CH. 10 THE MARGINAL PROPENSITY TO CONSUME 127

It is also obvious from the above that the employ-ment of a given number of men on public works will(on the assumptions made) have a much larger effecton aggregate employment at a time when there issevere unemployment, than it will have later on whenfull employment is approached. In the above example,if, at a time when employment has fallen to 5,200,000,an additional 100,000 men are employed on publicworks, total employment will rise to 6,400,000. Butif employment is already 9,000,000 when the additional100,000 men are taken on for public works, total em-ployment will only rise to 9,200,000. Thus publicworks even of doubtful utility may pay for themselvesover and over again at a time of severe unemployment,if only from the diminished cost of relief expenditure,provided that we can assume that a smaller proportionof income is saved when unemployment is greater; butthey may become a more doubtful proposition as a stateof full employment is approached. Furthermore, ifour assumption is correct that the marginal propensityto consume falls off steadily as we approach full employ-ment, it follows that it will become more and moretroublesome to secure a further given increase ofemployment by further increasing investment.

It should not be difficult to compile a chart of themarginal propensity to consume at each stage of a tradecycle from the statistics (if they were available) ofaggregate income and aggregate investment at suc-cessive dates. At present, however, our statistics arenot accurate enough (or compiled sufficiently withthis specific object in view) to allow us to infer morethan highly approximate estimates. The best for thepurpose, of which I am aware, are Mr. Kuznets’ figuresfor the United States (already referred to, p. 103 above),though they are, nevertheless, very precarious. Takenin conjunction with estimates of national income thesesuggest, for what they are worth, both a lower figureand a more stable figure for the investment multiplier