The Problem

For their premium terroir sake, the brewers need the right kind of rice. More specifically, they need (a) sake rice (far more difficult to grow than table rice) (b) of a variety optimized for their peculiar location (and, being an unusual variety, one without well-established best practices), (c) grown to the exacting standards required for premium sake, (d) but which standards are ones that the brewer can observe but often cannot specify at levels sufficiently verifiable for explicit contracting.

Ethnic Groups

The brewers need farmers whom they can trust—farmers on whom they can rely to produce rice at quality levels higher than that for which they could draft verifiable contract terms. To cultivate that trust, the farmers and brewers will choose to draw on those aspects of their respective reputations that convey information about their own trustworthiness. The modern iteration of this literature began with Stewart Macaulay (Reference Macaulay1963), of course. In his famous study of Wisconsin business practice, he (1963, 58) quotes an attorney who was “sick of being told, ‘We can trust old Max.’” His businessmen dealt with people they knew and trusted (1963, 63).

To increase the information about the trustworthiness of possible contract partners, some traders in some industries focus their trades within distinctive religious, ethnic, or familial lines. Within these densely intertwined communities, information about contractual trustworthiness travels. And to the extent that it travels, contractual partners learn about each other and trade on that reputation for probity. Modern scholars typically illustrate the informational nexus between reputations and ethnic ties through several sets of articles. Three of the better known involve those of Janet Landa, Lisa Bernstein, and Avner Greif. Landa (Reference Landa1981) examines Chinese middlemen in the Southeast Asian rubber market. There, she finds trade dominated by several clans bound together through “a tightly knit kinship structure” and “linked in complex networks of particularistic exchange relations” (350). She and Robert Cooter describe them as “ethnically homogenous middlemen groups” (1984, 15–16). They traded, Landa and Cooter continue, within “a repository of trust which reduces the probability of breach on a contract between insiders” (16). Because of their personal ties, they were able “to rely upon informal means of enforcement of contracts” (16).

Bernstein (Reference Bernstein1992) studies the New York diamond market. There, she observes that “the diamond industry has long been dominated by Orthodox Jews” (141). Their ethnic ties, she posits, gave them both access to relatively high levels of information about their contracting partners and an ability to punish those who reneged on their promises. The Orthodox Jewish community provided, she explains, “geographical concentration, ethnic homogeneity, and repeat dealing” (140). Within this world, traders organized their industry “to minimize the cost of obtaining information about dealers’ reputations” (133).

Greif (Reference Greif2012, 445) examines the agency contracts among eleventh-century Jewish Maghribi traders in the Mediterranean. These traders “deterred opportunism in bilateral agency relations [through] a credible threat of losing future profitable relations in the traders’ broader community,” writes Greif (445). The traders could rely on this informal punishment strategy because they traded through “coalitions,” Greif contends; Bernstein (2019) more plausibly argues that they could punish opportunism because they traded within “small-scale networks.”

Social Capital as an Ethnic Substitute

Landa, Bernstein, and Greif study the reputations of merchants located within strongly bonded ethnic groups. In Japan, brewers and farmers choose to replicate some of the effect of those ethnic connections through dense and deliberately cultivated networks of social ties. The concept is that of “social capital.” Political scientist Robert Putnam (Reference Putnam1995, Reference Putnam2000) popularized the concept two decades ago when he argued that that America suffered from its decline. As Putnam articulates the idea, mid-twentieth-century Americans had carefully built and maintained ties with each other. They had helped the PTA, joined the Lions Club, attended churches and synagogues, and coached little league. Now, they did so only haphazardly. They no longer lived within webs of intersecting relationships, and suffered for that loss.

Putnam had not invented the term, and sociologists quickly noted their own antecedents in Durkheim and Marx (Portes Reference Portes1998, 2). They cited more recent discussions by Pierre Bourdieu (Reference Bourdieu and Richardson1985) and James Coleman (Reference Coleman1988, Reference Coleman1990). As Coleman (Reference Coleman1988, 1990; see Burt Reference Burt2000, 351) explains it, people can hold each other most closely to communal norms—people can most reliably trust each other—when they maintain a network of overlapping ties dense enough to create “closure” (Coleman Reference Coleman1988, S105–07). Suppose, explains Coleman (Reference Coleman1988), that members in one society maintain ties with other members that do not intertwine. A knows B and C, but B and C do not know each other and neither do they know anyone else in common. Absent those mutual connections, they find it harder to punish people who break norms.

By contrast, if people live within a network of intertwined relationships, their world is “closed.” Should anyone deviate from the norms (e.g., should a farmer sell substandard rice to a brewer), many in the network will know, and many can punish (Coleman Reference Coleman1988, S106). “In a structure with closure,” Coleman continues, “B and C can combine to provide a collective sanction, or either can reward the other for sanctioning A.” As Ronald S. Burt explains, “network closure facilitates sanctions that make it less risky for people in the network to trust one another” (Reference Burt2000, 351). It “enables local monitoring,” write Hillmann and Aven (Reference Hillmann and Aven2011, 486). It helps “ensure that members know how their exchange partners behaved in the past, whether the behavior complied with community norms or not, and if these partners should be trusted in future transactions” (486).

Vertical Integration

Rather than buy their rice, the brewers could—hypothetically—grow it themselves. In the language of the law, they could acquire their rice by contract, or integrate vertically into rice production. In the language of economics workshops, they could either “make or buy.” As early as 1937, Ronald Coase noted that rational firms will make component parts internally when the costs of contracting across the market exceed the costs of administering a transaction within the firm (Coase Reference Coase1937).

Coase left open the question of what those market and administrative costs might be. During the 1980s and 1990s, scholars focused on whether a transaction involved assets whose value was specific to the transaction. Such a “transaction-specific investment” would generate “quasi rents,” they explained. Should the two parties try to contract for the assets across the market, those rents would create incentives for them to try to “hold up” each other (for example, Klein, Crawford, and Alchian Reference Klein, Crawford and Alchian1978; Williamson Reference Williamson1985). Unfortunately, the question lacked an extensive application. Scholars had initially focused primarily on one historical example (the GM-Fisher Body merger). That example, however, turned on factors later shown to be entirely orthogonal to the “transaction-specific asset” hypothesis (Coase Reference Coase2000; Casadesus-Masanell and Spulber Reference Casadesus-Masanell and Spulber2000). When scholars did find other applications, many of them involved clearly nonstandard venues like government procurement (for example, Masten, Meehan, and Snyder Reference Masten, Meehan and Snyder1991).

In his own theory of the firm, Oliver Hart (Grossman and Hart Reference Grossman and Hart1986; Hart and Moore Reference Hart and John1990) suggests that assets should be owned by the institution whose involvement raises the value of the asset to its highest use. As Halonen-Akatwijuka (Reference Halonen-Akatwijuka2016) summarizes the argument:

If the land is most valuable when used for the production of sake, it should be owned by the entity whose role is most important in producing that sake.

In fact, however, Klein, Williamson, and Hart may be selling lawyers (or the human imagination more generally) short. People can replicate much of the administrative immediacy of a single firm by contract. Writing in this journal, Thomas Dietz (Reference Dietz2012, 44) notes that German companies negotiating software procurement contracts across national boundaries did the same: They used reputational mechanisms, social capital, and contracts with short-term termination options to obtain much the same control that vertical integration would give.

Efficiency Wages

When transacting across the market, a firm can increase the incentive of its contracting partner to keep its word by paying it extraordinarily high prices. Obviously, firms do not generally increase profitability by raising the prices they pay. They raise profitability through higher prices only if those prices induce a contracting partner to so increase its reliability that the firm can produce in a way otherwise not possible.

Consider it the contractual analogue to the concept of “efficiency wages.” Sometimes, an employer can raise profitability by raising wages. It can do so when those higher wages raise productivity by an amount that more than offsets the higher wage costs. The classic example is Henry Ford. He had conceived a new manufacturing technique: the assembly line. To make the technique work, he needed workers who would stay on the job. He needed workers who would accumulate experience, and then draw on that experience to make the hundreds of changes necessary to make the assembly line work. As long as he paid the going rate in Detroit—$2.50 per day—workers quit as they pleased. Ford doubled the pay to $5.00, and workers now stayed on the job. They studied the assembly line and made it work (Raff and Summers Reference Raff and Summers1986, 1; see also Shapiro and Stiglitz Reference Shapiro and Stiglitz1984).

The Drink

At least some version of sake dates to prehistory. Migrants from Korea or Sakhalin (no one knows which) brought wet rice agriculture to Japan sometime around the third century BCE. Probably Japanese began to drink a fermented rice drink soon thereafter.

In the two millennia since, sake has played a prominent role in Japanese culture. Seventh-and eighth-century poets sang its praises. The fifteenth-century bakufu elites paired the “way of tea” with a Zen-ish “way of sake.” No one knows how it tasted back then, but everyone drank it. Farmers drank it, merchants drank it, swordsmen drank it.

Over time and across the country, the brewers doubled as bankers. Fans of classic cinema will remember Mifune playing off the village brewer against the village brothel owner in Kurosawa’s 1961 Yojimbo. In premodern villages, the sake brewers were often among the richest. They financed paddies, irrigation networks, and entrepreneurship more generally.

Sake brewing is easy to describe, but uncommonly hard to execute well. Through the fermentation process, yeast turns sugars into alcohol. Yet rice contains no sugar. To create the sugar, brewers need first to add mold (koji) to a small amount of steamed rice (koji mai). The mold then transforms the starch in the kernel into a sugar. To this starter, the brewers add additional rice (shubo mai, or moto mai) and yeast (kobo). This creates a culture (shubo), to which the brewers will add more mold, steamed rice (kakemai), and water. They take this last step multiple times. They precipitate out the solid particles. And they typically close the process by heating the mixture to kill the active ingredients.

The Industry

Before the 1868 Meiji coup, most sake production was local. Most major towns (and many small ones) had their local brewer. Out of the local rice, the brewers brewed the sake for their local customers.

For a few years after the coup, production remained local. In 1876, brewers operated 26,000 breweries (Ninomiya Reference Ninomiya2012, 58; Suzuki Reference Suzuki2015, 71). Soon, however, this changed—as the incipient transportation network allowed the most efficient producers to exploit economies of scale. By 1899, nearly half the breweries had disappeared, and only 15,000 remained. By 1942, the number had fallen to 7,000, and by 1958 to 4,000. In 2017, barely 1,378 still survived (Osake Sutatisutikkusu n.d.).

Today’s 1,378 breweries vary enormously. The 11 largest each brew more than 10,000 kl of sake. Together, they employ over 4,300 employees and produce 47.5 percent of all sake (2017 data). Another 838 firms brew less than 100 kl each. Together, they employ 7,600 people and produce only 5.1 percent of all sake. Indeed, 830 of the breweries are seasonal firms that run only during the winter (Kokuzeicho 2018c) (fiscal 2018 survey).Footnote ²

Most of the biggest brewers operate from the Nada area of Kobe or the Fushimi area of Kyoto (see Figure 1). Now a center of 1.5 million and the capital of Hyogo prefecture, Kobe also serves as port to the massive city of Osaka to its east. Nada lies along the eastern coast of Kobe, where the brewers enjoy access to an unusually clear stream of underground water. They traditionally bought their rice inland, brewed it with the local water, and—even during the Tokugawa period—shipped the sake to ports as far away as the Edo (now Tokyo) capital (Ninomiya Reference Ninomiya2012, 51–53).

FIGURE 1.

Japan.

Source: https://www.freemap.jp/itemFreeDlPage.php?b=japan&s=japan2.

Nada and Fushimi remain the base for the very largest producers. In Table 1, I give the market shares of the ten largest modern firms. Of these ten, only Koyama Honke does not operate from either Nada or Fushimi.

Table 1.

Largest Breweries, 2018

Note: Premium—Junmai ginjo

Sources: Yoshio Ono, ”Shuzogyo keieisha no henkaku kodo [Transformation of Brewery Managers].” Shiga daigaku keizaigakubu kenkyu nenpo 26 (Reference Ono2019): 13–38; ”Shuin naki fushin ni ochiitta seishu shijo [The Sake Market That Is Thrown into Instability without Reason].” Shurui shokuhin tokei geppo, February 2019.

The Rice

Japanese farmers grow about 8.6 million tons of rice a year (Norin suisan sho n.d.). Most of this they grow as table rice, and for 35.9 percent of the table rice use the variety known as koshihikari. It is an enormously popular rice. Even the second and third most popular varieties are koshihikari derivatives. Farmers grow the largest amount of table rice in Niigata prefecture along the Japan Sea coast (K.K. Kaneko shoten 2016; Okome no chishiki 2016).

For their high-quality sake, brewers use more expensive rice varieties optimized for the drink. This sake rice is not a different species. Like the table rice, it still falls within the Japonica rice used in Japan, Korea, and China. But within Japonica, the varieties that farmers grow specifically for sake will not double for food. Sake rice simply is not a grain that consumers would choose to eat (Suzuki and Takada Reference Suzuki and Osamu2017, 140).

For their very best sake, brewers use only this sake-optimized rice. Difficult to raise and producing a smaller yield, sake rice costs substantially more than table rice. Because of this cost, brewers sometimes skip the sake rice entirely for their cheapest sake. Given that the taste of the sake depends most critically on the first steps in the process, even brewers who use some sake rice may use it only for the initial starter (as koji mai) and culture (as shubo mai). They then add cheaper table rice for the rest of the process (as kake mai; this is 70 percent of the total) (Koike Reference Koike1995, 162; Maikoku kiko 2014, 12).

In 2018, Japanese farmers grew only 93,000 tons of sake rice. Sake rice has a relatively large kernel (which brewers need in order to remove the outer layers that muddy the taste). Sake rice will absorb the mold (koji) into the heart of the kernel more readily than table rice, and will often contain a large starchy (shimpaku) core. It will have low levels of fat and protein (Koike Reference Koike1995, 161; Kadono et al. Reference Kadono2011; Daiginjo jidai no ishizue 2014, 2; Saito Reference Saito2015, 2).

Yamada nishiki is the koshihikari of sake rice—indeed, 36.8 percent (34,059 tons) of all sake rice (Norin suisan sho 2018). It is particularly hard to grow. Developed in the 1930s, it is unusually susceptible to disease, and with a long stem is especially vulnerable to late season typhoons.

Yet where koshihikari grows well across most of the country, yamada nishiki thrives only in a few areas. By latitude, the main islands of the Japanese archipelago stretch roughly from San Diego to Portland. The northern island of Hokkaido contains enormous ski resorts; the southern island of Kyushu was once a semitropical tourist destination. The island chain is relentlessly mountainous, and the area near Nagano hosted the 1998 winter Olympics. Rain falls massively on the western side as the winds sweeping off Siberia pick up moisture over the Japan Sea and then drop it when they hit the mountains. The eastern side is drier. Given that the mountains are essentially volcanic, soil content varies widely.

Yamada nishiki grows best in Hyogo prefecture (not coincidentally, near the Nada brewers). It needs moderate temperatures—ruling out both the northeast and the south. It does best in valleys (ideally, positioned east to west), with good sun exposure, with a large temperature variation from day to night, with good drainage, and with a clay base (Takeyasu et al. Reference Takeyasu2018, 69). More than any other prefecture, Hyogo offers these conditions and in 2018 Hyogo farmers grew 58.5 percent of the yamada nishiki crop (Norin suisan sho 2018).

Outside of Hyogo, farmers grow other varieties of sake rice. At 21,000 tons (22.7 percent of the total), gohyakuman goku is the next most common sake rice. Farmers in Niigata prefecture produced 46.1 percent of the variety. Miyama nishiki (6,000 tons, 6.9 percent), of which Nagano farmers produced 61.0 percent, is third (Norin suisan sho 2018).

Relative Decline

As the twenty-first century opened, Japanese were drinking less alcoholic beverages. In 1990, they had drunk 100.1 liters per adult capita per year.Footnote ³ By 2000, they drank only 95.5 liters, and by 2010 only 82.3. Middle-aged men and women drink more than younger men and women, but they also drink more than the elderly. As the Japanese population aged, the fraction in middle age declined, and with it fell per capita alcohol consumption (Kokuzeichio 2018b).

And sake accounted for less of what Japanese were drinking. For over a century, they had been steadily switching from sake to beer (Table 2). In 1902, they had drunk 92 percent of their alcohol as sake, and only 2 percent as beer. By 1952, they drank 31 percent as sake and 34 percent as beer. By 2000, they drank only 10 percent of their alcoholic beverages as sake and 71 percent as beer.Footnote ⁴ Some urban professionals were beginning to drink wine, and some were returning to the traditional drink of shochu (once a distilled rotgut for impoverished peasants and factory hands) as a mark of retro-cool. Having peaked in 1973 at 1.4 million kl, annual sake consumption fell by 2012 to 439,000 kl (Maikoku 2014, 2; Ono Reference Ono2019, 13).

Table 2.

Consumption of Alcoholic Beverages

*The total volume of sake consumed peaked in 1973.

Sources: Fujio Oana, “Shuzogyo no suii [Trends in the Brewing Industry].” Jokyo 65 (1970): 307–11; Kokuzeicho no “sake no shiori” … [The “Guide to Sake” from the National Tax Office …], Hatena blog, November 9, 2014. http://longlow.hatenablog.com/entry/20141109/p1.

The decline hit brewers everywhere. As noted earlier, across the country as a whole the number of brewers fell from 4,000 in 1958 to 1,378 in 2017. In Nada itself, the number fell from 67 breweries in 1973 to 29 firms in 2011. In 1973, those 67 had made 582,000 kl of sake. In 2011, the remaining 29 made only 163,000 kl (Nihon seisaku toshi ginko 2013).

Historical Context

The Nada-Fushimi Response

Deregulation would eventually come, but only after sake brewers (in response to the regulation) had so badly depreciated sake quality as to alienate most of their sophisticated potential customers. It was not until 1969 and 1974 that the government finally dismantled the regulatory strictures on the sake and rice markets (Ito Reference Ito1981; Ozeki Reference Ozeki1996, 425–26; Ninomiya Reference Ninomiya2015, 492; Suzuki Reference Suzuki2015, 199; Sando Reference Sando2018, 2019; Ono Reference Ono2019, 22, 25). Over the four preceding decades, brewers had learned to weather regulatorily created rice shortages. Now they needed to survive in a competitive market. To that goal, they would find their learned strategies badly mismatched.

The brewers had focused almost entirely on making sake as cheaply as possible. In 1922, a chemist had developed a synthetic sake that used very little (if any) rice (Ishiguro Reference Ishiguro2015a; Ninomiya Reference Ninomiya2015, 472; Ono Reference Ono2019, 23). With this technology, brewers had developed what would become the infamous “treble sake.” Although they started with rice, mold, and water, they added a variety of other ingredients that let them triple the amount of drink they could produce from a given quantity of rice (Yamakata Reference Yamakata1977, 344; Ishiguro Reference Ishiguro2015a; Ninomiya Reference Ninomiya2015, 472, 482; Ono Reference Ono2019, 23).

Synthetic sake remains on the market, and is often used for cooking. The treble-sake drink it spawned constituted a large part of the market in the 1950s and only disappeared in 2006. It had been the very worst sake. By 2006 too few people wanted to drink it to warrant production anymore (Nishinomiya shuzo Reference Nishinomiya shuzo1989, 264–65; Ishiguro Reference Ishiguro2015b, Reference Ishiguro2015c).

In this desultory market, the Nada and Fushimi brewers did have name recognition. When during the 1950s and 60s they lacked high-enough rice quotas to meet the market demand, they simply filled the resulting shortage by buying additional sake. They bought it from less well-known regional brewers in bulk and resold it under their own names (Ono Reference Ono2019, 24). They did not buy this bulk sake from brewers with whom they maintained long-term relations. Instead, they merely bought the available sake on the spot market: One brewery needed more sake; another had more than it could sell. In 1975, the Nada giant Hakutsuru brewed 109,000 koku (1 koku = 180 l) of sake; it sold 404,000 (Yamakata Reference Yamakata1977, 390; Ninomiya Reference Ninomiya2015, 488).

Note the obvious implication of these bulk sales: By midcentury, cheap sake was everywhere the same. By deliberately controlling the chemical transformations involved—standardizing yeast quality, introducing lactic acid, adding alcohol, and scientifically controlling quantities and timing—brewers had eliminated nearly all random variation. In the course of shifting to mechanized and sanitized processes, they had cut most of the variation not only within their own output, but from brewer to brewer as well (Ninomiya Reference Ninomiya2015, 473, 490).

Although the government deregulated much of the rice and sake markets in the early 1970s, the Nada-Fushimi brewers continued to position their sake as a low-cost mass-market beverage. They focused first on price. Japan had not been rich in 1960, and they had positioned their drink as a low-cost beverage for ordinary Japanese. By the 1970s, Japanese were no longer poor, but the mass-market brewers continued to make the same drink.

In fact, the Nada-Fushimi brewers kept this strategy through the end of the century. The number two Gekkeikan firmly controlled a solid presence as the Budweiser of Japanese sake: ubiquitous, cheap, and defiantly working class (see generally Kawaguchi and Fujimoto Reference Kawaguchi and Masayo2007). The number four Ozeki found its own niche for cheap sake in a one-drink-quantity bottle that it distributed the country over: “One Cup Ozeki,” they called it. In 1965, One Cup was 0.9 percent of Ozeki output. In 1967, Ozeki loaded the One Cup into vending machines, and by 1991, it constituted 34.0 percent of Ozeki output (Ozeki Reference Ozeki1996, 334–35).

Consistent with this mass-market strategy, in the 1970s the Nada makers began to sell sake in paper cartons like those used for milk. The cartons save money, fit in refrigerators—and look aggressively antigourmet. Ozeki introduced its first paper-carton sake in 1976. By 1983, carton sake constituted 18.7 percent of its output (Ozeki Reference Ozeki1996, 368). Gekkeikan introduced paper cartons in 1980 (Gekkeikan Reference Hart and John1987, 178). By 2000, paper-carton sake was about 40 percent of all sake volume; by 2016, it constituted more than half (Osake Sutatisutikkusu n.d.).

The Premium Niche

The Industry Transformation

Steadily, one-time local breweries like Dassai have begun to transform the sake industry. Overall, Japanese continue to drink less alcohol (Table 2). They also continue to drink less mass-market sake (Figure 2). From 2007 to 2018, annual sales of ordinary sake fell from 339,000 kl to 222,000 kl. Those of the intermediate category honjozo (described in the subsection on certification, below) fell from 67,000 kl to 35,000 kl.

FIGURE 2.

Annual Sake Production in kl.

Source: Kokuzeicho, Shuzo nendo ni okeru seishu no seizo jokyo to ni tsuite [Regarding the Production Circumstances, Etc., of the Brewing of Sake], various years.

Over the same period, however, premium junmai and ginjo sales rose from 98,000 kl to 125,000 kl. This is growth that the regional breweries have driven. By total 2018 production, nine of the top ten breweries were in either Nada or Fushimi (Table 1). Yet of the top producers of junmai ginjo sake, only four of the top ten were from Nada or Fushimi, and none of the top five were.Footnote ⁶

During the war-impoverished and heavily regulated midcentury, Nada-Fushimi breweries had cultivated a competitive advantage in low-cost production. Exploiting their economies of scale, they had focused on the mass-market sector. Hyogo producers still brew 85 percent of their sake as general (futsu or ippan) sake (2014 numbers). The Kyoto producers brew 90 percent as general. By contrast, in Dassai’s Yamaguchi prefecture producers brew barely 21 percent of their sake as general (Yamagata ken kogyo gijutsu sentaa 2015; see generally Shibuya Reference Shibuya and Shibuya2020, 86).

The regional brewers have deliberately crafted a new image. For decades, sake was a drink that middle-aged men (especially working-class men) drank in dirty bars on their way home from work. They drank it heated in small ceramic cups. By contrast, the regional brewers have nurtured an image of sake drunk by urbane thirty- and forty-something gourmets. They drink it cold, in long-stemmed wine glasses (Ono Reference Ono2019, 13).

The regional brewers present sake as a distinctively Japanese drink, but one to be savored by these successful young professionals. On its Web site, Niigata brewer Kikusui describes sake as a drink for all seasons. There is a sake to drink when relaxing with friends at a restaurant. There is a sake for men to drink while flirting with a waitress. And there is a sake for women to drink while recalling their first crush, continues Kikusui, while commiserating with each other over the sadnesses that only women can know, or while (alluding to a well-known passage in the Tale of Genji) trashing the actual men in their lives (Kunifuda Reference Kunifudan.d.).

Certification

In 1990, the government introduced a standard certification regime for the terms junmai, ginjo, and daiginjo. It had graded sake by a different metric for much of the midcentury, though nominally by quality. It had also taxed the brewers according to those quality grades, however, so brewers began to manipulate the grades to lower their tax bill. By the end the twentieth century the earlier grades had ceased to reflect quality, and the government repealed the system (Iga Reference Iga2008, 152; Ishiguro Reference Ishiguro2015c; Ono Reference Ono2019, 22).Footnote ⁷

Under the new 1990 certification regime, the tax office defined several of the new premium categories.Footnote ⁸ The rules have changed modestly over the years, but as of 2020:

Junmai: sake with only rice and rice-based mold (koji).

Ginjo: sake in which each kernel is milled to 60 percent or less of its initial size, and in which any brewers’ alcohol constitutes less than 10 percent of the weight of the rice.

Daiginjo: sake in which each kernel is milled to 50 percent or less of its initial size, and in which any brewers’ alcohol constitutes less than 10 percent of the weight of the rice.

Honjozo: sake in which each kernel has been milled to 70 percent or less of its initial size, and any brewers’ alcohol constitutes less than 10 percent of the weight of the rice.

Junmai offers classic nineteenth-century-style sake. Brewers use it to showcase their distinctive flavors. Ginjo and daiginjo offer smooth, clear tastes. Because differences attributable to soil, weather, fertilizers, and rice variety tend to concentrate in the outer layers, however, ginjo and daiginjo offer less variation across breweries. A ginjo with no added alcohol would be called a junmai ginjo. Honjozo represents a sake intermediate both in quality and in price between the other certified sakes and ordinary uncertified sake (futsu or ippan shu).

The Competitive Logic

The local breweries had sparked a renaissance in sake. By cultivating a taste for quality, they had created a market where none had existed before. The Nada-Fushimi brewers had responded to the wartime regulatory regime by using technology to lower cost and generate consistent products, but ignored the potential market for premium sake. The smaller local breweries now exploited the resulting opportunity and succeeded.

Unfortunately for the local breweries, the Nada-Fushimi breweries can compete in the market for ginjo and daiginjo if they choose. They will need better rice. They will need to shift production technique. They will need expertise. But they already have sophisticated equipment. They have capital. And if they lack the requisite experience and technique, they can hire it.

They can, and they have. All of the top mass-market producers (see Table 1) have premium products. Ozeki still stocks its One Cup Ozeki in vending machines, but it also offers an award-winning junmai daiginjo at 5,118 yen. Hakutsuru sells a 200 ml can of ordinary sake for 218 yen, but also offers an award-winning junmai daiginjo for over 10,000 yen. Gekkeikan sells its perennial working-class drink, but produces an award-winning junmai daiginjo besides. Takara offers enough varieties in cardboard cartons to fill a refrigerator, but sells a junmai daiginjo to boot.

What is more, a local brewery looking to move upscale will worry about more than Nada-Fushimi. It will face direct competition from Dassai and the Niigata-Nagano-Akita-Fukushima breweries that already dominate the “best sake” rankings. These breweries are local in name only. For the young Matsuo president in Shinano village, his competitors include the many premium breweries striving to replicate the Dassai success.

Crucially, the more successful two daiginjo rivals might be, the closer they will taste. Rice does indeed acquire flavors that reflect the mineral content of the water, the character of the soil, and the weather. But these differences are located almost entirely in a kernel’s outer layers. Grind off 40 to 50 percent of the kernel, and rice loses most of that distinctiveness (Sugihara Reference Sugihara2012; Ninomiya Reference Ninomiya2014, 469–70; Bumu 2015; Miyajima Reference Miyajiman.d.). All too often, as one brewer cruelly put it, ginjo will be to sake what the “valedictorian” was to the high school class—never wrong, but never interesting either (Miyajima Reference Miyajiman.d.).

Hence the appeal of “terroir.” If the Matsuo president in Shinano village and his peers around the country can convince consumers to distinguish and enjoy regional variation, they can create a niche that the other breweries cannot readily fill. Perhaps small Niigata breweries like Watanabe can fashion a taste that reflects the windswept Nechi valley, the brutally cold winters, and the massive snowstorms. Perhaps Matsuo can fashion a taste distinctive to the Arasebara hamlet (population 219) on the foothills of Mt. Madarao. To be sure, even connoisseurs of the best French and Napa wines roll their eyes at some terroir claims; many sake fans retain a similar suspicion too. As one reader of an earlier draft put it, mostly this shows that people have too much money.

But what the Matsuo and Watanabe presidents and their allies hope to create is an appreciation for a wide variety of very different, but each appealing sakes. In New York, a connoisseur might choose a white wine one night, a red another. Among the reds, he might choose a pinot noir one night, a cabernet sauvignon another. Among the pinot noir, he might choose a Cote de Nuits one night, a Givry another. If the Matsuo and Watanabe presidents and their allies have their way, the sake industry will offer connoisseurs choices just as rich.

Certification Problems

Give him half a chance, however, and President Takahashi will complain about his certification problems. He (and others championing terroir sake) (see, for example, Nihonshu 2019; Shibuya Reference Shibuya and Shibuya2020, 94; Miyajima Reference Miyajiman.d.) would like to adopt something close to the French appellation d’origine controlee (AOC). Under AOC, a winery can advertise its offerings as coming from a given district—a system in which, as Orley Ashenfelter (Reference Ashenfelter2007, 2) put it, the “best wines of Bordeaux are made from grapes (typically cabernet sauvignon and merlot) grown on specific plots of land and the wine is named after the property, or chateau, where the grapes are grown.” Under AOC, the size of the certified district can vary, depending on how selective the winery chooses to be. A winery can market its product as a Bordeaux wine, as a Haut-Medoc (within Bordeaux), or as a Pauliac (within Haut-Medoc). Some Nagano breweries sell distinctively Nagano sake. Matsuo offers a sake distinctive to Arasebara.

In fact, Takahashi would like to go further, and certify rice variety as well. The French chateaux blend wines, but the AOC rules specify the grapes they can blend (for example, Ashenfelter and Storchman Reference Ashenfelter and Karl2016, 39–40). Premium California wineries usually specify the grape. For his nonterroir sake at the Matsuo brewery, Takahashi sometimes uses yamada nishiki from areas outside Nagano. But for his local sakes, he uses several different varieties. For some, he uses miyama nishiki, the standard Nagano premium sake rice. For others, he celebrates the more recent local alternative, sankei nishiki.

In 2015, the Ministry of Agriculture, Forestry and Fisheries (MAFF) bureau for northeastern Japan launched the “Tohoku Sake Terroir Project” (it too uses the French word). The surviving local breweries in the region were trying to move into premium sake, much like the Niigata breweries had done four decades earlier. The bureau hoped to facilitate their shift by encouraging a distinctively northeastern taste. The MAFF bureau sponsored a study group. It met several times. It assembled a roster of eighty breweries that hoped to market sake with the new designation. Largely, however, it seems to have gone inactive (Matsuo Reference Matsuo2017, 42; Tohoku n.d.).

In any case, for breweries like Matsuo the project did not go far enough. To qualify under the project, a brewery had to use local water, use at least 50 percent local (prefectural, or at least northeastern) rice, and communicate directly with its supplying farmers (Tohoku n.d.). For advocates of a narrower terroir approach, this was far too broad. The project proposed certifying the sake’s prefecture. Yet Nagano has a population of 380,000, more than the entire Bordeaux area (population 250,000). In effect, the project would preclude all of the finer AOC partitions.

When Japan joined the World Trade Organization, it adopted a completely separate, national program for geographical certification (Kokuzeicho 2020a). As of mid-2020, it had haphazardly approved four sake designations under the program. In 2005, it had certified the sake from the town of Hakusan on the Japan sea coast: Local breweries could use the designation if they used local water and only Japanese rice (i.e., rice from anywhere in Japan) of grade 1. In 2016, they certified sake from the prefecture of Yamagata: Prefectural breweries could use the designation if they used local water and—again—Japanese rice. In 2020, they certified a much more finely partitioned designation for the Harima area of Hyogo: Harima breweries could use the designation if they used local water and yamada nishiki rice from Hyogo prefecture.

Alas for terroir brewers like Matsuo, the government saved its ultimate insult for 2018. That year, under its World Trade Organization geographical certification, it designated “Nada Gogo.” To advertise the certification, the Nada firms had to use local water. They had to use Japanese rice (again, from anywhere in Japan). And that rice had to be of at least grade 3 (rarely does anyone designate sake rice below grade 3).

Takahashi and his fellow terroir brewers have not constructed (either publicly through the government, or privately among each other) a certification program—yet. But this does not mean too much. They want a program. And they have only begun to try.

The JA Cooperative Network

Most firms buy their sake rice through their local agricultural cooperative. More precisely, in recent decades sake brewers have been buying about 70 percent of their sake rice through their prefectural sake brewers’ association and the JA, the national network of Japanese agricultural cooperatives (Saito Reference Saito2015, 3; Suzuki and Takada Reference Suzuki and Osamu2017; Hirogaru sakamai seisan (ka) 2018; Keiyaku saibai de hinshitsu kojo 2018; Nihon okome kyokai 2018).

Under this arrangement, each year the brewers’ association solicits rice orders from its members for the season two years hence. It places the orders with the agricultural cooperative. The cooperatives then distribute the orders among their membership (Koike Reference Koike1995, 163; Maikoku kiko 2014, 13; Hayashi Reference Hayashi2017, 55).

The brewers will use this sake rice mostly for their initial steps. Recall that they add rice in three broad steps: They mix a small amount of rice with the mold (the koji mai); they add more rice with the yeast (the shubo mai); and after the fermentation has progressed they add the rest of the rice (the kake mai). Many (if not most) brewers use table rice for the third, high-volume step. They use this costlier sake rice primarily for the first two steps.

Muramai

Direct Contracts

The Nature of the Contracts

Brewers pay a generous price

Regional breweries pay their farmers more than what they would pay the coop—sometimes substantially more. In his recent study of Akita brewers, Futoshi Hayashi (Reference Hayashi2017, 62) examines the way they set the price. Often, he finds, they simply add a premium to the coop’s price. They do not use formulas that turn on the results of quality inspections or on the methods that the farmers use to cultivate.

Instead, Akita brewers take what the farmer could obtain from the coop and pay him substantially more. “We want a price where the farmers will want to raise sake rice even if table rice is selling for 20,000 yen [per 60 kg],” said one brewer. To do this, he needed to pay two or three times the usual price for table rice (Hayashi Reference Hayashi2017, 62).

Prior to the war, the Hyogo brewers paid high prices under their muramai contracts as well. In 1933, Hyogo sake rice sold for an average 23.30 yen per koku (180 l); the Kamikume muramai rice sold for 27.30. In 1935, the Hyogo sake rice sold for 32.63, while Kamikume muramai went for 35.60. In 1937, the Hyogo average was 36.64 while Kamikume sold for 40.00. And in 1938, the Hyogo sake rice sold for 37.73, while Kamikume muramai sold for 43.10—a 14.2 percent premium (Ninomiya Reference Ninomiya2014, 474; see also Mori Reference Mori1983, 124).Footnote ¹¹

Brewers use renewable (and therefore terminable) contracts

A simple high price presents a risk of moral hazard, of course. A farmer can renege on the deal in myriad observable but not legally verifiable ways (a distinction at the core of Hart and Moore (Reference Hart and John1990)). The breweries mitigate this risk straightforwardly: They offer only one-year contracts. They renew the contracts if a farmer cooperates and look elsewhere if he or she does not. Should a farmer adopt opportunistic strategies, he or she will find himself or herself next year without a high-priced contract.

The history of the muramai contracts in Hyogo indicates that brewers do terminate contracts. If farmers failed to live up to brewers’ hopes, prewar brewers did indeed refuse to renew the contracts, and turned to other growers (see Hyogo 1961, 98–105; Nishinomiya shuzo Reference Nishinomiya shuzo1989, 197; Ozeki Reference Ozeki1996, 432; Ninomiya Reference Ninomiya2014, 471–72; Toku A chiiki shitei no yurai Reference Tokun.d.). Given the quick termination, however, one would not necessarily expect the parties to litigate disputes. At least in the Westlaw database, no record of any litigation over premium sake-rice quality remains.

Brewers select for the person

Over and over, the brewers insist that they “need to be able to see the farmer’s face” (for example, Hirogaru sakamai seisan (ka) 2018; Keiyaku saibai de hinshitsu kojo 2018). To obtain consistently high quality, they need to contract with men and women of character and ability. Were they obtaining their rice from the cooperative, brewers could specify the graded quality they wanted. When obtaining rice from a farmer under contract, matters are more complex. They will take the rice the farmer delivers—and the quality of their sake for the year will turn in part on that rice.

Whether through opportunism or simple incompetence, a farmer can easily sabotage a brewer’s product, and a lawsuit will not help. To be sure, brewers can specify some verifiable quality metrics. If a farmer failed to meet a metric, the brewer could sue for breach. But if any brewer ever did—again—no record of it remains on the Westlaw database. For the most part, Japanese farmers are low-income retirees with a tiny paddy of little economic value and a dilapidated two- or three-cylinder truck. Seventy-two percent of the farmers own less than a hectare of paddy (2.47 acres; Hatsuki Reference Hatsuki2015, 128), and the trucks are fifty-horsepower machines that sell for as little as $7,000 new. If a farmer has few assets to attach, the brewer gains little from any claim for breach.

The brewers and farmers draft a contract to build a relationship. As Thomas Dietz (Reference Dietz2012, 39) puts it in his study of international transactions (where state enforcement is obviously problematic), a contract serves as “a communication document.” It facilitates “mutual communication, but not the enforcement, of legal claims.” Lisa Bernstein (Reference Bernstein2015, 562) makes the same point about Midwestern heavy machinery manufacturers: Their contracts simply “are not designed to create incentives for performance.” The brewers negotiate the contract to build a collaborative relationship. Toward that end, they pick their suppliers carefully. They pick by talent. They pick by character. And in the brutal candor of small towns the world over, they pick by family: Know a farmer’s last name, explained the muramai chair for Yokawa village, and you know the quality of his rice (Nihonshu no korekara 2017).

Brewers select for the land

And almost as often, the brewers insist that “to buy the rice, you need to see the land” (for example, Mori Reference Mori1983, 124; Ozeki Reference Ozeki1996, 432). Brewers do not offer contracts on just any paddy; they select for the quality of the land. To raise sake rice properly, “you need to see the rice paddy,” explained the Yokawa muramai chair (Nihonshu no korekara 2017). “You need to talk to the yamada nishiki.” By soil, by location, by drainage, some farms produce better rice than others. The same is true in Burgundy and Bordeaux, of course. Even within Yokawa village—the very origin of the muramai contract—some paddies yield better rice than others (Nihonshu no korekara 2017). The Miyajima brewery (Miyajima Reference Miyajiman.d.) in Nagano stresses the importance of water: The fields upstream must not use heavy pesticide; the homes upstream cannot discharge their sewage. It stresses the importance of sunlight, and it stresses the importance of the wind patterns.

Brewers monitor observable inputs

Verifiable inputs can correlate with rice quality, and many brewers mandate the levels of the inputs that do. Brewers often specify cultivation methods. Masaya Iga examines closely the contracts that one regional premium brewer maintained. Should a farmer insist on using pesticides and chemical fertilizers at ordinary levels, the brewer would offer no contract at all.Footnote ¹² Should he or she agree to use them at only low levels, the brewer might offer one price. Should he or she agree to use none, the brewer would pay a 50 percent premium (Table 3, Panel A; Iga Reference Iga2008, 157).

Table 3.

Two Regional Brewers and Supplying Farmers

Brewer A is Yamana Shuzo, a brewer in Tamba city, Hyogo, specializing in premium sake. In 2006, Yamana consumed 42.4 tons of rice per year. Of these, it purchased 18.8 by direct contract. It promises to buy all rice produced on designated fields.

Brewer B is Taketsuru, a brewer in Takehara city, Hiroshima. Taketsuru only advertises versions of junmai sake. In 2006, it consumed 59.9 tons of rice, of which it acquired 12.8 tons by direct contract. Like Yamana, it promises to buy all rice produced on designated fields.

Notes:

Exp: Years of experience in sake rice production

Age: Age of male household head, deciles

Kg: Kg of sake rice supplied

ha: hectares of land devoted to sake rice

Price (A): yen per 60 kg

Price (B): Market price plus adjustments plus number in table, per .1 hectare

Pest.: Pesticides use—none, or low

Fert.: Chemical fertilizer use—none, or low

Contact: Number of regular contacts with brewery per year.

Source: Masaya Iga. “Seishu kyokyu taikei ni okeru shuzogyosha to sakamai seisansha no teikei kankei [The Relationship between the Sake Breyers and the Sake-Optimized Rice Producers under the Sake Supply Structure].” Chiri gaku hyoron 81 (2008): 150–78.

Other brewers focus on yield—as in, they want low yields. Wine quality tends to correlate inversely with vineyard productivity, and so it is with sake.Footnote ¹³ As a result, both wineries and sake breweries contract to buy the entire yield of a field. They then discourage farmers from ramping the productivity by adopting pricing formulas that track yields only up to a given volume, and pay nothing for any excess (Iga Reference Iga2008, 157; Washington 2015, 10 (grapes); Hayashi Reference Hayashi2017, 62 (rice)).

Sometimes (but only sometimes), brewers contract over observable outputs

Brewers can observe rice quality, and can verify that quality—in part. They can measure the percentage of broken kernels. They can measure the level of protein in a kernel. And they can adjust the price they pay by these verifiable measures (Hayashi Reference Hayashi2017, 56).

And some do—but not all. The problem with setting price by verifiable quality is precisely that it works, and may work too well: It may skew an agent’s effort toward the verifiable measures, but away from other possibly more important aspects of quality. Quality is a subtle thing, and only some measures are verifiable. As Holmstrom and Milgrom (Reference Holmstrom and Paul1991) point out, a principal who pays his agent only by verifiable indices of quality may cause the agent to bias his efforts away from aspects of quality not incorporated into the pricing formula. If the cost of that bias is sufficiently high, the principal may rationally choose to offer instead a contract with only muted incentives. Wineries face the same problem. Grape quality turns in part on observable characteristics, but “more subtle characteristics are harder to measure” (Goodhue Reference Goodhue2002).

And so it is that many premium sake brewers omit quality measures from their pricing formula. Instead, they meet regularly with the farmers and discuss how better to raise that quality. If a farmer produces defective rice, they bring samples with them and talk to him or her about what to do (Hayashi Reference Hayashi2017, 60). But when so many of the distinguishing measures of rice quality are observable but not verifiable, many of them rationally decide not to key their contracts to the verifiable measures of quality. Instead, they monitor and talk: They follow the progress of the rice, they talk with the farmer, they watch for potential problems, and they try to forestall a disaster before it arrives.

Leasing, Ownership, and Vertical Integration