Revisiting the Balassa-Samuelson model with markup variations

The Balassa-Samuelson model (BS hereafter) has achieved workhorse status in the analysis of trends observed in industrialized countries regarding the relative price of nontradables (Balassa [1964] and Samuelson [1964]). The model delivers two important testable implications. First, the relative price of nontradables is strictly proportional to the productivity differential between traded and non traded goods sectors. In other words, an increase by 1% in relative productivity raises the relative price by 1%. Second, the relative price being totally fixed by supply-side considerations, government spending shocks leave it unchanged in the long run. This paper empirically revisits the relationship between the relative price of nontradables and exogenous shocks to productivity and government spending. Our empirical evidence, by confirming earlier studies, casts doubt on the BS model's predictions. We find that a rise in the cross-sectoral productivity differential by 1% raises the relative price of nontradables by only 0.70%, providing little support for the proportionality hypothesis implied by the BS model (see Canzoneri et al. [1999] and Kakkar [2003]). Importantly, we also report an increase in the relative price by around 1% in response to a positive government spending shock equal to 1% of output, in line with evidence documented by De Gregorio et al. [1994b] and Balvers and Bergstrand [2002] among others. We propose an explanation for these observed effects based on variations in the composition of demand for nontradables which provide an incentive for firms to change markups