We investigate the ability of workhorse structural change models in accounting for business cycle properties of an economy. These models reproduce the long-run shift of resources towards services by imposing a large income elasticity of services consumption. However, this large elasticity often makes the services sector highly volatile, which is at odds with the stylized fact that manufacturing consumption is more volatile than services consumption. We consider four different preferences specifications: i) Herrendorf, Rogerson and Valentinyi (2014), ii) Boppart (2014), iii) Comin, Lashkari and Mestieri (2021) and iv) Alder, Boppart and M¨uller (forthcoming), paired with standard sectoral production functions with random TFP shocks. In each case, we estimate preference parameters using long-run structural change data, and common TFP processes that depends on observed relative prices. Our findings confirm that in most cases services is more volatile than manufacturing, and only Comin, Lashkari and Mestieri (2021) can account for a volatility of service smaller than manufacturing.
