HE SUBCOMMITTEE ND STATISTICS ON POST OFFICE ERVICE EPRESENTATIVES 8.78395 Department of Commerce OLLER GENERAL COMPTROLLER GENERAL OF THE UNITED STATES WASHINGTON. D.C. 2QS48 B-78395 Dear Mr. Chairman: In your letter of March 26, 1971, you advised us of your Subcommit- tee's recent investigation of the 1970 decennial census conducted by the "? 4-- Bureau of the Census and stated that during the investigation State and local officials recommended alternatives>.to,the.current practice-of con- ducting __,_ a- rz-~. census every decade. ~-?7~%,.Pi-ran-,c :aMey?w~~ Specifically these alternatives were (1) the~blishment of a mid-decade census and (2) the use of special sample surveys to be taken on a regular basis that would utilize in-depth and de- tailed interviews. You requested that we evaluate which method--a mid- decade census or the use of special surveys --would yield more fruitful re- sults, taking into consideration the costs and the final product. In accordance with discussions with your office, we agreed to provide you with (1) estimates of the costs--in current dollars--of four mid- decade census proposals presented during hearings held by the Subcommit- tee on Census and Government Statistics in 1961 and 1962 and (2) whatever information we could develop to indicate which method--a mid-decade census or special surveys --would yield more fruitful results, taking into consid- eration the costs and the final product. This letter transmits the cost estimates of the four mid-decade census proposals. We plan to transmit the information relative to the second phase of your request in September 1971. At our request the Bureau of the Census developed cost estimates for each of the four mid-decade census proposals. The estimates, shown below, were made available for our review on June 4, 1971. The Bureau's estimates show what the four mid-decade census proposals would cost as of January 1, 1971, on the basis of work-load projections to 1975. Estimated cost Proposal (millions) Level I $154 Level II 169 Level III 170 Level IV 228 A detailed presentation of these estimates is shown in enclosure I. A description of the four proposals and our comments on certain factors used by the Bureau in developing the cost estimates are presented in sub- sequent sections of this report. 50 TH ANNIVERSARY 7921- 1971 B-78395 In view of your urgent need for this report, we did ? very limited amount of testing and verification of the estimates and the underlying as- sumptions made by the Bureau. (See p. 3.) As evidenced by the comments presented below regarding certain assumptions made by the Bureau concern- ing price and work-load increases (see p. 5) and other factors noted rel- ative to the cost estimates (see p, 6), the Bureau's estimates should not be considered to be precise. However, they do appear to fall within rea- sonable limits. MID-DECADE CENSUSPROPOSALS The following descriptions of the four mid-decade census proposals were presented by the Director, Bureau of the Census, in his testimony before your Subcommittee on June 10, 1971. The level I proposal provides for (1) a head count of the entire popu- lation and the accumulation of the basic characteristics of name, address, relationship to head of household, sex, race, and age and (2) a collection of information concerning persons of Spanish cultural heritage. The Direc- tor indicated that data collected could be tabulated for every State, every county and its subdivisions, and every census tract1 in the metropolitan areas and for areas as small as enumeration districts.2 Also the data for enumeration districts could be arranged to provide data for larger areas, such as voting districts, school districts, and other special areas for which such data are needed. The Bureau's estimated cost for this proposal is about $154 million. The level II proposal would seek answers to all of those questions asked of 100 percent of the population in the 1970 decennial census. This would include the same type of information as the level I proposal plus additional information on basic housing characteristics, such as type of unit, condi- tion, occupancy, rooms, tenure, vacancy, and plumbing facilities. The Di- rector said that this information could be tabulated in the same degree of geographic detail as that under the level I proposal and could be tabulated also by city block for the urbanized areas within the metropolitan areas. The Bureau's estimated cost for this proposal is about $169 million. 1Census tracts are small areas into which large cities and their adjacent areas have been divided for statistical purposes. 2An enumeration district is a small area which is assigned to an enumerator and which must be canvassed and reported on separately. In most cases an enumeration district contains approximately 250 housing units. 2 B-78395 The level IV proposal is a replication of the 1970 decennial census and provides for the collection of the same type of data as the level II proposal plus data--to be gathered by a 20-percent sample of the popula- tion--on education, income, employment and unemployment, commuting, migra- tion, foreign origin, type of work, method of heating housing units, fuels used, age of structure, and air conditioning. Bureau officials informed us that the tabulations and publications program for this proposal could be similar to that executed for the 1970 decennial census. The Bureau's estimated cost for this proposal is about $228 million, The level III proposal provides for the collection of the same type of data as the level IV proposal. All the data, however, would be collected from a sample of 25 percent of the population. Bureau officials informed us that the data would not be tabulated for any area with a population of less than 25,000 because of the unreliability of the sample data collected in such areas. The Bureau's estimated cost for this proposal is about $,170 mil.lion. TESTING REASONABLENESS OF COST ESTIMATES The Bureau used the cost of the 1970 decennial census of $218 million-- comprising actual costs of $151 million for the period 1964 through fiscal year $970 and budgeted costs of $67 million for postcensus work to Decem- ber 31, 1972--as the basis for estimating the costs for each of the four mid- decade census proposals. The estimated cost for the level IV proposal (rep- lication of the 1970 census) was determined by adjusting the total of $218 million for (1) annual increases in prices and costs for personnel ser- vices from fiscal year 1964 to January 1, 1971 (applied against the actual cost portion of the total costs), (2) costs of increases in work loads pro- jected to 1975, (3) costs of nonrecurring activities, and (4) costs of in- creased personnel services for anticipated overtime requirements. The costs for the levels I, II, and III proposals were determined by eliminating from, or adding to, the estimated cost for the level IV proposal certain costs applicable only to such proposals. The assumptions as to price increases for updating the costs of the 1970 decennial census to January 1, 1971, are for the most part not well sup- ported. The Bureau was handicapped in this regard because of the lack of readily available historical information concerning price increases. The Bureau's assumptions as to work-load requirements for each of the proposals were based principally on personal judgment. Our tests of the cost estimates included (1) evaluation of the method- ology used by the Bureau in estimating the costs of the mid-decade census proposals, (2) verification of the accuracy of the cost of the 1970 decen- nial census by tracing the majority of fiscal year 1970 program costs to the Bureau's cost accounting summary documents, (3) examination of the Bureau's 3 B-78395 working papers supporting the costs attributable to three program activi- ties--mailing preparation, data collection, and data processing (these ac- tivities accounted for about $149 million, or 68 percent of the total costs of the 1970 decennial census), and (4) discussions with Bureau employees regarding the basis for the assumptions used in developing the cost esti- mates. Due to the limited time available for our review and the fact that many of the Bureau's assumptions were based on personal judgment, we were unable to evaluate the reasonableness of the assumptions. Our review resulted in several adjustments in the various costs compris- ing the total estimated cost for each of the mid-decade census proposals. Because these revisions were generally offsetting, however, the net effect on the total estimated cost for each of the mid-decade census proposals was negligible. FACTORSWHICH COULDSIGNIFICANTLY AFFECT VALIDITY OF COST ESTIMATES Several significant factors in relation to the Bureau's cost estimates for the four mid-decade census proposals should be noted by the Subcommit- tee in its evaluation of the costs of a mid-decade census. The cost estimates are based on January 1971 prices and therefore do not include price increases that have occurred, or can be expected to occur, be- tween January 1971 and the time that costs would be incurred for a mid- decade census. For example, the cost estimates do not provide for the in- crease in the cost of mailing census questionnaires resulting from postage- rate increases which occurred in May 1971 or for any Federal pay increases which may occur after January 1971. Another factor is the Postal Service charge for checking and updating Wy.: mailing addresses. For the 1970 decennial census, the Post Office Depart- . ment, under a negotiated contract, charged the Bureau $2.9 million on the basis of 8 cents for every address checked and updated. The Bureau esti- mated that these costs for a future Postal Service contract would amount to $3.9 million. The Postal Service is required by legislation to become self-sustaining by January 1, 1978. Bureau employees informed us that postal officials had not been contacted to obtain their views concerning what the costs may be for such work for a mid-decade census. B-78395 An additional factor is the variance in the method of data collection between the level IV proposal and the levels I, II, and III proposals. The costs for the level IV proposal were based on the assumption that 60 per- cent of the questionnaires would be returned by mail and that the remaining 40 percent would be collected by enumerators. This ratio was used in the 1970 decennial census. Costs for the other three proposals, however, were based on the assumption that 90 percent of the questionnaires would be re- turned by mail and that the remaining 10 percent would be collected by enu- merators. The estimates for the levels I, II, and III proposals are there- fore not comparable to the estimate of the level IV proposal because of cost variances associated with the different mail-back ratios. Bureau officials informed us that the cost estimates for the four mid- decade census proposals had originally been determined on the basis of the 90 to 10 mail-back ratio. Because of some concern over the indicated higher costs associated with the 90 to 10 mail-back ratio, the cost of the level IV proposal was recomputed on the basis of the 60 to 40 ratio--the same as that used in the 1970 decennial census. Bureau officials informed us further that because of time constraints they considered it impracticable to recompute the other proposals on this same basis. As indicated earlier the Bureau made a number of necessary assumptions in revising the costs of the 1970 decennial census to arrive at the esti- mate of the cost for each of the mid-decade census proposals. The recom- puted cost estimates for projected work-load increases amounted to about $19 million and for the price increases amounted to about $16 million. Fol- lowing are some of the principal assumptions made by the Bureau. --The number of housing units in 1975 would be 12.2 percent higher than in 1970 based on the estimated housing-unit starts and losses for the 1970-75 period. --The cost of personnel compensation for both permanent and temporary employees for the 1970 decennial census would be increased at the rate of increase in Federal pay for the June 1964 through January 1971 period. --The cost of personnel benefits was based on the application of sev- eral estimated percentages to the 1970 decennial census costs for personnel services as adjusted for salary increases. The estimated percentages were based on (1) increases in Federal Insurance Contri- butions Act taxes paid by temporary employees and (2) the ratio of permanent and temporary employees in the various census program areas. , . B-78395 --The cost of travel and transportation of persons was increased by the rates of increases in Government reimbursements for the use of pri- vately owned automobiles during the 1964-71 period. Other cost ele- ments, such as travel by common carrier and per diem, were not con- sidered. The costs incurred for reimbursements for use of privately owned automobiles accounted for the major part of the costs in travel and transportation of persons for the 1970 decennial census. --The costs of printing and reproduction and of other services were as- sumed to be higher than the actual costs of the 1970 decennial census by 25 percent and 75 percent, respectively, of the rates of Federal pay raises during the 1964-71 period. Bureau officials could provide us with no firm bases for these assumptions. --The costs of rents, communications, utilities, supplies and mate- rials, equipment, and the transportation of things were all in- creased over the actual costs of the 1970 census by 2 percent a year. A Bureau official informed us that this percentage had been arrived at by considering the yearly increases indicated by the consumer price index and the wholesale price index. IMPROVEMENTSRECOMMENDEDBY THE SUBCOMMITTEEON CENSUSAND STATISTICS The Bureau's cost estimates for the four mid-decade census proposals could be significantly understated should the Bureau implement, in a mid- decade census, the improvements recommended by the Subcommittee on Census and Statistics in a "Report on Accuracy of the 1970 Census Enumeration," dated December 17, 1970. At the time of our review, the Bureau had developed estimates for the costs of some of these improvements for each of the four mid-decade census proposals. The cost of the improvements, however, were not included in the cost estimates presented on page 1. The recommended improvements and the Bureau's estimated cost for some of the improvements are shown in enclo- sure II. We did not verify the accuracy of the estimates shown in enclosure II. Bureau officials advised us that the estimate of $20 million for additional publicity was suggested by the Subcommittee as the minimum amount required to achieve an adequate public awareness. The Bureau had not developed any information concerning the composition of such a publicity program. The Subcommittee recommended that the Bureau's temporary district of- fices remain open for a longer period of time after completion of the enu- meration, to enable local community officials who complain of an undercount 6 B- 78395 to deal directly with the Bureau official having responsibility for carry- ing out the census locally. The cost estimates developed by the Bureau for this improvement were based on keeping the district offices open 1 addi- tional month. We plan to make no further distribution of this report unless copies are specifically requested, and then we shall make distribution only after your agreement has been obtained or public announcement has been made by you concerning the contents of the report. Members of my staff will be available to discuss these matters with you further if you desire. Sincerely yours, Comptrolier'General of the United States The Honorable Charles H. Wilson, Chairman Subcommittee on Census and Statistics C Committee on Post Office and Civil Service House of Representatives . ENCLOSUREI BUREAUOF THE CENSUS COST ESTIMATES FOR MID-DECADE CENSUSPROPOSALS . . Level I Level II Level III Level IV Program activities proposal proposal proposal proposal (000 omitted) Program planning, direction, and review $ 4,431 $ 4,802 $ 7,439 $ 7,439 Pretests and dress rehearsals 720 800 1,100 1,100 Geographic work 13,077 13,077 13,077 13,077 Mailing preparation 14,706 16,139 14,090 13,882 Data collection 98,587 107,996 80,001 129,899 Data processing 7,272 8,920 32,844 34,116 Publications 554 869 2,033 3,730 Sampling, statistical studies, and evaluations 2,100 2,100 4,678 4,478 Outlying areas 1,603 1,972 1,306 2,866 Publicity 521 521 521 521 History 172 172 172 172 Improvements of data usability 310 402 1,458 1,548 Office space, Bureau general expenses, etc. 8,049 8,717 8,840 11,604 Capital outlay 2,215 2,461 2,506 3,413 Total $154.317 $168,948 $170.065 $227,845 ENCLOSUREII BUREAUOF THE CENSUS COST ESTIMATES FOR CERTAIN IMPROVEMENTS RECOMMENDED BY THE SUBCOXMITTEEON CENSUSAND STATISTICS Level I Level II Level III Level IV Improvements proposal proposal proposal proposal (000 omitted) Additional publicity (note a) $20,000 $20,000 $20,000 $20,000 Keep temporary district offices open for 1 additional month after completion of the enumeration (note b) 3,507 3,750 4,296 4,850 Expand precanvas survey 1,250 1,250 1,250 11250 Total a Amount suggested by the Subcommittee on Census and Statistics as the minimum .:mount required to achieve an adequate public awareness. b Includes release of preliminary population and housing counts. Note: The Bureau had not developed cost estimates for the following recommenda- t - tions by the Subcommittee. - 1. Increase of two or three regional field offices. 2. Eliminate unused census questionnaires. 3. Extend postenumeration check. 4. Improve postenumeration check. 5. Improve coordination between Census Bureau's and Post Office Department's population growth estimate program. 6. Arrange for Executive order making better coordination possible between the Census Bureau and other Government agencies.
Bureau of the Census Cost Estimates for Mid-Decade Census Proposals
Published by the Government Accountability Office on 1971-07-27.
Below is a raw (and likely hideous) rendition of the original report. (PDF)