Introduction
In 1938 the Mexican government nationalized all foreign and domestic oil companies to create Pemex, the Mexican state-owned petroleum company. Today, they are Mexico’s largest employer, and the number five oil company in the world by sales. They pay almost 100% of their profits as taxes, and contribute about one third of all taxes collected in Mexico. This high tax rate has put a heavy strain on Pemex. Pemex CEO Emilio Lozoya believes Mexico will require almost 60 billion per year in investment to maximize production in deep water reserves, Pemex has only been able to invest 23.9 billion per year (2012). In coming years there are plans to invest another 10 billion per year.
PEMEX suffers from many of the problems associated with most monopolies: they are inefficient and do not provide timely service. Despite its size and status as an oil producer, Pemex’s productivity is low, and has had major delays and budget overruns in its refineries as a state monopoly. As a result, Mexico currently imports half of its gasoline, mostly from the United States. To improve productivity, obtain the necessary investments, and increase the tax revenues from Mexican oil, the government has begun instituting reforms.
In addition to reforms on the taxation of Pemex, Mexico has plans to open the market to foreign competitors and allow foreign investment in oil production. Part of Mexico’s reforms are to bring Pemex’s productivity up to international oil company standards. These reforms impact Pemex in two ways:
- They must improve their efficiency in delivering to customers in order to show improvement.
- Foreign oil companies are going to be able to deliver oil products to their customers in Mexico.
The second item is important because it means a loss of revenues for delivered gasoline and diesel fuels. In 2015, the price for gasoline is set at 3.50/gallon, though that will float in the future. Lower oil prices have pushed gasoline futures prices to around 2.00/gallon[1]. For every gallon of gasoline not delivered, Pemex will lose out on 1.50 minus delivery costs.
In order to improve efficiency and customer service Pemex has decided to invest in a truck scheduling system. Pemex has been investigating the decision since 2009, and must now decide which software to buy, how to finance it, and how to implement it. The aim of this study is to help Pemex make the decision on purchasing, implementation, and the most important factors that will affect the success of the decision.
By investing in new truck scheduling software, PEMEX hopes to decrease its delivery costs, improve the throughput at the terminals, and improve customer service. But new software and a new project will cost PEMEX several million dollars and may not provide the benefits they were hoping for. Additionally, there may be some markets that do not justify the investment, and the price of oil may further impact the outcomes.
Frame
The decision-maker for this purchase is the sub-director for operations. His frame is to select a software package that is able to improve the use of resources, make on-time deliveries, and avoid the costs of an unsuccessful project. To that end, there are two objectives: the first is to improve the operational efficiency and on-time delivery of all products. The second is to avoid a project failure. Though a failed project will mean there were not enough benefits to justify the ongoing cost of the project, a failed project will also put Pemex in a bad light and generate negative publicity.
Influence Factors
In evaluating Pemex’s frame, there is an important goal, which is to maximize the NPV of the project. This takes into account the costs of the project as well as the savings. Because the project is also politically charged, there is a second element, which is avoiding failure. If the project is discontinued during the time horizon, then it is considered a failure and an additional penalty to the NPV is assessed. This is the cost of answering questions from politicians and the press regarding the project, finding a new vendor, and taking management focus away from other priorities.
The objective is to find the best purchase and implementation strategy that maximizes the failure-adjusted NPV.
| Factor | Description |
|---|---|
| Delivery Volume | The delivery volume is the at-risk volume delivered from the each terminal to the customers |
| Overtime Costs | Overtime costs are the costs spend on overtime. In a typical week Pemex schedules deliveries Monday – Friday. On Saturdays, Pemex terminals must open again for make-up deliveries. As a result they incur overtime expenses for their drivers, pump operators, and other personnel required to load a truck and make a delivery. |
| Operational Costs | Operational costs reflect the costs of energy and administration required to run a terminal. |
| Pilot Costs | This is the cost to develop, integrate, and test the scheduling software in the pilot location. |
| Implementation Costs | These are the costs to implement in each additional location after the pilot. |
| Operational Cost Savings | These are percentage savings of the operational costs. This reflects a savings due to not having to reschedule deliveries and not having to have machines operating for some or all of Saturday deliveries. |
| Cost of Failure | This is the political cost of failure. It reflects the costs of negative publicity and redirection of management attention. |
| Software Costs | This is the cost of the software as reflected in both up-front and ongoing charges. |
| Market Growth | Market growth is the volume growth in market demand. |
| Years to Maximum Recovery | The years to maximum recovery reflect the amount of time improved service will take to recapture lost customer deliveries. |
| Overtime Cost Savings | These are the overtime costs saved by operating more efficiently. |
| Years to Maximum Loss | The years to maximum loss reflect the eventual migration of customers to new vendors, either due to better service, lower costs, or some other effect. |
| Market Share Loss | The market share loss reflects the percentage of the delivery volume that will go to competitors as they enter the market. |
| Profit Margin | The profit margin reflects the difference in the cost to purchase fuels and deliver fuels to the sales price. |
| Market Share Recovery | The market share recovery is the percentage of the lost market share that Pemex is able to recover through improved customer service. |
Alternatives
Pemex needs to decide how to purchase and how to implement the software in all 77 of their terminals. They have already decided to make the purchase and to implement a pilot terminal.
In considering a software implementation, Pemex needs to make decisions regarding both the software purchase, and the pace of implementation. A typical software purchase involves paying an up-front amount for software, and every year afterwards, maintenance. Maintenance usually runs between 18 and 25% of the software cost.
Alternatively, a software company may offer to finance the software purchase. Instead of paying the license costs up front, the costs are spread out over two to three years with a low rate of interest. Software companies find they have a higher probability of closing deals by offering financing, so they readily finance purchases and provide attractive rates.
As a purchasing strategy, Pemex can decide to make an outright purchase or finance over two years. With an outright purchase, Pemex will pay 100% of the licensing costs in year 1, and 20% thereafter. If the licenses are financed, Pemex pays 50% of the license cost in year 1. In year two, they begin to pay maintenance and pay the other 50% of the software cost. A third alternative is to split the software purchase over multiple years. The software required for the pilot program is only a fraction of the total software required. By splitting the software purchase over multiple years, Pemex can delay payments for the software. Software vendors prefer to sell all the software at once, and will discount to do so. By splitting the software purchase over 3 years, the software vendors will be less inclined to volume price. In this strategy, the percentages for maintenance are increased in order to reflect a higher cost of software.
The benefits of each strategy depend on various uncertainties. The financing options determine the size and timing of the cash flows. The conservative purchase gives the flexibility to stop implementing if the savings don’t materialize. This could take place in the event of a failed implementation or defective product, or merely because another product provides more value. On the other hand, up-front and financed payments reduce the effect of inflation on future payments.
In considering implementations, Pemex needs to decide on the speed at which it will roll out the software to other terminals after the pilot program ends. A fast rollout can be led by the vendor. They will be able to implement quickly, but generally at a higher price. The other two strategies increase the participation of Pemex in the rollout. These strategies decrease in cost per terminal, but carry the higher uncertainty of cost. This Pemex will be able to save, but may run into some issues which will increase the costs. The more aggressive implementation strategies accelerate the benefits of the software.
The purchase and implementation alternatives are grouped into three distinct strategies. The strategies can highlight the advantages and disadvantages of accelerating or slowing down the purchase and implementation decisions. By analyzing these strategies, it is possible to learn the elements that have a higher chance of leading to success and which benefits to focus on in the purchase and implementation of software.
Deterministic Analysis
In order to understand the components that with affect the decision, there is an excel model that calculates the net present value of the cash flows of this project. Since there will be cash flows regardless of whether a project is implemented, there is a fourth strategy called the momentum strategy. This strategy does not purchase nor implement any software. It will lead to a loss in market share, but will not incur political failure costs, software and implementation costs, or reap any of the market recovery or cost savings.
The calculations of the deterministic model are the following:
The sales are based upon the market size and market share. Market size grows based on the original market size and the growth rate. From the market size, the market share loss due to competition is calculated. The market share recovered based on the timing and quantity of implemented terminals recovers a percentage of the total volume. The adjusted sales volume is multiplied by the margin to get the final sales dollar amount.
The software costs are based on the up-front software cost multiplied by the percentage paid in a given year. For years past the fifth year, the annual payment is assumed to be in maintenance mode and the fifth-year software cost is propagated for the rest of the horizon.
The operational costs and operational savings are a function of the cumulative number of terminals implemented the previous year. In the first year, there are no savings.
The continue variable is a variable that determines if the project is to continue. This determination is made in the third year. It is there to assess whether the pilot resulted in enough savings to justify the ongoing maintenance costs. If the recovered market share (partial at period 2), operational and overtime cost savings projected to all 77 terminals is higher than the software maintenance costs, the project continues. If the project does not continue, it is viewed as a failure. In a failed project, all implementations, software payments, benefits, and software recovery are set to zero and the political cost is incurred in the NPV.
The operating profits include adjusted revenues, subtract the net operating and overtime costs, and subtract the software an implementation costs. These operating profits are discounted to the present. To this discounted sum of cash flows, the political cost is added is the project is discontinued and viewed as a failure.
The following is an analysis of the deterministic model strategies. All values are at their base case estimate except implementation costs. These are set at their highest levels for the Vendor Led strategy, at the base case for the Combined strategy, and at the low level for the Conservative strategy.
| Strategy | Base Case NPV | Sources of Value |
|---|---|---|
| Momentum | 421.2MM | |
| Vendor Led | 456.7MM | Market share recovery, accelerated overtime and operational cost savings. |
| Combined | 454.4MM | Deferred expenses, market share recovery, operational and overtime savings. |
| Conservative | 452.5MM | Deferred expenses, market share recovery, operational and overtime savings. |
The value of the Vendor Led strategy breaks down as follows. In the regulated market, the present value of the fuel distribution system is 507.3MM. Once the market is deregulated, Pemex stands to lose 86.1MM in value due to market share losses. The software and services implementation will cost another 4.4MM. The operational and overtime savings resulting from the software implementation are 3.8MM. The cost savings almost make up for the implementation costs alone. The largest benefit from the project is the market share recovery. This adds 36.1MM to the strategy value.
The waterfall diagram showing the migration from the vendor led strategy to the financed strategy reveals several insights. The timing of the software costs only adds 60,000 in value. It will help to delay costs, but this will not drive noticeable value. Reducing the per-terminal cost for implementation will add an extra 1.03MM in NPV. Finally, deferring the benefits for 38 terminals for one year will reduce the total value by 3.44MM. This leads to the insight that software financing is not an important driver of value, while implementation costs and timing are more important drivers of value.
Probabilistic Analysis
With the deterministic analysis complete, the next step is to do a probabilistic analysis. In the deterministic analysis, most variables are set at their base case. In probabilistic analysis, the values of the inputs are allowed to vary. The first step in probabilistic analysis is to complete a sensitivity analysis of the inputs. In this sensitivity analysis, the values of the inputs are switched between their minimum and maximum values in turn. Some inputs will swing the results by orders of magnitude, while others will have almost no impact.
Because the value of the business is always positive, the sensitivity analysis is conducted using a relative NPV value. This relative NPV is in comparison to the Momentum strategy. With every combination of inputs, the Momentum strategy will have an NPV, which will vary with some of the parameters. The sensitivity analysis is done in comparison to the Momentum NPV. The tornado diagram below shows the variability of the NPV with the input parameters set at each of their extremes.
With the exception of there being no market share loss, the tornado diagram shows that it is always better to implement the Vendor Led strategy. 98% of the variability is explained with the market share loss, the sales margin, and the mitigated loss. From project success perspective, this means that saving costs on overtime and operations will not be very important. In order to stem market share loss, Pemex would need to move politically to keep competition out of Mexico. Leftist opposition to profit-sharing contracts and uncertainty surrounding government levies on foreign oil producers have so far kept foreign oil companies from entering Mexico. Though this seems to be effective, allying the interests of anti-capitalists will likely lead to a clash as other investments that result in greater efficiencies for Pemex clash with leftist policies. It is also unwise to encourage regulatory chaos as this can also work against Pemex when it decides to make future investments. Improving the sales margin means lowering the costs of exploration, production, and distribution. This project is able to help control at least a portion of this. Finally, loss mitigation is also influenced by the results of this project. By improving on-time deliveries and customer communications, Pemex can reduce the chance customers flee to other suppliers.
Though this sensitivity analysis is able to show the important causes of variability, none of these situation results in a project failure. Even in the one case where the NPV was slightly negative, the ongoing savings were greater than the ongoing software maintenance costs, and the results could not be considered a disaster. It takes several factors at once to create a situation with disastrous results. In order to explore this situation, probabilistic analysis continues with Monte Carlo simulation.
In Monte Carlo simulation each of the input parameters is allowed to vary randomly. With enough simulations, it is possible to plot a distribution of the NPV. The distribution shows the most aggressive strategy, the Vendor Led implementation to have the highest average NPV. In general, all three strategies share roughly the same probability distribution, but with the more conservative strategies having smaller extremes in terms of both losses and gains.
| Strategy | Expected NPV |
|---|---|
| Vendor Led | 46.3MM |
| Combined | 43.0MM |
| Conservative | 40.6MM |
It would help slightly to defer the largest portion of the software purchase to the second year, but as discussed above, it is not a major contributor to value. No one strategy is able to avert disaster, so the timing of the software and services costs, which were originally a small factor, only avoid small expenses, not major ones.
Appraisal
Given that no strategy offered a way to avoid major losses, and aggressive implementation is likely to provide the best results, it is beneficial to focus on the value of control. In the Probabilistic Analysis section, the two controllable inputs were the profit margin and the market share recovery. These two factors are not completely controllable, but determining the value of control is useful in ensuring the project implementation in these two areas will ensure overall project success.
In the case of both margin control, and loss mitigation control, the strategies are much more profitable. Under the Vendor Led strategy, the mean NPV is 46.3MM. The mean increases to 92.0MM under margin control, giving a value of control of 45.7 . Finally, under loss mitigation control, the mean NPV is 86.7MM, giving a value of control of 40.4. Adding control in these key variables still does not avert product disaster. In analyzing the scenarios where there was a product disaster, the major contributors were high disaster costs, which are triggered when there is no market share loss, and no savings in operational or overtime costs. This additional analysis means that even though mitigating costs is not a key driver to the overall NPV of the project, ensuring that the project saves enough costs to pay for the maintenance will keep the project from being considered a disaster and reduce the probability of a negative NPV from 17.87% to 5.86%. In contrast, the probability of a negative NPV when the margin is controlled is still 15.57%. The NPV when the operational savings and overtime savings are under control has a mean NPV of 52.2MM, giving a combined value of control of 5.9MM. As mentioned before, the value of controlling the savings rates will not improve the mean NPV very much.
The important lessons in this decision analysis for Pemex are that an aggressive implementation strategy is going to give the best results. This aggressive implementation strategy accelerates the benefits which are expected to exceed the costs of a rapid implementation. In order to improve the results of the investment, the most important controllable factors are the product margin and the loss mitigation. This means that the focus for Pemex should be to wring out costs from their supply chain to improve profitability, and to improve customer service to keep as many customers as it can. On the loss mitigation side, controlling the cost savings reduced the probability of a loss by just over 12%. Though it is not crucial, ensuring that the project is successful in eliminating operational and overtime costs will ensure at least a mild success.
[1] Price on Bloomberg.com/energy on May 13, 2015.