Firstable, yes, it's a better way to have statistical anaysis, based on montecarlo simulation to have predictions on your forecast sales, The most difficult issue is to get reliable data for your input for model selection to apply in your NPVs and IRRs prediciton.

Hi Jose,

Yes, i will be using a monte carlo simulation for my NPVs and IRRs respective prob distributions. That is also my concern in the first place, the reliable input data will be dependent on my correct assumptions on the environmental factors like distribution of future market shares, price behaviour, technological improvement as well as production efficiency of the enterprise. Maybe forming a model to capture the price, costs and production dynamics by incorporating the environmental factors.

If you have the time to enter seriously into the stochastic risk measurement of proyects of any kind, it would do you good to check Mr Cesari's book

http://www.amazon.com/gp/product/3642044530/ref=cm_li_v_cr_self?tag=linkedin-20

Hi Hernan,

Yes, I am really interested to pursue this research idea. Is my framework of analysis is a doable one?

I guess, as far as your risk producing assets are properly defined and simulated, a realistic exposure measure is possible. However, neither of that trivial even for experienced analysts. The suggested book requires though some good command of financial math. If you think you qualify, I would strongly encourage you to read it.

You can do it, but there is a possiblity of different result by npv and irr.

Until the moment in that the indicators are inferred, it is supposed that the flows of funds of the project know each other sufficiently and it is for this reason that all the analyses go to the mensuration of the financial merit of the project with the purpose of determining their acceptance or rejection. However, although the obtained results contribute certain bases for the taking of a reasonable decision, it is indispensable to consider another situation where the flows of funds are subject to the influence of factors of uncertainty and risk.

Before this event, the analyst won't be able to have precise information about the behavior of many of the variables, so one will have to appeal to the artificial generation of series of data of costs, revenues and future benefits, with the purpose of determining the relative probability of obtaining a certain yield of the investment. This method is known as Stochastic Analysis of Risk and for it execution the technique of Simulation of Montecarlo it is used.

Notice: The decision of carrying out a capital investment requires precise information on the flows of funds, during the life of the project, since the estimate of the profitability of the investment depends on its behavior, the one which, in general, it is uncertain. With this it is meant that all analysis of risk suffers of two inconveniences. That is:

1) The inference of the "better estimate", or, said otherwise, of the "more probable prospective Value", be this way product of an infinite number of feigned replicaciones, it doesn't guarantee that those values agree with the values that will be obtained in the real life.

2) Today in day, technique doesn't exist some by means of which you can determine the probability that the project generates earnings or produce losses of money.

Rather, there is not form, well-known, that allows to measure the real risk associated to a capital investment neither, neither, the call financial risk of the project.

Although a vision a so much pessimist of the matter has been presented, since the justification of using the method of the calculation of probabilities in the analysis of the risk obeys the application of the “law of the big numbers” that doesn't only require of a very big number of observations, but rather it demands countless intents of the experiment, when the project evaluations are unique events that don't repeat, in the practice and in absence of methods deterministic, one will have to depend on the calculation of subjective probabilities, by means of which the percentage is determined in that, probably, the result of the project will spread to reach the values inferred in the studies.

SIMULATION DE MONTECARLO

It is a method by means of which a series of values is generated that it should complete the condition of randomness and to possess

the properties probabilistic of the distribution type that it is specified, with the purpose of taking it as random sample of a population that, supposedly, it can exist in the real life.

To carry out a simulation analysis, the first step is to build an analytic model that represents the situation of decision that is wanted to analyze. In this case it is an equation to infer the VPN with base in the variables Value of the Investment, Value of the Expenses and Value of the Revenues. The second step consists on defining a distribution of probability, based on historical or subjective data, for each factor of present uncertainty in the pattern.

And the third step is to generate random sample results with the distribution of defined probability for each factor.

This way, the multiple repetition of this sampling process leads to the obtaining of a distribution of frequencies of those results and, this way, it is hoped to find the value of the VPN whose relative frequency is the biggest, being this value the one that has been given in calling the "better estimate" or the "more probable".

In conclusion, you can use the resulting distribution of frequencies to infer conclusions of type probabilistic.

For a better understanding of this topic, it is explained that the conclusions of type probabilistic are similar to those that they are deduced by means of opinion polls on a population's random samples.

Hi

If there is any other different approach than monto carlo ? I am currently in Phase of analysis risk

I have very interesting and novel developments that can be useful for their work. Please write me to the email [email protected] and I will send him the document.

Greetings,

I use instead of probabilistic models an neural Network, considering external factors to build a forecast.

First define your variables independent and dependent, you can consider variables as market index or inflation rate, any kind, define your Sales o any variable as output., if you have some fuzzy variables you need to transform those to some numerical values and add categories or change the model to fuzzy logic.

Train the network and You will have a better estimate, remember depending on the conditions you can define the variables that is playing a role in the output, define many historical data as you can get, the neural Network will weight each factor to have an impact on the result. I use to simulate with a back propagation model... If you want more information pleas send me an e mail.

@ Dalal A.S.

There is another simulation tool, the bayesian approach. You can try study this.

@ Jaime

Thanks, will send you an email.

I am sending you an analysis that I did for a stochastic modeling class where we took a look at two porfolios to determine which one to select based on issues like net present value, cash flow, etc. In a second answer, I will send you the spreadsheet that I developed for the analysis.

The spreadsheet is enclosed below.

@ Eric Hall, Thank you for the materials you share. Will read it.

I have developed the "risk sensitive value measure method" for the evaluation of investments.

@ Yoshio, is it possible if you can share your method to us? I would love to read your material.

Please see "Risk-Sensitive Value Measure Method for Projects Evaluation",

Yoshio Miyahara

Journal of Real Options and Strategy 01/2010; Vol. 3,(No. 2,):185-204

Rudy, I understand your problem is calculating the Expected Value of your Project's Future Cash Flows. In real situations, you can have many Sources of Risk (Stochasticity) in that Cash Flow and practical evaluation can drift easily into hardship, beyond what is normally explained in the basic industrial text books.

For example: Price can be Stochastic, Sales Volume can be Stochastic, either or both Variables can depend on more than one Risk Source, Price or Volume can be dependent on their Value at Prior Periods, there can be Correlation among Variables, etc.

Normally, to face honorably these Complex Calculations, you are required to defining some simplifications to what I mention above. But this is a one-by-one case task and requires more information about the Variables at stake. If you explain me little further, about how your Price and Volume evolve or relate, I can be of better help.

Rgds., Hernan

Hernan, thank you for your answer. Since i posted this question i have reviewed some materials already and have clearer understanding of my modeling problem. In fact i already tried some technique but still i find the result unsatisfactory. I wanted to have a simpler approach so that I can relate the results and analysis to my audience without some degree of complexity so they may appreciate my effort.

I will be conducting again an industry study, mostly agriculture farmers (cooperatives, Small and medium agricultural enterprises. This is to identify their financing needs and how to proceed with the financing agreement with them. Basically part of our framework of analysis will be to study their profitability level and try to find parameters that we can use to evaluate if they are eligible for a financing program.

Can you suggest some pointers on how to conduct the profitability analysis of the farmers? assuming inputs to the financial analysis will be price and volume of their produce.

Rudy, I'll try to give you more hints in some hours. I'm in the middle of something now. Please contact me at [email protected]. Send me your e-mail so we can talk. Rgds., Hernan

It seems we are speaking of 3 random processes here:

1. production volume "V" (risk drivers: climate/location, soil/location, human-resource/location: there must be large statistic data here)

2. price "P" (risk drivers: local or international market exchange prices: there must also be good data here) and

3. internal capacity "IC" of producers to afford debt (assets, capital + liability) (there is research data here as well)

First thing is to get the data. Second thing is analyzing correlation and trying to build up models. V has some interesting correlations with P through local-market and location. V has important correlations with IC through investment capital.

I suspect, than rather than calculating expected value of a future cash flow, here you are faced with calculating the expectation of a default during the term of the credit. This is 2 very different analyses, Rudy.

dear the sensitivity analysis in todays world may not be right as the sensitivity factors are too volatile hence sensitivity analysis should be used along with trend analysis of each such factors and than use the moving average of such factors. then apply adjusted moving average. I have used this in one of my corporate consultancy project 3 years back and it is proving good.

My preference is to use Monte Carlo simulation. I build a model of the business (in Excel using Simetar(c)) with a complete annual income statement, cash flow and balance sheet for 10 years. Then I simulate the model with stochastic input and output prices and production. If it is an agribusiness firm the stochastic production can come from weather induced risk on crop yields for the farm or for the local market area where ag inputs are purchased. After the 500+iterations of the model I get an empirical estimate of the probability distribution for net present value. If the NPV is positive it means the rate of return on equity is greater than the discount rate and if the discount rate is the owner's opportunity cost of capital it indicates that the business is an economic success. Because simulation gives me the full distribution for NPV I can calculate the probability that NPV is positive which gives me the probability of the business being an economic success. Next I focus on the probability that the business has a positive cash flow each year. Again the simulation model provides probability distributions for each year so it is just a matter of checking the probability that cash flows are positive in the empirical probability distributions. I provide a couple of references to articles I have written that demonstrate this approach. The oldest article was my first and best application of including risk into an investment decision. It was an analysis of a business my family was considering when I was a graduate student.

Simulation of a base scenario is educational but to really be useful to business we should provide them analyses of alternative scenarios for managing the business. This is where Monte Carlo simulation is most useful. I run multiple scenarios for a business or for multiple investments, say in a portfolio or different ways to make a renewable fuel. The simulation models generates an NPV probability distribution for each scenario and I rank them using stochastic dominance with respect to a function (SDRF) or better yet I rank the scenarios using stochastic efficiency with respect to a function (SERF). The SERF method does not require that we know the decision makers' risk aversion coefficient, just whether he/she is slightly, normal, moderate, or extremely risk averse. I include references to SERF.

I build risk models in Excel and use the add-in Simetar(c) you can learn more about Simetar at www.simetar.com. It is a very powerful econometric (regression, times series, exponential smoothing), forecasting, simulation, risk ranking (SERF and SDRF), validation (6 statistical validation tests), and report preparation tools that seamlessly works inside Excel. It simulates more than 50 univariate and multivariate distributions and includes several copulas for multivariate distributions. I use Simetar in my two graduate risk courses.

I record all of the lectures and labs for my Masters level course and you can download and watch my lectures, see my labs, etc. for free at http://www.afpc.tamu.edu/courses/622/ I had some technical difficulties with the recordings for some of the 2013 lectures so watch the 2012 lectures if the 2013 lecture has no sound. The course syllabus is on line so you can determine which lecture or lab you are interested in rather than watch all of them.

I am writing chapters for two books and have attached them as well.

Richardson, J.W. and H.P. Mapp, Jr. “Use of Probabilistic Cash Flows in Analyzing Investments Under Conditions of Risk and Uncertainty.” Southern Journal of Agricultural Economics, 8(December 1976): 19 24.

Richardson, J.W., B. K. Herbst, J. L. Outlaw, and R. C. Gill, II. “Including Risk in Economic Feasibility Analyses: The Case of Ethanol in Texas.” Journal of AgriBusiness, 25,2 (Fall 2007): 115-132.

Hardaker, J.B., J.W. Richardson, G. Lien, and K.D. Schumann. “Stochastic Efficiency Analysis With Risk Aversion Bounds: A Simplified Approach.” The Australian Journal of Agricultural and Resource Economics, 48:2, 2004, pp. 253-270.

Richardson, J. W., B. Herbst, A. Duncan, M. D. Besten, and P. V. Boven. “Location Preference for Risk Averse Dutch Dairy Farmers Immigrating to the United States.” Journal of Agricultural and Applied Economics. 39,3(Dec 2007): 735-748.

Dear James Richardson,

If you get the full distribution for NPV, then you can apply my evaluation method of "risk sensitive value measure method" to your problem.

Dear Mr. Richardson,

I am grateful of your generosity in providing us your valuable inputs. I am sure it will be a good read. Will check your site and read relevant materials. Will also try to research your publications. Thank you very much.

Hi,

You may have a look to this paper:

Changing Risk Perception and the Time-Varying Price of Risk.

By: Füss, Roland; Gehrig, Thomas; Rindler, Philipp B. Review of Finance. Jul2016, Vol. 20 Issue 4, p1549-1585. 37p. DOI: 10.1093/rof/rfv046.

Best regards