BUS 552 Module 4 Monte Carlo Analysis Example

Ten Thousand Possible Plants: A Monte Carlo Simulation of an Insect Protein Startup's First Commercial Facility

Student Name

MBA Program, Aspen University

BUS 552: Innovative Finance and Venture Capital

Instructor Name

Month Day, Year

Ten Thousand Possible Plants: A Monte Carlo Simulation of an Insect Protein Startup's First Commercial Facility

Bluegrass BioProtein, a composite startup in Louisville, Kentucky, raises black soldier fly larvae on pre-consumer food waste from grocery distributors and food processors, then dries and grinds the larvae into a protein meal sold as an ingredient for poultry and aquaculture feed. After two years running a pilot facility, the company is seeking $22 million from venture investors and a state economic development program to build its first commercial plant, processing about 30,000 tons of food waste a year. Investors want to know how much the plant is worth and how risky it is. This paper uses Monte Carlo simulation to answer.

The Base Case

The company's financial model uses single best estimates for each input: the plant runs at 80% of capacity, sells protein meal at $1,900 a ton, collects $30 a ton in tipping fees for accepting food waste, pays expected energy and labor costs and costs $22 million to build. Over a 15-year life, discounted at 12%, the base case net present value is $6.1 million.

Why One Forecast Is Not Enough

Hertz (1964) argued that investment decisions based on single best estimates hide the uncertainty in each input and can lead managers to accept projects that are much riskier than they appear. He proposed simulating many combinations of possible values, drawn from probability distributions, to produce a range of outcomes with their likelihoods. The technique itself traces to scientists working on physics problems in the 1940s; Metropolis and Ulam (1949) described the Monte Carlo method as using random sampling to solve problems too complex for direct calculation.

Savage (2009) named a related trap the flaw of averages: feeding typical values into a model seldom yields the typical result when the model bends. When a project's results respond unevenly to its inputs, for example when low utilization produces losses that high utilization cannot fully offset, the result calculated from average inputs differs from the average of the results.

Input Distributions

Meal price and utilization were modeled as independent after the team found no clear link in pilot data, a simplification noted below.

Results

The model was run for 10,000 trials, each drawing a value for every input and computing net present value.

The mean is $2.7 million below the base case, illustrating Savage's point. Construction overruns are skewed upward, low utilization hurts more than high utilization helps because fixed costs must be paid regardless, and these asymmetries pull the average down. The wide range from the 5th to the 95th percentile and a one-in-three chance of losing money show that the project is far riskier than the single forecast implied.

What Drives the Risk

A sensitivity ranking, measuring how much each input's variation contributes to the variation in net present value, shows the order of importance: meal price contributes about 48% of the variation, utilization about 27%, construction cost about 15%, tipping fees about 6% and energy about 4%. Most of the risk comes from what the meal sells for and how fully the plant runs.

How the Model Was Built

The simulation was built in a spreadsheet with a simulation add-in. Each trial draws a value for the five inputs, runs the 15-year cash flow model and records the net present value. The construction cost draw affects the initial outlay and depreciation; utilization affects both revenue and variable costs, while fixed costs for labor, insurance and maintenance remain constant across trials. Running 10,000 trials took a few minutes, and repeating the run with a different random seed changed the mean by less than $100,000 and the probability of loss by less than one point, suggesting the results are stable.

Reading the Distribution

The shape of the distribution matters as much as its center. Its left tail is longer than its right: the worst outcomes, in which construction overruns combine with low utilization and weak prices, lose more than the best outcomes gain. Investors care about this asymmetry because a venture fund can absorb a failed investment, but a startup's first plant that loses money may be its last. For the founders, the left tail describes the risk of losing the company itself.

Limits of the Model

The simulation is only as good as its distributions, which rest partly on judgment. Treating meal price and utilization as independent may understate risk if weak feed demand lowers both at once. The 12% discount rate, used for each trial, reflects the cost of capital rather than the project's full risk, which the distribution itself describes.

What Investors Should Require

The results point to specific conditions. First, the company should secure an offtake agreement with a feed manufacturer for at least half its output at a floor price, cutting the largest source of risk. Second, investors should stage their commitment, funding detailed engineering and a fixed-price construction bid before the full $22 million. Third, the company should sign multi-year waste supply contracts with minimum tipping fees. With an offtake floor of $1,700 a ton on half its output, rerunning the simulation reduces the probability of loss to about 19%.

Conclusion

The base case of $6.1 million suggested a sound investment. Monte Carlo simulation, following the logic Hertz described and correcting the flaw of averages, shows a lower expected value, a wide range and a one-in-three chance of loss. Its most useful product is the sensitivity ranking, which tells investors that an offtake agreement on meal price would remove more risk than any other step.

References

Hertz, D. B. (1964). Risk analysis in capital investment. Harvard Business Review, 42(1), 95-106.

Metropolis, N., & Ulam, S. (1949). The Monte Carlo method. Journal of the American Statistical Association, 44(247), 335-341. https://doi.org/10.1080/01621459.1949.10483310

Savage, S. L. (2009). The flaw of averages: Why we underestimate risk in the face of uncertainty. Wiley.