Capital Expansion, Rate of Growth, and Employment ⑧ Evsey D.Domar Econometrica, Vol. 14, No. 2. (Apr., 1946), pp. 137-147. Stable URL: http: //links.jstor.org/sici?sici=0012-9682%28194604%2914%3A2%3C137%3ACEROGA3E2.0.CO%3B2-9 Econometrica is currently published by The Econometric Society. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. jstor's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. P Please contact the publisher regarding any further use of this work Publisher contact information may be obtained at http: //www. jstor.org/journals/econosoc. html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support @jstor.org. http: //www.jstor.org/ Sun Sep1719:58:252006
CAPITAL EXPANSION, RATE OF GROWTH AND EMPLOYMENTI By EvSEY D. DOMAR I INTRODUCTION This paper deals with a problem that is both old and new-the rela- tion between capital accumulation and employment. In economic litera- ture it has been discussed a number of times the most notable contribu on belonging to Marx. More recently, it was brought forth by Keynes nd his followers a thorough analysis of economic aspects of capital accumulation is a tremendous job. The only way in which the problem can be examinee at all in a short paper like this is by isolating it from the general eco- nomic structure and introducing a number of simplifying assumptions Some of them are not entirely necessary and, as the argument pro- gresses, the reader will see how they can be modified or removed The following assumptions and definitions should be noted at the outset: (a)there is a constant general price level;(b)no lags are pres- ent;(c)savings and investment refer to the income of the same period (d) both are net, i, e, over and above depreciation;(e) depreciation is L measured not in respect to historical costs, but to the cost of replace lent of the depreciated asset by another one of the same productive capacity; 2 (f)productive capacity of an asset or of the whole economy is a measurable concept. he last assumption, on which(e)also depends, is not entirely safe Whether a certain piece of capital equipment or the whole economy is considered, their productive capacities depend not only on physical and technical factors, but on the whole interplay of economic and instit tional forces, such as distribution of income, consumers'preferences 1 This is a summary of a paper presented before a joint session of the Econo- metric Society and the American Statistical Association in Cleveland on January 24, 1946. It contains the logical essence of the argument with relatively little economic detail. I hope to develop the latter in a separate paper to be published n one of the other economic journals. Many thanks for help and criticism go to my fell mbers of the‘ Little Seminar": Paul Baran, Svend Laursen, Lloyd A. Metzler, Richard A. Musgrave Mary S Painter, Melvin W. Reder, Tibor de Scitovszky, Alfred Sherrard, Mary Wise Smelker, Merlin Smelker, and most of all to James S. Duesenberry 2 If the original machine worth $1,000 and producing 100 units is replaced by another one worth also $1, 000, but producing 120 units, only $833. 33 will be egarded as replacement, and the remaining $166.67 as new investment. A simi lar correction is made when the new machine costs more or less than the original one. The treatment of depreciation, particularly when accompanied by sharp technological and price changes, presents an extremely difficult problem. It quite possible that our approach, while convenient for present purposes, may give rise to serious difficulties in the future 137
138 EVSEY D. DOMAR wage rates, relative prices, structure of industry, and so on, many of which are in turn affected by the behavior of the variables analyze here. We shall nevertheless assume all these conditions as given and shall mean by the productive capacity of an economy(or an asset)its total output when all productive factors are fully employed under these conditions. 3 The economy will be said to be in equilibrium when its productive capacity P equals its national income Y. Our first task is to discover ne conditions under which this equilibrium can be maintained, or more precisely the rate of growth at which the economy must expand in order to remain in a continuous state of full employment The idea that the preservation of full employment in a capitalist economy requires a growing income goes back(in one form or another at least to Marx. It has been fully recognized in numerous studies (recently made in Washington and elsewhere) of the magnitude of gross national product needed to maintain full employment. But though the various authors come to different numerical results, they all approach their problem from the point of he size of the labor force. The labor force (man-hours worked) and its productivity full employment is to be maintained, national . ala or another, and if the combined rate For practical relatively short- run purposes this is a good method, but its analytical merits are not high, because it presents ally incomplete system: since an increase in labor force or in its productivity only raises productive capacity and does not by itself generate income(similar to that produced by investment), the demand side of the equation is missing. Nor is the difficulty disposed i of by Mr. Kalecki's method according to which capital should increase proportionally to the increase in labor force and its productivity. As) Mrs. Robinson well remarked, "The rate of increase in productivity of labor is not something given by Nature. 5 Labor productivity is not a function of technological progress in the abstract, but technological progress embodied in capital goods, and the amount of capital goods It should undoubted roductive capacity, but i prefer to leave the matter open, because a more pre- cise definition is not entirely necessary in this paper and can be worked out as See his essay," Three Ways to Full Employment""in The Economic of full Employment, Oxford, 1944, p. 47, and also his"Full Employment by Stimulating Private Investment?"in Oxford Economic Paper&, March, 1945, pp. 83 6 See her review of The Economics of Full Employment, Economic Journal, Vol. 55, april1945,p.79
CAPITAL EXPANSION, RATE OF GROWTH, AND EMPLOYMENT 139 general. Even without technological progress, capital accumulation in- reases labor productivity, at least to a certain point, both because more capital is used per workman in each industry and because there is a shift of labor to industries that use more capital and can afford to l pay a higher wage So if labor productivity is affected by capital accum- ulation, the formula that the latter should proceed at the same rate as the former(and as the increase in labor force)is not as helpful as it ap- The standard Keynesian system does not provide us with any tools for deriving the equilibrium rate of growth. The problem of growth is entirely absent from it because of the explicit assumption that employ ment is a function of national income. This assumption can be justified only over short periods of time; it will result in serious errors over a period of a few years. Clearly, a full-employment level of income of five years ago would create considerable unemployment today We shall? assume instead that employment is a function of. the ratio of national in- come to productive capacity. While this approach seems to me to be I superior to that of Keynes, it should be looked upon as a second ap- proximation rather than a final solution: it does not allow us to separate unused capacity into idle machines and idle men; depending upon various circumstances, the same ratio of income to capacity may yield different fractions of labor force employed Because investment in the Keynesian system is merely an instru-l ment for generating income, the system does not take into account the extremely essential, elementary, and well-known fact that invest- ment also increases productive capacity. 6 This dual character of the investment process makes the approach to the equilibrium rate of growth from the investment(capital)point of view more promising:if investment both increases productive capacity and generates income, it provides us with both sides of the equation the solution of which may yield the required rate of growth t investment proceed at the rate i per year, and let the ratio of F the potential net value added(after depreciation), i. e, of the produc- tive capacity of the new projects to capital invested in them, i.e., to 5: Pr I, be indicated by 8. 7 The net annual potential output of these projects will then be equal to I&. But the productive capacity of the whole econ Whether every dollar invested increases productive capacity is essentially a matter of definition. It can safely be said that investment taken as a whole cer ainly does. To make this statement hold in regard to residential housing, im- puted rent should be included in the national income. See also note 19 7 The use of the word"project"does not imply that investment is done by the government, or that it is always made in new undertakings. I am using"pro (in the absence of a better term) because investment can mean the act of in ing and the result of the act
EVSEY D. DOMAR omy may increase by a smaller amount, because the operation of these ew projects may involve a transfer of labor(and other factors) from ther plants, whose productive capacity is therefore reduced. We shall define o, the potential social average investment productivity as (1) The following characteristics of o should be noted 1. Its use does not imply that other factors of production and tech nology remain constant. On the contrary, its magnitude depends to a very great extent on technological progress. It would be more correct to say that o refers to crease in capacity which accompanies rather than one which is caused by investment 2. o refers to the increase in potential capacity. Whether or not this potential increase results in a larger income depends on the behavior of money expenditures 3. o is concerned with the increase in productive capacity of the whole society, and not with the rate of return derived or expected from investment. Therefore o is not affected directly by changes in distribu tion of income 4. 8 is the maximum that g can attain. The difference between them will depend on the magnitude of the rate of investment on the one hand, and the growth of other factors, such as labor, natural resources, and technological progress on the other. A misdirection of investment will also produce a difference between and o We shall make the heroic assumption that s and o are constant. From(1)it follows that dP dt It is important to note that, with a given o, dP/dt is a function of I, and not of dI/dt. Whether di/dt is positive or negative, dP/dt is always positive so long as o and I are positive. Expression (2)showing the increase in productive capacity is tially the supply side of our system. On the demand side we hav multiplier theory, too familiar to need any comment, except for an emphasis on the obvious but often forgotten fact that with any given marginal propensity to save, dY/dt is a function not of I, but of di/dt Indicating the marginal propensity to save by a, and assuming it to be constant, 'we have the simple relationship that I am disregarding the external economies and diseconomies of the older plants due to the operation of the new projects Over the period 1879-1941 the average propensity to save(ratio of net capital
CAPITAL EXPANSION, RATE OF GROWTH, AND EMPLOYMENT 141 Let the economy be in an equilibrium position so thatio To retain the equilibrium position, we must have dp dy dt dt Substituting(2)and (3)into(5) we obtain our fundamental equation the solution of which gives ao is the equilibrium rate of growth. So long as it remains constant the maintenance of full employment requires investment to grow at a con- If, as a crude estimate, a is taken at 12 per cent and o at some 30 per cent, the equilibrium rate of growth will be some 3. 6 per cent per The reader will now see that the assumption of constant a and o is not entirely necessary, and that the whole problem can be worked out with variable a and g formation to national income)was fairly constant and approximately equal to some 12 per cent. See Simon Kuznets, National Product Since 1869, National Bureau of Economic Research (mimeographed, 1945)p. II-89 and the Survey of Current Business, Vol. 22, May, 1942, and Vol. 24, April, 1944. In a problem of rclical character, an assumption of a constant propensity to save would be very bad. Since we are interested here in a secular problem of continuous full employ ment, this assumption is not too dangerous 10 The problem can be also worked out for the case when Po>Yo. oa After this paper was sent to the printer I found a very interesting article by E. H Stern, "Capital Requirements in Progressive Economies, "Economica, Vol. 12, August, 1945, pp. 163-171, in which the relation between capital and utput in the U. S. during 1879-1929 is expressed (in billions of dollars )as capital 3.55. My estimates gave roughly similar results. This would place s around 30 per cent, though this figure should be raised to account for the underutilization of capital during a part of that period. It is also not clear how che junking process(see p. 144)was reflected in these figures. Review, Vol 34, December, 1944
EVSEY D. DOMAR III. THE EFFECTS OF GROWTH Our next problem is to explore what happens when investment does grow at some constant percentage rate r, which, however, is not neces- sarily equal to the equilibrium rate ao. It will be necessary to introduce two additional concepts: average propensity to save I/y and the average ratio of productive capacity to capital P/K. To simplify the blem we shall assume that 2. P/K=8, i.e, the ratio of productive capacity to capital for the whole economy is equal to that of the new investment projects. We shall consider first the special simple more general case when 0s. Formally this case can be excluded by falling back on the definition of depreciation given in note 2. This, however, is not a very happy solution, but the approach used in this paper will hardly offer a better one. I think, however chat a in our society is sufficiently high to make o>s in a continuous state of full ployment more an exception than a rule
CAPITAL EXPANSION, RATE OF GROWTH, AND EMPLOYMENT 143 Substituting K=P/s into(11) we obtain Y p Since in the present case 0=8 (13) lim Yr P The expression may be called the coeficient of utilization. When the economy grows t the equilibrium rate, so that r=ao, 0=100 per cent and productive capacity is fully utilized. But as r falls below ao, a fraction of capacity a)is gradually left unused. 12 Thus the failure of the economy to grow e required rate creates unused capacity and unemployment Case 8:0<8. As investment proceeds at the rate I, new projects with a productive capacity of Is are built. Since the productive capac ity of the whole economy increases only by Io, it follows that some- where in the economy (not excluding the new projects) productive capacity is reduced by I(s-o). Therefore every year an amount of capital equal to I(s-a)/s becomes useless. The problem can now be approached from two points of view. The amounts I(s-o)/8, can be looked upon as capital losses, which are not taken into account in calculating income and investment. a In thi ase, I still indicates the rate of net investment, and all other symbols retain their old meaning, except that capital has to be redefined as the pital losse chunks of capital (over and above depreciation)are written off and junked. The annual addition to capital will then be I(8-a (15) d (16) +Io eidt= Ko+Io-(ert- 1). 1 It should be noted that if r, a, and o are constant, 0 is also a constant. Even though the to grow at the required rate, the relative disparity income does not become wider because its capital also grows not at the ut at the r rate hese losses are in the accounting sense See note 14
EVSEY D. DOMAR (17) K which is exactly the same result we had in(13) The second approach consists in treating the amounts I(s-a)/s not as capital losses but as a special allowance for obsolescence. Net invest- ment would then have to be defined not as I, but as Io/s. Other sym- bols would have to be redefined accordingly, and the whole problem could then be reworked out in the same way as on pp 142-143 In a sense the choice between these two methods is a matter of bookkeeping; depending upon the character of the problem in hand, one or the other can be used, though I suspect that the second method can easily become misleading. The nature of the process will be the same whichever method is used. The fact is that, owing to a difference between 8 and o, the construction of new investment projects makes certain assets(not excluding the new projects themselves ) useless, be- cause under the new conditions brought about by changes in demand, or a rise in the wage rates, or both, the products of these assets cannot be sold. 4 As stated on p. 140 the difference between s and o is created M either by misdirection of investment or by the lack of balance between the propensity to save on the one hand, and the growth of labor, dis- So long as mistakes are made or this lack of balance exists, the junking overy of natural resources, and technological progress on the other. process is inevitable From a social point of view, the junking pr undesirable. In this country, where saving involves little hardship, it may be perfectly justified. But it may present a serious obstacle to the achievement of full employment, because the owners of capital assets headed for the junk pile will try to avoid the losses. So long as they confine themselves to changes in their accounting practices,no special consequences will follow. But it is more likely that they will try to accumulate larger reserves either by reducing their own con- To be strictly true the statement in the text would require considera divisibility of capital assets. In the absence of such junking"should not be taken too literally. The fact that these assets may still be operated to some extent or that their products are sold at lower prices or that both these conditions exist, does not invalidate our argument, because o, being expressed in real terms, will be higher than it would be if the assets were left completely unused
CAPITAL EXPANSION RATE OF GROWTH, AND EMPLOYMENT 145 sumption or by charging higher prices(or paying lower wages). As a result, the total propensity to save may rise. This will be exactly the opposite measure from what is needed to avoid the junking process and will of course lead to greater trouble, though I am not prepared to ay to what extent capital owners will succeed in passing on these In so far as they are able to control new investment, they will try to avoid losses by postponing it. Consequently, the rate of growth may well be depressed below the required ao, and unused capacity will de- velop Our present model does not allow us to separate unused capacity into idle capital and idle men, though most likely both will be present. 5 Because of humanitarian considerations, we are more concerned with unemployed men. But unemployed capital is extremely important, because its presence inhibits new investment. 16 It presents a grave danger to a full-employment equilibrium in a capitalist society. IV GUARANTEED GROWTH OF INCOME In the preceding sections it was shown that a state of full employ-7 ment can be maintained if investment and income grow at an annual rate ao. The question now arises as to what extent the argument can I be reversed: suppose income is guaranteed to grow at the ao rate; will that call forth sufficient investment to generate the needed income? We are concerned here with a situation where spontaneous invest- ment (i. e, investment made in response to changes in technique, shifts in consumers'preferences, discovery of new resources, ete. )is not suffi cient, and therefore a certain amount of induced investment(made in esponse to a rise in income) is also required. 17 To simplify the argu ment, let us assume that spontaneous investment is absent altogether. It should also be made clear that the problem is treated from a theo- retical point of view, without considering the numerous practical ques- tions that the income guarantee would raise If an economy starts from an equilibrium position, an expected rise in income of Yao will require an investment equal to Yao/s. As before, two cases have to be considered he presence of unemployed men may be obscured by inefficient utilization of labor, as in agriculture. 16 It is true that a given capital owner may often have a hard time distinguish ing between capital idle because of a<s, and capital idle because of r<ao. The irst kind of idleness, however, is relatively permanent, and cannot be corrected by greater expenditures, while the second is temporary (it is hoped) and is due to poor fiscal and monetary policies IT Cf. Alvin H. Hansen, Fiscal Policy and Business Cycles, New York, 1944, Part Three, and particularly p. 297