Every contract, however carefully drafted, leaves out the part of the relationship that matters most.
This is not a flaw in any particular contract. It is the inherent limit of contracts as a form. A contract specifies what both parties anticipate. The world supplies what neither one did. Whatever fills the gap between those two things is governed by something else, and the something else is usually called a relationship.
Economists have studied this for forty years and named it precisely. In 1986, Sanford Grossman and Oliver Hart published the foundational paper on what came to be called incomplete contracts, demonstrating that no agreement can specify every possible eventuality at acceptable cost. Hart and Bengt Holmström received the 2016 Nobel Prize in Economic Sciences for the broader theory this paper opened. The Nobel committee’s own summary made the point in plain terms: because a contract cannot foresee every situation, it must allocate the right to decide in unforeseen ones. Someone, in other words, must be trusted.
The principle scales down to the household level intact. A standing agreement with a caterer specifies dates, quantities, and dietary restrictions. It does not specify what should happen when the family decides, on Thursday, that Saturday’s gathering will include twenty additional guests, none of whom were on the original count. The contract has nothing to say about that. The relationship, if there is one, supplies the answer immediately.
A vendor agreement with a landscape firm specifies frequency, scope, and budget. It does not specify the calibration that comes after a major storm, when the property has six days of work that needs to be triaged, sequenced, and absorbed without disrupting the principal’s schedule. The contract is silent. The relationship knows what to do.
This pattern is universal in long-standing arrangements. It is also, in the economic literature, well theorized.
What Replaces The Missing Clauses
The Stanford economist Jonathan Levin, in a 2003 paper in the American Economic Review, formalized what he called relational incentive contracts: self-enforcing arrangements that operate where formal contracts cannot. The mechanism is straightforward. Both parties value the continuation of the relationship more than they value any single transaction within it. That valuation, sustained over time, supplies the discipline that no document can.
Levin’s framework is technical. The intuition is not. A landscaper who has worked a property for a decade has more to lose by mishandling the storm response than any single contract could compensate them for. A housekeeper who has been part of the household for fifteen years has decision-making authority over hundreds of small matters that no agreement enumerates, and exercises that authority with a fidelity that no enforcement mechanism could produce. The relationship is doing the work the contract cannot.
This is why long-standing arrangements hold what new ones cannot, even when the new ones are more carefully drafted. The carefully drafted contract describes more contingencies. The long-standing relationship covers all of them, including the ones that have not arisen yet.
The asymmetry of failure is also worth noticing. When transactional arrangements break down, the failure is often visible: a missed deadline, a billing dispute, a vendor who fails to return calls. When relational arrangements break down, the failure is usually quiet: a small slippage in attention, a routine that is no longer maintained, a decision that gets made later than it would have been. The visible failures are easier to manage because they declare themselves. The quiet ones are not, and they are the ones that compound.
A household at scale runs on hundreds of arrangements at any given moment. Some are transactional. Some are relational. Most are both. The skill is not in choosing between them. It is in recognizing which is which, and ensuring that the relational ones have been allowed the time they require to become what they are.
Contracts cannot hold the parts of a household that matter most. The relationships have to.
