Monday, December 9, 2013

Successful Economic Forecasting with the Bayesian Method

Gupta and Kabundi (2010) started with an interesting question on which macroeconomic models are most likely to predict economic growth and success. Decision-makers that have tools are better able to make current decisions that are likely to foster greater growth in the future. The researcher used emerging markets of South Africa but these same models may apply to economic hubs and the factors that predict their success. 

Models are simply explanations that attempt to predict activities within the environment. Some models are more successful than others. Success is determined through a process of validity where multiple researchers over a period of time analyze the same phenomenon over and over in multiple ways to determine if the model makes sense. 

Common data points in measuring economic development include per capita growth rate, consumer price index (CPI), inflation, the money market rate, and the growth rate of nominal effective exchange rates. These data points often work their way into various models in an effort to create and develop some predictability. 

Bayesian VAR (BVARs) are based upon the Bayesian Method which is a subjective probability analysis used in a number of different fields. It is a rational decision making regression analysis for updating beliefs. In economics, the methods use monthly, yearly and other time based measurements to help determine the vector and trajectory of actions. It provides a method of blending new information with prior beliefs. 

BVAR models incorporate a greater amount of data than a number of other common models. The authors found that the BVARs have more predictability and would be beneficial for evaluating economic growth. Administrators that consider these models may find an additional tool for understanding and managing economic hubs. 

Gupta, R. & Kabundi, A. (2010). Forecasting macroeconomic variables in small open economy: a comparison between small-and large-scale models. Journal of Forecasting, 29 (2).

No comments:

Post a Comment