In a previous blog post, I criticized a study by the economics consulting firm NERA purporting to measure the costs companies would face as the Dodd-Frank reform of the OTC derivative markets is implemented. NERA is working on behalf of a group of energy companies lobbying to avoid some of the law’s mandates. Last week, NERA filed a “Briefing Note” with the CFTC specifically addressing my criticism and explaining the reasoning that leads them to stand by their original numbers.
What is at issue? Dodd-Frank forces companies to margin swap trades that previously could be executed without margins. Does this impose extra costs on those companies? If so, how large are these costs?
NERA calculated the cost per dollar of margin as the difference between (i) the cost of capital the company must pay to the capital markets to obtain the money needed for the margin account and (ii) the rate of return earned on a typical margin account. For (i) they used the 13.08% Weighted Average Cost of Capital (WACC) of the companies on whose behalf they are doing this lobbying work. For (ii) they estimated a rate of 3.49%, yielding a shortfall of 9.59%.
I criticized NERA for using the WACC as a measure of what it would cost to fund the margin account. A WACC reflects the risk of all of a company’s many assets, including many very risky assets. The right cost of capital to use in the calculation is one that is particular to the riskiness of the margin account. Funds in the margin account are invested in safe assets, and the right cost of capital will be much less than a company’s WACC.
NERA’s “Briefing Note” agrees with the principle that the right cost of capital is particular to the riskiness of the margin account. But they object to my claim that since the margin funds are invested in relatively safe assets such as Treasurys, therefore one should use a low cost of capital. They insist that the margin account is an indissoluble part of the swaps business, which is very risky. They write that,
Professor Parsons overlooks the risk the value of the collateral could change – a risk that stems from the fact that it is used to secure the ultimate settlement of inherently volatile energy swaps transactions. Price swings in oil and natural gas markets can be substantial. Natural gas swap prices over the past five years have ranged from below $3.00 per MMBTU to over $13 per MMBTU. In 2008, the Brent Crude oil price fell from over $140 per barrel to under $40 per barrel. Electric power swaps can exhibit even greater price volatility, as that commodity cannot be stored. Such price swings trigger exposures and variation margin that, under the CFTC’s proposed rules, could exceed the notional value of the swap.
Indeed, NERA doubles down, insisting that the WACC is a conservative number that is probably too low because the swaps business is among the riskiest of the company’s activities.
That quote from NERA’s Briefing is astonishing. The NERA authors are confounding different parts of the company’s operations. They erroneously attribute to the collateral account the risk that flows from the uncollateralized swap exposure. It isn’t “the value of the collateral” that changes when the price of natural gas or oil fluctuates. It is the value of the uncollateralized swap. The value of the collateral is unchanged. The collateral is there and available to be debited to cover the loss earned on the uncollateralized swap. Fluctuations in the balance of the margin account reflect deposits and withdrawals by the company. These fluctuations do not reflect a risk that contributes to the cost of capital any more than the pattern of my monthly contributions to my personal 401k plan contribute to the risk of the stocks bought with those contributions.
By the logic of NERA’s Briefing Note, the risk of the particular securities in the margin account is irrelevant to the cost of capital for the margin account. That cost of capital is derived exclusively from the riskiness of the swaps business with which the margin account is associated. If NERA’s logic were valid, then there would be no reason for the margin account to be invested in Treasurys. Why forego the extra return from alternative investments if the risk on the account derives from the fact that they are part of the swaps business?
NERA’s insistence that the margin account is an indissoluble part of the risky swaps business is beside the point. So long as one is consistent in the application of this principle, my original criticism of their calculation still holds. But NERA is not consistent in the application of this principle. When selecting the cost of capital they insist on looking at the swaps business in its entirety. But when evaluating the income earned, they look narrowly at the margin account in isolation. This inconsistency is a key element driving NERA’s exaggerated cost from the margin mandate. The high cost of capital they attribute to the margin account is due to the risks flowing from the swaps account. A consistent calculation, whether done for the margin account alone as my original blog post had proposed, or done for the combined swap+margin accounts, shows that NERA’s calculated cost of the margin mandate is hugely exaggerated.
For those who prefer the concreteness of numbers, here is an illustrative example, calibrated to the key figures in NERA’s original report:
We start with the hypothetical company SwapcoPre, which conducted its energy swap business in the pre-Dodd-Frank era. There was no margin mandate. Some of its swaps were margined, others were margined only in certain circumstances, and others were not margined at all. I bundle everything together and call this the “swap account”. The table below shows the key figures. The company, which was 100% equity financed, had total capital invested of $2,350 million. It earned a rate of profit on this business of 13.08%. I use the standard CAPM measure of the riskiness of the business, and ascribe to this business a Beta of 0.63. Assume a risk-free rate of 5.82%, and an expected market return of 17.32%, so that the market risk premium was 11.50%. Then SwapcoPre’s cost of capital would have been 13.08%, exactly the rate being earned. I purposefully set the differential between the rate of profit earned and the cost of capital equal to zero. This zero baseline simplifies all later comparisons when we evaluate the impact of mandated margin.
The table below shows some information on a similar company, SwapcoPost, which conducts its energy swap business in the post-Dodd-Frank era. The company’s swaps business is exactly the same as for SwapcoPre: it has $2,350 million invested and carries swaps with the exact same notional exposure. Therefore, the first line of the box below contains all of the same data as for the box above. But, there is now a margin mandate that compels the company to hold additional margin in order to run this same notional exposure. The amount of additional margin is $235 million, and information on this account is itemized in the second line of the box below. This additional margin is invested in riskless securities with a Beta of 0, implying a cost of capital of 5.82%. I have assumed that these securities generate cash earnings at a rate of 3.49%, which is below their own cost of capital. This differential is necessary in order to bake in a cost from mandating higher margins. The company’s total invested capital is $2,585 million. Its total Beta is 0.57, which is lower than SwapcoPre’s Beta due to the low risk investments in the margin account. The company’s overall cost of capital is 12.42%, which is lower than SwapcoPre’s cost of capital.
The simplest way to measure the cost of the margin mandate is to take the differential between the cost of capital on the margin account and the rate earned on the margin account, and multiply this differential times the incremental capital invested in margin:
The annual cost of $5.47 million is the straightforward consequence of investing an incremental $235 million in an account that only earns 3.49%, but that has a cost of capital of 5.82%.
What happens if we follow NERA’s insistence that the swaps business is indissoluble, and look to the cost of capital on SwapcoPost’s entire swaps business? So long as we apply this principle consistently, we arrive at the same answer.
NERA’s inconsistent methodology for calculating the cost of the margin mandate is shown in the following box:
Instead of using the 5.82% cost of capital assumed for the margin account alone, they insist on using the 13.08% cost of capital for the non-margined swaps business at SwapcoPre. However, to calculate the income earned, they use the rate for the margin account alone and not the rate for the swaps business. This is the fundamental inconsistency of their position. The differential of 9.59% is exaggerated because they mix and match the entities for which they get their data. For one line item they look to the swap account, while for the other item they look to the margin account, and they take the difference of these two unrelated numbers That’s why NERA’s inconsistent methodology produces a much larger annual cost of $22.54 million, approximately 4 times the cost calculated using the correct methodology.
My original blog post characterized NERA’s faulty use of the WACC as “one big error” in their calculation. It was hardly the only one. NERA purports to calculate the cost of posting the additional margin, but never asks what cost may have been paid to trade swaps without margin. NERA simply takes for granted that, absent a collateral requirement, companies are able to trade derivatives at no cost to their balance sheet. My colleague, Antonio Mello, and I have repeatedly highlighted this as the main weakness of the several studies on this issue commissioned by lobbyists seeking to loosen the regulations–see here and here. Our critique of the earlier Keybridge study applies equally to NERA’s study:
If you don’t back up your derivative trades with a cash collateral account, then you are backing them up with a promise that you are good for it, i.e. with credit. Companies have limited debt capacity, so using credit is costly, too. A regulation that requires using cash instead of credit costs the company on one side, but loosens its constraints on the other. The net effect on the company’s free cash flow is zero. Keybridge’s oversight here is a first order mistake. One could argue that the cash requirement is costlier than credit, but then you would have to figure out by how much. That would be an extra, very difficult step in the calculation, and any reasonable estimate for the differential would drive the headline number down enormously, possibly to zero.
NERA’s study, too, makes a first order mistake on this point, as well as many others.