A Call for Out-of-Sample Testing in Macroeconomics

Social science has come under a firestorm of criticism following the discovery that many findings cannot be replicated. This paper argues that increasing the use of out-of-sample tests would improve the reliability of findings in all fields, including “complex” fields such as macroeconomics. It illustrates the paucity of examples of this method by reviewing the literature on the government spending multiplier. Two out-of-sample tests in academic papers from the 1960s were identified, along with a procedure similar to an out-of-sample test in a forthcoming paper by Jorda and Taylor