Some new insights on the empirics of Goodwin's growth-cycle model

Abstract

Vector Autoregressive (VAR) models have been applied for a long time now to study profit-squeeze cycles, most of the time using problematic Hodrick-Prescott (HP) filtered time series. In a recent paper, Hamilton (2018) has provided a simple alternative that overcomes the main drawbacks of the HP procedure. In order to evaluate the empirical relevance of the profit-squeeze mechanism, we compare both methodologies using quarterly data for the United States after the Second World War. It is shown that using the Hamilton detrending method significantly increases the magnitude as well as the time length of the response. Furthermore, we present an extension of Goodwin’s (1967) growth-cycle model that includes employment rates, income distribution, and capacity utilisation as endogenous variables. We demonstrate analytically that the system always admits a family of periodic solutions. The model is estimated using the Autoregressive Distributed Lag (ARDL) approach. Through numerical simulations and making use of our estimations, we confirm that fluctuations are persistent and bounded.