LYNCH, Chief Judge.

Massachusetts officials appeal from an injunction against a 2006 Massachusetts statute establishing differential methods by which wineries distribute wines in Massachusetts, Mass. Gen. Laws ch. 138, § 19F. The district court enjoined enforcement of § 19F on the ground that the law discriminates against interstate commerce in violation of the Commerce Clause of the United States Constitution. See Family Winemakers of Cal. v. Jenkins, No. 1:06-cv-11682-RWZ at 17-28 (D.Mass. Nov. 19, 2008) (order granting summary judgment).

We briefly summarize the basis for the lawsuit, the issues presented, and our resolution of them before turning to the supporting analysis. Section 19F only allows "small" wineries, defined by Massachusetts as those producing 30,000 gallons or less of grape wine a year, to obtain a "small winery shipping license." This license allows them to sell their wines in Massachusetts in three ways: by shipping directly to consumers, through wholesaler distribution, and through retail distribution. All of Massachusetts's wineries are "small" wineries. Some out-of-state wineries also meet this definition.

Wines from "small" Massachusetts wineries compete with wines from "large" wineries, which Massachusetts has defined as those producing more than 30,000 gallons of grape wine annually. These "large" wineries must choose between relying upon wholesalers to distribute their wines in-state or applying for a "large winery shipping license" to sell directly to Massachusetts consumers. They cannot, by law, use both methods to sell their wines in Massachusetts, and they cannot sell wines directly to retailers under either option. No "large" wineries are located inside Massachusetts.

Plaintiffs, a group of California wine-makers and Massachusetts residents, assert § 19F was designed with the purpose, and has the effect, of advantaging Massachusetts wineries to the detriment of those wineries that produce 98 percent of the country's wine, in violation of the Commerce Clause. Massachusetts defends § 19F on the basis that its law has neither a discriminatory purpose nor a discriminatory effect. Massachusetts has not argued in its briefs that there are no legitimate alternative methods of regulation to serve § 19F's asserted purposes. Massachusetts also argues that under the Twenty-first Amendment, state laws are immunized from Commerce Clause scrutiny unless the laws discriminate on their face.

The primary question before us is whether § 19F unconstitutionally discriminates against interstate commerce in light of both the Commerce Clause, art. I, § 8, cl. 3, and § 2 of the Twenty-first Amendment.

It is clear that § 2 of the Twenty-first Amendment does not protect state alcohol laws that explicitly favor in-state over out-of-state interests from invalidation under the Commerce Clause. Granholm v. Heald, 544 U.S. 460, 489, 125 S.Ct. 1885, 161 L.Ed.2d 796 (2005). But § 19F is neutral on its face; it does not, by its terms, allow only Massachusetts wineries to distribute their wines through a combination of direct shipping, wholesaler distribution, and retail sales. Section 19F instead uses a very particular gallonage cap to confer this benefit upon "small" as opposed to "large" wineries.

We hold that § 19F violates the Commerce Clause because the effect of its particular gallonage cap is to change the competitive balance between in-state and out-of-state wineries in a way that benefits Massachusetts's wineries and significantly burdens out-of-state competitors. Massachusetts has used its 30,000 gallon grape wine cap to expand the distribution options available to "small" wineries, including all Massachusetts wineries, but not to similarly situated "large" wineries, all of which are outside Massachusetts. The advantages afforded to "small" wineries by these expanded distribution options bear little relation to the market challenges caused by the relative sizes of the wineries. Section 19F's statutory context, legislative history, and other factors also yield the unavoidable conclusion that this discrimination was purposeful. Nor does § 19F serve any legitimate local purpose that cannot be furthered by a non-discriminatory alternative.

We further hold that the Twenty-first Amendment cannot save § 19F from invalidation under the Commerce Clause. Section 2 of the Twenty-first Amendment does not exempt or otherwise immunize facially neutral but discriminatory state alcohol laws like § 19F from scrutiny under the Commerce Clause. We affirm the grant of injunctive relief.

I. Facts

We engage in de novo review both because the district court entered summary judgment and because the issues presented are ones of law. There is no disagreement on the material facts. See Fed.R.Civ.P. 56(c); see also Sullivan v. City of Springfield, 561 F.3d 7, 14 (1st Cir. 2009).

The ratification of the Twenty-first Amendment ended Prohibition and gave states substantial control over the regulation of alcoholic beverages. Most states, including Massachusetts, then imposed a three-tier system to control the sale of alcoholic beverages within their territories. The hallmark of the three-tier system is a rigid, tightly regulated separation between producers, wholesalers, and retailers of alcoholic beverages. Producers can ordinarily sell alcoholic beverages only to licensed in-state wholesalers. Mass. Gen. Laws ch. 138, §§ 2 and 19. Wholesalers then must obtain licenses to sell to retailers. Id. § 18. Retailers, which include stores, taverns, restaurants, and bars, must in turn obtain licenses to sell to consumers or to serve alcohol on their premises. Id. §§ 12, 15. Recently, as to wine, Massachusetts has adjusted the separation between these three tiers, as we describe below.

The structure of the usual three-tier system is commonly described as an hour-glass, with wholesalers at the constriction point. There are thousands of producers nationwide, a handful of licensed Massachusetts wholesalers, and approximately ten thousand licensed retailers in Massachusetts. See Commonwealth of Massachusetts Alcoholic Beverages Control Commission, Licensing, http://www.mass.gov/abcc/licensing/licensing.htm.

The three-tier system has had a particularly pronounced effect on wineries' access to the Massachusetts market. The economic incentives created by the three-tier system, in conjunction with the structure of the wine industry, severely limited certain wineries' ability to sell their wines in Massachusetts.

In 2006, the year § 19F was enacted, 5,350 registered wineries in the United States produced a total of 646,395,818 gallons of wine, which includes both grape wine and fruit wine production. Almost all of the country's wine production and sales come from a small number of wineries. In 2006, the five largest wineries in the U.S. produced approximately 70 percent of the country's wine. The country's thirty largest wineries comprised approximately 92 percent of the market, and each produced between 680,000 and 150 million gallons per year. The rest of the commercial market — the 3,540 wineries which produce between one and 680,000 gallons per year — competed for 8 percent of the market share. Finally, 1,780 wineries produced less than one gallon of wine per year and had virtually zero percent of the market share.

The concentration of wine production among the largest producers is driven by another feature of the wine industry: there are, broadly speaking, two categories of wine, high-volume, lower-cost wines and low-volume, higher-quality, higher-priced boutique wines. The largest wineries produce millions of gallons of wine per year because they have generally specialized in the former, but not to the exclusion of the latter. Wineries smaller than the largest producers have tended to specialize in low-volume boutique wines, which can be produced with a relatively small quantity of grapes and a much lower initial outlay of resources. At least until the current recession, consumer demand for boutique wines had grown exponentially, fueling a rise in the number of smaller U.S. wineries (which include many wineries producing more than 30,000 gallons annually). Fed. Trade Comm'n, Possible Anticompetitive Barriers to E-Commerce: Wine 6 (2003), available at http://www.ftc.gov/os/2003/07/winereport2.pdf (hereinafter FTC Report).

Under Massachusetts's former three-tier system, all wineries could only distribute their wines through licensed Massachusetts wholesalers, and 75 percent of the wine sold in Massachusetts went through five wholesalers. This gave wholesalers, not wineries, the balance of the bargaining power. Wholesalers do not necessarily distribute a winery's entire range of wines; they often distribute the wines most likely to be profitable to them, which are lower-priced, high-volume wines. Wholesalers make profits by selling wines to retailers at a markup. The more a wine sells out at retail, generating more requests for restocking, the more money a wholesaler makes. Wholesalers also face fixed costs that do not depend on the price they pay to the winery for the wine: they bear the costs of transportation, storage, and handling. For these reasons, wineries producing higher-priced, low-volume wines, whatever the gallonage output of the winery, are less profitable and less likely to attract wholesaler distribution.

The largest wineries, as the major producers of lower-priced, high-volume wines, have been best able to attract wholesalers. Only the country's fifty to one hundred largest wineries have consistently secured wholesaler representation. For most smaller wineries of whatever gallonage, which produce mostly boutique wines, obtaining wholesaler representation has been difficult, if not impossible. And even if a smaller winery obtained wholesaler representation, wholesalers were likely to distribute only one or two of its wines, limiting Massachusetts consumers' access to particular wines.

Wineries have heralded direct shipping as a supplemental avenue of distribution because of its economic advantages, especially for wineries that do not rank among the fifty to one hundred largest producers. Direct shipping lets consumers directly order wines from the winery, with access to their full range of wines, not just those a wholesaler is willing to distribute. Direct shipping also avoids added steps in the distribution chain, eliminating wholesaler and retailer price markups. See FTC Report at 22-23.

Before 2005, § 19B, Massachusetts's farmer-winery licensing law, on its face allowed only in-state wineries to obtain licenses to combine distribution methods through wholesalers, retailers, and direct shipping to consumers. Mass. Gen. Laws ch. 138, § 19B (2002). Five months after Granholm invalidated similar facially discriminatory state laws, § 19B was held to be invalid under the Commerce Clause. Stonington Vineyards v. Jenkins, No. 05-10982-JLT, slip op. at 1-2 (D.Mass. Oct. 5, 2005).

In 2006, the Massachusetts legislature enacted § 19F over then-Governor Romney's veto. Section 19F does not distinguish on its face between in-state and out-of-state wineries' eligibility for direct shipping licenses, but instead distinguishes between "small" or "large" wineries through the 30,000 gallon cap.

During floor debates, § 19F's sponsor summed up § 19F as follows: "[W]ith the limitations that we are suggesting in the legislation, we are really still giving an inherent advantage indirectly to the local wineries." Likewise, the state senator whose district included Massachusetts's then-largest winery explained his qualified support for § 19F by stating that "the agricultural industry here in Massachusetts is really strong and should be preserved. And we do this . . . because we produce these specialty goods, pick-your-own orchards and wineries." The senator had another concern — that the winery in question, which primarily produced fruit wine, "comes close to the 30,000 [gallon] production limit" for "small" wineries and would likely soon exceed it because "it's a winery that is growing . . . and certainly uses wholesalers in other states." The senator urged modifications to § 19F because "we should be promoting this kind of industry and not adopting regulations, however inadvertently, that might take away the advantage that the winery would have." The draft of § 19F was amended shortly thereafter to exempt non-grape fruit wine production from the 30,000 gallon cap, and that version was enacted.

To repeat, all wineries producing over 30,000 gallons of wine — all of which are located outside Massachusetts — can apply for a "large winery shipment license," which allows them to directly sell and ship wine to consumers, but only if "the winery has not contracted with or has not been represented by a wholesaler licensed under section 18 for the preceding 6 months." Mass. Gen. Laws ch. 138, § 19F(a). To the extent a choice is available at all, under § 19F(a), "large" wineries can either choose to remain completely within the three-tier system and distribute their wines solely through wholesalers, or they can completely opt out of the three-tier system and sell their wines in Massachusetts exclusively through direct shipping. They cannot do both. Wholesaler distribution is also the only way "large" wineries can distribute wines to retailers, including all Massachusetts restaurants and bars. To put it differently, "large" wineries cannot distribute directly to consumers except at the cost of giving up distribution to retailers. By contrast, "small" wineries can simultaneously use the traditional wholesaler distribution method, direct distribution to retailers, and direct shipping to reach consumers.

The practical effects of the distinctions Massachusetts has drawn are significant. In 2006, 637 U.S. wineries were "large" under § 19F(a)'s definition. They produced between 30,001 and over 100 million gallons per year and accounted for 98 percent of all wine produced in the United States. The thirty largest "large" wineries represented 92 percent of the national market, while the other 607 "large" wineries produced between 30,001 and 680,000 gallons per year, averaged slightly less than 60,000 gallons per year, and made up approximately 6 percent of the U.S. wine production market in 2006.

There were 4,713 "small" wineries in the United States in 2006, as the term "small" is defined by § 19F(b). Of these wineries, 1,780 — more than a third — produced less than one gallon per year and had virtually no market share. The remaining 2,933 "small" wineries accounted for 2 percent of the total annual wine production in the United States in 2006.

In 2007, there were thirty-one wineries in Massachusetts, all met § 19F(b)'s definition of "small," and approximately half of these wineries produced fruit wine in addition to or in lieu of traditional grape-based wines. Each produced between 200 and 24,000 gallons per year.

II. Whether § 19F Discriminates against Interstate Commerce

The Commerce Clause prevents states from creating protectionist barriers to interstate trade. See, e.g., Lewis v. BT Inv. Managers, Inc., 447 U.S. 27, 35, 100 S.Ct. 2009, 64 L.Ed.2d 702 (1980). Discrimination under the Commerce Clause "means differential treatment of in-state and out-of-state economic interests that benefits the former and burdens the latter," as opposed to state laws that "regulate[] evenhandedly with only incidental effects on interstate commerce," Or. Waste Sys., 511 U.S. at 99, 114 S.Ct. 1345 (internal quotation marks omitted). Plaintiffs bear the initial burden of showing discrimination. See Cherry Hill Vineyard LLC v. Baldacci, 505 F.3d 28, 33 (1st Cir. 2007) (citing Hughes v. Oklahoma, 441 U.S. 322, 336, 99 S.Ct. 1727, 60 L.Ed.2d 250 (1979)).

If plaintiffs meet their burden, then "a discriminatory law is virtually per se invalid . . . and will survive only if it advances a legitimate local purpose that cannot be adequately served by reasonable non-discriminatory alternatives." Dep't of Revenue v. Davis, 553 U.S. 328, 128 S.Ct. 1801, 1808, 170 L.Ed.2d 685 (2008) (citations omitted) (internal quotation marks omitted). The state bears the burden of showing legitimate local purposes and the lack of non-discriminatory alternatives, and discriminatory state laws rarely satisfy this exacting standard. See Camps Newfound/Owatonna, Inc. v. Town of Harrison, 520 U.S. 564, 581-82, 117 S.Ct. 1590, 137 L.Ed.2d 852 (1997).

We explain in more detail the arguments being made. Plaintiffs argue that Massachusetts's choice of 30,000 gallons as the demarcation point between "small" and "large" wineries, along with the production exception for fruit wine, has both a discriminatory effect and purpose. The discriminatory effect is because § 19F's definition of "large" wineries encompasses the wineries which produce 98 percent of all wine in the United States, all of which are located out-of-state and all of which are deprived of the benefits of combining distribution methods. All wines produced in Massachusetts, on the other hand, are from "small" wineries that can use multiple distribution methods. Plaintiffs also say that Section 19F is discriminatory in purpose because the gallonage cap's particular features, along with legislators' statements and § 19F's process of enactment, show that § 19F's true purpose was to ensure that Massachusetts's wineries obtained advantages over their out-of-state counterparts. Plaintiffs also argue that Massachusetts cannot meet its burden of justifying § 19F because the law neither advances the three-tier system nor effectively assists small wineries in ways that available non-discriminatory alternatives could not. Finally, in the alternative, plaintiffs contend that § 19F impermissibly burdens interstate commerce under Pike even if it is not discriminatory.

Massachusetts counters that § 19F is not discriminatory in effect because most "small" wineries are located out-of-state. It says this proves that § 19F disproportionately benefits out-of-state, not in-state, wineries, especially since there are far more "small" § 19F(b) wineries in the country than "large" § 19F(a) ones. Massachusetts argues that § 19F is not discriminatory in purpose because its aim is to level the economic playing field for all "small" wineries irrespective of where they are located, and the district court erroneously looked to comments by individual legislators, lobbyists, and intermediate steps in § 19F's process of enactment to find discriminatory purpose. Finally, Massachusetts says that § 19F poses no undue burden on interstate commerce under Pike and any such burden is surpassed by the local benefits of greater competition and consumer choice.

We explain below our reasons for rejecting Massachusetts's arguments. Because we hold that § 19F discriminates against interstate commerce, it is unnecessary for us to decide whether § 19F would also violate the Commerce Clause under Pike. See C A Carbone, Inc. v. Town of Clarkstown, 511 U.S. 383, 390, 114 S.Ct. 1677, 128 L.Ed.2d 399 (1994).

A. Section 19F is Discriminatory in Effect

A state law is discriminatory in effect when, in practice, it affects similarly situated entities in a market by imposing disproportionate burdens on out-of-state interests and conferring advantages upon in-state interests. Or. Waste Sys., 511 U.S. at 99, 114 S.Ct. 1345 (defining discrimination); see also Gen. Motors Corp. v. Tracy, 519 U.S. 278, 298, 117 S.Ct. 811, 136 L.Ed.2d 761 (1997) ("[A]ny notion of discrimination assumes a comparison of substantially similar entities.") (footnote omitted).

One such form of discrimination is plainly when "the effect of a state regulation is to cause local goods to constitute a larger share, and goods with an out-of-state source to constitute a smaller share, of the total sales in the market." Exxon Corp. v. Governor of Maryland, 437 U.S. 117, 126 n. 16, 98 S.Ct. 2207, 57 L.Ed.2d 91 (1978). State laws that alter conditions of competition to favor in-state interests over out-of-state competitors in a market have long been subject to invalidation. See, e.g., Hunt, 432 U.S. at 350-51, 97 S.Ct. 2434; Polar Ice Cream Creamery Co. v. Andrews, 375 U.S. 361, 376-77, 84 S.Ct. 378, 11 L.Ed.2d 389 (1964); Baldwin v. G.AF. Seelig, Inc., 294 U.S. 511, 519, 55 S.Ct. 497, 79 L.Ed. 1032 (1935); see also Baldacci 505 F.3d at 36 (explaining the doctrine); Houlton Citizens' Coalition v. Town of Houlton, 175 F.3d 178, 188-89 (1st Cir. 1999) (same).

Plaintiffs must present evidence as to why the law discriminates in practice. See Baldacci, 505 F.3d at 36-37. Here, the totality of the evidence introduced by plaintiffs demonstrates that § 19F's preferential treatment of "small" wineries that produce 30,000 gallons or less of grape wine is discriminatory. Its effect is to significantly alter the terms of competition between in-state and out-of-state wineries to the detriment of the out-of-state wineries that produce 98 percent of the country's wine.

Section 19F confers a clear competitive advantage to "small" wineries, which include all Massachusetts's wineries, and creates a comparative disadvantage for "large" wineries, none of which are in Massachusetts. "Small" wineries that obtain a § 19F(b) license can use direct shipping to consumers, retailer distribution, and wholesaler distribution simultaneously. Combining these distribution methods allows "small" wineries to sell their full range of wines at maximum efficiency be cause they serve complementary markets. "Small" wineries that produce higher-volume wines can continue distributing those wines through wholesaler relationships. They can obtain new markets for all their wines by distributing their wines directly to retailers, including individual bars, restaurants, and stores. They can also use direct shipping to offer their full range of wines directly to Massachusetts consumers, resulting in greater overall sales.

Combining these methods also lowers "small" wineries' distribution costs because they can choose which method or combination of methods will be most cost-effective for a particular wine. As the parties' briefs highlight, this produces important synergies. "Small" wineries' use of retail distribution increases brand recognition and makes wholesaler distribution more likely. Direct shipping can similarly increase consumer demand for a particular wine, increasing the prospects for further retail sales and wholesaler distribution.

Not surprisingly, Massachusetts's wineries have taken advantage of these benefits. Twenty-seven of Massachusetts's thirty-one wineries have obtained "small" winery licenses; in contrast, only twenty-six of the 2,933 out-of-state "small" wineries producing more than a gallon per year have done so. Massachusetts's wineries have also benefitted from their access to multiple distribution channels in practice. In 2007, the first year § 19F was in effect, Massachusetts's wineries distributed 29 percent of their annual production through wholesalers and 71 percent through retail out-lets, including direct shipping. See An Economic Snapshot of the Massachusetts Winery Industry.

The 637 out-of-state wineries that qualified as "large" under § 19F(a) in 2006 do not get these advantages and must instead choose between direct shipping and wholesaler distribution. Under § 19F(a), whether a "large" winery chooses wholesaler distribution or direct shipping, its choice carries a significant loss of potential profits, since using a single method results in a comparative loss of consumer sales. "Large" wineries also face comparatively greater distribution costs because they cannot always distribute a given wine through the most cost-effective method. And they cannot take advantage of the synergies that increase the net amount of demand for wines when multiple distribution methods are used together. These amount to considerable competitive disadvantages in an industry that Massachusetts's own evidence characterizes as one with indisputably slim profit margins and a highly competitive market.

Moreover, contrary to Massachusetts's assertions, § 19F does not level the playing field for all wineries unable to obtain consistent wholesaler distribution under the three-tier system. Section 19F's demarcation line between "small" and "large" wineries instead creates an especially acute competitive disadvantage for the wineries that are defined as "large" under § 19F(a) but which in practice face the same difficulties in distributing most of their wines as the "small" § 19F(b) wineries. Massachusetts's own evidence shows that only the largest 50 to 100 wineries can distribute most of their wines through wholesalers under the three-tier system. The remaining 537 or so "large" wineries each produce between 30,001 and 680,000 gallons per year of a mix of mass-market and boutique wines. In 2006, their percentage of the market share for wine production far exceeded that of § 19F(b) "small" wineries.

These smaller "large" wineries lose the most under the § 19F regime. Unlike the largest of the "large" wineries, which can distribute the vast majority of their wines through existing wholesale distribution, these smaller "large" wineries can only distribute a handful of their higher-volume wines through wholesalers. If they choose direct shipping, however, they are forced to terminate their existing wholesaler relationships, which also means that they lose all access to retailers in Massachusetts. Since this is a crucial way for a winery to build consumer awareness for the brand in Massachusetts, its unavailability means that these wineries are not able to compete on the same footing as § 19F(b) "small" wineries. Importantly, these are also the wineries that would otherwise be most competitive in the market for boutique wines: their size affords them otherwise considerable advantages in terms of marketing, volume, transportation, and brand recognition.

The ultimate effect of § 19F is to artificially limit the playing field in this market in a way that enables Massachusetts's wineries to gain market share against their out-of-state competitors. Section 19F(b)'s choice of a 30,000 gallon grape wine production cap helps Massachusetts wineries to improve their position in the market. At the same time, § 19F(a) burdens all the larger out-of-state competitors and impedes their ability to effectively use their natural advantages.

Massachusetts argues that there can be no discrimination because the favored "small" winery group created by § 19F(b) is almost entirely comprised of out-of-state wineries. Massachusetts claims this means that whatever the burden on out-of-state wineries under § 19F(a), § 19F(b) does not create an in-state benefit, since Massachusetts's "small" wineries are made no better off than their out-of-state counterparts. Without evidence of in-state benefits, Massachusetts concludes, the Supreme Court's decision in Exxon dictates that we find no discriminatory effect.

Massachusetts's argument ignores the effect of its statute. Section 19F(b)'s benefit to eligible "small" out-of-state wineries cannot be viewed separately from the much greater disadvantages that § 19F(a) imposes on out-of-state wineries. Massachusetts's wineries uniquely receive a net competitive gain under § 19F, while the law impairs out-of-state wineries' competitive position. It deprives "large" wineriesand especially those "large" wineries that have trouble obtaining wholesale distribution — of the competitive advantages of specialization and higher-volume production. These disadvantages exceed the benefits that out-of-state "small" wineries receive.

Exxon does not support Massachusetts's argument. Exxon held that a law that restricts a market consisting entirely of out-of-state interests is not discriminatory because there is no local market to benefit. 437 U.S. at 125-26, 98 S.Ct. 2207. Exxon is not apposite where, as here, there is an in-state market and the law operates to its competitive benefit. Massachusetts cannot apply Exxon only to "large" wineries as distinct from "small" wineries; the wine market is a single although differentiated market, and § 19F's two provisions operate on that market together.

Likewise, the fact that § 19F(b) benefits both in-state and some out-of-state "small" wineries does not prove that § 19F is non-discriminatory. We have previously rejected the notion that "a favored group must be entirely in-state for a law to have a discriminatory effect on commerce," reasoning that when a law burdened a group whose members were entirely out-of-state and benefitted a class whose members were largely but not wholly located in-state, it was still impermissibly discriminatory in effect. See Walgreen Co. v. Rullan, 405 F.3d 50, 59 (1st Cir. 2005).

B. Section 19F is Discriminatory in Purpose

We further hold that § 19F conferred a competitive advantage upon Massachusetts wineries by design.

In Alliance of Auto. Mfrs. v. Gwadosky, 430 F.3d 30 (1st Cir. 2005), we discussed the methodology for determining legislative purpose when a state statute is allegedly motivated by an intent to discriminate against interstate commerce. Under that methodology, we look to "the statute as a whole," id. at 37, including statutory text, context, and legislative history, but we also consider whether the statute was "closely tailored to achieve the legislative purpose" the state asserted. Id. at 38.

That § 19F discriminates against out-of-state wineries in its effects strengthens the inference that the statute was discriminatory by design. "'[L]ess deference to . . . legislative judgment is due . . . where the local regulation bears disproportionately on out-of-state residents and businesses.'" Id. at 39 (second and third alterations in original) (quoting Kassel v. Consol. Freightways Corp., 450 U.S. 662, 675-76, 101 S.Ct. 1309, 67 L.Ed.2d 580 (1981) (plurality opinion)); see also D.H. Regan, The Supreme Court and State Protectionism: Making Sense of the Dormant Commerce Clause, 84 Mich. L. Rev. 1091, 1144-47, 1206-45 (1986) (suggesting that the Commerce Clause is particularly concerned with deliberate discrimination, and that previous Supreme Court cases invalidating state statutes involved discriminatory effects in combination with, and as evidence of, discriminatory purpose); K.M. Sullivan G. Gunther, Constitutional Law 206 (16th ed. 2007).

As to statutory context, § 19F is a new addition to a provision that covers an array of alcohol licensing rules. While § 19 generally includes licensing rules for producers that are typical of the three-tier system, § 19F is one of a number of recently appended subsections that sets out special exceptions to that system for particular entities. See, e.g., Mass. Gen. Laws. ch. 138, § 19C (farmer-brewery licenses); id. § 19D (pub-brewery licenses); § 19E (farmer-distillery licenses). Many of these subsections were enacted for the express purpose of assisting Massachusetts's domestic industries, including but not limited to § 19B, § 19F's facially discriminatory and unconstitutional predecessor. Though § 19F contains no stated statutory purpose, its placement in a licensing law that grants exceptions to the three-tier system for the predominant purpose of benefitting local industry is pertinent evidence of discriminatory intent. Based on statements made by various Massachusetts legislators, it is also clear that Massachusetts intended to benefit its local wine industry, and that it did so in particular ways whose effects on out-of-state wineries could easily be foreseen.

The gap between Massachusetts's professed neutrality and § 19F's practical effects also underscores the conclusion of discriminatory purpose. See Hunt, 432 U.S. at 352, 97 S.Ct. 2434 (observing that the disparity between a law's asserted ends and its means was "somewhat suspect" and evidenced a likely discriminatory purpose).

Massachusetts has asserted various purposes behind § 19F: to facilitate direct shipment, to further the three-tier system, to make all small wineries, irrespective of their location, better able to compete, and to thereby provide Massachusetts consumers with greater choice. The 30,000 gallon cap and the fruit wine exception, Massachusetts claimed at oral argument, reflected the legislature's rational assessment of the kind of wineries that needed special assistance because they were suffering from the limitations of the three-tier system. But these general aims stand in stark contrast to § 19F's specific and highly irregular features.

The wine industry and federal law have developed definitions of "small," "medium," and "large" wineries in order to describe the way the industry produces and distributes wines and, in the case of federal law, to offer "small" wineries regulatory benefits. These definitions do not, of course, bind states to particular regulatory choices. But their lack of correlation to § 19F belies Massachusetts's claim that § 19F's features reflected an objective choice to remedy the purported competitive disadvantage faced uniquely by wineries producing 30,000 gallons or less of grape wine. That is particularly true given that this gallonage cap counts wineries as "small" even if they produce more than 30,000 gallons of wine when fruit wine production is counted. See Kassel, 450 U.S. at 675-78, 101 S.Ct. 1309 (questioning the legitimacy of the Iowa legislature's motives in enacting a statute that banned vehicles longer than 55 feet from using Iowa roads, when all other states in the West and Midwest had a 65-foot limit and the Iowa statute had a number of significant and irregular exceptions); Hunt, 432 U.S. at 350-52, 97 S.Ct. 2434 (finding a North Carolina apple-labeling law discriminatory in effect and, impliedly, in purpose when its requirements prevented Washington from using its apple-grading and labeling system, which had become the industry standard).

According to uncontested evidence in the record, the wine industry considers wineries that produce 120,000 gallons per year or less "small." "Medium" wineries produce between 120,000 and 600,000 gallons annually, and "large" wineries produce more than 600,000 gallons per year. The industry apparently does not differentiate between wineries that produce fruit as opposed to grape wine; relative size is the critical factor. The Federal Trade Commission largely adopted these definitions when it surveyed conditions of competition in the wine industry. See FTC Report at 6.

Nor, according to testimony from industry figures, does Massachusetts's 30,000 gallon demarcation point between "small" and "large" wineries correspond to the ability of the winery to obtain wholesaler representation. To the contrary, this choice prevents out-of-state, smaller "large" wineries from competing on equal terms with Massachusetts's "small" wineries even though these wineries faced similar difficulties in obtaining wholesaler distribution under the three-tier system.

Massachusetts's claim at oral argument that its definition of "small" wineries targets those wineries in need of competitive assistance also diverges considerably from the definitions the federal government and other states have developed for this same broad purpose. As we have said, there is no relationship to those wineries who are able or unable to obtain wholesalers. Beyond that, as a matter of federal tax policy, wineries producing 250,000 gallons or less of any type of wine, and not merely wineries that produce less than 30,000 gallons of grape wine per year, are deemed "small wineries" in need of competitive assistance in the form of a substantial tax break. See Alcohol Tobacco Tax Trade Bureau, Dep't of the Treasury, TTB Compliance Seminar for Bonded Wine Premises 73-74 (2008), available at http://www.ttb.gov/pdf/compliance-seminar.pdf (hereinafter TTB Compliance Seminar). No other state has defined a "small" wine producer and attached the same consequences to this definition as Massachusetts has. And no other state counts gallonage by excluding all fruit wine production; "wine" in these other states means wines made from any fruit or other agricultural product. See Ariz.Rev. Stat. Ann. § 4-101(36); Ark. Code Ann. § 3-5-202(4); Fla. Stat. § 564.01; Ky.Rev. Stat. Ann. § 241.010(55); Ind. Code 7.1-1-3-49; Ohio Rev. Code Ann. § 4301.01(3). Section 19F's definition of a "small" winery does not even correspond to the way Massachusetts previously classified wineries by size for the purpose of calculating a licensing fee.

Section 19F's unusual regulatory features do track one thing precisely: the unique attributes of Massachusetts's own wine industry. All of Massachusetts's thirty-one wineries are eligible for "small" winery licenses. All fall neatly within the 30,000 gallon cap, producing between 200 gallons and 24,000 gallons annually. And the record demonstrates — and Massachusetts does not contest — that legislators were well aware of these figures.

The fact that this gallonage cap excludes wines made from fruits other than grapes, no matter how many gallons a winery produces per year, is particularly probative. In past years, Massachusetts's largest winery produced more than 30,000 gallons of wine annually because between half and three-quarters of its production came from apple wines. The main effect of the fruit wine exception was to guarantee that this winery, like all other Massachusetts wineries, could take advantage of § 19F(b)'s beneficial distribution rules for "small" wineries. Massachusetts has offered no other explanation for the fruit wine exception, and there is no obvious reason why it would serve § 19F's ostensible purposes. This exception, like similar, facially neutral statutory exemptions apparently motivated by a desire to shield in-state interests, "weaken[s] the presumption in favor of the validity of the [general provision], because [it] undermine[s] the assumption that the State's own political processes will act as a check on local regulations that unduly burden interstate commerce." Raymond Motor Transp., Inc. v. Rice, 434 U.S. 429, 447, 98 S.Ct. 787, 54 L.Ed.2d 664 (1978).

We conclude that § 19F altered the competitive balance to favor Massachusetts's wineries and disfavor out-of-state competition by design.

C. Lack of Legitimate Local Purpose and Availability of Reasonable Non-Discriminatory Alternatives

Because plaintiffs have shown that § 19F discriminates against interstate commerce, Massachusetts bears the heavy burden of showing that the statute is nonetheless constitutional because it serves a legitimate local purpose that cannot be attained through reasonable non-discriminatory alternatives. Dep't of Revenue, 128 S.Ct. at 1808. The state can only carry this burden by presenting "concrete record evidence," and not "sweeping assertion[s]" or "mere speculation," to substantiate its claims that the discriminatory aspects of its challenged policy are necessary to achieve its asserted objectives. Granholm, 544 U.S. at 492-93, 125 S.Ct. 1885; see also Chem. Waste Mgmt., 504 U.S. at 342, 112 S.Ct. 2009. Massachusetts has not even attempted to do so here. Because the constitutionality of a state statute is involved, we nonetheless consider the issue.

The record shows that at least one viable non-discriminatory alternative existed when § 19F was under consideration: the Model Direct Shipment Bill, which the National Conference of State Legislatures adopted in 1997. The Model Bill does not define "small" or "large" wineries or regulate access to licenses depending on winery size. As an alternative to § 19F, then-Governor Romney proposed a version of the Model Bill which would have allowed all wineries to ship directly to consumers, sell to retailers, and distribute through wholesalers. But the state legislature rejected this proposal and overrode his veto.

Plaintiffs argue that this alternative would have helped small wineries without undercutting the three-tier system because it included limitations on the total volume wineries could ship to consumers. What ever the merits of this proposal, Massachusetts has never claimed it would be unworkable. Under similar circumstances, the Supreme Court has, as a rule, struck down the discriminatory state law in question. See Granholm, 544 U.S. at 491-92, 125 S.Ct. 1885; Camps Newfound/Owatonna, 520 U.S. at 582 n. 16, 117 S.Ct. 1590.

III. Whether the Twenty-first Amendment Immunizes Facially Neutral Alcohol Statutes from Commerce Clause Scrutiny

We now consider whether, as Massachusetts asserts, the Twenty-first Amendment protects § 19F from invalidation, notwithstanding the fact that it discriminates against interstate commerce in purpose and effect.

Whether the Twenty-first Amendment granted states the authority to enact even facially neutral but discriminatory alcohol laws that would otherwise violate the Commerce Clause was not decided by Granholm and the answer is not readily apparent from the text of the Amendment. Granholm holds the interpretation of this amendment instead turns on historical context. Section 2 of the Twenty-first Amendment granted the states the authority to regulate liquor only to the extent that they had done so before Prohibition under two federal laws: the Wilson Act of 1890 and the Webb-Kenyon Act of 1913. See Granholm, 544 U.S. at 484, 125 S.Ct. 1885.

The Supreme Court held in Granholm that through these Acts, Congress gave the states newfound powers to regulate alcohol that came within their borders, even if it had traveled in interstate commerce. The Wilson Act did this by allowing states to restrict or prohibit the sale of out-of-state alcohol "to the same extent and in the same manner" as alcohol that was produced in-state. 544 U.S. at 478, 125 S.Ct. 1885 (quoting 27 U.S.C. § 121) (internal quotation marks omitted). The Webb-Kenyon Act expanded states' regulatory authority by expressly authorizing states to regulate alcohol that traveled in interstate commerce even if it was being shipped solely for consumers' personal use. Id. at 481-84, 125 S.Ct. 1885. These Acts did not, however, exempt states from the Commerce Clause's existing prohibitions on state laws that discriminated against out-of-state goods and favored local interests. Id. at 484-85, 125 S.Ct. 1885.

The precise question in Granholm was what effect, if any, the Twenty-first Amendment has upon facially discriminatory state alcohol laws that would otherwise be subject to invalidation under the Commerce Clause. 544 U.S. at 471, 125 S.Ct. 1885. The question of whether the Twenty-first Amendment protects facially neutral laws like § 19F was not before the Court.

Massachusetts now contends that the Twenty-first Amendment protects facially neutral laws from invalidation under the Commerce Clause, even if they discriminate in purpose or effect, because it says such laws are distinguishable from facially discriminatory laws for the purposes of the Twenty-first Amendment. In the alternative, Massachusetts asserted at oral argument that the Twenty-first Amendment should lessen Commerce Clause scrutiny of such laws to mere rational basis review.

We reject these arguments. Based on our analysis of historical sources, we conclude that the Wilson and Webb-Kenyon Acts did not protect facially neutral state liquor laws from invalidation under the Commerce Clause if they were discriminatory. To hold otherwise, we would have to find that these Acts not only recognized the difference between facially discriminatory and facially neutral but discriminatory state laws, but also affirmatively intended to protect the latter and not the former. All evidence points to the contrary.

By the time the Wilson Act became law in 1890, it was well established that under the Commerce Clause, facially neutral state statutes that had a discriminatory effect on out-of-state interests constituted impermissible discrimination, just as facially discriminatory state laws did.

The Supreme Court had decided two major discriminatory effects Commerce Clause cases just before the Wilson Act passed. In Robbins v. Taxing Dist. of Shelby County, 120 U.S. 489, 7 S.Ct. 592, 30 L.Ed. 694 (1887), the Court had invalidated a facially neutral state tax on "drummers," individuals who "drummed up" sales by displaying samples, because, inter alia, the tax disproportionately disadvantaged out-of-state merchants and manufacturers. Id. at 490-91, 497-98, 7 S.Ct. 592. And in Minnesota v. Barber, 136 U.S. 313, 10 S.Ct. 862, 34 L.Ed. 455 (1890), the Supreme Court had invalidated a Minnesota statute that required in-state inspection of all meat before it could be sold within the state. Id. at 326, 10 S.Ct. 862. Its reasoning cut broadly: "Although this statute is not avowedly or in its terms directed against the bringing into Minnesota of the products of other states," this was the statute's "necessary effect." Id.

In a separate line of cases, the Supreme Court had also indicated that a state's asserted rationale for a statute would be viewed with skepticism if other evidence, including the statute's effects, pointed strongly to a discriminatory purpose. "[I]f the State, under the guise of exerting its police powers, should make such exclusion or prohibition applicable solely to articles, of that kind, that may be produced or manufactured in other States," the Court stated as early as 1879, "the courts would find no difficulty in holding such legislation to be in conflict with the Constitution of the United States." Guy v. City of Baltimore, 100 U.S. 434, 443, 25 L.Ed. 743 (1879); see also Austin v. Tennessee, 179 U.S. 343, 349-50, 21 S.Ct. 132, 45 L.Ed. 224 (1900) (suggesting that ostensibly neutral laws that were intentionally applied in a discriminatory manner were invalid in the Commerce Clause context).

When drafting the Wilson and Webb-Kenyon Acts, Congress was presumably aware that these types of facially neutral but discriminatory state laws were subject to invalidation under the Commerce Clause. See Edelman v. Lynchburg Coll., 535 U.S. 106, 117 n. 13, 122 S.Ct. 1145, 152 L.Ed.2d 188 (2002); see also N. Star Steel Co. v. Thomas, 515 U.S. 29, 34, 115 S.Ct. 1927, 132 L.Ed.2d 27 (1995). Yet Congress made no reference to the notion that the Wilson and Webb-Kenyon Acts would permit states to enact liquor laws with a discriminatory effect or motive. Although "Congress may authorize the States to engage in regulation that the Commerce Clause would otherwise forbid," courts can "exempt[] state statutes from the implied limitations of the Clause only when the congressional direction to do so has been unmistakably clear." Maine v. Taylor, 477 U.S. 131, 138-39, 106 S.Ct. 2440, 91 L.Ed.2d 110 (1986) (internal quotation marks omitted). The Wilson and Webb-Kenyon Acts do evince an unmistakably clear intention to permit states to regulate alcohol which traveled in interstate commerce the same way as they regulated in-state alcohol. But the two Acts cannot be construed to authorize anything more.

Supreme Court decisions and legal scholarship of the era confirm this interpretation. Scott v. Donald, 165 U.S. 58, 17 S.Ct. 265, 41 L.Ed. 632 (1897), involved a challenge to a state law that gave the state liquor commissioner control over all state sales of alcohol and included two other provisions that explicitly disfavored out-of-state manufacturers. Id. at 92, 17 S.Ct. 265. The Court compared the facts to other Commerce Clause cases, including various discriminatory effects cases involving goods other than alcohol, implying that alcohol regulation was not a unique category for the purposes of the non-discrimination rule. Id. at 93-99, 17 S.Ct. 265. The Court's ultimate holding was that "[the Wilson Act] was not intended to confer upon any state the power to discriminate injuriously against the products of other states." While states, under the Wilson Act, could enact laws to "forbid entirely the manufacture and sale of intoxicating liquors," they "cannot . . . establish a system which, in effect, discriminates between interstate and domestic commerce." Id. at 100, 17 S.Ct. 265.

Contemporaneous treatises on liquor law likewise concluded that the Wilson Act did not immunize any kind of discriminatory state law from scrutiny under the non-discrimination rule.

Against this background, we hold that the Twenty-first Amendment does not exempt facially neutral state alcohol laws with discriminatory effects from the non-discrimination rule of the Commerce Clause. Nor, of course, are such laws exempt when they also discriminate by design.

We also reject Massachusetts's alternate contention that the Twenty-first Amendment lessens the degree of Commerce Clause scrutiny for facially neutral but discriminatory state alcohol laws to mere rational basis review. The Supreme Court implicitly rejected this argument in Granholm when it applied the usual, searching degree of scrutiny to invalidate the facially discriminatory laws at issue. 544 U.S. at 489-90, 125 S.Ct. 1885. And there is nothing in the text, legislative history, or contemporaneous understandings of the Wilson or Webb-Kenyon Acts that supports Massachusetts's argument, let alone yields an unambiguous indication of congressional intent to reduce Commerce Clause scrutiny. In the absence of such evidence, Massachusetts's interpretation of the Twenty-first Amendment fails.

Finally, we need not address whether § 19F could escape invalidation on the ground that, despite its discriminatory effect and design, the "core purposes" of the Twenty-first Amendment "are sufficiently implicated . . . to outweigh the Commerce Clause principles that would otherwise be offended." Bacchus, 468 U.S. at 275, 104 S.Ct. 3049. Those purposes include "promoting temperance, ensuring orderly market conditions, and raising revenue." North Dakota, 495 U.S. at 432, 110 S.Ct. 1986. Massachusetts does not present any argument as to why § 19F serves any of these purposes. In any event, it is unclear that this balancing test survives Granholm.

IV.

We affirm the judgment of the district court.

So ordered.