Media coverage shifts and policy overreactions: Evidence on government serial processing and information saturation

There have been two competing schools of thoughts about how governments process information: serial and parallel processing. Serial processing implies that governments suffer from information-processing bottlenecks, and that may lead to erroneous decision-making. While much theory has been written in both schools of thought, no empirical test comparing them exists. In this paper, we collect extensive newspaper and government expenditure data to formally test both hypotheses. What is nice about this work is that we identified a natural experiment in Mexico where two consecutive earthquakes were geographically distant, produced roughly the same human and economic losses, but received differentiated media coverage. Under this setting, we show that the Mexican government overreacted in its expenditure decisions, favourably biasing the victims that received more media coverage.

We investigate the causal link between societal signals and policy changes through novel data and a unique empirical setting. Using a corpus with all the opinion columns published in 2017 in the nine major Mexican newspapers (the signals) and data on the universe of individual expenditure programs (the policies), we provide quantitative evidence about government overreactions (in terms of overspending) to differentiated media coverage across various policy topics. First, we test formal models of serial and parallel information processing that relate to the debate between punctuated equilibrium and incrementalism. We find empirical support for the serial processing hypothesis. Second, we frame a natural experiment exploiting two earthquakes that took place in September of the same year. By leveraging the disparity in coverage of opinion columns across both earthquakes, we employ a difference-in-differences design and find evidence of the causal relationship between coverage shifts and policy overreactions. Our study sheds new light on a central topic in the study of policy changes and agenda-setting, and provides quantitative evidence difficult to obtain under conventional empirical frameworks.