Adaptation towards reference values: A non-linear perspective

Using large-scale panel data, we examine the dynamics of adjustment towards reference points for key workplace attributes. We discover that an Exponential Smooth Transition Autoregressive (ESTAR) model is superior to a linear model in characterizing such a process. The speed of adjustment increases non-linearly with the distance from reference points and adjustment is faster for job satisfaction shocks compared to shocks in earnings or work hours. Our findings lend further credence to the relative utility hypothesis and highlight the inherently non-linear nature of the dynamic path of adjustment towards reference points, a previously neglected issue in the adaptation literature.