[BETA] Fortnightly

Forecasts Send ROEs Wide of the Mark

In a recent "Offpeak" ("Forecasting is Just That," Jan. 1, 1996, p. 54), David Foti and Clay Denton report data showing the percentage of error found in various seven-year forecasts of natural gas prices (1988-94) produced by the American Gas Association (A.G.A.), Energy Information Administration (EIA), DRI/McGraw-Hill (DRI), Gas Research Institute, and WEFA Group. These errors ranged from approximately 50 to 95 percent.

What does that say about utility ratemaking, where regulators sometimes rely on a 20-year horizon?

Until recently, the Federal Energy Regulatory Commission (FERC) has calculated just and reasonable rate of return on equity (ROE) using the traditional Discounted Cash Flow (DCF) model, which posits that investors require an ROE equal to the sum of the expected yield for the next period, plus the anticipated (by investors) long-term growth rate (g) in cash flow (dividends). The simplified constant-growth version of this model is: r = D1/Po + g, where Po is an estimate of the current price of the stock, D1 denotes the next-period expected dividend, and g indicates the expected annual growth rate in dividends. The FERC estimates the next-period expected dividend by adjusting the indicated dividend for one half-year of expected growth. The adjustment factor = 1 + g/2.

For many years, the FERC determined the requisite growth term g using financial forecasts (Value Line, Zack's, IBES, and others) of the rate of earnings growth for natural gas pipelines or electric utilities. It then converted these earnings-growth forecasts to

estimated growth rates for dividends through the model relationship g = b x r, where b represents the forecasted rate of earnings on book value and r marks the retention ratio. However, in a line of recent cases, the FERC has virtually abandoned its traditional use of the DCF model. (See, Ozark Gas Trans. Sys., 68 FERC ¶ 61,032 [1994]; NW Pipeline Corp., 71 FERC ¶ 61,253 [1995]; Panhandle Eastern Pipe Line Corp., 71 FERC ¶ 61,228 [1995]; Williston Basin Interstate Pipeline Co., 72 FERC ¶ 61,074 [1995].)

In calculating the g term, the FERC now skips over the step that converts earnings estimates to dividend estimates. Instead, the FERC relies directly on earnings-growth rate estimates, as though they were equivalent to the dividend-growth rates required by the DCF model. Further, with respect to these earnings-growth rate estimates, the FERC has decreed that "a projection limited to five years, with no evidence of what is anticipated beyond that point, is not consistent with the DCF model and cannot be relied on in a DCF analysis" (Ozark at ¶ 61,105).

In Ozark, FERC staff employed 20-year forecasts of natural gas consumption prepared by DRI/ McGraw-Hill as a proxy for the long-term earnings-growth rate for interstate natural gas pipelines. The FERC rejected an initial determination by a skeptical administrative law judge (ALJ): "The DRI data have no demonstrated accuracy, are related only in a tangential sense to the gas pipeline industry, and are not relied on by investment advisory services." Instead, the FERC accepted the position of the staff witness that "reliance on DRI