Will China Escape the Middle-income Trap? A Politico-economic Theory of Growth and State Capitalism

Transcription

1 Will China Escape the Middle-income Trap? A Politico-economic Theory of Growth and tate Capitalism Yikai Wang eptember 16, 2013 Abstract In this paper, we propose a theory of political-economic transition, to study growth and political development in China. We incorporate political constraints into a two-sector growth model to analyze endogenous dynamic interactions between economic and political systems. To sustain the oligarchy, the elite need to get political support from suciently many citizens. They use divide-and-rule strategy, which provides state workers with high wage and turns them into supporters. This policy results in distortions, i.e., wage premium and over-investment in favor of the state sector. In the shortrun, these distortions lead to rapid capital accumulation and economic growth. However, they will harm long-run growth. Our theory suggests that an oligarchy develops along a three-stage politicoeconomic transition: rapid growth, state capitalism, and the third stage with two possibilities - sustained growth or middle-income trap, depending on whether democratization occurs. The model's predictions are consistent with empirical facts in China, including investment-driven rapid growth, and persistent labor and capital market frictions. JEL: O41 P16 Department of Economics, University of Zurich, yikai.wang@econ.uzh.ch. I thank Daron Acemoglu, imeon Alder, Chongen Bai, Jimmy Chan, Jiahua Che, Heng Chen, Mikhail Golosov, Yasheng Huang, Xi Li, Michael Peters, Michelle Rendall, Dominic Rohner, Zheng ong, Kjetil toresletten, Wing uen, Jianrong Tian, Nico Voigtlaender, Yong Wang, Xiaodong Zhu, Fabrizio Zilibotti, and participants of Econometric society China meeting, HKU economics seminar, MIT macro lunch seminar, Third Annual Conference on Chinese Economy, University of Zurich PhD seminar, and Zurich Workshop of Economics for helpful comments. Financial support from wiss National cience Foundation (PBZHP ) is gratefully acknowledged. All errors are my own. 1

2 1 Introduction China has by now been growing at a stellar rate for over 3 decades. While this is generally acknowledged to be a great historical achievement, there is major controversy on how far in time and scope the Chinese success story can go. The optimists argue that China can provide a new model for growth as an alternative to the liberal democracy growth model Washington Consensus. For example, in a debate hosted by The Economist (see also Musacchio and Lazzarini (2012)), Aldo Musacchio argues that China's hybrid form of capitalism can become a new growth model for the 21st century. In his view, such a model oers three very attractive features: less pronounced recessions, focus on long-term investing and producing world champions. These considerations make him optimistic about the sustainability of China's future growth, and even about the possibility that China can become a role model for other developing and emerging countries. In contrast, critics predict that the growth rate will soon slow down. For example, Acemoglu and Robinson (2012) argue that China's extractive political institution is not compatible with innovation and high growth in the long run. In their view, although the growth process driven by catch-up, import of foreign technology, and export of low-end manufacturing products may continue for a while, it is deemed to come to a halt as soon as China reaches the living standards of a middle-income country. The pessimistic perspective of Acemoglu and Robinson raises a number of questions. If growth under the current regime slows down, as they predict, will this trigger changes in the political system? Will unsatised citizens oust the oligarchy and allow growth to be resumed under a more democratic system? Or, alternatively, will the oligarchy be able to retain sucient support even in a low-growth economy? On the one hand, modernization theory suggests that the rst scenario is likely to occur. But, then, one can argue that it may have been right for China to adopt its hybrid form of state capitalism to achieve high economic growth in the catch-up stage, and then switch to liberal democracy when state capitalism runs out of steam. The Chinese model, in other words, could be a model of transition, albeit not of long run growth for mature economies. On the other hand, this view may well be overly optimistic: at the time of transition, the political elite may be unwilling to give up state capitalism and stick to power in order to keep control on political power and economic resources, as we see in countries such as Kuwait and Venezuela. In the language of Acemoglu, Aghion and Zilibotti (2006), state capitalism 2

3 may be appropriate to promote growth at an early stage of development, but may become impossible to reform at a later stage when it becomes a burden on further economic development. Which of the two views on China's future growth is right? Under which circumstances will the elite choose to democratize or to maintain oligarchy at the cost of growth? To shed light on the discussion of the above questions, in this paper we propose a theory of the political-economic transition in oligarchy. We incorporate political constraints in oligarchy into a two-sector growth model to study endogenous dynamic interactions between economic and political institutions. The building blocks of the model are three-fold. First, it is a hybrid economy consisting of a competitive private sector and a state sector under the control of the political elite. The elite can distort resource allocation in favor of state rms, for their political and economic interest. econd, rapid economic growth driven by state investment is accompanied by high inequality, especially the articially low wage in the private sector. The low wage is the outcome of distortions, and it also contributes to fast capital accumulation and economic growth. Third, the oligarchic regime gets support from some citizens, including the middle class. Those middle class members consist of state sector workers who get high wage and secured jobs, and private entrepreneurs who benet from the articially low wage in the private sector. To be more specic, in our model, the interactions between the economic and political institutions are the following: to sustain oligarchy, the elite need political support from suciently large fraction of citizens. They adopt the divide-and-rule strategy and provide higher wage to state workers at the cost of private workers. The economic benet turn state workers into supporters of the oligarchic regime. To maintain high wage in the state sector, the elite over-invest in state rms to keep the capital-labor ratio high. In the short-run, the capital and labor market distortions lead to rapid growth, beneting from state investment and low private sector wage. This is the rst stage of the political-economic transition: rapid growth. However, the mid- and long-run eects on economic growth can be dierent. When the private sector grows larger, it squeezes the employment share of state sector and threats the supporter base of oligarchy. The elite start to restrict the private sector by limiting their access to the capital market. Meanwhile, a suciently large state sector is built up to match the private sector growth and keep enough workers in the state sector. We call this second stage state capitalism. In the long-run, if the elite are still willing to pay the cost of maintaining state capitalism and oligarchy, the distortions persist and the economy stops growing at a lower level than in democracy. The other possibility is that 3

4 the private sector capital growth makes the race between the state and the private sector too costly for the elite. Then they choose to democratize for their own economic interest. Distortions disappear and the economy grows again and converges to the level in democracy. This is the third stage. These two cases are called middle-income-trap and sustained growth, respectively. Our theory is related to three strands of literature. The rst one is on China's economic growth with resource re-allocation from the state sector to the private sector. ong et al. (2011) construct a twosector growth model to study how the capital and labor reallocation from the state to the private sector leads to economic growth. Brandt and Zhu (2000, 2010) document the contribution of private rms to growth and show that the government tries to maintain the state sector employment share. In these studies, the state sector employment share is either determined by the pure economic force or assumed to be exogenously set by the government. The government's political constraint introduced in our model provides the micro-foundation for the endogenous evolution of the state sector employment share, which gives a better match to the data and a richer prediction on the further growth trend. The second is the literature on the relation between the political and economic institutions. Acemoglu and Robinson (2012) focus how the political institution aects economic performance in the long run. The argue that the extractive political institution in a non-democratic country is detrimental to economic growth. On the other hand, Lipset (1959) and Fukuyama (1992) study how the long-term eect of economic development on the political development. Their modernization theory argues that the economic development will ultimately leads to political modernization, i.e., liberal democracy. Our contribution to them is that in addition to the eect of one on the other, we emphasize the inter-dependency of the economic institution and the political institution, to give a more complete picture both in the short- and the long-run. Third, our theory contributes to the study of middle-income trap, i.e., the signicant slowdown of economic growth when a country's GDP per capita reaches the middle level. ee Gill and Kharas (2007) and Eichengreen et al. (2013). Fatás and Mihov (2009) argue that this is because growth from low income to middle income doesn't require good institutions but only good policies, but good institutions are necessary for high income countries. Without improved institutions, rapid growing countries will hit the wall. This discussion is also heated in the public but there are in lack of theoretical frameworks to provide guideline for the discussion. With the model, we formally show how a country can grow within 4

5 the extractive institution but the growth stops at a middle level if there's no reform on the political institution. We also analyze in which circumstances a country can avoid the middle-income trap. The rest of the paper is organized as follows. ection 2 provides empirical facts about the features of the political-economic system in China. ection 3 uses a simple static model to illustrate the key interactions of the political and economic system in oligarchy. ection 4 presents the dynamic model that discusses political-economic transition in the short- and long-run. ection 5 concludes. 2 The Political-Economic ystem in China: Empirical Facts 2.1 The Political upport from tate Workers ince the reform of 1978, the strategy that Chinese government use to maintain political power has been changing. The once powerful ideology, propaganda, and repression have been gradually weakened, while the economic benets have been becoming more important and useful in buying support of citizens, as the country becomes more open and more practical. Nowadays, for many Chinese people, the legitimacy of the government and the oligarchic regime relies on sustained economic growth and improved living standards. However, in a rapid changing economy with increasing inequality, dierent economic groups benet dierently from the political-economic system. There are winners and losers, so there are also supporters of the existing oligarchy, and supporters of democratization. Who support the current political regime instead of democracy? Chen and Lu (2011) use a survey data of 2810 individuals, collected in three Chinese cities in late 2006 and 2007 to document which groups of people support democratic values and institutions 1. In their regression of support for democratic values and institutions on individual characteristics, state sector employment is negatively correlated with the support for democratic values. Its eect is even stronger than the party membership. In fact, employment in the state apparatus has the strongest eect, among all the control variables. In other words, state sector workers are more supportive for the current political system. 1 The data provide information on each individual's answers to questions about many aspects of democratic values, including right consciousness, valuation of political liberty, support for participatory norm and support for competitive election. The authors use factor analysis to nd the single dominant factor as the support for democratic values and institutions. 5

6 Ownership ector Wage Premium tate JF 0.6 Log Wage Differential Year Figure 1: tate sector wage premium (blue line) and JF rms wage premium (red line). ource: Ge and Yang (2012). 2.2 The tate ector Wage Premium Then why do state workers prefer the current oligarchic regime over democracy? The economic interests can largely shape their political preference. tate workers benet from the state support and enjoy more secured job, better working condition, and most importantly, higher wage. Ge and Yang (2012) use the Urban Household urvey to document that after controlling for worker characteristics such as age, education, sex and so on, the wage premium in the state sector over the domestic private sector has been more than 20% since the 90s. Figure 1 is reproduced from panel D of gure 1 in Ge and Yang (2012). The blue line shows the persistent labor market friction between the state and domestic private sectors. In contrast, as the red line shows, the wage dierence between joint-venture, stockholding, and foreign rms (JF) and domestic private rms has been decreasing dramatically, indicating that the over-all labor market friction has been declining. This implies that the persistent state sector wage premium can not be explained by the market incompleteness which has been decreasing. 6

7 1 0.9 tate ector s Investment and Employment hare Investment Employment hare Year Figure 2: tate sector's share of xed investment (blue line) and employment (red line). ource: Brandt and Zhu (2010), and tatistical Yearbook Misallocation of Capital Between the tate and the Private ector What keeps the state wage high? The answer lies in the misallocation of capital in favor of the state sector. Figure 2 shows that though the state sector's employment share has been declining since 1998, the state sector's xed investment share stay higher than 50%. This misallocation of capital in favor of the state sector results in high capital-labor ratio in the state sector (see Brandt and Zhu (2010); ong et al. (2011)), and high wage in the state sector. Also, the state capital return is signicantly lower than the private sector (5% v.s. 30% in 2007, see Brandt and Zhu (2010)). The misallocation of capital reduces the eciency, but is necessary for maintaining the state sector wage premium. The state sector's employment share is calculated as the number of urban OE workers over urban workers. We want to focus on the urban area throughout this paper, for two reasons. First, urban citizens are generally more politically mobilized and better organized for revolution than the rural citizens, so they are crucial for the elite to maintain oligarchy. econd, most data and gures available about the urban area, including those in subsection 2.1 and 2.2. The only data not available for the urban area is the investment share. Brandt and Zhu (2010) report the xed investment in state rms over the total 7

8 xed investment, in both urban and rural areas. Given that the in China xed investment in the urban area is almost all the total xed investment - 86% in 2007 and 95% in the state investment share in the urban area should be very close to the state investment share of the whole country. 2.4 low-down of tate Employment hare Decrease Figure 2 also shows that the urban state employment share declines after 1998, rapidly in the beginning but much more slowly later on, and it stays at around 20%. This suggests that signicant amount of citizens still rely on the state sector and the government. The process of privatization slows down and stops. 2.5 Linking the Empirical Facts to the Theory Putting the above four facts together, we propose a theory of political and economic transition in oligarchy. In our theory, the elite need political support from enough citizens to sustain oligarchy. They use divide-and-rule strategy and treat state and private workers dierently. tate workers are transformed into supporter of oligarchy (fact 1) using high state wage (fact 2). This high wage is maintained by distorting capital allocation in favor of the state sector (fact 3) to keep the capital-labor ratio high. To guarantee enough supporters, the elite strategically keep state sector alive and suciently large (fact 4). The key dierence of this theory from previous literature about China's economic transition, such as ong et al. (2011), is the political constraint. While a pure economic model usually assumes frictions to be exogenous and predicts that ineciency rms (OEs) gradually disappear, our theory with the political constraint can endogenize frictions in the labor market and the capital market as the choice of the elite, and explain the persistence of inecient state rms. 3 tatic Model In this section, we rst present a simple one-period model. The goal is to illustrate the political constraints that the oligarchic government faces and the immediate economic consequences. This static 8

9 model is also the cornerstone of our full dynamic model, which talks about the dynamic interactions between the political and economic institutions and the transition. 3.1 Preferences and Technology The model economy is populated by three classes of agents: elites, private entrepreneurs, and workers. The population of workers is normalized to measure 1, and the sizes of elites and private entrepreneurs are assumed to be relatively very small, i.e., of measure 0. Elites and private entrepreneurs run rms (to be dened below), and workers provide 1 unit of labor inelastically. In this static model, preferences of agents are simple: the higher income, the better. There are two sectors and two types of rms. tate () rms produce in the state () sector, while private (P) rms in the private (P) sector. There are innitely many of them. They use capital and labor to produce the same nal goods and maximize prots. They are dierent in two aspects. First, capital owners. rm capital are owned by elites (e) while P rms by private entrepreneurs (p). econd, the productivity. tate rms are less productive than private rms. Technology of and P rms are described by the following production functions, respectively: Y = (zk ) α L 1 α, Y P = KP α L 1 α P, where z < 1. In the static model, capital in and P sectors - K and K P - are predetermined, so there will be no decisions about investment or capital ow from rms to P rms. The labor is not predetermined, so it can be allocated in two sectors. 3.2 Political Equilibriums There are two types of political regimes: democracy and oligarchy. In democracy, the government is elected by majority vote. o workers, given their dominating population, determine the government policies. In oligarchy, elites control the government, but they still face a political constraint, i.e., support from a suciently large fraction of population. In oligarchy, rms owned by elites are called rms, because in the context of China, they are state rms. After democratization, they are still owned by elites but may not be called state rms anymore. However, they are still the same rms as they were 9

10 in democracy, in terms of technology and owners. o we still refer them as rms, hoping that this doesn't create any confusion. imilar for P rms. In both political regimes, the government can decide two policies: tax and wage regulation. First, the government chooses which groups of agents to tax, and the tax rate. We assume that tax payers can hide their income at the cost of τ fraction of their income. This implies that if the tax rate is lower than τ, tax payers choose to pay the tax. Otherwise they hide the income and pay no tax. This is a simple way to model the exogenous tax rate, as used in Acemoglu (2008) and referred as state capacity in Besley and Persson (2009). Then the tax income is shared among the ruling group - workers in democracy, or elites in oligarchy. We assume that the ruling group can not make costless direct transfer to the ruled. If they want to give benets to a sub-group of the ruled, they have to implement distortive policies, for example, resource misallocation and inecient projects in favor of that group. This phenomenon is discussed in literature, e.g., Acemoglu (2003) and Robinson and Torvik (2004). Acemoglu (2003) explains theoretically why transfer is not a credible tool for the ruling group to achieve Pareto optimal and satisfy the ruled class. o political Coase theorem doesn't apply and inecient policies emerge. Robinson and Torvik (2004) explicitly model how politicians build inecient white elephants projects to provide benet to certain group of people and in exchange, get their political support. There are also empirical evidences showing the extremely low eciency of transfer from the central government to citizens due to local capture. For example, Reinikka and vensson (2004) document that 87% of the transfer to schools in Uganda was not received during For this reason, we assume that the cost of transfer to the ruled is so high that the government does not use the transfer to buy political support from ruled. Though direct transfer is not available, the government can still use the second policy tool - wage regulation - to aect the income of the ruled. It can set the minimal wage in the sector. With this policy, the government can guarantee a high income level for workers if it wants. This policy distorts rms' employment decision. Also, it aects labor allocation and the income of and P workers, elites and entrepreneurs. We assume that the government can not set wage for P rms. In our theory, it is not useful for the government to set a binding private sector minimal wage, if it tries to gain support from state workers. In reality, the Chinese government has less control over private rms, compared to state rms. Private rms are in general free to set their wage and labor demand. Meanwhile, state rms are 10

11 also free to decide the labor demand to maximize prots, respecting the minimal wage constraint Democracy In democracy, the majority - workers - elect a representative worker into the government, then she decides government policies to maximize her income. The government policies in democracy are trivial. The government taxes elites and entrepreneurs, and it doesn't regulate the sector wage. The reasons are the following. First, the tax on workers is transferred back to them. o it is equivalent to no tax on them. econd, the government has no incentive to distort the wage in sector and the labor market allocation. etting sector wage lower than the P sector wage implies no workers join the state sector. etting it higher leads to inecient labor allocation, lower total output and less nal income for workers. o the government does not regulate wage in democracy. The labor market is competitive and ecient: and P rms compete for the labor. Given the above analysis, the timing of events is simplied to the following: 1. Capital available for and P rms are given as K, K P. 2. The government sets tax rate τ on elites and entrepreneurs' income. 3. and P rms hire workers in the competitive labor market and produce. and P rms produce. Capital incomes and wages are distributed. 4. Elites and entrepreneurs decide whether to hide income at the cost τ. 5. The government collects tax, and transfers it to workers. Now we can solve the equilibrium in democracy from backwards. Tax payers choose to pay the tax if τ τ, otherwise they hide the income 2. Competitive labor market implies that wage in and P rms are the same and equal to the marginal productivity of labor: w D = (1 α) (zk ) α ( L D ) α = (1 α) (KP ) α ( L D ) α P. 2 Without loss of generality, we assume that when tax payers are indierent between paying tax or hide the income, they pay the tax. 11

12 The market clearing condition L D + LD P = 1 helps us to pin down the labor allocation: L D = L D P = zk, zk + K P K P. zk + K P The pre-tax income for elites and entrepreneurs are the capital returns, respectively: π D = α (zk ) α ( L D ) 1 α, π D P = α (K P ) α ( L D P ) 1 α. Given the taxpayers' strategy on income hiding, the government decides τ = τ to get the maximum tax income, because in this static model without investment, tax is not distortive. The nal income of workers, elites and entrepreneurs are: ) yw D = w D + τ ( π D + πp D ( α = 1 + τ 1 α y D e = (1 τ) π D, y D p = (1 τ) π D P. ) w D, The transfer to workers is τ α 1 α wd simply because the tax base - capital income - is income. α 1 α times labor To sum up, in democracy, no distortive policy and competitive labor market imply rst best allocation. Elites and entrepreneurs get (1 τ) fraction of capital income, respectively. The nal income of workers is τ α 1 α wd times larger than their wage income Oligarchy In oligarchy, a representative elite controls the government to maximize her income. First, the government decides to democratize or not. If democratization is chosen, the economy goes to exactly the same as in democracy. Otherwise, the government makes two policies - tax rate and wage regulation in the state sector Given the policies, agents know their nal incomes in oligarchy. Comparing expected income in oligarchy and in democracy, an agent decides to support oligarchy or not. It is necessary for the oligarchy to maintain political support from large enough fraction of citizens, otherwise oligarchy collapses and democratization occurs. In the latter case, the economy is again the same as in democracy. 12

13 The minimal size of supporters is denoted as L, which is exogenous. Because the population of elites and entrepreneurs is measure 0, this condition is equivalent to that oligarchy is sustained if and only if L workers give political support 3. We call this minimal support constraint. Here we claim that the government taxes private entrepreneurs and private workers, but not elites and state workers. The logic is the following. Tax on elites goes back to themselves, so it is equivalent to no tax. Tax on state workers gives elites no benet but implies larger distortions. Because the government wants to increase state workers' after-tax income to a certain level to get their political support. It implies that the tax has to be compensated in other ways, which involves distortions. o for elites, there is no real benet to tax state workers. With this claim, timing of events in oligarchy is simplied and summarized below: 1. The government chooses to democratize or not. If yes, the economy is the same as in democracy. otherwise the following events occurs. 2. The government decides tax rate τ on entrepreneurs and private workers. 3. The government sets wage in sector as w. 4. rms and P rms hire workers. 5. Workers in and P sectors decide to support the current regime or democratization, simultaneously. (a) If more than L workers support the regime, oligarchy is sustained. (b) If less than L workers support the regime, democratization occurs. 6. If oligarchy is sustained, and P rms produce. Capital income, state and private sector wages are distributed accordingly. 7. Taxpayers decide whether to hide the income. 8. The government collects the tax and make the transfer to elites. 3 The above setting, that political support decides political regime, is based on the framework of Acemoglu et al. (2012). In their language, a coalition of some agents has certain level of political power, and if it is large enough, they can choose to change the political system. In oligarchy, elites as the ruling group are granted some level of political power, denoted as ω e. Each worker has political power ω w, and each entrepreneur has ω p. The aggregate political power of entrepreneurs, whose size is 0 is just 0. Workers can change the political regime from oligarchy to democracy if and only if they form a coalition of size L r and ωwlr ω w+ω e there must be at least 1 α ωw+ωe, where α is exogenous. In other words, the sustain a oligarchy, > α L r > α ωw+ωe ω w ω w workers supporting it, and we denote this size as L. 13

14 Given the same logic as in democracy, we know that the government sets the tax rate at the level of the state capacity, τ. Taxpayers pay the tax. o state and private workers' nal incomes are y w = w, y wp = (1 τ) w P. Given that oligarchy is our focus, we save the superscript O of variables in oligarchy to simplify the notation in the rest of the paper. Now we can solve the rest of the problem from backwards. We rst look at workers' political decisions. No matter what others choose, supporting oligarchy if and only if the income is higher than in democracy is the weakly dominant strategy of a worker. Without loss of generality, we assume that workers use the above strategy. In other words, a worker's political choice truthfully reects her economic benets, and and P workers support oligarchy if and only if y w y D w, y wp y D w, respectively. In the second step, we analyze rms' labor demand in oligarchy. rms, given the regulated wage w, choose labor demand L so that wage equals marginal productivity. imilar for private rms who face market wage w P. w = (1 α) (zk ) α L α, (1) w P = (1 α) K α P L α P. Third, the government sets the minimal wage w in sector. If the minimal wage is lower than the market wage with ecient labor allocation, i.e., w D, it's not binding and equivalent to setting minimal wage to w D, otherwise it is binding. o without loss of generality, we claim that w w D and the state sector wage equals the minimal wage. Facing too high wage, rms demand less labor than in democracy. More labor is left to the P sector and the wage there is lower than in democracy. Private workers always get less income than in democracy, y wp < w P w D and they never support oligarchy. o the only hope for the government to maintain enough support is to set w yw D so that state workers support oligarchy. We call this high state wage constraint. However, there is also an important tradeo 14

15 that restricts the government from setting too high state wage, i.e., the higher wage in sector, the less labor rms demand. Given the constraint of minimal number of supporters, the government needs to balance between the high wage and high state employment. Because equation 1 gives a one-to-one mapping from w to L, we can safely think setting w as setting L. This is convenient because the minimal support constraint is also about L. uciently high wage becomes suciently low employment in sector. tate sector jobs are under-supplied therefore precious. The highest level of employment in sector, given the high state wage constraint, is pinned down in the following: We denote L = w = (1 α) (zk ) α L α ( yw D α = 1 + τ 1 α ( α L 1 + τ 1 α ) 1 α L D. ) (1 α) (zk ) α ( L D ) α ( ) 1 ) 1 + τ α α 1 1 α L D (1 and ν = + τ α α 1 α. We can see that ν < 1 if τ > 0. This implies that there must be less sector labor than in democracy to guarantee this high wage. To sum up, we get the following two lemmas. The rst is the tradeo of high sector wage. Lemma 1. The tradeo of high sector wage for elites. Large enough w buys workers' political support. But a too high w means too low L, which may leads to less than enough supporters for oligarchy. The second is the workers' political choices. Lemma 2. Workers' political support. Workers in sector support oligarchy if income in sector is higher than in democracy, which is equivalent to small enough employment in sector : L L = νl D, where ν < 1. P sector workers always get lower income than in democracy and support democracy. If the government decides to sustain oligarchy instead of democratization, it maximizes the elites' income subject to the high wage and high state employment constraints: y e = max α (zk ) α L 1 α L + τ (K P ) α (1 L ) 1 α s.t. L L L. 15

16 The political constraints give an area of L [ L, L ] that the oligarchy can be sustained. If L L, this area is non-empty, otherwise no L satises the constraint and oligarchy can not be sustained. L is an exogenous parameter, determined by political power of workers and elites. If citizens are well-organized and have relatively high political power, elites need to buy o many workers to sustain oligarchy. If most citizens are not politically mobilized, the government can stay in power with a minority of supporters. In the latter case, L can be low. L = νl D in democracy. Fixing other variables, if L D = is endogenously determined by labor allocation and tax rate zk zk +K P is large enough, L can be larger than L. In other words, sustaining oligarchy requires sector strong enough, equipped with enough capital. The logic can be shown in the following gure. Then when does the government chooses to sustain oligarchy and which L does it set? First, we assume that τ α. This low tax condition means that elites get a large share - capital share α - from sector than P sector - tax share τ. This assumption guarantees that the government doesn't want to reduce sector employment to be lower than in democracy. o given the political constraints, the government still wants to minimize the distortion. In reality, this condition is easy to satisfy, considering that α is set to 0.5 or 0.66 in the literature and τ is always lower. Given this, the solution to the government's problem is summarized in the following proposition. Proposition 1. ustainability of oligarchy. If there is suciently high capital stock in sector, zk K P L ν L, oligarchy can be sustained. sector labor is set at the largest level that satises the high wage constraint. Elite income is higher in oligarchy than in democracy, so the government always choose to sustain oligarchy. If sector capital is small, zk K P < L ν L, democratization occurs. We use gure 3 to show the intuition for the above proposition. The blue line and red line are the marginal productivity in and P sectors, depending on sector labor. The intersection pins down the labor allocation in democracy, as well as the wage and income. Given worker income in democracy y D w, the oligarchic government set the sector wage higher than y D w. The high wages means low sector employment, lower than L. If L L, any L [ L, L ] sustains oligarchy, and the optimal level for elites is L = L because it leads to the minimal distortion, given the low tax condition satised. Larger sector capital pushes the curve of the state sector marginal labor productivity up, then L D and L move to the right. When zk K P L ν L, L L. 16

17 Figure 3: Labor allocation and marginal productivities. The nal income of elites is always higher in oligarchy. For elites, the cost in oligarchy is that they pay extra wage to each state sector worker (area 2 in gure 3), which is to compensate the tax that the elites anyway have to pay in democracy (area 2 and 3 and 4). o the cost in oligarchy is lower than in democracy. The benet in oligarchy, i.e. tax income from the P sector makes elite income higher in oligarchy. o the government choose to sustain oligarchy if zk K P L ν L. If zk K P < L ν L, the government can choose to democratize voluntarily in the rst step, or choose to sustain oligarchy but fail in getting enough support. In the later case, revolution leads to democracy. ince the outcomes are the same, democracy, everyone is indierent between voluntary or involuntary democratization. Not only elites, but also entrepreneurs get higher income in oligarchy, if tax rate in oligarchy is not higher than in democracy. The minimal wage in sector pushes abundant labor into the P sector and creates articially low wage there. o entrepreneurs get higher capital income. L P > L D P, w p < w D y p > y D p. 17

18 3.3 Discussions and Implications The divide and rule strategy, presented in this simple static model, can explain three empirical facts in section 2 - more political support, higher wage and higher capital labor ratio in the state sector. Furthermore, this model gives the following implications linking to the real world questions. First, fast capital accumulation and growth in the short-run. The income of capital owners is higher. Elites' political power is transformed to economic power, i.e. tax benet, and they get higher income in oligarchy. Entrepreneurs also indirectly benet from the policy that reduces private sector wage. We can't evaluate this with a simple static model, but intuitively, for example in a olow model with constant saving rate, capital owners can save more, and this high investment drives fast growth. We will formally show this with a dynamic model in the next section and evaluation the long-run growth trend in oligarchy, which can be dierent from the short-run. Moreover, income inequality is higher in oligarchy and the short-run growth benets from the inequality. The rich elites and entrepreneurs get even richer than in democracy, also workers are divided into two groups with unequal income. The inequality, provides abundant cheap labor in the private sector, increases the capital return and allows potentially faster capital accumulation and economic growth. But how long can the growth keep beneting from cheap labor? We leave the answer to the dynamic model. Third, the middle class are not necessarily pro-democracy, on the contrary of some traditional thinking based on modernization theory. As we see in the model, the middle income groups - private entrepreneurs and state workers - benet from the oligarchic regime, at least in the short-run, so they lend political support to the state. Especially lower middle class - state workers. They are created by and rely on the state for the privileged jobs. This helps us to understand in some emerging markets why economic growth and the burgeoning bourgeoisie (The Economist (2009)) do not push for democratization. For example, Tsai (2007) documents that there is no coherent middle class in China and the majority of private entrepreneurs accept the current political system. The middle class beneting and supporting the state are also very common in other countries. The Economist (2009) documents that 95% of adult Kuwaitis work for the government, usually in white-collar civil-service jobs - middle class jobs, and the emirate's private-sector middle class consists almost entirely of foreigners. The wage gap between the state and private sector is also large there. These distortions keep politically important local workers 18

19 in the state sector and is a smart way to maintain oligarchy. More examples are the anti-democracy middle class - the Yellow hirts- in Thailand and the growing state middle class linked with growing inequality in 1960's Brazil. Again, in the long-run, the status may be dierent. Though it is likely that state workers always support the state, the private entrepreneurs may not love the distortions forever. To sum up, this static model shows the interactions of the political and economic systems in oligarchy and leaves more interesting questions about the long-run. The interaction is the following: on the one hand, the political interests, largely shape the economic outcome. Taking into account political constraints, we can understand economic puzzles. On the other hand, economic power determines political outcome. When the state sector is economically powerful with enough capital, elites can keep a large enough power base and sustain oligarchy, otherwise they can not provide enough economic benets to enough political supporters. 4 The Dynamic Model We have seen that the relative size of state and private capital is crucial for the divide and rule strategy. Only with abundant state capital, can the government buy enough political support. o the oligarchic government is motivated to control the capital formation and allocation. In the following, we build a dynamic model to study the government's and entrepreneurs' investments in the state and private sectors, respectively, and also the government's policies in the nancial market, for example, allowing or restricting private rms' access to external nancing. In a dynamic world, the long-run economic growth and economic benets, e.g., of entrepreneurs, can be dierent from the short-run. 4.1 Environment In China, state owned enterprises (OEs) are well-connected to state banks and backed up by the state, so it's pretty easy for them to get loans from state banks. In contrast, the private owned enterprises (POEs) have limited ability in getting bank loans, partly due to the lack of connection with state banks, partly due to the government policies. In the model, we assume that the government can freely set the rm capital, while a private rm faces borrowing constraint, as in ong et al. (2011). To be more 19

20 specic, the upper bound that a bank is willing to lend to an entrepreneur is η fraction of her own asset 4. This is private rms' natural nancial constraint given the market structure. Additionally, in oligarchy, the government can restrict the borrowing limit of entrepreneurs to a even lower level η [ η, η ]. The government can create barriers in loans to POEs, or directly give administrative instructions to banks (see Brandt and Zhu (2000)) 5. In democracy, which is an ecient market, the government can but choose not to restrict the loans to the more ecient private sector, for the interest of workers. It is an open economy, banks can borrow and lend in the international bond market at the interest rate r. The nancial market is competitive, so the interest rates for loans to state and private rms are both r. We denote the gross return R = 1 + r. Besides the setting on the capital market, there is a second dierence compared to the static model, i.e., elites and entrepreneurs live for innite periods and are forward-looking. For simplicity, we still assume that each generations of workers live for only one period, so that they simply consume their nal income in every period. 6 Third, the income sources of elites and entrepreneurs, in additional to capital income share in the static model, now include after-depreciation capital and interest earning/expense. This may make a dierence for the tax. But to keep the notations and expressions simple and similar to the static model, we assume that the government can only tax the capital income share of elites and entrepreneurs, but not the two new sources. Modifying this assumption doesn't change the model qualitatively. To sum up, the timing of events within each period is very similar to the static model. There are only two dierences, in the beginning and the end of each period. In the beginning, the state and private capital is no longer predetermined. The government can set state capital K and private borrowing limit η [ η, η ]. P rm capital is constrained by K P (1 + η) a p, where a p is entrepreneurs' assets. In 4 This assumption comes from the following logic, similar to ong et al. (2011). Banks face moral hazard problem when lending to a private borrower with asset s. he could be an entrepreneur who has access to production technology and gets potentially higher capital return than return on bank savings R P > R or simply someone without entrepreneurship who can only save all the money in other banks and get return R. If it is the latter, the borrower can chooses to steal and leave. In this case, the bank can still take back ηr (l + s), where l is the loan. The borrower gets (1 η) R (l + s). To guarantee that the borrower won't steal and leave, the incentive constraint is (1 η) R (l + s) R (l + s) Rl l η s. 1 η We denote η = η. 5 1 η η is the lower bound of the borrowing limit that the government can set. For example, it can be 0 if the most the government can do is to order banks not to lend to private rms, but the private rms can nance their investment using entrepreneurs' asset. 6 Assuming forward-looking workers complicate the calculation a lot, but it should not add or change too much, especially when the economy is closer to the steady state. In the steady state, one period income is proportional to life-time income, so forward-looking workers and myopic workers are exactly the same. 20

21 the end, each elite and entrepreneur decides the next period asset a e and a p, respectively. Finally, we assume the logarithm form for the instantaneous utility. 4.2 Democracy In democracy, workers control the government to maximize their income. Due to the same reasons in the static model, workers do not want to distort the economy. The Laissez-faire policy achieves the rst best in each period. Both the capital and labor markets are ecient. The more productive P sector get the maximal external nancing η = η. The labor allocation is such as the wages are equal in both sectors. o the outcome in every period is exactly the same as the static model: ecient labor allocation, tax from elites and entrepreneurs and transfer to workers. The dynamics is also simple, because each elite member or entrepreneur takes the return to the asset as given. The optimal choice of a private entrepreneur is as follows: rst, she maximizes the current period income. he borrows as much as possible if P sector capital return is strictly higher than the global interest rate r, otherwise she puts some asset in the bank to get the return r. Then, given the income from the asset, she saves a constant fraction for the next period, due to the log utility. Assuming agents discount the future by the discount factor β, we have the evolution of the entrepreneur asset, a p = βy p (a p ). imilarly, each elite member chooses to put the asset in rms or in the bank, whichever gives higher return, then saves a constant fraction of income for the next period. Basically, the solution to democracy is similar to ong et al. (2011). The equilibrium is summarized in the following proposition. Proposition 2. In democracy, elites get return on their asset at interest rate r, and entrepreneur asset yields return greater or equal to r. If β is large enough, entrepreneurs accumulate more and more asset over time, the relative size of sector over P sector, measured by k = zk K P, decreases over time to 0. The intuition for the results is the following: Ecient allocation of labor implies the same wage in and P sectors. rms are less productive and they exist when entrepreneur asset is small and P sector is small. Because of competition among rms, the capital return is equal to the nancing cost, i.e. r. P rms are more productive than rms, so entrepreneurs get higher return on their asset. If β is large enough, entrepreneur asset and P rm capital increase over time, and nally P rms hire all workers and rms all exit. The market force is decisive in such a model without political constraints. 21

22 4.3 Oligarchy In oligarchy, the representative elite controls the government and makes four policies - minimal wage in sector, tax, nancial constraint on P rms and sector investment - to maximize her life-time utility. To simplify the analysis, we rst pin down those tax rates that are very obvious. The same logic in the static model applies to tax rates on elites, workers and P workers. The rst two are zero because the government maximizes elite income and wants to guarantee certain income for workers to get their political support. The last is the maximum since it is not distortive. However, tax rate on private entrepreneurs may be dierent from the static model. It is not necessarily the maximum. In the dynamic model, tax on entrepreneurs distorts their future investment, which may harm elites' interest by reducing future tax income. o τ p can not be pinned down in this stage. Given the tax rate, workers' political decisions are exactly the same as in democracy. If a worker's nal income is higher than in democracy, she chooses to support oligarchy. The minimal wage policy in sector increase sector wage but always reduces P sector wage. o P sector workers always choose to support democratization. The political constraints for sustaining the oligarchy are: high state wage constraint w yw D and minimal support constraint L L. Given K, K P, wage in P sector w P and labor allocation L, L P are determined in the same way as in the static model. o given K, K P, w, the equilibrium is almost the same as in the static model. If oligarchy is sustained, expressions for wages w, w P, nal income of workers y w, y wp, and income share of capital π, π P are the same as before, therefore they are omitted here. The nal income of elites and entrepreneurs have dierent expressions, because now there are depreciation at the rate δ and interest earning/expense at the rate r. y e = π + (1 δ) K R (K a e ) + τw P L P + τ p π P, y p = (1 τ p ) π P + (1 δ) K P R (K P a p ). a e contributes to y e only in the form of Ra e, but a p contributes to y p more than that, because K is not restricted by a e, but K P is bounded by (1 + η) a p. o we can write elite income net of interest return on asset as ŷ e = y e Ra e and ŷ e doesn't depend on a e. But we can't claim that for y p. K is decided by the government, then how is K P determined? Given the government policies K, η, w, τ p, 22