Externalities, incentives and strategic complementarities: understanding herd behavior in IT adoption

Abstract

Herd behavior arises in many instances of information technology (IT) adoption. This study examines the economic and behavioral bases for herd behavior and decision conformity. We investigate the roles of payoff externalities, observational learning and managerial incentives in influencing IT adoption decision-making. Our study underscores the benefits of viewing various drivers of IT adoption herding in a unified framework focusing on equilibrium coordination under strategic complementarities. Motivated by the recent advance in behavioral economics and behavioral game theory, our study relates IT adoption herding to a range of individual-level problems, including managerial incentives, managerial behavioral biases and limited rationality. We develop a coordination game of IT adoption within the unified framework. Our analysis of the game demonstrates that, under strategic complementarities, behavioral biases or incentive problems of a small minority of decision-makers may dramatically impact aggregate outcomes.

Appendix: propositions and proofs

Proposition 1 (Two IT Adoption Scenarios)

When α > 1, every decision-maker in the IT adoption game adopts the technology in the payoff-dominant equilibrium, and when  α < 1, no one adopts the technology in the unique equilibrium. This equilibrium is socially-inefficient when 0.5 < α < 1.

Proof

If a decision-maker with the adopt cost $\hat{\beta }$ adopts the technology in any equilibrium of this coordination game, all other decision-makers with adoption cost below $\hat{\beta }$ will adopt the technology in the equilibrium. Otherwise some decision-makers will play the game in an irrational manner. However, when α < 1, we cannot construct any equilibrium with an adoption threshold of $0 < \hat{\beta } < 1$. In this scenario, every decision-maker in the IT adoption game will adopt the technology in the payoff-dominant equilibrium. Similarly, we cannot construct any equilibrium with an adoption threshold of $0 < \hat{\beta } < 1$ in situations where  α < 1. It is also easy to verify that adoption by all decision-makers is not a viable equilibrium. So the unique equilibrium entails no adoption by any decision-maker under this scenario. If all decision-makers adopt the technology, the average payoff of all decision-makers is α − 1/2. So this unique non-adoption equilibrium is not socially-efficient when 0.5 < α < 1. □

Proposition 2 (Fully-Rational Decision-Makers’ IT Adoption)

When 0.5 < α < 1 and a small portion of decision-makers, 0 < η < 1, adopt the technology because of their limited rationality, a portion, 0 < λ < 1, of the fully-rational decision-makers will also adopt the technology in equilibrium, when λ satisfies α [η +(1 − η)λ] = λ. Compared to the benchmark scenario where no rational decision-makers adopt the technology, rational adopters’ average payoff will be higher by λ/2, and the average payoff of adopters with limited rationality will increase by αλ(1 − η).

Proof

As illustrated in Proposition 1, any fully-rational decision-maker’s equilibrium strategy entails the characterization of an adoption threshold. Under this scenario, we can construct an equilibrium with an adoption threshold λ that satisfies α[η + (1 − η)λ] = λ. This equilibrium condition guarantees that any rational decision-maker with adoption cost less than λ will be better off by adopting the technology, and that those with adoption cost above this threshold will have no incentive to adopt the technology. Moreover, it is easy to verify that 0 < λ < 1 when 0.5 < α < 1. Consequently a portion, 0 < λ < 1, of the rational decision-makers will adopt the technology in the equilibrium. On average, these rational adopters will be better off by λ − λ/2 = λ/2, and those adopters with limited rationality are better off by the strictly positive value, α [η + (1 − η)λ] − 1/2 − (αη − 1/2) = αλ(1 − η). □

Proposition 3 (Conditions for Excess Inertia)

While the payoff-dominant equilibrium entails adoption by every decision-maker whenever α > 1, excess inertia—when everyone stays with the old technology—may occur when it is known that a small portion, 0 < δ < 1, of decision-makers will stay with the old technology because of their incentive problems. This inefficient situation arises when either α is sufficiently close to 1 or δ is sufficiently large, that is, whenever α (1 − δ) < 1.

Proof

As illustrated in Proposition 1, any equilibrium strategy entails the characterization of an adoption threshold. Suppose that we can construct an equilibrium with an adoption threshold $\hat{\beta }$ in this scenario. To guarantee that any rational decision-maker with adoption cost less than $\hat{\beta }$ is better off by adopting the technology, we must have α (1 − δ) $\hat{\beta }$ >$\hat{\beta }$. We can verify that this threshold is not obtainable whenever α (1 − δ) < 1. The payoff-dominant equilibrium entails adoption by every decision-maker when this condition is satisfied. This situation involving excessive inertia is caused by some decision-makers’ incentive problems. This leads to overall inefficiency because the average payoff for all decision-makers decreases by α − 1/2. □