SciELO - Scientific Electronic Library Online

vol.52 issue3Reproductive cycle of Parauchenipterus striatulus (Pisces- Auchenipteridae) in the Ribeirão das Lajes reservoir - Brazil author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Arquivo Brasileiro de Medicina Veterinária e Zootecnia

Print version ISSN 0102-0935On-line version ISSN 1678-4162

Arq. Bras. Med. Vet. Zootec. vol.52 n.3 Belo Horizonte June 2000 

Economies of scale in the production of swine manure

[Economias de escala na produção de dejetos de suínos]


W.C. Losinger, R.K. Sampath

Department of Agricultural and Resource Economics
Colorado State University
Fort Collins, Colorado 80523 USA


Recebido para publicação, após modificações, em 15 de outubro de 1999.



Manure production on grower/finisher swine operations in the United States was examined using data from 184 grower/finisher swine operations that participated in the United States National Animal Health Monitoring System's 1995 National Swine Study. Two methods were used: one, assuming that pigs produced 8.4% of their body weight in manure each day; another using the difference between feed fed and weight gained as a proxy variable to study manure production. Using this latter approach, a production function was developed. The function exhibited diminishing returns to scale when food waste was not fed to pigs, but constant returns to scale when food waste was included in their diets. The difference between feed fed and weight gained was lower on operations that restricted entry to employees only.

Keywords: Swine, waste production, manure management, pork production, returns to scale



A produção de dejetos em granjas de crescimento e terminação de suínos nos Estados Unidos foi avaliada utilizando dados de 184 granjas participantes de um estudo nacional de 1995 do "United States National Animal Health Monitoring System". Dois métodos foram usados: um considerando que suínos produzem 8,4% do seu peso corporal de dejetos por dia e o outro usando a diferença entre o alimento ingerido e o ganho de peso como um indicador para o estudo da produção de esterco. Através desse último procedimento, desenvolveu-se uma função de produção.

Palavras-chave: Suíno, produção de dejetos, economia de escala




Farmers have traditionally valued animal manure as a fertilizer for crops, and frequently integrated animal and crop production by feeding crops that they grew to their livestock and using livestock manure to fertilize crops (Westenbarger & Letson, 1995). With increasing number of pigs being produced on large specialized operations, problems associated with the large quantity of manure produced have intensified (Honeyman, 1996; Letson & Gollehon, 1996). Public health and environmental concerns include adverse impacts on both ground water and surface water, effects on air quality, and effects on soil quality (Integrated..., 1996). In arid locations (such as in many of the western states), high rates of manure application can result in the accumulation of salts in the soil (Integrated... 1996). Regional differences in swine manure management have been noted, with operations in Iowa tending to apply manure on land to recycle nutrients to grow corn, and operations in North Carolina tending to use anaerobic lagoons to treat manure (Hoag & Roka, 1995). Uses for manure (in addition to land application) include recycling manure in livestock and fish feed, conversion to fuel and energy, synthesizing organic compounds, and commercial mushroom production (Integrated.., 1996).

In the past, economic research on livestock production often ignored the value of manure (Hoag & Roka, 1995). Since alternative uses (other than land application) for swine manure are not well developed in the United States, swine manure probably has more value to producers who raise relatively few pigs and also grow crops than for large, specialized operations that raise large numbers of pigs in confined settings.

Relatively little work has been done on the quantity of manure produced by pigs raised under different circumstances. According to standards published by the ASAE (1997), pigs produce 84 kg of total manure per 1,000kg of live mass per day, with a standard deviation of 24. Roka & Hoag (1996) assumed that manure production equaled 8.5% of pigs’ body mass per day, and concluded that economies of size caused manure value to increase with herd size. Bridges et al. (1992) developed a computer model which simulated the physiological growth of pigs and which computed dry matter waste as the sum of material passing through the cecum-colon at each time step. Clanton et al. (1991) reported a large variation in urinary and fecal qualities as pigs were exposed to different environmental conditions.

Honeyman (1993) and Bridges et al. (1995) reported that the amount nitrogen and phosphorus excreted by pigs could be controlled by altering the diet. They did not report any findings on the total quantity of manure produced, although this may be less important for the environment than the amount of nitrogen and phosphorus contained in the excreta.

According to the MidWest Plan Service (1985), manure properties depend on several factors, including type of ration, animal age, productivity, and the environmental conditions. The MidWest Plan Service (1985) listed 125.0 kg gestating sows as producing an average of 4.1 kg of manure per day, compared to 5.9 kg per day for 90.7 kg finisher pigs. Hatfield et al. (1998) stated that better efficiency of converting feed to meat leads to a lower rate of converting feed into waste. Hatfield et al. (1998) concluded that, given recent advances in production efficiency, applying previous estimators of swine-waste production on modern operations will generally result in overestimating the quantity of swine-waste produced.

The United States National Animal Health Monitoring System (NAHMS), which is a program of the United States Department of Agriculture (USDA), was designed to provide timely information about interactions between animal health, animal welfare, animal production, human health, and the environment (Bush & Gardner, 1995). The NAHMS 1995 National Swine Study involved producers who were randomly selected from operations in 16 states to permit inferences to be drawn to over 90% of the swine inventory and over 70% of the swine producers in the United States (Losinger, 1997). In a producer survey, to derive directly estimates of the quantity of swine manure produced would be impossible. Roka & Hoag (1996) concluded that the marginal change in manure value was quite small relative to the marginal change in returns from pork production, and that the value of pork therefore clearly dominated a producer’s decision-making process. Probably few producers keep accurate records on the quantity of swine manure produced on the operation. Therefore, a researcher wishing to use survey data to make inferences concerning the quantity of manure produced on swine operations generally must do so indirectly through records that relate more specifically to pork production. One may assume, as Roka & Hoag (1996) did in their analysis, that the amount of manure produced is directly proportional to the weight of the pigs. However, this would preclude the possibility of analyzing manure production in isolation from pork production. Essentially, any variable that applied to pork production would apply identically to manure production.

Another approach, which we propose here, is to examine manure production through a proxy variable developed from production parameters on which pork producers more commonly focus attention (e.g., feed-conversion ratio, average daily gain, and average number of days in the grower/finisher unit). The proxy variable represented the difference between the amount of feed fed and amount of weight gained by finisher pigs marketed at slaughter weight, culled and died over a six-month period. By means of this proxy variable, it was possible to develop a production function that identified factors influencing swine-manure production and that examined returns to scale for the production of manure in the grower/finisher phase of swine production in the United States. For comparison, the quantity of manure produced was also calculated using the assumption that this was directly proportional to the body weight of the pigs in the grower/finisher unit.



Data used in this study were collected during the NAHMS 1995 National Swine Study. The study took place in 16 major pork-producing states that accounted for over 90 percent of the country’s swine inventory and about three-fourths of the country’s pork producers (Losinger et al., 1998). In the first stage of data collection (1-23 June, 1995), enumerators from the USDA: National Agricultural Statistics Service interviewed 1,477 producers involved in all phases of swine production, and gathered information on general management practices. In the second stage of data collection (limited to first-stage participants with > 300 finisher pigs), state and federal veterinary medical officers and animal health technicians collected data from 418 producers (once between 17 July and 15 September, 1995, and again between 6 November, 1995, and 17 January, 1996). At the final interview, producers reported the number of finished pigs marketed at slaughter weight (n1), the number of pigs that had been culled and marketed prior to reaching slaughter weight (n2), and the number of pigs that had died in the grower/finisher unit (n3) during the previous six months. Producers were also asked the average number of days in the grower/finisher unit (x3), the average daily gain (a), and the feed-conversion ratio (f) in the grower/finisher production phase.

Manure production was initially computed directly assuming that all pigs produced 8.4% of their body weight in manure per day, following standards published by the ASAE (1997). Pigs were assumed to have a body weight of 20.5 kg upon entering the grower/finisher unit, and, at the end of the grower/finisher production phase, to weigh 20.5 kg plus the product of the average daily gain (a) and the average number of days in the grower/finisher unit (x3). For the n1 pigs marketed at slaughter weight, the kg of manure produced was estimated as

m1 = 0.084 ´ n1 ´ (41 + ax3)/2

No survey information was collected on the age at which finisher pigs were culled or died in the grower/finisher unit. Therefore, for the n2 pigs culled and marketed prior to reaching slaughter weight, the kg of manure produced was estimated thus:

m2 = 0.084 ´ n2 ´ (41 + ax3/2)/2

Similarly, for the n3 pigs that had died, the kg of manure produced was estimated as:

m3 = 0.084 ´ n3 ´ (41 + ax3/2)/2

Total manure produced was then derived by adding m1, m2 and m3.

Number of pigs (x1) was the sum of n1, n2 and n3. Kilograms of feed fed (x2) was estimated thus:

x2 = (n1 + n2/2 + n3/2)x3a f

Kilograms gained by pigs (y) was estimated from a, x3, n1, n2 and n3:

y = (n1 + n2/2 + n3/2)x 3a

The proxy variable for the amount of manure produced (w) was then computed by subtracting y from x2.

Over 90 continuous and dichotomous variables (previously listed, Losinger, 1997) were initially considered for inclusion in a production function with w as the dependent variable. To evaluate possible collinearity among explanatory variables, the SAS CORR procedure (SAS, 1990) was used to obtain Spearman rank correlation coefficients (Hogg & Craig, 1978). Spearman correlation coefficients, which are based on ranks, were computed rather than Pearson correlation coefficients because many of the variables were dichotomous (and took values of zero or one).

To screen potential explanatory variables, all variables considered were offered for stepwise regression using the SAS REG procedure (SAS, 1989). Logarithms were computed for continuous variables. A significance level of 0.10 was required for a variable to enter and remain in the model (for screening purposes). Variables in this model were deemed "screened," and were considered for inclusion in the final production function.

Interaction terms between logarithms of screened continuous variables and screened dichotomous variables were created by multiplying the value of the logarithm of the continuous variable by the value of the dichotomous variable for all of the screened variables. To allow for the eventuality of a translog (or transcendental) production function (Christensen et al., 1973), multiplicative terms were created by multiplying together the logarithms of screened continuous variables, and squared terms were created by squaring the logarithms of screened continuous variables. To permit the development of a generalized power production function (de Janvry, 1972), the values of screened continuous variables were multiplied by the logarithm of all other screened continuous variables. Variable selection was run with stepwise variable selection of the SAS REG procedure (SAS, 1989), using a significance level of 0.05 for variables to enter and remain in the model.

Since a Cobb-Douglas type of production function resulted, the determination of returns to scale was straightforward. In an ordinary Cobb- Douglas type of production function, the returns-to-scale parameter (R) is determined by summing the coefficients of the continuous explanatory variables in the linear-logarithmic form (Henderson & Quandt, 1980). A returns-to- scale parameter less than/equal to/greater than one indicates decreasing/ constant/increasing returns to scale. In the presence of dichotomous variables which interact with the continuous variables, the returns-to-scale parameter may be computed thus:

where b1 represents the coefficient for the i-th continuous variable (in linear-logarithmic form); zj = 1 if the operation possesses the j-th dichotomous characteristic, 0 otherwise; and lij represents the coefficient for the interaction term between the i-th continuous variable (in linear- logarithmic form) and the k-th dichotomous variable. If such interaction terms are present, returns to scale may vary depending on the values of lij. In addition, the returns-to-scale parameter may vary depending upon whether one believes particular inputs to be variable (if an input is believed to be fixed, then it should not be used in the computation of the returns-to-scale parameter).

The SAS GLM procedure (SAS, 1989) was used to obtain estimates and standard errors of R. T-tests were used to test the null hypothesis that R = 1 (i.e., that constant returns to scale exist).

The marginal physical product (MPP) refers to the change in output associated with an incremental change in the use of an input (Henderson & Quandt, 1980):

The average physical product (APP) was computed simply as

For each of the operations used in the development of the production function, marginal physical products and average physical products were calculated for all continuous explanatory variables in the production function. Then, the SAS UNIVARIATE procedure (SAS, 1990) was used to obtain summary statistics of the results.



Of 418 operations that participated in the second stage of data collection for the 1995 National Swine Study, 184 (44.0%) provided enough data to examine parameters of manure production. Table 1 summarizes descriptive statistics for manure production computed assuming that all pigs produced a fixed 8.4% of their body weight in manure per day. Most of the manure was produced by pigs marketed at slaughter weight.



Table 2 presents summary statistics on the difference between kg of feed fed and kg of body weight gained in the grower/finisher unit, which was proposed as a proxy variable for examining swine-manure production. On average, this was 35.8 times higher than the total manure production estimate which assumed fixed manure production per body weight of pig. The variable x1 was negatively correlated with the ratio of w to the total manure production estimate that assumed fixed manure production per body weight of pig (Spearman correlation coefficient = –0.279, P<0.001). Therefore, as the number of pigs increased, the value of the proxy variable w fell relative to the estimate of manure production which would have been obtained under the assumption that all pigs produced the same percentage of manure relative to their body weights.



Table 3 provides the coefficients for the production function. The model indicated that, holding other variables constant, increasing the number of pigs resulted in a reduction in the difference between the amount of feed fed and body weight gained; increasing the amount of feed resulted in an increase in the difference between feed fed and body weight gained; and increasing the number of days in the grower/finisher unit resulted in a decrease in the difference between feed fed and body weight gained. The difference between feed fed and weight gained was reduced on operations that restricted entry to employees only. Feeding food waste (garbage) to pigs was associated with an increase in the difference between feed fed and body weight gained, which increased as the number of pigs increased.



Table 4 lists the Spearman rank correlation coefficients for the model variables. As one would expect, x1 (number of pigs) was highly correlated with x2 (amount of feed).



Table 5 shows that, if days in the grower/finisher unit (x3) is considered a fixed input, the returns-to-scale parameter (R) was significantly below one (indicating diminishing returns to scale) for operations that did not feed food waste to pigs, but not significantly different from one for operations that did not feed food waste. If x3 is considered a fully variable input, R is significantly below one regardless of the practice of feeding food waste to pigs.



Table 6 presents summary statistics on marginal physical products and average physical products for the continuous explanatory variables in the production function. On average, an extra pig would have resulted in a reduction of 28.9 kg in the difference between feed fed and body weight gained, while an extra kg of feed would have resulted in an extra 0.79 kg in the difference between feed fed and body weight gained. A rather wide variation was observed for MPP3 and APP3 (Table 6). As one would expect, the difference between feed fed and body weight gained resulting from an extra day in the grower/finisher unit was related to the size of the operation. MPP3 was highly correlated with x1 (rho = -0.978) and x2 (rho = -0.979). In addition, APP3 was highly correlated with x1 (rho = 0.977) and x2 (rho = 0.978).




A rather large number of participants (56.0%) did not report figures for both average daily gain and feed conversion, and, therefore, their data could not be used in this study. In addition, many of the producers gave figures for average daily gain and feed efficiency that were estimated or guessed rather than calculated accurately (USDA, 1996). In any sample survey, some nonsampling error is destined to occur (Sukhatme & Sukhatme, 1970). Quality of reported average daily gain and reported feed efficiency were offered for multivariable modelling for the production function, but never entered the model. Thus, one probably does not need to be overly concerned about potential biases presented. Still, producers can undeniably receive benefits from maintaining an accurate record-keeping system (Muirhead, 1976; Edwards et al., 1989; Boland & Patrick, 1994). Producers who knew the average daily gain and feed-conversion ratios of their pigs most likely computed these figures from the quantity of pork produced and amount of feed fed over a particular time frame for a particular number of pigs. In this study, the figures which the producers gave for average daily gain and feed-conversion ratio were used to compute estimates of the amount of pork produced and amount of feed fed for the pigs included in the study period.

Literature on animal growth generally assumes that growth of animals follows a sigmoidal shape rather than the linear projection used here to estimate manure production. Roka & Hoag (1996) used an empirical model that approximated an S-shaped curve to estimate growth and manure production. However, the model which they used did not allow for differences in average daily gain of pigs. In the NAHMS 1995 National Swine Study, different producers reported different values for average daily gain in the grower/ finisher unit. Average daily gain certainly varies depending on the age of the pig. In our analysis, we assumed that both the average daily gain and the feed-conversion ratio which the producer reported were the average for the whole grower/finisher production phase.

In epidemiologic and socio-economic analyses, to use a proxy variable as a surrogate for a variable of interest when the variable of interest cannot be measured directly is a fairly common practice (Maddala, 1988). For example, Frisbie et al. (1997) used several variables (maternal education, annual household income, public assistance, etc.), that fell under the broad category of "socio-economic status", to examine determinants of compromised birth outcomes. "Socio-economic status" cannot be measured by itself, but only indirectly through variables which stand for socio-economic status. Nerrie et al. (1990) used annual capital recovery charge as a surrogate for capital in analyzing catfish production. Losinger & Heinrichs (1997) found a relationship between rolling herd average milk production and mortality among preweaned dairy calves, and noted that, while herd milk production in itself had no obvious impact on preweaned calves, a higher level of milk production was indicative of more careful management, which could have led to reduced mortality among preweaned calves.

Certainly not all of the feed fed to pigs is turned into pork or manure. Some is used in metabolic processes including respiration and transpiration. Muirhead (1989) stated that approximately 35% of dietary energy was used by pigs for maintenance, 25% for growth, and 40% was lost from the body as heat. Some feeds result in greater feed efficiency than other feeds (Muirhead, 1989). Unconsumed feeds may be disposed of together with the manure. Although little is known about actual quantities of manure (and other waste) produced from feed not converted into pig flesh on farms in the United States, the quantity and characteristics of swine manure are known to be affected by the diet, environmental conditions (including ambient temperature), and body weight of the pigs (Clanton et al., 1991). For many purposes (for example, to project capacity requirements for an operation’s waste-storage facility) one might reasonably assume that all pigs produce a quantity of manure equivalent to a certain percentage of their body weight per day. However, to examine the economics of swine waste production on the national level, one would need to move beyond this assumption. This study demonstrated that decreasing returns to scale existed in the difference between feed fed and body weight gained by pigs in the grower/finisher production phase among operations not feeding garbage to their pigs. Further research is required to determine how well this variable applies as a proxy for swine-manure production. Although the relationship between manure production and the difference between feed fed and body weight gained is unknown, it seems reasonable to postulate that the amount of manure produced increases as the difference between the kg of feed fed and the kg of body weight gained increases.

Intercorrelations among explanatory variables occur commonly in economic analyses (Neter & Wasserman, 1974). When the independent variables are highly correlated, estimated regression coefficients can become imprecise (Neter & Wasserman, 1974). However, Maddala (1988) argues that high intercorrelations among explanatory variables do not necessarily create a problem, and that many suggested remedies may do more harm than good. The stepwise regression procedures used here served to screen out independent variables too highly correlated with explanatory variables already in the model (Neter & Wasserman, 1974).

An important criterion for selecting from among possible functional forms to describe an economic relationship is that the model should be useful in predicting outcomes (Lau, 1986). One well-known limitation of the Cobb-Douglas type of production function is that elasticities of substitution between all pairs of inputs are one (Lau, 1986). Variable transformations (that would have turned the model into a transcendental or power production function) did not enter the model.

Many larger operations are highly specialized in pork production, do not raise crops, and thus have little use for the manure. However, the efficiency of converting feed into meat (as measured by the feed-conversion ratio) is widely recognized as an important determinant of profitability for pork producers (Edwards et al., 1989; Muirhead, 1989; Henry, 1992). Feed costs have a very strong impact on profits for swine producers (Edwards et al., 1989). Therefore, the desirability of reducing feed costs may have been a more compelling factor influencing returns to scale in the production function than the need to reduce manure. Manure reduction may have been a complementary benefit for some producers.

Biosecurity measures are important to reduce the risk of introducing disease to a swine operation (Van Arsdall & Nelson, 1985). People may serve as vectors for disease organisms in pigs (Muirhead, 1976). Permitting only employees to enter the operation may be emblematic of a high level of biosecurity on an operation. The 1995 National Swine Study showed that 40.5+2.1 % of operations restricted entrance to employees only (USDA, 1995). It is possible that operators who restricted entry to employees either successfully prevented the introduction of diseases that could have induced an increase in manure production, or managed their operations in other ways that engendered an overall reduction in the difference between feed fed and body weight gained.

Sharma et al. (1997) found that although feed costs were lower on operations that fed garbage, total costs per kg of live pig produced were higher, in part due to higher feed and labor costs.



The common assumption that pigs produce a fixed percent of their body weight in manure each day may not hold well for economic analyses relating to this product of the swine industry. An analysis of the difference between feed fed and body weight gained (proposed as a proxy variable for waste production) indicated the presence of diminishing returns to scale when garbage was not fed to pigs.

The results of this study showed that operations where meticulous biosecurity was practised (as attested by restricting entry to employees only) had a reduced difference between feed fed and body weight gained during the grower/finisher phase of swine production. Although feeding food waste was implicated in increased returns to scale in the difference between feed fed and body weight gained, the social costs of this may have been offset to some extent by being able to recycle the food waste.

While the presence of diminishing returns to scale suggests that less manure per pig may be produced as the scale of the enterprise increases, the chief problem is aggregate waste relative to what a farm can utilize. A principal concern is the impact of waste from large, intensive swine operations overwhelming the assimilative capacity of the air, land and water and conducing to pollution damages. This is an area where further knowledge is required and increased study is warranted.



ASAE. American Society of Agricultural Engineers, ASAE Standards 1997, 44.ed., St. Joseph, Michigan: ASAE, 1997. p.649.        [ Links ]

BOLAND, M.A., PATRICK, G.F. Measuring variability of performance among individual swine producers. Rev. Agric. Econ., v.16, p.75-82, 1994.        [ Links ]

BRIDGES, T.C., TURNER, L.W., STAHLY, T.S. et al. Modeling the physiological growth of swine part I: model logic and growth concepts. Trans. Am. Soc. Agric. Eng., v.35, p.1019-1028, 1992.        [ Links ]

BRIDGES, T.C., TURNER L.W., CROMWELL, G.L. et al. Modeling the effects of diet formulation on nitrogen and phosphorus excretion in swine waste. Appl. Eng. Agric., v.11, p.731-739, 1995.        [ Links ]

BUSH, E.J., GARDNER, I.A. Animal health surveillance in the United States via the National Animal Health Monitoring System (NAHMS). Epidém. Santé Anim., v.27, p.113-126, 1995.        [ Links ]

CHRISTENSEN, L.R., JORGENSON, D.W., LAU, L.J. Transcendental logarithmic production frontiers. Rev. Econ. Stat., v.5, p.28-45, 1973.        [ Links ]

CLANTON, C.J., NICHOLS, D.A., MOSER, R.L. et al. Swine manure characterization as affected by environmental temperature, dietary level intake, and dietary fat addition. Trans. Am. Soc. Agric. Eng., v.34, p.2164-2170, 1991.        [ Links ]

INTEGRATED Animal Waste Management. Task Force Report No. 128. Ames, Iowa: Center for Agricultural Science and Technology, 1996.        [ Links ]

De JANVRY, A. The generalized power production function. Am. J. Agric. Econ., v.54, p.234-237, 1972.        [ Links ]

EDWARDS, W.M., VAN DER SLUIS, G.T., STEVERMER, E.J. Determinants of profitability in farrow-to-finish swine production. North Central J. Agric. Econ., v.11, p.17-25, 1989.        [ Links ]

FRISBIE, W.P., BIEGLER, M., TURK, P. et al. Racial and ethnic differences in determinants of intrauterine growth retardation and other compromised birth outcomes. Am. J. Publ. Health, v.87, p.1977-1983, 1997.        [ Links ]

HATFIELD, J.L., BRUMM, M.C., MELVIN, S.W. Swine Manure Management. In: WRIGHT, R.J., KEMPER, W.D., MILLNER, P.D. et al. (Eds.) Agricultural Uses of Municipal, Animal and Industrial By Products. USDA-ARS Conservation. Research Report Number 44, p.78-90, 1998.        [ Links ]

HENDERSON, J.M., QUANDT, R.E. Microeconomic Theory: A Mathematical Approach, 3.ed. New York: McGraw-Hill, 1980.        [ Links ]

HENRY, Y. Influence of diet composition on feed efficiency and utilisation in growing-finishing pigs. World Rev. Anim. Prod., v.27, p.75-92, 1992.        [ Links ]

HOAG, D.L., ROKA, F.M. Environmental policy and swine manure management: waste not or want not? Am. J. Altern. Agric., v.10, p.163-166, 1995.        [ Links ]

HOGG, R.V., CRAIG, A.T. Introduction to Mathematical Statistics. New York: Macmillan. 4.ed., 1978.        [ Links ]

HONEYMAN, M.S. Environment-friendly swine feed formulation to reduce nitrogen and phosphorus excretion. Am. J. Altern. Agric., v.8, p.128-132, 1993.        [ Links ]

HONEYMAN, M.S. Sustainability issues of U.S. swine production. J. Anim. Sci., v.74, p.1410-1417, 1996.        [ Links ]

LAU, L.J. Functional forms in econometric model building. In: Grileches, Z., Intriligator, M.D. (Eds.), Handbook of Econometrics Amsterdam: Elsevier, 1986, v.3, p.1515-1566, 1986.        [ Links ]

LETSON, D., GOLLEHON, N. Confined animal production and the manure problem. Choices, Third Quarter, p.18-24, 1996.        [ Links ]

LOSINGER, W.C. Production of Finisher Pigs in the United States. Fort Collins: Colorado State University, 1997. (M.S. thesis).        [ Links ]

LOSINGER, W.C., BUSH, E.J., HILL, G.W. et al. Design and implementation of the United States National Animal Health Monitoring System 1995 National Swine Study. Prev. Vet. Med., v.34, p.147-159, 1998.        [ Links ]

LOSINGER, W.C., HEINRICHS, A.J. Management practices associated with high mortality among preweaned dairy heifers. J. Dairy Res., v.64, p.1-11, 1997.        [ Links ]

MADDALA, G.S., Introduction to Econometrics. New York: Macmillan Publishing Company, New York, 1988.        [ Links ]

MIDWEST PLAN SERVICE. Livestock Waste Facilities Handbook, Second Edition. Ames: Iowa State University, p.2.1-2.8, 1985.        [ Links ]

MUIRHEAD, M.R. Veterinary problems of intensive pig husbandry. Vet. Rec., v.99, p.288-292, 1976.        [ Links ]

MUIRHEAD, M.R. Factors affecting efficient growth rate in the feeding pig. Vet. Annual, v.29, p.103-109, 1989.        [ Links ]

NETER, J., WASSERMAN, W. Applied Linear Statistical Models. Homewood, Illinois: Richard D. Irwin, 1974.        [ Links ]

NERRIE, B.L., HATCH, L.U., ENGLE, C.R. et al. The economics of intensifying catfish production: a production function analysis. J. World Aquac. Soc., v.21, p.216-224, 1990.        [ Links ]

ROKA, F.M., HOAG, D.L. Manure value and liveweight swine decisions. J. Agric. Appl. Econ., v.28, p.193-202, 1996.        [ Links ]

SAS User’s Guide, Version 6, 4.ed, Cary, North Carolina: SAS Institute Inc., 1989. v.2.        [ Links ]

SAS Procedures Guide, Version 6, 3.ed. Cary, North Carolina: SAS Institute Inc., 1990.        [ Links ]

SHARMA, K.R., LEUNG, P.S., ZALESKI, H.M. Economic analysis of size and feed type of swine production in Hawaii. Swine Hlth Prod., v.5, p.103-110, 1997.        [ Links ]

SUKHATME, P.V., SUKHATME, B.V. Sampling Theory of Surveys with Applications. Ames: Iowa State University Press, 1970.        [ Links ]

USDA 1995. Swine ‘95 Part I: Reference of 1995 Swine Management Practices. N186.995. Fort Collins, Colorado: United States Department of Agriculture: Animal and Plant Health Inspection Service: Veterinary Services, Center for Epidemiology and Animal Health, 1995.        [ Links ]

VAN ARSDALL, R.N., NELSON, K.E. Economies of size in hog production. United States Department of Agriculture Economic Research Service. Technical Bulletin Number 1712. United States Department of Agriculture, Washington, D.C., 1985.        [ Links ]

WESTERNBARGER, D.A., LETSON, D. Livestock and poultry waste control costs. Choices, Second Quarter, p.27-30, 1995.        [ Links ]

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License