A Spatial Assessment of Possible Water Quality Trading Markets in Tennessee

The idea of using marketable or tradable permits to control water pollution was first proposed over 40 years ago (Crocker; Dales). However, only recently has this idea generated much interest from policymakers. The United States Environmental Protection Agency (EPA) began publicly mulling over the idea of a particular type of trading program known as “offset trading” in 1996, eventually endorsing it in 2003 (EPA 1996, 2003). EPA quickly followed this endorsement with specific guidance on implementing emissions trading schemes (EPA, 2004). EPAs endorsement has encouraged a number of states to adopt water quality trading (WQT) programs (Breetz et al.; Woodward and Kaiser).

Increased interest in WQT can be attributed to a number of factors, not the least of which is the success of trading programs in air quality regulation (Colby). The costs savings attributed to air quality trading schemes stand in stark contrast to the perception that large expenditures on water pollution control have failed to address ambient water quality problems adequately.¹ Another motivating factor is that the sources that are increasingly responsible for water quality degradation, such as agricultural non-point sources, are also ones over which EPAs regulatory options have, to date, proven somewhat limited.²

However, the extent to which WQT programs are capable of addressing the Nation's lingering water quality problems remains unclear. For example, the WQT programs that have been implemented have been plagued by a paucity of trades (Farrow et al.; King, 2005). What is clear, though, is that WQT programs face a number of constraints that may limit their ability to address these impairments. Some of these constraints involve the geographic or spatial dimensions of potential markets. In simple terms, the feasibility of a market depends quite crucially upon the relative location of discharge sources both to one another and to water quality impairments. This paper evaluates the potential for WQT programs to address oxygen-related impairments in the State of Tennessee by considering the spatial relationship between these impairments and sources of oxygen-related discharges, focusing initially on nitrogen discharges.

The method employed in this evaluation is somewhat similar to that used in an assessment of the potential for WQT in U.S. coastal watersheds (Crutchfield, Letson, and Malik; Letson, Crutchfield, and Malik). Although there were a number of methodological differences, the largest difference between this study and the coastal study is in terms of focus. By focusing solely on coastal zones, the coastal study concentrated solely on the potential for WQT in the mouth of, what are typically, large, intensively used watersheds. Thus, all sources within the watershed could be deemed contributors to the degradation of ambient water quality in the coastal zone. The focus here was on the potential for WQT to address specific water quality impairments wherever they may occur in the watershed. Thus, location of sources relative to impairments (i.e., upstream or downstream) and the resulting definition of the potential WQT market were of critical importance.

The remainder of the paper is organized as follows. The next section provides a brief overview of WQT, specifically focusing on the offset trading program endorsed by EPA. An analysis of the spatial dimensions of offset trading and a description of the way in which these spatial dimensions are used to delineate potential WQT markets follows. The extent to which markets for nitrogen offsets can address Tennessee's nitrogen-related water quality impairments is then analyzed. These markets are also ranked to identify those that are most likely to be capable of supporting an offset market. The analysis is then broadened to include other oxygen-related pollutants. Finally, some concluding thoughts are provided.

Water Pollution Control and Emissions Trading

The passage of a series of amendments to the Federal Water Pollution Control Act in 1972, collectively known as the Clean Water Act (CWA), represented a sea change in water-pollution control policy in the United States. Prior to the CWA, these policies primarily consisted of a network of state-based ambient water quality-based regulations (Houck, pp. 11–12). The CWA consolidated much of the authority for water pollution control at the federal level and imposed technologically based discharge limits on all “point sources” (PS).³ As a result, all PS discharging a pollutant to the waters of the United States were required to obtain a permit from the ambitiously titled National Pollution Discharge Elimination System (NPDES). To obtain an NPDES permit, point sources were required to satisfy the technology-based effluent limits (TBELS) imposed as a condition of the permit. Much of EPAs water-pollution control efforts over the next 30 years went into promulgating, defending, and enforcing these permits and the TBELS they imposed. Although much of the actual responsibility for administering the NPDES program has been devolved to the states over time, EPA retains substantial oversight of the state programs (Gallagher, pp. 279–81).

The CWA took a different approach to non-point source (NPS) discharges, exempting them from the NPDES permitting requirements. Instead, Section 319 of the CWA requires that states undergo a series of planning and reporting requirements for reducing NPS pollution, but includes no penalty for states that fail to meet these requirements (33 United States Code §1329.). States, in turn, have responded to the challenge of regulating NPS pollution with a patchwork of initiatives that vary widely from one state to another. Individual initiatives aside, the principal legacy of Section 319 has been a series of voluntary programs for agricultural NPS—primarily education, technical assistance, and cost sharing for agricultural best management practices (Malik, Larson, and Ribaudo).

The results of the differing approaches of the CWA to PS and NPS pollution have hardly been surprising. The regulation of PS has dramatically reduced discharges from these sources and often led to a substantial improvement in ambient water quality downstream of these PS (Houck; Boyd; Smith, Alexander, and Wolman). The voluntary programs directed at discharges from NPS have not generally enjoyed the same level of success, and NPS are now considered the most prevalent contributors to water quality impairment.⁴ For example, agriculture now contributes to more miles of river and stream impairment than any other single source of pollutants (EPA, 2002).⁵ The failure to meaningfully reduce NPS pollution has also contributed to a failure to achieve the CWAs stated goal of “restor[ing] and maintain[ing] the chemical, physical, and biological integrity of the nation's water.” At last count, approximately 39% of the Nation's assessed river and stream miles were impaired for one or more uses (EPA, 2002).

The imposition of mandatory controls on PS but not NPS also appears to have contributed to a divergence in abatement costs. Recent estimates suggest that the marginal abatement costs of NPS may be significantly lower than those of PS, perhaps as much as an order of magnitude or greater (Faeth). This difference is in part due to the fact that PS have largely exhausted the less expensive technologies for abating pollution while many NPS still face a full range of abatement options. That said, variability in abatement costs for both PS and NPS, due to such factors as the stringency of the abatement requirements (e.g., Jiang et al.), is likely to create circumstances in which NPS enjoy no abatement-cost advantage over PS. In any event, it seems clear that attempting to address the country's lingering water quality problems through increasingly stringent regulation of PS will be an excessively expensive, if not fruitless, proposition.

Under pressure to reduce water pollution further, but lacking clear statutory authority to regulate NPS directly, EPA has resorted to a vestige of the old state-based ambient water quality standards retained in the CWA. Section 303(d) of the CWA requires that states: (i) identify waters that fail to meet water quality standards after all relevant PS have complied with their NPDES permit requirements, (ii) for each such water body, calculate the total maximum daily load of pollutants (TMDL) that can be discharged into the water body without causing it to fail to meet applicable water quality standards, and (iii) allocate this pollutant load among all sources of discharges to this body of water (33 United States Code 1313(1)(A)). Thus, Section 303(d) bases water pollution control on the amount of the pollutant the receiving waters can assimilate, rather than the level to which PS could “reasonably” be expected to lower their discharges. In allocating this pollutant load, EPA can reduce discharges from PS by imposing water quality-based effluent limitations (WBELs) that are more stringent than the TBELs imposed by an NPDES permit and/or impose WBELs on discharges not subject to a pre-existing TBEL (EPA, 2003). In addition, the TMDL program may provide EPA the means of imposing effluent limitations on NPS, although the extent of the statutory authority for these limitations is unclear (e.g., Tobin). In either event, the TMDL program is viewed as a mechanism for implementing WQT (EPA, 2004).

EPAs endorsement of WQT is limited in a number of important ways. For example, it extends to a limited set of pollutants, primarily nutrients, “oxygen-related cross-pollutants,” and sediment, although EPA is also willing to consider the trading of other pollutants on a “case-by-case basis” (EPA, 2003). This paper focuses solely on nitrogen and other oxygen-related pollutants. Another important limitation is that EPA opted for an “offset” program over a watershed-based “cap and trade” system. Under an offset program, a polluter under regulatory pressure to make additional discharge reductions—such as a PS confronted with WBELS—would be allowed to purchase the required reductions, or offsets, from other sources in the watershed. In a cap and trade system, a maximum quantity of discharges of a particular pollutant for a particular watershed would be determined. Then, individual discharge quotas, or “rights,” summing to this maximum would be distributed to all dischargers within the watershed. Individual dischargers would be free to trade these rights among themselves.⁶ Thus, under EPAs offset program, the only sources with a regulatory incentive to purchase these rights or offsets would be PS upon which a WBEL has been, or likely will be, imposed, i.e., those PS that have “a reasonable potential to cause or contribute to” a violation of water quality standards. If EPA is judged to have the authority to impose effluent limitations on NPS under the TMDL program, then NPS subject to a TMDL could face a similar incentive (and the resulting program would more closely resemble a “cap-and-trade” system). However, for the purposes of this paper it is assumed that only PS facing a WBEL have the regulatory incentive to purchase offsets.⁷ Potential sellers of these offsets would be both NPS and PS contributing to one or more of the same impairments.

The Spatial Dimensions of Water Quality Trading

The spatial dimensions of emissions trading programs revolve around the physical relationship between sources and receptors (geographic locations where ambient environmental quality is measured). For the type of WQT programs envisioned by EPA, these receptors are impaired water bodies. The physical relationship between a source and a receptor is captured in the notion of a transfer coefficient or a numerical or functional representation of the extent to which discharges from a particular source affect a particular receptor. The process of translating transfer coefficients into trading ratios, or the rate at which discharges from one source will trade for discharges from another source, has been the subject of a number of studies (e.g., Farrow et al.; Hung and Shaw; Horan and Shortle; Woodward; McGinnis). Hung and Shaw, for example, propose a trading system that utilizes predetermined trading ratios to minimize transactions costs. Under this system, a stream would be divided into zones in which the discharge of one unit of a pollutant from anywhere within the zone would have approximately the same environmental impact. Each zone would have a maximum discharge cap imposed by quotas that could be traded at a ratio of 1:1 within the zone. The inter-zone trading ratios would be determined by the rate at which the pollutant is transferred from upstream to downstream zones. This trading ratio defines the incentives that sources in different zones have to trade with one another, essentially defining the geographic extent of the market.

The ability of a WQT program to address water quality impairments will depend upon these spatial dimensions. Under EPAs offset program, a feasible market will typically require one or more PS contributing to one or more impairments (thus being subject to WBELS) and one or more NPS or PS with offsets to sell, i.e., the ability to reduce their discharges below any regulatory requirements (such as a TBEL for a PS), contributing to one or more of the same impairments. Thus, evaluating the feasibility of WQT to address impairments requires defining the relevant markets based on the spatial relationship between sources and impairments and the existence of potential buyers and sellers of credits. The remainder of this paper focuses on defining these markets for Tennessee, identifying potential market participants, and assessing the implications for WQT as a means of addressing Tennessee's water-quality impairments.

Data

The geographic delineation of a market begins with the notion of a watershed, or land area that drains all water that falls within it to a common point. In the United States, watersheds, or hydrologic units, are identified by a unique hydrologic unit code (HUC). HUCs range from two to twelve digits with an increasing number of digits signifying increasing spatial resolution. Thus, each two digit HUC is comprised of a number of four digit HUCs, each four digit HUC is comprised of a number of eight digit HUCs, and so on. This analysis will focus on all of the 12-digit HUCs (HUC12s) that are contained, either partially or completely, within the State of Tennessee.⁸

Digital representations of HUC12 boundaries were downloaded from the U.S. Department of Agriculture's Natural Resources Conservation Service Geospatial Data Gateway. Geospatial data on the location and direction of flow of Tennessee's rivers and streams was extracted from the U.S. Geological Service's National Hydrography Dataset Geodatabase. Data on water quality impairments was obtained from Tennessee's 2002 303(d) List (Tennessee Department of Environment and Conservation, 2004), which lists all impaired water bodies in the State along with their respective causes (i.e., pollutants). The geospatial location of these impairments was obtained from a shape file downloaded from EPAs Watershed Assessment, Tracking & Results website. Information on PS was obtained from EPAs online Permit Compliance System.⁹ Finally, land use data came from the 2001 National Land Cover Dataset developed by the Multi-Resolution Land Characteristics Consortium.

Methods and Procedures

Assessing the feasibility of WQT in Tennessee essentially involved applying a set of screens or hurdles to Tennessee's watersheds to identify potential WQT markets. The characteristics used to delineate a WQT market are summarized in table 1. To begin, impairments due to nitrate, nutrients, organic enrichment, or low dissolved oxygen (Nitrogen-Related Impairments) were selected from the entire set of Tennessee river and stream impairments.¹⁰ Similarly, PS with quantitative limits for nitrogen (Nitrogen PS) were selected and geospatially located. HUC12s that neither contained nor were upstream of a Nitrogen-Related Impairment were eliminated from further consideration. Next, Nitrogen-Related Impairments that were not downstream of one or more Nitrogen PS were eliminated on the basis that there were no potential buyers of offsets for these impairments, leaving the “Tradable Nitrogen Impairments.” Similarly, Nitrogen PS that were not upstream of a Tradable Nitrogen Impairment were eliminated, leaving the “Contributing Nitrogen PS.” HUC12s that neither contained nor were upstream of a Tradable Nitrogen Impairment were then eliminated. The NPS included in this analysis were the agricultural lands located within these remaining HUC12s.

The remaining HUC12s (and the Tradable Nitrogen Impairments, Contributing Nitrogen PS, and agricultural lands that they contained) were then grouped into “Nitrogen Markets” based on their contribution to a water quality impairment. Thus, all remaining HUC12s containing or upstream of a particular impairment were grouped into the same Nitrogen Market. Figure 1 illustrates this process with a map of the North Fork Forked Deer watershed. In this watershed, there is a Nitrogen-Related Impairment at the mouth of the watershed, so the entire watershed forms a single Nitrogen Market and all of the Nitrogen PS and NPS in the watershed are potential participants in the market. However, the watershed also contains a number of Nitrogen-Related Impairments that are not downstream of a Nitrogen PS and, thus, are not a part of the market (i.e., are not Tradable Nitrogen Impairments).

These markets were scored and ranked based on: aggregate length of Tradable Nitrogen Impairments within each market; aggregate permitted discharges of nitrogen from all Contributing Nitrogen PS; estimated nitrogen discharges from NPS (agricultural lands); distance between discharges and impairments; and relative balance between permitted PS discharges and estimated NPS discharges. The aggregate length of Tradable Nitrogen Impairments was included as a measure of the expected “benefits” associated with a trading program (both in terms of environmental improvement and potential cost savings). PS and NPS discharges were included as a measure of the likelihood that a trading program would be capable of addressing the impairments. PS discharge limits were used instead of actual discharges because these limits represent the baseline for a trading program (WBELS will reduce discharges below these baselines). Nitrogen loads from NPS were estimated by using land use data to quantify the amount of land in either “pasture/hay” or “cultivated crops” and multiplying this acreage by emissions factors of 5.53 pounds of nitrogen per acre per year for pasture/hay and 23.2 pounds per acre per year for cultivated crops, based on empirical estimates of such loads (Bhaduri et al.).

PS and NPS discharges were discounted by distance from impairment(s) because pollutants such as nitrogen dissipate during in-stream transport (Hung and Shaw). Thus, to obtain scores that account for the likely effects of the pollutant loads, NPS and PS scores (S_N and S_P, respectively) were calculated for each Nitrogen Market as

(1)

(2)

where L_i is the total estimated nitrogen load from NPS for the ith HUC12; D_i is the total permitted nitrogen discharges from all of the PS in the ith HUC12; M_j is the aggregate length of the Tradable Nitrogen Impairments in the jth HUC12; x_ij is the distance (in number of HUC12s) between the ith and jth HUC12s, such that, if i = j then x_ij = 1, if i and j are contiguous then x_ij = 2, and so on; and n is the number of HUC12s within the Nitrogen Market.

The relative contribution of PS and NPS discharges is also an important factor in evaluating the Nitrogen Markets, since PS are likely to be the sole purchasers of offsets and NPS may often be the only potential sellers for which the difference in abatement costs are large enough to justify a trade. If either set of discharges contributes little to the impairment(s), it is unlikely that a market would be able to generate enough trades to have a meaningful effect on the impairment(s). On the other hand, if these contributions were more balanced so that the reductions necessary to achieve water quality standards could be made by either PS or NPS, it is more likely that there would be a large enough supply of, and demand for, offsets to eliminate the impairment. Calculating accurate estimates of these relative contributions would require extensive water quality modeling efforts. Thus, for the purposes of this analysis, the ratio of the NPS and PS scores is used to measure relative contributions. We assume that the closer the ratio is to 1:1, the more favorable the conditions are for market development. This assumption should not be construed as assuming a trading ratio of 1:1, as these scores are weighted both for distance and extent of impairment (in meters) to which the discharges contribute.

Standard deviations for the NPS and PS scores, and the ratio of the two were calculated and each Nitrogen Market was scored based on the number of standard deviations above zero its NPS and PS scores and ratio were. One point was awarded for each standard deviation. Thus, a Nitrogen Market with a NPS score that was greater than two, but less than three, standard deviations above was awarded two points. The points earned for the NPS and PS scores and ratio of the two were then summed to give an overall score.

While the identification, analysis, and ranking of Nitrogen Markets is a useful exercise it is also limited in three important ways. First, it assumes that trading is limited to nitrogen offsets even though EPA explicitly allows for the possibility of “cross-trading of oxygen-related pollutants” (EPA, 2003). Second, some PS that do not currently have a quantitative limit for nitrogen discharges, are likely to receive one in the event an oxygen-related TMDL is developed for an impairment to which they contribute. Third, to capture all impairments potentially contributed to by nitrogen emissions, the types of impairments considered Nitrogen-Related (impairments due to nitrate, nutrients, organic enrichment, or low dissolved oxygen) is quite broad. For all these reasons, the potential for trading may be underestimated by looking exclusively at the Nitrogen PS. To provide a more accurate upper bound on the extent to which trading might be able to address oxygen-related impairments in Tennessee, the analysis was broadened to add impairments due to phosphorous to the Nitrogen-Related Impairments to create the “Oxygen-Related Impairments” and a set of Oxygen-Related PS was identified and geospatially located. The Oxygen-Related PS were comprised of all PS with NPDES-permits for nitrogen, phosphorus, dissolved oxygen, or any kind of oxygen demand as well as any other PS for which the SIC code was listed as “4952 sewerage systems.” No effort was made to rank the resulting markets as was done with the Nitrogen Markets, in large part, due to the difficulty of comparing discharges of different pollutants.

Results

Analysis of Nitrogen Trading Program

The search of Tennessee water quality impairments identified 2,300 Nitrogen-Related Impairments totaling 3,815,042 meters. Of these impairments, 2,037, totaling 3,427,687 meters, were impaired due to a single nitrogen-related cause, while the remaining 263 impairments, covering 387,355 meters were impaired due to two nitrogen-related causes. The most prevalent of the nitrogen-related causes was “low dissolved oxygen” or “organic enrichment/low dissolved oxygen,” which together were a cause of 1,231 impairments covering a total of 2,076,767 meters. The next most prevalent causes were “nutrients” with 813 impairments for 1,340,028 meters, and “nitrate” or “nitrate/nitrite,” which was a cause of 519 impairments covering 785,602 meters. In addition to the oxygen-related causes, a large number of these impairments was also attributed to non-oxygen-related causes, such as “siltation,” which was listed as a cause of 1,274 impairments, totaling 1,881,637 meters. The number of all causes for each impairment ranged from one to six with an average of 2.64 causes per impairment. The locations of the Nitrogen-Related Impairments are shown in figure 2.

The search of Tennessee's approximately 1,511 NPDES-permitted PS identified 161 with a quantitative limit on nitrogen discharges. The aggregate total of these permitted discharges was equivalent to 18,751,312 pounds of nitrogen per year. Thus, the mean discharge limit for the Nitrogen-Related PS was equivalent to 116,468 pounds per year.

Eliminating the HUC12s that neither contained nor were upstream of a Nitrogen-Related Impairment eliminated 699 of the 1,085 HUC12s partially or wholly within Tennessee, leaving 386. Eliminating impairments that were not contributed to by a Nitrogen-Related PS eliminated 1,528 of the Nitrogen-Related Impairments (totaling 2,531,565 meters) and another 93 of the HUC12s. Similarly, 79 of the 161 Nitrogen-Related PS were eliminated, as they were not contributing to a Nitrogen-Related Impairment. The spatial location of the remaining 772 Tradable Nitrogen Impairments covering 1,283,477 meters, 293 Nitrogen-Contributing HUC12s, and 82 Nitrogen-Contributing PS were grouped into the 40 Nitrogen Markets shown in figure 3.

The Nitrogen Markets covered approximately 29% of the State's land mass, but almost 34% of the State's pasture/hay and cultivated cropland acres and about 31% of the estimated Statewide nitrogen discharges from these land uses. About 51% of the State's Nitrogen-Related PS and 57% of the permitted nitrogen discharges from these PS were contained in these markets. However, only about 34% of the State's Nitrogen-Related Impairments, by length, were included, suggesting that about two-thirds of the State's Nitrogen-Related Impairments could not be addressed with this type of WQT program. Table 2 contains summary statistics for all 40 Nitrogen Markets.

Table 2. Summary statistics for nitrogen markets

N = 40	Minimum	Maximum	Mean	Total	Share of State Total
HUC12s	1	51	7.325	293	27.0%
Nitrogen-Related
Impairments (meters)	4,011	133,442	32,154	1,286,167	33.7%
Nitrogen-Contributing
PS	1	11	2.05	82	50.9%
Permitted Nitrogen
Discharges from PS (lbs/yr)	450	4,818,780	266,497	10,659,868	56.8%
Total Area (acres)	11,281	1,545,334	210,349	8,413,951	29.0%
Hay/Pasture Land (acres)	1,016	417,604	50,449	2,017,974	37.8%
Share in Cropland (acres)	0	369,977	23,631	945,257	28.0%
Estimated Nitrogen
Discharges from NPS (lbs/yr)	6,549	8,809,864	827,234	33,089,368	30.7%

The number of HUC12s that comprised these markets ranged from a low of one to a high of 51, with a mean of 7.33. However, a majority of the markets (21 out of 40) were comprised of a single HUC12, while another four were comprised of only two HUC12s. This finding indicates that many of the Tradable Nitrogen Impairments occur in the headwaters of rivers and streams where stream flow is low. This suggests that many of these markets may prove to be “thin,” or suffer from a low number of potential market participants.

The number of PS contained within any one Nitrogen Market ranged from a low of one to a high of 11, with the average being 2.05. Evidence of thin markets is provided by the finding that 28 of the 40 markets contained a single Nitrogen-Contributing PS, while 32 of the 40 contained two or fewer such PS. Of the 28 markets with a single PS, 19 were comprised of a single HUC12. These findings have important implications for the structure of Nitrogen Markets. For example, offset trades from one PS to another were not feasible in 70% of the markets. Second, the low number of potential offset buyers implies that these markets are more likely to function as a series of bilateral negotiations than as a clearinghouse or exchange market (Woodward and Kaiser).

Analysis of Oxygen-Related Trading Program

The Oxygen-Related Impairments (the Nitrogen-Related Impairments plus impairments caused by phosphorous) were comprised of 2,407 impairments covering 3,954,517 meters. Again, a majority of these impairments (1,995 covering 3,340,138 meters) was impaired due to a single oxygen-related cause. A minority, (412 impairments covering 614,379 meters) was impaired due to two oxygen-related causes. “Phosphorous” was the least prevalent of these causes accounting for only 252 impairments and 357,949 meters. The locations of the Oxygen-Related Impairments are shown in figure 2.

The search of Tennessee's approximately 1,511 NPDES-permitted PS for Oxygen-Related PS (i.e, PS with an NPDES permit for nitrogen, phosphorus, dissolved oxygen or any kind of oxygen demand, or an SIC code of “4952 sewerage systems”) yielded 554 PS. All but 11 of these had an NPDES permit for either nitrogen (N), phosphorous (P), dissolved oxygen (DO), biochemical oxygen demand (BOD), chemical oxygen demand (COD), or carbonaceous oxygen demand (CBOD). The average number of permits for these pollutants per PS was 2.6, with 62 PS having one, 173 having two, 233 having three, 69 having four and six PS having a permit for five of the six pollutants. There were 333 permits for N, 85 for P, 434 for DO, 275 for BOD, and 256 for CBOD.

Purging the HUC12s that neither contained nor were upstream of an Oxygen-Related Impairment eliminated 682 of the 1,085 HUC12s partially or wholly within Tennessee. Eliminating Oxygen-Related PS that did not contribute to an Oxygen-Related Impairment identified 241 Oxygen-Contributing PS out of the 554 Oxygen-Related PS. Eliminating impairments that were not downstream of an Oxygen-Contributing PS removed another 60 of the HUC12s. The remaining impairments and the associated 343 HUC12s were then grouped into 57 different “Oxygen-Related Markets” as shown in figure 4 based on whether the impairments shared one or more discharge sources.

Summary statistics for the Oxygen-Related Markets are provided in table 4. While the number of HUC12s per market ranges from a low of 1 to a high of 51, a majority of the Oxygen-Related Markets (30 of 57) are comprised of a single HUC12, with another six comprised of two HUC12s. While the average number of PS within the markets is 4.23 per market, 29 of the 57 Potential Markets contain a single Oxygen-Contributing PS, and 40 of the 57 contain two or fewer such PS. Of the 29 markets with a single PS, 23 are comprised of a single HUC12.

Table 4. Summary statistics for oxygen-related markets

N = 57	Minimum	Maximum	Mean	Total	Share of State Total
HUC12s	1	51	6.02	343	31.6%
Oxygen-Related
Impairments (meters)	4,011	150,466	35,744	2,037,418	51.52%
Oxygen-Related PS	1	42	4.23	241	43.50%
Total Area (acres)	9,276	1,545,334	172,061	9,807,463	33.80%
Hay/Pasture Land (acres)	2.4	417,589	38,224	2,178,778	40.8%
Cropland (acres)	0	369,964	23,297	1,327,977	39.31%

These 57 Oxygen-Related Markets can be thought of as representing the full range of possibilities for the trading of oxygen-related offsets under an offset-type trading program as envisioned by EPA. Thus, these markets represent an upper bound on the extent to which this type of trading program would be able to address Oxygen-Related Impairments in Tennessee's rivers and streams. Even as an upper bound, these markets could address only slightly more than half of the State's Oxygen-Related Impairments. Further, a more detailed consideration of the suitability of the discharges from individual sources for participation in a trading program (including the use of water quality modeling to determine the extent to which these discharges actually affect specific water quality impairments) is likely to decrease this percentage, perhaps significantly. No effort was made to rank these markets due to the absence of quantitative data on emissions and the difficulty of determining appropriate trading ratios for different oxygen-related pollutants.

Conclusions

The analyses of the geospatial dimensions of potential WQT markets in Tennessee presented in this paper highlight a number of important factors to consider when evaluating the feasibility and possible design of WQT markets. First, WQT is unlikely to be a potential solution for the majority of waters impaired due to nitrogen or other oxygen-related pollutants in Tennessee. Simply looking at the most basic requirements for a possible market—the existence of potential buyers and sellers of offsets—eliminates the possibility of trading for roughly half of these impairments. Considering other requirements for trading would surely eliminate a significant number of the other impairments. Second, even when these minimal requirements are met, often a “thin market” would exist with no more than one or two potential buyers of offsets. Similarly, the possibility of trading among PS is likely to be relevant for a small number of watersheds. In large part, this paucity of potential buyers comes from the oxygen-related impairments often being located in or near the headwaters of rivers and streams. The low number of potential buyers increases doubt over whether these markets would be able to overcome the problem of low or non-existent trading activity that has hindered many previous efforts to establish water and other environmental trading programs.

These findings also have important implications for policy design. For example, the paucity of potential buyers suggests that an agency promoting WQT in Tennessee would be well advised to focus on helping individual purchasers locate and contract with individual sellers, rather than on creating elaborate infrastructure to support a more formal market. Fortunately, or unfortunately, the opportunities for a more formal market are limited by the low number of areas with more than one or two potential buyers of credits. In addition, in designing a WQT program, results indicate that the more restrictive the possibilities for trading, the less likely the program will be successful in addressing many of the existing water quality impairments. Thus, the temptation to design highly restrictive programs to ensure that trading activity will result in substantial water quality improvement should be understood in light of the obvious tradeoff in program relevance. Nowhere is this tradeoff more evident than in the distinction between an offset program, as analyzed here, and a more inclusive cap-and-trade program. The structure of the offset program reduces the set of potential buyers of discharges reductions to NPDES-permitted PS that contribute to an existing impairment, thus, eliminating a wide array of other sources, including NPS. Thus, it is perhaps not surprising that effectiveness of this type of market in Tennessee is likely to be severely limited by low numbers of potential buyers.

The ranking system employed in the analysis, however, has identified a limited number of areas where WQT would appear to be most feasible. Further study of WQT in Tennessee should be focused on one or more of these areas. Such study should include some of the factors that are important to WQT suitability, but which were not considered in this analysis, such as marginal abatement costs, transfer coefficients and transactions costs. In addition to providing a better picture of the feasibility of trading in these particular areas, this more in-depth examination could also provide information that would be helpful in interpreting the scores and rankings produced by this exercise and, possibly refining these processes for future studies. Perhaps more importantly, the large gap between those markets deemed most suitable for trading and the rest of the markets provides a clear geographical guide for policymakers interested in implementing a WQT policy in Tennessee.