©f 
 
 AT UBB^SS« 
 
NOTICE: Return or renew all Library Materials! The Minimum Fee for 
 each Lost Book is $50.00. II 141 o r» 
 
 The person charging this matenaMs fespSuQble for 
 its return to the library from which it was withdrawn 
 on or before the Latest Date stamped below. 
 
 sons for discipli- 
 University. 
 
 IANA-CHAMPAIGN 
 
 Theft, BMtilation) and underlining 
 nary attioff add rn'ay result 
 To renfw ia^T|ltto|c^*<tiii'nt 
 
 UNIVERSITY OFiMnOIS 
 
 LI61— O-1096 
 
Digitized by the Internet Archive 
 
 in 2012 with funding from 
 
 University of Illinois Urbana-Champaign 
 
 http://archive.org/details/uncertaintyin196191bull 
 
v\$in 
 
 ■ ENGINEERING LI3RARY 
 
 CONFERENCE R90M ""ZZSUXT * 
 
 enter for Advanced Computation 
 
 UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 
 
 URBANA. ILLINOIS 61801 
 
 CAC DOCUMENT WO. 191 
 
 UNCERTAINTY IN THE 1967 
 U. S. INPUT-OUTPUT DATA 
 
 By 
 Clark W. Bullard 
 
 April 1976 
 
 
CAC DOCUMENT NO. 191 
 
 UNCERTAINTY IN THE 1967 
 U. S. INPUT-OUTPUT DATA 
 
 By 
 
 Clark W. Bullard 
 
 This volume Is bound without 
 
 hd.tfS' 
 
 which is/are unavailable, 
 
 on 
 ampaign 
 
 12-1-ait 
 
 r Research Institute 
 nistration. 
 
CAC DOCUMENT NO. 191 
 
 UNCERTAINTY IN THE 1967 
 U. S. INPUT-OUTPUT DATA 
 
 By 
 Clark W. Bullard 
 
 Center for Advanced Computation 
 University of Illinois at Urbana- Champaign 
 Urbana, Illinois 6l801 
 
 This work was supported by the Electric Power Research Institute 
 and the Energy Research and Development Administration. 
 
ABSTRACT 
 
 Parameters for input - output models of the U.S. economy are based on 
 data drawn from a variety of goverment and private sources. This report pre- 
 sents subjective estimates of the uncertainty of that data, based on inter- 
 views with federal agency personnel. 
 
1. INTRODUCTION 
 
 Input-output data form the basis for most structural analyses of the 
 U.S. economic system. The massive tables of data provide a complete and 
 internally consistent set of linear production functions for all sectors 
 of the economy. But surprisingly, these analytical objectives and appli- 
 cations do not guide the efforts to acquire the data and compile the 
 tables. Actually the input-output tables are constructed as a bridge be- 
 tween the national income and produce accounts for selected base years in 
 order to provide a "benchmark GNP" estimate for those years. 
 
 It is important for the analyst using input-output (10) data for 
 structural analyses to view the data from this perspective. Since it was 
 not acquired primarily to support structural economic analyses, it places 
 additional burdens on the analyst to verify the data' s usefulness and 
 relevance to his particular application. 
 
 Consider the most general type of application, where the analyst 
 wants to predict sector outputs X needed to produce a final bill of 
 goods Y. To do this he premultiplies Y by the Leontief inverse matrix 
 (l_-A)~ which is calculated from a matrix of direct coefficients A for a 
 (prior) base year input-output table. The problem the analyst must 
 address is: How much uncertainty AX on the result X is attributable to 
 
 Data for 368 sectors are published by the U.S. Department of Commerce (l97^a) 
 For a more detailed discussion of input-output analyses see Leontief (19^1 )• 
 
uncertainty AA in the input-output coefficients? Actually AA may result 
 from l) base-year measurement error and 2) changes in the actual A since 
 the base year. In this paper we are concerned only with the former. 
 
 Quantitative methods for treating this general error analysis problem 
 are of three types. The first uses the condition number of (_I-A) to obtain 
 a bound on the norm of (l_-A) , resulting in an extremely conservative 
 upper bound on parametric error magnification. An expression for true 
 maximum upper bound on (i-A) '" was derived by Sebald (1973), and the 
 relative importance of certain parameters to specific applications was 
 determined. Even the true upper bound, however, was quite conservative 
 in that it did not account for the (likely) possibility of error can- 
 cellation. A stochastic analysis to quantify these effects is currently 
 underway at the Center for Advanced Computation. 
 
 This report is limited to presentation of uncertainty estimates on 
 10 data used in calculating the direct coefficients A. 
 
 1.1 Sources of Error 
 
 Uncertainty in the 10 coefficients is related directly to several sources 
 of error in estimating interindustry transactions for the base year. Due 
 to the exhaustive nature of 10 data, it originates from a variety of sources 
 ranging from census questionnaires to judgemental guesses. Morganstern (1950) 
 has categorized the various sources of error in economic data and most of 
 his observations are relevant here. The total uncertainty on a particular 
 transaction "measurement" will include effects of incomplete census cover- 
 age, reporting errors due to misunderstandings or outright lying, campling 
 errors inherent in surveys of firms, transcription or key punching errors, 
 
 -2- 
 
the possibility that forms are lost, classification errors (matching firms 
 and products to SIC codes), and last but certainly not least, the problem 
 of separating companies from establishments in processing returns from 
 surveys or censuses. 
 
 1.2 Effects of Scale 
 
 The scale of this problem is what makes it unique. Due to the size 
 and complexity of the system being modelled (the U.S. economy) measurements 
 can be taken only at infrequent intervals and at great expense. Moreover, 
 it takes whole institutions to take the measurements (e.g. the U.S. Census 
 Bureau) so the user of the data is generally not the one who acquired it. 
 Thus the burden borne routinely by many analysts who play the roles 
 of data-taker and analyst is now split amoung bureaucracies. Part of that 
 burden — responsibility for estimating parametric uncertainty and its effects 
 on analyses — is sometimes never borne because of the way the roles and 
 responsibilities of the bureaucracies are defined. 
 
 The mission of the Census Bureau is to produce statistics; the Bureau 
 of Economic Analysis (BEA) takes these and others and produces accounting 
 tables supporting a benchmark GNP estimate. The analyst would like to 
 take these statistics and interpret them as observations of a physical sys- 
 tem whose structure he would like to model. The statistics are often 
 published in terms of 5 or 10 significant figures, but none of the hundreds 
 of thousands of persons involved in deriving a statistic are responsible 
 for estimating and documenting its uncertainty. 
 
 -3- 
 
2. METHOD 
 
 Recognizing that actual measurements of interindustry transactions 
 and other variables are made in the presence of "noise" (error sourcps), 
 and that frequent measurements are impractical, we must rely on subjective 
 estimates of uncertainty. Such estimates are best obtained at the level 
 of detail at which the measurements are taken, but here too a compromise 
 must be made. A single transaction in an 10 table may be the sum of millions 
 of individual measurements of physical quantities; this report is based 
 on interviews with personnel at BEA, Census, and other agencies near the 
 top of this statistical pyramid. 
 
 2.1 Quantities Estimated 
 
 Uncertainty estimates were obtained on the three basic constituents of 
 the interindustry transactions matrix. These were direct allocations, mar- 
 gins on domestic transactions, and transfers* Independently, estimates were 
 obtained for final demands, gross domestic outputs, and imports and exports. 
 
 In the next section, uncertainty estimates will be given for each of 
 these categories of data. 
 
 2.2 Degree of Detail 
 
 Within the scope of this study it was possible to consider data inputs 
 to the 10 tables at the U&U+-sector level of detail in many cases; and at 
 the 368-sector level for the remainder. At the more detailed level, a 
 magnetic tape was available from BEA which included notes for various direct 
 allocations indicating the source of the data and the magnitude of the 
 
 * 
 
 Precise definitions of these terms are given by the U.S. Department of 
 Commerce (l9T 1 +b). 
 
 -k- 
 
figure obtained from that source. This tape was scanned for notes identi- 
 fying entries from the Census Bureau or other sources deemed equally ac- 
 curate. If more than 75% of the entry in the 368-order Direct Allocations 
 matrix was from one of these sources, it was assigned the same uncertainty 
 as census data. Estimates of uncertainty for all other data were made at 
 the 368- level of detail as described in the next section. 
 
 2. 3 Interview Techniques 
 
 Many agency personnel seemed well-prepared and sometimes even anxious 
 to assign quantitative estimates of uncertainty to the statistics for which 
 they were responsible. Others were quite reluctant, citing the fact that 
 the "correct" answer was not known and only one measurement had been taken 
 so there was inadequate information on which to base an answer. While this 
 latter group was probably more correct in their assessment of the situation, 
 it should be remembered that such a statement could be used as an "excuse" 
 for covering up error levels that might reflect badly on one's job perfor- 
 mance. In virtually every case, those interviewed responded with a quanti- 
 tative answer to a question of the form "If God appeared and told the 
 correct number to the commander of a firing squad, and if that commander 
 asked you to estimate error bounds for your published figure and threaten- 
 ed to kill you if the correct figure lay outside the bounds ... What would 
 you estimate?" 
 
 During the course of interviews with persons relying on the same data 
 sources, and with persons responsible for producing that source data, I was 
 able to arrive at what I believe to be an internally consistent set of uncer- 
 tainty estimates. All results presented in this report may be attributed to the 
 
 These sources are Minerals Yearbook , Census of Mineral Industries , Census 
 of Manufactures Table TA , Census of Transportation , Census of Business , 
 Interstate Commerce Commission, and Civil Aeronautics Board publications. 
 
 -5- 
 
author, although footnotes are used to identify the persons -whom I inter- 
 viewed to obtain information and impressions. Given the nature of the 
 strong institutional pressures for downward bias in these estimates, I 
 do not expect that the pressures for conservatism that I offered in phras- 
 ing my interview questions provided a significant counteracting force. 
 
 2.k Bias 
 
 It is expected that uncertainty estimates obtained from such "top of 
 the pyramid" interviews will be biased downward, since a BEA employee (say) 
 will be reluctant to question the Census Bureau's estimate of the total 
 U.S. steel production unless he has conflicting statistics from somewhere 
 else. Since the Census Bureau has a virtual monopoly on such statistics, 
 the latter situation is impossible; since the BEA employee has hardly the 
 resources to do his own job, he cannot begin to duplicate the efforts of 
 the Census Bureau so the former situation never arises either. Simply 
 stated, if one bureaucracy publishes a seven-significant-figure statistic 
 that cost a million dollars to derive, the humble bureaucrat in another 
 agency with his own problems to worry about is unlikely to seriously 
 challenge the figure. 
 
 Possible treatments for this problem of bias will be discussed in 
 the last section. 
 
 2. 5 Effect of Numerical Magnitude 
 
 Development of 10 data involves much work within established, or 
 relatively well-known, control totals. For this reason, and since the 
 work is done primarily within an accounting framework , the largest numbers 
 
 * 
 
 See Morganstern (1950) 
 
 -6- 
 
usually receive the most attention and are the best known, and the "residual" 
 between the well-known components and the control total is often distributed 
 among other categories using some kind of estimation algorithm. The only 
 exception to this general "rule" occurs when the figure involved has a signi- 
 ficant impact on the value of GNP; then, though small, the figure may become 
 the subject of further analysis and refinement. 
 
 3. UNCERTAINTY ESTIMATES 
 
 In this section, estimates will first be presented at the 368-sector 
 level of detail. This was the level of disaggregation at which most of the 
 persons interviewed were most comfortable in assigning their subjective 
 estimates of uncertainty. 
 
 As indicated earlier, all estimates of upper and lower bounds presented 
 here may be attributed to the author. The discussion and footnotes indicate 
 the source of my impressions and information. 
 
 Estimates of upper and lower bounds are given in two ways. The first 
 is a fraction 6 which denotes symmetric bounds around the published value 
 of + 1006 %. The second, applied in cases where the published value is 
 less well known, is the factor D which when multiplied by the published 
 value gives the upper bound, and whose inverse determines the lower bound. 
 All bounds should be taken to represent a 99.7% confidence level. 
 
 3. 1 Direct Allocations 
 
 "G ood" Census- grade entries. All transactions from one manufacturing 
 sector to another are assigned 6 = .05, as are all other interindustry 
 direct allocations obtained from Census Bureau sources. This figure is 
 
 This information based primarily on interviews with Kenneth Hanson, Richard 
 Chassey, Ruth Runyan, and Patrick Duck of the Census of Manufactures, 
 Industry Division. 
 
 -7- 
 
based on interviews with Census Bureau personnel who feel their techniques 
 for circumventing problems associated with less than 100$ coverage are 
 well within these limits, and that internal cross-checks minimize report- 
 ing and related errors. The largest source of error here is suspected to 
 be classification error; matching products and firms to SIC codes. 
 
 Agriculture sector rows . Based largely on crop reporting surveys; 
 estimate 6 = .10 except for certain transactions noted elsewhere 
 (e.g. government final demand). 
 
 Agriculture sector columns . Inputs from real estate, chemicals, and 
 chemical fertilizer mining are known best from surveys and other sources; 
 estimate 6 = .10. Directly allocated inputs from transportation and trade 
 sectors were treated the same as margins, as described in sections 3,U. All 
 other entries are based at least in part on farm expenditure surveys taken in 
 1955; assume D = 2 for all entries greater than 1% of gross domestic out- 
 put for the sector. All smaller nonzero numbers scaled from D = 2 -> D = 10 
 as described in Appendix A. 
 
 Federal government purchases . For both defense and non-defense 
 purchases, the following assumptions apply: new construction inputs are 
 based on a good data source, so assign 6 = .05; maintenance and repair 
 construction is more subject to classification errors, so 6 = .10. 
 All entries between $10 million and $50 million are assigned 6 = . 30 
 unless otherwise specified below. Purchases less than or equal to 
 $10 million are assigned D = 2 -*■ 10 as discussed in Appendix A. 
 
 **Based primarily on interviews with Jerry Schluter, U.S. Department of 
 Agriculture. 
 
 Based primarily on interviews with Roy Seaton, Bureau of Economic Analysis, 
 
Defense purchases are generally better known, due to more complete 
 source data. Inputs from manufacturing sectors are assigned 6 = .10 
 if they exceed $50 million. Transportation inputs vere derived from 
 outdated formulae that applied poorly to the Southeast Asia situation 
 in 196T and are assigned 6 = .50. Other non-manufacturing inputs were 
 assigned 6 = .10 if they were above the $50 million threshold. 
 
 Non-defense purchases of inputs from non-manufacturing sectors were 
 
 less well known, and were assigned D = 3 if they exceeded one percent of 
 total inputs and D=3->10 if they were smaller. Manufacturing inputs below 
 the $50 million threshold were treated the same. Transportation inputs 
 were assigned 6 = .30. 
 
 State and local government purchases . For health, welfare, education, 
 and sanitation purchases, new construction and real estate inputs are as- 
 signed 6 = .05 since they are obtained from census sources. Together with 
 wages, these inputs account for nearly 75% of all inputs. Other inputs 
 are assigned 6 = .25 if they exceed 1% of total inputs, and D = 1.5 ^ 10 
 as per Appendix A if they are equal to or smaller than 1%. 
 
 For public safety purchases, new construction and real estate are 
 assigned <5 = .05. Maintenance construction is known poorly; D = 1.5. 
 Manufactured inputs greater than $2 million are assigned D = 1.5, and 
 smaller inputs D = 1.5 "► 10. Non-manufactured inputs are assigned D = 1. 5 for 
 those greater than $10 million, and D = 1.5 ■+ 10 for the smaller ones. 
 
 Other state and local government purchases are also assigned 5 = .05 
 for new construction and real estate, but also 6 = . 05 for maintenance 
 construction since it is primarily highway maintenance which is a Census 
 
 Based primarily on interviews with John We alt y, Bureau of Economic Analysis 
 
 -9- 
 
number. Manufactured inputs greater than $5 million are assigned D = 1.5, 
 and smaller figures D = 1.5 -> 10 as per Appendix A. Non-manufactured inputs 
 greater than $50 million are assigned D = 2 and D = 2 -> 10 for smaller inputs. 
 
 Imports and exports . Trade data for commodities (BEA sectors 1.00 - 
 64.00) are obtained from Census sources and are assigned 6 = .05. Trans- 
 portation and wholesale and retail trade data, including margins, were assigned 
 6 = .25. Data on other items (services, etc.) involved in international 
 trade were assigned D = 2, since they were obtained from balance of pay- 
 ments sample data. Small entries at the 368-sector level of detail, repre- 
 senting less than 1% of gross imports or exports were assigned D = 2 -> 10 
 as per Appendix A. 
 
 Inventory change . These figures are in general the least accurate 
 of all final demand entries, and were assigned 6 = .20 for manufactured 
 goods and 6 = .Uo elsewhere. 
 
 "All other" direct allocations . Within the scope of this study it was 
 impossible to identify those responsible for most entries in the input- 
 output tables. Having taken care of most entries through interviews des- 
 cribed above, the remainder were handled as a group. The algorithm was 
 designed to assign very tight tolerances to any transaction comprising a 
 high percentage of total outputs or inputs, and to any sector's output 
 which "by definition" had to be assigned to a particular cell. For example, 
 the algorithm had to assign a very tight tolerance to sales from new resi- 
 dential construction to gross private capital formation, so it would be 
 compatible with the tolerance assigned to that sector's gross domestic out- 
 put. There are numerous other instances where census data might identify 
 
 Based primarily on interviews with Robert Mangen, Bureau of Economic Analysis, 
 
 -10- 
 
sales of butter to food processors or bakers , and the remainder is attri- 
 buted to personal consumption expenditures. On the other hand, very small- 
 magnitude transactions were assigned high uncertainty for the reasons dis- 
 cussed earlier. 
 
 The algorithm defined two fractions for each direct allocation: 
 an input fraction, by normalizing with respect to the gross domestic 
 output of the consuming sector; and an output fraction, by normalizing 
 with respect to the gross domestic output of the producing sector. The 
 algorithm proceeds with these tests in the following order, and assigning 
 6 or D when the first condition is satisfied: if both fractions exceed 
 .95 then 5 = .01, if only one exceeds .95 then 6 = .02; if both exceed 
 .80, 6 = .05, if only one exceeds .80, 6 = .10; if either fraction ex- 
 ceeds .05, then 6 = .20; if either exceeds .01, then D = 1.5. If 
 both are smaller than .01 it assigns D = 2 -+ 10 as per Appendix A. 
 
 Energy Sector Rows . Estimates of relative uncertainty were made by 
 Knecht (1975). Appendix A. 3 gives these estimates, interpreted consistent 
 with the other estimates in this document. These do not apply directly to 
 the dollar transaction data published by BEA; for that information the pre- 
 ceding subsections apply. These estimates are for transactions measured in 
 physical units, assembled from primary sources by Smith and Simpson (197^). 
 Since the University of Illinois model relies on this physical data rather 
 than the dollar transaction, all analyses based or. the uncertainty esti- 
 mates presented here will employ the modified model. For this reason the 
 additional information is included in Appendix A. 3 for completeness. 
 
 -11- 
 
3. 2 Gross domestic output . 
 
 These figures are the best known because they are from the Census 
 or other equally reliable sources (e.g., IRS) and are assigned 6 = .01. 
 The largest errors here probably stem from classification problems and 
 possible confusion between company and establishment-based data. 
 
 3. 3 Transfers . 
 
 If both the row and column sectors were manufacturing sectors, the 
 source of this data was the Census Bureau, but the accuracy was less than 
 that of direct allocations; assign 6 = .20. All other transfers were 
 assigned upper and lower bounds in the same manner as the corresponding 
 cell in the direct allocations matrix. 
 
 3.J+ Margins 
 
 Transportation margins, by product type and mode, are obtained as 
 totals and then prorated proportional to producers' prices across all pur- 
 chasers of that commodity. Then margins in each input are summed for 
 each purchaser and added to the directly allocated inputs. For all 
 transport modes, 6 = .25 was assigned to the margins. Wholesale and 
 retail trade margins may be expected to be more variable, and are some- 
 times computed as percentage markups over the already estimated trans- 
 port margins. Therefore they are assigned <S = .35. 
 
 * 
 
 Based primarily on interviews with Gene Roberts and Phil Ritz , Bureau of 
 Economic Analysis, and with Kenneth Hanson, Census of Manufactures Industry 
 Division. 
 
 ** 
 
 Based primarily on interviews with Kenneth Hanson, Census of Manufactures 
 Industry Division. 
 
 -12- 
 
CONCLUSIONS, APPLICATIONS, AND LIMITATIONS 
 
 As mentioned at the outset, this work was performed to support on- 
 going error analysis projects at CAC. Earlier work using maximum-upper- 
 bound analyses had shown the dangers that might he encountered using re- 
 sults of input-output analyses. Therefore, these estimates of uncertainty 
 on the actual data were needed to check the maximum error hounds on the 
 particular results we were interested in using (e.g.., elements of the 
 energy sector rows of the 196? Leontief inverse matrix). It soon became 
 evident from the magnitude of the uncertainties in the parameter esti- 
 mation process that our maximum upper bound analysis would yield un- 
 satisfactory results. 
 
 The same information is also needed for a stochastic analysis of 
 uncertainty, and such an analysis is currently underway. It is based 
 on the uncertainties estimated here, and incorporates additional assump- 
 tions regarding the distributions of the random variables. Since virtu- 
 ally no information of this type could be obtained in the interviews, 
 it was arbitrarily decided to characterize all data having symmetric 
 upper and lower bounds by normal distributions and those with specified 
 dispersion factors by lognormal distributions. The primary result of 
 this analysis is expected to be a determination of the magnitude of 
 error cancellation. 
 
 The above information is given to further illuminate the context in 
 which these uncertainty estimates were made, and hopefully will dis- 
 courage inappropriate applications of the results. 
 
 *See for example the results presented by Bullard ana Sebald (1975). 
 
 The final report on this work, sponsored by the Electric Power Research 
 Institute, should be available in June 1976. 
 
 -IV 
 
Finally, I repeat that the uncertainty estimates presented here 
 are my own. I have listed many of the persons whom I interviewed, but 
 they have not endorsed my interpretations of those interviews. If the 
 absolute levels of the estimates are widely disputed (and I expect they 
 will be) perhaps at least the relative levels will be accepted. On 
 this basis we plan to perform stochastic error analyses on the 196? U.S. 
 input-output model for several cases ; including perhaps halving and doub- 
 ling error margins presented here, to determine the sensitivity of the 
 results to systematic bias in the estimates. 
 
 -Ik- 
 
APPENDIX A 
 A.l DISPERSION FACTORS FOR SMALL-MAGNITUDE FIGURES 
 
 Often the uncertainty on a column of figures, x. would be described 
 in terms of "a dispersion factor D for x. = B increasing to a dis- 
 persion factor D for the smallest value reported." Taking this lover 
 bound to be X- = A where A = $10 for the 1967 U.S. input-output 
 tables, and assuming a linear dependence of D(x) on log(x) we obtain 
 the following expression for D as a function of x-« Let D(x) = a log(x) 
 + b where a = (D - D )/(log A - log B) and b = (D 1 log A - D 2 log B)/ 
 (log A - log B). It is easy to verify that D(x) takes values D and D g 
 at x - B and X = A respectively. 
 
 Obviously this is a crude approximation, but it actually may be 
 even too refined when viewed from the perspective of the person estima- 
 ting the uncertainty. 
 
 -15- 
 
A. 2 PROBABLE VALUES OF "ZERO" ELEMENTS 
 
 Since few transactions can be defined to be zero, the published 
 figures truncated at $10 dollars may be misleading. It is probably 
 true that if we examined in detail the transactions of all firms in the 
 U.S. defined by a particular transaction cell in the 10 table, we would 
 find at least one nonzero transaction. Therefore the following approxi- 
 mation was used to estimate the probable distribution of nonzero values 
 between the lower and upper bounds [0, $10 ]. 
 
 Let X be the absolute value of a normal random variable Y with 
 mean and a = 10/ . Then X takes nearly all its values between 
 and 10 . By truncating X at 10 , in the sense that larger values are 
 discarded and resampled, the resulting random variable takes all of 
 its values in [0, 10 ] with the great bulk of its unit probability ac- 
 cumulated near zero. 
 
 -16- 
 
A. 3 UNCERTAINTY OF DIRECT ENERGY ALLOCATIONS 
 
 Knecht (1975) estimated error tolerances on all physical-unit energy- 
 transactions. These are coded in Table A. 3-2, and the codes are explained 
 in Table A. 3-1 below. 
 
 Table A. 3-1 
 ENERGY TRANSACTION TOLERANCE CODES 
 
 Code 
 
 6 
 
 00 
 
 (u=0 and 36=10 10 Btu) 
 
 01 
 
 .05 
 
 02 
 
 .05 
 
 03 
 
 .15 
 
 Oil 
 
 .10 
 
 05 
 
 .15 
 
 06 
 
 .15 
 
 07 
 
 .20 
 
 08 
 
 .30 
 
 09 
 
 .05 
 
 10 
 
 .30 
 
 11 
 
 .30 
 
 12 
 
 .20 
 
 kl 
 
 .10 
 
 *Note that instead of the 368-sector level of aggregation, the results pre- 
 sented here are consistent with the slightly aggregated 357-sector breakdown 
 described by Bullard & Herendeen (1975). Dummy sectors consuming no energy 
 have been deleted and public and private sectors producing the same primary 
 product have been combined. 
 
 -17- 
 
TABLE A. 3-2 
 TOLERANCE CODES FOR DIRECT ENERGY USE DATA CELLS 
 
 Sector 
 
 
 
 Energy 
 
 Supplies 
 
 
 Number 
 
 Sector Name 
 
 Coal 
 
 Crude 
 
 Oil 
 
 Electric 
 
 Gas 
 
 1 
 
 COAL MINING 
 
 02 
 
 00 
 
 0? 
 
 02 
 
 02 
 
 2 
 
 CRUDE PETROL./ GAS 
 
 11 
 
 02 
 
 02 
 
 02 
 
 02 
 
 3 
 
 RE F *D PETROL . PROo. 
 
 01 
 
 01 
 
 01 
 
 01 
 
 01 
 
 4 
 
 FLECTRIC UTILITIES 
 
 01 
 
 00 
 
 01 
 
 01 
 
 01 
 
 5 
 
 NATURAL CAS UTIL'S 
 
 11 
 
 01 
 
 11 
 
 11 
 
 01 
 
 6 
 
 Dm JRY 
 
 00 
 
 00 
 
 OH 
 
 Of- 
 
 00 
 
 7 
 
 POULTRY, EGGS 
 
 11 
 
 00 
 
 OP 
 
 OP 
 
 11 
 
 8 
 
 MEAT S ANIMAL PROD. 
 
 00 
 
 00 
 
 Oo 
 
 08 
 
 00 
 
 9 
 
 COTTON 
 
 GO 
 
 OJ 
 
 Or 
 
 08 
 
 00 
 
 10 
 
 FEED GRAINS 
 
 00 
 
 00 
 
 OP 
 
 08 
 
 00 
 
 11 
 
 TOBACCO 
 
 00 
 
 00 
 
 8 
 
 08 
 
 
 
 1? 
 
 FRUI TS 
 
 GO 
 
 00 
 
 or 
 
 08 
 
 00 
 
 13 
 
 V E G * S 5 ^1 S C - CROPS 
 
 00 
 
 00 
 
 08 
 
 08 
 
 00 
 
 14 
 
 OIL ->3 EARING CROPS 
 
 00 
 
 00 
 
 1 1 
 
 11 
 
 00 
 
 15 
 
 FOR./ GRNHSE/ NURS. 
 
 11 
 
 00 
 
 OP 
 
 OS 
 
 11 
 
 16 
 
 FORESTRY ? FISHERY 
 
 00 
 
 CO 
 
 11 
 
 11 
 
 11 
 
 17 
 
 AG./ FOP. g FISH 
 
 11 
 
 00 
 
 11 
 
 11 
 
 11 
 
 18 
 
 IRON ORE MINING 
 
 11 
 
 00 
 
 02 
 
 02 
 
 02 
 
 19 
 
 C P P t R MINING 
 
 11 
 
 00 
 
 0? 
 
 02 
 
 02 
 
 20 
 
 OTER NON-FER. M 1 N . 
 
 11 
 
 00 
 
 02 
 
 02 
 
 2 
 
 21 
 
 STONE/- CLAY/ R M I N . 
 
 02 
 
 00 
 
 02 
 
 G2 
 
 02 
 
 22 
 
 C H E V . MINERAL M I N . 
 
 11 
 
 00 
 
 02 
 
 02 
 
 02 
 
 23 
 
 NEW CONST./ RESID'L 
 
 no 
 
 00 
 
 
 11 
 
 11 
 
 24 
 
 NEW CONST./ N N - R E S 
 
 CO 
 
 00 
 
 
 11 
 
 11 
 
 25 
 
 NEW CONST./ PJ3 UT. 
 
 00 
 
 00 
 
 
 11 
 
 11 
 
 26 
 
 NEW CONST./ hi 1 WAYS 
 
 GO 
 
 00 
 
 
 11 
 
 11 
 
 ?7 
 
 NEW CONST./ OTHER 
 
 GO 
 
 
 
 
 11 
 
 11 
 
 28 
 
 MINT. CONST./ RES. 
 
 GO 
 
 00 
 
 
 11 
 
 11 
 
 29 
 
 MAINT. CONST./ OTHR 
 
 00 
 
 00 
 
 
 11 
 
 11 
 
 30 
 
 GUIDED MISSILES 
 
 11 
 
 00 
 
 
 41 
 
 11 
 
 31 
 
 AMMUNITION 
 
 11 
 
 00 
 
 
 41 
 
 11 
 
 32 
 
 T A N < S 
 
 11 
 
 00 
 
 
 41 
 
 11 
 
 33 
 
 FIRE CONTROL EQPMNT 
 
 
 
 00 
 
 
 41 
 
 00 
 
 34 
 
 SMALL AR*S 
 
 11 
 
 00 
 
 
 41 
 
 11 
 
 35 
 
 SMALL ARMS AMMJNIVN 
 
 11 
 
 00 
 
 
 41 
 
 11 
 
 36 
 
 OTHER ORDNANCE 
 
 11 
 
 00 
 
 
 41 
 
 11 
 
 37 
 
 MFAT PRODUCTS 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 38 
 
 3UTTER 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 39 
 
 CHEESE 
 
 02 
 
 • 00 
 
 06 
 
 02 
 
 02 
 
 40 
 
 CONDENSED MIL< 
 
 2 
 
 00 
 
 06 
 
 02 
 
 02 
 
 41 
 
 ICE CREAM 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 42 
 
 FLUID MILK 
 
 03 
 
 00 
 
 07 
 
 03 
 
 03 
 
 43 
 
 CANNED SEA FOODS 
 
 03 
 
 00 
 
 07 
 
 11 
 
 11 
 
 44 
 
 CANNED SPECIALTY 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 45 
 
 CANNED FRUITS/ V E G . 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 46 
 
 DEHYDRATED PROD'S 
 
 03 
 
 00 
 
 07 
 
 11 
 
 11 
 
 47 
 
 PICKLES/ DRESSINGS 
 
 03 
 
 00 
 
 0? 
 
 11 
 
 03 
 
 48 
 
 F R F S H & FROZEN FISH 
 
 03 
 
 00 
 
 07 
 
 11 
 
 03 
 
 49 
 
 FROZFN FRUITS/- VEC. 
 
 H2 
 
 00 
 
 06 
 
 02 
 
 02 
 
 50 
 
 FLOUR/ CEREAL PREP. 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 51 
 
 PREP'D ANIMAL FEEDS 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 52 
 
 RICE MILLING 
 
 00 
 
 00 
 
 11 
 
 11 
 
 03 
 
TABLE A. 3-2 (Continued) 
 
 Sector 
 
 
 
 cner yy 
 
 oupp 1 1 
 
 Ci 
 
 
 Number 
 
 Sector Name 
 
 Coal 
 
 Crude 
 
 Oil 
 
 Electric 
 
 Gas 
 
 5 3 
 
 WET CORN! V, I L L I M G 
 
 02 
 
 3 3 
 
 06 
 
 02 
 
 C2 
 
 54 
 
 BAKERY PRODUCTS 
 
 02 
 
 03 
 
 04 
 
 02 
 
 02 
 
 55 
 
 S U P i A u 
 
 02 
 
 
 
 0^ 
 
 02 
 
 02 
 
 56 
 
 CONFECTIONERY PROD. 
 
 02 
 
 3 3 
 
 04 
 
 02 
 
 02 
 
 57 
 
 ALCOHOLIC BEVERAGES 
 
 02 
 
 03 
 
 Oc 
 
 02 
 
 02 
 
 58 
 
 SOFT DRINKS 
 
 02 
 
 03 
 
 0& 
 
 2 
 
 02 
 
 59 
 
 FLAVORINGS < SIRUPS 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 60 
 
 COTTONSEED MILLS 
 
 02 
 
 00 
 
 6 
 
 02 
 
 0? 
 
 61 
 
 SOYBEAN MILLS 
 
 02 
 
 00 
 
 Ob 
 
 02 
 
 02 
 
 62 
 
 VEGETABLE OIL MILLS 
 
 r '3 
 
 3 
 
 11 
 
 11 
 
 03 
 
 63 
 
 A H I M A L FATS 
 
 2 
 
 00 
 
 06 
 
 02 
 
 02 
 
 64 
 
 C F F E L 
 
 00 
 
 03 
 
 7 
 
 03 
 
 11 
 
 65 
 
 COOKING OILS 
 
 02 
 
 03 
 
 Oh 
 
 02 
 
 02 
 
 66 
 
 MANUFACTURED ICE 
 
 11 
 
 33 
 
 11 
 
 03 
 
 11 
 
 6? 
 
 MACARONI <. S P A 5 H E T . 
 
 03 
 
 00 
 
 07 
 
 11 
 
 03 
 
 68 
 
 FOOD PPEP'NS/ ^EC. 
 
 02. 
 
 03 
 
 06 
 
 02 
 
 32 
 
 69 
 
 CIGARETTES/ C13ARS 
 
 02 
 
 03 
 
 04 
 
 02 
 
 02 
 
 70 
 
 TOBACCO STEMMING... 
 
 02 
 
 03 
 
 4 
 
 02 
 
 02 
 
 71 
 
 BROAD FABRIC * I L L S 
 
 02 
 
 00 
 
 04 
 
 02 
 
 03 
 
 72 
 
 NARROW FABRIC MILLS 
 
 03 
 
 33 
 
 5 
 
 1)3 
 
 11 
 
 75 
 
 YARN MILLS 
 
 03 
 
 00 
 
 07 
 
 03 
 
 11 
 
 74 
 
 THPEmD mills 
 
 3 
 
 03 
 
 0/ 
 
 33 
 
 33 
 
 75 
 
 FLOOR COVERINGS 
 
 03 
 
 3 
 
 OS 
 
 03 
 
 1 1 
 
 76 
 
 FELT GOODS/ N . E . C . 
 
 03 
 
 03 
 
 07 
 
 11 
 
 03 
 
 77 
 
 LACE GOODS 
 
 CO 
 
 03 
 
 11 
 
 11 
 
 11 
 
 78 
 
 UPHOLSTERY FILLINGS 
 
 03 
 
 33 
 
 1 1 
 
 11 
 
 11 
 
 79 
 
 PROC. TEXTILE WASTE 
 
 00 
 
 03 
 
 or 
 
 11 
 
 11 
 
 80 
 
 C JATED FABRICS 
 
 02 
 
 3 
 
 06 
 
 02 
 
 02 
 
 81 
 
 TIRE CORD i F A3RIC 
 
 03 
 
 03 
 
 11 
 
 03 
 
 11 
 
 82 
 
 SC0URIN3 PLANTS 
 
 00 
 
 03 
 
 07 
 
 11 
 
 33 
 
 83 
 
 CORDAGE S TWINE 
 
 03 
 
 00 
 
 07 
 
 11 
 
 03 
 
 84 
 
 TEXTILE GOODS/ NFC. 
 
 03 
 
 00 
 
 07 
 
 11 
 
 03 
 
 85 
 
 HOSIERY 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 86 
 
 KNIT APPAREL MILLS 
 
 03 
 
 33 
 
 07 
 
 03 
 
 3 
 
 87 
 
 KMT FABRIC 'MILLS 
 
 03 
 
 03 
 
 7 
 
 OS 
 
 03 
 
 S& 
 
 APPAREL/ P U P C H . M £ T 
 
 03 
 
 00 
 
 5 
 
 03 
 
 11 
 
 89 
 
 CURTAINS '-DRAPERILS 
 
 11 
 
 3 
 
 11 
 
 11 
 
 11 
 
 90 
 
 H0USEFURNISHIV5 NEC 
 
 3 
 
 03 
 
 7 
 
 3 
 
 11 
 
 91 
 
 F A 3 . TEXT. PROD. N E 
 
 11 
 
 03 
 
 11 
 
 11 
 
 11 
 
 92 
 
 LOGGING CAMPS * CON 
 
 03 
 
 33 
 
 05 
 
 3 
 
 3 
 
 93 
 
 SAWMILLS 
 
 3 
 
 03 
 
 07 
 
 03 
 
 03 
 
 94 
 
 H m R D W L D FLOORING 
 
 03 
 
 03 
 
 07 
 
 03 
 
 33 
 
 95 
 
 SPEC. PROD. SAWMILL 
 
 30 
 
 3 3 
 
 11 
 
 11 
 
 11 
 
 9 6 
 
 MiLL WORK 
 
 3 
 
 03 
 
 G7 
 
 03 
 
 11 
 
 97 
 
 VFNEER * PLYWOOD 
 
 2 
 
 30 
 
 06 
 
 02 
 
 02 
 
 93 
 
 PREFAB. JOOD STRUC. 
 
 11 
 
 00 
 
 11 
 
 11 
 
 11 
 
 99 
 
 WOOD PRESERVING 
 
 2 
 
 03 
 
 06 
 
 02 
 
 02 
 
 130 
 
 WOOD F ROD'S/ N.E.C. 
 
 02 
 
 00 
 
 U6 
 
 02 
 
 02 
 
 101 
 
 WOOD CONTAINERS 
 
 11 
 
 00 
 
 5 
 
 03 
 
 11 
 
 102 
 
 WOOD H'HOLO FURN'PE 
 
 03 
 
 00 
 
 07 
 
 03 
 
 03 
 
 103 
 
 UPHOL'D H'HOLD FUP. 
 
 03 
 
 03 
 
 07 
 
 03 
 
 11 
 
 104 
 
 MET. H'HOLD FURN'RE 
 
 11 
 
 03 
 
 11 
 
 11 
 
 11 
 
 105 
 
 MATTRESSES *, 3FDSP. 
 
 03 
 
 00 
 
 07 
 
 11 
 
 03 
 
TABLE A. 3-2 (Continued) 
 
 Sector 
 Number 
 
 Sector Name 
 
 Coal 
 
 Energy Supplies 
 Crude Oil Electric Gas 
 
 10:, WOOD OFF. F U R N I T U P E 
 
 137 M t T . OFF. FURNITURE 
 
 138 P U 3 . & L D G . F U R N I ' R E 
 
 139 WOOD FIXTURE S 
 
 110 METAL F I XTURES 
 
 111 BLINDS 5 SHADES 
 
 11? F U R N . 5 FJXURES/ \ F 
 
 113 PULP .SILLS 
 
 114 PAPER MILLS 
 
 115 PnPERaJARD SILLS 
 
 116 ENVELOPES 
 
 117 SANITARY PAPER PROD 
 
 118 WALL i 3LD&. PAPER 
 
 119 CONT'D PAPER PROD'S 
 
 120 PAPERBOARD 30XCS... 
 
 121 NEWSPAPERS 
 
 122 PERIODICALS 
 
 123 ^00< PRINT. 5 P U 3 L . 
 
 124 MISC. PUBLISHING 
 
 12 5 C0MMEKC1AI PRINTING 
 
 126 B U S I N fc S S F R * S . . . 
 
 1 2 7 GREETING CARD P U 3 L . 
 
 12 8 MISC. P R I N I M S S E R V . 
 
 129 INDUSTRIAL C H E *! I C * L 
 
 130 FERTILIZERS 
 
 151 A3. CHEMICALS/ NEC. 
 
 132 MISC. CHEM. PROD'S 
 
 133 PLASTICS i RENINS 
 
 154 SYNTHETIC RU33ER 
 135 CtLL. MAN-MADE F 1 3 . 
 156 ORGANIC FI3ERS 
 
 13 7 D * US S 
 
 158 CLEANING PREPARA'.JS 
 
 139 TOILtT PREPARATIONS 
 
 HD PAINT S aLL'D PROD. 
 
 141 PAVING MIX. i 3 L C K 
 
 142 ASPHALT FELTS/ COAT 
 
 14 5 TIRES '« J\uR T U 3 F S 
 
 144 R U 3 3 F R FOOTWEAR 
 
 145 MISC. RUBBER PROD'S 
 
 146 MISC . PLASTIC PROD. 
 
 147 LEATHER TANNING... 
 
 148 FOOTWEAR CUT STOCK 
 
 149 'FOOTGEAR EXC. RUBOR 
 
 150 OTHER LEATHER PROD. 
 15": GLASS «< GLASS PROD. 
 
 152 GLASS CONTAINERS 
 
 153 CEMENT/ HYDRAULIC 
 
 15 4 ciRICK >; STRUC. CLAY 
 
 155 CtRAMIC WALL. ..TILE 
 
 156 CLAY REFRACT0R1FS 
 
 157 STRUC. CLAT PROD. N 
 
 158 V1TR . PLUM3. FIXT 'S 
 
 03 
 
 00 
 
 11 
 
 11 
 
 00 
 
 3 
 
 00 
 
 7 
 
 11 
 
 03 
 
 03 
 
 00 
 
 05 
 
 03 
 
 1 1 
 
 11 
 
 00 
 
 07 
 
 11 
 
 11 
 
 11 
 
 00 
 
 07 
 
 03 
 
 03 
 
 03 
 
 03 
 
 7 
 
 11 
 
 03 
 
 03 
 
 00 
 
 0/ 
 
 11 
 
 03 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 00 
 
 00 
 
 07 
 
 11 
 
 05 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 00 
 
 04 
 
 02 
 
 3 2 
 
 03 
 
 00 
 
 0/ 
 
 03 
 
 03 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 02 
 
 00 
 
 04 
 
 2 
 
 2 
 
 03 
 
 00 
 
 1 1 
 
 03 
 
 11 
 
 3 3 
 
 
 
 05 
 
 03 • 
 
 03 
 
 n 3 
 
 03 
 
 05 
 
 11 
 
 11 
 
 03 
 
 03 
 
 5 
 
 03 
 
 3 
 
 11 
 
 00 
 
 0/ 
 
 11 
 
 11 
 
 11 
 
 3 
 
 1 1 
 
 11 
 
 11 
 
 r '0 
 
 00 
 
 7 
 
 11 
 
 11 
 
 '■2 
 
 3 
 
 04 
 
 02 
 
 37 
 
 00 
 
 00 
 
 Do 
 
 02 
 
 02 
 
 11 
 
 00 
 
 11 
 
 03 
 
 11 
 
 02 
 
 02 
 
 04 
 
 02 
 
 02 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 03 
 
 06 
 
 02 
 
 02 
 
 02 
 
 00 
 
 6 
 
 02 
 
 02 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 11 
 
 03 
 
 05 
 
 03 
 
 11 
 
 03 
 
 00 
 
 07 
 
 03 
 
 03 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 11 
 
 00 
 
 06 
 
 02 
 
 0? 
 
 32 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 03 
 
 04 
 
 02 
 
 02 
 
 00 
 
 00 
 
 04 
 
 02 
 
 02 
 
 ^2 
 
 00 
 
 04 
 
 02 
 
 02 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 3 
 
 00 
 
 5 
 
 11 
 
 11 
 
 11 
 
 00 
 
 11 
 
 11 
 
 1 1 
 
 "3 
 
 00 
 
 5 
 
 03 
 
 3 
 
 00 
 
 00 
 
 1 1 
 
 11 
 
 11 
 
 02 
 
 03 
 
 6 
 
 02 
 
 02 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 00 
 
 04 
 
 02 
 
 02 
 
 02 
 
 03 
 
 06 
 
 02 
 
 02 
 
 00 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
TABLE A. 3-2 (Continued) 
 
 Sector 
 Number 
 
 15 9 
 
 160 
 
 161 
 
 162 
 
 163 
 
 164 
 
 165 
 
 166 
 
 167 
 
 1 68 
 
 169 
 
 170 
 
 171 
 
 172 
 
 1/3 
 
 174 
 
 175 
 
 176 
 
 177 
 
 178 
 
 179 
 
 ISO 
 
 131 
 
 132 
 
 133 
 
 184 
 
 185 
 
 136 
 
 1.37 
 
 138 
 
 189 
 
 190 
 
 191 
 
 192 
 
 193 
 
 194 
 
 195 
 
 196 
 
 197 
 
 193 
 
 199 
 
 2 00 
 
 231 
 
 2.12 
 
 203 
 
 204 
 
 205 
 
 2 06 
 
 207 
 
 208 
 
 209 
 
 210 
 
 211 
 
 FOOD 
 
 P R C 
 
 POTT 
 
 CONC 
 
 CONC 
 
 READ 
 
 LIME 
 
 GYPS 
 
 CUT 
 
 A3PA 
 
 AS3E 
 
 6*S< 
 
 TRF « 
 
 v» J M F 
 
 NONC 
 
 3 a. SI 
 
 IRON 
 T RON 
 PR IM 
 PR I "I 
 PRIM 
 PR I vi 
 PR] V! 
 PR I M 
 SEC. 
 COPP 
 A L J M 
 N N F 
 M N F 
 ALU^ 
 bPAS 
 NONF 
 NONF 
 ^ETA 
 MET A 
 M E T . 
 PLUM 
 HEAT 
 FAB. 
 M £ T A 
 F A3. 
 SHEE 
 ARCH 
 -11 SC 
 SCRE 
 M E T A 
 CUTL 
 HAND 
 HARD 
 COAT 
 M1SC 
 SAFE 
 
 Sector Name 
 
 UTENSILS, POT. 
 . ELECT. SUPP "S 
 LRY PROD'S/ NEC 
 RhTt dLOC< I BR 
 ^ E T F PROD'S .NEC 
 Y - M 1 X E D CONCRET 
 
 Energy Supplies 
 Coal Crude Oil Electric Gas 
 
 U A p 
 
 STON 
 
 SI\/E 
 
 ST JS 
 
 ETS 
 
 TED 
 
 RAL 
 
 LAY 
 
 FT. 
 
 C 3T 
 
 t S 
 
 f. S 
 . Mt 
 A S Y 
 ARY 
 ARY 
 A FY 
 . NO 
 
 NON 
 fR R 
 . R ? 
 ERR . 
 CPR. 
 1 N U M 
 S, . . 
 tRR . 
 ERR 
 L CA 
 L dA 
 
 S *N 
 -IN-, 
 I f\ G 
 
 STR 
 L DO 
 
 PL A 
 TIE 
 "L M 
 . IE 
 U MA 
 L ST 
 ERY 
 
 fc E 
 WARE 
 IN 3/ 
 . FA 
 S 5 
 
 R D U C T S 
 
 r s pro 
 
 PRODUC 
 P R D D J C 
 •', I N S U L 
 •i I N E R .«. L 
 W L 
 P r F R A C T 
 V, IN. PR 
 EEL PRO 
 T E E L F N 
 TEEL FO 
 T. PROl; 
 C P P L R 
 LEAD 
 71NC 
 A L U M I *i J 
 N F E R R . 
 F E P R . M 
 } L L I N G 
 L L 1 N G S 
 ROLLIN 
 WIRE D 
 CASTIN 
 .CASTIN 
 CASTIN 
 'IS FORG 
 NS 
 
 RRELS. . 
 I T A R Y 
 
 FITS .. 
 EQUIP./ 
 U C . S T E 
 ORS / SA 
 T E *J R < 
 T A L W R 
 lTAL A3 
 T&L *0R 
 C H . PRO 
 A .vj P 1 N G S 
 
 DGE TOO 
 / M.E.C 
 ENGRV. 
 i3. WHE 
 VAULTS 
 
 D ■ S 
 
 TS 
 
 TS 
 
 A'NS 
 
 S 
 
 OR *S 
 OD *S 
 'S 
 DR "S 
 RG 'S 
 . NE 
 
 M 
 
 IE T . 
 ETAL 
 4 DR 
 uPA 
 o • . . 
 
 R A ,; . 
 GS 
 GS 
 
 G S . . 
 INGS 
 
 J A RE 
 
 r E X C . 
 
 EL 
 
 SH . . 
 
 < 
 R< 
 
 < 
 
 D '?/ 
 
 LS 
 
 / . . . 
 
 PRO 
 
 -21- 
 
 11 
 
 02 
 00 
 2 
 2 
 03 
 02 
 "2 
 05 
 02 
 02 
 2 
 02 
 02 
 02 
 0.0 
 02 
 02 
 02 
 2 
 02 
 '?. 
 02 
 02 
 T2 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 00 
 02 
 02 
 02 
 3 
 3 
 2 
 3 
 00 
 00 
 2 
 02 
 03 
 f )2 
 02 
 03 
 03 
 11 
 
 00 
 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 
 00 
 00 
 00 
 00 
 
 OJ 
 00 
 00 
 00 
 00 
 00 
 00 
 
 00 
 00 
 
 00 
 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 
 00 
 00 
 00 
 
 00 
 00 
 00 
 00 
 00 
 00 
 00 
 
 11 
 
 06 
 06 
 6 
 6 
 07 
 6 
 06 
 05 
 6 
 6 
 6 
 06 
 6 
 6 
 06 
 04 
 04 
 
 06 
 6 
 06 
 6 
 6 
 06 
 06 
 4 
 6 
 06 
 06 
 06 
 06 
 06 
 06 
 06 
 04 
 06 
 6 
 06 
 6 
 07 
 7 
 6 
 0/ 
 07 
 07 
 04 
 04 
 07 
 06 
 06 
 05 
 05 
 11 
 
 02 
 02 
 02 
 02 
 02 
 n3 
 02 
 02 
 3 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 03 
 03 
 02 
 03 
 03 
 03 
 0? 
 02 
 03 
 11 
 02 
 03 
 03 
 11 
 
 02 
 02 
 02 
 02 
 03 
 02 
 02 
 1 1 
 2 
 02 
 2 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 2 
 02 
 02 
 2 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 02 
 03 
 03 
 02 
 03 
 03 
 3 
 02 
 02 
 03 
 02 
 02 
 03 
 03 
 03 
 
TABLE A. 3-2 (Continued) 
 
 Sector 
 
 Number 
 
 212 
 213 
 214 
 215 
 216 
 21 7 
 218 
 219 
 220 
 221 
 222 
 223 
 224 
 225 
 226 
 227 
 228 
 229 
 230 
 231 
 232 
 '33 
 254 
 235 
 236 
 237 
 238 
 259 
 240 
 241 
 242 
 243 
 244 
 245 
 246 
 247 
 248 
 249 
 250 
 251 
 252 
 253 
 254 
 255 
 256 
 257 
 258 
 259 
 2b0 
 261 
 262 
 263 
 264 
 
 Energy Suppjies 
 
 Sector Name 
 
 STTEL SPRINGS 
 PIPES, VALVES/ FIT. 
 COLL^PSI^LF TU3ES 
 METAL FOIL * LEAF 
 F A 3 . MET. PROD'S N E 
 S T E A A E N G I N E S ? T U R 
 INT. COM3. FNG1MFS 
 FApq *i4CKI?JERY 
 CONSTRUC. MACHINERY 
 M I H I H G MACHINERY 
 OIL FIELD VI A C H I H E R Y 
 ELEVATO D S & MOV . ST 
 CONVEYORS i EQUIP. 
 HOISTS, CRANES K WO 
 INDUSTRIAL TRUCKS.. 
 MACHINE TOOLS/ CUT. 
 MACHINE TOOLS/ FOR 1 * 
 SPEC. DIES i TOOLS 
 METAL WORK. MACH./ N 
 FOOD PROD. MACH. 
 TEXTILE MACHINERY 
 WOODWORK. MAC^INE&Y 
 PAPFP INDUS. v A C H • Y 
 p RINT. TRADES MACH. 
 SPEC. INDUS. v A C H . / 
 PUMPS J COMPRESSORS 
 BALL t ROLLER SEAR. 
 
 3 L *' f o S 
 
 FAN^ 
 
 INDUSTRIAL PATTERNS 
 POWER T R A \ S . taUIP. 
 INDUS. FURNACES R . . 
 GEN. INDUS. MACH./. 
 MACH. SHOP. PROD ■ S 
 
 COMPUTING MACH 'S 
 
 TYPEWRITERS 
 SCALES S, BALANCES 
 OFFICE MACH 'S/ MEC. 
 AUTO'C M E R C H . MACH. 
 C M M ' L L ' U N D . E 3 II I P 
 REFRIGERATION MACH. 
 McAS. \ DISP. PUMPS 
 S L P V I C t I N i> . w A C H . / 
 E L E C . A E ' S . I \ S T R U . 
 
 SWlTChGE AR APPAP. 
 
 M T W S <. GENERATORS 
 INDUSTRIAL CONTROLS 
 WELDING APPARATUS 
 CA30N £ GRAPHITE PR 
 ELfcC. IND. APP. NtC 
 HOUSE 'D COOK. EQUIP 
 
 HOUSE»D PFFRIGER 
 
 HOUSE* D LAUNDRY EQP 
 
 oal 
 
 Crude 
 
 Oil 
 
 Electric 
 
 Gas 
 
 00 
 
 OJ 
 
 07 
 
 11 
 
 0? 
 
 03 
 
 00 
 
 7 
 
 11 
 
 05 
 
 00 
 
 00 
 
 11 
 
 11 
 
 03 
 
 00 
 
 00 
 
 1 1 
 
 11 
 
 11 
 
 3 
 
 00 
 
 07 
 
 03 
 
 11 
 
 2 
 
 00 
 
 6 
 
 02 
 
 02 
 
 2 
 
 00 
 
 Co 
 
 C2 
 
 02 
 
 2 
 
 00 
 
 4 
 
 2 
 
 2 
 
 02 
 
 00 
 
 Of. 
 
 02 
 
 02 
 
 U2 
 
 00 
 
 Ob 
 
 02 
 
 02 
 
 00 
 
 00 
 
 1 1 
 
 3 
 
 11 
 
 11 
 
 00 
 
 07 
 
 11 
 
 11 
 
 03 
 
 00 
 
 07 
 
 11 
 
 03 
 
 2 
 
 00 
 
 06 
 
 02 
 
 02 
 
 03 
 
 00 
 
 07 
 
 11 
 
 3 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 03 
 
 00 
 
 0/ 
 
 P3 
 
 3 
 
 02 
 
 00 
 
 U6 
 
 02 
 
 02 
 
 11 
 
 00 
 
 7 
 
 11 
 
 03 
 
 3 
 
 00 
 
 0/ 
 
 03 
 
 11 
 
 11 
 
 00 
 
 0' 
 
 11 
 
 03 
 
 11 
 
 00 
 
 7 
 
 11 
 
 3 
 
 11 
 
 00 
 
 07 
 
 11 
 
 3 
 
 2 
 
 00 
 
 6 
 
 02 
 
 02 
 
 03 
 
 00 
 
 07 
 
 03- 
 
 03 
 
 02 
 
 00 
 
 6 
 
 02 
 
 02 
 
 02 
 
 00 
 
 06 
 
 02 
 
 2 
 
 00 
 
 00 
 
 11 
 
 11 
 
 11 
 
 U2 
 
 00 
 
 06 
 
 02 
 
 02 
 
 11 
 
 00 
 
 11 
 
 11 
 
 03 
 
 3 
 
 00 
 
 7 
 
 03 
 
 3 
 
 3 
 
 00 
 
 05 
 
 03 
 
 03 
 
 11 
 
 
 
 06 
 
 02 
 
 02 
 
 00 
 
 00 
 
 06 
 
 02 
 
 02 
 
 11 
 
 00 
 
 6 
 
 02 
 
 2 
 
 ^2 
 
 00 
 
 06 
 
 02 
 
 02 
 
 03 
 
 
 
 7 
 
 11 
 
 03 
 
 11 
 
 no 
 
 1 1 
 
 11 
 
 11 
 
 2 
 
 00 
 
 6 
 
 02 
 
 02 
 
 00 
 
 00 
 
 07 
 
 11 
 
 03 
 
 3 
 
 
 
 7 
 
 11 
 
 03 
 
 11 
 
 00 
 
 7 
 
 03 
 
 1 1 
 
 03 
 
 
 
 07 
 
 ^3 
 
 1 1 
 
 3 
 
 00 
 
 07 
 
 03 
 
 1 1 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 11 
 
 00 
 
 07 
 
 03 
 
 03 
 
 03 
 
 00 
 
 07 
 
 11 
 
 03 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 11 
 
 00 
 
 07 
 
 11 
 
 03 
 
 03 
 
 00 
 
 7 
 
 11 
 
 03 
 
 02 
 
 00 
 
 06 
 
 02' 
 
 02 
 
 03 
 
 00 
 
 07 
 
 03 
 
 11 
 
TABLE A. 3-2 (Continued) 
 
 Sector 
 
 
 
 
 — t r 
 
 
 
 Number 
 
 Sector Name 
 
 Coal 
 
 Crude 
 
 Oil 
 
 Electric 
 
 Gas 
 
 265 
 
 ELEC. HOUSE WAPES & 
 
 11 
 
 00 
 
 7 
 
 03 
 
 11 
 
 266 
 
 HOUSEHOLD VACUUM CL 
 
 00 
 
 00 
 
 11 
 
 11 
 
 3 
 
 267 
 
 SEWING MACHINES 
 
 00 
 
 00 
 
 07 
 
 11 
 
 11 
 
 26S 
 
 H U S c • D A P F L ' S / NEC 
 
 03 
 
 00 
 
 07 
 
 11 
 
 05 
 
 269 
 
 ELECTRIC LAMPS 
 
 03 
 
 00 
 
 7 
 
 3 
 
 1 1 
 
 270 
 
 LIGHTING FIXTURES 
 
 02 
 
 00 
 
 oc 
 
 02 
 
 02 
 
 271 
 
 WIRING DEVICES 
 
 02 
 
 03 
 
 0(> 
 
 02 
 
 02 
 
 272 
 
 RADIO i TV SETS 
 
 3 
 
 00 
 
 7 
 
 03 
 
 11 
 
 273 
 
 PHONOGRAPH RECORDS 
 
 00 
 
 00 
 
 7 
 
 11 
 
 3 
 
 274 
 
 TELEPHONE & TFLEGP. 
 
 2 
 
 00 
 
 6 
 
 02 
 
 02 
 
 275 
 
 RADIO i TV COM. EQP 
 
 02 
 
 00 
 
 06 
 
 02 
 
 02 
 
 276 
 
 ELECTRON TUdES 
 
 2 
 
 00 . 
 
 Qt- 
 
 02 
 
 02 
 
 277 
 
 SEMICONDUCTORS 
 
 00 
 
 00 
 
 06 
 
 02 
 
 2 
 
 27Z 
 
 ELEC. C *t P . , M . E . C . 
 
 03 
 
 OJ 
 
 07 
 
 33 
 
 3 
 
 279 
 
 STORAGE 3ATTERIES 
 
 3 
 
 03 
 
 7 
 
 03 
 
 11 
 
 280 
 
 PRIM. BATTtRIES/... 
 
 3 
 
 03 
 
 7 
 
 11 
 
 03 
 
 2*1 
 
 X-RAY APPARATUS < T 
 
 11 
 
 00 
 
 1 1 
 
 11 
 
 03 
 
 282 
 
 ENGINE ELEC. E B U I r . 
 
 03 
 
 
 
 07 
 
 3 
 
 11 
 
 283 
 
 ELEC. F G V I P . , N . F . C 
 
 00 
 
 03 
 
 1 1 
 
 11 
 
 11 
 
 284 
 
 TRUCK *, sUS BODIES 
 
 03 
 
 03 
 
 7 
 
 3 
 
 3 
 
 285 
 
 TRUCK TRAILERS 
 
 02 
 
 00 
 
 06 
 
 02 
 
 2 
 
 ?S6 
 
 MOTOR V f H . i P A P T S 
 
 0.2 
 
 00 
 
 06 
 
 J 2 
 
 2 
 
 287 
 
 AIRCRAFT 
 
 2 
 
 03 
 
 6 
 
 02 
 
 02 
 
 288 
 
 AIRCRAFT ENo. <; PTS 
 
 ii 2 
 
 3 
 
 06 
 
 02 
 
 02 
 
 7 S9 
 
 AIRCRAFT PROP r LLERS 
 
 00 
 
 00 
 
 1 1 
 
 11 
 
 1 1 
 
 290 
 
 A I P C P A. F T EQUIP. NEC 
 
 02 
 
 OJ 
 
 06 
 
 r 2 
 
 02 
 
 291 
 
 SH1P_ J LDG •; REPAIR* G 
 
 2 
 
 03 
 
 0( 
 
 02 
 
 02 
 
 ?92 
 
 bJAToLiiG S REPAIR' G 
 
 3 
 
 3 
 
 0/ 
 
 03 
 
 03 
 
 293 
 
 LOCOMOTIVES 3 PARTS 
 
 02 
 
 00 
 
 06 
 
 02 
 
 0? 
 
 294 
 
 RAILROAD s S R * T CAR 
 
 > c 
 
 00 
 
 06 
 
 ^2 
 
 02 
 
 295 
 
 MOTORCYCES/ BICYCLE 
 
 11 
 
 03 
 
 07 
 
 11 
 
 03 
 
 296 
 
 TRAILER COACHES 
 
 00 
 
 03 
 
 7 
 
 11 
 
 11 
 
 297 
 
 TRANSP'N c U I D . NEC 
 
 11 
 
 3 
 
 07 
 
 11 
 
 3 
 
 298 
 
 ENG'G i SCI'C INST. 
 
 03 
 
 
 
 05 
 
 "'3 
 
 11 
 
 299 
 
 MECH. v|£ AS * G DEVICE 
 
 03 
 
 03 
 
 07 
 
 03 
 
 11 
 
 300 
 
 AUTO'C TEMP. CONTR. 
 
 00 
 
 03 
 
 07 
 
 11 
 
 05 
 
 301 
 
 SURG. J MFD. I^STP. 
 
 03 
 
 03 
 
 7 
 
 11 
 
 03 
 
 302 
 
 SURG. APPL'S S SUPP 
 
 3 
 
 00 
 
 07 
 
 11 
 
 11 
 
 3 03 
 
 DENTAL EQUIP, s SUP 
 
 no 
 
 00 
 
 07 
 
 11 
 
 3 
 
 304 
 
 WATCHES/- CLOCKS/ PT 
 
 11 
 
 03 
 
 05 
 
 11 
 
 11 
 
 305 
 
 OPTICAL INSTRUMENTS 
 
 TO 
 
 3 
 
 5 
 
 11 
 
 3 
 
 306 
 
 OPTHALMIC GOODS' 
 
 00 
 
 00 
 
 05 
 
 11 
 
 5 
 
 307 
 
 PHOTO. ETUIP. ^ SUP 
 
 00 
 
 03 
 
 04 
 
 2 
 
 02 
 
 308 
 
 J E W E L F Y , 
 
 03 
 
 03 
 
 7 
 
 03 
 
 33 
 
 309 
 
 MUSICAL INSTRUMENTS 
 
 03 
 
 03 
 
 05 
 
 11 
 
 03 
 
 310 
 
 GAMES/ TOYS/ FTC. 
 
 03 
 
 
 
 07 
 
 3 
 
 11 
 
 311 
 
 SPORTING GOODS 
 
 03 
 
 03 
 
 07 
 
 03 
 
 11 
 
 312 
 
 PENS/ PENCILS/ ETC. 
 
 03 
 
 03 
 
 0:> 
 
 03 
 
 11 
 
 313 
 
 ARTIFICIAL FLOWERS 
 
 3 
 
 03 
 
 1 1 
 
 11 
 
 00 
 
 314 
 
 8UTT0NS/ NFEDLES 
 
 03 
 
 00 
 
 07 
 
 11 
 
 03 
 
 315 
 
 3R00MS s BRUSHES 
 
 11 
 
 00 
 
 11 
 
 11 
 
 11 
 
 316 
 
 HARD FLOOR COVER. 
 
 03 
 
 03 
 
 11 
 
 11 
 
 03 
 
 317 
 
 MORTICEANS GOODS 
 
 3 
 
 03 
 
 07 
 
 11 
 
 03 
 
 -23- 
 
TABLE A. 3-2 (Continued) 
 
 Sector 
 
 
 
 tnei 
 
 rgy Suf 
 
 jplies 
 
 
 Number 
 
 Sector Name 
 
 Coal 
 
 Crude 
 
 on 
 
 Electric 
 
 Gas 
 
 313 
 
 SIGNS i AD DISPLAYS 
 
 00 
 
 00 
 
 11 
 
 03 
 
 
 319 
 
 MISC. lANUF'S/ NEC. 
 
 C3 
 
 00 
 
 07 
 
 11 
 
 
 320 
 
 RmILPOADS 
 
 f;9 
 
 00 
 
 C 
 
 09 
 
 
 321 
 
 LOCAL. . .PASSENGER. . 
 
 
 00 
 
 09 
 
 09 
 
 
 322 
 
 MOTOR FRT TPANSP 
 
 
 00 
 
 09 
 
 11 
 
 
 323 
 
 WATER TRANSPORTAT 'N 
 
 09 
 
 03 
 
 09 
 
 11 
 
 
 32 A 
 
 AIR TRANSPORTATION 
 
 
 00 
 
 09 
 
 11 
 
 
 525 
 
 PIPE LINE TRA\SPO'N 
 
 
 00 
 
 
 11 
 
 9 
 
 326 
 
 TRANSPORT SERVICES 
 
 
 
 
 00 
 
 11 
 
 
 327 
 
 COMMUNICATIONS/ EYC 
 
 
 00 
 
 
 11 
 
 
 328 
 
 PADIO 5 TV a D C A S T ' G 
 
 
 03 
 
 
 11 
 
 
 329 
 
 WATER & SANIT. S F R V 
 
 
 00 
 
 
 11 
 
 
 33 
 
 WHOLESALE TRADE 
 
 
 no 
 
 
 11 
 
 
 331 
 
 RETAIL TRADE 
 
 00 
 
 03 
 
 
 11 
 
 
 332 
 
 BANKING 
 
 
 03 
 
 
 12 
 
 
 53 3 
 
 CREDIT AGENCIES 
 
 
 00 
 
 
 12 
 
 
 334 
 
 SECJPITY i COvtvjOD I T 
 
 (J 
 
 f i3 
 
 
 12 
 
 
 335 
 
 IVSURANCE CARRIERS 
 
 00 
 
 03 
 
 
 12 
 
 
 536 
 
 INS. AStNTS * 3 * < R S 
 
 .'0 
 
 J 3 
 
 
 12 
 
 
 33 7 
 
 W NEK-OCCUPIED D W E L 
 
 10 
 
 03 
 
 00 
 
 00 
 
 00 
 
 33 3 
 
 REAL ESTATE 
 
 no 
 
 JO 
 
 
 11 
 
 
 339 
 
 HOTELS >: LODGING 
 
 11 
 
 03 
 
 
 11 
 
 
 340 
 
 PtRS. <* REPAIR ScPV 
 
 11 
 
 3 
 
 
 11 
 
 
 341 
 
 3AR3FR % 3EAUTY SHP 
 
 PO 
 
 00 
 
 
 11 
 
 
 542 
 
 MISC. 3 0^. SERVICES 
 
 11 
 
 CO 
 
 
 12 
 
 
 343 
 
 ADVERTISING 
 
 00 
 
 03 
 
 
 12 
 
 
 . 344 
 
 MISC. PRO. SERVICtS 
 
 1 1 
 
 3 
 
 
 12 
 
 
 345 
 
 AUTO REPAIR ft SERV. 
 
 11 
 
 00 
 
 
 12 
 
 
 346 
 
 MOTION PICTURES 
 
 ^0 
 
 00 
 
 
 12 
 
 
 347 
 
 AMUSEMENT I REC.StR 
 
 11 
 
 03 
 
 
 12 
 
 
 34 3 
 
 DOCTORS % DENTISTS 
 
 'JO 
 
 33 
 
 
 12 
 
 
 549 
 
 HOSPITALS 
 
 11 
 
 00 
 
 
 11 
 
 
 350 
 
 OTHER 4ED. g HEALTH 
 
 00 
 
 30 
 
 
 11 
 
 
 351 
 
 EDUCATIONAL SERVICE 
 
 11 
 
 03 
 
 
 11 
 
 
 352 
 
 NON-PROFIT ORGAN* MS 
 
 11 
 
 3 3 
 
 
 12 
 
 
 353 
 
 POST OFFICE 
 
 11 
 
 03 
 
 
 12 
 
 
 354 
 
 OTHER FED. GOV. ENT 
 
 no 
 
 03 
 
 
 12 
 
 
 355 
 
 OTHER ST. &, LOC. EN 
 
 11 
 
 03 
 
 
 12 
 
 
 356 
 
 BUS. T R A V . , ENTER* T 
 
 00 
 
 3 
 
 
 JO 
 
 30 
 
 357 
 
 OFFICE SUPPLXFS 
 
 no 
 
 03 
 
 
 30 
 
 
 
 35 8 
 
 PERSONAL CO VS. F*'*». 
 
 11 
 
 33 
 
 
 12 
 
 01 
 
 359 
 
 GROSS PR1V. FIXED C 
 
 no 
 
 00 
 
 00 
 
 00 
 
 
 
 350 
 
 NET INVtNTDRY CHNbE 
 
 01 
 
 01 
 
 01 
 
 00 
 
 31 
 
 361 
 
 NET EXPORTS 
 
 01 
 
 01 
 
 01 
 
 01 
 
 01 
 
 562 
 
 FED. GOVT/ DEFENSE 
 
 11 
 
 00 
 
 
 11 
 
 
 363 
 
 FED. GOV'T/ OTHFR 
 
 n 
 
 00 
 
 
 11 
 
 
 364 
 
 ST. K LOC. GOV./ ED 
 
 11 
 
 00 
 
 
 11 
 
 
 36 5 
 
 ST. % LOC./ H,W*S 
 
 11 
 
 00 
 
 
 12 
 
 
 366 
 
 ST. < LOC./ SAFETY 
 
 11 
 
 00 
 
 
 12 
 
 
 567 
 
 ST. *. LOC./ OTHER 
 
 u 
 
 00 
 
 
 12 
 
 
 _o), 
 
REFERENCES 
 
 W. Leontief , The Structure of the American Economy 1919-1939 ? Oxford Univer- 
 sity Press, New York, 19^1 . 
 
 A. Sebald, "An Analysis of the Sensitivity of Large Scale Input-Output Models 
 to Parametric Uncertainties," CAC Document No. 122, Center for Advanced 
 Computation, University of Illinois, Urbana, IL. 6l801, November 197*+. 
 
 0. Morganstern, On the Accuracy of Economic Observations , Princeton University 
 Press, Princeton, N.J., 1950. 
 
 U. S. Department of Commerce Bureau of Economic Analysis, Input-Output Struc- 
 ture of the U.S. Economy 1967 > U. S. Government Printing Office, Washington, 
 D. C, 191k. 
 
 U. S. Department of Commerce Bureau of Economic Analysis, Definitions and Con- 
 ventions of the 1967 Input-Output Study , (mimeo) 197*+. 
 
 C. Bullard and A. Sebald, "Effects of Parametric Uncertainty and Technological 
 Change in Input-Output Models," CAC Document No. 156, Center for Advanced 
 Computation, University of Illinois, Urbana, IL. 6l801, March 1975. 
 
 C. Bullard, A. Sebald, D. Putnam, D. Am ado , "Stochastic Analysis of Uncertainty 
 in a U.S. Input-Output Model," CAC Document (forthcoming). 
 
 R. Knecht, "Reliability Measures of CAC-ERG Direct Energy Use Data," CAC Tech- 
 nical Memorandum No. 67, Center for Advanced Computation, University of 
 Illinois, Urbana, IL. 6l801, November 1975- 
 
 D. Smith & D. Simpson, "Direct Energy Use in U.S. Economy, 1967," CAC Technical 
 Memorandum No. 39, Center for Advanced Computation, University of Illinois, 
 Urbana, IL. 6l801, January 1975. 
 
 -25-