Yongjian Ke

Psychological contracts: the unwritten side of construction procurement

· 6 min

A construction contract lists what each party must do. Alongside it sits a second set of terms that nobody signs: that information will be passed on before it has to be, that a problem will be raised early rather than stored up, that doing good work will count for something later. None of it is enforceable and none of it is written down, but people act on it and they notice when it fails. Two papers written with Peter Davis and Marcus Jefferies at the University of Newcastle, funded by a pilot grant from Newcastle's Faculty of Engineering and Built Environment, set out to describe that second set of terms and to measure it (Ke, Davis, & Jefferies, 2016; Davis, Jefferies, & Ke, 2017).

A concept borrowed from human resources

The idea is the psychological contract, developed in human resource research during the 1990s and associated mainly with Denise Rousseau, who described it as individual beliefs about reciprocal obligations between an employee and an employer. What makes it useful is that it is subjective and held by one person. It does not require both sides to agree. It only requires one side to believe an undertaking was given, which is enough to change how that side behaves.

We moved the concept out of the employment relationship and applied it between a construction firm and the parties immediately upstream and downstream of it, a client on one side and a specialist contractor on the other. Relational contract theory is the more familiar tool in construction research and was the obvious alternative. We used the psychological contract instead because relational contract theory explains conduct through norms sustained by the prospect of future exchange, while the question here was narrower: what do parties believe they have been undertaken to, and what do they then do about it?

One survey, four focus groups

Both papers rest on the same evidence. A questionnaire went to 109 construction managers in Australia, selected purposively for direct experience of project delivery, and 53 valid responses were returned. The respondents were senior: over 22 per cent had more than 30 years in project delivery, and 42 per cent described themselves as project managers. Around 62 per cent worked in the private sector. The projects they reported on were mostly commercial buildings, then infrastructure and education, and most were valued between A$1 million and A$100 million. Only about one in ten used a collaborative form of procurement; the rest were traditional or integrated.

Four focus groups sat around the survey. The first two, with Tier 1 contractors in New South Wales, shaped the instrument. The last two reviewed the results and are the source of most of what is interesting in the 2017 paper. One feature of the design matters throughout: each respondent rated their perception of a nominated partner, so every figure below is one party's reading of the other rather than a matched pair.

The model

The 2016 paper sets the structure out in three blocks.

Enabling conditions come first: how long the relationship has been in place, whether future tangible outcomes such as further work appear to be forthcoming, whether resources and support will be provided, and whether the role on offer carries autonomy and responsibility.

The psychological contract sits in the middle, in two halves. The unwritten agreement is what a party believes it is owed: trust, commitment, and good faith and fair dealing. The unwritten contracting behaviours are what parties do about it: initiating (taking the trouble to understand the other party's business), signalling (giving early notice of changes to schedule, team or quality), and disclosing (revealing sensitive information about oneself). Satisfaction with the partner sits at the end. Five hypotheses connect the blocks, and all five were supported by Pearson correlations.

Which links are strong

That all five were supported is not the interesting part, since with a single instrument of this kind the general result was always likely. The useful reading is in which correlations are strong and which pairs turn out not to be related at all.

Trust is the hinge. It has the strongest association with the partner's satisfaction of anything measured (r = 0.759) and the strongest association with disclosure (0.755). Commitment and good faith and fair dealing sit well below it on both counts.

Of the four enabling conditions, only one is associated with every component of the model. It is not the length of the relationship and not the promise of more work, but the character of the role offered: whether it is meaningful, whether it is challenging, and whether it carries real responsibility. That is largely within a client's gift and costs little to provide.

The length of the relationship is associated with trust, commitment and disclosure, and not with good faith and fair dealing.

Time in a relationship produces candour. It does not produce fair dealing.

The prospect of further work behaves differently again. It is associated with commitment and with initiating behaviour, and with nothing else. Parties who expect more work will invest in understanding their partner, but they do not become more trusting or more forthcoming about themselves.

What was present, and what was not

The 2017 paper asks the prior question: are these things present at all? Each item was tested against the midpoint of the scale, and two results are worth separating.

Undertakings about conduct scored highest in the survey. Items on a collaborative working environment, candid feedback, honest treatment and professional collegiality returned means between 3.72 and 3.96. Trust scored much lower. Of seven trust items, only two cleared the midpoint significantly: that the partner is reliable and does what it says (3.38), and that the partner has high integrity (3.36). Stronger statements, such as being able to trust the partner completely, did not. Undertakings about how parties will behave are readily perceived; confidence that they will be honoured is a good deal thinner.

Two items fell significantly below the midpoint, meaning respondents actively disagreed with them. Their partners had not undertaken to provide incentives linked to their own performance (2.70), and had not undertaken to fund training for the respondent's employees (2.62). The unwritten agreement covers conduct. It does not extend to money, or to building the other party's capability.

The focus groups then rejected the vocabulary. Participants held that "promise" was too strong a word for the industry and preferred "implied undertaking". One observed that offering a good price is a long way from any guarantee of future work. They also described a three tender rule, under which a client will periodically put work back out to a more stringent tender as a check on value for money, whatever the state of the relationship. That is where the psychological contract stops, and the parties know it.

The relationships were short. Nearly two thirds had been in place for three years or less, reported by people whose own careers are measured in decades. Whatever is being built between firms is being rebuilt often.

What follows for practice

Three things follow, and none of them requires a change of contract.

When selecting a partner, past dealings and the role on offer both count. A track record of commitment indicates how a firm is likely to behave next time, and a role carrying autonomy and responsibility is the condition most consistently associated with unwritten agreement and behaviour on both sides.

When managing a live relationship, treat disclosure as the exposed position. It carries the strongest link to trust and a strong link to satisfaction, but a firm that opens up its methods, its pricing logic or its internal weaknesses has made itself vulnerable, and misuse of what it disclosed is the classic breach. Disclosure that is not reciprocated does not last.

Do not expect fair dealing to accumulate on its own. Because it is unrelated to the length of the relationship, it has to be established deliberately and early, through conduct rather than through time served.

What the papers do not settle

The sample is 53 responses drawn purposively from one country, which limits how far the results travel. The analysis is correlational, so direction is untested: unwritten agreement may produce satisfaction, or satisfied partners may simply report their relationships warmly. The outcome variable is the respondent's estimate of the partner's satisfaction rather than the partner's own report, which is the weakest point in the design. There is no cost, time or quality data, so the model connects perceptions to perceptions and stops there.

Two further qualifications should be stated plainly. The instrument returned a Cronbach's alpha of 0.961, which is usually reported as high reliability but at that level also indicates substantial overlap between items, so part of the correlation between constructs is a property of the questionnaire rather than of the industry. And the 2016 paper reports that the hypotheses are proved. Pairwise correlations are consistent with the model; they do not test it as a whole, and a structural model on a larger sample would be needed before that word is warranted.

The data were collected around 2015. Collaborative procurement has become more common since, and a repeat today would draw on a different mix of projects.

Journal papers discussed on this page