l LEC-C(7)M RF(,OR!) SHEET DEPARTME~NT O1' AGRIC I I IUR A L( N I A\D RU RAL SOC:IOLOGY- 1.- X I'RN I . N' RII 4rDnRFS ' (HANGFD T Ni -y - 9 _ EVALUATIN9_ 1 2 3 4 8 5 _62?-5 i- 7 g. 1 27 _5 _ : 3 4 - 5 :- - = ELEC- COM 8 2- 3 4 5 68 -r 6 9 1 ' .o - 2 3 4-- S2- 3 4 $ 6 _. a = 3 AN EXPERIMENTAL ACCOUNTING SYSTEM 87WB9 I: 2 3 5 6 :.3t 8 9 _' 1 2 3 4 5 8 8 9- :2 g 4 5 8 3' 8 i _ 4 5 -8 7 8 9 1 - , 2 -3- -4 5 8 : 8 2K lo 2 5 4- 5 6- ? g 3Q i E 3 4 : 8 9 _ t 2 3 4 5 -4 5 86 3 8= 9 - ______AGRICULTURAL ECONOMICS ____ __ I 2 3 SERIESJ 18 74 6 ? S2. 3 4 -4 6- 7 8 9: W 2 3 4 -5 _8 - -2 3 4: 5: 6 ? 8_ g i 2 el 4 5 68 7 8 _9 -0 2 &O 4: 5: 8: 7 8 9 3 4 I 2 3 4 8c -w 8 ._, S23- 4 AGRICULTURAL EXPERIMENT STATION 68 7, 8 9 _ 1i 2 3 _OF AUBURN UNIVERSITY 8_ 7 8 9 ka 2 3 6 7i & 8 91 _ t 2 3 rE. V. SMITH, DIRECTOR AUBURN, ALABAMA -s ?- 8 9 -- 2- 3- DECEMBER 1968 ag- 2 3 6i 7? 8 ? 3 4 8 7 8 f: 9 - I ad 3 4 5 - - ? 8 9= Q 2 l &pr 3 4: . 5- 6 7 8- - 2" 3 : 4 - 5 8 7 : 8 9 - I 3 4 -$ a. = 5 q ' - 9 3 - I I 1 TABLE OF CONTENTS INTRODUCTION . . . . . . . W . . . . . . . . . COSTS AND RETURNS IN CENTRAL RECORD KEEPING PROGRAMS COSTS OF INPUT CONVERSION OTHER DATA PROCESSING COSTS USE BY FARMERS AND EVALUATION C CHARACTERISTICS OF COOPERATORS UNFAVORABLE COMMENTS BY COOPERA FAVORABLE COMMENTS BY COOPERATC DESIGN OF THE ELEC-COM SYSTEM AUTOMATED INPUT AND CODE STRUCT[ PROCESSING * ,., * . . ERROR DETECTION AND CORRECTION MONTHLY ACCOUNTING .* . .. . QUARTERLY SUDMMARIES . . *.. YEARLY SUMMARY AND TAX OUTPUT SUMARY AND CONCLUSIONS *... LITERATURE CITED .+9 .* .**.*. " " " " " I I 0 " *R 0 0 00 0 6 0 SYSTEM " " " 0" "0"0f0" * 0 @0"9 0009"fw * " " f "0 " " 6" 00090 a9" " 00 "990 "0 "0" 0 " 00" "9" " 0f "0"00 90" 0 " f U"9S9"9"9"9" 0 0"9"9"0"S"0" U * *." " " 9 0 0 00" 0 * 000 09I "0 " "0 " f 00 " " 00"90 " 0 2 3 4 U 14 14 16 17 18 20 2v 30 32 33 36 42 47 EVALUATION OF ELEC COM: AN EXPERIMENTAL AUTOMATED ACCOUNTING SYSTEM* Victor M. Yellen and Bill R. Miller** Programmer nd Assistant Professor, Department of Agricultural Economics and Rural Sociology Modern management practices in today's expanding farm business require an increased amount of management information. The size of the individual farm business in the South is growing rapidly. According to the 1964 Census of Agriculture for 10 Southern States, there were 20 per cent fewer farms in 1964 than in 1959, but commercial farms with more than 5,000 dollars gross sales increased in number. The number of farms with more than 20,000 dollars gross sales almost doubled. Data resulting from larger businesses speedily summarized into a form suitable for decision making will greatly enhance manage- ment efficiency. Electronic record keeping at a central location is one answer to improving management skills needed for growth. Advances have been made in automating centralized accounting systems. Central processing of farm accounts has progressed from hand computation, to tabular machine computation, to today's electronic processing. Growing interest among farm leaders, farmers, and bankers in Alabama for knowledge about the use of computers in management resulted in a project named Elec-com initiated to provide a source of information about cost, use, and design of a computerized system. The project was accomplished in three phases: (I) A survey was made of all known central record keeping programs, 26, to obtain cost and returns data that could be compared with the experimental system Elec-com. (2) The experimental system, Elec-com, was tested in use among 30 farmers in the Sand Mountain area of Alabama. A field agent was hired to assist the voluntary group of farmers. *Research in which this report is based was carried out under Ala. 1-033. **Resigned September, 1968 3 (3) The Elec-Com system, including computer programs, was designed specifically for using automated input via an optical scanner or an optical reader. Costs and Returns in Central Record Keepine programs The survey of all known central record keeping programs conducted in Fall, 1966 gave results useful in an economic comparison of Elec-Com to other systems. Eighteen replies were received from 26 program supervisors. Thir- teen programs were under the direction of various university personnel and five were directed by private enterprise. Most prominent of the private groups were programs now under the direction of the Farm Bureau and the Farm Journal. The largest programs were university related, up to 1,367 farmers in one state; but some private groups were forecasting 2,000 cooperators in 1968. One important interested private group was banks. The American Bankers Association has estimated that 100 banks are now offering some type of com- puter service.(5) Several banks were participating in university and private programs included in the survey. Services offered by programs surveyed were in five categories: (1) tax records, (2) tax filing service, (3) cost and return for each enterprise on the farm, (4) general farm management analysis, and (5) research and education. Commerical services tended to be highly oriented to tax filing whereas the university-related programs tended to be management analysis-research oriented. The oldest programs were university related; three in the survey were more than 35 years old. New programs were closely related to availability of computers. Ten new programs, all less than 5 years old, began as computerized systems. Results of the survey were used in evaluating the attainment of study objectives. As stated in the introduction, objectives were to minimize two kinds of costs: (1) processing costs and (2) cost of professional and clerical staff. Costs of Input Conversion Automation figured on an hourly basis was expensive. An optical scanner rented for approximately $425 per month. To determine an hourly charge, the monthly rental was divided by 160 hours (40 hours per week for 4 weeks) and $2.65 was obtained as the hourly cost of the scanner. The same calculations were computed for a keypunch ($50 per month) and an hourly charge of $.31 was obtained. An optical scanner was thus 755% costlier per machine hour of use than a 026 keypunch. A clerk operator was paid a minimum wage of $1.50/hour and raised the total cost per hour of scanning to $4.15. Table 1 gives the comparative rate for keypunching and verification of transactions obtained from the survey of central record keeping program. Transactions per hour of key punching were highly variable because of differences in amount of coding. Table 2 gives the cost of key punching and verification from the production rates in Table 1. The cost of computer cards, electricity and repair, aze deleted since they are incurred by each system. Simple economic analysis was applied to determine whether a system should change from its present non-automated input processing to optical scanning. Assuming that a system's code could express a transaction in less than 26 numbers: Monthly Added Value value of key $.31 (320,000 - Monthly Transaction) = punch 2000 .availability Monthly Added Costs $425 = machine rental costs $1.50 (number of transaction) = total labor cost 2000 Table 1. Hours of Keypunching and Verifying in Ten Electronic Record Systems, 1966 Trans- Hr. key Trane./hr. Hr. Trans./hr. ver- actions! punching key verify- ifying Stat moe punching ing C1) (2) (1)/(2)(3 4 1)()5 A 2,199 10. 219.9 10. 219.9 D 6,985 35. 199.6 15. 465.7 E 19,110 262.5 72.8 262.5 72.8 H 132,736 300. 442.4 300. 4424 I 11,666 100. 116.7 100. 116.7 J 9,250 200. 46.2 120. 77.1 L 8,000 30. 266.7 30. 266.7 0 7, 425 100. 74.25 76. 97.7 R 10,600 21. 504.8 20. 530.0 T 23 ,862 92.6 257.8 78. 305.9 Summary 231,833 1151.1 312-.1 1011.5 3 56.8 Alabama 601 .3 200OO. .. __---2/_ 1/Keypunching and verifying done simultaneously by an optical scanner. Table 2. Costs of Keypunching and Verifying in Eleven Electronic Record Systems, 1966 Total hr. Cost trans. Cost Av. trans. Trans. keypunch. Cost keypunch Hr. Ver. Cost trans. ver. cost State month month hr. keypun. (2)(3)/(1)= month hr. ver. (5)(6)/(1)= (4) + (7) = (1) (2) (3) (4) (5) (6) (7) (8) Hr. Hr. Hr. A 2,199 10.0 $3.50 $.016 10.0 $3.50 $.016 $.032 D 6,985 35.0 $3.511/ $.017 15.0 $3.50 $.007 $.024 E 19,110 262.5 $2.20 $.030 262.5 $2.20 $.030 $.060 H 132,736 300.0 $3.65 $.008 300.0 $3.65 $.003 $.011 I 11,666 100.0 $3.501/ $.030 100.0 $3.501/ $.03 $.06 J 9,250 200.0 $1.812/ $.039 120.0 $1.812/ $.023 $.062 L 8,000 30.0 $3.00 $.011 30.0 $3.00 $.011 $.022 0 7,425 100.0 $1.96 $.026 76.0 $2.01 $.021 $.047 R 10,600 21.0 $2.50 $.005 20.0 $3.50 $.007 $.012 T 23,862 92.6 $3.50 $.014 78.0 $3.50 $.011 $.025 Alabama 601 .33/ $4.154/ $.002 a$.002 1/Average commercial rate. 2/$.31/hr. for keypunch + $1.50/hr. labor. 3/2,000 transactions/hr. or 1,000 sheets/hr. punched and verified. 4/$2.65/hr. + $1.50/hr. labor = $4.15/hr. for punching and verifying. a1 Monthly Reduced Value $0.00 Monthly Reduced Costs Average Number of cost k transactions = Present kpv 2000 total costkpv Value of keypunch availability is equal to the number of hours per month not utilized by the scanner's keypunch multiplied by $.31 (cost of keypunch/hour). A keypunch is obtained as a part of the rented rate of the scanner. Number of hours per month is equal to 320,000 transactions (number of transactions which could be processed in 160 hours) minus number of transactions processed by the scanner in a month divided by 2000 transactions/hour. Change-over point (COP) was defined as that volume of transactions per month at which optical scanning's total cost (TCos) equals the total cost of keypunching and verifying (TCkp.v) for that volume of transactions presently to be processed by the account system. TCos = Monthly added costs - Monthly added value. TCkp.v = Monthly reduced costs At the change-over point: TCkp.v = TCos = COP TCos = Number of transactionscop . ACkp-v Number of transactionsco p = TCos = COP ACkp-v Grouping constants from the analysis resulted in the following equation for change-over-point: COP = $375.40 - $.000905 (volume transactions/month) average cost/transaction 8 Table 3 indicates the change-over points for the systems responding to the previously mentioned mail questionnaire. The code structure of many of the systems may not fit in less than 26 numbers required by an Elec-Com transaction. The change-over points in Table 3 are thus biased downward in some relationship to size of code.. The amount of this bias is one representation of the efficiency of the Elec-Com code. In other words, there would be no bias in the change-over-points if each system had a code as efficient as Elec-Com. Based on Table 3, 50 per cent of the reporting States could have been utilizing scanning; and one State could have changed over when a 4.3 per cent increase in volume was obtained. Other Data Processing Costs Processing costs other than costs of input conversion were divided into 3 types: (1) variable computer processing costs, (2) variable labor costs, and (3) fixed labor costs. Computer processing costs were determined by an assembly language timing subroutine in which computer time was evaluated at two dollars per minute. Table 4 is an enumeration of the computer processing costs including scanning by individual program. Variable labor costs were for a clerk who was responsible for the physical handling of the transactions. This included opening of the envelopes, correction of returned transactions and mailing of processed transactions. It was estimated that a full-time clerk could process 1,000 transactions per hour or 160,000 transactions per month. The other expenditure of variable labor cost was expense of an assistant county agent. The assistant county agent's principal duty,after five hours of initial instruction, was to evaluate, with the cooperator, the quarterly and yearly farm management and tax output. This was approximately one hour per quarter or four hours per annum. 9 Table 3. Change-Over Points in Ten Electronic Record Systems, 1966 Trans- Average cost! Hr. scanner Number of Should be System actions! transaction available for transac- utilizing MOO key punching tions at scanning COP A D E H J L 0 R T Average. No, ?x199 6,985 19,110 132,736 11,666 9,250 8,000 7,425 10,600 23,862 23,235 $ .032 .024 .060 .016 .025 .062 .022 .047 .013 .025 $ .026 Hr. 158.9 156.5 150.4 93.6 148.0 155.4 156.0 156.3 154.7 148.0 148 .4 No. 11,669 15,378 5,968 15,955 5,921 5,920 16,734 7 ,844 28,139 14,152 13,630 No No Yes Yes Yes Yes No No No Yes Yes -mob 10 Table 4. Variable Processing Costs of the Computer and Scanner in Elec-Comn's Accounting System, 1966 Dollars! Dollars/788 Minutes of Operation computer time action per 1000 transactions Scanning .003 2.36 1.5 Checking & correcting .002 1.58 1.0 Addition of master data .001 .79 .5 Sort 1 .0006 .47 .3 Journal listing .005 3.94 2.5 Sort 2 .0006 .47 .3 Merge .0006 .47 .3 Income statement .003 2.36 1.5 Schedule F .002 1.58 1.0 Depreciation schedule .023 .572! 11.5 Annual data processing costs 14.59 I/Farms nationally sent in annually, on the average, 788 transactions. aVFarms averaged 25 depreciable items. Table 5. Total Variable Cost Per Cooperator of Elec-Corn's Accounting System, 1966 Average data processing cost Clerk Supplies Assistant county agent Average variable costs 14.59 1.181/ 4.*00 27.08?/ 46.85 1/160,000 transactions/month at $1.50/hour. 2VFour one hour visits per year at $6.77 per visit concerned with record keeping procedures and explanation of record output. Time does not include management counseling. "R~ Variable costs also included four dollars per cooperator for scan sheets, postage, envelopes, and reporting forms. All variable costs are enumerated in Table 5. Fixed costs to be borne by the system are salaries of two state office personnel: (1) a state supervisor and (2) a computer programmer. The state supervisor would devote his time to the supervision of the processing pro- cedures, supervision of county personnel, and to the determination of any changes that would improve the system. The programmer would be responsible for accomplishing any changes determined necessary by the state supervisor. The salary of the state supervisor would be approximately $10,000 per annum and that of the programmer approximately $8.00 per hour. If the number of cooperating population is 250, the average fixed cost becomes $40.19. Costs of initial programming and initial contact with cooperators is excluded. How do these costs compare to other systems? The comparison of 14 farm processing systems to Elec-Com showed that total costs of Elec-Com were less than the reported full costs of all other systems except one. One system not yet in operation estimated all costs to be less than Elec-Com, but the pro- posed system is to offer only tax accounting as a service, Table 6. Table 7 is the summarization of professional and clerical time spent by 15 farm accounting systems for checking, editing and transposing. The average cost for all systems responding to the survey was $24.79 per cooperator per annum. This figure was compiled by evaluating professional time at $5.20 per hour ($10 M per annum) and clerical time at $1.50 per hour. Utilization of Elec-Com in which the cooperator performs these functions provided the system with a reduced cost. 12 Table 6 Annual Cost to Cooperators of 15 Farm Record Systems, 1966 System Cost to coopeator cooperators Cost covered?/ cooperatoro Dol. No. A 100 30 Data processing costs C 75 800 Forms & data processing D 221 55 Processing costs F 120 50 All costs G 168 295 Processing costs H 105 1367 All processing cost I 50 300 Forms & data processing J 75 200 Processing costs L 100 160 2/3 of total costs M 112 150 All costs N 54 All costs 0 130 90 Office processing S 250 93 1/5 of total cost T 60 830 Processing Alabama 87.043/ 250 All fixed & variable costs iFrom statements of supervisors of the systems. /Undetermined as system was not in operation. 3/Fixed costs of $10,048 allocated for 250 cooperators or $40.19 per/ a Nnum.tr Table 7. Professional and Clerical Time and Cost for Checking, Editing and Transposing Transactions of 15 Farm .Accounting Systems, 1966 Coopera- System tors Checking & Editing Profes- Clerical sional hrs. per hrs. per farm per farm per month month Transpos ing Profes - s ional hr s. per farm per month Cler ical hrs . pe: farm pemot Emonth Checking, Editing & Transposing SProfes- r siTonal ,r hrs. per farm per month Pro fes" s ional hrs. per farm per annum Clerical hrs. per farm per month Cle rical hrs. per farm per annum Total Time Profes- Clerical sional costs per cost per farm per farm per annum 2/ annum 1/ Hr. Hr. 1.00 .19 .91 .08 1.25 .50 12.00 2..2 8 10-.92 .96 15.00 6.00 .50 6.00 1.58 .15 .041 18.96 1.80 .12 1/ Evlae at 5.2 per hour for 1920 hours at $10,000 per annum. YEvaluated at $1.50 per hour. Hr. .30 Hr . 1.00 1.*30 A B C D E F H I J L 14 0 Q T No. 30 30 800 55 525 50 2 95 1367 300 200 160 150 90 100 830 Total cost per farm per annum Hr. .70 .19 .91 .08 1.00 .50 .20 .83 .03 .01 Hr. 1.30 3.10 .86 .40 .53 .10 1.00 .25 1.38 .30 .06 .25 Hr. 1.*00 2.60 3.10 .86 1.90 .53 .55 1.60 .50 2.63 1.20 .06 1.50 .45 .60 .25 1.25 .90 .30 .75 .12 Hr.' 12.00 31.20 37.20 10.32 22.80 6.36 6.60 19.20 6.00 31.56 14.40 .72 Dol. 0.00 63.40 11,.86 56.78 4.99 78.00 31.20 0.*00 0.00 0.00 31.20 0.00 98.59 9 *36 0.62 Do 1. 18.00 46.80 0,00 55.80 15.48 0.00 0.00 34.20 9.54 9.90 28.80 9.00 47.34 21.60 1.08 Do l . 18.00 109.20 11.86 112.58 20.47 78.00 31.20 34.20 9.54 9.90 60.00 9.00 145.93 30.96 1.70 r r ~ rrr r _ 1 _r _ _ ~ m + ~ _ h_ _ 1- ~C __- R -! + ~ m >W ,, __ _ __ __ _ __ __ _ 14 Use by Farmers and Evaluation of the System The transactions of each cooperator of 30 in the test group were ob- tained from January 1, 1966, and were received by the state office in April, 1966. The system continued throughout the year with bimonthly journal listings, quarterly income summaries, a yearly depreciation schedule, and a yearly schedule F. Thirty cooperators received year-end reports although all appeared to have not reported enough information for a complete report. Characteristics of Cooperators An average cooperator could be described as a 48 year old high school graduate whose gross income was $19,000 in 1965 and who spent $24 last year for preparation of income tax forms. Two-thirds of the cooperators would keep records if they were not required to do so for income tax preparation; and one-fourth had kept records broken down by enterprise, All cooperators at time of initiating the Elec-Com system kept some kind of records. Elec-Com cooperators were above average in most respects, Table 8. In terms of gross farm income they were similar to the class of commercial farmers whose numbers have been increasing in recent years. Data in Table 8 were collected by personal interview with 23 Elec-Com cooperators. During the interview, four yes-no questions were asked and the results appear in Table 9. In general farmers thought the system was easy to learn and use. They were equally divided on the use of non-cash inputs that are essential for management information. What types of enterprises were operated by the 23 cooperators? During 1966, 28 enterprises were offered by Elec-Com. All were requested by coopera- tors except for peanuts and horticultural crops: 15 Table 8. Socio-Economic Characteristics of 23 Elec-Com Cooperators V. Alabama Farmers, 1966 Average Average A abama Characteristic cooperator farme1r Age 48 years old 52 years old Education 12 years 7.9 years Value of farm products 19,000 dollars 804 dollars Row cropland operated 80 acres 40 acres Improved pasture 50 acres 65 acres I/Includes commercial, part-time and part-retired. Source: U.S. Bureau of the Census, 1964 United States Census of Agri- culture. Preliminary Report, AC 64-P1, pp. 2-3. Table 9. Responses of 23 Elec-Com Cooperators to Four Yes-No Questions Concerning Farm Records, 1966 Question Yes No No response No. Pct. No. Pct. No. Pet. Was the Elec-Com 2 8.7 20 87.0 1 4.3 code difficult to use? Were non-cash in- puts necessary for 11 47.8 11 47.8 1 4.4 proper accounting? After practice can the coding be done 21 91.3 2 8.7 0 0.0 without help? Should the Farm Bureau offer Elec-Com as a 5 21,7 11 47.8 7 30.5 continuing program? 16 1. Cotton 15. Independent pullets 2. Corn 16. Contract layers 3. Beef cattle 17. Independent layers 4. Hogs 18. Dairy 5. Capital goods 19. Johnson grass 6. General farm 20. Peanuts 7. Field tomatoes 21. Oats 8. Southern peas 22. Wheat 9. Sweet potatoes 23. Grain sorghum 10. Snap beans 24. Soybeans 11. Pimentoes 25. Alfalfa 12. Lima beans 26. Coastal burmuda 13. Pickling cucumbers 27. Temporary grazing 14. Bell peppers 28. Home gardening Tape recorded interviews to determine cooperator acceptance were made with six cooperators and the responses were transcribed from magnetic tape. Some representative statements follow: Unfavorable Comments by Cooperators In response to the question: Are monthly journal listings easy to check against source documents? One cooperator felt that in the beginning it was difficult but he had devised a system to aid in checking. "I just write my code number in the corner of the check somewhere and just turn it down and take the next one and then come back through and (mark) your sheets." Concerning the same question another cooperator said: "Well, I couldn't have checked that one (monthly output). In fact, I could have made a new record quicker than I could have checked that one out." Use of the previous comment would have made the checking task easier. One of the problems that made the task more difficult was that most of the cooperators waited two to thcee months to send in entries and this made the volume of output large and echecking more difficult. The same cooperator admitted: "Well, my mistake was I let it pile up and I mailed too many at one time." A problem encountered by several cooperators was to find an enterprise on the journal listing that was not on the farm. This error prompted the following comment from one cooperator: "One of the biggest ones that was really standing out was where they (Elec-Com) had broilers. We don't have a chicken on the farm." This type of error was difficult to explain to a cooperator. It was found upon rechecking this cooperator's scan sheets, as well as others, that the error was one of the cooperator incorrectly entering the account number. Poor erasure or light marking also explained a number of these types of errors. One solution to this problem was to incorporate a routine into the journal listing program to point out any transactions in enterprises not on the cooperator's farm. One cooperator said of the farmer acceptance of Elec-Com coding: "ell, they can do it but it wasn't simple at first. I mean it was too easy to make mistakes." Questions 1 and 3 in Table 9 summarize the position of most cooperators. Eighty-seven per cent of the cooperators felt that the Elec-Com code was easy to use and 91.3 per cent felt that after practice the coding could be done without assistance. Favorable Comments by Cooperators In response to the question: Was Elec-Com beneficial to you in record keeping? The following reply was recorded. "It was a great benefit to us; we just started farming and the first year our records were a mess and with this system it helped us work our problem out quite a bit." This cooperator was enrolled the previous year in another state's central record keeping project. He felt that the Elec-Com's code was much easier to comprehend than his previous system's code structure. How had this individual felt concerning the utilization of optical scan sheets for an input media? "I believe anybody could use them. All it takes is to be careful when you are marking them. After you learn how, why, anybody, I believe, could mark them." When another cooperator was asked if the 1967 Elec-Com User's Manual was self-explanatory a favorable response was obtained. When another cooperator was asked if the use of scan sheets were time consuming in respect to your present record system the following reply was obtained: "No, I don't think it takes as much time to do that as it would to write it out on the ledger." Do farmers use records for management decision-making? One set of records showed a cooperator that the $417.00 per year spent for labor to sweep out the broiler houses was more than enough to justify purchase of sweeper attachment for his tractor. This type of decision was made possible by keeping detailed accounts, but most farmers were still against this type of breakdown. Design of the Elec-Com System A system that minimizes costs to the cooperator is necessary to help low and middle income farmers in management decisions. This system would also minimize the amount of cost borne by the processing organization John Doneth, extension economist at Michigan State University, has indicated some weaknesses in electronic mal-yn records. Two weaknesses cited were: 19 (1) Cash cost of the mail-in system to the cooperator is usually greater than his present accounting costs. Low and middle income farmers have, in general, indicated no motivation to keep records except for tax purposes. They have seen no need for management analysise (2) A central system requires skilled personnel to make the program successful. (3) The problem of this study was to overcome these weaknesses by minimizing two kinds of costs: (1) cost of converting cooperator records to an input that can be digested by electronic data processing equipment, and (2) cost of a pro- fessional and clerical staffs One readily available answer to costs of converting records to computer input is an automated input device. The second cost factor, reducing the use of professional and clerical staff, is closely associated with type of code system, the person doing the Six possible objectives form of a code: (I) Objectives seen by (2) (3) (4) (5) (6) program. Objectives Objectives Objectives Objectives Objectives seen seen seen seen by by by by coding, and where the coding takes place. have been listed by Vincent (3) as dictating the the user cooperating in a group accounting accountants serving agriculture. farm management specialists. research workers. programmers and systems personnel. seen by coding personnel. Objectives of the cooperators fall into two areas: (1) recording transactions needed for income tax accounting and (2) recording transactions needed for management decision-making. The entries required for management decision- making are more numerous than entries necessary for a tax system. Entries for management include unpaid family labor, opportunity cost of interest on investment, land charges, and many types of inter-farm transfer of resources. Automated Input and Code Structure Some of the automated input devices available today include: (1) Mark sense cards (2) Port-a-punch cards (3) Optical scanner (4) Optical reader (5) Magnetic input character reader (MICR) (6) Voice interpretation device Devices one and two were not considered since the using and mailing of the input forms (computer card) would have created proceEing problems at the processing center. A card reader is less tolerant than other devices in acceptance of folded and swelled input forms and port-a-punch cards create a problem as they are non-correctable. An optical reader and voice interpre- tation device were commercially unavailable in March, 1966. MICR has been utilized by the commercial banking industry for many years and its success has been established. An IBM 1232 optical scanner was selected as an input conversion device that was to be used in conjunction with a tape oriented IBM 7040-1401 data processing system. A code for use in an IBM 1232 optical scanner used in the study placed several restrictions on the code: (I) it had to be all numeric; (2) it had to be uniform for all entries; and (3) it had to conform to a scan sheet capable of containing two columns of 50 numbers each. Efficient use of this space indicated use of less than 26 numeric digits for an entire transaction entry or four transactions per sheet. 21 The code system originally developed for use in the Elec-Com system met all requirements set forth by Vincent except double-entry bookkeeping. This feature was added to Elec-Com at the end of the study. The code provided tax accounting and enterprise analysis for the farmer. Uniform code and com- parable enterprise analysis were provided for the management specialist. Unlimited breakdown of production inputs was provided for research workers. Identical input for all firms was provided for the programmer and analyst. In the final version of Elec-Com, a three-step decision process was provided the cooperator to determine a transaction code number. The code originally consisted of an account number that identified the following questions: (1) Was the transaction a variable or fixed cost? (2) In what enterprise should the transaction have been debited or credited? (3) Was the account a production, harvesting, or storage function? (4) Was the transaction an expense, income, or interfarm transfer? (5) What was the account name? (6) In what units was the amount reported? (7) Where on a tax return did the item belong? This information was internally coded by the computer using a specific account number. Figure 1 illustrates graphically the original Elec-Com code structure. As Figure 1 is read from left to right, branch one indicates fixed or variable cost. The second set of branches indicates enterprise number. The third set of branches indicates production, harvesting and storage input by enterprise . The fourth set of branches identifies an input as an expense, VARIABLE FOR ENTERPR IS E PRODUCTION FUNCTION COMPLETE TRAN SAC TION DESCRIPTION VARIABLE FOR FARM I FIXED FOR FARM HARVESTING FUNCTION MARKETING & STORAGE FUNCTION RECEIPT INTER FARM TRANSFER A(V, E,S,F,T) FIGURE 1. ELEC-COM CODE TREE EXPENSE 1 .. I.. -.. r.v..---- ............... .I... ... n.. 23 income or non-cash transfer to another enterprise. The final set of branches identifies the subscripted account name A(V, E, F, S, T) where V identifies branch one, costs; E identifies branch two, enterprise; F identifies branch three, stage of production; S identifies branch four, income or expense flow; and T converts the account number to a function of reporting farm income or expenses for tax purposes. The cooperator made two coding decisions and reported only step two. Step one was accomplished by selecting one of the pages in his code book labeled E 1 , E 2 ... En, Table 10. Step 1 is a sample page from the coopera- tor's code book from which he obtained the code numbers. This system had fewer decisions requiring a code than had been employed by other farm accounting systems because a code number is not required at every branch of the tree in Figure 1. The decision of an account number was recorded on a scan sheet, Figure 2, along with the cooperator's social security number (SSN) the date of the transaction, the number of units sold, bought, or used, and the cash value of the transaction. As an example of its use, the cooperator filled in his social security number, 123456789, the date December 5 (12/05), the account number, (0051), as looked up previously in a code book supplied to the farmer, and the number of cwt. of fertilizer bought for the corn enterprise. He had to convert one ton to 20 cwt. (0200) since the code book specified this to be the units used with this corn fertilizer transaction. He recorded the cash value $40.27 (0004027) of the fertilizer and the transaction record was com- pleted, Figure 2. As indicated in Figure 1 all other information was coded in the computer as a part of the account number. Every month the TABLE 1 0 PORTION OF AN ELEC-COM CODE BOOK LI STPG THE[- CORN ACCT TAX ACCOUNT NAMEK. 848 3 7 ACS PAYMENT RECD 55 1 G CU;STOM B REAK [120 54 1 '?U CUSTOM! CULTIVATE 6 1 1 20 CUSTOM DUSTING 56 1 2C CUS1Cfr FERTILIZING 59 1 20 CUSTOM HARVESTING 85 2 21 CUSTOM HAUL SUPPLY 84 2 21 CUSTOM0 HAUL TO MKsT 57 1 20 CUS TOM PLANT ING 6C 1 2 CUSTOM, SIDE DRESS 58 1 20 CUSTOM SfPAYIN 1- 8C 3 16 CUES 51 1 19 F-ERTILIZE.-k 930 0 #L FOP teF?AKING 71 0 C FL FOR CULTIVATING 99 0 0 FL FOR UISKING 97 0- 0 FL FUR DOUCT ING 94t 0 0 FL F~R FERTILIZING 100 0 0 F L FO.R. HARROWING 825 0 C) FL FOR =F IC ijE 826 0 0 FL F(UR HOEING 95 0 0 FL FOR PLANTI1 G 98 0 0 r1_ FJP SIDE .R SS 967 0 0 FL FOR SPRAYING 827 0 0 FL HAND HARVEST 828 C) 0 FL MLCH HARVEST 1116 0 0 HAING SUPPLIES 72 0 0 HA~ULING; TO MARKET 92 1 14 -L FOR HERI3ICIDE 91 1 .14 HL FOR HOELING 66 1 14 HL FOR FLANFTIN 8 7 1 14 F L [u S ID E DR E SS 67 1 14 F-L FOR SPRAYING 10 1 21 INSECTICIDE 787- 0 v LABOR TOTAL A 83G0iC LAND PLAN TED 76 3 12 LAND RLNTAL R EC' D 52 1 1 ) LIME 74 1 21 M ISCELLANEOUS SUP 1031 0 0 PLACL IN STORGAG;E 5 / i '1 18 SF fED r HNI[:- 0 0 7 4 Z - Z) 4 O Fig. 2.--Half of an Elec-Corn scan sheet __Z. 1=2 3-= =4-_5--- 8: 7- __-8: _9- f0:-1527-3-_-4-M -8:: - 9 0:1 27 3 - : 7 8 1Q -EW _2_ 3- 4_5_8: 7 _ 8: =9 -=-- 3-= A: 5-= 8:_ =_ __e_-9 8 : -9: :0: 1_ -2- _ _ =x =6 = 3-= -A:_= 8M5: =f: -f 8:-- .~ 2 3- : 5_-8:_- 8: 9= :0s - -- 3 45 : 7 8 m m m z n . 9 C E 0 _ _ - a _ vas _ .- s _ s . _ ram a _ s a _ a _ o cue a _ -w a sr e w r _ +- s .s? a _ -a _ _ s _ - e _? _ -? sv J+ H- 26 cooperator mailed his scan sheets to the Agricultural Economics and Rural Sociology Department, Auburn University, Auburn, Alabama. The cooperator-computer interaction is shown by means of five sets of flows in Figure 3. Flow I was accomplished by the cooperator when he selected an account number and recorded this decision on a scan sheet. The State office accomplished Flow II when the scan sheets were converted into punched cards and read by the computer. The computer completed the remaining three flows. In Flow III the computer found the proper master code information for the account on a magnetic tape. This master code information was transferred to the computer in Flow IV. The computer out- put in Flow V was the complete transaction described by Figure 1 plus the date, social security number, cash value and amount. The output in Flow V was contained on a magnetic tape from which accounting and management information could be summarized and printed in tabular form for mailing to the cooperator. The program was reviewed at the end of the first year and it was decided that double-entry bookkeeping and financial position was a desirable, if not necessary, feature of an electronic farm accounting system. Farmers seeking loans indicated a trial balance would be helpful in talking to creditors. O 3 4 z?I Q 2 3 4 CO S -1 1 2 3 4 _ 2 3 4 5 6 7 8 9 5 6 7 8- 9 5 6 7 S 9 5 6 7 8 9 I k~ivw m Fig. 3 ,--System flow of E12c-Comn code structure 28 A vendor number field was added to obtain a financial position by use of the account's trial balance and the account number was used to accom- plish all the remaining functions of a total accounting system. Vendor number field was used in the following manner. The field was left blank for a cash transaction and the bank (cash assets) was either credited or debited as determined by the account number. The appropriate enterprise was likewise either debited or credited. If a number appeared in the vendor number field$ this was a credit transaction and identified the vendor to which the individual has either sold orpurchased' an item as indicated by the account number. The appropriate enterprise or bank (cash assets) was then likewise debited or credited. The Elec-Com transaction block with vendor number added as a field is compared with the original block in Figure 4. One more decision requiring a code number was required for the vendor number, but the additional flexibility gained was necessary for full accounting. The account codes A(V, E, F, S, T) did not have a subscript added for vendor number, but were thought of as being on one of two basic branches, cash or credit. Processing Processing was divided into four areas: (1) error detection and correction, (2) monthly accounting, (3) quarterly summaries, and (4) yearly summary and tax output. Upon arrival of the coded material at Auburn a clerk opened the envelopes and checked for missing social security numbers, blank account number fields, and lightly 18 ~ 5 7 & 9- K_ 2 :45- 6 -7= * [9 9 f ~ -WITHOUT IC) 2 4 s ? 7 a aVENDOR 013 1 2 4 5 7 a aNUMBERS Q .. 1.3 45 L o3243- 4 7 8 9 QQI 2a3 4 . .63 d7 . 4 2= 45. &5;: ! 2 3 4 5 6 s Via VENOR3 4 6 79 1 2 3 .561 3 4 5 L? 0 'g r 30 marked transactions. If any blank account number fields were found, the sheet was returned to the cooperator for correction. The missing social security number or lightly marked data. were rectified by: the clerk. The data were then converted in flow II from scan sheets to card output by the IBM 1232 Optical Scanner and placed on tape. Error Detection and Correction In the process of going from card to tape, transactions were checked for three types of errors. The number and kinds of errors are summarized in Table 11. The errors were: (1) errors that were correctable by the system, 11.4 per cent, (2) errors that were uncorrectable by the system, 1.3 per cent of total entries, and (3) errors that were undetectable by the system as determined by an analysis at the end of the year, 1.4 per cent of processed transactions. About 1.6 per cent of type three errors were eventually corrected by the cooperator. Type 1 errors consisted of two kinds: (1) incorrect permutation of the digits in the cooperator's social security number and (2) incorrect or non-reporting of the date. An error routine was developed and programmed for the 7040 in which all valid social security numbers were stored in the computer and the reported social security number was checked against the valid list. Consider the following example: A cooperator reported the following:social security information: 421579603. The correct cooperator social security number was 421569713. The cooperator has three incorrect digits. If and when the computer would find a stored and valid social security number with at least five correct digits in the correct sequences, it generated the following printed line of output. SSN 421579603 CHANGED TO 421569713 Table 11. Errors in the Use of Elec-Com by Type and Number for 7300 Transactions, 1966 Per cent error Per cent error Per cent error Type I of all Type 2 of all Type 3 of all Field in correctable submitted errors submitted errors' processed error errors transactions transactions transactions SSN 426 5.8 37 .5 N/A Date 431 5.9 .. 2/ Account code --- 2/ 47 .6 94 1.2 Amount in or out -.. / 7 .1 8 .2 Cash value -- 2/ 8 .1 1 Total 857 99 104 Number of transactions 3 / 831 11.4 94 1.3 104 1.4 r-1 /Seventeen of the transaction errors were returned by mated by system personnel by means of a questionnaire. 2 !Unchecked by the computer. cooperators and the remaining were esti- 2/More than one error of type one or type two can occur per transaction. 32 The computer tied this correct number to the transaction as shown above. If a social security number had not been found the following message would have been generated: 421579603 EEEEEEEEE The transaction would have been deleted for processing at this time. In the case of an incorrect date, the computer corrected the error. No correction was made if the date was missing. The most common error when reporting dates can be illustrated: by the following example. The month of August may have been reported as 80 instead of 08. In such a situation the computer simply reversed the two columns and proceeded. All fields were checked for type two errors that were of three types: (1) unidentifi able social security number, (2) an alphabetic character in a field (caused by making two marks on the same line), and (3) black account number field. Type 3 errors may only be discovered by the cooperator upon checking monthly transaction sheets. Monthly Accounting After incoming transactions had been checked for errors and transferred to tape, they were ready for the addition of master data. Corresponding to Flows III and IV, the account numbers were sequentially stored on tape and the following information added to the scanned transaction: (1) Enterprise number (2) Tax code (3) Account descriptive name (4) Units of amount reported by cooperator (5) Type of transaction: income, expense, non-cash 33 The journal listing in Figure 5 was produced after obtaining the final transaction tape from Flow V. The primary purpose of this output listing was for the cooperator to check and determine if he had any type three errors or errors of omission. If a type three error was discovered, the cooperator made the correction on a carbon copy and returned it to the central processing office for transaction updating. One revision made on the system was to introduce an aid in checking. A list of enterprises the farmer had on his farm was checked during journal listing against the enterprise coded to each transaction. If the enterprise was not found in the valid enterprise list, an error message was generated to warn the cooperator of a possible error. This occurrence was usually caused by incorrectly recording an account number. The warning must have been cleared by the cooperator before the transaction was fully processed. Any transactions which were returned by the cooperator were processed by a program to correct the Y-T-D transaction type by inserting, deleting, or replacing an incorrect transaction. The list of cooperators who had transactions updated was stored "