That would be long indeed (100 years or so for the US)
Keynesian theories of the business cycle start from the notion that the changes in income equilibrate savings to investment, and the level of activity is determined by effective demand. In that sense, the economy can fluctuate in the long run, with wage and price flexibility, around a normal position that is below full utilization of labor and capital. Unemployment is the norm. Keynes was not concerned in the GT with business cycles per se, even though he discussed the issue at the end of the book. His main concern was with what he referred to as unemployment equilibrium. It is clear in his terminology, unemployment equilibrium, instead of disequilibrium as Patinkin suggested would be more appropriate, that Keynes meant that unemployment was not the result of some type of imperfection, wage or interest rate rigidities, for example, which became the leading explanation for unemployment among the mainstream Neoclassical Synthesis or New Keynesian authors.
Keynes suggested that it was the cyclical changes in investment, which he associated with the marginal efficiency or productivity of capital, in marginalist fashion, that determined business cycle fluctuations. His correspondence with Harrod on the growth model, and Tinbergen's method, suggests that he disliked the accelerator, which was probably too mechanical for him. In that sense, cycles in Keynes own conception depended on shocks, which were associated to the state of long-term expectations, and the state of confidence.
Kalecki, which independently from Keynes advanced the Principle of Effective Demand (PED), developed an early theory of the cycle based on the interaction of Keynes’ multiplier process with the concept of the accelerator. Kalecki emphasized the role of time lags between the placing of investment, the demand for new equipment, and the delivery of the new equipment, indicating the dual role of investment as part of demand, but as creating productive capacity in the future. Further, Kalecki suggested that investment orders are a positive function of autonomous demand and a negative function of the existing capital stock.
Assuming that expectations about future demand are high, then investment would increase and through the multiplier effect it would have a reinforcing effect on income. The increase in income, in turn, would lead, according to the accelerator, to an increase in investment, leading to an economic boom. However, as investment increases, eventually new investment orders would exceed the replacement requirements, and the capital stock would also rise. This would have a negative effect on the rate of increase in investment, and new investment orders would slowdown first, and then decrease. The decreasing orders would have a negative impact on demand, and through the multiplier, lead to a reduction in the level of income, creating the conditions for a recession. The falling income would imply lower investment, following the accelerator, and even further collapse of income. In the depression, investment orders would collapse and at some point they would fall below the replacement requirements associated to depreciation, leading to a reduction in the stock of capital. Finally, the falling stock of capital would make the need for investment inevitable, and that would lead to more demand and a recovery.
The essential mechanism of the business cycle in the Kaleckian model was connected to the lags between the demand effect and the capacity effect of investment. Kalecki assumed that shocks would provide the initial spark for the business cycle, and the multiplier-accelerator mechanism would keep it going. He noted also that only under very specific circumstances would the cycle recur, and that additional shocks would be necessary to avoid a dampened cycle.
Kaldor developed a model, later formalized by Goodwin and Hicks, which allowed for the economic cycle to recur even in the absence of external shocks. The central difference in the Kaldorian model was the introduction of non-linear investment and savings functions. The idea was not to deny the existence of stochastic shocks or time lags, but to demonstrate that the economic system would also fluctuate in their absence, and that in a broad sense the capitalist system was inherently unstable.
This view of the cycle was dominant and basically accepted by the mainstream Neoclassical Synthesis in the 1950s and 60s. Yet, the notion of an endogenous cycle was one of the first victims of the attack on Keynesian economics in the 1970s. Eventually leading to the Real Business Cycles (RBC) school. Today most macroeconomic textbooks do not even mention the accelerator, even though the empirical evidence for it is overwhelming.