This paper demonstrates that the Kennedy Goodwin macroeconomic model of capital accumulation (KGM) does not reflect direct increasing return. The author presents its two versions: KGM-I with weakening roundabout increasing return and KGM-II with reinforcing roundabout increasing return. Both have the common intensive form and same asymptotically stable stationary state. KGM-II is changed to allow for direct increasing return to scale, whereby the growth rate of employment ratio positively influences the growth rate of labour productivity. If the latter effect is strong enough, the dynamic equilibrium is locally repelling and bifurcates into closed orbits. Their period is estimated. This paper supposes a closed loop control that stabilizes the oscillatory dynamics of the main macroeconomic variables, main-taining profitability and employment under direct and roundabout increasing returns. It is proved that the supposed policy would be destabilizing if the direct scale effect were power-fully negative that is not supported empirically. Simulation runs maintain analytical findings. This paper yields insights for public debate on competent pro-growth stabilization policy. Key words: growth cycle, primary income distribution, employment, stabilization policy.