Complete pass through of the taxes is necessary but not sufficient to conclude that the policies operate effectively and efficiently. Section 1.4 presented an empirical framework for estimating pass through of taxes implicitly levied on ULSD through RIN and LCFS deficit obligations and found they are fully passed through to diesel prices. This raises the price of the petroleum product so that blenders and, if passed through to retail prices, consumers demand less of it. This suggests effectiveness of one prong of the two-pronged approach of these policies. The implicit taxes have made petroleum more expensive, but have the implicit subsidies made the alternatives cheaper? In this section, I shift focus to the second prong; pass through of bio-diesel subsidies from the RFS, LCFS, and CFP is estimated. Racks provide an ideal setting to study pass through because the marginal cost of producing the blended fuel is observed daily. The marginal revenue for the blender is the rack price and the marginal cost is the wet cost of the component fuels. Spot prices are a good measure of the true marginal cost because they reflect the cost of replacing a gallon of fuel on a given day. I use the previous day’s spot price when calculating the wet cost because that is the information available to rack participants on the day of the transaction. Let ? denote the city of the rack market, ? denote days, and ? denote the diesel blend. One important confounding factor is the Blender’s Tax Credit , which acts as an additional implicit subsidy for bio-diesel realized by the rack seller. The nature and timeline of the BTC is described in Section 1.1.4. As mentioned,cannabis grow equipment within the sample used in this paper, the BTC was in place some years and retroactively reinstated in others.

In years that it wasn’t in place, the market formed expectations around the likelihood it would be retroactively reinstated which led to risk-sharing contracts between rack sellers and buyers. These year-to-year changes directly affect the observed margin in and may be correlated with RIN prices. To see this, consider a scenario where the BTC is taken away and the market forms expectations around its reinstatement, and the bio-diesel producer and the blender form a 50/50 sharing contract. This means that implied subsidy pass through is identified from within-year and within-blend variation in rack margins. The identification strategy outlined above requires additional assumptions about how the BTC affects margins and RIN prices. The first assumption I make is that blenders and bio-diesel producers expect the tax credit to be reinstated with probability one throughout each year that it’s not in place. The other assumption is that sharing contracts are 50/50 split throughout the year. If either are violated, the resulting impacts on margins will be attributed to the RIN subsidy. These assumptions seem reasonable since the tax credit had already been retroactively reinstated three times prior to the beginning of my sample, in 2010, 2012, and 2014. Similarly, I assume that pass-through of the BTC to bio-diesel spot prices is complete in years when the BTC expired. Irwin , looking at a sample of bio-diesel prices from Iowa plants in the months before and after the BTC expired , suggests that it hadn’t been passed through in previous years. In the example above, if none of the BTC was passed through to bio-diesel spot prices, we would see no change to the bio-diesel cost and the observed margin, and an increase in the D4 RIN subsidy. In my sample, however, bio-diesel spot prices and rack margins do appear to respond to the BTC expiration in 2017, and the RIN subsidy remains constant . Similarly, in 2020 when the BTC was reinstated, observed rack margins fell, consistent with the retroactive BTC being completely passed through.

The RIN subsidy fell at the same time, but likely reflects the decline in ULSD prices rather than the BTC. In addition to the confounding effects of the BTC outlined above, anticipation of changes to the BTC may create similar issues. The spot price of B100 rose starkly at the end of 2016, which may have resulted from blenders purchasing and blending excess bio-diesel before the tax credit expired. A similar but more modest pattern emerged at the end of 2019 prior to the 2020 reinstatement. Therefore, I also include blend-specific dummy variables for these two anticipation periods for robustness. Results are not sensitive to the inclusion of these variables. Outside of the anticipation periods, I assume nothing about the BTC is changing within years. Like jet fuel, blended bio-diesel is nearly a perfectly substitute to petroleum diesel. Therefore, since the RIN tax is fully passed through to ULSD prices, B100 and ULSD prices should only differ by their net RIN obligation, which is 1.5 D4 RINs, if the subsidy is fully passed through at the wholesale level . Figure 11 plots the B100-ULSD spread in Chicago, Gulf Coast, and New York Harbor Barge and the D4 RIN price multiplied by 1.5. The two series are nearly identical outside of the years with the BTC in place, which is to be expected. Figure 11 also highlights the fact that SME bio-diesel is the marginal fuel for compliance in the D4 category, meaning that D4 RIN prices should reflect the marginal cost of D4 compliance.Table 4 presents estimates of short- and long-run RIN subsidy pass through for California, Oregon, and the rest of the U.S. – which consists of Dallas, Trenton, St. Louis, and Wood River. The first three columns utilize the full sample, while the last three drop observations in the RIN Shock Period outlined in Section 1.3.1. The long-run coefficients suggest regional heterogeneity of RIN subsidy pass through; using the full sample, only around 60 cents/gal are passed through on the West Coast compared to 95 cents/gal in ROUS. When dropping the RIN Shock Period, ROUS RIN subsidy pass-through falls to 77 cents/gallon. Sensitivity of results to inclusion of the RIN Shock Period are discussed later in this section.

Short-run estimates in California and Oregon are imprecise and not statistically different from zero. Columns 3 and 6 show that it takes more than one week for the cumulative pass through of the RIN subsidy to reach its long-run average in ROUS. Only 34 cents/gal are passed through one day after a shock to the RIN price, 21 cents/gal when dropping the RIN Price Shock Period.Table 4 highlights some of the regional heterogeneity in pass through of bio-diesel RIN subsidies, however, heterogeneity is present within the ROUS as well. Figure 12 presents point estimates and 95 percent confidence intervals of long-run pass through of the RIN subsidy for each region in the sample. Regions are presented in ascending order of long-run pass through rates using the full sample. Rates in California and Oregon are the lowest nationwide at about 60 cents/gal on average. In the ROUS, average pass-through rates are 0.8, 0.95, and 1.07 in the East Coast, Gulf Coast, and Midwest, respectively; however, the 95 percent confidence intervals include 1 for all three regions. The estimates in Figure 12 are robust to controlling for both 5 lags and 30 lags, except for California. In California, confidence intervals for the long-run RIN subsidy pass through fall to [0.22, 0.61] and [0.25, 0.67] for the full sample and dropping the RIN Price Shock period, respectively, when increasing the number of lags to 30 days. In both cases, the point estimates fall below 0.5, suggesting less than half of the RIN subsidy has been passed through in California. Despite the quantitative differences in results between the two specifications, the qualitative conclusions remain: the RIN subsidy pass through has only been partially passed through in the state.Long-run RIN subsidy pass through results are qualitatively different when ignoring the RIN Price Shock Period and the ordering of regions changes. Pass through in Oregon becomes very imprecise since CFP prices begin in 2017,vertical grow rack leaving a small sample once dropping the period from the analysis. The lowest levels of RIN subsidy pass through levels now occur in the East Coast, where only half is passed through on average and the upper bound of the 95 percent confidence interval lies below three quarters of complete pass through. Using the restricted sample, pass through in the Gulf Coast is 67 cents/gal on average and the confidence interval no longer includes complete pass through. Incomplete pass through in the Gulf Coast is economically significant, as previous studies have consistently found complete pass through of implicit gasoline taxes and ethanol subsidies from the RFS . One concern regarding the results from the Gulf and East Coast is the effect of the Colonial pipeline shutdown in May of 2021 in response to ransomware attack. 21 The Colonial pipeline runs from Texas to New Jersey supplies a substantial amount of fuel to both Dallas and Trenton. The pipeline shutdown on May 7th, 2021, and continued operation on May 13th, 2021. Estimates for the two cities served by the pipeline aren’t sensitive to the inclusion of a blend specific dummy for the month of March in 2021, therefore I don’t control for the event moving forward and differences between the results from the full sample and dropping the RIN Price Shock period shouldn’t be attributed to the shutdown. Another concern about the results presented in Table 4 and Figure 12 is that blend offerings vary across regions , which raises the question of whether or not I am attributing differences in pass through among blends to regional differences. The portfolio of bio-diesel blends exhibits similar characteristics to ethanol, in that there are lower-percentage blends that are commonly used by retail consumers around the U.S. and higher blends that are only used in certain types of engines and have limited availability nationwide. Previous literature is mixed in its findings regarding high- vs low-blend RIN pass through.

A body of work has demonstrated lower pass-through rates of RIN subsidies for E85, gasoline with 85 percent ethanol, than the more common blend with less ethanol content, E10 . This work generally finds E85 pass through is incomplete. However, more recent work has found that it had been completely passed through . To test for heterogeneity in the pass through of RIN subsidies across blends, I estimate separately for each blend in each region.22 The long-run coefficients from those regressions are depicted in Figure 13, showing that long-run RIN subsidy pass through is generally consistent across blends within each region. When point estimates differ in a meaningful way, one of them tends to be much more imprecise than the other. Generally, lower blends are estimated less precisely because variation in the subsidy is smaller in magnitude than for higher blends. Most notably, B5 estimates are much less precise than other blends in most regions.In the East Coast, however, the results for B5 are qualitatively different. This could result from the fact that markets for B5 are fundamentally different in some cases . It could also be that some B5 is blended above the rack. The PNW, for example, has a spot market for B5, so the subsidy there would be passed through to the spot price rather than the rack price and this action could arise in other regions. In California, the pattern is similar, however only blends above B20 have 95 percent confidence intervals that exclude complete pass through. Additionally, although imprecise, pass-through point estimates tend to be lower for higher blends. This is discussed further in conjunction with LCFS subsidies in Section 1.5.2.The RIN subsidy pass through results exhibit some consistencies with previous findings in the literature studying pass through of RIN subsidies to blended gasoline and some important differences. The finding of complete pass through in the Midwest and incomplete pass through on the East Coast is generally consistent with Pouliot et al. . However, the cities used in my sample differ from theirs for each region, and their finding are sensitive to looking at branded and unbranded products, and only unbranded fuels are available here.