Economic Feasibility of Large Modern Confined Swine Farrowing Operations in North Dakota

Annual livestock revenues in North Dakota significantly lag behind those reported in neighboring states such as Minnesota, South Dakota and Iowa. Expansions in corn and soybean acres, steady increases in per-acre yields and large-scale investments in grain processing facilities — such as ethanol and soybean crush plants — are contributing to a large local supply of livestock feedstuffs, including soybean meal, pelleted soybean hulls, feed corn and dried distillers grains (DDG). As a result of this increase in feed products, crop and livestock producers and other North Dakota agricultural stakeholders have expressed interest in understanding the economic feasibility of new, long-term investments in alternative modern livestock enterprises at the farm level.

To understand the economic feasibility of modern livestock production systems at the farm level, we evaluated the expected economic feasibility of three alternative large, modern, confined swine farrowing enterprises. The three operations vary in size in terms of the number of sow spaces. For this study, the three sizes of operations evaluated include 1,500-sow, 2,500-sow and 5,000-sow. Based on national trends, we assume that investors would likely be associated with existing firms located outside of North Dakota that are highly integrated (e.g., Christensen Farms, Cargill, etc.). However, results from the study would be valid for interested investors who currently reside in North Dakota. Note, however, that we do not consider the potential risk and corresponding costs associated with a total eradication of sows from a disease or virus for any of the three operation sizes. This is one of the key reasons why most farrowing operations are highly integrated.

The specific objectives of the study were (1) to determine the expected 20-year net present value (NPV), expected return on investment (ROI) and expected internal rate of return on investment (IRR) for the three alternative farrowing systems, and (2) to determine how sensitive the base-case values of NPV and IRR are to variations in input and output prices that vary higher and lower over time (i.e., 12-pound weaned pig, individual feed ingredient (e.g., corn, soybean meal, vitamins and mineral) prices); veterinary/health; fuel, repairs and utilities; interest rates on operating and investment capital; discount rates; quantity and price of labor; and transportation costs for cull sows, weaned pigs and replacement gilts.

Expected NPV on a $/litter of 12-pound weaned pig basis for the three operation sizes were found to be $8, $345 and $576/litter, respectively. ROI for the three operations was determined to be 2.3%, 35.8% and 58.1%, and the IRR was 28.7% and 51.7% for the 2,500-sow and 5,000-sow operations. IRR for the operation with 1,250 sows could not be computed because the value was less than zero. The expected NPV on a per-operation basis was found to be $43,926, $1,899,652 and $3,170,255 for the three operations, respectively. Results of the sensitivity analysis indicate that the probabilities of the 1,250-, 2,500- and 5,000-sow operations realizing an NPV above 0 over the 20-year investment period are 52.6%, 96.9% and 99.8%, respectively. This indicates that the two larger-sized operations are expected to be more economically feasible than the smaller operation.

Overall, the results suggest that the profitability associated with a long-term (20-year) investment in owning and operating large, modern, confined swine farrowing operations is expected to be economically feasible for North Dakota. Moreover, the results suggest that the expected profitability (i.e., NPV) of swine farrowing is expected to be greater for operations with a higher scale, which follows what has been observed over time in other states that have swine farrowing operations, such as Iowa, Minnesota and South Dakota.

Over the past 25 years, the U.S. swine industry has undergone significant structural changes (USDA-NASS, 2024). There has been a notable shift towards large, modern, confined operations — also referred to as confined animal feeding operations (CAFOs) — particularly for operations greater than 2,000 head spaces. In fact, between 1997 and 2022, producers with operations between 1,000-head and 2,000-head increased their share of production from 3.46% to 7.52%, and producers in operations with more than 5,000 hogs increased their share from 1.48% to 5.82% (USDA-NASS, 2024). During the same period, the total number of swine operations decreased by more than half, from 124,889 to 60,809, with much of the decrease realized by smaller operations with fewer than 1,000 head (USDA-NASS, 2024).

Compared to national and regional trends, the North Dakota swine industry has shown a distinct decline in the total number of hogs on farms over the past few decades (USDA-NASS, 2024). Between 1975 and 2023, the number of hogs on farms declined by 57.14% from 350,000 to 150,000. This change involves a 33.33% reduction in breeding hogs and a 62% reduction in market hogs. In contrast, during this same 47-year period, the neighboring states of South Dakota, Minnesota and Iowa realized significant increases in hogs on farms by 65%, 210% and 100%, respectively, and the hog industry at the national level realized an increase of 54% (USDA-NASS, 2024).

Several agricultural stakeholders in North Dakota, including crop and livestock producers, policymakers, commodity groups, agriculture lenders and rural community leaders, have expressed interest in developing and growing the swine industry in North Dakota for several reasons. First, there is a growing disparity between swine revenues in North Dakota and neighboring states. In 2023, neighboring states realized an increase in their gross income from swine production compared to 1988. Between 1988 and 2023, South Dakota, Minnesota and Iowa realized substantial increases of 189%, 313% and 287% in gross revenues, while North Dakota realized an increase of 48% during the same period (USDA-NASS, 2024).

A second reason is that expanding the swine industry would add value to the increased supply of locally produced coproduct animal feed sources (e.g., corn, dried distillers grains and soybean meal and hulls) that are currently being produced in the state via corn ethanol and soybean crush plants. Expansions in corn and soybean acres over the past two decades (mostly due to advances in seed technologies), steady increases in per-acre yields and large-scale investments in grain processing facilities — such as ethanol, corn sugar and soybean crush plants — are contributing to a sizeable local supply of livestock feedstuffs, including soybean meal, pelleted soybean hulls, feed corn and dried distillers grains (DDG). The case of two new soybean meal crush plants in North Dakota provides a good example. The two new soybean crush plants located in Spiritwood and Casselton produce a combined total of more than 2.2 million tons of soybean meal and approximately 150,000 tons of soybean hulls per year. Due to the lack of livestock enterprises that consume soybean meal and hulls (e.g., poultry, swine and dairy cows), the majority of these coproduct feedstuffs are being exported out of the state.

Expansion in swine would also provide much-needed economic activity in rural communities where expansion would occur. For instance, tax revenues from expansion would provide an influx of income that could be used to help renovate public infrastructure, such as schools, fire departments, hospitals and public roads and utilities. The development of a modern swine industry would also provide employment opportunities for underutilized rural labor; it would provide opportunities for people who have left their communities for work but wish to return, allowing them to be closer to family and friends. Moreover, the development of a swine industry would lead to increased private sector businesses directly serving swine operations, as well as businesses that indirectly support the swine industry. The indirect effects of this circulation of funds will also help improve rural economies.

Lastly, a number of potential investors — including integrated swine farrowing and finishing businesses located outside of the state — view North Dakota as a unique and ideal natural environment suited for farrowing and finishing hogs. The cooler temperatures (especially in the summers), wide open spaces and sparsely populated rural areas provide a biosecurity advantage that would lead to improvements in costs associated with morbidity and mortality rates (viruses, diseases, heat stress) that are becoming more common in states with large concentrations of hogs (e.g., Iowa, North Carolina, South Dakota and Minnesota). In addition, North Dakota has a very good transportation infrastructure (roads, highways, interstates and railways) and cheap, abundant, reliable energy resources necessary for supporting a large, modern, confined swine production industry.

At present, investment in large, modern, confined swine farrowing and finishing operations is being promoted primarily to North Dakota row crop farmers for the aforementioned reasons. However, information about the economic feasibility (e.g., expected profitability) of long-term investments in facilities, equipment and machinery, as well as the operation of a large, modern, confined hog finishing operation at the farm level, is lacking. This is likely impeding investment and adoption.

In response to this issue, we conducted a study that sought to determine (1) the expected 20-year net present value (NPV), expected return on investment (ROI) and expected internal rate of return on investment (IRR) for the three alternative swine farrowing systems; and (2) how sensitive the values of NPV and IRR are to variations in influential variables, such as weaned pig prices, feed prices (e.g., feed corn, soybean meal, vitamins and mineral), interest rates on operating and investment capital, quantity and price of labor and transportation costs of cull sows, weaned pigs and replacement gilts.

We utilized investment appraisal analysis (also known as capital budgeting) to evaluate the economic potential of large, modern, confined swine farrowing operations at the farm level. Using this method, we determined the expected profitability in the form of net present value (NPV) on a $/litter basis associated with three sizes of swine farrowing operations. These include 1,250-sow, 2,500-sow and 5,000-sow operations. We also calculated the expected return on investment (ROI) and the expected internal rate of return (IRR) associated with NPV for each operation. In addition, we calculated a number of other important results that are tied to the NPV ($/litter) for each operation, including NPV ($/operation/year), labor (hours/litter, hours/sow and hours/operation/year) and the number of employees per operation per year. Through sensitivity analysis, we calculated the probability of the NPV ($/litter) being greater than zero for each operation size.

Data for this study were obtained from various sources, including public databases (e.g., USDA, Federal Reserve Bank), university and Extension publications, member-only databases (e.g., Livestock Information Marketing Center) and public and private sector swine industry experts. Details about the data are included in sections that describe the specific assumptions used in the calculations below.

Several assumptions underpin the methodology and data used in the base-case economic analysis and its corresponding sensitivity analysis. We outline assumptions below into categories for the base-case analysis for (1) initial investment costs (land, facilities, machinery and equipment and breeding herd), (2) sources of revenue, (3) sources of annual production costs (i.e., feed, veterinarian/health, repairs, labor, transportation and others) and discount factor.

Assumptions used to calculate the initial investment for land and facilities:

1. We assume that an integrated swine farrowing company (e.g., Christensen Farms, Cargill, etc.) would make the initial investment and would own the land, facilities and breeding herd for the operation, which is typical for a swine farrowing operation in the U.S. They would also hire and manage employees needed to manage and conduct day to day operations of the swine farrowing operation.
2. We assume that in North Dakota, we would be promoting our comparative advantage of cooler summer climates that provide increased biosecurity compared to states with large populations of swine operations.
3. Because of biosecurity issues, we calculate initial investment costs for the three sizes of swine farrowing operations that are equipped with farrowing and gestation facilities and a nursery. That is, in North Dakota, we would specialize in producing weaned 12-pound pigs and 40-pound feeder pigs that would be transported to operations specializing in finishing pigs to their final slaughter target weights.
4. For this study, we only focus on the economic feasibility of producing 12-pound weaned pigs.
5. The facilities, machinery and equipment (i.e., the swine operation) associated with each sized operation are purchased new at the time of investment.
6. Land for the footprint for each sized operation is included in the investment at $5,000 per acre. Regardless of who owns the land, there is an opportunity cost associated with it.
7. Acres required for each sized operation is assumed to be 50, 100 and 150 acres for the 1,250-sow, 2,500-sow and 5,000-sow operations, respectively.
8. We assume that each operation will have all feedstuff delivered, so no land will be purchased to grow/produce feed.
9. To simplify the analysis, we assume each sized operation will be built prior to the first year of operation, so annual interest will start being paid back to the investment in year 1 of operation and continue over the 20-year life of the investment.
10. Initial investments for each sized operation are based on investment data obtained from Iowa State University Extension swine publications (Christensen and Schultz, 2004).
11. It was assumed that building each sized operation would be 20% more expensive in North Dakota than in Iowa, due to the cost of transporting materials for facilities, machinery and equipment over further distances and higher labor costs.
12. It was also assumed there would be 20% cost savings from efficiencies in scale moving from the smaller (1,250-sow) operation to the medium-sized (2,500-sow) operation and from medium-sized to the largest (5,000-sow) operation.
13. Based on assumption 8 above, the initial total investment for facilities is $3.4 million, $5.7 million and $9.1 million for the 1,250-sow, 2,500-sow and 5,000-sow operations, respectively.
14. We used straight-line depreciation and assumed an interest rate of 8% to calculate the fixed investment costs on a $/litter basis for the 20-year investment for each sized operation, resulting in investment (depreciation and interest) costs of $126, $105 and $84/litter (i.e., $346,501, $577,502 and $924,003/operation).
15. We assume that after eight years of operation (i.e., years 9 and 18), additional investments will be required to update the facilities with repairs and update obsolete technologies with new, advanced technologies (e.g., feed and water delivery systems, individual animal measurement, computer and software, etc.). We assume the investment in these upgrades will be incurred at the beginning of years 9 and 18 and will be calculated as 20% of the initial investment made prior to year 1.
16. We also assume a cost for taxes and insurance equal to 1% of total initial investment costs. These costs, plus annual interest, would be paid each year and are therefore included in the study as variable expenses.

Assumptions used to calculate investment costs of the breeding herd:

1. Investment into the breeding herd includes breeding costs for boar semen and replacement gilts.
2. We assume boar semen is $13/litter and replacement gilts are $155/head and that each sized operation requires 0.28 gilts per litter (Christensen and Schultz, 2024).
3. We assume the cost of interest and insurance on sows is equal to 10% of the price of the replacement gilt for 166 days, which is the time needed to produce one litter of nine weaned pigs.
4. Total investment cost for the breeding herd was found to be $59/litter for each of the three sized operations, or $162,154, $324,308 or $648,615 per operation for the 1,250-sow, 2,500-sow and 5,000-sow operations.

Assumptions used to calculate annual sources of revenue:

1. Three sources of revenue ($/litter) were assumed, including (1) the sale of one litter of 12-pound weaned pigs, (2) the sale of cull sows and (3) a swine manure credit from the sale of manure produced in the operation.
2. One litter of weaned pigs was assumed to be nine piglets (Christensen and Schultz, 2024).
3. Each sow for all three sized operations was assumed to produce 2.5 litters of weaned 12-pound pigs per year (Christensen and Schultz, 2024).

Assumptions used to calculate feed, vet/health, repairs, manure and marketing costs:

1. Average 10-year (2013-2024) prices were used for feed ingredients (feed corn, soybean meal, vitamins, mineral) in the base-case analysis (LMIC, 2024).
2. Base-case prices for feed processing and feed delivery; veterinary and health care products; swine manure application; fuel, repairs and utilities; and marketing were obtained from Iowa State University Extension publications (Christensen and Schultz, 2024).
3. Ten-year (2013-2024) average interest rates on operating capital were obtained from the Federal Reserve Bank of Saint Louis (FRSTL, 2024).
4. Quantities of inputs for feed (lbs/litter) were obtained from Iowa State University (Christensen and Schultz, 2024).

Assumptions associated with transportation costs:

1. Transportation expenses were calculated assuming the cost of transporting (1) weaned pigs, (2) cull sows and (3) replacement gilts using a $3.50/mile price for round trips using a truck and three-level pot-belly trailer with a total transport space of 400 square feet and a 50,000-pound weight limit.
2. Three products require transportation: (1) cull sows that we assume would be transported 405 miles from our hypothetical sow operation near the community of Spiritwood, North Dakota, to the pork processor in Sioux City, Iowa; (2) replacement gilts that would be transported approximately 150 miles from a gilt development facility in southern Minnesota; and (3) delivery of 12-pound weaned pigs from the sow operation near Spiritwood to finishing operations located approximately 100 miles. Distances reported were taken from Google Maps (2025).
3. We assumed round-trip distances of 810 miles for cull sows, 300 miles for replacement gilts and 200 miles for weaned feeder pigs.
4. We used pork quality assurance (PQA) recommendations for animal spacing per pig per load, including 7.69 square feet for cull sows, 5.48 square feet for replacement gilts and 0.75 square feet for 12-pound weaned pigs.

Assumptions associated with labor costs:

1. We utilized industry experts to calculate the cost of labor for the three operation sizes.
2. We used 12 (11.5 FTE) employees for our base-case labor cost for the 2,500-sow operation, which includes an operations manager, part-time assistant to do miscellaneous tasks (e.g., cooking, laundry, floating, etc.), farrowing lead (manage daily operations), three farrowing assistants (processing, power washing, floater), gestation lead (heat check), two gestation assistants (moving/feeding, heat check, gilt management) and three floaters managing miscellaneous tasks.
3. We assumed that employees are restricted to 40-hour workweeks and the swine operator prefers not to pay overtime. Because animals need labor daily, schedules are developed in a way that allows workers to trade off weekends and holidays to limit overtime.
4. We assumed a 20% efficiency of labor associated with scale. That is, the 1,250-sow operation would be 20% less efficient than the 2,500-sow operation, and the 5,000-sow operation would be 20% more efficient than the 2,500-sow operation.
5. Based on assumption 3 above, we assumed that the 1,250-sow, 2,500-sow and 5,000-sow operations would require 4.83, 4.35 and 3.91 hours of labor per litter.

The central challenge with assessing the economic feasibility of long-term investments is the difficulties associated with predicting future output and input prices. To do this for the three swine farrowing operations, we utilize various time-series forecasting techniques to place future values on those prices, with some more complex than others.

Certain inputs in our model do not tend to vary higher and lower over time, including vitamins and mineral, feed processing, feed delivery, veterinary/health, manure application and marketing, Instead, these operating inputs tend to increase over time by some market rate of inflation, so we do not have access to their price history. For these inputs, we assume a 2% rate of inflation, starting in year 2 and continuing through year 20. The year 1 price for these inputs was obtained from Christensen and Schultz (2024).

For the other inputs (feed corn; soybean meal; repairs, fuel and utilities; and interest rates) and outputs (weaned pigs, cull sows and the manure credit), prices are expected to fluctuate over time due to world supply and demand forces. For these inputs (and outputs), we have access to past 15-year (2010-2024) prices for each input and output for which we utilize different, more challenging time-series forecasting techniques to starting prices (year 1) and trend values over the remaining 19 years. Part of this technique required us to conduct econometric hypothesis testing for each variable to determine the time-series model that best fits the relationship between the mean and variance within the historical data.

In all cases, our data either fit one of three general time-series forecasting models, including (1) an autoregressive (AR) model, (2) a moving average (MA) model or (3) some combination of both models (ARMA). An AR model predicts the future prices based on past prices, an MA model predicts future prices based on past random shocks that affected a past price and the ARMA model forecasts future prices using a combination of past prices and past random shocks affecting a past price. There are a number of variations of each type of forecasting model; however, the point to note is that the AR, MA and ARMA models used in this study are well documented and accepted for use for many years, with early concepts going back to the 1920s and 1930s and more formally combining both (ARMA) in 1970.

The cost category that uses fuel is repairs, fuel and utilities. The cost for repairs and utilities do not tend to vary significantly over time, but fuel costs do. We have a 15-year history of fuel data, and we use that data with time-series forecasting to place a forecasted value on repairs, fuel and utilities for years 2 through 20.

Sensitivity analysis was conducted in tandem with the forecasting procedures to determine how sensitive the base-case net present values (NPVs) are to variations in input and output prices that fluctuate over time. This is the data for which we have a long-term (15-year) history, allowing us to calculate a distribution of data that enables us to determine a mean and standard deviation for each price variable. This provides the opportunity to conduct what is known as a Monte Carlo Integration (MCI). The idea is fairly straightforward.

First, we programmed a computer to generate one iteration of sample prices from each of the individual input and output price distributions described above for each of the 20 years of the investment period (one sample price for a weaned pig, one for a cull sow, one for a ton of feed corn, one for a ton of meal, one for fuel and one for an interest rate on operating capital). It then uses that sample of prices in tandem with the fixed prices that do not have variation but do have forecasts for each year, and uses them to calculate the corresponding average NPV (our primary measure of profit for the long-term investment). The computer then repeats/simulates this process for a large number of iterations. We used 5,000 iterations in this study. The results are an expected NPV. A small fraction of the 5,000 iterations provides really attractive input (low price) and output (high price) combinations, resulting in a very attractive NPV for those iterations. Conversely, a similar fraction of the 5,000 iterations results in unattractive input (high) and output (low) prices and a corresponding unattractive (negative NPV). The majority of iterations sample a mixed bag of input and output prices. In the end, we have an average computed from 5,000 iterations of all possible input and output price combinations, ranging from great to terrible.

To do the MCI, we utilized a special type of statistical/risk management software known as “@Risk” and a very fast computer.

Values for expected measures of profitability (NPV, ROI and IRR) and corresponding labor requirements for each operation size are reported in Table 1. Estimated values of NPV are $8, $345 and $576/litter for the 1,250-sow, 2,500-sow and 5,000-sow operations, respectively. The $/litter values adjusted on a $/operation/year basis result in NPV of $43,926, $1,899,652 and $3,170,255 for each operation size. Expected NPV is our primary measure of expected profitability of swine farrowing operations at the farm level and should be considered an expected net return to investors after accounting for the annual cost of land, labor and facility management. It is important to note, however, that additional expenses have not been included in the calculations due to a lack of data. These expenses include the cost of upfront planning and permitting, which usually takes a minimum of two years. These costs are often in the form of time (billable hours) provided by construction engineers, environmental engineers, attorneys, accountants, information and technology specialists. Other costs may include expenses associated with road maintenance that are not part of the land footprint of the facility/operation, but are county or township roads adjacent to the operation for which the integrator has agreed to help maintain once the farrowing operation is built and operational. Additional expenses for hired labor could also have a claim on NPV, such as partial stipends for housing and unforeseen overtime pay. These types of expenses can vary depending on location, but would indeed have a claim to the NPV reported here, which could reduce them significantly.

Table 1. Expected Net Present Value, Return on Investment, Internal Rate of Return and Measures of Labor by Size of Operation

NPV, ROI and IRR all increase with operation size, which is in line with industry trends over time. There are operations now that are being built that have 10,000 sow spaces.

It required 4.83, 4.53 and 3.91 hours of labor per litter produced for the 1,250-sow, 2,500-sow and 5,000-sow operations. This resulted in the need for an average of 6.39, 11.5 and 20.7 employees per operation per year. Note, even though the total employees rose with the size of the operation, the labor efficiency in terms of hours needed per litter declined with the size. By 20%, per our assumption, the labor hours needed to decrease with operation size.

The results of our Monte Carlo simulations that we used for the sensitivity analysis revealed that the probability that the NPV would be greater than zero was much greater (96.9% and 99.8%) for the two larger-sized operations compared with the smaller operation, which realized 52.6% NPV>0. This result also helps strengthen the rationale for why we observe new swine confined farrowing operations trending towards larger sizes, rather than smaller sizes.

The purpose of this study was to gain an understanding of the potential economic feasibility of investing and operating large, modern, confined swine farrowing operations in North Dakota. In the study, we evaluated three alternative-sized operations, including a 1,250-sow, 2,500-sow and 5,000-sow operations. The results suggest that integrated swine farrowing operations are in fact economically feasible, and larger-sized operations tended to be more feasible than smaller ones. These results support the growth in size trends of swine farrowing operations reported in published data at the national and regional levels.

Several limitations exist. First, the expected NPV and ROI reported in this study are likely higher than they should be due to an inability to place defendable estimates for some nonoperational expenses that would likely be incurred by investors for large farrowing operations. Some of these may include partial expenses for maintaining county/township roads, highways and other infrastructure; expenses for unexpected employee overtime due to an unexpected production event; partial compensation for employee housing; upfront costs for planning, zoning and permitting; and donations to community charities and events designed to improve goodwill.

In addition, it is possible that the costs of employee compensation (wages and fringe) used in the analysis are underestimating the true cost of labor. This is primarily due to the lack of public information available for more precise estimates of the quantity of full-time equivalent employees (FTEs) needed for each of the three operation sizes. Further research that addresses alternative estimates for FTEs and the effect on economic feasibility (NPV and ROI) is warranted.

Additionally, this study did not consider the economic feasibility of larger-scale farrowing operations. That is, we did not consider the economics of building and operating more than one large, modern, confined swine farrowing facility/operation on a single footprint in a single location. Multiple operations at the same location could provide production synergies with labor, equipment and machinery to achieve cost savings and greater profitability. Of course, operation with increased scale would require a greater amount of upfront investment capital land, facilities and breeding herds.

Christensen, T., and L. Schulz. 2024. “Livestock Enterprise Budgets for Iowa–2024.” Ag Decision Maker (B1-21), Ames, IA: Iowa State University Extension and Outreach. https://www.extension.iastate.edu/agdm/livestock/pdf/b1-21.pdf

Federal Reserve Bank of Saint Louis. 2024. FRB Rates-discount, fed funds, and primary credit. Found at:
https://www.extension.iastate.edu/agdm/livestock/pdf/b1-21.pdf

Google Maps. 2025. www.google.com/maps

U.S. Department of Agriculture, National Agricultural Statistics Service (USDA, NASS). 2024. Quick Stats Database. Found at: https://quickstats.nass.usda.gov

Biermacher, Bullock and Hossain received funding for this study from a Market Develop Program grant provided by the North Dakota Soybean Council. Biermacher also received funding from North Dakota State University Extension. We thank William Wilson for constructive input and feedback.