Predicting spatial processes often involve using many, many parameters. That approach requires using Bayesian methods — or something with the same effect — to shrink the predictions back to something more reasonable. I’m going to use something simpler, regression. No, not a ridge estimator either. Rather, by constructing a particular explanatory variable, I can achieve much the same effect at the cost of just a few parameter estimates. My talk will cover this trick as well as show a variety of maps of the evolution of default rates in the US. Hope this is enough to lure you back inside from the beach next week!