By definition, dividends are the classic method for remunerating equity holders. Therefore, one of the expectations is a steady payout from the free cash flow that the company produces.
However, it’s quite common to only release money to the investors after analyzing possible projects – with positive NPV, of course. It’s quite rational to finance part or totally that project with the FCF capital, not only because it’s the cheapest form of financing but also because more return will be left for the shareholders ( at a greater risk, obviously).
One interesting point about this theory is that since the money is implicitly reinvested, the management requires a much tighter control. Managers might be compelled to invest in projects that would not meet the return rate required for a bond or equity finance. This would result in decisions that the investors wouldn’t likely agree – the risk that they could incur would not be compensated with enough remuneration.
Another interesting point is based on the signaling theory of the dividends. It’s much easier for managers to increase dividends than to lower them. Markets have a very bad outlook on lowering dividends, when combined with a poor performance year might result on a tighter effort to release more funds from the company – possibly even skipping a few good investment opportunities.
There’s also another important factor to consider in any dividend policy – the taxation system. Investors, as rational agents, will prefer a payout solution that will be the most tax-effective possible. This might mean that some will prefer no payout at all – the free cash flow will remain in the company and the result will be incorporated in the share price (and, consequently, in the portfolio valuation). They might also prefer a stock repurchase, where they’ll get part of the free cash flow as well.
Years after writing this essay, retail investors are increasingly flocking to ETFs, many with a clear preference for Accumulating funds - meaning that any dividend payouts will be reinvested within the fund, thus not creating a taxable event for the investor. This allows for a supposedly greater compound growth opportunity by deferring the tax payment into the future.
But at the end of the day all of this will be somewhat subjective to the shareholder’s preference – commonly referred as the clientele effect. The company exists to please the investors, so this particular shareholder will have a lot of pressure in the dividend policy. This is easily observable when you jump from industry to industry and from country to country.