The Sarbanes-Oxley Act has cost public companies millions, if not billions, of dollars in extra costs since the epic corporate meltdowns in 2000 and 2001. It has required a great deal of time and effort in the implementation of internal controls and processes in the hopes that these steps would prevent the high profile abuses of Enron and other companies. Energy trading, in particular, has been criticized highly for its abuses.
Before Sarbanes-Oxley, mark-to-market (MtM) processes included procedures that aggregated trader input and external market data in the forward curve-building process. However, the instances where traders took advantage of their position in the market to influence prices have created an environment where all trader insight is suspect. Sarbanes-Oxley rules and the audit firms have encouraged organizations to build processes for forward curve-building and MtM that circumvent trader insight altogether. Extensive controls and processes around data gathering, storage, and processing have occurred in this controlled environment.
There is an endless debate around whether this law has had a positive or negative impact on public companies. Certainly, the emphasis on proper controls and processes needed to be addressed after so many well-respected companies were caught abusing the existing system. But, one must ask: What is the total cost of Sarbanes-Oxley, and is it worth it?
A recent survey issued by Financial Executives International (FEI) quantifies some of these costs. The respondents of their survey (172 “accelerate filers,” companies with market capitalizations of over $75 million) indicated that they spent an average of $2.9 million on Sarbanes-Oxley in 2006, which was a decrease from $3.8 million and $4.5 million in 2005 and 2004, respectively. At the same time, average financial statement audit fees increased from $4.1 million to $4.4 million from 2005 to 2006.
Those numbers appear surprising, necessitating a closer look at the energy trading sector. These organizations typically take on additional risk through gas and power trading and are therefore susceptible to the spectacular derivative trading debacles that make for good news reporting.
It turns out that the largest energy merchants have had to double the fees paid to their financial statement auditor since 2001. Table 1 shows the annual audit fees paid by 25 of the largest investor-owned utilities and natural-gas traders increased from $5.4 million to $10.7 million between 2001 and 2006.
In general, professional service fees are only one portion of the total cost of supporting Sarbanes-Oxley. Internal staff time can be as significant as the external costs. The FEI survey addresses this fact. It notes that internal costs amount to 31 percent of the total cost of compliance (including audit fees). We can use these proportions to add nearly $5 million of implied internal costs to the documented audit fees to estimate the average annual cost for Sarbanes-Oxley at $15.5 million per year for the typical energy trader.
While these costs are significant, a potentially higher cost is embedded in how the typical integrated energy merchant has had to change its behavior and possibly become less effective in its core operations.
These unintended consequences of Sarbanes-Oxley compliance could far exceed the specific costs of additional audit procedures and internal compliance staff. Energy merchants make most of their revenue on the assets they operate. Investor-owned utilities own generation plants, pipelines, and gas storage. Oil companies own refineries. These assets operate at varying degrees of financial efficiency depending on the operator’s understanding of the markets. Depending on the accuracy of price forecasts over the operating period, a great deal of opportunity exists for additional profits.
In fact, one primary argument for taking on the risk of a sophisticated energy trading function is that it increases the price discovery for the entire organization. If the traders are active in the market, they know where the markets expect prices to be over the next few months based on where the forward curve is trading. They also know the level of volatility in the market based on the implied volatilities derived from the options trading. Likewise, if traders are knowledgeable about their organizations’ assets, they have a better understanding of the supply and demand fundamentals in their markets and can derive a better forecast of these markets.
Unfortunately, there is a disturbing trend within these organizations. Traders have abused their position in an attempt to prop up their own profitability by providing market prices that were not a reflection of the market as much as they were an attractive sliver of additional profitability on their own position. Sure, traders have been doing this forever. Some have gone to jail for it, but that has not stopped others from trying to accomplish the same goal. The reaction from the Sarbanes-Oxley crowd has been to build a “Chinese wall” between traders and the MtM function. Many have gone so far as to exclude any trader insight from the curves generated for MtM purposes.
Given the history of trader abuse, this seems like a prudent move. Go out to the brokers, get their quotes, and use them to mark the books to market. Put all sorts of controls around this process, make it independent from the front office, and build IT infrastructure around it. Then, archive the data. You now have a stable, well-controlled process and a database full of data.
The problem is, this process and the resulting data ignores what the company knows about the market’s fundamentals. The data set is completely external and does not take advantage of the idiosyncratic position of the integrated energy merchant. Furthermore, all the energy put into creating this process and maintaining the data leaves little time to create a second set of prices that can be used for internal asset planning and operations. Typically, the company may recognize the flaw, but have no capacity to change it. The result is that the asset operators have to settle for the external curves in their planning processes. They miss out on the additional profits gained by having a better view of the markets and operating their assets more effectively as a result.
There is a better way. Instead of excluding trader forecasts altogether, organizations simply should evaluate these forecasts analytically and identify and correct for systematic trends of over or under forecasting across time. Consider an example.
Suppose a risk-control function is in place that generates MtM values based on curves derived from external sources only. This risk function also generates value, earnings, and cash flow at risk metrics derived from these external curves. Part of this process includes the simulation of spot prices into the future. These spot-price scenarios are key inputs into all of the risk metrics. For example, Figure 1 illustrates a set of simulated spot-price paths for Henry Hub (HH) natural gas. This figure includes each individual price path (the thin multicolored lines) and the daily percentiles for 5, 50, and 95 percent (the bold red lines).
Sarbanes-Oxley created the incentive to put in processes that would generate this information. It is invaluable for MtM accounting and risk reporting. However, it explicitly excludes the trader’s perceptions of market prices in the future. Any organization that is integrated in the market, owning assets across the value chain, could have a better view of the direction of the market based on supply and demand fundamentals that they contribute to through operating their own assets. This information often is available to the traders, which gives them a competitive advantage when trading against banks and hedge funds. However, for this information to flow back to the organization, it needs to be integrated into the price forecasting, forward-curve building, and risk-measurement infrastructure. If Sarbanes-Oxley has created a process that explicitly prohibits the trader’s participation, the entire organization will pay through decreased efficiency in operations.
Suppose the traders in this organization have a specific view of the average price in July 2007 that is different from what the market is telling risk control? Figure 2 provides the histogram of average prices in July 2007 derived from the information provided to Risk Control. The vertical blue line represents the expected price of $8.09/MMBtu that is also the executable forward price in the market. However, in this instance, the trader has a view that this price is higher than expected given knowledge of the production, pipelines, and storage assets that affect the market. The trader’s view of the market leads to an expectation of an average price for July 2007 of $6.92/MMBtu as seen by the vertical green line in Figure 2. As a trader, one would expect to sell forward contracts at $8.09 and then expect to settle those commitments at the expected price, netting $1.17/MMBtu.
In the pre-Sarbanes-Oxley era, this trader could have executed the trade at $8.09 and then reported the expected price of $6.92 and generated an MtM gain and the end of the next reporting period. Furthermore, the price forecast would have fed into the planning infrastructure throughout the organization, and the asset operators would have optimized their operations based on that forecast. However, now that the trader’s insight is excluded from the process, one would have to wait until the trade settles. If the trader is accurate, profits still would be made as expected. However, the rest of the organization would fail to benefit from this superior insight. If any of the asset operators that rely on natural-gas prices made an operating decision based on the curves derived from risk control, they would have operated inefficiently even when an alternative forecast is better.
The counter argument that the trader is biased and does not consistently have a better view of the market can be controlled by setting up systematic back-casting tests that constantly evaluate the trader’s forecasts over time. If the distribution around that forecast is tighter than that of the historically derived risk-control forecasts, then the organization can be confident in the accuracy of the trader’s curve and include it in the operation of its assets.
Figure 3 compares an illustrative distribution around the trader forecast compared with the risk-control derived histogram from Figure 2. In this example, a very small portion (the area shaded in red) of the trader’s distribution is higher than the expected price derived from risk control. If this area is small, the organization can conclude with a high degree of confidence that market prices likely will fall below current levels.
Sarbanes-Oxley has been expensive for energy merchants. It has required the redesign of processes, the addition of controls, and an overall decrease in operational and strategic flexibility. Internal processes have been redesigned with the thought of risk and control and not with operational efficiency and shareholder-value optimization. Controls and processes are positive as long as they do not cripple the organization.
Communication between traders and the rest of the organization should not be prohibited. Traders should simply be monitored with an analytic framework that is constantly evaluating their effectiveness. In this way, the organization will make both shareholder and the auditor happy.