From the just prior preceding footnote in that same chapter:
I will admit that the terms here used are a bit weird to us (I can’t remember the last time I used a shilling, myself), so let’s reduce this to a unit we’ll call Simoleons, and figure this out. BTW, I quoted the whole thing so as to demonstrate that an honest person honestly arguing without any bias would grant the second part of that note’s veracity as well as the first part.
But getting to our Friedman beef, let’s say that at first at each stage, from raw material miner to component maker to assembler, some widget is priced at a cost of 100 at each value-added stage, and that each company is seeking a 5% profit, and so prices its product to the next stage up the chain accordingly. Thus, the total price at the end would be 100+(100*.05)=105, first, then 105+100+((105+100).05), or 215.25, and finally 215.25+100+((215.25+100).05), or 331.0125.
Now increase the wage cost such that the all-in cost at each stage rises by 5 simoleons, thereby increasing the all-in cost at each stage by 5%. So, you get 105+(1050.05)=110.25, then 110.25+105+((110.25+105)0.05)=226.0125, and finally 226.0125+105+((226.0125+105).05)=347.56313.
Increase profit by 5% to 10%, however, while leaving the cost at each stage at 100, and things work out quite differently: 100+(100.1)=110, first, then 110+100+((110+100).1), or 231, and finally 231+100+((231+100).1), or 364.10.
That’s the difference between an arithmetic and a geometric increase in price, and the difference in effect between (assuming that entire cost of 100 was wages) a 5% increase given to labor, as opposed to capital deciding it needed to raise its rate of return from 5% to 10%.
Finally, as a general observation, the rate of return demanded by capital tends to rise and fall with interest rates. As interest rates fall, so too does the rate of return demanded on risk capital, which will always be, in the US, the Treasury rate for the time horizon in question plus some premium. Vice versa as interest rates rise. So, generally, rates of return rise and fall at different stages of processing together, which gives rise to the compounding effect you see above. Labor is not nearly as educated about getting a return on its investment in sweat and artifice, and even if it were it wouldn’t really be capable, except under the most extreme conditions, of demanding that its rate of return rise and fall with the general rate of interest, or inflation, or anything else. Thus, in real life, the miners might get a raise, but the manufacturing workers working on the components to our widgets won’t, or those workers will, but the final assemblers won’t. So in the above two examples, it’s far more likely that the general rate of return demanded by capital will rise at all stages of the value-added process than that labor at all stages will demand a similar increase in its return.
Of course the reverse is also true.