Stata does econometrics. Stata does Bayesian statistics. Stata 17 combines both to do Bayesian econometrics.
Bayesian econometrics uses Bayesian principles to study economic relationships. How likely is it that an extra year of schooling will increase wages? Are those who participate in a job-training program more likely to stay employed for the next five years? What is the probability of default for a low default portfolio? Such questions can be answered naturally within the Bayesian paradigm, which can estimate the probability of any hypothesis of interest.
One of the appeals for using Bayesian methods in econometric modeling is to incorporate the external information about model parameters often available in practice. This information may come from historical data, or it may come naturally from the knowledge of an economic process. For instance, income elasticity may be known to be less than 1 for some countries, or autocorrelation is known to be between -1 and 1. Either way, a Bayesian approach allows us to combine that external information with what we observe in the current data to form a more realistic view of the economic process of interest.
You may ask: What if I do not have any external information? No problem. Without any informative prior knowledge, the results will be similar to those you would have obtained using classical econometrics methods. But their interpretation may be more intuitive. For instance, a 95% credible interval—a Bayesian counterpart of a confidence interval—can be interpreted as a range in which a parameter lies with a 0.95 probability.
There is another reason Bayesian econometrics may be appealing in the absence of strong external knowledge. Econometrics models often describe complex economic theories and thus tend to have many parameters—often so many that it becomes infeasible to fit the models without incorporating some information about model parameters. In such situations, Bayesian econometrics modeling can provide a balance between what’s observed in the data and what’s reasonable to assume about model parameters to obtain reliable inference.
For instance, vector autoregressive (VAR) models are known to have many parameters relative to the data size. Bayesian analysis of these models introduces specialized priors that allow you to obtain more stable parameter estimates.
Also, dynamic stochastic general equilibrium (DSGE) models are known to have parameters that have direct economic interpretations and often have logical bounds that can be easily incorporated by a variety of prior distributions.
You can find Bayesian VAR models and Bayesian linear and nonlinear DSGE models among the new Bayesian econometrics features of Stata 17, and much more:
Bayesian VAR models
Bayesian IRF and FEVD analysis
Bayesian dynamic forecasting
Bayesian longitudinal/panel-data models
Bayesian linear and nonlinear DSGE models
You may also be interested in other existing features, including
Bayesian linear models
Bayesian generalized linear models
Bayesian zero-inflated models
Bayesian survival (duration) models
Bayesian sample-selection models
Bayesian multilevel models
Bayesian autoregressive models
Extensive Bayesian inference is available after estimation, including Markov chain Monte Carlo (MCMC) diagnostics, posterior summaries of linear and nonlinear functions of parameters, interval hypothesis testing, and model comparison using Bayes factors