Issues in Growth Accounting: a Reply to Edward F. Denison
Total Page:16
File Type:pdf, Size:1020Kb
May 1972 Part II By DALE W. JORGENSON and ZVI GRILICHES Issues in Growth Accounting A Reply to Edward F. Denison Page 1. Introduction.. ___ 65 2. Measurement of Output 67 2.1. Introduction. 67 1. Introduction 2.2. Consumption, investment, labor, and capital. 68 2.3. Price and quantity of output. 68 3. Measurement of Capital Input 69 IN our paper, "The Explanation of cially valuable since his objectives are 3.1. Introduction. ___ 69 Productivity Change" [60], we examine similar to ours and his approach is 3.2. Perpetual inventory method- 69 the measurement of total factor pro- carefully articulated with national in- 3.3. Price of investment goods._ 70 ductivity from tlie perspective pro- come and expenditure accounts. 3.4. Price of capital services 72 3.4.1. Introduction...... 72 vided by tiie economic theory of Although Denison's objectives and 3.4.2. Household sector. 72 production. From the accounting point our objectives are similar, any attempt 3.4.3. Noncorporate sec- of view the major innovation in our to integrate his approach to produc- tor______________ 72 approach is in the integration of tivity measurement into national ac- 3.4.4. Corporate sector__ 73 productivity measurement with na- 3.5. Price and quantity of capital counts for saving and wealth gives services 74 tional accounts for income, saving, rise to serious difficulties. The first 4. Relative Utilization of Capital________.. 74 and wealth. Our main substantive important difficulty arises from a basic 4.1. Introduction.... _ 74 conclusion is that growth in real confusion between depreciation and re- 4.2. Measurement of relative uti- factor input rather than growth in placement that underlies all of Deni- lization- ____ 75 total factor productivity is the pre- 4.3. Actual and potential capital son's work. Denison measures net services.-. 76 dominant source of growth in real national product as gross product less 5. Measurement of Labor Input 77 product. replacement; the correct definition is 5.1. Introduction.-., :__ 77 Both our approach to productivity gross product less depreciation. The 5.2. Hours of work. 77 measurement and our substantive con- error in measurement of total product 5.3. Price and quantity of labor clusions require much further analysis services 78 carries over to Denison's measure of 6. Measurement of Total Factor Produc- and testing. Edward F. Denison has total factor input, since the vdue of tivity _ .__'_ 79 made an important contribution to total product is equal to the value of 6.1. Introduction 79 this further analysis and testing in total factor input as an accounting 6.!2. Alternative measures of pro- his paper, "Some Major Issues in ductivity change__________ 80 identity. 6.3. Sources of U.S. economic Productivity Analysis: An Examination A second important difficulty in growth, 1950-62__________ 80 of Estimates by Jorgenson and Gril- Denison's approach arises from an in- 7. Major Issues in Growth Account ing___ 80 iches" [25]. In this paper Denison consistency between his treatment of 7.1. Introduction 80 examines our approach from the van- depreciation in the measurement of 7.2. Scope of product 80 tage point of methods developed in 7.3. Index numbers 83 total product and his treatment of re- 7.4. Capital and labor weights... 84 his study, Why Growth Rates Differ placement in the measurement of cap- 7.5. Weights for components of [28]. Denison's contribution is espe- ital input. This inconsistency results in capital and land--- 84 a contradiction between the income 7.6. Measurement of capital and NOTE.—Professors Jorgenson and Griliches land__________ 87 are both members of the Department of accounts that underlie productivity 7.7. Utilization adjustment 88 Economics, Harvard University. A version measurement and the wealth accounts 7.8. Labor input 89 of this paper was presented at the 12th that underlie the measurement of cap- 7.9. Conclusions and sugges- Conference of the International Association ital input. Although Denison's measure tions for further research, 89 for Research in Income and Wealth in Footnotes __ 90 Ronneby, Sweden, August 30-September 4, of total factor productivity is con- References 92 1971. sistent with national income and ex- 65 Digitized for FRASER http://fraser.stlouisfed.org/ Federal Reserve Bank of St. Louis May 1972 Part II penditure accounts, it is impossible to as part of factor input. An important measure employed by Denison. Even integrate his measure into national objective of our approach to pro- though the details of the measurement saving and wealth accounts. ductivity measurement is the integra- procedure are quite different for the Further difficulties arise in Denison's tion of capital input into national two estimates, the empirical results are allocation of property income among accounts for income, saving, and wealth. very similar. Both measures of labor assets. First, Denison employs nominal Our estimates of product and factor input differ substantially from measures rates of return rather than real rates input, consumption and investment, based on unweighted man-hours, such of return in measuring income from the and labor and capital services are as those of Abramovitz [1], Kendrick supply of capital services. As a con- presented in Section 2 below. [61, 62] and Solow [70]. In Section 5 sequence his allocation of property In Section 3 we present estimates of we compare our measure of labor input income among assets is inconsistent capital input implementing our ap- with alternatives incorporating addi- with the integration of property in- proach in much greater detail than in tional detail. come into accounts for saving and our original study. The new estimates In Section 6 we present revised es- wealth. Second, Denison's classification permit us to distinguish among com- timates of total factor productivity. of assets ignores important differences ponents of property income correspond- Revised estimates of capital input in direct taxation of property income ing to sectors of the economy that require data on property income by by legal form of organization. His differ in legal form of organization. legal form of organization, an analysis allocation of property income fails to These estimates provide for a much of the tax structure for property in- reflect the impact of the tax structure more staisfactory integration of direct come, and the incorporation of measures on rates of return of different types of taxation of property income into factor of relative utilization of capital stock. assets. input accounts. Estimates of capital stock already The purpose of this paper is to com- We have attempted to validate incorporated into productivity studies pare our approach to productivity our original measures by checking provide an important part of the measurement with Denison's. For this our data against a more comprehensive empirical basis for revised estimates of purpose we present a new set of esti- body of supplementary evidence—es- capital input. Ultimately, satisfactory mates of total factor productivity for pecially evidence on investment goods estimates will require the integration the period 1950-1962 covered in Deni- prices in Section 3 and data on changes of productivity measurement with ac- son's study, Why Growth Rates Differ in the relative utilization of capital counts for income, saving, and wealth. [28]. These estimates, prepared by in Section 4. In constructing a new Productivity measures of this type are Christensen and Jorgenson,1 implement set of estimates Christensen and Jorgen- available for the United States for the our approach in much greater detail son have been able to incorporate new period 1929-67,4 but much further than the estimates given in our earlier data. In the most difficult area of work remains to be done in refining study. The new estimates and the empirical research, the measurement and extending these estimates. methods employed in obtaining them of relative utilization, they incorporate Section 7 summarizes the results of are presented in Sections 2-6 below. In cyclical as well as secular changes these revisions, compares them with Section 7 we compare these results with in relative utilization into their measure our original estimates, reviews Denison's and our own earlier ones and of capital input.2 In reviewing their Denison's objections to them, and ex- assess the quantitative importance of work in Section 4 and in response to plores some of the remaining unresolved the differences. Denison's comments we have reached issues. Our original conclusions are The first step in productivity the conclusion that the scope of our changed somewhat, primarily as the measurement is to define measures of original adjustments for changes in result of the reduction in the magnitude product and factor input in current relative utilization should be reduced. and scope of the relative utilization prices. Product is divided between con- In the measurement of real factor adjustment. The resulting estimates of sumption and investment; factor input input, rates of growth of labor and growth in total factor productivity are is divided between labor and capital capital input are averaged to obtain the closer to Denison's estimates than our input. Investment and capital input are rate of growth of total factor input, original ones, but still significantly linked through national accounts for using relative factor shares as weights. lower. Our revised estimates meet, we saving and wealth. Investment in The measurement of aggregate labor believe, all of Denison's valid objections reproducible tangible capital assets is input as developed by Denison, Gri- to our original procedures. We have part of the national product and also liches, and others,3 amounts to applying preserved, however, the major con- part of saving. Investment less deprecia- the same principle of aggregation to the clusion of our original paper: Growth tion plus capital gains is equal to the individual components of labor input.