Perhaps one of the flaws is less with the structure of the models and more with the stochastic specifications used to give the models empirical content. Most of the models I have seen are fed shocks that follow a gaussian distribution. But we have known for half a century now (John Tukey, Benoit Mandelbrot) that real-life economic and financial shocks are fat-tailed. Extreme events will occur far more frequemtly than a gaussian distribution will predict.
The primary problem in my view is not that economic models are flawed (although that too) but rather that the average economist has a rather poor mastery of statistics. That leads to serious underestimation of volatility and to parameter estimates that are very unstable.
Economics in Denial
Perhaps one of the flaws is less with the structure of the models and more with the stochastic specifications used to give the models empirical content. Most of the models I have seen are fed shocks that follow a gaussian distribution. But we have known for half a century now (John Tukey, Benoit Mandelbrot) that real-life economic and financial shocks are fat-tailed. Extreme events will occur far more frequemtly than a gaussian distribution will predict.
The primary problem in my view is not that economic models are flawed (although that too) but rather that the average economist has a rather poor mastery of statistics. That leads to serious underestimation of volatility and to parameter estimates that are very unstable.