Are VAR models evidence for a profit-led Goodwin pattern?

Vector auto-regressive (VAR) models remain highly influential in the macroeconomic literature regarding the existence and explanation of the macroeconomic Goodwin pattern – a counter-cyclical movement in capacity utilization × labor share space. The study by Barbosa-Filho and Taylor (2006), which provided the theoretical backbone of the neo-Goodwin model, demonstrated through vector auto-regression that the theoretically required profit-led nature of demand was empirically prevalent in the Goodwin pattern in the US, and further studies follow its approach. This is often cited in the debate about whether capitalism has profit-led or wage-led demand characteristics, with implications for macroeconomic policy. In this paper, I replicate the VAR approach and extend it to recent years. Through additional econometric techniques not currently employed in the literature, I demonstrate that the supposed profit-led demand derivatives are statistically insignificant. Noting further issues with robustness, I critically analyze the use of VAR models for this purpose, suggesting that other methods need be employed in the debate between profit-led and wage-led demand regimes and the Goodwin pattern.