1. Introduction
Does learning drive the process of policy diffusion? Do governments actively observe
policy outcomes in other jurisdictions and incorporate this evidence into their own poli-
cymaking? If so, what are the mechanics of this process? A fundamental problem facing
observers of multi-state systems is the difficulty of differentiating policy adoption based upon
learning from policy adoption based upon idiosyncratic but correlated preferences, changing
political, social and economic conditions, and other factors. The problem is all the more
complicated in politics because learning agents (governments) care not only about the gen-
eral value of policies (the net revenue generated by them, say) but also about their ideological
or political implications (whether the policy is sufficiently conservative or liberal). Hence
widespread adoption of a policy in a population of states may demonstrate its general value,
its relative ideological appeal, or some combination of these factors.
Scholars have long been interested in the idea of policy diffusion, particularly within
political science. Within American federalism, studies of how policies spread across the states
were stimulated by the early work of Walker (1969) and Gray (1973), and accelerated by the
insights and approach of Berry and Berry (1990). Dozens, if not hundreds, of subsequent
studies focus on different policies and various aspects of the diffusion process across the
American states. In comparative politics, the spread of similar policies across countries
has recently received significant attention (e.g., Gilardi 2005, Simmons and Elkins 2004).
Outside of political science, the diffusion of innovations has been relevant for everyone from
farmers to firms to physicians (Rogers 1995).
Crucial to the concept of diffusion is the idea of learning from the actions of others.
1
“Innovators” experiment with new policies and processes while others “watch.” “Successful”
innovations then spread while “failures” are abandoned. To the extent that external learning
is taking place, in-depth studies of policy diffusion are very important. Diffusion across
states and localities in federal systems underscores one the benefits of decentralization—
multiple simultaneous experiments lead to better outcomes for all, without a risky new
nationwide experiment. Extensive diffusion across countries means that scholars must study
intergovernmental relations rather than isolating policymaking within one country alone, if
they are to accurately understand the public policy process.
Unfortunately, much of the previous work on policy diffusion has not provided rigorous,
differentiating evidence of learning-based diffusion, emphasizing instead the adoption of sim-
ilar policies within regions or by geographic neighbors. While consistent with learning-based
1
We set aside non-informational causes of policy diffusion here.
2
Convention