Chen's cage theory

Historians tend to trace the origin of Chen Yun's cage theory to the period of 1961–62, when Chen was sidelined by Mao, who personally preferred a more radical approach to socialization.Footnote ²⁵ An early study on Chen Yun and Chinese politics noted,

Chen's restricted view of the market economy contrasted with Deng's more liberated view. The political as well as ideological position differences between these perspectives underlies the tension of the marketization and stabilization debate in the 1980s. Following the Tiananmen tragedy, the promotion of Shanghai party secretary Jiang Zemin was widely seen as Deng's concession to Chen. However, Deng was not pleased with stalled economic reforms and worried about the historical legacies of his economic liberalization policies.Footnote ²⁷ The collapse of the Soviet Union offered Deng the exact moment for a political maneuver to win back the party's support for his version of market socialism.

Deng's political maneuver during the Southern Talks

The identification strategy in this study relied on Deng's Southern Talks to be credible through an immediate change in formal institutions (e.g., the forthcoming party congress). While Deng's message could generate differential expectations in potential entrepreneurs, if not their beliefs about possible future policy reversals, it is essential to note that Deng's political venture and its decisive impact on the planned versus market debate have added a strong sense of irreversibility. This was achieved in a difficult situation beset by a confluence of factors: the collapse of the Soviet Union a month before his visit, the failing health of the older statesmen among the party conservatives,Footnote ²⁸ the fading of memories about Tiananmen, and finally, the urgency that a political report would be drafted at the forthcoming party congress.

Meeting the challenges of the party and new situations globally, Deng Xiaoping went to Guangdong Province, in the southern part of the country, where he spent five days in Shenzhen and another six days in Zhuhai, two of his signature cities in the Special Economic Zone exemplified by his reform and opening programs.Footnote ²⁹ In each place he visited along with his family, Deng Xiaoping reasserted the reformist spirit, exhorting local cadres to experiment with bolder steps and kill their way out for a better future. In particular, Deng looked at Shenzhen and hoped it would one day speed up its development by catching up with the four little dragons: Singapore, South Korea, Hong Kong, and Taiwan.

The most difficult challenge facing the party leadership was ideological obstacles, given the collapse of the Communist Party of the Soviet Union and the tragedy of Tiananmen. The public mood was low and there was an expectation of more crackdowns as policy programs of the anti-color revolution unfolded. The theoretical issues over whether, to what extent, and how a genuine market economy could be compatible with the ideals and theories of classical socialism, however, were deliberatively brushed away, as Deng famously asserted in one of his Southern Talks, which was circulated first as a formal party document and then crystalized posthumously (around 1997) as part of the CPC's official ideologies and guiding principles, Deng Xiaoping Theory:

While giving informal sessions to local party officials, reporters from Hong Kong spread the notes of Deng's speech, an act that was difficult for the conservatives in Beijing to ignore. Incumbent party secretary Jiang Zemin remained alert to the elders’ reaction and the active shots fired by Deng. Meanwhile, the Hong Kong newspaper and television began to spread Deng's message. By the time the Southern Talks ended on 30 January 1992, the official party media was silent on Deng's trip to Guangdong and made no political announcements as rumors spread from the southern part of the country. Immediately afterward, Deng Xiaoping went to Shanghai with his family to enjoy the Chinese New Year for about three weeks before heading to the capital for a political cleanup, anticipating the outcome of his finale.

The immediate and lasting impacts on China's policy regime regarding the market economy

It was no easy task, even for paramount leader Deng Xiaoping, to solidify his reform program committed to marketization and command the support of the party apparatus. During the celebration of the Chinese New Year, party secretaries and high officials paid visits to Chen Yun for advice, waiting for the changing winds to come. Deng Liqun, a notable party hard-liner, shot back, clearly reflecting the planner's message: “People who stubbornly cling to their liberal beliefs are using anti-leftism as a pretext to oppose the Communist Party leadership and the socialist system.”Footnote ³¹

However, by then the political tide had shifted entirely in Deng's favor, partly because the conservatives, confronted with new challenges, had nothing new to offer. A notice prepared by Beijing on Deng's talk was circulated to all Central Committee members by the end of February. At the March Politburo meeting, President Yang Shangkun and Party Secretary Jiang Zemin fully supported Deng's statements. The party center implemented Deng's political spirit into concrete policy packages in late May. Jiang, on 9 June, delivered a speech on Deng's address to high-level students at the Central Party School. By then, the reformers had won. As his conservative colleague, Li Xiannian, reached the end of his life, Chen Yun visited Shanghai Pudong and approved further economic development efforts, despite his usual cautious tone. As Deng Xiaoping consolidated his victory and captured the political agenda, the 14th Party Congress held on 12–18 October put a striving goal into the party's politico-economic agenda: a socialist market economy.

The year of Deng's Southern Talks was fascinating as an example of organizational transformation and institutional adaptation marked by a combination of elite-driven policy debates with entrepreneurial responses to the society at large. Fighting over Maoist and quasi-Maoist (e.g., Hua Guofeng), Deng fought back to high-level politics through a Maoist critique of Maoist errors seeking truth from facts. Reigning in party liberals Hu Yaobang and Zhao Ziyang, two successors he had handpicked, Deng Xiaoping scapegoated them for failing to handle strategic criticisms hurled against the party because of liberating policy throughout the reformist path. In the case of Hu Yaobang, a good-hearted comrade indeed, Deng considered him too soft on the party's political enemies, who disagreed with the fundamental lines of the party and could not be reasonably convinced. For Zhao Ziyang, the failed pricing reforms almost invalidated the “dual-track” approach by escalating student protests and public anger over profiting from the gap between “planned price” governed by state quotas and market prices. The moment the policymakers were viewed as both promoters and beneficiaries of reform, the political logic and moral appeals crumbled into tears.

The tragedy of Tiananmen and the subsequent reassertion of power by the party conservatives, not to mention the promotion of Jiang Zemin, Shanghai party chief, seemed to have scored for the conservatives: the markets were allowed but would be kept within strict cages. Biding his time, this is where Deng Xiaoping's Southern Talks came in, taking on the conservatives when the two elders were ailing (Li Xiannian died in 1992 and Chen Yun in 1995). How Deng Xiaoping turned a losing game into a stunning victory for the reformist spirits amid a pessimistic and depressing mood after the Tiananmen tragedy and capitalized on that victory through formal party documents was a major historical event for the CPC. It encapsulated the phrase “market economy” firmly into the ideology of socialism with Chinese characteristics, eventually sidelining the theoretical debate over the relationship between capitalism, socialism, and markets. As such, Deng was rather sincere in saying that “China must watch out for the Right, but mainly defend against the Left.”Footnote ³²

Thus, from the perspective of leadership-driven institutional change, Deng's Southern Talks were recognized as a critical watershed event not only because of what he said during the Southern Talks, but also, importantly, how the battle of ideas paved the way for a final consolidation of a political victory over the competing view proffered by the cage theory of market socialism. With Deng Xiaoping and Chen Yun leaving the political podium, China's institutional transition into full-fledged market socialism was the outcome of an inevitable end of a set of conflicting ideological and political battles in the 1980s.Footnote ³³ This rather dramatic resolution of political conflicts offered a unique opportunity for estimating its impact on entrepreneurial outcomes.

Model setup and initial allocation

Consider an extended political economy model with occupational choicesFootnote ³⁴ and three kinds of agents: public officials, entrepreneurs, and workers. Each person i lives for only one period. There is a stable distribution of entrepreneurial abilities in the population such that it is common knowledge that this distribution has stable support [1, ∞), density function f (A), and cdf F (A). There is a single output y in the private economy. Agent i with ability A_i can become an entrepreneur by exploiting existing technology r(.) by hiring labor units h.

In addition, the quality of the public sector G is taken to be fixed. Upon the occupation choice between workers and entrepreneurs, the talents face a sectoral choice between the public and the private sector, where a mixture of public motivations and rent-seeking opportunities drives the leading officials. The ability of public officials can be allocated between providing good public service for the common good and capturing higher-status rents for themselves.

The quality of the public sector G(.) = G(h^P) is assumed to be a function of the number of public sector workers h^P, as opposed to bureaucratic elites (i.e., those being able to share rents), where G(.) is imposed with standard concavity and Inada restrictions. The public-mindedness of the agent i is given by public motivation p_i, which has a stable support [1, ∞), density function q(p), and cdf Q(p). The variable can be interpreted as the highest gap in materialistic rewards that the public sector can offer less to keep the worker i with the ability A_i participate. A choice governs a trade-off among the elites of the society, i ∈ {P, E}, where P stands for the public sector and E the private economy as described in the previous section. The payoff for workers in the public sector and the private markets is w^P and w^E, respectively.

This model further assumes that becoming a political elite is more challenging than opening a small business, and therefore the ability cutoff for becoming a political elite (Ā) is higher than the ability cutoff for an entrepreneur (A).Footnote ³⁵ In addition, the linkage between the economy's private and public sector shares is technically bounded such that an additional s unit goes into the state's treasury for each unit surplus generated by the private sector.

The timing of events is as follows:

• • The social guardians (clarified in the next section) set the entry criteria for being a political elite Ā.

• • Agent i solves their optimization problems given their innate ability A_i and public motivation p_i and the economy's objective environment {w^P, w^E, Ā, A, s, r(.), G(.)} and other's optimal decision.

• • Under the equilibrium of this economy, no person i has an incentive to change their choices.

This simple model features a simple static equilibrium characterized in the following proposition, which is proven in the Appendix.

Theorem 1 pins down the initial allocation of talents in this model economy. This also gives the initial size of firms, the fraction of agents becoming workers, and the generated rents for political elites in this economy.

Debating the political economy of reform: elite competition and entry into entrepreneurship

Consider the institutional impact of Deng's Southern Talks political maneuver on talent allocation and, eventually, the size of firms. Because neither Deng Xiaoping nor Chen Yun was keen on promoting their heirs to the leadership position, the debate on the political economy of reform transcended dynastic politics and was consistent with a social guardian position.Footnote ³⁶ Notice that the entry criteria, Ā, is made exogenous for being a political elite in the prior discussion. This policy variable is controllable through the Organization Department of the CPC apparatus, which makes choices about personnel policies and selection criteria. Therefore, this is no wonder, viewed from the proposed framework, that the most crucial organizational policy is to find persons with a combination of high ability A_i and high motivation pi, who are well known as “red and technical” (you hong you zhuan) individuals capable of being technically competent as well as loyal to the party leadership. Now, the debate on the political economy of reform was to decide whether to impose a cage on the socialist market economy. This is modeled as setting a policy handle, $ \,\bar {\!h} $, on the maximum size of the private firm such that equilibrium labor hiring from the entrepreneurial side cannot exceed, $ h^{\ast} \le \ \bar {\!h} $. Comparing the size of gross domestic product (GDP) with and without such imposition on the size of private firms, the immediate and long-term impact of Deng's political moves can be formalized in the following theorem.

The proof of this proposition is straightforward from the setup of the model. Chen Yun's vision requires the imposition of a cage, $ \,\bar {\!h} $, over labor hirings. For those entrepreneurs such that their labor demand $ D> \,\bar {\!h} $, they are thus prevented from hiring labors beyond the cage limit. The private sector, viewed from the entrepreneurial surplus, will be smaller than what could be achieved efficiently. With a fixed s relating back to the public sector, the total compensation for public workers and political elites will be smaller. Unless h is so high as to be irrelevant, this implies that GDP_cage < GDP_market.

An intuition of this result can be provided. Chen Yun's blueprint requires the imposition of a cage, $ \,\bar {\!h} $, on maximum labor hirings. For those high-ability entrepreneurs with greater labor demand above this cage limit, there is a surplus to be gained, but they are prevented from being able to do so. The size of the private sector measured in terms of the entrepreneurial surplus will therefore be smaller. Consequently, it would translate into a smaller pie for public sector workers and political elites unless $ \,\bar {\!h} $ is so high that there is no policy difference between Chen Yun and Deng Xiaoping.

However ad hoc this political economy model is, the formal presentation offers two testable predictions from the credible removal of political as well as ideological barriers and consolidation of Deng's blueprint for the empirical work. Thus, switching from Deng and Chen's political impasse to Deng's victory allowed an expansion of the private economy measured by the number and the size of private businesses. This is listed as hypothesis H1.

In addition, the formal model predicts that around the impacting time point, noticeable entry into entrepreneurship shall be observed as political uncertainty gave away to business opportunities. In other words, the empirical data should witness the presence of a 1992 cohort of entrepreneurship, and this is listed as hypothesis H2.

Two stylized facts

The development of China's private sector was initiated in 1978 when the Cultural Revolution ended. Figure 1 plots two curves presenting the changing numbers of private firms and small owner firms (getihu) during China's reform era, as documented by the National Bureau of Statistics. While both curves witnessed an upward trend, the growth rate of private firms was more significant and accelerated after Deng's Southern Talks. This piece of evidence offers some support for H1.

Figure 1.

The number of private firms and small owner firms (getihu). The red curve represents the evolution of the number of small owner firms (getihu), and the blue curve represents the evolution of the number of private firms.

Source: National Bureau of Statistics.

Regarding the size of the private firms, a proxy measure from the wealth of the super-rich persons is considered. Building on the Hurun China Rich List (HCRL), a British publication, Lu, Fan, and Fu described the construction of the CSRP as a four-step process.Footnote ³⁷ First, they defined the super-rich as those who had ever ranked higher than 100 on the HCRL. Second, they collected basic personal information and the rankings on the on list for each year from 2000 and 2018. Third, they constructed a list of questions regarding the families of the super-rich and asked their research assistants to fill in those items using corporate websites, media reports, and other official datasets. Finally, the final dataset was formed with cross-checking, cleaning, and integration.

Figure 2 offers a picture of the wealth dispersions from the fifty super-rich Chinese persons from 2000 to 2018. Initially, the gap among the wealthiest persons was negligible before 2007, and that dispersion rose to a situation where a few superstar billionaires outpaced other super-rich persons. This documentation demonstrates increasing wealth from these super-rich persons and growing divergence of wealth dispersions.

Figure 2.

Wealth distribution of super-rich Chinese persons. This plot graphs the evolution of wealth distribution of super-rich persons in China from 2000 to 2018. For each boxplot, the dispersion is shown for a particular year.

Source: CSRP (Lu et al. Reference Lu, Fan, Han and Xiao2019).

Taking these two stylized facts together, the evolution of critical macro statistics on China's private sector development confirms that Deng's Southern Talks victory reaffirmed the public's confidence in the party's continuing development commitments. In addition, given the deaths of Chen Yun and Li Xiannian, the policy return to a political impasse seen in the 1980s became less likely.

The empirical part of the article concentrates on the second point of the prediction. Using Deng Xiaoping's Southern Talks as an exogenous political shock with credible commitment impacts, the effect on the changing social composition of the super-rich persons in China can be estimated. This novel approach has a number of benefits. First, the regression discontinuity kink design exploits a nationwide shock to gauge evidence for any likely structural changes. This research design contrasts with studies using local variation from specific policy experiments. Second, the regression discontinuity kink design offers some insights into the prediction of the models and addresses the role of generation shifts in changing the composition of super-rich persons. The intuition is that if politically induced economic uncertainty has been removed, the chance of becoming superstar wealth creators would have a time dimension associated with large entry into entrepreneurship around that period. In the absence of the event of the Southern Talks (i.e., with cage continuously being imposed or political and ideological debate unresolved), the effects from mechanical generation shifts could be much smaller than what can be captured from the data as predicted by the proposed theory. Third, this does not imply that generation turnover does not matter. Indeed, the entry into entrepreneurship induced by Deng's Southern Talks could facilitate the subsequent generation turnover.

A regression kink design

The research design using an event shock like the Southern Talks suggests an empirical strategy based on a regression discontinuity kink design using semi-parametric techniques:

(1)$$y_{ipc} = {\rm \alpha } + {\rm \rho }_1T_i + {\rm \rho }_2T_i \times f( {{\rm time\ distanc}{\rm e}_i} ) + {\rm Z}^{\prime}{\rm \beta } + f( {{\rm time\ distanc}{\rm e}_i} ) + \Phi _{\,pc} + \it \$ _{ipc}$$

where y_ipc is the outcome of the research interest for a super-rich person i based in province p with cohort c, and T_i is an indicator variable equal to 1 if the person i entered into entrepreneurship after the Southern Talks, with zero otherwise. The company's founding year by person i, a running variable, is used to determine whether it was affected by the Southern Talks. Z_i belongs to a set of socioeconomic variables (i.e., characteristics of the entrepreneur) independent of the assignment variable T_i, and f (.) is the regression discontinuity polynomial function, controlling for the shape around the neighborhood of discontinuous shock. Lastly, Φ_pc is a set of interaction terms between province and cohort fixed effects. In some specifications, Φ_pc is simply modeled as Φ_pc = θ_p + γ_c.

Regression discontinuity design has been used widely in spatial models and election studies. In practice, f (.) represents a polynomial function of the gap to the year of the Southern Talks, 1992. The treatment assignment variable, T_i ≡ 1(Found Year_i > 1992), designates the post–Southern Talks binary. The treatment impacts, ρ_s where s ∈ {1, 2}, capture the impact of Deng Xiaoping's Southern Talks on the set of entrepreneurial composition under the assumption that the suitable window offers the right comparison. If ρ ₁ = 0, then the regression discontinuity design captures the structural change induced by the Southern Talks. If ρ ₁ ≠ 0, not only the outcome of interest change following the Southern Talks but also jumps at the discontinuous point.Footnote ³⁸

The estimating framework based on sharp regression discontinuity design relies on two critical identifying assumptions.Footnote ³⁹ One, the assignment variable, the time location to the Southern Talks, cannot be fully controlled by those choosing entrepreneurship before the Southern Talks yet affects later comers. This implies that the suitable choice of the window for the comparison between those entrepreneurs joining prior to the Southern Talks and immediately following the design approximates a good experiment. Formally speaking, let y_{i 1} and y_{i 0} denote two potential outcomes under treatment and control. The entry year of entrepreneur i determines not only the value of T_i, but also the time distance_i ≡ (the year of entry into entrepreneurship_i − 1992) from the Southern Talks. This value can be used as input for f (.). The running variable in the context can thus be the time distance defined above or an entrepreneur's raw year of entry. Identification requires at the discontinuous threshold, conditional expectation functions for E[y_{i 1}|td_i] and E[y_{i 0}|td_i] are continuous for an approximately good random assignment.

To assess the foregoing concern, I examine the summary statistics table (see Table 1) of two groups apart from the Southern Talks with a suitable choice of windows: full sample, 1980–2000, and 1985–1995.Footnote ⁴⁰ Across five target measures of social composition, none is significantly different in terms of difference-test and distributional test across the three window periods. I take this as a confidence check for a plausibly good random assignment. The case of a college degree of the person is quite interesting. Among the full sample and the window period of 20 years (1980–2000), the level of education increases significantly, indicative of substantial generational change, consistent with significant changes in distribution. Yet, column (9) of Table 1 showcases that the changes are less effective in a narrow window. This signal of evidence could support the story: Deng's Southern Talks somewhat facilitates generational changes by increasing the share of college graduates.

Table 1.

Summary statistics.

Source: CSRP (Lu et al. Reference Lu, Fan, Han and Xiao2019). Columns (1)–(3) represent the whole sample. Columns (4)–(6) represent the subsample in which entry into entrepreneurship was made between 1980 and 2000. Columns (7)–(9) represent the subsample in which entry into entrepreneurship was made between 1985 and 1995. Columns (3), (6), and (9) give the t-test in parentheses and the distributional test in square brackets.

In addition, the continuity of the running variable at the kink needs to be ensured against manipulation. To assess the validity of the assumptions underlying the regression discontinuity and regression kink design, Figure 3 offers a histogram and kernel density for the year of entering the first job (Panel A) and becoming an entrepreneur (Panel B). Most of the super-rich persons in our database were employed far earlier than the time of the Southern Talks. In contrast, the timing of becoming an entrepreneur was rather balanced across the two sides of the Southern Talks. Notably, the entry years of the super-rich persons seem to be concentrated around the timing of the Southern Talks and gradually fade out, suggesting that the high chances of becoming super-rich are somewhat associated with a small window period around the Southern Talks. Combining a potentially ample market space or opportunities and removing political as well as ideological barriers worked to the advantage of future billionaires.

Figure 3.

Histogram and kernel density: Entering year of first job and becoming an entrepreneur. Panel A depicts the histogram and kernel density of the entering year of the first job. Panel B displays the histogram and kernel density of the entering year as an entrepreneur. The red vertical line denotes the year 1992, when Deng Xiaoping made his Southern Talks.

Source: CSRP (Lu et al. Reference Lu, Fan, Han and Xiao2019).

Manipulation testing for falsification of regression discontinuity design requires that those entrepreneurs could preempt their investment projects knowing that Deng Xiaoping would do something to overturn the gloomy mood after Tiananmen. Or, potential entrepreneurs could set up their businesses before the political uncertainties were removed. This possibility was rather unlikely because it would have required knowledge of the timing of the collapse of the Communist Party of the Soviet Union and Deng's rather unconventional response via the Southern Talks. To address this concern, I falsify the continuity assumption of the running variable around the cutoff.Footnote ⁴¹ Appendix Figure A.2 presents a density continuity test near the cutoff.Footnote ⁴² A local polynomial density estimator turns out a robust T value of 1.3304 with an associated p-value of .1834, suggesting that manipulation of entry was rather unlikely.Footnote ⁴³

Graphical evidence

In this section, I estimate the impact of the Southern Talks on the changing social composition of entrepreneurs using several measures. In short, the aim is to look for plausible structure turning points and immediate impacts induced by Deng's political moves. While generational shifts might threaten the identification strategies, this lurking factor coevolves with rather than causes the composition change in the space of super-rich persons. In other words, Deng's Southern Talks facilitated the differences in these variables. In all graphical and regression evidence that follows, cohort fixed effects are included to minimize the unobservable generational shifting factors.

I begin my discussion by plotting three outcome variables for the CSRP against the running variable, as shown in Figure 4. These are binned scatterplots, a nonparametric method of plotting the conditional expectation function by absorbing unobservable provincial and cohort-level fixed effects. Among CSRP, Panel A documents evidence that the share of college degree holders increases until the Southern Talks. This kink significantly confirms that Deng's political victory marked promising entry opportunities for those entrepreneurs. Consistent with Panel C, where those well-educated persons with state sector jobs leave their careers around that window of opportunity, this supports a case in which high-ability, highly educated former state employees are changing their career paths and venturing into entrepreneurship, eventually making it to the top. In both cases, the highest share is reached at or around the time of the Southern Talks. However, Panel B documents that the percentage of party members does not change drastically before and after the Southern Talks.

Figure 4.

Graphical evidence: The impact on the person's attributes. This plot provides the regression discontinuity plot based on binned scatterplots after absorbing provincial and cohort fixed effects. The outcome variable of interests from Panels A–C: the share of a college degree, the share of party membership, and the share of state sector experience by the entrepreneurs. The red vertical line denotes the year 1992, when Deng Xiaoping made his Southern Talks.

Source: CSRP (Lu et al. Reference Lu, Fan, Han and Xiao2019).

Suppose market opportunities were correlated with political connections and family backgrounds, then similar changes might have happened to the attributes of the parental father of these super-rich persons. Accordingly, I plot these three measures for the parental father of the CSRP in Figure 5. The picture does not support a strong correlation between market opportunities and social position. The share of party membership does not change significantly before and after the Southern Talks (Panel E). Interestingly, the parental father's share of a college degree reaches a low point as the corresponding share for the person reaches the highest point (Panel D). In Panel F, the trend of a declining share of state sector experience for the parental father of the CSRP continues to drop after the Southern Talks. In short, these graphical results indicate that the pro-market political reform initiated by Deng benefits entrepreneurs with actual market abilities in contrast with sons or daughters of elite officials.

Figure 5.

Graphical evidence: The impact on father's attributes. This plot provides the regression discontinuity plot based on scatterbins, absorbing provincial and cohort fixed effects. The outcome variable of interests from Panels D–F: the share of a college degree, the share of party membership, and the share of state sector experience by the entrepreneurs by the entrepreneur's father. The red vertical line denotes the year 1992, when Deng Xiaoping made his Southern Talks.

Source: CSRP (Lu et al. Reference Lu, Fan, Han and Xiao2019).

Additionally, Deng's Southern Talks might have inadvertently facilitated generational shifts in the space of super-rich persons. Binned scatterplots are provided by Figure 6 on the age structure (Panel A) and gender composition (Panel B) of the entrepreneurs. Panel A documents a slight change in the age structure of entering CSRP: the cohort, on average, becomes younger following the Southern Talks, with a flatter slope. By contrast, the share of female entrepreneurs in the CSRP reversed from a growing trend and experienced a decline. These observations reveal a general trend that the billionaires in the dataset are driven by market success as opposed to parental connections.

Figure 6.

Graphical evidence: Other covariates. Panel A and panel B provide binned scatterplots of the age of the entrepreneur and gender composition (male = 1). The red vertical line denotes the year 1992, when Deng Xiaoping made his Southern Talks.

Source: CSRP (Lu et al. Reference Lu, Fan, Han and Xiao2019).

These patterns are informative in at least two aspects. First, there is a structural kink in the space of super-rich persons induced by the flow of events generated primarily by Deng's Southern Talks. Second, given ongoing political debates and the probability of a policy reversal, this window of business opportunity does not privilege too much those with political connections in general, which suggests that the policy impact of forthcoming pro-market reform does not particularly favor children of political elites or other privileged circles, as divided as they were during Deng's political maneuvers. In other words, this implies that Deng's message was transmitted in a leveled national information platform instead of targeted segments of society.

Regression estimates

Table 2 reports the regression kink estimates for the changing social composition of super-rich persons following the Southern Talks, with standard errors clustered at the provincial and cohort levels. Across all specifications, the estimated coefficients for the jump (ρ ₁) cannot significantly reject zero. Three specifications using second-order polynomials—columns (2), (5), and (8)—do not increase explanatory power, giving credence to an underlying linear trend. The kink effect (ρ ₂) on the share of college degrees—columns (1), (2), and (3)—offers an estimate between –0.027 and –0.037, confirming the graphical evidence of Panel A of Figure 4. This evidence shows that the share of college degree holders among CSRP dropped by about 10 percent within three years of the South Talks, without other changes.

Table 2.

The impact of the Southern Talks on the social composition of super-rich persons.

Notes: Standard errors in parentheses are clustered at the provincial and cohort levels. T stands for the incidence of the Southern Talks in 1992. TD refers to the time distance to the Southern Talks. An unreported constant is included in each specification. Columns (1)–(3) use the person's college degrees as a binary dependent variable. Columns (4)–(6) use the person's party membership employment as a binary dependent variable. Columns (7)–(9) use the person's state employment as a binary dependent variable.

A large number of higher education persons joining entrepreneurs, among whom quite a few stand out, suggests a window opportunity around the Southern Talks. The fact that these highly educated people flocked into founding their own businesses around that period is coupled with another fact that quite a few of these career switchers previously worked in the state sector. Based on specifications (7) and (9), the estimated kink effect is around –6 percent. Both values reach close to the top of their values at or nearby the Southern Talks. In the absence of Deng's political maneuver and the dampening effect of the Tiananmen tragedy, the sudden flocking into entrepreneurship driven mainly by good-quality talents within the party apparatus needs to be properly explained. Therefore, this is a telling sign of a 1992 cohort of entrepreneurs.

The kink effect on the share of party membership is somewhat less effective, a point confirmed by Panel B of Figure 4. The kink effect is estimated to be nonsignificant, failing to indicate that this value peaked during the event of the Southern Talks. In other words, this statistic, in particular, is less strongly related to the share of college degree holders and that of state sector experience among the CSRP sample. These estimates indicate that the presence of party membership is strongly associated with neither market abilities nor political background.

Table 3 reports the estimated kink effects for the parental father of these super-rich persons following the Southern Talks using the same outcome variables as for the persons themselves. Following results from columns (2), (5), and (8), the estimated effects on the second-order polynomial, as in the previous case, fail to generate meaningful results. Across all specifications, the estimated jump effects and the kink effects cannot significantly reject the zero thresholds. It meant that during the window of money-making opportunity around the Southern Talks, parental characteristics among CSRP were only marginally related to the core story. Those who flocked into entrepreneurship with a college degree and prior state sector experience had all sorts of parental fathers, making detectable changes in these attributes less visible. In contrast with the story of targeted information and political connection, the window of opportunities around the Southern Talks can be viewed as a relatively fair race open to all children of family backgrounds.

Table 3.

The impact of the Southern Talks on the social composition of parental father of the super-rich persons.

Notes: Standard errors in parentheses are clustered at the provincial and cohort levels. T stands for the incidence of the Southern Talks in 1992. TD refers to the time distance to the Southern Talks. An unreported constant is included in each specification. Columns (1)–(3) use the parental father's college degrees as binary dependent variables. Columns (4)–(6) use the parental father's party membership employment as a binary dependent variable. Columns (7)–(9) use the parental father's state employment as a binary dependent variable.

Holding constant the main specifications, The following exercise tests whether the main results are robust to different window periods against control variables. Table 4 deals with issues of multiple windows since regression kink designs seek to capture local effects. In the restricted sample for the 1980–2000 window—columns (1) and (3)—the estimated kink effects remain robust to the magnitude of a college degree share and the percentage of state employment at −3 percent and −6 percent, respectively. Further shrinking the sample size to a window period of 1985–1995—columns (2) and (4)—seems to coincide with the window of opportunity unleashed by the Southern Talks. None of the estimated kink effects can significantly reject zero at 5 percent, indicative of a period of sea jumping during which state employees switched their jobs and became entrepreneurs as their expectations over return prospects (adjusted for risk preference) began to shift among ideological as well as political debates over planned economy versus market economy.Footnote ⁴⁴ One could raise a concern that Deng's Southern Talks merely confirm underlying trends from the 1980s. If this were the case, it would be difficult to reconcile with a changing composition of entering entrepreneurs, as high-quality talents might wish to stay in their careers had political uncertainties not been resolved.

Table 4.

Robustness checks: Multiple windows.

Notes: Standard errors in parentheses are clustered at the provincial and cohort levels. The standard errors in the brackets are based on the Huber-White robust errors. T stands for the incidence of the Southern Talks in 1992, and TD refers to the time distance to the Southern Talks. An unreported constant is included in each specification. Columns (1)–(2) use the person's college degrees as a binary dependent variable. Columns (3)–(4) use the person's state employment as a binary dependent variable.

Table 5 examines this consistent relationship against two control variables: gender and ethnicity. Using the complete sample, the kink effect for the share of college degrees stands at 3.7 percent in the negative for both variables—columns (1) and (2). Similarly, under this complete sample, the kink effect for the share of state employment is estimated at 6.7 percent in the negative for these two controls—columns (3) and (4)—showing the variability of ethnicity and gender is small compared with the main effect.

Table 5.

Robustness checks: Control variables.

Notes: Standard errors in parentheses are clustered at the provincial and cohort levels. The standard errors in the brackets are based on the Huber-White robust errors. T stands for the incidence of the Southern Talks in 1992, and TD refers to the time distance to the Southern Talks. An unreported constant is included in each specification. Columns (1)–(2) use the person's college degrees as a binary dependent variable. Columns (3)–(4) use the person's state employment as a binary dependent variable.

One might argue that the year of the Southern Talks might not be the “correct” shock year for a regression kink design. Table 6 tests the timing of the kink effects using four alternative timing indicators. Looking at Panels A and B as the timing of the event “moves” one or two years ahead, respectively, the kink effects on the share of college degree holders remain at similar magnitudes but with greater standard errors. On the other hand, the extent of the kink effect becomes more considerable. One explanation is that the share of college degree holders reaches its peak earlier than state employment. In addition, following the Southern Talks, those with state employment jobs continue to flow into the private marketplace, thus delaying the highest point for that particular value. As a result, the kink estimates become larger in absolute values. This explanation is consistent with the estimated coefficient at Panels C and D when the event's timing “moves” backward at one and two years, respectively. The kink effects on the share of college degree holders go up in absolute terms from the original 3.7 percent (column [3], Table 2) to 4 percent (Panel D, Table 6). The estimated coefficient on the share of state employment remains roughly the same at around 6 percent in absolute values. Combining this finding with the political context, one interpretation is that the cohort of 1992 entrepreneurs might not be limited to 1992. The overall picture could be broadly conceived as those talents leaving their bureaucratic careers between 1992 and 1994.

Table 6.

Further examination of the kink.

Notes: Standard errors are clustered at the cohort level. All regressions use interacted fixed effects. The kink effect refers to the estimated coefficient ρ ₂ in equation (1). An unreported constant is included in each specification. Columns (1)–(3) use the person's college degrees as a binary dependent variable. Columns (4)–(6) use the person's state employment as a binary dependent variable.

Taken as a whole, the regression estimates show that the timing of 1992 is an unusual impacting time point consolidating Deng's political vision. Not only did the paramount leader talk the talk, but also he made it true that his victory was institutionalized through formal party documents. Although the window of opportunity might have already opened before 1992, perhaps going back to 1985–1988, the flocking into the private economy with an energized entrepreneurial spirit could not have had such a substantial impact on China's political and economic landscape had Deng failed his epic maneuvers. That is why those individuals eventually become super wealthy by utilizing that window period of opportunity, confirming the concentration of entrepreneurial entries in Panel B of Figure 3. Viewed from the sample of CSRP, Deng's Southern Talks facilitated and consolidated the trends that had already been lurking in the late 80s had it not been abruptly interrupted by the Tiananmen tragedy. Accompanied by Chen Yun's death and the failure of the party conservatives to come up with a competing alternative, the ideological and political fight between the cage theory and the full-fledged market theory of socialism reached the end of its life cycle. Knowing that a policy reversal was unlikely, entrepreneurs and the private businesses understood that Deng's preferred version of market socialism was consolidated.