Ethanol Production and Gasoline Prices: A Spurious Correlation.

• Author: Knittel, Christopher R.

"As a result of our biofuel industries, consumers across America are paying about $0.90, on average, less for gas than they would otherwise pay. So, it's a great opportunity for consumer choice, it's a job creator, and it improves income opportunities for farmers." --Secretary of Agriculture Thomas Vilsack, 10/24/11. 1. INTRODUCTION

The median American household spent over 8 percent of its income on gasoline in 2011. Gasoline price fluctuations therefore significantly affect household budgets, and government policies that affect gasoline prices resonate widely. The most prominent recent policy has been to promote the use of ethanol as an ingredient in gasoline. This year, 10 percent of finished motor gasoline in the United States will be comprised of ethanol made from corn, up from 3 percent in 2003. The main forms of government support have been explicit subsidies through the Volumetric Ethanol Excise Tax Credit (VEETC) and even larger implicit subsidies through such policies as the Renewable Fuel Standard and state-level blend mandates. (1) The benefits of ethanol over gasoline are that it diversifies our fuel mix, can have lower emissions, and increases farmer and landowner wealth. An additional potential benefit is that it may relieve gasoline refining capacity constraints during peak demand periods; this would in turn lead to lower gasoline prices.

The national trade association for the U.S. ethanol industry, the Renewable Fuel Association (RFA), recently launched an advertising campaign claiming ethanol production lowered gasoline prices by 89 cents in 2010 and $1.09 in 2011 (see Figures 1 through 3). The U.S. Secretary of Agriculture Thomas Vilsack has endorsed these claims in several speeches (see the opening quote of this paper for one example). These estimates are derived from regression models that use monthly regional data to estimate the relationship between ethanol production and the profit margin for oil refiners in text. (2) Thus, the effect comes not from a reduction in world crude oil prices, but rather from a decline in the relative price of gasoline to crude oil.

Given the importance of these estimates, we investigate their robustness. We show that they are driven by implausible economic assumptions and spurious statistical correlations. Put simply, the empirical results merely reflect the fact that ethanol production increased during the sample period whereas the ratio of gasoline to crude oil prices decreased. These trends make the empirical analysis extremely sensitive to model specification; however, we find that empirical models that are most consistent with economic theory suggest effects that are near zero and statistically insignificant.

Several other papers have addressed the effect of ethanol policy on gasoline prices. In contrast to Du and Hayes, none of them focused on the oil refining margin. McPhail (2011) estimates, using a structural VAR, that ethanol demand shocks have no effect on oil prices in the first few months, but a significant negative effect after a year. McPhail identifies this effect from ethanol price shocks by assuming that short-run (i.e., within-month) ethanol supply is perfectly inelastic. She does not report an effect on gasoline separately from that on crude oil. Serra et al. (2011) fit a smooth-transition vector error correction model to the U.S. corn-ethanol-oil-gasoline price system and find that ethanol price increases Granger cause oil price increases. Thus, the model predicts that a policy-induced increase in demand for ethanol would slightly increase gasoline prices. Drabik (2012) argues using a simulation model that the expansion of ethanol production has reduced gasoline prices by $0.05-$0.10 per gallon. Rajagopal et al. (2007) and DeGorter and Just (2009)) report similarly small effects using similar simulation methods.

Because ethanol production increased smoothly during the sample period, statistical analysis with this variable is fraught with danger. It is strongly correlated with any trending variable. To illustrate this point, we take the same empirical models in Du and Hayes (2011) and Du and Hayes (2012) and use them to "explain" variables that have no material relationship to U.S. ethanol production: the U.S. price of natural gas and unemployment rates in the U.S. and the European Union. Our resulting estimates suggest that increases in ethanol production "cause" reductions in natural gas prices but increases in unemployment. The estimates imply that, had we eliminated ethanol in 2010, natural gas prices would have risen by 65 percent and unemployment would have dropped by 60 percent in the U.S., 12 percent in the EU, and 42 percent in the UK. To further underscore this point, we provide a silly example. Again, using the same empirical models in Du and Hayes (2011) and Du and Hayes (2012), we show that ethanol production "causes" our children to age. Obviously, anyone using these models to advocate eliminating ethanol production to end the Great Recession or make children age more slowly would be greeted by extreme skepticism. We encourage similar skepticism about the estimated effect of ethanol on gasoline prices generated from these models.

The remainder of the paper is organized as follows. Section 2 discusses the economics of how ethanol production may influence gasoline prices. Understanding these basic economic concepts puts useful bounds on the effect. Section 3 discusses how these basic concepts can guide the choice of the empirical model. In Section 4 we discuss the empirical models we employ. The data are discussed in Section 5. Section 6 reports the estimated results from the models used in Du and Hayes and alternative specifications. Section 7 offers some concluding remarks.

2. BASIC ECONOMICS OF OIL REFINING

   We begin with a basic discussion of how ethanol production might influence gasoline prices. In doing so we discuss the channels through which this is possible and stress the difference between short-run effects--those that might last one or two months--and long-run effects--those price effects that can be sustained in the industry. Simple economic calculations allow us to place loose bounds on the impact ethanol production has on the price of gasoline in both the short and long run.

   The largest component of the price of gasoline is the cost associated with crude oil. Oil refineries produce numerous products from each barrel of crude, but we can obtain a rough approximation to the contribution of the cost of crude to the price of gasoline by allocating it proportionally to each product. For example, when oil is $100 per barrel (i.e., $2.40 per gallon), approximately $2.40 of the price of a gallon of gasoline could be assigned to the cost of the crude in that gallon. (3) Ethanol production has a minimal impact on the price of crude oil. In the world market for crude oil, an individual country's supply and demand decisions are small relative to the market as a whole--even for a country the size of the U.S. To put this into perspective, the U.S. consumes roughly 20 percent of world oil. Roughly half of the U.S. oil consumption goes toward gasoline, and ethanol comprises roughly 10 percent of our gasoline-blend fuel. Thus, on a volu-metric basis, U.S. ethanol constitutes about 1 percent of world oil use. However, ethanol has 33.3 percent less energy than gasoline and thus engines require more ethanol than gasoline to go the same distance. So, U.S. ethanol replaces just 0.67 percent of world oil. Crude-oil supply and demand would need to be very inelastic before such a quantity had a noticeable effect on price (see Rajagopal et al. (2007) and DeGorter and Just (2009)).

   In analyzing the oil price spike of 2007-08, Hamilton (2009) argues that a short-run demand elasticity of -0.06 and a short-run supply elasticity of zero fit the observed price patterns. This demand parameter is consistent with that estimated by Hughes et al. (2008) using regressions with time series data. However, the demand elasticity is much larger over longer horizons because consumers adapt to high prices by driving less and buying more fuel-efficient vehicles. For example, Hamilton notes a large switch in consumer's purchases from SUVs to smaller cars when oil prices peaked in summer 2008. This adaptation was followed by a steep decline in prices, a decline that accelerated when the financial crisis hit world markets several months later. Thus, although a -0.06 demand elasticity implies that removing 0.67% of global crude oil demand would decrease oil prices by 11%, this effect only applies in the short run. Over a year or longer the demand elasticity is much greater and suppliers would respond by changing production. It is difficult to estimate reliable long-run elasticities, and we are unaware of such estimates, but some introspection is useful. If the long-run supply elasticity were 0.25 and the long-run demand elasticity -0.25, then the effect of U.S. ethanol production on world crude oil prices would be 0.67/(0.25--0.25) = 1.33% or about $0.03 per gallon on $100 oil. We see these elasticities as likely to be too small and so we view 1.33% as an upper bound of the effect.

   Ethanol production may affect gasoline prices through other channels, however. Retail gasoline prices typically exceed crude oil prices by $0.70-$1.20 per gallon, although this price spread can spike much higher for short periods of time. About 45 cents of this premium represents state and federal taxes and the remainder is the margin associated with the refining and transportation of gasoline. Du and Hayes focus on the refining margin. They estimate the relationship between ethanol production and two measures of the refining margin: the crack spread and crack ratio. The crack spread equals the weighted average price of the two main refined products (gasoline and distillate fuel oil) minus the price of crude oil. Du and Hayes define the crack ratio as the price of...