The Rise and Fall of Imperial China: The Social Origins of State Development By Yuhua Wang. Princeton: Princeton University Press, 2022. 322 pp. $120.00 (cloth), $35.00 (paper)

Why did the Chinese empire last for two millennia, only to decline at the end of the nineteenth century? And what fundamental principles can China's example suggest to explain why some states in world history have prospered and others have failed? Wang Yuhua, in his new book, The Rise and Fall of Imperial China, seeks to answer these questions. He explores China's history from the Tang Dynasty (618–907) to the fall of the Qing in 1911, basing his analysis on datasets and sophisticated statistics, including regression analysis. Wang is a specialist of government; his book is thus primarily meant to be read by political scientists. Nevertheless, Chinese historians will find the overarching thesis, as well as many of the arguments in individual chapters, to be both engaging and illuminating. Notably, Wang offers a new way of conceptualizing the “localist turn”—usually associated with the Southern Song—and its impact on China's state–society relations. The author can also be commended for his critique of the eurocentrism of his own discipline, and for placing China's history into the political science conversation.

The central contention of the book is that state development (both the forms states assume and whether they are strong or weak) depends at the most fundamental level on the “elite social terrain”—meaning here the marriage and kinship ties connecting “central elites” to “local social groups” (7). Wang envisions three ideal types of elite social terrains, each associated with a particular state–society relationship (14). In a “star network,” which existed according to Wang during the Tang Dynasty, central elites have close ties both to each other and to local elites in the provinces. This particular configuration produces a weak monarch who faces a powerful elite, but a strong state, because that state's interests align with those of the elite. In a “bowtie network”—which Wang argues characterized China from the Song Dynasty until the nineteenth century—the central elite is divided into factions, each with its own network of connections to local society. The monarch is strong vis-à-vis the fragmented central elite, but the state is relatively weak in relation to an entrenched local elite. Wang refers to the paradox of the monarch gaining authority at the expense of state strength as the “sovereign's dilemma.” Finally, in a “ring network,” factions within the central elite have close ties neither to each other nor to local elites. According to Wang, this configuration reflected the situation in post-Opium War China, when state authority crumbled as government functions were entirely taken over by local private-order institutions.

Clearly, in Wang's model, state development is determined not by the economic foundation of society (as Marxists and other materialists would have it), nor by the quality and strength of government institutions (as traditional historians in China have tended to assume). There is also no significant place for cultural particularities, nor for historical contingency. Instead, the structure of the elite determines state development. My own inclination is in fact to agree with Wang: when assessing a state's stability and its capacity to exert authority, I tend to see the ways in which a bureaucrat fits into broader elite social networks to be more consequential than how the bureaucrat's office fits into institutional hierarchies. I would thus expect elite kin networks to have a far-reaching impact on the political world. I also find Wang's analysis of longue durée trends on the basis of large datasets to be an appealing methodology. Some of the clever metrics the author devises would be well worth adapting to other scholarly projects. Nevertheless, I also have some fundamental disagreements with Wang, and will propose ways in which his model might be refined to better fit the historical data.

The Rise and Fall of Imperial China is divided into nine chapters. Chapter 1 is an introduction that gives a synopsis of the argument and its theoretical foundation, framed for political scientists. Chapter 2 provides a big-picture overview of China's state–society relationship over the two millennia of imperial rule, and introduces the most important metrics used in the book. To assess the elite social terrain, Wang measures both the “social fractionalization” of “major officials” (i.e., the degree to which these officials were connected to each other by a single, large marriage network), as well as the major officials’ “kinship network localization” (i.e., the geographic dispersal of their marriage ties). To evaluate state power, he looks at the state's fiscal system, most notably by assessing per capita tax revenue. To evaluate the monarch's authority, he looks at the average reign length of emperors, as well as the frequency of palace coups. Finally, the chapter attempts to establish correlations between climate anomalies, mass rebellions, and external invasions.¹

Following the introductory material, the next six chapters deal in turn with historical periods from the Tang through the Qing, each chapter offering a historical overview focusing on questions of elite social structure, state–society relations, and the fiscal and military foundations of state power. These overviews are primarily for political scientists who do not specialize in China. Historians may quibble with points of detail and with some of the oversimplification. But such criticisms would have little bearing on Wang's overarching argument. Each chapter also supports some of its main contentions with one or more “original” datasets.² Chapter 6, for example, demonstrates that the number of years needed for a prefecture to implement the late-Ming Single Whip reform increased with the number of major officials who came from that prefecture. Chapters 7 and 8 use the publication dates of lineage genealogies to quantify collective lineage activity by time and place. They then show how this metric correlates with both jinshi success and the frequency of local uprisings. Finally, Chapter 9 shifts course to explain how an understanding of elite social terrain might benefit policymakers seeking to shape state-building projects in Africa, Latin America, and the Middle East.

Let me now turn to three specific points of disagreement relevant to my own areas of expertise.

1. A “Star Network” or a “Bull's Eye” Network?

Elite marriage networks offer a powerful tool for understanding state–society relations. A case in point is Wang's notion of a “bowtie network,” which I find to be exceptionally useful for conceptualizing Chinese state–society relations after the “localist turn” of the Southern Song. Localism incorporated a number of interconnected elements, including a change in family strategy, as members of the elite came to see that their future lay in provincial society and not in the service of the central government; a concomitant shift from a “court-oriented” to a “shi [literati]-oriented” elite culture; a new “social contract,” in which a provincially based elite took over many state functions, including local defense and social welfare; a novel ideology promoting “voluntarism” as an alternative to government service; and a much broader geographic dispersal of political power.³ Finally, and of particular relevance here, localism entailed a shift in marriage patterns, such that members of the central elite ceased to intermarry with each other in order to entrench themselves in local elite society.⁴ The consequence was a dissolution of the horizontal bonds between members of the central elite, and the strengthening of vertical links connecting them to social groups in the provinces—vertical links defined first and foremost by kinship ties, as the “bowtie” model suggests.⁵

Wang's analysis of the earlier Tang period, however, is much more problematic. Specifically, the “star network” misrepresents the basic structure of Tang elite society. The Tang bureaucracy was dominated by a tightly knit marriage network of capital-based clans, which—contrary to Wang's claim that capital officials maintained “marriage ties with multiple families with home bases in the provinces” (41)—in fact included the clans of both officials and their in-laws. It is true that the capital region by my definition included the “Capital Corridor” connecting Chang'an to Luoyang, and that there were small pockets of the national elite in two or three of the southern metropolises.⁶ But Wang is simply incorrect to conclude that “while the central officials were concentrated in the capital area, their kin were dispersed across the country” (42). To assess the claim (depicted graphically in Figure 2.5b on p. 41) requires careful examination of the R code and datasets in the “replication package”—which Wang has made available for download to allow the reader to replicate his results.⁷ There are difficulties (e.g., some tables identify individuals only by pinyin names), so I consulted the China Biographical Database (CBDB) and my own very extensive database of Tang individuals.⁸ My conclusion upon scrutinizing a sample of Wang's data is that the non-capital “kin” appearing in Figure 2.5b are erroneously localized to their ancestral places of clan origin.⁹ This is a fundamental error (which many historians also make), as the Tang great families had almost entirely relocated to the capital region by the end of the seventh century.¹⁰

One can take the capital official Li Zhengchen 李正臣 as an example. In the replication data, Wang correctly identifies his hometown as the Eastern Capital of Luoyang (and not as Jiangxia 江夏, his place of clan origin).¹¹ But the first ten of Li's kin listed in Wang's dataset consist of family members of his son-in-law Cui Yuanlue 崔元略 (771–830)—including Cui's grandfather, father, and several brothers—and Wang assumes that their hometown was Anping 安平 County, the place of origin of this branch of the Boling Cui 博陵崔 clan.¹² In fact, from other available data, we know that Cui Yuanlue's ancestors had left Anping generations earlier. We know the names of at least thirty-six Cui clan members who were buried in Luoyang beginning in the mid-seventh century.¹³ Indeed, Cui's own epitaph and that of his grandfather, both excavated in recent decades, confirm that the two men were buried in Luoyang.¹⁴ One can point to other such examples in Wang's replication data.¹⁵ Simply put, I see no evidence of a nationwide marriage network in Tang times. What distinguished the Tang sociopolitical elite from its counterpart after the localist turn in the Southern Song was not the geographic dispersal of its kinship network (41–42), but rather its heavy concentration in the capital region.

As an alternative to the “star network,” I would propose a “bull's eye network,” in which dense marriage ties connect central elite families, which, however, maintain no familial connections to local elites. The Tang court in alliance with a strongly cohesive, capital-based aristocracy (i.e., the central elite) dominated provincial society. But it did so not via the aristocracy's marriage network, but rather because local elite families in different parts of the vast provincial periphery could not join forces and stand up to the alliance of imperial power and capital elite families. Under these circumstances, Wang's measure of the “fractionalization” of the marriage network of major officials is considerably more useful than his kinship localization score for contrasting the Tang aristocracy with the political elite after the localist turn. To distinguish the “bull's eye network” from other types of networks, one might also devise metrics quantifying an elite man's probability of attaining high office, or, as I have suggested elsewhere, the degree of concentration at the capital of the clans that produced major officials.¹⁶

2. The Timing of the Elite Transformation

When did the elite transformation take place? Wang explains in Chapter 4 that “the transition happened in the late Tang era” when “a mass rebellion … destroyed the medieval aristocracy” (94). To demonstrate that a new order was in place by the beginning of the Northern Song, he deploys graphs, including one depicting the dramatic spike in importance of the jinshi exam (Figure 4.5 on p. 98)—a phenomenon that has traditionally been treated by historians as marking the appearance of a new elite.¹⁷ Given that Wang's account of the demise of the aristocracy paraphrases my own book's conclusion, I have no fundamental objections to this particular element of his account.¹⁸ But the fall of the aristocracy should not be conflated with the “localist turn”: the transformation of the elite entailed a long, multi-step process. The generally accepted model, first articulated by Robert Hartwell, envisions a two-stage transition from aristocracy in the Tang to a national civil-bureaucratic elite in the Northern Song, then to a local elite in the Southern Song.¹⁹ Moreover, as I will suggest, the second stage was itself a complex and drawn-out process.

In contradiction to this model, Wang argues in Chapter 5 that localism had already manifested itself in the eleventh century amid the well-known political factionalism surrounding Wang Anshi's New Policies. The model derived from Hartwell's thesis treats this factionalism as discord between members of the national elite; the value of “action from the political center” was taken for granted across the factional divides.²⁰ But Wang argues for a more fundamental transformation. He envisions a rivalry in the Northern Song between two groups. The first was a “patronage-based elite,” whose members tended to gain office through the hereditary yin privilege, to engage in cross-regional marriage alliances, to have interests that aligned with those of the state, and to support the New Policies. The second group consisted of “newly-risen scholar-officials,” who were much more likely to gain office through the civil service examinations, to have localized social networks, to focus on local interests and on “keeping the state weak,” and so to oppose the New Policies in favor of local, private-order institutions (106–7). Wang's evidence to support the existence of this particular dichotomy, however, is problematic.

To give readers a sense of the differences in geographic scope of the marriages of the two elite types, Wang compares the kinship networks of Wang Anshi (1021–1086) and Lü Gongzhu (1018–1089) (Figure 5.2 on p. 122). Wang Anshi's network is indeed empire-wide, including individuals from the north, from Huainan, as well as from Jiangnan and Sichuan. Lü's network, by contrast, includes individuals only from the north and from Huainan. But I would point out that even this second network was rather widely dispersed, spanning at least 1100 kilometers—comparable to the distance between Philadelphia and Chicago, or Paris and Rome. Lü's kinship ties were thus a far cry from the elite marriages after the localist turn of the Southern Song, which tended to be restricted to a single county or prefecture. Scale matters here. It was only prefecture- or county-wide marriages that created a kin network that was sufficiently geographically concentrated to permit an elite family strategy of local entrenchment. Despite the author's claim, Wang Anshi and Lü Gongzhu both very much exemplified the national elite of the Northern Song.

In fact, there is an alternative set of correlations more in line with what historians have observed in the past. To explore further the contrast between supporters and opponents of the New Policies, we can once again examine Wang's own R code and datasets. From a list of 137 Northern Song officials, the author identifies—on the basis of his research team's own reading of Northern Song chronicles—63 supporters and opponents of Wang Anshi's reforms.²¹ It is on the basis of this relatively small sample size that Wang concludes that kinship network localization negatively correlates with support for the reforms.²² As an alternative to kinship network localization, I used Wang's data to assess the prefecture of origin of the 22 consistent supporters and 34 consistent opponents of the New Policies whose hometown is identified in the table.²³ I then classified each prefecture into either north (including Shannandong) or south (including Huainan), and also determined the number of Northern Song-era jinshi recipients in each respective prefecture.²⁴ I found that 64 percent (14/22) of opponents of the New Policies were northerners, while 71 percent (24/34) of supporters were southerners—reflecting a regional dichotomy that others have previously observed.²⁵ Given that the north remained the center of political power in the Northern Song, one would imagine that upwardly mobile southern elites were more likely to seek northern marriage partners than vice versa. The regional dichotomy thus can explain why Wang found that supporters of the New Policies had more geographically dispersed kin networks. On the question of the exams, it turns out that a substantial percentage of high officials on both sides of the issue had entered the bureaucracy through the exams.²⁶ But there is one notable point of contrast: I found that only 27 percent (6/22) of opponents of the New Policies came from prefectures producing over 100 jinshi in the Northern Song, whereas 56 percent (19/34) of supporters came from such high-performing prefectures. It was thus proponents of the New Policies who more likely emerged from a prefectural elite that benefited from the exams. In brief, I see few grounds to conclude that a “newly-risen” examination elite with “local” kinship networks constituted the opposition to Wang Anshi's reforms, as the author argues.

One can better account for the data by treating the demise of the aristocracy and the rise of localism as two distinct steps occurring at different moments in time. On this issue, Wang's data makes a useful intervention. Robert Hymes’ classic work on localism unambiguously dates the phenomenon in Fuzhou, Jiangxi Province, to the Southern Song, when one witnesses a much greater prevalence of intra-county marriages.²⁷ Peter Bol's data from Wuzhou (now Jinhua, Zhejiang Province) complicates the picture, showing both that intra-county marriages there became prevalent somewhat later in the Southern Song, and also that intra-county networks during the Yuan became further geographically constricted to clusters of neighboring villages.²⁸ Wang's data complicates the picture in yet another way. His visualizations of the major officials’ marriage networks show that Northern Song and Southern Song networks look very similar (Figure 4.4 on p. 93)—corroborating Beverly Bossler's argument that there survived into the Southern Song a stratum of locally based national elites who continued to specialize in officeholding and to marry nationally.²⁹ But by the late Ming, there was a much greater fractionalization of the marriage network, as shown in Figure 6.3 (150), suggesting that the stratum of national elites described by Bossler had by then disappeared. In sum, Wang's data helps us to see how the phenomenon of localism, first apparent in certain regions of China in the early Southern Song, continued to evolve in complex and interesting ways in subsequent centuries.

3. The Strength of the State in Late Imperial China

Finally, let me address the issue of the “fall” of Imperial China. Did the post-Song state–society relationship produce a fiscally and politically weakened state, culminating in the collapse of the Qing at the turn of the twentieth century? This question is at the heart of the “sovereign's dilemma.” Wang's view is unambiguous: “China's state weakened during the Song-Ming times” (24). His contention aligns with numerous classic accounts of China's long-term development. Nineteenth-century theories of “oriental stagnation” and Naitō Konan's vision of “precocious modernity” in the Song posited that the Chinese empire had become too stable for its own good.³⁰ In more recent decades, theories of post-Song stagnation have taken the guise of modern “social scientific” analysis, as exemplified by Mark Elvin's “high-level equilibrium trap,” as well as the Maddison Project's problematic dataset of historical GDP per capita.³¹

But historians of Late Imperial China nowadays generally disagree with this premise. Indeed, one could argue that much of the work on Ming–Qing history in English dating to the 1980s and 1990s made it a goal to demonstrate the enduring vitality of China's state and society. Kenneth Pomeranz, for example, has famously argued that China's economy did not begin to diverge from the West until around the year 1800, suggesting that “stagnation” did not set in until a much later date.³² Wang counters Pomeranz by citing in a footnote specific scholarship that implies that “stagnation occurred much earlier—in the fourteenth to fifteenth centuries” (24). But my sense is that the scholarship Wang cites on this question in fact remains a minority position.

China's enduring strength in the eighteenth century is best embodied by the extraordinary success of the Qing Dynasty's empire-building project. It was in this century that the Qing built up one of the largest land empires in world history, perhaps second in size only to the Mongol Empire of the thirteenth century. Wang's useful graph depicting “external wars” by decade (Figure 2.2 on p. 34) is illustrative of this phenomenon. One finds almost no external wars either at the height of the Mongol Yuan Dynasty (late twelfth through early thirteenth centuries) or at the height of the Qing (during the “long” eighteenth century). As a regime with northern origins, the Qing (like the Yuan) benefited from its mastery of political and military techniques necessary to subjugate the forest and steppe tribes of the north. Yet Qing imperial expansion also depended on the eighteenth-century state's ability to secure a wide range of vital resources.³³ Although the Qing did ultimately “fall,” China's recent “rise” suggests that the post-Opium War decades may have simply constituted a particularly tumultuous political transition, not fundamentally different from the post-Huang Chao century (880–979)—with the resurgence of a “new China” in the mid-twentieth century analogous to the resurgence of the Chinese empire under the Song.

How does one explain the endurance of China's strength and vitality following the “localist turn”? Wang treats the fragmented elite of the “bowtie” model as an inherently centrifugal force, pulling the state apart by advocating first and foremost for local interests at the expense of national concerns. But there is another way of looking at it if one's point of contrast is not the “star network,” but rather the “bull's eye network,” in which the state forcibly imposes its hegemony upon the provinces. From this alternative perspective, the “bowtie network” constitutes a centripetal force that draws in and stabilizes local societies. More precisely, in the case of China, because every prefecture could at any given moment claim several native sons in the central bureaucracy, the prefectural elite acquired a new stake in the system. Meanwhile, by taking over some state functions (local defense, social welfare, etc.), this prefectural elite helped to make it possible for the state to sustain itself on smaller tax receipts.³⁴ This devolution of state authority might be seen as analogous to the devolution of authority in a federal system (such as one finds in the United States and Switzerland today). The “sovereign's dilemma” might be recast thus not as the choice to strengthen the monarch at the expense of the state, but rather as the choice to distribute political power and institutional responsibility more evenly across the state's territory.