A CfiMFUTEREZEB PREDYCTEGN MODEL FDR EGG PRQDUE’FEON MANAGEMEKT BECESEORS A Dissertation Yo:- i’ins Degree 95 pi. D. MICHIGAN STAYE UNIVERSE” James. Leslie Daie {974 LIBRARY lmlrwmulumulmn lllllllllfllallglflfil I) 293 10788 3096 Michigan State University This is to certify that the thesis entitled A Computerized Prediction Model For Egg Production Management Decisions presented by James Leslie Dale has been accepted towards fulfillment of the requirements for Ph.D. Poultry Science degree in Major profe o flat/6% Date August 5, 1974 0-7639 m “a .n s“ H 0-:- ~» .2 =2 mm ”a l. mm nu .m mw ~_ ”u n mama a r e .. e 3 z 3 s a 2 z a r. an 3 a n I: nummmm mmmmmwmmc so i .1 ~ , L- p— 5“ p. ‘ r I'D u‘ “I «I n an .n on m ’1 fl a II, 0" .\ J‘I .0-3 .11 m a: .. Oi O”! m m a: a: an m U, m =5 a: Cl 3‘) 0. =5 6') 5) ‘3) C‘) C..- \D C) :4 C) CD CD CD CL.) 6-.) (n LJ CL'J cu (a) mamammuaaouaamoawau. mammomumammwwwmwwwmmmnm f. (s p... o- l- ‘ 1‘ pm Is 0-— h [a 7—. h 7‘ --------pfihhhpfifisrpfippfihphfihfis new”mmmoaauommummagwammwmmummmmmmammwumamuoammc.momggf \l) ‘1') W m m ch ‘23 u? so If! if. I!) u- t!" u“! U’) .— nnmmnnmmnmnnmmnmmmmmm mmmwmmnmmmmmmmmn cec..v~.v.w.~e.cc~eecvvwovvqavvvcvvvvvaqqgceacv evquv. M m ('2 (“'4 m m m mammnnnnnnmmnnnnmnnn an”mmmmnmmnmmmmmmnmmmnpnmm --N----~N--N-NNNNNNNNNNNNNNNNNNNN--N--N~Nn ____________—_—__________F__—_______________1_._____ .— 3 a. n~ .~ - a“ mp .— H_ - ._ oh on .n F. an m. qu a. - _u an aw an “n m” an .w nu mm in an n¢ ma 5. m. n. *4 n. «a a. mm mm mm on an ~a an an .w an «N n~ - n~ ecaeaaceaaaaaa.aeo====amenaoaaaaaccapacmacaec acaamaaa ABSTRACT A COMPUTERIZED PREDICTION MODEL FOR EGG PRODUCTION MANAGEMENT DECISIONS BY James Leslie Dale The poultry industry is continuing through a period where producer numbers are declining and number of birds per farm and capital investment per farm are increasing rapidly. Large farm egg production complex and egg production con- tractor managers are wanting more sophisticated and accurate management tools, particularly tools for planning purposes. The purpose of this research was to 1.) test the hypothesis that egg production, egg size distribution and mortality could be predicted and 2.) if the hypothesis was true, to incorporate the prediction equations into a dynamic simulation model to aid farm managers in their flock replacement decisions. Performance records of 210 commercial egg laying flocks representing about 2.5 million birds were obtained. Infor- mation such as flock size and numbers of birds per cage pertaining to each flock was also obtained. Recycled flocks and flocks with incomplete data input were excluded from the statistical analyses. The data were analyzed by the least squares method. The dependent variables for all analyses were mortality, bméobk James Leslie Dale total eggs produced per hen housed, and the total number of large size, medium size, small size and undergrads eggs produced per hen housed. Initially, the effects of the total number of 28-day periods of production, starting age of production and flock size were measured. The results indicated that, generally, these three independent variables significantly affected the dependent variables. The data were then adjusted (standar- dized) for the three independent variables. A second least squares analysis was undertaken to measure the effect of five other independent variables and seven interactions on the adjusted data. The total number of eggs produced per hen housed and the total number of small size eggs produced per hen housed were the only depen— dent variables significantly affected by the independent variables and interactions of this study. The results of the statistical analyses were used as the bases for the development of multiple variable, multiple regression prediction equations for each of the dependent variables. The equations were developed to give 28-day, period by period predictions. The equations that were developed did not perform satisfactorily and were abandoned. The Gavora-Parker-McMillan model for egg production prediction was introduced at that point. Following their stated procedure, decreasing lay rate coefficients were James Leslie Dale obtained. The model was slightly modified and tested. The results were very favorable. The egg distribution coefficients developed for pre- dictive purposes were the average distribution rates per period by strain. The mortality rate for predictive pur- poses was determined to be a constant rate: an average of the first three periods of production. The simulation model developed, using the prediction equations and the manager's expected cost-revenue relation- ships, performed very satisfactorily. Like any other forecasting and planning tool, the results obtained, compared to actual results, were only as good as the input data. A COMPUTERIZED PREDICTION MODEL FOR EGG PRODUCTION MANAGEMENT DECISIONS BY James Leslie Dale A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Poultry science 1974 ACKNOWLEDGMENTS The author wishes to acknowledge the assistance of and express appreciation to: Dr. C. C. Sheppard for his sincere interest, guidance and counseling during the course of graduate study. Dr. H. H. Vincent, Dr. H. C. Zindsl, Dr. T. H. Coleman and Dr. T. R. Pierson for their interest and assistance as members of the guidance committee. The Poultry Science Department for research financial assistance. Dr. J. L. Gill and Dr. R. R. Neitzel for their statistical analysis expertise. Ms. Jacqueline Grossman and Mr. Bruce Johnson for their computer programing expertise. My wife, Judy, for her support and encouragement during the course of graduate study. 11 TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . v LIST OF APPENDIX I TABLES . . . . . . . . . . . . . . . vi INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . 1 LITERATURE REVIEW 0 O O O O O O 0 O O O O O O O O O 0 O 4 RESEARCH METHODOLOGY . . . . . . . . . . . . . . . . . . 9 Preface . . . . . . . . . 9 Statistical Analysis . . . . . . . 12 Procedures . . . . . . . . 12 Results and Discussion . . . . 16 Mortality . . . . . . . . 24 Egg production . . . . . 24 Large eggs . . . . . . . 26 Medium eggs . . . . . . . 26 Small eggs . . . . . . . 27 Undergrade eggs . . . . 28 Regression prediction e on 28 q m PM* a Gavora-Parker-McMillan d The Simulation Model . . . eoeeeeeeeeeofieeeeeesooo .o.........a¢+.o.......o .o........o.m.......... oeooeeeeeeeeooeoosoeoee eeeeeeoeoeoeoeeeeoeoeoe oesoesseoosoooooseeoooe o...................... OOOOOOOOOOOOOOOOOOOOOOO {3 Procedures . . . . . . . . . . 39 UPDATE 0 O O O I O O 0 0 4O TCYCLE . . . . . . . . . 42 PREDICT O O O O O O O O O 42 BIRDINV . . . . . . . . . 43 EGPROD O O O O O O O O O 43 CASH O O O O O O I O O O 45 EXPENSE O O O O O O 0 O O 45 NETWH O O O O O O O O O 47 Results and Discussion . . . . 48 SUWY O O O O O O O O O O O O O O O O O O O O O O O O 52 CONCLUSIONS 0 O O O O O O O O O O O O O O O O O O O O O 58 111 BIBLIOGRAPHY APPENDIX I . iv 3. 4. LIST OF TABLES Base statistics on the collected flock data . . The effects of the total number of 28-day periods of production, starting age and flock size on mortality, total eggs pro- duced and egg size distribution . . . . . . . Selection of independent variable or inter- action effect from the least squares analyses 0 O O O O O O O O O O C O O O O O O 0 Potential maximum egg production and rate of decrease of lay by strain . . . . . . . . . . Potential maximum egg production and rate of lay decrease by quarter of start for strain 1 The average egg size distribution by strain . . The average egg size distribution by quarter of start for strain 1 . . . . . . . . . . . . Sample output of the computer model . . . . . . A comparison of predicted and actual flock performarlce O I O O O O O O O O O O O O O O O Page 17 19 22 31 32 36 37 44 51 LIST OF APPENDIX I TABLES Table Page 1. Actual raw data - example . . . . . . . . . . . 63 2. Rate of lay decrease coefficients for each period for strains 2, 3, 4 and 5 . . . . . . . 64 3. Rate of lay decrease coefficients for each period by quarter of start for strain 1 . . . 65 4. Egg size distribution coefficients for each period for strain 2 . . . . . . . . . . . . . 66 5. Egg size distribution coefficients for each period for strain 3 . . . . . . . . . . . . . 67 6. Egg size distribution coefficients for each period for strain 4 . . . . . . . . . . . . . 68 7. Egg size distribution coefficients for each period for strain 5 . . . . . . . . . . . . . 69 8. Egg size distribution coefficients for each period for strain 1 for quarter of start 1 . . 7O 9. Egg size distribution coefficients for each period for strain 1 for quarter of start 2 . . 71 10. Egg size distribution coefficients for each period for strain 1 for quarter of start 3 . . 72 11. Egg size distribution coefficients for each period for strain 1 for quarter of start 4 . . 73 12. Two examples of the egg prediction model . . . . 74 13. An example of the measurement of an indepen- dent variable's effect on a dependent variable from the least squares analysis . . . 75 14. The data adjustment formula . . . . . . . . . . 76 vi INTRODUCTION The egg industry in the United States has been and is expected to continue through a reduction in producer num- bers. The same number of birds are being controlled by fewer and fewer individuals or firms. A recent example of this is the arrival of the egg production complex; 300,000 or more birds in side-by-side houses of 50,000 or 60,000 birds each. The larger production units have lower building and equipment costs per bird than their smaller predecessors. However, the overall total investment is greater, possibly reaching into the millions-of-dollars category. As total investment increases, the need for accurate management information that is readily available becomes more important. This, of course, includes all aspects of the operation. Paramount, is the need for information on how the flock will perform. Flock performance then serves as the base from which overall flock financial performance is derived. Generalized cash flow and production simulation models exist that can be utilized to make flock projections using a standard production curve given current input-output cost 2 relationships. Accounting programs also exist that give an indication of how the current flock is performing compared to past or standard flocks. However, in the literature, there is no program currently available that gives the pro- ducer an accurate, comprehensive projection of his current flock's overall performance. It was the hypotheses of this research that 1.) egg production, egg size distribution and mortality could be predicted on a 28-day period by period basis reasonably accurately using multiple regression equations, and 2.) that a dynamic simulation model could be developed that would incorporate the prediction equations along with a flock managers expectations to generate data useful in making flock management decisions. Therefore, the objectives of this research were: 1. to determine the factors that affect mortality, egg production, and egg size distribution based upon recent commercial flock data (flocks started and completed in the 1970-1973 period). 2. to develop prediction equations that, on the basis of 12 weeks of actual data, will project reasonably accurately, the future patterns of egg production, mortality and egg size distri- bution over the remaining periods of production of the flock. 3. 3 to develop a computer simulator to incorporate the prediction equations as a function and simu- late all cost-revenue relationships and flock financial performance over the remaining periods of production to help poultry producers in their flock replacement decision actions. LITERATURE REVIEW Bell (1972) reported that "hen day production was significantly affected by...cage density..." He also reported that "the four-bird cages had a significantly higher rate of mortality when compared to the three-bird cages." Bell's experiment involved 1,120 SCWL birds housed half in three birds per 12" x 18" cage and half in four birds per cage (same dimensions). Adams and Jackson (1970) measured the effect of cage size and bird density on the performance of six commercial strains of layers. They found that the differences between the strains' hen housed egg production were significant. The birds housed at the lower densities performed better than the birds housed at the higher densities (less square inches per bird). In addition, birds housed at two and three birds per cage performed better than birds housed at more birds per cage. The effects of cage density, cage size and strain on mortality were mixed with no definite pattern evolving. Marks, Tindell and Lowe (1970) reported that signifi- cant differences in egg production existed between the six commercial strains under study. They reported that the 5 only significant effect of cage density was the 560 day body weight. Hill and Binns (1971) measured the effects on the performance of varying densities and numbers of birds per cage. They reported that egg production, feed efficiency, mortality, increased body size and net egg returns tended to be linearly related to size of cage; the greater the space per bird, the better the performance. Plumart, Carlson and Holmquist (1972) housed birds at 552, 414, 331 and 276 square centimeters per bird. This represented 3. 4. 5 and 6 birds per cage, respectively. They reported that the 414 cm2 per bird density (4 birds per cage) was "...detrimental to egg production compared to the other groups..." Feldkamp and Adams (1973) measured the effects of rearing relationships, cage size, and bird density on the performance of two commercial strains of egg-type chickens. They reported that as the square centimeters per bird decreased, egg production significantly decreased. Mor- tality, however, was not significantly affected. Latimer's and Bezpa's (1970) projections and cash flow for a 30,000 bird commercial table egg operation utilized a ”standard" production curve and a "standard" egg size distribution for calculation purposes. These respective values were derived from the 1966-67 New Jersey Random Sample results. 6 Ruggles (1971) published a handbook to help North- eastern poultry producers prepare cash flow projections. The egg production pattern used a "standard" curve. The producer would develop his best estimate of egg pro- duction based on his flock's historical annual dozens of eggs produced. Egg size distribution was calculated by a similar procedure. Bell (1971) published a handbook to help Southern California poultry producers prepare cash flow projections. Bell utilized a "typical production curve for White Leghorn chickens" to project egg production. Egg size distribution was calculated utilizing a table of typical egg weights (as a percent) by age of bird. Muir (1972) developed a computerized cash flow for market egg farms. The program projected production on the basis of five input values supplied by the producer. These values were flock age at the start of production, flock age at peak production, flock age at the date of sale, flock peak production level and flock production at the date of sale. The model used the five points to establish two linear lines (one increasing and one decreasing) which it moved along, simulating the egg production level per pro- duction period. Projections simulated after the flock's peak production required estimates of egg production at the date of sale and flock age at the date of sale. The model established a linear line between the known peak egg 7 production and the anticipated ending egg production. It then simulated the egg production per period and calculated costs and revenues per production period for the remaining periods of the flock. Gavora, Parker and McMillan (1971) applied a mathema- tical model of egg production originally developed by McMillan, 33 $1., for predicting egg production in the "Drosophila Melanogaster" to chickens and turkeys. The model was described as follows: N(t) - m (1-e “€(t'to)) e'“t where N(t) a the number of eggs laid in a 28 day period M - the potential maximum egg production in a 28 day period tO - the initial age of egg laying E - the rate of increase in egg laying a - the rate of decrease in egg laying They reported that the a value was -0.0340 for egg-type chickens, groups of hens, based on 17 records. When the formula was applied to one example flock (12,000 birds) the model explained 96.5 percent of the total variation in egg production for that flock. Ruggles (1968), Vincent and Sheppard (1970) and Dale, Vincent and Sheppard (1974) developed poultry farm manage- ment game-simulators. Each game functioned somewhat differently regarding the answers given. However, in the 8 area of egg production and egg size distribution, all three utilized "standard" production curves and egg size distribution tables. The Dale, 23 gl., model allowed a limited amount of prediction by asking the player to specify an estimated yearly average production rate. The model responded by raising or lowering the "standard" egg production curve to obtain the desired results. Bird mortality in all of the above mentioned cash flow handbooks and programs and simulator games, was calculated as a constant rate throughout the production cycle. Latimer and Bezpa (1970), Ruggles (1968), Ruggles (1971) and Bell (1971) calculated feed requirements for the flock on the basis of a "standard" amount of feed consumed per bird per period of production. Dale, 35 gl., (1974) and Vincent and Sheppard (1970) utilized a formula incorporating a maintenance level of feed intake and a variable amount of feed intake directly related to egg pro- duction. This sum amount was considered to be the amount of feed consumed per bird per period. RESEARCH METHODOLOGY Preface The flock data that was considered necessary to develop the prediction equations was available from two sources. One was from controlled experimental units. The second was from commercial flocks. The decision to select either of the sources was dif- ficult. If controlled experimental flocks were sampled, most of the intangible factor of flock-care management effect may have been removed. However, it was questionable that enough flocks and birds within flocks could be obtained to develop credible equations for direct application to commercial flocks. It was believed that if a large cross-sectional sample of commercial flocks could be obtained, the flock-care management factor would be reduced to a minimum. In addi- tion, prediction equations developed from actual commercial data would be more readily accepted by industry for manage- ment planning use. Thus it was determined that as much recent commercial flock data as possible would be obtained. The periods of 1970-1973 were selected as the sample years, with the 10 emphasis on 1972-1973 flocks. The emphasis was on those two years because of the introduction of Marek's vaccine and continuing breeder development of improved commercial strains. Many organizations were contacted as possible data sources. The following organizations were most cooperative and supplied this research with data: DeKalb AgResearch, Inc., DeKalb, IL. Michigan Farm Bureau Services, Inc., Lansing, MI. Landmark Foods, Inc., Columbus, OH. Babcock Poultry Farms, Inc., Ithaca, NY. Hyline International, Inc., Des Moines, IA. The first three organizations listed above maintain up-to-date egg production, egg size distribution and mor- tality on many commercial flocks as part of their organizations' business or as a service to their customers. The other two organizations had limited amounts of data available. The actual collection of data from the first three sources required the researcher to visit their respective corporate headquarters and physically transcribe data from 28-day period reports to summary forms. The following information was sought from each set of records: 1. starting age of egg production, 2. number of birds per cage, 11 3. starting month of egg production, 4. windowed or windowless house, 5. commercial strain of layers (e.g., DeKalb), 6. location (state), 7. total number of 28-day periods of production, 8. size of flock, and 9. for each 28-day period of production: a. mortality (total number of birds), b. total eggs produced, and c. egg size distribution. Information items 1 through 8 were sought because of their believed possible effect on mortality, total egg production and egg size distribution. The information gathered was taken from each set of records as presented by the records. It was not possible to obtain square inches per bird from the record sets surveyed. Number of birds per cage, however, was obtained and therefore used. The general assumption was that as the number of birds per cage increased, the area per bird declined. The total number of record sets surveyed was 210, representing more than 2.5 million birds. Table 1 in Appendix I contains the results of one surveyed flock's record set. 12 Statistical Analysis Procedures The start of egg production ranged from 18 weeks to 26 weeks and was coded from 1 to 9 (recycled flocks were coded 0). The recycled flocks were not included in the final analyses. Numbers of birds per cage ranged from two birds to ten birds. The five or more birds per cage category was established because of the few number of records in the individual categories beyond five birds per cage. All 12 months were used as starting months. The months were grouped into quarters (quarter I being January, February and March) for analytical purposes as a method to reduce the total number of variables involved. Windowed houses were coded as 1 and windowless houses were coded as 2 for analytical purposes. The data included both types, however, the windowless was predominate. Data were collected on seven strains. They were DeKalb, Byline, Babcock, Heisdof and Nelson (H&N), Shaver, Kimber and Arbor Acres. All strains were egg-type chickens. No attempt was made to differentiate specific identifications within the commercial name. Nearly 70 percent of the total sample was accounted for by one strain. It was difficult to obtain information for some strains. 13 Blocks surveyed were located in seventeen states. The states were divided into regions for statistical analyses purposes. The majority of the flocks surveyed were in region 1. The following was the region break down by state: Region 1 - East North Central Michigan Illinois Indiana Wisconsin Ohio Region 2 - West North Central Minnesota Iowa South Dakota Missouri Wyoming North Dakota Colorado Region 3 - Northeast New Jersey Pennsylvania Region 4 - Southeast North Carolina Maryland Virginia Initially, the data were subjected to least squares multiple regression, measuring the effects of flock size, the total number of 28-day production periods and the starting age on the following dependent variables: mortality, the total number of eggs per bird housed and the total number of large, medium, small and undergrads eggs, per bird housed. 14 Based on the significance levels, the data were adjusted (standardized) to reduce the number of variables involved in later analyses. The three variables had a significant or near significant affect on most of the dependent variables listed. Thus, as a matter of con- venience, all the data were adjusted before further analyses. Table 14, Appendix I, presents the adjustment formula utilized. The adjusted data were analyzed utilizing the general least squares multiple regression method. The dependent variables were again those listed above. The independent variables were numbers of birds per cage (four groups--2, 3, 4. 5 or more), quarter of start (four groups), windowed or windowless house, strain, region (four groups--West North Central, East North Central, Northeast, and Southeast) and the following selected interactions: number of birds per cage by quarter of start, number of birds per cage by strain, number of birds per cage by region, quarter of start by windowed-windowless house, quarter of start by strain, quarter of start by region, windowed-windowless house by region, and strain by region. 15 These interactions that were believed to have no in- fluence were excluded to facilitate a more manageable matrix for the analysis. The variables and interactions having a near signifi- cant or a significant effect on each dependent variable were noted. A third general least squares analysis was under- taken. The dependsnt variables were the above listed, on a 28-day period by period basis beginning in the fourth period of production and continuing through period fifteen. Beyond the fifteenth period, the amount of data was too small to allow a meaningful analysis. The independent variables included were those that had a near significant or significant effect on the dependent variable under study. A Other independent variables included the summed data for the dependent variable category through the i-1 period. An example would be predicting egg production in the fourth period. The summed data would then be the total number of eggs produced through the third period. This process was undertaken to establish a periodized prediction equation for mortality, egg production, and egg size distribution. The formula outlined by Gavora, 23 §;.. for estimating the egg production rate of decrease (a) and potential maximum monthly egg production was applied to the data collected. The formula was a log-linear regression equation as follows: 16 In N(t) - ln M-at where N(t) - the number of eggs produced per hen housed in time period t (28-day period) M a the potential maximum monthly (28-day period) egg production per hen housed a a rate of decrease t a time period. The regression was performed on the data in periods where t 3 tomax, tomax being the time period (number) of actual maximum (peak) production. The analysis was performed on each strain. Analyses on strain 1, which composed 70% of the data, were done by quarter of start. The other strains did not have sufficient amounts of data to allow analyses by quarter of start. Results gnd Discussion When the data were initially collected, recycled flocks and flocks with no answers to some of the questions (birds per cage, for example) were included. During the analyses recycled flocks and those with missing data were not included. This action reduced the sample size to 163 flocks with a total of 2.2 million birds. Table 1 presents the minimum, maximum and means of the data asked for in the initial survey. The average lay rate of 229.01 eggs, per bird hen housed, in 397 days (14.18 17 Table 1. Base statistics on the collected flock data Item Minimum Maximum Mean Total number of 28-day periods of production 8 18 14.18 Starting age (weeks) 18 26 21.18 Flock size (000) 1 41 12.85 Mortality (total percent) 6.8 41.8 17.90 Total eggs per bird housed 119.02 365.99 229.01 Large eggs per bird housed 55.74 234.44 145.07 Medium eggs per bird housed 7.7 111.16 49.64 Small eggs per bird housed .22 41.22 10.30 Undergrade eggs per bird housed 3.11 42.69 21.65 18 28-day periods) expressed as a percentage, was 57.68 percent. The egg size distribution was as follows: large eggs were 63.35 percent, medium eggs were 21.68 per- cent, small eggs were 4.50 percent, undergrade eggs were 9.45 percent and lost eggs (unaccounted, loss eggs, etc.) were 1.02 percent of those produced. The average mor— tality rate for the flock data was 1.26 percent per period. The information in Table 1 was not analyzed by strain. It was noted that a very wide range existed for some of the variables. An example was the 246.97 difference in total eggs produced per hen housed. Similar ranges were evident for all the items except starting age. The purpose of the following statistical analysis was to determine how the data should be adjusted for further analyses. The effects of the total number of 28-day periods of production, starting age and flock size (inde- pendent variables) on mortality, total egg production and egg size distribution are presented in Table 2. The regression coefficient and significance level of each independent variable are listed. The total number of periods of production had a highly significant (p<.01) positive effect on all the dependent variables except on the total number of small size eggs produced per hen housed. One would expect this type of response due to the increased production cycle length and the natural response_of more eggs and increased mortality. Since the vast majority of 19 mamcp condom mom 9 pemaoawuooo scammosmomm mmapcwsw> pemceoaon mm0.0 40?.0 P0.0 ¢F0.0 F00.0 v P00.0v ooqsoaeeemam mam.0 mFF.0a am¢.0u mem.0 u 0mm.r u m00.0 m.mooo .msm seam Mooam 000.0 mmr.0 F00.0v mm~.0 400.0 mFF.0 mossoauaamsm www.cu sa¢.0 me..m mm¢.0 me.m 400.0 s.umoo .mom ems mnwpnmum 80.0v >00.0 F00.0v e00.0v F00.0v 000.0 cosmonessmhm m0.a N0F.0 mem.~ Nnm.0a nam.ea s00.0 a.uooo .msm muowhmm pmwmm pmmwm pmmmm pmmmm nmmwm hpaacpuoz canmunm> ocwnmnoosb Hamam Beans: ownmq Hmpoa pnmusommunH soapsnwspmfic seam wmo new cooscosn mmmm Hmpop .hpfiampaos so seam mooau 0mm 0mm manHMpm .nofipozwoum no snowmen hmnnmm mo Hones: Hmpop esp Ho mpoommo one .m manna 20 small size eggs would be produced in the first few periods of production, one would also expect the insignificant effect of a longer production cycle. Recent literature did not refer to the effects of production cycle lengths on the dependent variables studied. Starting age had a mixed effect on the dependent variables. The delaying of egg production (delaying sex- ual maturity) seemed to slightly increase the amount of flock mortality. However, the effect was not significant. Delaying sexual maturity had a highly significant (p<.01), positive effect on the total number of eggs per hen housed and on the total number of medium size eggs produced per hen housed. However, the starting age did not significantly affect the total number of large size or small size eggs produced per hen housed. The negative effect on the number of undergrads eggs produced per hen housed, as the starting age increased, was nearly significant (p a .096). Recent ‘literature did not refer to starting age and its effects on the dependent variables. Increasing flock size had a negative effect on bird performance. Mortality and the total number of undergrads eggs produced per hen housed were significantly (p<.05) increased. The total number of eggs per hen housed, the total number of large size eggs per hen housed and the total number of medium size eggs produced per hen housed were significantly (p<.05) decreased. Increasing flock 21 size also had a negative effect on the total number of small size eggs produced per hen housed. However, the effect was not significant (p = .104). Flock size was measured in 1,000 bird increments. Recent literature did not refer to the effects of flock size on the dependent variables analyzed. The purpose of the above analysis was to determine how the data should be adjusted for further analyses. The criterion was that if an independent variable had a signi- ficant or near significant effect on any dependent variable, the data would be adjusted for that independent variable. The independent variables met that criterion in most of the individual analyses. Thus, for ease of handling, all the data were adjusted for the total number of 28-day periods of production, starting age and flock size. As previously mentioned, general least squares was used to analyze the adjusted data. Table 3 presents the selection of independent variable or interaction effect from the least squares analyses. This table presents the information base for the discus- sions below from mortality through undergrads eggs. The information in Table 3 is presented as an i/j value with a superscript. The i . the number of separate variables with a significance value of .2 or less. The j . the total number of separate variables in the group. The superscript indicates a significant, or a near 22 M\m M\m n\o n\o m\o M\o mmoazoeeaz unsoeeaz mp passe no newneso m\s m\m m\a m\a m\m m\a ammo you means do Hogans an pumps no Hosanna m\0 on\m M\0 n\a n\0 n\0 mmsflsoasaaumsoesfiz an soammm m\m m\m m\r m\m m\r m\m peeps so assumes an soamom 0\F om\0 om\m om\e om\m om\m means so managesWMo:MMMsm 0\0 0\0 0\0 s\0 ns\0 0\0 cashew w\o nr\w oF\r F\o "\o «\o mmmazouna3lm3ocnfiz n\a M\0 m\0 n\0 pm\m M\0 ammo use means no smnasz m\o m\o n\o M\o pn\n n\o peeve no nopnmso M\0 nm\n om\m M\F om\m n\0 sesame mmmm mmmm mmmm mmmm mmmm hwaampuoz soapowsoch no mumnmumczo HHcEm suave: swung Hmpoe memHHm> pnmcnmmmunH mcapmaum> vnocaomon mmomhamnm monmsvm pmmma as» Scam pomwwm nofipomnmpna no mammasm> pnmuammmona Mo soapooaem .m capes 23 Anowmmsomac nmnpnsw you H Maoeeaa< ..mr canes mom .asonm on» ad mosacb monmoamHamHm Hesefisacnfi on» no hpaewpmama one one mafia» n\« can M0 coupons“ m mm: nofipcnfisnopmc ceasefimaewam “maozv moeosamefi as no voomwo pqmoauaewam name so poomwc pamoawanmfimn macaw on» ma moapdwhm> opmnmgmm Ho amass: one a n "N.v enamb moemoawasmam a spa: macaw on» a“ mmHanum> opmnmmom mo mopeds on» u L n\fim a.\m 02\0 0a\0 0a\m noa\m 0a\r assess mp mwmo you woman no nonsdz .P\N Fa\0 PF\0 ra\0 paa\m FP\N assess an peeps no saunaso mmwm mmmm mmwm mmmm mmmm mafiaepnoz coapomuman no oomumnmuep Hamam suavez mmnmq Hmpoe oHanHm> psocsonman mmapmanm> pzmcnomcn A.v.v:oov m manme 24 significant influence effect. The determination of significance was a function of the i/j value and the rela- tivity of the separate significance values within each group. Table 13 in Appendix I presents an example of the determination of the superscript. Mortality. The R2 value (variation or causative value) obtained was .48, thus indicating that the independent variables (region, etc.) and the interactions included, accounted for less than one half of the variation in mor- tality. The only independent variable or interaction group to have a near significant effect on mortality was the interaction of the number of birds per cage by region. However, the base independent variables (number of birds per cage, region) from which the interaction arose were not statistically significant or near significant. These findings concurred with those reported by Adams and Jackson (1970), Marks, 23 3;. (1970) and Feldkamp and Adams (1973). These findings did not concur with Bell (1972) and Hill and Binns (1971), each of whom reported significant effects on mortality due to decreasing area per bird. Egg production. The variables and interactions of this study accounted for about 77 percent of the variation (R2) in total eggs produced per hen housed. Region tended to influence total egg production per hen housed. The regression coefficient indicated that the 25 North Central states had a negative effect on total produc- tion, while the Northeast and Southeast regions had a positive effect. The quarter of start of production had a highly signi- ficant effect on total egg production per hen housed. The regression coefficients indicated that quarter three (July, August, September) was the most favorable quarter to start while quarter two (April, May, June) was the most unfavor- able. The number of birds per cage significantly affected total egg production per hen housed. This concurred with Bell's (1972) findings. The regression coefficient ranked the effect of the number of birds per cage as follows (on the basis of preferential performance): three, two, four and five or more. These results concurred with Adams and Jackson (1970), Hill and Binns (1971) and Feldkamp and Adams (1973). Each of these teams reported that as the area per bird decreased, egg production significantly decreased. The reversal of the two and three birds per cage in ranking in this study tended to counter the above findings. However, the overall general trend of reduced space per bird: reduced egg production was intact. Strain highly significantly affected total eggs produced per hen housed. This result concurred with the published reports of Adams and Jackson (1970) and Marks, gt_§l. (1970). 26 Ranking the strains according to regression coeffi- cients (most positive coefficient ranked first) established the following order: 5, 3, 1, 4, 6, 2, and 7. The region by number of birds per cage interaction tended to affect total eggs produced per hen housed. However, the region by quarter of start interaction and the quarter of start by number of birds per cage interaction had no effect. These results were somewhat confusing given the significant effects of their component independent varia- bles. 0n the other hand, the quarter of start by strain and number of birds per cage by strain interactions signifi- cantly affected total eggs produced per hen housed. Large eggs. The R2 value of this study was .5 indi- cating that the independent variables and interactions accounted for about half of the variation in the total number of large size eggs produced per hen housed. No independent variable or interaction significantly influenced the total number of large size eggs produced per hen housed. The region by number of birds per cage interaction had a very slight influence. Medium eggs. The independent variables and inter- actions of this study accounted for about 46 percent of the variation in the total number of medium size eggs produced per hen housed. 27 Region tended to affect the total number of medium size eggs produced per hen housed. The North Central and Southeast states had a negative impact while the Northeast states had a positive impact. The absence of windows in the house tended to have a negative effect on the total number of medium size eggs produced per hen housed. The presence of windows tended to have a positive effect. Similar to the total number of large size eggs pro- duced per hen housed, the region by number of birds per cage interaction had a slight influence on the total number of medium size eggs produced per hen housed. Again, the com- ponent independent variables had no effect. Small eggs. The independent variables and interactions of this study accounted for about 52 percent of the varia- tion in the total number of small size eggs produced per hen housed. Region had a significant effect on this dependent variable. As with the medium size, the North Central and Southeast states had a negative impact while the North- east states had a positive impact on the total number of small size eggs produced. The presence or absence of windows had a significant effect on this dependent variable. The presence of windows had a positive effect. The absence of windows had a negative effect. 28 The region by number of birds per cage interaction tended to affect the total number of small size eggs produced per hen housed. A similar effect was indicated for the region by windows-windowless house interaction. Undergrade eggs. The R2 value for this dependent variable was .58. However, none of the independent varia- bles or interactions included in this study indicated any influence on the total number of undergrads eggs produced per hen housed. None of the literature searched referred to the inde- pendent variables that might influence the egg size distribution studied here. Some of the results of this study were rather confusing. Particularly the question of why a specific interaction exhibited tendencies to in- fluence egg size when its component independent variables clearly had no influence. Other results tended to follow the line of commonly accepted knowledge of climatic and light effects on egg size distribution and egg production. Regression prediction equations. The next step in the analytical process was an attempt to establish multiple variable, multiple regression prediction equations based on the statistical information obtained. These equations would be for mortality, total egg production per bird and egg size distribution per bird. The equations would give an estimate for each value for each 28—day period from 29 period four (assuming peak production in period three) through the expected number of production periods for a given flock. The first equation developed was for total egg pro— duction per 28-day period, from period four through the production cycle. The equation was developed by strain by region. Results of the equation were very poor. The equation accounted for about 32 percent of the varia- tion in egg production in period four. The percent of variation accounted for increased with each period to about 80 percent in periods thirteen and fourteen. Equations for mortality and individual egg size distri- bution were also developed. Results of these equations were similar to those found from the egg production equation. The percent variation in the dependent variable accounted for by the equation was very low in period four. However, as the production cycle progressed, the percent variation that the equation accounted for increased. Gavora-Parker-McMillan model. The results of the multiple regression equations were not acceptable to use for prediction purposes. At this point, the Savors-Parker- McMillan model was introduced to this study. The rate of decrease of lay from the peak period of production through the end of production was the concern of this study. Gavora's, 33 gl., procedure to obtain the rate of 30 decrease (a) was followed within each strain. Different a values for each quarter of start for strain 1 were also obtained. Table 4 presents the a values and potential maximum egg production on a hen housed basis for each strain. The rate of decrease was the slope of the production curve from peak production through the end of production. Strain 2 had the least declining rate of production, but also had the lowest potential peak. On the other hand, strain 4 had the highest potential peak production, but its rate of decrease was the greatest. Gavora, gt gl., (1971) reported the potential maximum 28-day period egg production to be 26.41 eggs per bird housed. None of the values from this study were that specific value. Three of the strains' values were higher than 26.41 while two values were below that level. The a value from Gavora's, 33 23., work was -.034. All of the a values from this study were more negative. The closest to Gavora's reported value was the -.051 of strain 2. Table 5 presents the a values and potential maximum production for the four quarters of start for strain 1. Those flocks started in quarter 2 had the lowest potential peak, and the least rapid decrease in egg production. Those flocks started in quarter 1 had the second highest 31 Table 4. Potential maximum egg production and rats of decrease of lay by strain Rate of Potential Decrease Number of Straina Maximumb ( a ) Flocks 1 25.517 -.055 115 2 240674 -0051 8 5 27.628 -0063 9 aTwo strains included in the statistical analyses were dropped from this analysis due to an insufficient number of flocks b Hen housed basis 32 Table 5. Potential maximum egg production and rate of lay decrease by quarter of start for strain 1 Rate of Potential Decrease Number of Quarter Maximuma (a ) Flocks 3 25.086 -.O497 35 4 25.866 -.0570 36 aHen housed basis 33 potential peak production and the most rapid decline in egg production. The prediction equation utilized for this study was as follows: P(t) a MAXEG(1 - INC(t))DEC(t) where P(t) a predicted eggs per 28-day period per hen housed in period t, where t 3 4(assume peak production in period 3) MAXEG - actual peak production per hen housed INC(t) a rate of lay increase coefficient for period t DEC(t) 3 rate of lay decrease coefficient for period t Gavora, 23 gl., specified the rate of increase of lay to be 2.9011 for egg-type chicken flocks. The coefficient referred to above (INC(t)) was e'2'9011 (t'to) , where t a current period of production and to a 1. The coefficient for period four was 1.68x10'4. The coefficient for periods five and forward was set at zero. The rate of lay decrease coefficient for each period -a(t-3) was s , where t - current production period. This equation was different than its Gavora, gt gl., (1971) counterpart. The difference in the formulas correspond to the great differences in the a values previously discussed. If the a values found in this research were applied to Gavora's formula (e'at) to obtain the lay rate decrease coefficients, erroneous prediction results were obtained. 34 These results were rather confusing since Gavora's pro- cedure, as reported, was duplicated. Thus, the adapted formula developed by this research was utilized later for simulation procedures. Tables 2 and 3 in Appendix I list the rate of lay decrease coefficients for each strain and by quarter of start for strain 1 for each 28-day period. The results of the above prediction equation were very favorable when applied to actual data. It did not predict accurately the actual 28-day period egg production for every flock tested. The reason being that every flock was not an average flock. However, the amount of error was about the same for each period of production. That is, if the equation predicted one egg per hen below actual production for a 28-day period, every period was about one egg per hen below actual production. Some deviation occurred from this pattern very late in the production cycle. The great majority of the differences between pre- dicted production per hen housed and actual production per hen housed per 28—day period ranged from virtually zero eggs per hen to about 1.5 eggs per hen. Table 12 in Appen- dix I offers some examples of actual results. The success of the slope of the production curve pro- cedure in defining an accurate egg production prediction equation gave rise to speculation that the same type analysis could be applied to egg size distribution and 35 mortality. Such analyses would hopefully generage similar successes. Such research, however, was beyond the means of this project. It was originally believed that viable regression prediction equations could be developed. How- ever, attempts to develop such equations failed and in the process engulfed much of the resources of this study. Therefore, to obtain a prediction model for egg size dis- tribution, the following was undertaken. The actual flock data were segregated by strain. An average egg size distribution per period (expressed as a percentage) psr strain was calculated. The same procedure was followed by quarter of start for strain 1. Tables 4 through 11 in Appendix I contain the period by period egg size distri- butions for each strain and by quarter of start for strain 1. Table 6 presents the average egg size distribution by strain. Strain 1 had the lowest average undergrads percen- tage (10.04 percent) and the highest average percentage of large size and medium size (84.4 percent). Strain 4 had the lowest average percentage of large size and medium size eggs with 77.7 percent. Strain 4 also had the lowest average percent of large size eggs. Strain 5 had the highest average percent of undergrads eggs. Table 7 presents the average egg size distribution by quarter of start for strain 1. The flocks started in quar- ter 2 had the highest average percent of large size eggs. 36 Table 6. The average egg size distribution by strain Strain Large Medium Small Undergrade No. % 9% % 1 63.32 21.08 5.56 10.04 2 63.40 15.66 5.86 15.08 3 65.44 17.30 5.87 11.39 4 55.21 22.50 8.97 13.32 5 60.89 17.13 6.63 15.35 37 Table 7. The average egg size distribution by quarter of start for strain 1 Quarter Laggs Megium Smgll Undeigrads 1 61.99 22.13 5.81 10.07 2 66.52 19.64 4.24 9.60 3 62.58 20.47 6.81 10.14 4 62.17 22.10 5.39 10.34 38 Those flocks started in quarter 1 had the lowest percent of large size eggs. Comparing the average percent of large size eggs with the a values of Table 5, one finds an asso- ciation. The flocks started in quarter 2 which had the highest average percent of large size eggs also had the least rapid rate of lay decrease (smallest a value). Similarly, those flocks started in quarter 1 had the lowest percent of large size eggs and the most rapid rate of lay decrease (highest a value). Comparison of the percentages of large eggs and a values among the strains did not yield the same association. Strains 4 and 5, the two lowest large size egg percentage strains, had the correspondingly two most rapid rates of lay decrease. However, strain 3 with the highest percent of large size eggs, did not have the least rapid rate of lay decrease. It is conceivable that the slope of the line procedure can be effectively utilized to predict mortality. The prediction process for mortality selected for use, however, would give a constant percentage value for each 28-day period. This type function concurred with methods reported by Ruggles (1971), Bell (1972), Muir (1972), Ruggles (1968), Vincent and Sheppard (1970) and Dale, Vincent and Sheppard (1974). The process to specify the actual rate, used the following equation: 39 MRT . MORTD/NOB/3 where MRT . the mortality rate, MORTD . total mortality through period 3, and NOB . the number of birds housed. The Simulation Model Procedures The third objective of this study was to develop a computer simulation model that would project overall flock performance after the peak production period. The model would predict mortality, egg production and egg size dis- tribution. The overall projection would involve the application of expected costs and revenues to those pro- jectsd production estimates. The resultant data would be useful for management planning and decision making. The model itself became an accounting exercise once the predictions were made. Excluding the prediction equations, virtually all the data that were used in making calculations were supplied by the individual manager for his specific flock. The specific flock flexibility did, however, require a large amount of input data. The more information that was available to the manager allowed him to make a better deci- sion. This section will explain the subroutines of the com— puter model. 40 UPDATE. The UPDATE subroutine was responsible for bringing the model on line with the farm's position at the end of the third production period (12 weeks of production). The manager supplied the specific aspects of working capital loans during the first three periods and the specifics of loans to pay for birds, house and equipment. The specifics for each loan included the amount of the loan, the interest rate, the total number of 28-day periods to repay the loan, the number of periods remaining to be paid, the number of periods previously paid and the monthly repayment amount. The model established a loan repayment table for each specific loan. This allowed the model to separate the 28— day period repayment into interest paid and principal paid. 0n long term loans (more than one production cycle), the model applied interest paid and principal paid in the amount applicable to the current flock. The interest calculations were based on the simple interest on the unpaid balance method. If all or a portion of the birds, building and equip- ment were owned, the manager supplied prices per bird for each, the expected life of the building and equipment, and the present age. The model then charged the current flock a non-cash depreciation charge on the owned assets. The manager supplied the total medication costs ex- pected, maintenance, utilities, taxes, insurance and any 41 mixcellaneous expenses, each on an annual basis. These expenses were prorated evenly to each production period. The labor calculations required the total operator hours per week, the total hired labor hours per week and the hourly rate for each. The number of extra man hours needed to place the birds in the house and that hourly rate was also required. Egg production data supplied included peak egg pro- duction per hen housed, total mortality, total dozens of eggs produced, total dozens of eggs sold and the total value of those eggs sold. The three previous items were also supplied on an egg size distribution basis. The average prices received per size per period were also needed. The relationships between the average prices per size to date, as a percent of the large size price, became the relationships between prices for the remainder of the production cycle. The model required the manager's ex- pected price for large size eggs per quarter for the next four quarters. The model then generated the anticipated prices for the remaining egg sizes. The manager was to then submit the spent hen price that he anticipated. If the firm had "booked" its feed, the model would use that price in its calculations. If not, the model re- quired the manager's anticipated feed price by quarter for the next four quarters. 42 The model also needed the current interest rate. The model was constructed to handle interest rates by 1/4 percent increments from five percent to fifteen percent. The strain of birds used, the quarter of start, the calen- dar period (1-13) of start, the number of birds housed, the expected number of 28-day production periods and the amount of cash on hand were also included in the input data. The model used all the above information to update itself to the farm's production, operating income and ex- penses and net worth position at the end of the first three periods of production. TCICLE. This subroutine was the time keeper for the model. It functioned to keep account of the production period number and the calendar period number. If the pro- duction period number was less than or equal to the expected production period number, TCYCLE called the remaining sub- routines and, thus, simulated another period of production. If the opposite case was true, TCYCLE terminated the simu- lation. PREDICT. PREDICT was responsible for estimating the total egg production level, the egg size distribution levels and the mortality rate for each period of the remaining periods of the production cycle. The procedures of calcu- lating the estimates were explained in the Results and 43 Discussion portion of the Statistical Analysis section of this report. Table 8 presents a sample computer print-out. The reader may find it helpful to refer to Table 8 while reading the description of each subroutine beginning with subroutine BIRDINV. BIRDINV. This subroutine kept the physical count of the flock size. The estimated mortality rate was applied to the beginning inventory of each period. The subroutine then calculated the period's mortality, the ending inventory, mortality to date and the average number of live birds during the period. EGPROD. This subroutine took the projected total eggs per hen housed per period from PREDICT and calculated the total dozens produced during the period. The egg size dis- tribution from PREDICT (expressed in large, medium, small and undergrads size eggs per hen housed per period) was expanded to dozens of each size produced. The model assumed that 98 percent of the eggs produced would be sold. The two percent difference was caused by loss or unaccounted for eggs. The price received per size (calculated in UPDATE) was applied to the dozens of eggs sold per size to arrive at a value of eggs sold per size. The four egg size values were FEED 80.00 PULLETS 1.65 .260“ HSU POULTRY SIflULATOR II YEAR PERIOD 7 E56 OLENO SOLO 0. PERIODS 13 1697. OF HORTALITY R100 9 pi- ”TORY a- p . FARM HSU rssr 3 1"... area va FJNU” Ndomm (DOnuao CHUCK: O O O O ONJCHJO HOI~U'Tr‘)u\ (CC)MVHDQ h.) I’MDU'S) o o o O U) H o o o o IJNN-fl'm r-cu'w "No F—xD'ONM OCH (0 b—H m m 2 m 0.. X m m 0 t- Z d 0-4 arc.»- .— OOHZ Id Irwacnq I0! UM-Jdldd OUJ Z O H .— L) D m D U] O()O-J ”I'll...“ LLFLQU'D ommvun NHC’fl'O nt’hDN otDDuN . O O O O . O O C . O‘UMI‘RIQ \DJDJ'C‘J 0:0!“an hmuhnc No-AC‘HCO OUQUQ UUL)D\§ e e e o o O O 0 I O cor-3K0 ocrmcsln 130 Y TOTAL ON or) o crumb-4 cocoon-4 07") ONNCO .JM 4 00‘ v4 (3 H J J '4 Nt—JHJ 0000 t! DLL U)LL> ($00660 (DDDL‘OO thTlokDO UL'ICIOL‘DL’ . o O s O 0 0000013 000000 UCNZ'QCDO ULJQDOO Or)u()LJ° O O O O O 0 00.3006 0 0. so 0 00000:: .1 P4 UN— 0 (HO wm (LII?— mO HuJ| b- HJCZ: I) UHDDLUHV) sees. Y T emcee F 44 000 0.4.4 I‘L’Q N00 0 O C o o O 0600‘ ANN U‘O‘O‘ OU)\D InNN 690d.) (av-4'4 \OO‘U‘ “(‘10 O C . DCJCJ Lu-» Oouu 0-22 132.399 TONS CONS 1531.991 ED E FACTS 163. BUS/HAN HR .U... LABOR HOURS cease CASH o e o fiC‘O t)?» C): Md [Dev-4 NO‘ 0‘ U) a. MOQH t—(KCJD Lay-4.10 mcncmu 177527. 2 O hJIZ 10d.) m (r‘I‘r LfihO? H L9QH:?O OIL/MUG..— — D - INCOH 51879. 1Ck3h9. Ab791. 107421. CASH TOTAL 000°C 0 o 0 .0 0000:) 0-0 Hit!!!) (1.] .JCZD O—(d 1053~90 -3071. 1C7h21. NET NORTH .I-U‘m COO O o o O O OHM ODD Ho. 0". RTH (no FI V) LU 2! RATE OF RETURN Sample output of the computer model Table 8. 45 summed and divided by the total number of dozens sold to arrive at the blend price for all eggs sold per period. gagg. The CASH subroutine maintained the cash inflow to the checkbook of the flock. The two sources of income were egg sales and spent hen sales. The subroutine checked the flock's cash balance at the end of the previous period and, if necessary, it established a working capital loan. The loan amount would bring the flock's checkbook cash level back to zero for the start of the new period. This action insured that all cash outlays were paid each period. EXPENSE. The EXPENSE subroutine in addition to its named functions calculated the lay rate percentages, the labor hour situation, and the feed conversion information. The lay rate percentages were a direct function of information that originated in the egg production prediction equations and the bird inventory information. The labor hour situation and cost was a direct function from input information. Feed consumed per period was a function of a mainte- nance level intake, an intake amount directly related to egg production level and the average number of birds per period. The formula used for feed intake was as follows (as adjusted from Vincent and Sheppard (1970)): 46 FDC . (.1537*28) + (10.75*28/1CO) * LAYRATE * AVGBD where FDC - the pounds of feed consumed, LAYRATE - the percent hen housed egg production, and AVGBD - the average number of birds. The (.1537*28) term gave a base feed intake of 4.3036 pounds per bird. The middle term of (1o.75*28/100) * LAY- RATE gave a feed intake level directly related to egg production. This equation gave an average feed conversion of 3.85 pounds per dozen at a 72 percent rate of production (hen housed basis). The feed expense varied by quarter. The labor, medi- cation, maintenance, utilities, taxes, insurance, and miscellaneous expenses were constant each period. The loan repayment and interest expenses were taken directly from the loan repayment tables established in UPDATE or CASH. Net income over cash costs resulted from cash income minus summed expenses minus short term loan repayment. If the short term loan repayment value was included in the expense figure, a double charge arose since cash borrowed was used to meet other cash expense commitments. Net cash flow was the difference between cash income (egg sales and spent hen sales) and cash outlay. This procedure was used to determine the net effect of each pro- duction period's operations on the firm's cash balance. 47 Non-cash depreciation was calculated only on owned assets. If the assets were older than the anticipated length of the useful life, depreciation was not charged. The model calculated depreciation using the straight- line method. The second non—cash cost calculated was an opportunity cost-interest on owned equity. It was calculated at an annual rate of six percent. The equity value used in any period was the ending net worth value from the previous period. NETWTH. This subroutine maintained the flock's balance sheet. The three sections were assets, liabilities and net worth. Within each section a beginning, change and ending balance was maintained for each item per period. The flock asset was depreciated (the change value) over the expected periods of production. The building and equipment were depreciated in value on a straight-line basis over their expected life. The change in the cash balance was equiva- lent to the net cash flow of the current period. The change values in the liability accounts were directly related to their respective loan repayment amounts portrayed in EXPENSE. Net worth was equal to assets minus liabilities. This subroutine also calculated rates of return on total assets and net worth, for both the current period and 48 for the production cycle to date. The rate of return was calculated by dividing net income by either total assets or net worth. If net worth was negative, its two rates of return were not calculated. Results and Discussion Projecting the future is extremely difficult. The degree of difficulty is increased when biological functions are added as a parameter of the decision model. The simulator developed from this research could not predict accurately the outcome of every flock. There were two major factors that affected the outcome. Each had sub- factors that increased the complexities of the problem. One of the major factors was predicting flock egg production performance. The other was the accuracy of the prices anticipated by the flock's manager. The egg production prediction equation coefficients and the egg size distribution coefficients were developed from a series of independent flocks having a spectrum of production. As a result, the coefficients may be considered to have represented the average of each strain's performance. If the simulation model was used to project an average or near average flock, it performed well in projecting egg produc- tion and size distribution. 0n the other hand, if the flock was actually non-average, the model performed poorly. In those poor performance circumstances, the model tended to 49 "miss" by a near constant amount each production period (unusual events excluded--power failure, sudden disease, etc.). That feature (constant miss) gave the user an added benefit. Analyzing production of the fourth period and the fifth period predictions against actual production data gave an average adjustment factor. Quick hand calculated updates of the predicted information then gave the manager an even more accurate projection of his flock. The other major factor involved in the accuracy of the projections was the anticipated prices of the individual manager. They included expected annual flock expenses, such as medication costs and maintenance, among others, the anticipated price to be received for large eggs and the anticipated price to be paid for feed if feed was not "booked” at a constant price. That area posed the greatest uncertainty. If the price information supplied was "good", relative to actual occurrences, the output would similarly be good and vice versa. A sub-factor of the price question was the relationship between the large size price and the other sizes' prices. The model calculated the relationships between the prices during the first three periods of production. Those rela- tionships were then applied throughout the remainder of the production cycle. Those relationships were also subjected to the uncertainty of the future. If they held true, the model's performance in that regard was excellent. 50 Table 9 presents an example of actual farm results compared to predicted results. The dozens sold and dozens of large size eggs sold were both accumulated annual values. The model predicted egg production and dozens sold reasonably accurately. However, the model did not do well on egg size distribution, indicating that the specific flock was not a near average flock compared to others of the same strain. The manager's anticipated prices for eggs and feed were different than actual values. These differences along with the production and egg size differences led to the 83,446 difference in net income. 51 Table 9. performance A comparison of predicted and actual flock Dozens Sold Predicted 450.403 Actual 478.775 Price of Large Eggs Quarter Predicted Actual (cents per dozen) Large Size Dozens Sold Predicted 293.420 Actual 3430760 Price of Feed Quarter Predicted Actual (dollars per ton) 1 50 52 1 115.00 115.50 2 45 43 2 110.00 112.50 3 45 42 3 105.00 107.50 4 42 4O 4 100.00 105.00 Price of Spent Hens Predicted Actual (cents per bird) 6O 55 Predicted Actual Income $209,107 $206,901 Expenses 194,889 196,122 Net Income 8 14,218 8 10,772 SUMMARY Recent records (1970-1973) on 210 commercial laying flocks were obtained from primary breeders and contract and independent producers. The total number of birds involved was in excess of 2.5 million. Information on each flock included egg production, mor- tality and egg size distribution (each by 28-day period), starting age of egg production, strain, number of birds per cage, size of flock, starting month of egg production, loca- tion, total number of 28-day periods they produced and whether they were in a windowed or windowless house. The data were collected to meet three objectives. They were: 1. to determine the factors that affect mortality, egg production and egg size distribution, 2. to develop prediction equations that, on the basis of 12 weeks of actual data, would project reason- ably accurately, the future patterns of egg production, mortality and egg size distribution, and 3. to develop a simulator to incorporate the predic- tion equations as one function and simulate all 52 53 cost-revenue relationships and flock financial performance over the remaining periods of produc- tion to help poultry producers in their flock replacement decision actions. The number of birds involved in the statistical analysis was 2.2 million from 163 flocks. Flocks were eliminated due to an incomplete data list or because they were recycled. Seven strains were involved. They were DeKalb, Hyline, Babcock, H&N, Shaver, Kimber and Arbor Acres. The Kimber and Arbor Acre strains were later dropped because of too few flocks. Two separate least squares multiple regression analyses were conducted. Mortality, total eggs produced per hen housed, and total eggs produced per hen housed as distri- buted among large, medium, small and undergrads sizes were the dependent variables for both analyses. The first analysis tested the effect of flock size, total 28-day periods of production and starting age of production. The total number of 28-day periods of production significantly (P <.05) affected all the dependent variables except the total number of small size eggs produced per hen housed. Starting age of production significantly (P <.05) affected only the total number of eggs produced per hen housed and the total number of medium size eggs produced per hen housed. Flock size significantly (P‘<.05) affected all the 54 dependent variables except the total number of small size eggs produced per hen housed. The significance value for that dependent variable was 0.104, a near significant effect. The data were adjusted for flock size, the total number of 28—day periods of production and starting age of produc- tion. The second analysis had as independent variables the remaining input factors requested and selected interactions. The determination of a significant effect from the least squares results was primarily intuitive. None of the independent variables had a significant effect on mortality. However, the interaction between number of birds per cage and region seemed to have some effect on mortality. The total number of eggs produced per hen housed was significantly affected by quarter of start, number of birds per cage, strain, quarter of start by strain interaction and number of birds per cage by strain interaction. Region and region by number of birds per cage interaction seemed to have an effect although it was not significant. The total number of large size eggs produced per hen housed was not significantly affected by the independent variables and interactions of this study. Region and the absence of windows had some effect on the total number of medium size eggs produced per hen housed. 55 The total number of small size eggs produced per hen housed was significantly affected by region and the presence or absence of windows. The region by number of birds interaction had some effect. None of the independent variables or interactions of this study indicated any influence on the total number of undergrade eggs produced. An attempt was made to develop multi-variable, multiple regression prediction equations for each of the dependent variables. The hope was to develop equations to predict the values from period four of the production cycle through the end of the cycle. The attempt failed. The procedure outlined by Gavora, 23.21- (1971) was utilized to determine a values for each strain for use in the Gavora, £1.21- model. The a value measured the pro- duction curve's slope from the peak through the end of the cycle. Strain 2 had the least slope while strain 4 had the steepest production curve slope among the five strains. Strain 1 was separated by quarter of start and a values were determined for each quarter. Quarter 1 had the steepest slope while quarter 2 had the least slope. Gavora's model was slightly modified and tested. The results were very favorable. The success of the production curve slope concept gave rise to speculation that a similar process could be applied 56 to egg size and mortality. This research, however, could not undertake that venture. The egg size distribution model for simulation purposes became the actual average (per 28-day period) percentages of large, medium, small and undergrade sized eggs. The mortality model became a constant rate as determined by the following equation: mortality rate - total mortality through period 3 of production/number of birds housed/3. The three prediction models became one of the major inputs to the simulation model. The other major input was specific to-date flock production and financial charac- teristics, plus, expected egg and feed prices as supplied by the flock manager. The simulator was designed to project flock production and financial performance by 28-day period from period four through the end of the production cycle with 15 periods being the maximum. The simulator was basically an accounting routine with the exception of the three prediction models. The results of the simulator when compared to actual results were heavily dependent on the accuracy of the manager‘s input data and whether or not the flock involved was average or near average. The egg production and egg size distribu- tion models performed at a near constant deviation from the actual of the non-average flocks, assuming they followed that strain's pattern. The models were highly inaccurate 57 compared to the actual performance of flocks subjected to short-term abnormalities, such as a power stoppage. If the manager's expectations were near accurate, the financial results of the model were very favorable. If his expectations were erroneous, the simulator's financial results were erroneous. CONCLUSIONS The results of the statistical analyses and the simu- lation modeling of this study led the researcher to the following conclusions: 1. 3. 4. Wide variations existed in the performance of the actual commercial flocks surveyed. The average total number of 28-day periods of production for the commercial flocks was 14.18. The average hen housed rate of egg production was 57.68 percent. Total mortality was significantly effected by the total number of 28-day periods of production and by flock size. The total number of eggs produced per hen housed was significantly effected by the total number of 28-day periods of production, starting age of production, flock size, quarter of start, number of birds per cage, strain, quarter of start by strain interaction and number of birds per cage by strain interaction. The total number of large size eggs produced per hen housed was significantly affected by the total number of 28-day production periods and flock size. 58 10. 59 The total number of medium size eggs produced per hen housed was significantly effected by the total number of 28-day periods of production, starting age of production and flock size. The total number of small size eggs produced per hen housed was significantly effected by region and the presence or absence of windows. The total number of undergrade eggs produced per hen housed was significantly effected by the total number of 28-day periods of production and flock size. Multiple variable, multiple regression prediction equations were not acceptable for predicting period by period values (from period four of pro- duction through the end of the production cycle) for mortality, total egg production and egg size distribution. The Gavora-Parker-McMillan model, as adapted, was an excellent predictor of period by period total egg production for average or near average flocks within a given strain. If it was inaccurate, it was by a generally constant amount. The potential peak number of eggs produced per hen housed was different for each strain and for each separate quarter of start for strain 1. 11. 12, 60 The rate of lay decrease was different for each strain and for each quarter of start for strain 1. The rate of lay decrease was less for those strains with the lower potential peak number of eggs compared to the other strains. The simulation model developed gave excellent planning results if (1) the prediction equations' results were reasonably accurate and (2) if the manager reasonably accurately projected cost- revenue relationships. BIBLIOGRAPHY BIBLIOGRAPHY Adams, A. W. and M. E. Jackson, 1970. Effects of cage size and bird density on the performance of six commercial strains of layers. Poultry Sci. 49: 1712. Bell, D., 1971. Cash flow for commercial egg farms. Uni- versity of California, Agricultural Extension Service, Riverside County, California. Bell, D., 1972. Cage density, housing and nutrition. Poultry Scratch, University of California Agricultural Extension Service, Riverside County, California. Dale, J. D., W. H. Vincent and C. C. Sheppard, 1974. Eggman--A computerized egg production management game. Unpublished. Feldkamp, J. F. and A. w. Adams, 1973. Effects of rearing relationships, cage size, and bird density on the performance of two commercial strains of egg-type chickens. Poultry Sci. 52: 1329-1334. Gavora, J. S., R. J. Parker and I. McMillan, 1971. Mathe- matical model of egg production. Poultry Sci. 50: 1306-1315. Hill, A. T. and M. R. Binns, 1971. Effects on the per- formance of varying densities and numbers of birds per cage. Poultry Sci. 50: 1585. Latimer, R. G. and J. Bezpa, 1970. Projections and cash flow for a 30,000 bird commercial table egg operation. No. 1/71/700, Cooperative Extension Service, Rutgers University, New Brunswick, New Jersey. Marks, H. L., L. D. Tindell and R. H. Lowe, 1970. Per- formance of egg production stocks under three cage densities. Poultry Sci. 49: 1094. Muir, F., 1972. A computerized cash flow for market egg farms. No. 570, Cooperative Extension Service, Uni- versity of Maine, Orono, Maine. 61 62 Plumart, P. E., C. W. Carlson and C. E. Holmquist, 1972. Cage density and laying hen performance. Poultry 801. 51: 18500 Ruggles, L. H., 1971. Projecting a cash flow for an egg production firm. No. 68, Department of Agricultural and Food Economics, University of Massachusetts, Amherst, Massachusetts. Vincent, W. H. and C. C. Sheppard, 1970. An application of computer simulation to the feed-integrated poultry industry. Abstracts of Scientifique Communications, XIV World's Poultry Congress, Madrid, Spain. APPENDIX I 63 A.mmwa mHmB mo mmmoomm azmmmmmHQ OPN mmm3 mmmmav Nam NN ram smo.m oem.mo. For ea Poe mm mme mew.m mee.mma mm me rmo.r mm emm moe..P mmm.>eF mm we mmo.a on mom mrF.o. mm>.Fme me we «$0.? mm. mom.m 4mm.ee oeo..m. we or mmm cm? me.n pmm.m 0mm.nmr me o nmm.. emu mee.e Pem.m mmm.>mP Fe m ome.. pom enm.m Fem.ow mmm.eom cor a eme.. own eem.m nmm.m rmm.emm mmm o mmm mew moo.» e>>.> cm>.mmm an, m ene.w ore." mom.mP sem.m omm.eem we? 4 000.? neo.m mom.> emm.n mno.mmm m.m n moo.e mmm.m «me.m omn 0n>.o.m mp. m Pmo.. mmm em mm www.mm om r mumamumosp Hamsm segue: swung umowmoam haaddpnoz voanmm m m uHom ammuon oen.or “needs He eunm zen needeem no» ”mkocsa3 m "ammo mom acnfim mamzunom ”npmoz mmapamvm ammo: om “omd mnfipnmpm mamamxm I mpmw zen Hmspo< .P magma 64 Table 2. Rate of lay decrease coefficients for each period for strains 2, 3, 4 and 5 Strain Period 2 3 4 5 4 .950283 .943654 .931463 .938949 5 .903038 .890484 .86763 .881622 6 .858139 .840304 .80816 .827791 7 .81547 .79295 .75277 .77725 8 .774925 .74826 .70119 .7298 9 .736398 .706106 .65313 .685234 10 .699783 .666318 .60836 .643399 11 .664984 .6268768 .56667 .604118 12 .631915 .593336 .52782 .567234 13 .6005 .5599 .49164 .53259 14 .570642 .52835 .45795 .500083 15 .542268 .498586 .42656 .46955 65 Table 3. Rate of lay decrease coefficients for each period by quarter of start for strain 1 Quarter Period 1 2 3 4 4 .94195 .954414 .95153 .944222 5 .88998 .910894 .910476 .89155 6 .83577 .86936 .861575 .841814 7 .787262 .829719 .81975 .7948496 8 .741564 .7198835 .78 .750516 9 .69852 .75578 .742308 .7086548 10 .657974 .7213132 .70623 .66891194 11 .619775 .6884217 .67199 .631788 12 .5838 .65705 .6394184 .59655 13 .5499 .6270724 .608421 .563278 14 .517994 .5984848 .5789278 .531946 15 .487924 .57121 .5508635 .5021848 66 Table 4. Egg size distribution coefficients for each period for strain 2 Large Medium Small Undergrade Period (Percentages) 4 56.92 31.66 1.96 9.46 5 72.26 16.80 1.03 9.92 6 75.25 12.87 1.06 10.83 7 80.50 7.60 .26 11.65 8 81.97 5.25 .18 12.61 9 84.41 3.69 .09 14.82 10 79.10 3.50 .09 17.32 11 78.31 3.22 .11 18.36 12 75.56 3.12 .13 21.19 13 76.46 3.96 .12 19.45 14 79.01 3.30 .14 17.56 15 79.01 3.30 .14 17.56 67 Table 5. Egg size distribution coefficients for each period for strain 3 Large Medium Small Undergrade Period (Percentages) 4 61.78 29.06 2.35 6.82 5 71.44 18.98 1.07 8.51 6 76.28 13.73 .57 9.41 7 77.34 11.84 .51 10.31 8 78.60 10.05 .37 10.98 9 79.49 8.82 .32 11.38 10 80.63 7.78 .24 11.35 11 81.20 6.22 .19 12.39 12 81.80 6.73 .24 11.23 13 80.15 7.30 .25 12.31 14 77.92 7.80 .28 14.00 15 77.92 7.80 .28 14.00 68 Table 6. Egg size distribution coefficients for each period for strain 4 Large Medium Small Undergrade Period (Percentages) 4 41.34 47.40 4.45 6.82 5 57.36 33.72 1.94 6.98 6 64.04 25.50 1.49 8.97 7 67.54 20.52 .84 11.10 8 69.17 17.64 .65 12.54 9 72.99 14.25 .49 12.26 10 72.16 12.41 .50 14.93 11 72.80 10.50 .47 16.24 12 75.10 8.75 .43 15.72 13 77.26 8.74 .38 13.61 14 76.84 10.70 .55 11.92 15 76.84 10.70 .55 11.92 69 Table 7. Egg size distribution coefficients for each period for strain 5 Large Medium Small Undergrade Period (Percentages) 4 56.58 32.79 2.31 8.32 5 66.50 22.72 1.34 9.42 6 70.43 17.02 1.43 11.12 7 72.66 15.08 .99 11.28 8 75.90 11.37 .58 12.15 9 79.02 6.75 .42 13.82 10 79.21 5.31 .68 14.80 11 79.18 4.27 .10 16.46 12 78.47 3.77 .14 17.63 13 75.02 5.36 .16 19.46 14 71.50 3.39 .11 24.99 15 71.50 3.39 .11 24.99 70 Table 8. Egg size distribution coefficients for each period for strain 1 for quarter of start 1 Large Medium Small Undergrade Period (Percentages) 4 54.53 34.43 3.70 7.33 5 61.22 28.80 2.16 7.82 6 66.82 23.08 1.69 8.41 7 70.22 19.56 1.24 8.98 8 73.64 16.06 .63 9.66 9 77.28 12.37 .54 9.81 10 78.15 11.08 .40 10.37 11 79.13 9.14 .29 11.44 12 78.62 8.12 .21 13.05 13 78.63 7.78 .37 13.22 14 78.00 7.56 .32 14.12 15 78.00 7.56 .32 14.12 71 Table 9. Egg size distribution coefficients for each period for strain 1 for quarter of start 2 Large Medium Small Undergrade Period (Percentages) 4 55.35 35.00 2.13 7.52 5 64.67 26.18 1.56 7.59 6 73.22 17.70 .76 8.32 7 76.92 13.80 .54 8.74 8 78.62 10.40 .21 10.77 9 78.47 10.16 .17 11.20 10 82.17 7.67 .34 9.82 11 82.94 6.07 .13 10.87 12 81.54 6.16 .25 12.05 13 80.92 7.58 .16 11.34 14 78.53 8.04 .28 13.15 15 78.53 8.04 .28 13.15 72 Table 10. Egg size distribution coefficients for each period for strain 1 for quarter of start 3 Large Medium Small Undergrade Period (Percentages) 4 55.45 34.09 3.53 6.94 5 65.36 25.67 1.47 7.50 6 73.09 18.14 .72 8.05 7 77.28 13.80 .43 8.50 8 77.51 11.74 .42 10.33 9 76.18 13.22 .35 10.25 10 78.71 9.33 .30 11.67 11 78.57 8.95 .42 12.06 12 76.94 9.68 .45 12.93 13 76.86 8.75 .38 14.01 14 76.74 8.25 .43 14.58 15 76.74 8.25 .43 14.58 73 Table 11. Egg size distribution coefficients for each period for strain 1 for quarter of start 4 Large Medium Small Undergrade Period (Percentages) 4 55.23 34.64 2.48 7.65 5 64.88 25.68 1.19 8.26 6 70.91 19.22 .89 8.98 7 73.91 15.29 .82 9.98 8 75.67 13.44 .79 10.09 9 75.83 13.03 .77 10.38 10 75.47 12.77 .76 11.01 11 75.44 11.96 1.10 11.51 12 75.58 11.17 .86 12.39 13 77.02 9.05 .65 13.28 14 78.72 7.36 .38 13.54 15 78.72 7.36 .38 13.54 74 Table 12. Two examples of the egg prediction model Actual Estimated Difference Period Eggs Per Hen Housed Producer Number 1 4 19.96 19.58 .38 5 19.13 18.49 .64 6 18.28 17.46 .82 7 17.24 16.48 .76 8 16.28 15.56 .72 9 15.45 14.69 .76 10 14.68 13.87 .81 11 13.79 13.10 .69 12 12.47 12.37 .10 13 11.73 11.68 .05 14 11.13 11.03 .10 Producer Number 2 4 16.52 18.09 -1.57 5 15.81 17.19 -1.39 6 13.98 16.34 -2.36 7 11.84 15.53 -3.69 8 11.62 14.75 -3.13 9 11.91 14.02 -2.11 75 Table 13. An example of the measurement of an independent variable's effect on a dependent variable from the least squares analysis (Dependent Variable - Total Eggs Produced Per Bird Housed) Variable Group Variable Number Significance Level Region: 15 .017 16 .218 17 .150 Region by Quarter of Start: 41 .609 42 .951 43 .149 44 .956 45 .222 46 .915 48 .429 49 .604 Explanation: Region: Two of the three listed variables had signifi- cance levels less than .2, however, one of the two was relatively high (.15). The third variable was outside the specified significance range, although close. Thus, the variable region was determined to have some effect, although not a significant effect. Region by Quarter of Start: A quick scan of the signi- ficance levels indicated that this interaction had no effect on the dependent variable. 76 nude xooau mama on» u nu soaposooan Ho ems maavumpn sees one u NH soaposeonm Ho mcoanom mdclmm aspen no panama sacs on» u «H A390» nose op copwaon Macaw no mean on» u Anvnx A390m memo op copmaon moaposcoum no ems maapnmpm one I ASNH A390m nose on cmpwflon moaposwoam Ho muoauom hmclmm no Hones: Hmpov on» u Afivww ouam Macaw mo essays on» no opmsapmo on» I nm& 832693 No ems $53um on» No vacuum 23 mo 3883mm 6:» a «Q 3623 83269.3 hmolmw no 9355 Hmpop 65. no pommmm on» no 395»: on» .- PQ osz> moapznwapmfic mafia mwo one sowvozwoag mwo .hvaampnos oobummpo moms n A329m msam> soapspwpvmac oufim mmo cam moaposoonm mwo .hvaamvuoa covmsncm Home a Anvcmh smog) mfllanvmk m%l NHIAnvNN N% I $7.2”va Ffil Anvnoh I Anvcwh essence pdeaemdned seen one .e. magma MICHIGAN STATE UNIV. LIBRARIES 1|”“1""WWWIHIMNIIIIWIIWHIIIHI 31293107883096