Chain indexing in a macro model: Aggregation and irreversibility

We investigate a switch from fixed-weighted national accounts data to chain-type data in the quarterly macroeconometric model Mona. As in other countries the idea behind this change in the national accounts data has been to reduce the substitution bias, which has grown with the amount of high-technology goods with a weak price development deviating from the average. In line with Lasky (1998) we demonstrate that the use of the chain index can also reduce some of the typical aggregation problems in macro econometric models based on traditional fixed-weight relations. Chain-type national accounts data are not perfect as e.g. their lack of additivity makes them more complex to work with. We demonstrate that also the lack of reversibility in the chain index may complicate some model results by leading to permanent effects from temporary chocks. It is not clear a priori if a switch to chain weights in the definitional relations on balance improve the model.