Andrew Atkeson

UCLA Department of Economics

COVID-19 Working Papers and Code are available on this page

Behavior and the Dynamics of Epidemics Brookings Papers on Economic Activity March 2021

This paper examines the impact of behavior and further non-pharmaceutical interventions in reducing the death toll from COVID in the United States over the short and long term. The page linked above offers a summary of the paper and access to the conference draft.

My presentation at the BPEA meeting is the first one in this session video.

A summary of the conference paper is here

Matlab files are here


Summary of my research on COVID from the NBER Reporter


A Parsimonious Behavioral SEIR Model of the 2020 COVID Epidemic in the United States and the United Kingdom

I present a behavioral epidemiological model of the evolution of the COVID epidemic in the United States and the United Kingdom over the past 12 months. The model includes the introduction of a new, more contagious variant in the UK in early fall and the US in mid December. The model is behavioral in that activity, and thus transmission, responds endogenously to the daily death rate. I show that with only seasonal variation in the transmission rate and pandemic fatigue modeled as a one-time reduction in the semi-elasticity of the transmission rate to the daily death rate late in the year, the model can reproduce the evolution of daily and cumulative COVID deaths in the both countries from Feb 15, 2020 to the present remarkably well. I find that most of the end-of-year surge in deaths in both the US and the UK was generated by pandemic fatigue and not the new variant of the virus. I then generate fore- casts for the evolution of the epidemic over the next two years with continuing seasonality, pandemic fatigue, and spread of the new variant.

Matlab code for this project is available in these files. The first two files below are MATLAB live scripts. The third is a MATLAB script.

US code, UK code, function file, US data, UK data


Behavior and the Transmission of COVID-19 (Long Version)

Short version is forthcoming in the AEA Papers and Proceedings

with Karen Kopecky and Tao Zha

We show that a simple model of COVID-19 that incorporates feedback from disease prevalence to disease transmission through an endogenous response of human behavior does a remarkable job fitting the main features of the data on the growth rates of daily deaths observed across a large number countries and states of the United States from March to November of 2020. This finding, however, suggests a new empirical puzzle. Using an accounting procedure akin to that used for Business Cycle Accounting as in Chari et al. (2007), we show that when the parameters of the behavioral response of transmission to disease prevalence are estimated from the early phase of the epidemic, very large wedges that shift disease transmission rates holding disease prevalence fixed are required both across regions and within a region over time for the model to match the data on deaths from COVID-19 as an equilibrium outcome exactly. We show that these wedges correspond to large shifts in model forecasts for the long-run attack rate of COVID-19 both across locations and over time. Future research should focus on understanding the sources in these wedges in the relationship between disease prevalence and disease transmission.


How should we interpret the cross country or region relationship between cumulative deaths and lost economic activity from COVID?

comment on "Macroeconomic Outcomes and COVID-19: A Progress Report" by Jesus Fernandez Villaverde and Charles I. Jones

How should we interpret empirical findings on the cross-country or cross-regional relationship between cumulative deaths and lost economic activity from COVID and/or the Spanish Influenza of 1918-19? Should we expect to see higher cumulative deaths associated with higher cumulative lost activity? Or should we expect the reverse: higher cumulative deaths associated with lower cumulative lost activity? I use a simple behavioral SIR model to show that the answer to this question is driven by the source of heterogeneity across regions or countries. If regions differ in the transmissibility of the disease holding behavior fixed due to natural or cultural factors determined pre-pandemic, then the relationship between cumulative deaths and lost activity should be upward sloping. If regions differ instead in the elasticity of their behavioral response to prevalence of the disease, then the reverse should be true; higher cumulative deaths associated with lower cumulative lost activity.


Four Stylized Facts About COVID-19

with Karen Kopecky and Tao Zha

We document four facts about the COVID-19 pandemic worldwide relevant for those studying the impact of non-pharmaceutical interventions (NPIs) on COVID-19 transmission. First: across all countries and U.S. states that we study, the growth rates of daily deaths from COVID-19 fell from a wide range of initially high levels to levels close to zero within 20-30 days after each region experienced 25 cumulative deaths. Second: after this initial period, growth rates of daily deaths have hovered around zero or below everywhere in the world. Third: the cross section standard deviation of growth rates of daily deaths across locations fell very rapidly in the first 10 days of the epidemic and has remained at a relatively low level since then. Fourth: when interpreted through a range of epidemiological models, these first three facts about the growth rate of COVID deaths imply that both the effective reproduction numbers and transmission rates of COVID-19 fell from widely dispersed initial levels and the effective reproduction number has hovered around one after the first 30 days of the epidemic virtually everywhere in the world. We argue that failing to account for these four stylized facts may result in overstating the importance of policy mandated NPIs for shaping the progression of this deadly pandemic.


Estimating and Forecasting Disease Scenarios for COVID-19 with an SIR Model

With Karen Kopecky and Tao Zha

This paper presents a procedure for estimating and forecasting disease scenarios for COVID-19 using a structural SIR model of the pandemic. Our procedure combines the flexibility of noteworthy reduced-form approaches for estimating the progression of the COVID-19 pandemic to date with the benefits of a simple SIR structural model for interpreting these estimates and constructing forecast and counterfactual scenarios. We present forecast scenarios for a devastating second wave of the pandemic as well as for a long and slow continuation of current levels of infections and daily deaths. In our counterfactual scenarios, we find that there is no clear answer to the question of whether earlier mitigation measures would have reduced the long run cumulative death toll from this disease. In some cases, we find that it would have, but in other cases, we find the opposite — earlier mitigation would have led to a higher long-run death toll.