The models show that the asset should go to the party whose investment matters more, and that mutual veto is the worst arrangement; that written milestones are most valuable in cross-cultural, moderately uncertain environments, but may weaken the relationship itself when the environment changes fast. The next chapter looks at speed from another angle: when is the speed of adjustment itself an advantage, and when a cost?
References
Fan, Y. (2026). Cross-Cultural Contracts, Real Options and Game Theory of Southeast Asian Market Entry: Relational Contracting, Optimal Stopping Thresholds and Institutional Distance, Read Beside the Art of War [跨文化契约、实物期权与东南亚出海博弈论] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23753261 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).