The layman attack against economists is that they can't do macroeconomic predictions very well. But if you look at what it would require, I don't think anyone can expect them to do so.
What macroeconomic modeling attempts to do is to look what are the effects of policy changes in fiscal or monetary policy __if everything else would stay the same__. Central banks use DSGE as reasoning tool, not as accurate forecasting tool for the future. The signal (their actions) is mostly buried in the noise (all other variables).
Macroeconomic models exaggerate individual rationality and foresight, and understate heterogeneity, because that's correct thing to do with current level of knowledge. Knowledge reburied for modeling limited rationality and heterogeneity at the level where predictions become more accurate would turn economics into psycohistory in Isaac Asimov's Foundation series.
One of the most important variables affecting the future is what consumers think future will be. Consumer sentiment and its relation to changes in their income and savings is accurate only to one or two quarters. After five quarters it's just noise. Consumers don't have the data or reasoning ability that macroeconomic models have. They only have gut feelings formed by events around them and narratives in media.
Another important variable is what investors think about future investment needs based on future demand linking their predictions to consumer predictions. They are not very accurate either. Consensus forecasting and other means are just hints.
To correctly predict the macroeconomic future would require that consumers and investors could predict the future with more accuracy.
The best possible future model is agent-based computational macroeconomic model that simulates collective behaviour of future predicting and speculative agents who can read the predictions from the models predicting their own behavior. It must include several empirically discovered mental models for layman to model heterogenic agents.
Once the accuracy of the model becomes widespread, it would change the world. Speculative agents would use the model as base of their actions and forecasts. Saying accurate would require stable dynamic interaction between the model and the environment.
What macroeconomic modeling attempts to do is to look what are the effects of policy changes in fiscal or monetary policy __if everything else would stay the same__. Central banks use DSGE as reasoning tool, not as accurate forecasting tool for the future. The signal (their actions) is mostly buried in the noise (all other variables). Macroeconomic models exaggerate individual rationality and foresight, and understate heterogeneity, because that's correct thing to do with current level of knowledge. Knowledge reburied for modeling limited rationality and heterogeneity at the level where predictions become more accurate would turn economics into psycohistory in Isaac Asimov's Foundation series.
One of the most important variables affecting the future is what consumers think future will be. Consumer sentiment and its relation to changes in their income and savings is accurate only to one or two quarters. After five quarters it's just noise. Consumers don't have the data or reasoning ability that macroeconomic models have. They only have gut feelings formed by events around them and narratives in media.
Another important variable is what investors think about future investment needs based on future demand linking their predictions to consumer predictions. They are not very accurate either. Consensus forecasting and other means are just hints.
To correctly predict the macroeconomic future would require that consumers and investors could predict the future with more accuracy.
The best possible future model is agent-based computational macroeconomic model that simulates collective behaviour of future predicting and speculative agents who can read the predictions from the models predicting their own behavior. It must include several empirically discovered mental models for layman to model heterogenic agents.
Once the accuracy of the model becomes widespread, it would change the world. Speculative agents would use the model as base of their actions and forecasts. Saying accurate would require stable dynamic interaction between the model and the environment.