Unified knowledge based economy hybrid forecasting

Many synthetic composite indicators have been developed with the aim to measure micro- and macro-knowledge competitiveness, however, without any unified, easy to visualise and assessable forecasting capability, their benefits to decision makers remain limited. In this article, a new framework for forecasting knowledge based economy (KBE) competitiveness is proposed. Existing KBE indicators from internationally recognised organisations are used to forecast and unify the KBE performance indices. Three different forecasting methods including time-series cross sectional (TSCS) (also known as panel data), linear multiple regression (LMREG), and artificial neural network (ANN) are employed. The ANN forecasting model outperformed the TSCS and LMREG. The proposed KBE hybrid forecasting model utilises a 2-stage ANN model which is fed with a panel data set structure. The first stage of the model consists of a feed-forward neural network that feeds to a Kohonen's self-organising map (SOM) in the second stage of the model. A feed-forward neural network is used to learn and predict the scores of nations using past observed data. Then, a SOM is used to aggregate the forecasted scores and to place nations in homogeneous clusters. The proposed framework can be applied in the context of forecasting and producing a unified meaningful map that places any KBE in its homogeneous league, even when considering a limited data set. •Knowledge economy and competitiveness indicators are very similar and complement each other.•Forecast knowledge economies from highly related composite indicators•Unify five complex, multi-dimensional macro-knowledge indicators•Examine three different forecasting models to forecast knowledge economies•Neural network forecasting model outperformed the linear multiple regression and panel data regression techniques.•Size and locations do not matter when it comes to competitive knowledge economies.