Beyond state and market: the role of labor regimes and class dynamics in the East Asian miracle

While conventional wisdom focuses on the roles of the state and the market in achieving high growth in East Asian countries, this paper offers an account focused on class dynamics. First, the paper connects post-World War II class dynamics in Japan and South Korea to the distributional outcome, i.e., the gap between labor productivity growth and real wage growth. Statistical evidence and econometric tests suggest a positive relationship between a balance of power in favor of workers and a higher growth rate in output. Secondly, the paper puts forward a labor regime framework for explaining the connection between the class dynamics and the distributional outcome, and finds it to have played a central role by using historical data and an autoregressive distributed lag (ADL) model. We further analyze China’s economic growth and conclude by stressing the importance of class dynamics for better understanding growth and distribution in capitalist development.


Introduction
The post-WWII rapid economic growth of East Asian countries is an astonishing phenomenon in the history of world development.This growth occurred first in Japan, beginning in the mid-1950s.It was followed by the rapidly industrializing economies of Hong Kong, Singapore, South Korea, and Taiwan (the "Four Asian Tigers") in the 1960s.Later, beginning in the 1970s, Indonesia, Malaysia, the Philippines, and Thailand (the ASEAN Four) also experienced swift economic growth.Similarly, since the 1980s, China has experienced remarkable growth, becoming a global powerhouse in the past two decades.There can be little doubt that the growth of the economies of East Asia remains unparalleled among developing countries (Arrighi et al. 2003).
Focusing on the three largest economies in East Asia -Japan, South Korea, and China -this paper intends to map the link between economic growth and class dynamics during their industrialization process.Table 1 demonstrates that Japan led the rapid growth in the post-1960 era, followed later by South Korea and eventually by China, in terms of both GDP and GDP per capita.These achievements were largely driven by rapid growth in labor productivity.As many studies in the development literature point out, labor productivity growth in this region was primarily the result of rapid industrialization, in particular, the deliberate and sustained upgrading of manufacturing (Stiglitz 1996;Storm and Naastepad 2005).
TABLE 1 ▪ GDP and per capita GDP growth rates in three East Asian economies   (annual average, %) 1961-1970 1971-1980 1981-1990 1991-2000 2001 The conventional wisdom used to explain the superior performance of East Asian economies focuses on the roles of the market and of the state.According to the market-oriented view, these economies were able to achieve high growth because market forces were released and factor prices were set around the market-clearing level (Balassa and Williamson 1987;Krueger 1990;World Bank 1993).At the same time, the state-interventionist view attributes their growth performance to the active role of state in setting prices "wrong" and orienting business in late development (Amsden 1992(Amsden , 2001;;Fishlow et al. 1996;Johnson 1982Johnson , 1995;;Wade 1990;Walter and Zhang, 2012).Despite the important differences between these competing arguments, they both dismiss the fundamental class underpinnings upon which market and state forces operate to generate and share productivity gains.Within the tremendous discussion of East Asian growth and the crisis of the late-1990s, there is still a noticeable lack of integrative theoretical and empirical work on the underlying capital-labor dynamics of East Asian capitalism (Hart-Landsberg and Burkett 1998;Hart-Landsberg et al. 2007).
In this essay, we pursue an alternative, labor-centered analysis of East Asian economic growth, examining shifting class relations as a key determinant.According to the social structure of accumulation (SSA) theory, capitalist economies experience periods of relatively rapid and stable growth once a set of socioeconomic institutions have been established.Among these institutions is a set of capital-labor relations.As time passes, capital-labor relations evolve along with other economic conditions.They together generate stress and eventually begin to erode the SSA, leading to a structural crisis, which ends the period of vigorous and stable accumulation.In its wake, a new set of capital-labor relations emerges, along with other endogenously generated and external factors, which determine the growth prospect.This process entails an institutional restructuring necessary to establish a new accumulation regime (Gordon et al. 1982;Bowles et al. 1986;Kotz et al. 1994;McDonough et al. 2010;Kotz, et al. 2019). 1 An example can be found in the U.S. economy: the two decades following WWII saw the formation of a class compromise and an economic boom, both of which eroded over time and were eventually replaced with a neoliberal institutional structure (McDonough et al. 2010).As Boyer, Uemura, and Isogai (2011) pointed out, institutional restructuring and transformations are also crucial for East Asian economies when we consider their evolution, diversity, and interdependence.
Concerned with the central importance of class dynamics, we first examine the empirical relationship between class relations and economic growth in the East Asian economy.From a political economy perspective, when the working class becomes increasingly concentrated and more powerful in the bargaining process, it tends to demand a larger share from productivity gains and is more likely to force capital to make concessions.If this class explanation has merit, we would expect that the consolidation of a capital-labor compromise leads to distribution in favor of workers and the demise of it leads to distribution in favor of capital.We would also expect the rapid growth periods to be associated with a class compromise, and that the slow growth periods are brought on by the demise of it.To test these expectations, in Section 2 we share Boyer et al's (2011) emphasis on a mix of historical, comparative, and statistical analyses, and present statistical and historical evidence to reveal the evolving class dynamics in East Asia.Then, in Section 3, we explain the reasoning behind these 1.A similar theory of long-run capital accumulation has been developed, since the end of the 1970s, by the Regulation School, which holds that capitalism has experienced a sequence of "regimes of accumulation", monitored by an adequate regulation mode (Aglietta 1979).The patterns of evolution also involve institutional configuration and reconfiguration, including the nature of labor organization (Lipietz 1986;Boyer 1987;etc.).The detailed differences of the regulation school and the social accumulation of accumulation theory (Kotz 1990), however, are beyond the scope of this paper.This paper adopts the shared approach by those two theories in conducting a mixed method of historical, comparative, and statistical analyses of the East Asian economies.
expectations and empirically test them by examining the link between output growth and distributional outcome in Japan since the mid-1950s and South Korea since the mid-1960s.
The more challenging task is to locate the class dynamics over the past fifty years in the historically specific settings of labor relations.To this end, in Section 4, we propose a labor regime framework that highlights the importance of labor process, labor distribution, and labor protection and reproduction to account for the distributional outcome.Then we empirically test the crucial role of the labor regimes by developing operational measures and estimating an econometric model using the autoregressive distributed lag (ADL) approach.This allows us to empirically assess the explanatory power of labor regime variables in accounting for the struggle over distribution while not excluding the strong influence of state intervention.The central argument will be that the labor regime framework is highly powerful in explaining the distribution over productivity gains and in signaling the class dynamics in these societies.Distribution is not only relevant in itself, but also has direct impact on the growth performance.Section 5 provides suggestive analysis for understanding China's economic growth since the 1980s.Section 6 concludes.

Class dynamics in the postwar East Asian economy
Following the postwar demobilization and subsequent land reforms of the late-1940s, an enormous number of people in East Asia began to be employed for economic recovery. 2Industrialization first featured labor-intensive manufacturing of products such as textiles, and later extended to the manufacture of chemical and machine products.The changing economic structure inevitably led to a changing class composition and dynamics in the East Asian societies.For instance, in Japan, the proportion of wage workers was still only 39.3 percent in 1950 but increased to 64.2 percent in 20 years (Japan Statistic Bureau, n.d.).In South Korea, the number of wage workers doubled from 1963 to 1975 and increased fourfold in the next ten years (South Korea Statistical Information Service, n.d.).Because capital-labor relations rose to be the core social relations of production in these societies, their analysis becomes crucial for understanding class dynamics.
In this paper, we develop an analysis of class dynamics in Japan and South Korea by examining capital-labor relations.Class contestation and compromise can be abstracted and conceptualized in a variety of ways (Wright 2000).
2. Marx defined relative surplus population in the rural agricultural areas as taking a latent form, and that in urban domestic or small industry, with conditions of life below the average level of the working class, as taking a stagnant form (Marx 1967).
We utilize here an alternative measure to capture the distributional outcome of this dynamics, represented by a growth gap: that is, the difference between labor productivity growth and real wage growth.Labor productivity is defined as the ratio of output (real GDP) to labor input (total working hours), and real wage is defined as total nominal employee compensation per working hour compared to the consumer price index (CPI).In fact, the aggregate data of these two measures have been widely used since the 1930s and 1940s as a key element of national income accounting and social policy setting (Block and Burns 1986).If real wage growth lags behind labor productivity growth (a positive gap), capital is gaining more than labor from rising productivity.On the other hand, if real wage keeps pace with or even rises faster than labor productivity (a zero or negative gap), a concession or compromise has likely been made between capital and labor.Therefore, a comparison of growth in labor productivity and real wage can help us in tracking the underlying class dynamics.
Figure 1 graphs the behavior of our key measure of capital-labor class dynamics -the growth gap between labor productivity and real wage -on an annual basis for Japan from 1956 to 2014 and for South Korea from 1966 to 2014.The series in both countries display short-run cyclical fluctuations, but they also reveal a long-term pattern.The gap plunged in the 1970s, representing the power of the working class to defend their interests, even amidst high inflation resulting from the oil crisis.However, in Japan real wage growth began to lag behind labor productivity growth in the late 1970s.Similarly, in South Korea, the gap began to shrink in the early 1980s, though it remained negative until the Asian financial crisis of 1997.Particularly after the financial crisis, most years saw labor productivity grow more rapidly than real wage.
We believe the changes in the movement of the growth gap can be attributed to the early consolidation and the late demise of a class compromise in postwar East Asia.The dynamic class relation has strong implications for the economic takeoff and slowdown there.Burawoy andWright (1990, 2000) point out that, as long as conflicts are contained within capitalism itself, coercion and consent both exist in the asymmetric power relations; therefore, we should expect shifts in the balance of power between them to generate different distributional consequences.This complex dynamic has been evident in the capital-labor relations associated with East Asian development.The economies of East Asia did not emerge or transform in a vacuum but were nurtured in an evolving constellation of social and historical forces.If we want to truly understand economic growth, we must examine class dynamics within this historical perspective.This salient feature is largely missing in dominant accounts of a free-market approach or a state-intervention approach, and has been increasingly called upon recently to decipher industrialization and development (for example, H. Song 2019; Ziai 2019).Japan 1956-2014and South Korea 1966-2014 Sources: Data from Japan Statistics Yearbooks and South Korea Statistical Database, various years.Notes: When the growth gap is below zero, the power of workers is increasing, as they acquire a larger share of productivity gains.LPG indicates labor productivity growth, while RWG indicates real wage growth..2 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 year Japan, 1956-2014South Korea, 1966-2014year 19651970197519801985199019952000200520102015 Real Wage Growth Labor Productivity Growth the Growth Gap = LPG -RWG

Real Wage Growth
Labor Productivity Growth the Growth Gap = LPG -RWG When examining the historical contours of capital-labor relations, a threephase development from formation to consolidation to subsequent demise of a class compromise in Japan and South Korea after WWII can be found.
Phase 1 runs from the end of WWII to the mid-1950s for Japan, and to the early-1960s for South Korea.In the tremendous upheavals of the immediate postwar era, labor militancy and state repression became the dominant themes.Japan had a nationwide struggle over production control supported by Japan's Communist Party in 1945-1947 and a remarkable increase in the unionization rate from zero in 1945 to nearly 60 percent in 1949.South Korea's popular political movement -which included occupation of factories by workers organized by strong left-wing unions -was characterized by massive confrontation with the police and the U.S. occupation forces.In both countries, labor movements not only challenged the economic interests of the capitalist class, but also threatened the whole system (Hamilton 1983;Moore 1985Moore , 1997;;Sugimoto 1978).As the cold war intensified, the governments of Japan and South Korea launched several rounds of political attack on labor militancy and developed a highly repressive labor regime.Japan's planned general strike of 1947 was suppressed.Public servants' right to strike was eliminated in 1948, and the "red purge" took place in 1949.The conservative All-Japan Confederation of Labor was established in 1964 and confined itself to the private sector.In South Korea, the 1947 railroad strike was crushed ruthlessly by the state, and right-wing unions were established after 1948 to replace the communist leadership in the labor movement.After the establishment of the Park military government in 1961, union political activity was banned, and labor unions were reorganized into the conservative Federation of Korean Trade Unions (FKTU) and were only allowed to function at the company level.At the same time, individual capitalists held a direct alliance with the state in order to get access to U.S. aid, raw materials, and other necessities for production.The significant labor militancy seen in both countries was politically terminated and the labor movement was dictated by the state with an almost exclusive focus on economic demands.The wage labor population saw a marked increase, especially among the female population (Koo 2001;You 1998).With the help of state repression and a relatively loose labor market, industrial capitalists were able to impose strict labor control on the shop floor.As a result, big labor disputes continued ceaselessly over the demands for improvements in labor conditions, and over opposition to "rationalization" and dismissals.Capitalist firms and managers could easily be ruined once a long labor struggle occurred (Itoh 1990).
Phase 2 stretches from the mid-1950s to the mid-1970s for Japan and from the mid-1960s to the early-1990s for South Korea.Class compromise was achieved in Japan and South Korea; neither labor nor capital was strong enough to definitively vanquish the opponent.The labor market gradually began to tighten with less surplus labor available to draw upon.A waged, urbanized, organizationally concentrated proletariat came to the fore.Along with that, the control and discipline of industrial labor was accomplished through the corporate control of labor union activities.In Japan, the defeat of the militant strikes that occurred from 1950-1960 was followed by a longterm dominance of the Liberal Democratic Party and a cooperative "economism" among trade unions.This economism offered workers lifetime employment and regular wage increases within a seniority system.The "spring wage offensive" [Shuntō], as a collective bargaining mechanism, was established in 1955 to guarantee an annual rise in basic wage rates.In South Korea, running in tandem with the transformation from labor-intensive to capital-intensive industrialization, male manufacturing workers were largely incorporated into enterprise unions, which prioritized industrial peace and cooperation.Employers sought to enhance worker loyalty through paternalistic arrangements, such as providing lifetime employment and various company-sponsored welfare schemes (Amsden 1992).More than half of all large-scale firms used the seniority wage system (Korea Employers' Association).In both Japan and South Korea, large business conglomerates -Keiretsu in Japan and Chaebol in Korea -dominated the economic activities, and the state employed tactics of division, co-optation, and repression of the working class, yet the end result was a wage structure that reflected the needs of workers over productivity gains (Ito and Kang 1989).In both Japan's cooperative regime (Lazonick 1995) and South Korea's authoritarian regime (You 1995), real wages had an even-higher growth compared to the already rapid growth in labor productivity (see Table 2).As a rising share of economic growth was delivered to the working class, both countries experienced remarkable increases in living standards.In the case of South Korea, the better economic position of workers helped intensify citizens' struggle against the military dictatorship for more political freedom in the process of democratization.Fairly volatile throughout the 1970s, the Korean labor movement in the 1980s became more threatening to the government: labor protests were frequently organized by independent, grass-roots labor unions with linkages to other popular activism.
Phase 3 begins in the late 1970s/early 1980s for Japan and in the mid-1990s for South Korea.In phase 3, neoliberal transformation has led to the demise of the class compromise with a capitalist assault over labor.Japan's high growth ended with the inflationary crisis of the 1970s and went through two unstable recoveries in 1976-1979 and 1983-1985.Privatization of state-owned enterprises was conducted in the name of solving the state's deepening fiscal crisis, but the end result was the diminishment of the public sector's once-powerful trade unions.For example, following railway privatization, the number of workers was cut by half, and membership in the Japan National Railway (JNR) trade union declined by 80 percent.With declining membership, and under pressure from the country's continuous recession, the bargaining power of Japanese trade unions was significantly weakened.Meanwhile, the introduction of new automation systems into the workplace made the experience and skills of regular male workers increasingly unnecessary.This led to further fracturing of the labor force and eroded the Japanese style of labor management (Itoh 2000).In South Korea, the neoliberal doctrine of privatization, liberalization and deregulation started to proliferate in the mid-1980s, and accelerated under pressure from the International Monetary Fund (IMF) after 1997 (Kim et al. 1994;Crotty and Lee 2002).In both countries, we see a pattern in which capital becomes more mobile while labor becomes more atomized and vulnerable.In response to increasing global competition, capital begins to adopt systems of informal and irregular employment -a situation which the state allows, and sometimes even facilitates with legislative changes.This results in a significant rise in the numbers of part-time, temporary and contract workers 3 (Mari and Kotosaka 2012).In Japan, irregular employees accounted for 15 percent of total employees in 1984, but rose to nearly 40 percent by 2016 (Miyamoto 2016).In South Korea, despite the implementation of the Minimum Wage Act in 1988, there has been a massive increase in non-compliance (C. Lee 2012).The practice of hiring irregular workers was even allowed in the public sector after 1997.By 2002, 51.6 per cent of South Korea's total workforce was comprised of irregular workers (South Korea Statistical Information Service, n.d.), many of whom suffer from low wages, lack of social insurance and low job security (Song 2012).Among these 3. Irregular employment existed in both Japan and South Korea at an earlier time; however, both countries had significant regulatory changes in the 1990s to remove or relax many restrictions on the use of temporary and fixed-term contract workers (J.Song 2008).irregular workers, more than 65 percent are women (South Korea Statistical Information Service, n.d.).Mainstream labor unions have not effectively mobilized the fragmented workforce for collective interests.Thus, the post-crisis era has been characterized by the dominance of capital over labor.As a result, real wage growth has most often lagged behind labor productivity growth, as presented in Table 2.This historical transition from formation to consolidation to demise of a class compromise in Japan and South Korea has had a profound impact on the economic performance of these two countries.To illustrate this, we conduct an empirical test, described in the next section.

The growth impact of class compromise in East Asia
There are several theoretical arguments concerned with the positive macroeconomic dynamics associated with high wage growth (Lavoie and Stockhammer 2013).First, from the supply side, higher wages tend to induce labor-saving technology and thus lead to high productivity growth.Second, from the demand side, higher wages are likely to create higher aggregate demand, thus stimulating more investment and growth.Alice Amsden (1990), when discussing South Korea's development strategy, also points out that higher wages work as a reward to workers who exercise their intelligence in the workplace to facilitate technology transfer.
Combining the growth performance and the key indicators of distribution in these economies, as summarized in Table 2, we find that real wages rose faster than labor productivity during their rapid growth periods (Japan 1956(Japan -1975;;South Korea 1966-1997), and then the trend reversed when growth slowed (Japan 1976(Japan -2014;;South Korea 1998-2014).At the same time, rapid real wage growth is associated with rapid productivity growth and high output growth.In sum, the regime that provides a higher share of productivity gains to workers tends to be associated with greater economic growth.
The relationship between a negative growth gap (real wage grows faster than labor productivity) and high economic growth in Table 2 can also be seen in the scatter plots below. 44.This is, however, not unique in the economic history of other countries.Both the U.S. and Europe experienced higher wage growth and higher economic growth in the post-WWII regulated era, and lower wage growth and lower economic growth in the neoliberal period starting from the late 1970s/early 1980s.Gavin Wright (2006) assembles a number of lines of historical evidence to suggest that higher wages created sufficient wage pressure to channel technological possibilities into productivity-boosting application in the 1920s and 1990s, while the productivity slowdown in the 1970s was at least in part a result of the deceleration of wage growth.The main results still hold when the growth gap is replaced with a lagged value to capture the adjustment process.A more detailed explanation of the forces behind the growth gap requires further unpacking the dynamics behind the power (im)balance between capital and labor.

Labor regimes in East Asia
We put forward a labor regime framework for explaining the connection between distributional outcome and class dynamics reflected by the relative strength of capital and labor, and test it against both historical and empirical data.

A labor regime framework
Frederic C. Deyo (2012) proposes a labor regime framework to analyze key labor components of a social transformation, in which social reproduction and protection, labor allocation and labor process can be differentiated and articulated to identify the internal tensions and dynamics in the transformation 5 (Figure 3).A labor regime comprises those variably institutionalized social processes and activities through which potential labor is mobilized and transformed into actual labor, useful services and products and -in capitalist economies -profits.The labor regime, whether imposed or agreed upon, represents the interplay between buyers and sellers of labor, under the formal or informal regulation of the state.Therefore, it is embedded in the broader regimes of accumulation that define, constrain, and contextualize the labor components and, together with other institutional arrangements, serves the core foundation of economic growth.
5. This is not to deny possible overlap and interpenetration among these phases.Labor reproduction and protection entails an investment in the labor force to maintain its necessary level of availability and productivity while keeping it from the risk of income loss and market uncertainties.Regarding the economies of East Asia, the relevant literature has emphasized two critical phenomena: the first is the size of the state's welfare program, particularly its contraction during the transition from phase 1 to phase 3.This contraction leads to an externalizing of the cost of labor reproduction and protection with responsibility moving from employers and states to families and communities.The second phenomenon is the productionist orientation of East Asian welfare systems, which tend to promote labor productivity over labor protection.Governments generally provided early and strong support for the expansion of primary education, 6 but took a minimalist approach to the provision of social insurance with respect to the extent of public commitment, the nature of financing, and the breadth and depth of coverage (Aspalter 2001;Haggard and Kaufman 2008;Holliday 2000).
The labor market works to mobilize and allocate labor force.In the process of structural change, effective labor supply is most likely from rural-urban migrants and other non-capitalist sectors, such as households.Therefore, the share of the economically active population absorbed in wage employment reflects the condition of the labor market, especially considering the gender division.Moreover, the difference between the economically active population and employment corresponds approximately to the concept of a reserve army of labor, which is regarded as the specifically capitalist mechanism to keep wages from rising beyond the limits conducive to the profitability of capital (Marx 1867).In more recent decades, the labor market has experienced massive reforms in Asia and the rest of the world with an increasing share of temporary and irregular work force, leading to new interaction with the traditional segmentation of the labor market.
The labor process is the domain of practices of managerial efforts to coordinate, motivate and control workers at the site of production and to capture the economic surplus they produce.Employers and employees contest working hours, choice of technology, the organization of work, and the allocation of profits.It is in this arena that production becomes a contested ground and the politics of production dominates the managerial and labor agendas, and actual work effort becomes the focus of labor productivity (Burawoy 1983(Burawoy , 1985)).Bowles and Gintis (1990) suggest that power relations between classes is the critical factor explaining labor productivity, by proposing a "labor extraction model".The bargaining power of workers depends not only on the fallback positions in a broader market setting, but also on their 6.One influential argument in the literature on the region's economic growth is to highlight the role of education (for example, World Bank 1993).
relative ability to handle industrial conflicts in a collective manner.Thus, labor organization and collective action are both shaping and shaped by the practices in the labor process and in the broad industrial relations.
In a capitalist economy, this power relation is expressed directly through a market mode of economic integration.Thus, the proportion of national product set aside for wages and profits (which includes rents and interest) is determined by the outcome of class contestation and compromise between workers and capitalists.We can measure labor productivity as the goods and services that an average hour's labor provides to employers, while employee compensation is what the workers are given back.While more and better machines, better education, and harder and faster labor efforts raise productivity, whether the labor share of increased productivity also grows depends upon the power-induced distribution between capital and labor.

Modeling and econometric analysis
Is this labor regime framework able to shed light on the distributive dynamics underlying the rapid economic growth in East Asia?To what extent can the transformation of labor regimes over time account for the power struggle between capital and labor reflected by the growth gap?To answer these questions, we test the empirical model below with variables constructed based on the three labor regime components discussed in the previous section.
Drawing insights from Bowles et al's (1986) model on capitalist power and social regimes of accumulation, we conduct a time series analysis of the distributional outcome of class dynamics, as measured by the gap between labor productivity growth and real wage growth.The following long-run relationship is established to empirically test the relationship between the growth gap and the labor regimes: In Equation (1), GAP is the dependent variable -the difference between labor productivity growth and real wage growth.A higher gap reflects a higher share of benefits from the growth in labor productivity being allocated to capital.LM is the labor market variable.Two measures are used separately.One is the absorption rate, measured by the ratio between employment and the working-age population.It captures the wage pressure from the market, which in the Marxian term represents the relative size of the reserve army of labor.A higher labor absorption rate corresponds to a relatively smaller reserve army of labor and thus greater wage pressure on capital (the expected sign with respect to GAP is negative).The other is the share of regular workers in total employment, which captures labor market dualism and institu- tional changes in terms of deregulation and liberalization.A higher regular worker share corresponds to higher power of the working class, thus a smaller or even negative growth gap is expected (the expected sign is negative).IR is the variable for industrial relations corresponding to the labor process.Depending on data availability, we have three specifications: average weekly hours of all work force (HOURS), as a proxy for the labor intensity in the workplace; union density (UNION) as a proxy for the organizational power of the working class, and the number of labor-management disputes (DIS-PUTES).To avoid the problem of multicollinearity, we use union density and number of labor disputes separately selected according to data availability.The expected signs for HOURS, UNION and DISPUTES are positive, negative and negative, respectively.WS is a proxy for the cost of labor reproduction and protection.It is measured by the ratio of social welfare expenditure on medical care, pensions, worker compensation insurance, etc., to national income or GDP.A higher ratio reflects a higher investment in labor reproduction and protection, thus signaling a relatively stronger worker position at the national level.Therefore, the expected sign is negative.α 0 is a constant.μ is the error term.
As Equation (1) shows, the three aspects of labor regime -i.e., labor market, industrial relations and labor reproduction and protection -are all considered here.All data are from official statistics of Japan and South Korea.
To avoid spurious results, a formal stationary test is performed before proceeding to estimation.Table 3 summarizes the results of the augmented Dickey Fuller test for both economies.The test reveals that the dependent variable of the growth gap is stationary in levels, and all other variables are nonstationary in levels but stationary in first differences, i.e., GAP is I(0) and all other variables are I(1).Given that the variables are a mix of I(0) and I(1), the autoregressive distributed lag (ADL) approach to cointegration developed by Pesaran et al. (2001)when it is not known with certainty whether the underlying regressors are trend-or first-difference stationary.The proposed tests are based on standard F-and t-statistics used to test the significance of the lagged levels of the variables on a first-difference regression.The asymptotic distributions of these statistics are non-standard under the null hypothesis that there exists no level relationship between the dependent variable and the included regressors, irrespective of whether the regressors are I(0 is appropriate.This widely-used approach in recent macroeconomic literature adopts the bounds test to examine whether a cointegration relation exists among variables that are either stationary or integrated with order one I (1).
According to the ADL approach, equation (1) should incorporate the short-run adjustment process: In equation ( 2), ∆ means first difference, p and q i are lag lengths, β i is a coefficient, and v t is the error term.The bounds test uses the OLS result of Equation ( 2) to test the hypothesis β 5 = β 6 =β 7 =β 8 = 0 by comparing the F-statistic with the lower bound and the upper bound of critical values.If the F-statistic is greater than the upper bound, then there is a long-run relationship between levels of the variables; if the F-statistic is smaller than the lower bound, then there is no long-run relationship between levels of the variables; if the F-statistic falls between the two bounds, then the result is inconclusive (Pesaran et al. 2001).
Lag lengths of choice (p, q 1 , q 2 , q 3 ), based on the Schwartz-Bayesian information criterion (SBIC), are chosen as (1, 1, 0, 2) when IR is specified as union density in the case of Japan, and (1, 1, 1, 2) when IR is specified as number of labor-management disputes in the case of South Korea.A key assumption Notes: *** and ** denote rejection of the null hypothesis of a unit root at 1% significance level and 5% significance level, respectively.GAP is the dependent variable -the difference between labor productivity growth and real wage growth.LM is the labor market variable (absorption rate or the share of regular workers in the total employment).IR is the variable for industrial relations.Depending on data availability it can be (1) average weekly hours of all work force (HOURS), (2) union density (UNION or (3), the number of labor-management disputes (DISPUTES).WS is a proxy for the cost of labor reproduction and protection.It is measured by the ratio of social welfare expenditure to national income or GDP.
of the bounds test is that the residuals of Equation ( 2) must be serially independent.With these lag lengths, there is no serial correlation in the residuals. 7 The results of the bounds test in Table 4 support the idea that there is a longrun relationship between levels of the variables, no matter which IR specification is used.The F-statistic is greater than the upper bound at the 1 percent level of significance.The t-statistic of the error-correction term is also greater than the upper bound at the 1 percent level of significance.Notes: ** and *** denote 5% and 1% significance levels, respectively.The null hypothesis is there is no cointegration.Critical values are from the appendix of Narayan (2005).
Table 5 gives the ADL regression results for both countries.All the coefficients of the explanatory variables are statistically significant and their signs are as expected, except the variable for average weekly hours (the coefficient 7. The p value of the Breusch-Godfrey LM test is 0.0041. of which is not statistically different from zero).Both the labor absorption rate and the welfare share have negative effects on the gains of capital.Union density or number of labor-management disputes, reflecting labor's organizational power in collective actions or workplace conflict, also have negative effects.The unit root test shows that the residual of the regression is stationary at the 5 percent level of significance, which confirms that there is cointegration among the variables.Notes: ** and *** denote 5% and 1% level of significance, respectively.The t-statistics are in parentheses.The lag lengths of regressors are selected according to Schwarz-Bayesian information criterion (SBIC).According to the ADF test, the residual of the regression is stationary.According to the Breusch-Godfrey LM test, there is no serial correlation.According to the Breusch-Pagan test, there is no heteroskedasticity.is the dependent variable -the difference between labor productivity growth and real wage growth.is the labor market variable (absorption rate or the share of regular workers in the total employment).is the variable for industrial relations.Depending on data availability it can be (1) average weekly hours of all work force (HOURS), (2) union density ( or (3), the number of labor-management disputes (DISPUTES). is a proxy for the cost of labor reproduction and protection.It is measured by the ratio of social welfare expenditure to national income or GDP.
After the ADL regression, the following error correction model is further estimated: Equation (3) replaces the level variables in Equation ( 2) with the error correction term EC t-1 lagged one period.The error correction term represents the long-run relationship.The coefficient δ 5 in Equation (3) should be negative, which means that ∆GAP t will respond to the deviation from the long-run relationship by offsetting the deviation.Thus, a negative δ 5 will indicate a longrun relationship between levels of the variables.Table 6 gives the regression results of Equation ( 3), which indicates that δ 5 is negative and its absolute value is smaller than one.It also gives the long-run effect of each factor.The long-run effects of labor market improvement, strengthening of organizational power, and increases in welfare share are negative on the growth gap both in Japan and South Korea.When the labor regime becomes less friendly to workers from all three aspects, the gains of productivity growth to workers decline.Notes: ** and *** denote 5% and 1% level of significance, respectively.The t-statistics are in parentheses.The lag lengths of regressors are selected according to SBIC.According to the ADF test, the residual of the regression is stationary.According to the Breusch-Godfrey LM test, there is no serial correlation.According to the Breusch-Pagan test, there is no heteroskedasticity.is the dependent variable -the difference between labor productivity growth and real wage growth.is the labor market variable (absorption rate or the share of regular workers in the total employment).is the variable for industrial relations.Depending on data availability it can be (1) average weekly hours of all work force (HOURS), (2) union density ( or (3), the number of labor-management disputes (DISPUTES). is a proxy for the cost of labor reproduction and protection.It is measured by the ratio of social welfare expenditure to national income or GDP.
The evidence in the regression analysis shows clearly that the distributional outcome can be well explained by the changes taking place in the labor market, in the workplace, and at the level of national welfare policy.We conclude that labor regimes appear to have had an important impact on the class dynamics in East Asia and that the results of this econometric exploration are fully consistent with the historical and institutional analysis of the trajectory of that movement presented in Section 2.

Implications for understanding China's growth
Beginning in the 1980s and accelerating in the 1990s, China's market reforms turned it into a late-coming Asian miracle.Researches have shown that the Chinese growth experience is, to a large extent, a replication in extreme form of an earlier East Asian model (Chan and Unger 1995;Goodman et al. 1998;Hung 2009Hung , 2013)).
A closer look at the distributional outcome reflected by the employee compensation share of GDP in China reveals a similar trajectory to the case of Japan and South Korea. 8Figure 4 reveals that the early 1980s marked an improved distribution in favor of workers. 9While the share declined after the mid-1980s, it remained high and picked up again in the early 1990s.This trend signals a possible class compromise achieved both by the deprivation of political rights of workers and the adoption of material incentive schemes.For many, the decade of 1980s witnessed an improvement without losses (Lau et al. 2000;Naughton 2006;Rawski 1994).However, beginning with the intensification of market reform in the early 1990s, and extending through the massive privatization of state-owned enterprises (SOEs) in 1995, the class compromise eroded with distribution increasingly disfavoring workers.Labor has been increasingly subjected to the assault of neoliberal reform perpetuated by pervasive state and capital collusion (for a detailed analysis of class struggles within China since the 2010s, see Chan and Hui 2016).
8. Given the lack of available data pertaining to working hours in China, we present the employment compensation share of GDP in Figure 4 rather than the growth gap between labor productivity and real wage.9.A more detailed data explanation can be found in Qi (2015).Employee Compensation Share of GDP Though a detailed examination of historical trends in China is beyond the scope this paper, the labor regime framework allows us to understand the possible explanations from the proposed three aspects.In terms of labor process, Li and Qi (2014) suggest that the wide adoption of computer numerical control (CNC) plays a key role in weakening the working class by creating lowskill, low-wage work.Additionally, long/uneven working hours and more labor discipline on the shop floor lead to rising discomfort and stress among workers.At the same time, corporate trade unions are ineffective at wage bargaining, and the government has so far done little to correct the frequent violation of existing labor laws (Friedman and C. K. Lee 2010).In terms of the labor market, rural areas have supplied a large reserve army to industrial development.In many industries, a tiered employment system is the norm, featuring full-time contract workers alongside dispatch, temporary, or "intern" workers who receive markedly lower wages and benefits (C.K. Lee 2012).This rise in the informal economy and demise of the socialist employment guarantee increasingly erodes job security and the power of the working class (Piovani 2014).Neoliberal reforms have also replaced traditional social policies with more market-friendly policies, such as pensions (1997), health (1998), unemployment (1999) and work injury insurance (2004).Despite the more recent expansion of coverage, the overall welfare program has a strong productionist orientation (Choi 2012).
What does differ China from the experience of Japan and South Korea is that the Chinese GDP has been able to keep growing rapidly as labor's share declines.This is largely due to the high and growing dependence on exports since the late 1980s, as well as on both exports and investment since the early 1990s (Zhu and Kotz 2011).However, the weak position of workers reflected by the distribution against them and the declining share of domestic consumption in GDP has imposed great threats to the sustainability of China's growth pattern.

Conclusion
An examination of East Asian economies cannot be complete without an analysis of the evolving power dynamics between workers, the capitalist class, and the state (Deyo 2012).This paper proposes a labor regime framework and empirically tests its explanatory power to complement our understanding of East Asia's rapid growth and slowdown.
Three general conclusions can be drawn here.First, over the past fifty years, East Asian economies went through a historical process of formation, consolidation, and demise in class compromise.This evolution reflected the class dynamics among capital, labor, and state when the economy, in Japan and South Korea, grew rapidly and later suffered from economic crisis and stagnation.Second, the experience of East Asian economies reflects the high growth potential of a class compromise in capitalist development.Once the class compromise began to erode, however, the high growth more likely became unsustainable.Third, the labor regime framework integrating labor market, labor process, and labor reproduction and protection, sheds much light on the distributional outcome in East Asian growth.Labor regimes, as one of the key social institutions in a capitalist economy, are not the product of disinterested design.Rather, they are outcomes of economic and political struggles among various interested parties with unequal degrees of power in the industrialization process.
Labor regimes and class dynamics in East Asia are not the only force determining the growth performance or the development trajectory in these economies, nor would we want to make such a claim.In fact, ample studies have shown the importance of other historical differences, such as colonial experience, public schooling, etc., in tracing the growth discrepancies across East Asian economies (see, for example, Harada and Tohyama 2011;Boyer et al. 2011;Go and Park 2019).Therefore, this paper is limited by its focus on one particular institution -the labor regimes and class dynamics in industrialization.Nevertheless, the neglect or exclusion of them creates an incomplete or even misleading picture of the rapid East Asian economic development.
To better interpret and reinterpret the history of this development, experience can offer lessons for us to understand the contradictions of our own times.Central to capitalist development processes, labor regimes and class dynamics deserve more academic and political attention than they have received thus far.Future research may well expand to focus more deeply on the labor-centered economic history of each individual country in this region, interrogate the social dynamics behind the tendencies of convergence or divergence in its overall economic development, and explore general lessons for labor struggles and social transformations in today's whole world.

Más allá del estado y el mercado: el papel de los regímenes laborales y las dinámicas de clase en el milagro de Asia Oriental
Resumen: Mientras el conocimiento convencional explica el alto crecimiento a través de los roles del estado y el mercado, este artículo plantea una hipótesis centrada en el trabajo y en los determinantes de clase sobre los que operan las fuerzas del mercado y el estado para generar y distribuir las ganancias de la productividad.La primera parte del artículo conecta las dinámicas de clase posteriores a la Segunda Guerra Mundial en Japón y Corea del Sur con la distribución de la riqueza, medida como el diferencial entre el crecimiento de la productividad laboral y el crecimiento de los salarios reales.La evidencia estadística y los ejercicios econométricos sugieren una relación positiva entre el equilibrio de poder a favor de los trabajadores y una más elevada tasa de crecimiento del producto.En la segunda parte, el artículo propone un marco de régimen laboral para explicar la conexión entre las dinámicas de clase y la distribución de la riqueza, realizando un test de esta relación con los datos históricos y empíricos y una aproximación temporal mediante Autoregressive Distributed Lag (ADL).Además, analizamos el crecimiento económico de China y concluimos remarcando la importan-

FIGURE 2 ▪
FIGURE 2 ▪ Growth and gap in Japan and South Korea

FIGURE
FIGURE 3 ▪ A labor regime framework

TABLE 2 ▪
Real wage growth, labor productivity growth and GDP growth (compound  annual growth rate, %) Source: Japanese Statistics Bureau and Korean Statistical Information Service.

TABLE 3 ▪
Unit root tests

TABLE 5 ▪
ADL regression results

TABLE 6 ▪
Error-correction representation results