Outsourcing to distant countries used to be about tapping into the lower costs and specialized skills of foreign labor markets, such as by making textiles in Bangladesh and electronic parts in Taiwan or Singapore. Now, with new information technologies and the digitization of the economy, it has become increasingly common to outsource services as well, including the development of information systems. India in particular, with Bangalore as its Silicon Valley, has become the leading destination for such projects, catering to nearly 60 percent of global requirements for offshore information systems development. And yet, an astounding 50 percent of such projects fail outright or do not meet client requirements. However, HEC researcher Shirish Srivastava noticed that the rate of failure was going down in the U.S., which is a more mature market than France for offshore development, and turned to control theory to analyze this improvement in performance.
Distinguishing control mechanisms
It is through the exercise of control that companies can motivate vendors to deliver applications that meet the needs of their clients. HEC’s Shirish Srivastava worked alongside the National University of Singapore’s Thompson Teo to fill in what they saw as gaps in existing control theory. While the academic literature had already examined formal and informal modes of control (such as contracts, shared norms, and so on), the two co-authors sought a more nuanced understanding of control. They described the process mechanisms at work within different modes of control; i.e. how the control is enacted as opposed to what control is used. Almost all outsourced client-vendor relationships function under the broad framework of a formal contract (the structural control mechanism), which defines the type of work expected, the delivery date, and so on. The contract may go into great detail or not, but the involved parties choose how closely they will follow its specifics. If it is implemented to a tee, then the contract governance is mechanistic. If the parties instead rely on mutual trust to deal with operational situations as they emerge, then the mechanism is one of relational governance. In that case, if for example the project is delayed, says Shirish Srivastava, then client and vendor negotiate to potentially charge less or extend the deadline. In other terms, the contract may go into great detail, but the parties can choose to implement it selectively. “It’s a very simple idea, but it had not yet been conceptualized and examined in the academic literature,” adds the researcher.
The effects of governance on performance
Based on a study of 160 information systems development projects carried out in India, the researchers looked at the effect on performance of mechanistic governance, first in relation to contract specification, then coupled with relational governance. Their first finding was that mechanistic governance (the kind that does not allow deviations from pre-specified outcomes and procedures) serves as a good complement to contract specificity. As Shirish Srivastava explains, the quality of information systems tends to be defined with objective criteria (the number of allowed bugs for example). Costs, on the other hand, perhaps counter-intuitively, are a “fuzzier concept” and consequently less easily defined by contract. So where the contract is detailed, mechanistic governance, by ensuring close adherence to the contract, enhances quality and cost performance. However, relational governance is somewhat more effective to improve cost performance, because as a process it allows for more leeway, such as in the negotiation of rates for example. “In such a fast-moving industry as information systems development, a client can suddenly see that competitors are offering better prices than his vendor, which will prompt him to bargain,” explains Shirish Srivastava. “Conversely, if a vendor sees that his client is not enforcing penalties despite delays, he can see it as a favor and work harder in exchange,” he adds.
Focus on cost or performance
Last but not least, the researchers fine-tuned their results by distinguishing between the two main types of contracts in use in the industry:
1) fixed price contracts, whereby an agreed-upon sum is paid for the final product, regardless of the actual costs incurred by the vendor; and
2) time & material contracts, whereby the client pays a daily (or hourly) rate for the work provided by the vendor. In the latter case, the client bears most of the risk.
When examining the effects of relational and mechanistic governance on performance, the results were broadly similar to those of the full model. The only major difference was that relational governance seemed to have an adverse effect on cost performance for time & material projects. The reason is that such contracts tend to be used in two very different cases: either when clients have clear knowledge of the work requirements, and for all practical purposes manage the work, merely using vendors as additional human resources; or, alternatively, in pure development work, with no precise idea of the outcomes or standards. In the latter case, process control (either mechanistic or relational) does not add any value to the process.
