A Statistical Framework

Central to our assessment is the notion of revealed preference: If a legal template for organizing business is widely adopted, we presume that it offers something valued by those taking it up (at least) and (possibly) by society at large. However, creditors might reasonably view limited liability less favorably than incorporators, and Guinnane et al.’s tentative agnosticism about whether higher take-up of a particular form is efficient or socially beneficial is well taken. It is also conceivable that a system of total liberty would be so abused as to undermine faith in the capitalist system, leading to fewer corporations, but in fact this did not happen, even in the most liberal regime. ^{Footnote 12} Flexibility may nonetheless bias the type of company being formed. For example, allowing directors to entrench their position to the detriment of outside shareholders might encourage incorporation by family oligarchs, while discouraging public offerings (although legislation reserved for quoted companies or stock exchange listing rules might separately promote the latter). Guinnane et al., interestingly, suggest that flexible internal governance rules drove high take-up of the private company, allowing businesses to limit the threat of untimely dissolution inherent in partnerships without taking on the full risk of minority oppression facilitated by classic corporations. They also point to other possible encouragements offered by legal forms, including cheaper registration costs, the ability to entrench senior management or minority veto rights, and lighter regulatory burdens (which in some countries were more attractive than in Germany). Other possible attractions—also varying internationally—included an administrative process capable of registering a company in hours, not weeks; minimal publicity; paralegal intermediaries selling “off-the-shelf” companies; regionally devolved registration offices; the absence of minimum required capital; tax privileges; not requiring an expensive notary to endorse changes in ownership or charter conditions; minimal annual reregistration requirements; and (in the extreme) not having secret police and a cowed judiciary endorsing their expropriations. Our measure of the take-up of various business forms captures only the net effect of the whole package of liberal flexibility (combined with any other determinants); other means (econometric or archival) must be found to distinguish their relative importance.

Comprehensive annual statistics on the numbers and capitals of extant stocks of all multi-owner business forms are not available for most countries for the twentieth century as a whole. Several series, however, are available annually for Japan from 1883 to the present for all forms in the commercial code and can be linked to form one continuous series (see the online appendix). For the four other countries in Table 1, we have similarly comprehensive statistics only for fully corporate forms (defined here as Japan’s kabushiki kaisha (KKs) ^{Footnote 13} plus yugen kaisha, in the UK public plus private companies, in the United States quoted plus close corporations, in Germany AGs plus GmbHs, and in Switzerland SAs plus SARLs). ^{Footnote 14} Partnership forms are necessarily excluded, because outside Japan, their numbers are rarely reported annually with precision. To assess the level of corporatization in these nations of vastly different—and differentially changing—sizes, we have divided the numbers of corporations by their national populations at all dates.

Table 1

Extant stocks of (private and public) corporations per million people, 1899–1999

Source: Online appendix to this article.

The most obvious feature of Table 1 is that the common-law countries (the United States and United Kingdom) were ahead of both Germany and Japan at the beginning of the twentieth century and remained so at the end. If the corporate form (as defined in this table) had advantages, these two common-law nations apparently benefited earlier and more abundantly. However, the example of civil-law Switzerland suggests that legal family was no barrier to rapid catch-up. Swiss corporate law, owing something to German stems, was on a federal rather than cantonal basis and (although Swiss lawyers debated developments elsewhere) developed independently. The number of corporations per capita in Switzerland (already ahead of Germany from the earliest available statistical counts) overtook the United Kingdom during World War I and the United States in the 1930s; it remained the leader of this group for the rest of the century. Moreover, Switzerland was not unique among northwest European countries with civil law regimes: Norway, Sweden, Denmark, Finland, the Netherlands, and Luxembourg were all (at various stages of the twentieth century) nearer to the high Swiss than to the modest German levels of corporatization. Although common-law nations led corporatization at the beginning of the century, civil-law nations experienced more rapid growth in corporate numbers, so there was some convergence. It was quite rare for (non-communist) societies to experience a reduction in the numbers of corporations per capita over a sustained period, but some countries managed it. Germany appears to hold the record for the longest sustained early decline (which can be more precisely dated from 1923 to around 1948), but even the United States experienced a temporary check during the 1930s depression. Factors other than the (unchanging) legal family evidently determined these variations. Foreman-Peck and Hannah’s (“Diffusion”) econometric analysis of an international cross section before 1914 suggests that more fundamental determinants of corporatization were living standards (proxied by GDP per capita) and the degree of liberality and openness in economic policy (proxied by openness to international trade). The latter is closely correlated with more corporation-friendly legal systems, in both common and civil law nations.

How does Japan fit into this picture? As Table 1 shows, in the early twentieth century it had relatively few corporations. Japan was still an early-stage industrializer but it had also come late to allowing incorporation without individual state sanction (as had become the norm in western Europe, the United States, and some of the developing world by the 1870s). Nonetheless, in the postwar miracle years, Japan came to have corporate numbers comparable with other leading nations, and at the end of the century it had numbers of corporations per million people almost as high as the United States and well ahead of Germany. Japan had already overtaken Germany’s corporations per capita in the 1930s (when it still had much lower living standards), confirming that adopting a civil law system—even one based on the German code but without the GmbH—did not especially inhibit incorporation.

However, Germany was by then a very soft target. Its declining corporate numbers, beginning a decade before Hitler’s 1933 election victory, were further reduced by Nazi policies under his dictatorship. These condemned outside shareholders as leeches on managerial creativity, reduced shareholder protections, restricted dividends and stock issues, expropriated Jewish companies, and provided strong incentives for Aryans to convert their corporations to partnership form. ^{Footnote 15} Other right-wing authoritarian regimes, such as those of Japan and Italy, shared only some anti-corporate prejudices and did not emulate the more extreme of these policies. Like many combatants, the Japanese government (on a full war footing in 1937–1945) naturally restricted dividends and concentrated existing corporations on armaments production, but its overall numbers of corporations consistently increased, with only slight and temporary setbacks. ^{Footnote 16}

Alternative Business Forms

The statistics for corporations in Table 1 arguably understate Japan’s early position. Statisticians and lawyers (even within the same legal family) draw the dividing lines among enterprise forms in different places, not always using identical logic or language in doing so and sometimes blurring lines of division other nations consider clear. As Elinor Ostrom insists in different contexts, many cooperative, multi-agent contractual relations may function in a coherent manner, without conforming to a market/hierarchy/state trichotomy. ^{Footnote 17} Table 2 illustrates some of the issues in multi-owner organizations, although forms masquerading as linguistic equivalents in the organizational menus of this table conceal considerable variations (anyone who has ordered spaghetti bolognese in Bologna, Los Angeles, and Tokyo will understand the problem of menus). ^{Footnote 18} Corporations—the first two columns—encompass two types (although their legal separation in many countries came only in the second half of the twentieth century): public companies, which could be (but were not necessarily) quoted; and private companies, not offering stock to the public. Some countries exempted private companies from requirements to publish accounts and restricted their shareholder numbers (for example, Japan and the United Kingdom, but not Germany, imposed maximum numbers of fifty). By this period, stockholders normally (but not invariably) had limited liability, whereas partners in ordinary partnerships (column 4) did not. Limited partnerships (column 3) occasionally had fully limited liability, but usually (as in Japan) followed the old European model of the commandite: Managing partners retained full liability and only outside investors had limited liability. This table includes only capitalist enterprises; state and municipal businesses (unless they were organized under the general law for capitalist enterprises) are excluded, as are some older legal forms once common in Europe but approaching extinction ^{Footnote 19} and minor variants that contemporaries treated together with the cases shown. ^{Footnote 20} The order of the six forms shown in the table can be thought of as a ranking: It is broadly true that progressing from public companies to sole proprietors entails a movement toward smaller enterprises, usually with fewer owners and greater personal liability, yet giant enterprises have at times existed in all categories and it is possible for a corporation (de jure or de facto) ^{Footnote 21} to be owned by one person, uniting the extremes.

Table 2

Capitalist enterprise forms in the first half of the twentieth century

* In addition to kaisha, there were many kumiai (cooperatives or associations) in Japan besides those set up under special legislation from 1900 (the “special” kumiai in column 5). Many similar entities existed in Western countries—sometimes set up as cooperatives, sometimes as corporations, but also with a wide variety of names, liability regimes, profit distribution rules. and legal status (savings banks, credit unions, mutuals, Vereine, Raffeisen, etc.).

† From 1940. The godo kaisha, introduced in 2006 to succeed the yugen kaisha, is irrelevant for this historical discussion.

Cooperatives and mutuals are generally excluded because, though they made and distributed profits, control rights were generally not proportional to the capital subscribed or dividends paid (by then the capitalist norm); instead, each member normally had one vote. However, in Japan, some kinds of kumiai—variously translated as cooperatives, associations, or partnerships—were close counterparts to certain collective business firms in the West. ^{Footnote 22} Kumiai was the term used in the civil code to describe partnerships that had neither separate legal personality nor limited liability. ^{Footnote 23} Many were very similar to Western (unlimited) business partnerships, and they had similarly wide freedom to contract and modify contracts when circumstances changed, but responsibility for debts was not (as in the West) joint and several, but proportional to participations in the enterprise (or, if creditors had not been apprised of that, equal liability per partner). ^{Footnote 24} Other kumiai (in the fifth column of Table 2) set up under special legislation in Japan from 1900 (partly modeled, like its corporate law, on German precedent) did have separate legal status; some had limited and some unlimited liability. ^{Footnote 25} Some undertook functions (trade associations, cartels, joint purchasing or credit for farmers or homeowners) that in the West might have been constituted under special legislation, but were sometimes organized as corporations or cooperatives. ^{Footnote 26}

In Japan’s distinctive terminology, “kaisha” describes all the first four columns in Table 2 (under the commercial code) but none of the remainder (under special legislation or the civil code). This was not the case for similar words for generic types in Western languages. The corporate form was usually confined to the first two columns (public and private companies) in the Anglosphere, whereas Gesellschaft, société, aktiebolag, and the like in continental European languages (and “company” in Victorian British English and modern American English) cover the first four columns but additionally include further types. “Firm” (or even “company”) in English can sometimes include sole proprietors, as do non-legal terms like kigyō, “enterprise” or “business” in most languages. Kaisha is usually translated as “company” (in its more restricted modern sense) or “corporation,” but is that correct? Not to twentieth-century Anglo-American lawyers. Common-law partnerships did not then have separate legal personality in the United States and England, ^{Footnote 27} but in Japan some types—partnerships (ordinary or limited) under the commercial code (both kaisha) and some (but not all) kumiai—did and even in civil law Scotland, France and Germany partnerships had some characteristics of separate legal entity. ^{Footnote 28}

Guinnane et al. (Reference Guinnane, Harris, Lamoreaux and Rosenthal2007) describe the GmbH as a “hybrid”; there was initial contemporary uncertainty about how to describe it. Some, noting the GmbH’s corporate form and fully limited liability, translated it as private limited company (close corporation); others—noting that its ownership stakes were named Anleihen (quotas) rather than Aktien (stocks) and that, as with partnership stakes, they were not tradable on stock exchanges—translated it as limited partnership. The first usage predominated, but the latter has recently been revived in American English, where, from the 1970s, newly legally defined “limited liability companies” (private companies), though sharing most legal characteristics of corporations, are technically considered partnerships. The historically pervasive Anglo-American use of that same term as a synonym for corporation is accordingly falling into disuse in the United States. ^{Footnote 29} Confusingly, “limited liability companies” in British English (in this context, more widely spoken worldwide) ^{Footnote 30} are definitely not partnerships, but corporations proper.

Economists might wish to define the boundaries between enterprises differently from lawyers; the legal categories in Table 2—even if not coinciding exactly with their needs—will sometimes provide serviceable approximations. For example, to measure enterprises that were corporate (in the Anglosphere’s sense) and in which everyone, including the top managers, had limited liability, one combines the types in the first two columns, making small adjustments. ^{Footnote 31} To include registered enterprises in which outside investors—but not designated principals or managing partners—had limited liability, one needs to add limited partnerships and some cases from column 5. To include all capitalist multi-owner entities, one further adds unlimited partnerships and more cases from column 5, again correcting for odd exceptions. ^{Footnote 32} Practicalities are more difficult than principles. Ordinary partnership data are full for Japan (at least for commercial code kaisha, though not for civil code kumiai), but are often only partial or nonexistent for other countries. Such gaps in the statistical record leave some inevitable lacunae in comparative measures.

Japan has fuller data for its kaisha (the first four columns) than the equivalents in most countries, facilitating examination of some variations resulting from alternative definitions. In Table 1, following a common convention, we counted as corporations only the first two columns. For Japan (before 1940) this includes only the first column, the KK, though one might view this with equanimity. Japan as yet had no formal private company form, but many KKs—as with the basic corporation in other countries—were de facto private, ^{Footnote 33} in the sense that they had few shareholders and their shares were not traded publicly, so companies in some respects resembling the German GmbH are actually included. Moreover, the differentiation is logically uniform: Both limited and unlimited partnerships are also excluded from all countries’ totals. Thus, as a measure of fully limited kaisha entities, the totals in Table 1 serve reasonably well for international comparisons. ^{Footnote 34}

However, there are indications that Japanese limited or ordinary partnerships (in addition to being distinctively described as kaisha or “corporate” in Japan’s idiosyncratic legal terminology) and some kumiai were prone to fulfill functions sometimes undertaken by corporations proper in other countries, despite many of them having (partially or fully) unlimited liability. Table 3 puts some numbers on the issue for 1910. It is clear that counting KKs alone only just captures most entities with fully limited liability in 1910, and captures less than half of those with any limitation of liability or with separate legal personhood. It is the cooperatives (mainly special agricultural kumiai) that make up most of the numerical deficit in 1910 and that would increase if we had statistics on other kumiai. Moreover, even omitting kumiai (as other countries’ statistics omit cooperatives, which they appear to have used less intensively than Japan), ^{Footnote 35} kaisha partnerships (combining goshi and gomei—i.e., those with and without some limited partners) also outnumber KKs. However, corporations (KKs) are on average much larger, so they account for 85 percent of paid-up kaisha capital in 1910. Neither would their dominance be much diminished if we added the paid-up capital of special kumiai, whose numbers grew more slowly and were smaller than kaisha and many of which lost their limited liability status in the 1930s. ^{Footnote 36} Goshi kaisha proliferated at twice the rate of KKs in the quarter century after 1910 (their numbers peaked in 1936), with gomei kaisha numbers also growing faster than KKs, but both kinds of partnership kaisha remained mainly smaller firms, ^{Footnote 37} while the average paid-up capital of KKs more than doubled. Thus the KK share in all kaisha paid-up capital remained at 85 percent in 1935, as in 1910. ^{Footnote 38}

Table 3

Main Japanese multi-owner enterprise forms in 1910

Source: Department of Finance, Twelfth Financial and Economic Annual of Japan 1912, pp. 83, 86.

* Sangyo kumiai under the 1900 Industrial Association Law (see Sangyo kumiai yoran), which had a choice of liability status: limited (yugen kumiai) and unlimited (mugen kumiai) are self-explanatory; the “partly limited” in the table (hosho sekinin) had chosen something in between e.g. double liability). In addition to the 7,259 shown in the table there were 49 whose liability was not classified. Many of these kumiai were closer to Western rural cooperatives than to companies, but shigaichi shinyo kumiai (city credit associations, which took deposits from nonmembers and were able to discount bills) and shigaichi kobai kumiai (city consumer cooperatives) were also formed under this law. Other kumiai seem to have been less numerous—for example, there were 3,421 fishermen’s associations (suisan kumiai) in 1910, but their numbers then rapidly declined; and 916 juyo bussan dogyo kumiai (trade associations) in 1912 (the earliest year), numbers that had slightly declined by 1930.

† Paid-up capital data refer to 3,527 kumiai: less than half the total numbers.

This numerical expansion and stable capital share of partnerships differs from continental European experience in this period, where the commandite was in relative decline in both numbers and paid-up capital, particularly, as in the cases Guinnane et al. emphasize, in countries that adopted the GmbH/SARL (private company) form. ^{Footnote 39} Shimizu argues that limited partnerships (goshi kaisha) and private companies (yugen kaisha) were more similar in Japan than their European equivalents. ^{Footnote 40} Goshi kaisha had separate legal personalities, like corporations, and might more readily survive the departure of a partner. However, even in Japan, dissatisfied partners in a gomei kaisha could still compel premature dissolution through withdrawal (at the end of a business year and with six months’ notice), although limited partners (or their heirs) in a goshi kaisha could withdraw only if all general partners agreed, ^{Footnote 41} whereas withdrawal (as opposed to share sale) was forbidden in yugen kaisha, providing fuller entity shielding. Goshi kaisha also differed from KKs or yugen kaisha in that at least one partner had to have unlimited liability (and, in respect to debts incurred while the person was a partner, that liability subsisted for two years after withdrawal), but in practice, until a change in the law in 1938 that made it easier to enforce judgments on unlimited partners, the procedure for outside creditors to attach assets for partnership debts was onerous. Goshi kaisha had other advantages for small family enterprises. They were cheaper to register than KKs, ^{Footnote 42} did not have to appoint auditors or publish accounts (secrecy was prized by family companies, as well as by group holding companies ^{Footnote 43} ), required only two rather than seven members, and had no minimum capital requirement. ^{Footnote 44} Partnership agreements could also entrench family control, protect minorities (through withdrawal rights of unlimited partners for reasonable cause) and allocate rights and duties more flexibly than shareholders (who had to have equal votes per share in KKs, with only a 50 percent majority required for major changes).

Goshi kaisha were the smallest of registered kaisha in Japan, with less than a tenth the average capital of KKs in 1910 (less than a twentieth by 1935) and only around one-third the average capital of ordinary partnerships at both dates. They thus occupied a position in the size hierarchy different from the ostensibly similar commandite limited partnerships in the West, which recruited limited partners to amass greater resources than unlimited partnerships. We do not have comparable capital measures for all German enterprises (only for AGs and GmbHs) but, in employment terms, German KGs (limited partnerships) were about eight times the size of OHGs (unlimited partnerships) and somewhat larger than GmbHs (private companies). ^{Footnote 45} In the United States, also, limited partnerships were of intermediate size. ^{Footnote 46} Japanese KKs, like AGs in Germany or corporations in the United States, occupied the top of the firm size hierarchy, but unlimited partnerships (gomei kaisha), in opposition to the norm in Western countries, ^{Footnote 47} were larger than limited ones (goshi kaisha). It was the latter—with outside investors’ liability limited and freedom from accounts publication—that fulfilled a role similar to GmbHs in Germany. That explains both why goshi kaisha substantially declined (in numbers and share of kaisha capital) ^{Footnote 48} with the introduction of the yugen kaisha (formally the GmbH equivalent) ^{Footnote 49} in 1940 and also why Shimizu reports that the advantages of the GmbH identified by Guinnane et al. were not a primary reason for the subsequent rising popularity of the yugen kaisha in Japan. ^{Footnote 50} In the 1920s the Tokyo Chamber of Commerce and Industry lobbied for a GmbH for small companies, but were successful only when a tightening up of newspaper publicity requirements and other rules for KKs seemed unreasonably onerous for small firms. Yugen kaisha were introduced in 1940 and were initially nearer in size to KKs than goshi kaisha, ^{Footnote 51} but smaller firms later predominated and, just as Guinnane et al. show for Germany and France, the use of limited partnerships declined. ^{Footnote 52} One advantage of the yugen kaisha was extending limited liability to the top owner-managers, at least one of whom had to accept liability in a goshi kaisha. ^{Footnote 53} However, other advantages of the GmbH—notably, the freedom from compulsory accounts publication and flexibility in governance arrangements—were already widely adopted by Japanese SME entrepreneurs in the goshi kaisha. It is nonetheless possible that partnerships in Japan were less effective than KKs. Nicholas shows that their returns on equity were lower than KKs in 1922–1938, ^{Footnote 54} an observation compatible with a dual economy interpretation of enterprise forms. SMEs, often using the partnership forms, perhaps occupied the less profitable niches in a developing economy with many backward sectors, though all kaisha forms were present in most industries. ^{Footnote 55}

Thus, the limited partnership (goshi kaisha) was more popular in its heyday in 1930s Japan than its German equivalent (the KG), despite Nazi policy favoring the conversion of AGs and GmbHs to KGs. ^{Footnote 56} To find proportions of extant registrations in the limited partnership form as high as in 1930s Japan in Germany (or other civil-law countries), one has to go back to the early nineteenth century—that is, to before general registration laws made individual authorization of corporations unnecessary ^{Footnote 57} —or to countries like Russia, where the tsar’s authorization was still required at the time of the 1917 revolution. ^{Footnote 58} There was, however, nothing uniquely Japanese in flexible responses to the absence of the specific GmbH corporate form that Guinnane et al. praise. Although commandite partnerships were not used in twentieth-century Europe as widely as the Japanese goshi kaisha, ^{Footnote 59} KK equivalents (the AG in Switzerland, the NV in the Netherlands, or the AS or AB in Scandinavia) were more commonly used to achieve what was offered by GmbHs in Germany and goshi kaisha in Japan. This was possible because their laws for this corporate option were less stringent on matters like publicity of accounts and governance rules than Germany’s for AGs. The Netherlands most closely resembled Japan, though, significantly, only after its NV law was tightened up in 1928, without permitting the besloten vennootschap (BV, equivalent to a GmbH) until 1971. In the intervening decades there was, though less markedly than in Japan, extensive recourse to the commandite form. This was unnecessary in Switzerland and many other small European countries, because their basic corporate form continued to have less stringent requirements, so that form could still be widely used by SMEs and family firms. Similar recourse to the KK form was known in Japan (in fact, most KKs were small family firms) ^{Footnote 60} but was less favored there. This was because Japan’s fuller disclosure requirements, the wider scope of decisions that required ratification at a general meeting, and shareholder fears of calls on unpaid capital by directors at an inconvenient time (which had largely died out in Europe in the nineteenth century) discouraged some SMEs from registering as KKs. ^{Footnote 61} However, the numbers of corporations (KKs) in Japan overtook the number of AGs in Germany in 1912 and then expanded even more rapidly. Even after the yugen kaisha was introduced in 1940, KKs remained massively more numerous than German AGs. The KK was perceived to have more prestige than the yugen kaisha, and the provisions for protecting shareholder interests in KKs long remained less stringent than those in German AGs. The KK also allowed some choices, such as restrictions on share transfers by family owners, which were not possible in a German AG, but were allowed in GmbHs. ^{Footnote 62}

Stock Exchanges

As the KK form was widely used by family SMEs, we cannot assume that most KK capital was publicly traded on stock exchanges, as was the case for AGs in Germany. ^{Footnote 63} In Japan—as in the United States, United Kingdom, and Switzerland ^{Footnote 64} —de facto close corporations (with shares rarely [if at all] traded and held by relatively few people) dominated the corporate population numerically throughout the twentieth century. ^{Footnote 65} Japan did not regulate KKs as stringently as Germany’s 1884 code regulated AGs, nor did it emulate the tough additional German stock exchange regulation of 1896–1897, which affected quoted AGs. ^{Footnote 66} Such laws reinforced the gatekeeper position of German banks and required AG accounts, prospectuses, and other investor aids facilitating the divorce of ownership from control. ^{Footnote 67} With fewer such legal protections for investors, one might expect Japanese stock exchange development to lag behind Germany’s, at least until postwar Americanization promoted investor protection. This increased Japan’s score on the “anti-director” rights index (which some modern analysts consider the major driver of stock exchange development) from 1 to 5 (out of a possible score of 6).

As Table 4 reports, however, Japanese equity markets were not behind Germany’s before World War I, by Rajan and Zingales’s ^{Footnote 68} widely used metric of the ratio of quoted equity capitalization to GDP, and they also forged ahead of Germany’s in the interwar years in quoted company numbers. One of the reasons may have been competition both among exchanges and from off-exchange trading, whereas Berlin (and Frankfurt) gradually dominated Germany’s provincial exchanges and off-exchange securities trading was illegal. Unlike major exchanges abroad, Japanese exchanges encouraged trading by not charging listing fees. Japan’s forty-six exchanges of 1898 fell to thirteen by 1911 ^{Footnote 69} and to eleven by 1943, when they were compulsorily amalgamated into the Japan Exchange (with a main exchange in Tokyo and ten branch exchanges, of which two soon closed). Outside these exchanges, over-the-counter (OTC) unregulated trading of both listed and unlisted stock long remained strong (but is excluded from the table), so the precise size of the market is difficult to define, but when nine exchanges reopened in 1949, they abolished forward trading in return for a ban on this OTC trading of listed stocks. ^{Footnote 70} (Low) listing fees were introduced; by 1950, much of the postwar reorganization of stock corporations was complete. Paradoxically, Japan’s quoted capital/GDP ratio reached a nadir at this time, when postwar Americanization increased statutory investor protection, although within a few decades Japan restored its lead over Germany and now has one of the most widely dispersed stock holdings in the world. ^{Footnote 71} As elsewhere, Japanese corporate size distributions were highly skewed; it seems likely that most Japanese KK capital at an early stage was in the small minority of companies that were quoted on exchanges, ^{Footnote 72} as was the norm in Europe, rather than (as in the United States before 1914 or the global norm today) most corporate capital being in close (unquoted) corporations. ^{Footnote 73}

Table 4

Stock markets: Germany and Japan 1913–1999

Source: Rajan and Zingales, “Great Reversals,” pp. 15, 17.

* Japan’s stock exchanges were closed from 1945 to 1949 and recovery was slower than in Germany.

Apart from a tightening of the commercial code in 1911, which strengthened directors’ liabilities for negligence, ^{Footnote 74} it is difficult to see any early German-style positive impact of corporate law in protecting shareholders in Japan. It has been argued that both law and Nazi politics shifted Germany from its strong initial (pre-1914) shareholder value culture to one that stunted stock exchange growth (as shown in Table 4). ^{Footnote 75} Why did Japan, with an apparently similar commercial code, develop so differently? Historians, noting the widespread development of stock exchanges before legal protections for investors were seriously implemented, have speculated that there were alternative mechanisms protecting shareholders, such as information signaling and monitoring by investment bankers, high dividend distribution disciplining boards, and private-order regulation of corporate charters by stock exchanges of corporate charters (such as anti-director rights or transparent accounts), which promoted shareholder confidence. ^{Footnote 76} As far as we are aware, there is no study of whether interwar Japanese corporate charters voluntarily adopted good governance rules. Okazaki et al. acknowledge that shareholders held ultimate control of corporate activities through the appointment and dismissal of directors at shareholder meetings, but argue that one of the reasons for the remarkable post-1918 growth of the Tokyo Stock Exchange, far from a strengthening of investor protection, was a relaxation of the rules, enabling the exchange to trade securities already traded on the outside spot market that had not applied for listing, thus improving market liquidity. ^{Footnote 77}

Another possible candidate for explaining Japan’s remarkable early growth in the quoted capital/GDP ratio is the voluntary emergence of gatekeepers and information signallers trusted by investors. Signaling by influential Meiji reformers such as Eiichi Shibusawa may have been an important supplement to legal regulation in Japan, and the zaibatsu may later have been an institutional substitute for weak legal investor protections. ^{Footnote 78} We thus provisionally view the strong development of Japanese stock exchanges in the interwar years as a result of intense monitoring of professional managers by large investors (in some cases, zaibatsu families), rather than wide dispersion of holdings driven by strong investor protections. It was postwar Americanization that placed stronger investor protections on the statute book, but the occupying authorities also dissolved the zaibatsu and expropriated large shareholders. If this caused a loss of valuable signaling and/or monitoring, it provides a neatly compatible explanation of the lower postwar penetration of Japanese stock exchanges. The later recovery may have been caused as much by alternative institutional innovations by main banks and keiretsu as by formal investor protections. ^{Footnote 79}

An Asian Model?

Was Japanese corporate law—and the flexibility with which Japanese businesspersons used it —in some sense part of a different Asian or Asia–Pacific model of corporate development ^{Footnote 80} than its plainly European origins suggest? Not on the dimensions we have so far discussed. Japan was precocious in using corporations (and developing stock exchanges) compared with most Asians, but was by no means the only case. Singapore and Hong Kong (which adopted Western corporate laws decades before Japan) consistently ^{Footnote 81} had (and still have) more corporations per capita than Japan, and Hong Kong had the highest ratio of quoted capital to GDP in the world as early as 1910. ^{Footnote 82} India also adopted English corporate law decades before Japan adopted the German model in 1899, yet still at the latter date had fewer corporations (though initially more corporate capital) than Japan, even though Japanese corporations until that time required individual government authorization or lacked legal clarity. India’s corporate numbers per capita grew steadily but slowly under British rule and, after independence in 1947, actually declined for several decades. Despite China’s adoption in 1904 of a commercial code very similar to Japan’s, ^{Footnote 83} China had fewer corporations even than India and, after Mao’s triumph in 1949, corporations were more comprehensively suppressed than under India’s disastrous “license Raj.” Korea and Taiwan were well behind India in the early decades of the century, but when the Japanese authorities extended free incorporation (on their own 1899 model) to their colonies (in 1920 and 1923, respectively) there was catch-up. ^{Footnote 84} After their unexpected independence in 1945, they followed the Singapore and Hong Kong rather than Indian trajectory, continuing to develop the relatively liberal corporate regime inherited from their colonial masters.

The main differentiator of the future tigers of the Asia–Pacific region from other underdeveloped regions in the early twentieth century was the exceptionally small size of their corporations, compared with those in the underdeveloped economies of eastern Europe, Latin America, or Africa. ^{Footnote 85} The average size of Japanese KKs in 1910 was only ¥247,612 ($125,819), in Korea $152,431, and their equivalents in the Netherlands East Indies (modern Indonesia) $128,260, and in the Straits Settlements (modern Singapore and parts of Malaysia) $138,678. If such countries had any advantage over other underdeveloped economies (where average corporate sizes of more than $1 million were not uncommon), it seems to be the absence of crony capitalism. This limited incorporations and access to finance to favored groups in Latin America and elsewhere and depended more on states “picking winners.” ^{Footnote 86} A multiplicity of small corporations perhaps encouraged Schumpeterian creative destruction—multiple foundations and multiple bankruptcies—and a more successful, risk-taking growth model than crony capitalism. However, the largest Asian populations suffered similarly to unsuccessful non-Asian models: The average size of Chinese corporations in 1910 was as high as $1,314,851 and, in India, $410,339. That structure left less room for creative destruction there.

Similar relativities are observable for companies quoted on stock exchanges. The average size of 389 Japanese companies quoted on Tokyo, Osaka, and some other exchanges in 1913 was, in dollar terms, ^{Footnote 87} only $3.2 million, compared with $34.9 million for the 298 listed on the New York Stock Exchange about the same time, $4.8 million for the 910 listed on the Berlin Bourse and $14.1 million for the 1,198 listed on the London Stock Exchange. Japanese quoted KKs were also smaller than the $16.4 million average for the 117 listed Argentine companies, $9.6m for the 75 listed Egyptian companies, or $7.1 million for the 113 listed South African companies. ^{Footnote 88} This was partly because of Japan’s low minimum listing size: paid-up capital of only ¥75,000 ($37,500) had been required since 1894 and, even when this was raised for new listings for forward trades, many small companies continued to be traded on the spot market. Furthermore, after railway nationalization in 1905, Japan (like Germany and South Africa) lacked this large formerly quoted sector, which still dominated many other stock exchanges.

The average corporate size in Japan will, to some extent, be understated because neither the overall KK statistics nor the stock exchange statistics group corporate subsidiaries into zaibatsu holding company groups, the major ones until the war typically controlled by (unquoted) family partnerships, not KKs. ^{Footnote 89} Seven major zaibatsu accounted for 17.5 percent of the paid-up capital of all KKs in 1928, ^{Footnote 90} nine zaibatsu controlled several hundred KKs with 15.1 percent of paid-up capital in 1937, ^{Footnote 91} and their dominance was then accentuated under wartime controls. ^{Footnote 92} However, some 98 percent of KKs with 85 percent of KK capital were not in major zaibatsu in 1937; these included most of Japan’s fifty largest enterprises, many listed on stock exchanges, with shares more widely dispersed than closely held zaibatsu subsidiaries.

Moreover, Japan was not unique in these dimensions. All countries’ corporate size distributions are highly skewed (with several giants and numerous small companies) and many enumerated in other countries were also subsidiaries of interlocking enterprise groups and corporate pyramids. ^{Footnote 93} Gardiner Means, now celebrated mainly for his work with Adolph Berle on the decline of American insider ownership, actually concluded that four interlocking business groups dominated by the Rockefeller, Du Pont, and Mellon families and Morgan financial interests in 1935 controlled more than 23 percent of U.S. corporate assets. ^{Footnote 94} They thus dominated the U.S. economy more than the leading family zaibatsu in Japan and, in absolute terms, were much larger. There were also smaller interlocking groups in both countries. The American New Deal bureaucrats occupying Japan in 1945 faced few constraints and forcibly dismantled its enterprise groups, whereas in the United States they had to make do with anti-holding company laws, taxes on intercorporate shareholdings, and antitrust measures against similar pyramided groups. The U.S. groups shared features with, but were not identical to, the zaibatsu, ^{Footnote 95} although the Mellon interests perhaps came closest in terms of degree of diversification and nature of interlocking controls. In Germany, also, similar holding company groups (Konzerne) with subsidiary and associate companies increased the average size of grouped companies above that shown in the unadjusted corporation or stock exchange statistics. ^{Footnote 96}

One aspect of Anglo-American commercial law that initially struck many Asians as odd (in addition to fully limited liability joint stock corporations) was joint and several liability in partnerships, by which all partners—or any one of them—were liable for all partnership debts to the full extent of their assets inside and outside the partnership. In much traditional Asian law, it seemed more equitable when a partnership failed for the active partners to share liability for its debts in proportion to their shares in capital or profits, although in China creditors might bargain to adjust the normal settlement according to the relative wealth of the partners, and in the Japanese civil code equal (but not joint and several) liability was accepted when partners had not disclosed their respective shares. A conflict with traditional notions of equity may be one reason why the take-up of Western-style partnerships in China’s 1904 commercial code was so low. Even in Hong Kong, where Chinese entrepreneurs more enthusiastically espoused Western laws, the British authorities proved more flexible than Manchu legislators on the mainland, adding to the enterprise menu in 1911 a partnership form for the Hong Kong Chinese that conformed more closely to native preconceptions. ^{Footnote 97} Although there was no similar modification of the Japanese commercial code, this was arguably because the contractual options for liability and governance in goshi kaisha and in civil code kumiai partnerships already left considerable scope for traditional-minded Japanese to mold their enterprises to a culturally congenial form and, for a time, to ignore newfangled Western models that the Meiji elite were keen to promote (and that eventually predominated).