American Political Science Review Vol. 100, No. 1 February 2006

Economic Backwardness in Political Perspective DARON ACEMOGLU Massachusetts Institute of Technology JAMES A. ROBINSON Harvard University e construct a simple model where political elites may block technological and institutional development, because of a “political replacement effect.” Innovations often erode elites’ in- W cumbency advantage, increasing the likelihood that they will be replaced. Fearing replacement, political elites are unwilling to initiate change and may even block economic development. We show that the relationship between blocking and political competition is nonmonotonic: elites are unlikely to block development when there is a high degree of political competition or when they are highly entrenched. It is only when political competition is limited and also when their power is threatened that elites will block development. Blocking is also more likely when political stakes are higher, for example, because of land rents enjoyed by the elites. External threats, on the other hand, may reduce the incentives to block.

overnment policies and institutions shape eco- also marked differences: while Britain and the U.S. nomic incentives, and via this channel, have a adopted new technologies and industrialized rapidly, Gfirst-order impact on economic development. Russia, Austria-Hungary, and Spain lagged behind. Why, then, do many societies adopt policies Why did these countries fail to adopt new technologies that discourage investment and maintain institutions that would have increased their incomes? that cause economic backwardness? Perhaps, politi- These differences in performance motivated cally powerful groups (elites) are not in favor of eco- Gerschenkron’s famous essay, Economic Backward- nomic growth. But why? It would appear that eco- ness in Historical Perspective (1962), which focused on nomic growth would provide more resources for these how relatively backward economies lacking the eco- groups to take over or tax, increasing their economic nomic prerequisites for industrialization could com- returns. So why don’t powerful groups always support pensate in different ways. However, in later work economic development? Gerschenkron recognized that the desire to promote In this paper, we develop a theory of inefficient the institutions necessary for industrialization varied government policies and institutions. All else equal, considerably across countries. Indeed, in the coun- politically powerful groups would welcome superior tries that lagged the most, rather than actively pro- institutions and technologies. But in practice all else is moting industrialization, political elites opposed it. not equal, because superior institutions and technolo- Gerschenkron argued that in the case of Austria- gies may reduce their political power and make it more Hungary, not only did the state fail to promote indus- likely that they will be replaced. At the center of our trialization but also theory is therefore the idea that changes in institutions or the introduction of new technologies often create ...economic progress began to be viewed with great suspi- turbulence, eroding the political advantages and future cion and the railroads came to be regarded, not as welcome economic rents of incumbent elites. Alternatively, new carriers of goods and persons, but as carriers of the dreaded technologies may enrich competing groups, increasing revolution. Then the State clearly became an obstacle to their threat to incumbents. These considerations make the economic development of the country. (1970, 89) politically powerful groups fear losing power and op- pose economic and political change, even when such So the problem of understanding why industrializa- change will benefit society as a whole. tion was rapid in some countries, whereas in others it To understand the mechanism at work and its poten- did not get off the ground, is closely related to under- tial applications, consider a concrete example: indus- standing why in some countries the state encouraged trialization in the nineteenth century. Bairoch (1982) industrialization, whereas in others it did not. More ex- estimates that between 1830 and 1913, world manufac- plicitly: why did the state and the political elites in some turing output increased by a factor of 5 (see Table 1 societies not only fail to encourage industrialization, Panel A). Nevertheless, this process was highly uneven but also go even as far as blocking the introduction of across regions and countries. Bairoch also calculates new technologies and economic institutions necessary that over the same period manufacturing output in for industrialization, such as the production of well- developed countries (Europe and North America) in- functioning factor markets, property rights, and legal creased by a factor of over 10, whereas it declined in the systems? Third World. Among developed countries, there were Our answer emphasizes the political replacement ef- fect: political elites will block beneficial economic and institutional change when they are afraid that these Daron Acemoglu is Professor, Department of Economics, changes will destabilize the existing system and make Massachusetts Institute of Technology, 50 Memorial Drive, it more likely that they will lose political power and Cambridge MA 02142 ([email protected]). James A. Robinson is Professor, Department of Government, future rents. More specifically, everything else equal, Harvard University, 1737 Cambridge Street, Cambridge MA 02138 political elites are less likely to be replaced when ([email protected]). they adopt technologies and institutional changes that

115 Economic Backwardness in Political Perspective February 2006

masses. For example, in Russia, the Tzar and political TABLE 1 elites were initially strongly opposed to industrializa- Panel A: Per-Capita Levels of Industrialization a tion, or even to the introduction of railways. When (UK 1900 is 100) industrialization in Russia finally got underway after 1750 1800 1860 1913 the Crimean War, the fears of the elites were confirmed: Austria-Hungary 7 7 11 32 France 9 9 20 59 industrialization brought social turbulence in urban Germany 8 8 15 85 centers, and political and social change, culminating Italy 8 8 10 26 in the 1905 Revolution. This is the idea underlying our Russia 6 6 8 20 political replacement effect. Even though the political Spain 7 7 11 22 elites in Russia may have preferred industrialization if United Kingdom 10 16 64 115 they could be sure of maintaining power and taxing the United States 4 9 21 126 proceeds, in practice they did oppose it because they Japan 7 7 7 20 were afraid of losing their political power. China 8 6 4 3 In addition to proposing a mechanism for why coun- Panel B: Per-Capita GDP Levelsb tries may fail to adopt superior technologies and in- 1820 1870 1900 1913 stitutions, our framework also gives a number of com- Austria 28416376 parative static results that provide interpretations of France 27406275 the historical evidence. The impact of political com- Germany25436886 petition on blocking is nonmonotonic. Both political Italy 24 32 38 55 elites that are subject to competition and those that are Russia 16232733 highly entrenched are likely to adopt new technologies. Spain 23284449 With intense political competition, elites prefer to United Kingdom 38 71 100 110 innovate, because otherwise they are likely to be United States 28 53 89 115 replaced. With a high degree of entrenchment, in- Japan 16 16 25 29 cumbents are willing to innovate, because they are China 12121415 a Source: Bairoch 1982. not afraid of losing political power. Instead, it is b Source: Maddison 1995. Maddison’s data covers only elites that are somewhat entrenched but still fear Austria. Because Austria and Vienna were by far the richest replacement that will block innovation. This non- parts of the Empire, these numbers severely overstate the pros- monotonicity result provides an interesting interpre- perity of Austria-Hungary. tation of the cross-country differences in industrial- ization. New technologies were rapidly adopted in Britain and subsequently Germany where the political increase output (e.g., because they will be more pop- elites—–the landed aristocracy—–were sufficiently en- ular). Moreover, such changes will also benefit elites trenched. In contrast, in Russia and Austria-Hungary, by increasing future output and their revenues. So all where the monarchy and the aristocracy controlled else equal, the elites prefer technological change. All the political system, but feared replacement, they else is not equal, however, because such change may were firmly against industrialization. Instead, they erode their political advantages relative to other groups continued to rely on the existing system of produc- that are benefiting from the changes or weaken their tion, including the feudal relations between lords and ability to control political challenges. As a result, in serfs. certain circumstances, institutional and technological Our analysis also shows that economic change is change will increase the likelihood that the elites will more likely to be blocked when there are greater lose power, creating the political replacement effect. rents to political elites from staying in power. This sug- This effect introduces a tradeoff for elites between the gests that another factor contributing to stagnation in likelihood of maintaining power and rents conditional Russia and Austria-Hungary may have been the sub- on maintaining power, and may induce them to block stantial rents obtained by the landed aristocracy from change. The presence of the political replacement ef- the more feudal labor relations in the agricultural sec- fect implies that a Coase Theorem type of logic, main- tor of these countries. Rents were also influenced by taining that investments that increase the size of the the political institutions. At the dawn of the nine- social pie will always be carried out, does not apply. teenth century, both Russia and Austria-Hungary were There is no (credible) way of compensating ex post the ruled by absolute monarchies who were unconstrained political elites who lose their power (Acemoglu 2003). by representative political institutions. In Britain, po- In the context of nineteenth-century industrializa- litical institutions were very different. The absolute tion, our model suggests that the elites, the monar- monarchy had been to a large extent emasculated chy and landowning interests, opposed industrializa- by the Glorious Revolution of 1688 and had lost tion and the necessary institutional changes, precisely many prerogatives. Therefore, an important determi- because these changes were likely to erode their po- nant of attitudes toward change will be the preexist- litical power. In fact, in most cases, the rise of mar- ing political institutions: when these institutions limit kets and industrialization have been associated with political rents, elites will be more favorable toward a shift of political power away from traditional rulers change. and landowners toward industrial and commercial in- This reasoning on the role of rents also suggests terests, and ultimately to popular interests and the that differences in the level of human capital may be

116 American Political Science Review Vol. 100, No. 1 important in shaping the elites’ attitudes toward THE BASIC MODEL industrialization; because human capital is comple- mentary to industrial activity, a high level of hu- We now discuss a simple model to illustrate the politi- man capital makes future gains from industrializa- cal mechanism preventing the introduction of superior tion larger relative to the rents from preserving the technologies and institutions. existing system, thus discouraging blocking by the elites. The Environment Finally, we show that external threats often make in- cumbents more pro-innovation, because falling behind Consider an infinite horizon economy in discrete time technologically makes counties vulnerable to foreign consisting of a group of citizens, with mass normal- invasion. This insight may explain why the Russian de- ized to 1, an incumbent ruler, and an infinite stream feat in the Crimean war or the American blockade of of potential new rulers. All agents are infinitely lived, Japan changed political elites’ attitudes toward indus- maximize the net present discounted value of their trialization and modernization. income and discount the future with discount factor, Our paper is most closely related to the literature on β. Whereas citizens are infinitely lived, an incumbent interest group politics, where existing powerful interest ruler may be replaced by a new ruler, and from then groups may block the introduction of new technologies on receives no utility. in order to protect their economic rents. In the context There is only one good in this economy, and each of development economics, this idea was first discussed agent produces: by Kuznets (1968), developed at length by Olson (1982) y = A , and Mokyr (1990), and formalized by Krusell and Rıos-´ t t Rull (1996) and Parente and Prescott (1999). Although where At is the state of “technology” available to the the idea that monopolists, or other interest groups, may citizens at time t. At should be thought of as technology block technical change at first appears similar to the broadly construed, so that it also captures the nature of thesis in this paper, it is fundamentally different. We economic institutions critical to production. For exam- argue that what is important is not economic rents that ple, a change in the enforcement of property rights such will be destroyed by the introduction of new technolo- as the creation of new legal institutions, or the removal gies, but the erosion in the political power of the elites. of regulations that prevent productive activities, or any After all, if the groups that have the political power kind of political and economic reform that encourages to block change were to maintain their political power investment would correspond to an increase in At.In after the change is implemented, why wouldn’t they light of this, we use the terms “innovate” and “adopt be able to use the same power to redistribute some of new technologies and institutions” interchangeably. the gains to themselves? This reasoning suggests that When a new technology is introduced or there is ben- whether certain groups will lose economically or not eficial institutional change, A increases to αA, where is not as essential to their attitudes towards change α>1. The cost of adopting the new technology or as whether their political power will be eroded.This initiating institutional change is normalized to 0. In view is consistent with the fact that British landed addition, if there is political change and the incumbent aristocracy, which maintained its political power, sup- ruler is replaced, this also affects the output potential ported industrialization despite its adverse effects on of the economy as captured by A. In particular, when land values. Issues of “blocking” also arise in the litera- the incumbent does not adopt a new technology, the ture on economic reform, for example, Fernandez´ and “cost of political change”—–that is, the cost of replacing Rodrik (1991), Alesina and Drazen (1991), or Hellman the incumbent—–is zA, whereas this cost is zA when he (1998) though with nothing analogous to the analysis introduces the new technology. Notice that this “cost” here. can be negative; it may be less costly to replace the Finally, this paper is closely related to our previ- incumbent ruler than keep him in place, for example, ous work, Acemoglu and Robinson (2000a,b), and to because he is “incompetent.” Bourguignon and Verdier (2000) and Robinson (1997). Therefore, more formally: In particular, Acemoglu and Robinson (2000a) sug- = − + − + + −
gested the idea that the greater impediment to eco- At At−1((1 pt)(1 (α 1)xt)) pt(1 (α 1)xt nomic development was not groups whose economic − x z − (1 − x ) z)), (1) interests were adversely affected by economic change, t t but those whose political power were threatened. This where xt = 1 or 0 denotes whether the new technology paper formalizes this idea by introducing the polit- is introduced (xt = 1) or not (xt = 0) at time t by the ical replacement effect in the context of a forward- incumbent ruler, whereas
xt = 1 or 0 refers to the in- looking dynamic political economy model, and obtains novation decision of a new ruler. Also, pt = 1 denotes the novel result that the relationship between polit- that the incumbent is replaced, while pt = 0 applies ical competition and the desire of political elites’ to when the incumbent is kept in place. (Notice that if the block innovation can be nonmonotonic (see Bueno incumbent is replaced, what matters is the innovation de Mesquita and Smith 2004 for a related nonmono- decision of the newcomer. So even when the incumbent tonicity). It also uses this framework to interpret cross- has chosen xt = 1, if the newcomer chooses
xt = 0, the country differences in industrialization during the nine- new technology will not be introduced. This assump- teenth century. tion is inconsequential, however, because we will see

117 Economic Backwardness in Political Perspective February 2006 in the following that the newcomer will always choose higher Z leads to a first-order stochastically dominated
xt = 1.) shift of the function F. Therefore, new technologies will Because this equation is crucial for the analysis, it is enrich new groups, increase Z, and reduce expected worth dwelling on it. First focus on the case when xt = cost of replacing rulers. This model directly maps into 0, so that the incumbent ruler has not innovated. In this our setup. This approach is also consistent with the case, we have At = At−1((1 − pt) + pt(1 + (α − 1)
xt − instances in which landed elites opposed industrializa- z)). Therefore, if citizens choose pt = 0 and do not re- tion because they feared political power moving away place the ruler, there is no change in productivity; that from themselves to the bourgeoisie. is, At = At−1. Alternatively, citizens could choose pt = Typically, political elites maintain their position by 1, incurring the cost z. In this case, whether there is in- using military power. Imagine that there are K regions novation or not then depends on the decision of the new (or groups), and the ruler will be replaced if anyone of ruler,
xt. Alternatively, when xt = 1, the cost of replac- these regions has a successful insurrection. The ruler ing the ruler is z and we have At = At−1((1 − pt)α + has some budget (or military personnel) M to allocate pt(1 + (α − 1)
xt − z )). for control at the beginning of each period, and it will The most important feature of this equation is that allocate this to minimize the probability of successful it allows us to model the notion that costs of polit- insurrection anywhere. Moreover, suppose that N < K ical change depend on whether the new technology of these regions pose a threat, and the probability of has been adopted. When the new technology is not insurrection in the other regions is 0. In the N regions, introduced, the position of the incumbent is relatively this probability is Q(m) where m is military resources secure, and it will be more costly to replace him. When allocated to that region, and naturally Q < 0. If a re- the new technology is adopted, there is political uncer- gion poses a greater threat, more resources will be al- tainty and turbulence, and part of the advantages of the located there. As long as the society has the same orga- incumbent are eroded. As a result, the cost of replacing nization of production and institutional features, there the incumbent may be lower. will be a high degree of persistence in which regions More explicitly, we assume that z and z are random and groups oppose the greatest threat. For simplicity, variables, enabling stochastic changes in rulers along suppose that if no new technology is introduced, the the equilibrium path. The distribution function of these same regions continue to pose the threat, so the ruler two shocks differ: z is drawn from the distribution F N knows which regions are dangerous at the time of al- and z is drawn from F I , which is first-order stochasti- locating its resources. Then, he will allocate m = M/N cally dominated by F N, capturing the notion that the to each region that poses a threat, and the probability N introduction of a new technology creates turbulence of replacement will be qNI = 1 − [1 − Q(M/N)] .In and erodes part of the incumbency advantage of the contrast, the outcomes of new technologies, in terms initial ruler. of which segments of the society and which regions The assumption that costs (or opportunities) of re- will be losers and winners and therefore in which re- placing rulers are different following certain economic gions there will be additional turbulence, are uncertain. and institutional changes is essential for our argument. Assume that after the new technology is introduced, Without this, voters are always less likely to replace there is a totally new draw of which regions pose the a ruler who adopts the new technology and provides threat. Thus, the optimal strategy for the ruler after greater output, and the ruler obtains greater returns the introduction of new technology is to evenly dis- with the new technology; therefore, there would never tribute its resources across regions, that is, M/K in each be any blocking. There are a number of reasons why region. Hence, the probability of replacement will be N costs of replacing rulers will indeed be affected by qI = 1 − [1 − Q(M/K)] > qNI because M/K < M/N, economic and institutional change in practice, and and the ruler is not concentrating its resources in the there are various ways of modeling these issues. For dangerous areas. Therefore, economic and institutional example: change will create turbulence and increase the proba- Often, there is a contest for political power, and those bility that the ruler will be replaced. It is straightfor- with greater economic power are more likely to obtain ward to generalize this model by making the probabil- it. New technologies often benefit not the rulers, but ity of replacement also a function of citizens’ income, some of the competing groups. This is clearly the case as in our baseline model, for example, by making Q(m) in the example of nineteenth-century industrialization, not directly the probability of replacement, but the where the bourgeoisie was the main beneficiary of mean of the distribution of the cost of replacement, economic growth and naturally used its new economic F(z). With this generalization, this approach maps ex- powers to gain political strength. This feature can be actly into our current model. Notably, this approach incorporated into our model in a number of ways. The is consistent with the historical instances, like Russia simplest would be to assume that a disproportionate and Austria-Hungary, where the main fear of the polit- share of the benefits from new technology go to the can- ical elites was revolutionary challenges to their power didate new ruler, who uses all of its income to contest coming from the emergent proletariat in new industrial power (this would be the case as long as returns from areas. coming to power are sufficiently large). Then we can Rulers are in power because, by definition, they have think of the cost of replacing rulers, z, as drawn from a managed to come to power and successfully fought cumulative probability distribution F(z | Z), where Z challenges. We can think of some characteristic, ζ, which is the total amount spent by the potential new ruler. A corresponds to the ability of the rulers to deliver public

118 American Political Science Review Vol. 100, No. 1 goods to citizens or fight challenges. This characteristic on whether there are changes that erode the incum- will typically depend on the ruler and the environment; bency advantage of rulers. for example, ζr(e), where r denotes the ruler and e We also assume that if an incumbent is replaced then the environment. Suppose that the ruler will remain whether or not innovation takes place in that period in power as long as ζr(e) ≥ ζc(e), where c denotes the depends on what the new ruler does. Thus, if the in- challenger and ζc(e) can be thought of as a draw from cumbent innovates but is replaced the new ruler can a distribution Gc. Plausibly, ζr(e) remains constant as decide not to innovate and this implies that there is no long as e remains constant, but when there is an inno- innovation (though as we shall see along the equilib- vation and the environment changes, say to e, there is rium path a new ruler always innovates). a new draw from a distribution Gr, and the ruler main- Finally, rulers levy a tax T on citizens. We assume tains power if ζr(e ) ≥ ζc(e ). This model also maps into that when the technology is A, citizens have access our reduced form setup with a different distribution to a nontaxable informal technology that produces of costs (or probability) of replacement, because by (1 − τ)A. This implies that it will never be optimal for virtue of having previously survived in power, ζr(e) rulers to impose a tax greater than τ. is already above a certain threshold (i.e., it can be It is useful to spell out the exact timing of events thought of as a draw from a truncated-below version within the period. of the distribution function Gr), so is greater than the expected new draw from Gr. Intuitively, the ruler is 1. The period starts with technology at At. in power because he is “selected” as a good match to 2. The incumbent decides whether to adopt the new the current environment, and by innovating, he would technology, xt = 0or1. destroy this advantage. To make this model map into 3. The stochastic costs of replacement, zt or zt, are re- our current setup, we need to make the probability of vealed. replacement not simply a function of the innovation 4. Citizens decide whether to replace the ruler, pt. decision, but also of the incomes of the citizens, which 5. If they replace the ruler, a new ruler comes to power is again straightforward. and decides whether to adopt the new technology Here, rather than pursue any one of these three,
xt = 0or1. or other possible, specific models, we simply adopt the 6. The ruler in power decides the tax rate, Tt. reduced form assumption that the costs of replacement come from the two different distribution functions, F I and F N, depending on whether there is innovation or Social Planner’s Solution not. Furthermore, to simplify the algebra, we assume We start by characterizing the technology adoption I − 1 + 1 N that F is uniform over [µ 2 ,µ 2 ], whereas F is and replacement decisions that would be taken by an − 1 + 1 ≥ uniform over [γµ 2 ,γµ 2 ], where γ 1. In this for- output-maximizing social planner. This can be done mulation, µ is an inverse measure of the degree of by writing the end-of-period Bellman equation for the political competition: when µ is low, incumbents have social planner, S(A). As with all value functions, we use little advantage, and when µ is high, it is costly to re- the convention that S(A) refers to the end-of-period 1 value function (after step 6 in the timing of events place the incumbent. Note that µ can be less than 2 , and in fact, we will focus much of the discussion on the case above). By standard arguments, this value function can in which µ< 1 , so that citizens sometimes replace be written as: 2 =   rulers. The case of µ 0 is of particular interest, be-   cause it implies that there is no incumbency advantage, S(A) = A + β xS 1 − pS(z) S(αA) + pS(z) and z is symmetric around zero. I I On the other hand, γ is a measure of how much the  ×
S − + −
S − I incumbency advantage is eroded by the introduction (x S((α z )A) (1 x )S((1 z )A)) dF  of a new technology: when γ = 1, the costs of replac-   + − S − S + S ing the ruler are identical irrespective of whether a (1 x ) 1 pN(z) S(A) pN(z) new technology is introduced or not. A new entrant  becomes the incumbent ruler in the following period  after he takes control, and it will, in turn, be costly to × (
xSS((α − z)A) + (1 −
xS)S((1 − z)A)) dF N , replace him. A higher γ, all else equal, also implies a lower probability of replacement for the ruler without (2) innovation, thus makes the position of noninnovating rulers more secure. where xS denotes whether the social planner dictates Citizens replace the ruler if a new ruler provides that the incumbent adopts the new technology, while them with higher utility. This assumption is made for
xS denotes the social planner’s decision of whether simplicity, and similar results are obtained if citizens’ to adopt the new technology when he replaces the S ∈{ } replacement decision translates into stochastic replace- incumbent with a new ruler. pI (z ) 0, 1 denotes ment of rulers (e.g., via revolution, coup, or simple whether the planner decides to replace an incumbent shifts of power). The important feature is that citizens who has innovated when the realization of the cost of S ∈{ } are more likely to replace a ruler that provides them replacement is z , whereas pN(z) 0, 1 is the decision with less income, but their decisions are also influenced to replace an incumbent who has not innovated as a by costs of replacing rulers. These costs, in turn, depend function of the realization z.

119 Economic Backwardness in Political Perspective February 2006

Intuitively, when technology is given by A,theto- rate
T. The strategy of the citizens consists of a replace- tal output of the economy is A, and the continuation ment rule, p(x, z, z) ∈{0, 1},withp = 1 corresponding value depends on the innovation and the replacement to replacing the incumbent. The action of citizens is decisions. If xS = 1, the social planner induces the in- conditioned on x, because they move following the cumbent to adopt the new technology, and the social technology adoption decision of the incumbent. At this value when he is not replaced is S(αA). When the plan- point, they observe z, which is relevant to their payoff, ner decides to replace the incumbent, there is a new if x = 0, and z,ifx = 1. An MPE of this game consists ruler and the social planner decides if he will adopt of a strategy combination {x, T,
x, T
, p(x, z, z)}, such the new technology,
xS. In this case, conditional on that all these actions are best responses to each other the cost realization, z , the social value is S((α − z )A) for all values of the state A. or S((1 − z )A) depending on whether the new tech- We will characterize the MPEs of this game by writ- nology is adopted. Notice that if
xS = 1 and the new- ing the appropriate Bellman equations. Let us denote comer innovates, this affects the output potential of the the end-of-period value function of citizens by V(A) economy immediately, hence the term (α − z )A.The (once again this is evaluated after step 6 in the timing second line of (2) is explained similarly following a of events), with A inclusive of the improvement due to decision by the planner not to innovate. The important technology adoption and the losses due to turbulence point in this case is that the cost of replacement is drawn and political change during this period. With a similar from the distribution F N not from F I . reasoning to the social planner’s problem, we have: By standard arguments, S(A) is strictly increasing   − in A. This immediately implies that S((α z )A) > V(A) = A(1 − T) + β x [(1 − p (z))V(αA) + p (z) S((1 − z)A) since α>1, so the planner will always I I choose
xS = 1. The same reasoning implies that the I social planner would like to replace an incumbent × (
xV((α − z )A) + (1 −
x)V((1 − z )A))] dF  who has innovated when S((α − z )A) > S(αA), that is, when z < 0. Similarly, she would like to replace an + (1 − x) [(1 − pN(z))V(A) + pN(z) − incumbent who has not innovated when S((α z)A) >  S(A), that is, when z <α− 1. Substituting for these N decision rules in (2), the decision to innovate or not × (
xV((α − z)A) + (1 −
x)V((1 − z)A))] dF , boils down to a comparison of (3) Value from innovating   where pI (z ) and pN(z) denote the decisions of the µ+ 1 0 2 citizens to replace the incumbent as a function of his = S (αA) dz + S ((α − z ) A) dz 0 µ− 1 innovation decision and the cost realization. Intuitively, 2 the citizens produce A and pay a tax of TA. The next and two lines of (3) give the continuation value of the citizens. This depends on whether the incumbent in- Value from not innovating novates or not, x = 1orx = 0, and on the realization   γµ+ 1 α−1 of the cost of replacing the incumbent. For example, 2 = S (A) dz + S ((α − z) A) dz . following x = 1, citizens observe z and decide whether α−1 γµ− 1 to keep the incumbent. If they do not replace the 2 incumbent, pI (z ) = 0, then there is no cost, and the In the Appendix, we show that the first expression is value to the citizens is V(αA). In contrast, if they decide always greater. Therefore, the social planner always pI (z ) = 1, that is, they replace the incumbent, then the innovates. Intuitively, the society receives two benefits value is V((α − z)A) when the newcomer innovates, from innovating: first, output is higher; second, the ex- and V((1 − z)A) when he doesn’t. The third line is pected cost of replacing the incumbent is lower. Both explained similarly as the expected continuation value of these benefits imply that the social planner always following a decision not to innovate by the incumbent. S = strictly prefers x 1. For future reference, we state The end-of-period value function for a ruler (again (for a formal proof see Acemoglu and Robinson, 2002): evaluated after step 6 in the timing of the game, so once Proposition 1. The social planner always innovates, he knows that he is in power) can be written as   that is, xS = 1. I W(A) = TA + β x (1 − pI (z ))W(αA) dF Equilibrium   + − − N We now characterize the decentralized equilibrium (1 x) (1 pN(z))W(A) dF . (4) of this game. We will limit attention to pure strat- egy Markov Perfect Equilibria (MPE) of this repeated The ruler receives tax revenue of TA, and receives a game. The strategy of the incumbent in each stage game continuation value which depends on his innovation is simply a technology adoption decision, x ∈ [0, 1], and decision x next period. This continuation value also a tax rate T ∈ [0, 1] when in power, the strategy of a new depends on the draw z or z, indirectly through the entrant is also similarly, an action,
x ∈{0, 1} and a tax replacement decisions of the citizens, pI (z ) and pN(z).

120 American Political Science Review Vol. 100, No. 1

Standard arguments immediately imply that the where the function P is defined as follows: P[h] = 0 value of the ruler is strictly increasing in T and A. if h < 0, P[h] = h if h ∈ [0, 1], and P[h] = 1ifh > 1, Because, by construction, in an MPE the continuation making sure that the first term is a probability (i.e., it value does not depend on T, the ruler will choose the does not become negative or greater than 1). Similarly, maximum tax rate T = τ. the value to the ruler of not innovating is Next, consider the innovation decision of a new ruler. Here, the decision boils down to the comparison of Value from not innovating − − W((1 z)A) and W((α z)A). Now the strict mono- = [1 − F N(α − 1)]W(A) tonicity of (4) in A and the fact that α>1 imply that  
x = 1 is a dominant strategy for the entrants. 1 = P + γµ − (α − 1) W(A), (8) The citizens’ decision of whether or not to replace 2 the incumbent ruler is also simple. Again by standard arguments V(A) is strictly increasing in A.Thisisanim- which differs from (7) for two reasons: the probability portant point and emphasizes that citizens will be hap- of replacement is different, and the value conditional pier with and less likely to replace a ruler that provides on no-replacement is lower. 1 + − them with greater income. Thus all else equal, greater It is straightforward to see that if P[ 2 γµ income (and better technology) further helps the ruler − 1 + (α 1)] < P[ 2 µ], so that the probability of replace- by increasing his chances of remaining in power. Never- ment is higher after no-innovation than innovation, theless, adoption of new technology not only increases the ruler will always innovate—–by innovating, he is income but also creates greater uncertainty, eroding increasing both his chances of staying in power and the political advantages of the ruler. This erosion is his returns conditional on staying in power. Therefore, the source of the tradeoff facing the ruler in deciding there will only be blocking of technological or institu- whether or not to innovate. tional change when More specifically, citizens will replace the incumbent     ruler whenever V(A) < V(A), where A is the output 1 1 P + γµ − (α − 1) > P + µ ; (9) potential under the incumbent ruler and A is the out- 2 2 put potential under the newcomer. Now consider a ruler who has innovated and drawn a cost of replace- that is, when innovation, by creating “turbulence,” in- ment z. If citizens keep him in power, they will receive creases the probability that the ruler will be replaced. V(αA). If they replace him, taking into account that For future reference, note that by the monotonicity the new ruler will innovate, their value is V((α − z)A). and continuity of the function P[.] there exists aγ ¯, Then, their best response is: such that (9) holds only when γ>γ¯. Therefore, as long as γ ≤ γ¯, there will be no blocking of new technologies pI (z ) = 0ifz ≥ 0 and pI (z ) = 1ifz < 0. (5) or institutional change. To fully characterize the equilibrium, we now conjec- Next, following a decision not to innovate by the incum- ture that both value functions are linear, V(A) = v(x)A bent, citizens compare the value V(A) from keeping the and W(A) = w(x)A. The parameters v(x) and w(x)are incumbent to the value of replacing the incumbent and conditioned on x, because the exact form of the value − having the new technology, V((α z)A). So: function will depend on whether there is innovation. p (z) = 0ifz ≥ α − 1 and p (z) = 1ifz<α− 1. (More generally, one might want to write v(x, xˆ) and N N w(x, xˆ), but we suppress the second argument, because (6) in any MPE, we havex ˆ = 1.) Note however that w(x) It is noteworthy that replacement rule of the citizens is and r(x) are simply parameters, independent of the identical to the one used by the social planner above. state variable, A. It is straightforward to solve for these This shows that the only source of inefficiency in the coefficients (see the Appendix). Here, the condition for model stems from the innovation decision by the in- the incumbent to innovate, that is, for (7) to be greater than (8), can be written simply as: cumbent ruler.     Finally, the incumbent will decide whether to in- 1 1 w (x) αAP + µ ≥ w(x)AP +γµ − α−1 novate by comparing the continuation values. Us- 2 2 ing the decision rule of the citizens, the return to µ+ 1 ⇐⇒ innovating is 2 (1 − p (z)) · W(αA)dF I , and the     µ− 1 I 2 1 1 value to not innovating is given by the expression αP + µ ≥ P + γµ − (α − 1) . (10) γµ+ 1 2 − · N 2 2 − 1 (1 pN(z)) W(A)dF . Now incorporating the γµ 2 decision rules (5) and (6), and exploiting the uniformity When will the incumbent adopt the new technol- of the distribution function F I , we obtain the value of ogy? First, consider γ. A higher γ always discourages innovating as innovation. This is intuitive, because a higher level of γ increases the security of the ruler in the absence of Value from innovating innovation; hence, it implies a greater erosion of en-   trenchment. 1 = [1 − F I (0)]W(αA) = P + µ W(αA), (7) Next, consider µ, which affects the extent of en- 2 trenchment both before and after change and hence

121 Economic Backwardness in Political Perspective February 2006 can be thought of as to measure of the degree of politi- block change, all future incumbents will also do so. cal competition. Now imagine the case µ = 0, where There will still be increases in A, as incumbents are there is no incumbency advantage (i.e., the cost of sometimes replaced and newcomers undertake inno- replacing the incumbent is symmetric around 0). In vations, but there will never be a transition to a political this case, there is “fierce” competition between the in- equilibrium with no blocking. cumbent and the rival. Condition (10) then becomes We now summarize the main results of this analysis: 1 1 − − αP[ 2 ] > P[ 2 (α 1)], which is always satisfied be- cause α>1. Therefore, when µ = 0, the incumbent Proposition 2. A higher γ always discourages innova- will always innovate; that is, x = 1. By continuity, for tion. The effect of µ is ambiguous: when µ is sufficiently µ low enough, the incumbent will always innovate. In- small or large (political competition very high or very tuitively, when µ is low, because the rival is as good low), the elites will always innovate. For intermediate as the incumbent, citizens are very likely to replace values of µ, economic change may be blocked. an incumbent who does not innovate. As a result, the incumbent innovates in order to increase his chances of As emphasized above, elites will block change be- staying in power. The more general implication of this cause of the political replacement effect: in the region result is that incumbents facing fierce political com- where blocking is beneficial for the incumbent ruler, petition, with little incumbency advantage, are likely the probability that he will be replaced increases when to innovate because they realize that if they do not there is economic change. This implies that the incum- innovate they will be replaced. bent ruler fails to fully internalize future increases in Next, consider the polar opposite case where µ ≥ output, making him oppose change. 1/2; that is, there is a very high degree of incumbency Perhaps the most interesting results is the nonmono- + 1 = ≥ 1 + − tonic relationship between political institutions, as cap- advantage. In this situation P[µ 2 ] 1 P[ 2 γµ − tured by µ, and economic change. Because µ is the (α 1)], so there is no advantage from not innovating only measure of political competition in our model, because the incumbent is highly entrenched and can- a low level of µ corresponds both to limited incum- not lose power. This establishes that highly entrenched bency advantage and to lack of political threats, and incumbents will also adopt the new technology. more generally, to an environment where the masses The situation is different, however, for intermedi- have some degree of control over political elites. For ate values of µ. Inspection of condition (10) shows example, we think of a society like the United States in that for µ small and γ large, incumbents will prefer the nineteenth century with weak political elites tightly not to adopt the new technology. This is because of constrained by institutions and high levels of politi- the political replacement effect, which operates in the cal participation as corresponding to low µ, whereas case where γ>γ¯: the introduction of new technology Germany and Britain where the landed aristocracy, increases the likelihood that the incumbent will be re- through the House of Lords and the Coalition of Iron placed, effectively eroding his future rents. As a result, and Rye, were highly entrenched correspond to a high the incumbent may prefer not to innovate in order value of µ. Moreover, changes in domestic or foreign to increase the probability that he maintains power. situations can correspond to changes in µ. For example, The reasoning is similar to the replacement effect in the defeat of Russia in the Crimean War is likely to have industrial organization emphasized by Arrow (1962): increased the political threat to the monarchy, and via incumbents are less willing to innovate than entrants this channel, to have reduced µ. because they will be partly replacing their own rents. Political competition is often viewed as a guarantee Here this replacement refers to the rents that the in- for good political outcomes, and this view has moti- cumbent is destroying by increasing the likelihood that vated many constitutions to create a level playing field. he will be replaced. Our model partly confirms this view; with low µ new To determine the parameter region where blocking technologies will be adopted, because citizens will re- happens, note that there can only be blocking when move incumbents who do not innovate. And yet, it 1 + 1 + − − both P[ 2 µ] and P[ 2 γµ (α 1)] are between 0 also highlights potential costs from political competi- 1 + 1 + − and 1, hence respectively equal to 2 µ and 2 γµ tion under different conditions (for other ranges of pa- (α − 1). Then from (10), it is immediate that there will rameters): whereas highly entrenched political elites, be blocking when that is, those with very high levels of µ adopt bene- ficial economic change, and those with intermediate 3 α − 1 γ>α+ . (11) values of µ fear replacement and are likely to resist 2 µ change. Hence, as α → 1, provided that γ>1, there will always be blocking. More generally, a lower gain from innova- POLITICAL STAKES AND DEVELOPMENT tion; that is, a lower α makes blocking more likely. Clearly, because the social planner always adopts So far we have considered a model in which the only new technologies, whenever the incumbent ruler de- benefit of staying in power was future tax revenues cides not to adopt, there is inefficient blocking of ben- from the same technology that generated income for eficial technological and institutional change. the citizens. There are often other sources of (pecu- Finally, also note that condition (10) does not depend niary and nonpecuniary) rents for political elites, which on A. Therefore, if an incumbent finds it profitable to will affect the political equilibrium by creating greater

122 American Political Science Review Vol. 100, No. 1

“stakes” from staying in power. This relates to an Again let us start with the decision of citizens. As intuition dating back to Madison, Hamilton, and Jay before,
V(Ah) is strictly increasing, so the citizens will (1987) that emphasizes the benefits of having limited use the same replacement rules as before, (5) and (6). political stakes. We now introduce these additional Then, with a similar reasoning, the value to the incum- sources of rents for political elites, enabling us to for- bent ruler of innovating and not innovating at time t malize this intuition: when rents from political power, are given by: “political stakes,” are large, for example, because of rents from land and natural resources, or because ex- Value from innovating isting political institutions do not constrain extraction I
= [1 − F (0)]W(αAth, R) by rulers, political elites will be more likely to block   1 development. We will also use this extended model to = P + µ W
(αA h, R) (13) discuss the importance of human capital in affecting the 2 t political equilibrium and show that high human capital will make blocking less likely. and We model these issues in a simple way by allowing Value from not innovating income at date t to be Ath, where h represents the N
exogenous stock of human capital. We assume that the = [1 − F (α − 1)]W(Ath, R) structure of taxation is as before so that now the ruler   1
gets tax income of τAth, and we additionally assume = P + γµ − (α − 1) W (Ath, R), (14) that a rent of R accrues to the ruler in each period. 2 The two important assumptions here are: first, the po- Notice that although, via the effect of A , the value litical rent to the incumbent does not grow linearly t functions, the W
’s, depend on time, the probabilities of with technology, A. This implies that a higher A makes staying in power do not, because the decision rules of the political rent less important. This is reasonable in the citizens do not depend on time. our context, because in the applications that follow Next, again by standard arguments W
is strictly in- we would like to think of R as related to rents from creasing in both of its arguments. This implies that preindustrial production relations. Second, human cap- − I
− N −
ital is more complementary to technology than to the if [1 F (0)]W(αAth, R) > [1 F (α 1)]W(Ath, R), I
other assets in the economy that are generating the then it is also true that [1 − F (0)]W(αA h, R) > [1 − N
political rents. This is again plausible since we con- F (α − 1)]W(A h, R), for all A ≥ At. Because innova- sider Ath as income from new sectors, industry and tions increase At, this implies that once an incumbent commerce. starts innovating, both that incumbent and all future Let us now write the value function for the citizen, incumbents will always innovate.
denoted V(Ah): This implies that we can characterize the condition   for innovation as follows: first determine the value
= − + −
+ function for the ruler under the hypothesis that there V(Ah) Ah(1 T) β x [(1 pI (z ))V(αAh) pI (z ) will always be innovations in the future, and then check whether the one-step ahead deviation of not inno-

I × (
xV((α − z )Ah)+(1 −
x)V((1 − z )Ah))] dF vating in this period is profitable. To do this, let us  make the natural conjecture that
V(Ah) =
v(x)Ah and

+ (1 − x) [(1 − pN(z))V(Ah) + pN(z) W(Ah, R) =
w(x)Ah +
r(x)R, where we have explicitly  allowed the coefficients of these value functions to de- pend on whether there will be innovation in the future.

N × (
xV((α − z)Ah) + (1 −
x)V((1 − z)Ah))] dF . By a similar reasoning to that above, the incumbent ruler innovates if (12)   1 P + µ (
w(x = 1)αAh +
r(x = 1)R) Equation (12) is very similar to (3). The value for the 2
  incumbent ruler, W(Ah, R), is 1 ≥ P + γµ − (α − 1) (
w(x = 1)Ah +
r(x = 1)R),   2 W
(Ah, R) = TAh + R + β x (1 − p (z)) I (15)

× W
(αAh, R) dF I + (1 − x) where
w (x = 1) and
r (x = 1) are the coefficients of the   value functions when there will always the innovation × (1 − p (z))W
(Ah, R) dF N , in the future and are simple functions of the underlying N parameters. Let us first focus on the main comparative statics whose interpretation is immediate from (4). A major of interest. As before, condition (15) can only be vi- 1 + − − 1 + difference from before is that whether blocking is pre- olated when P[ 2 γµ (α 1)] > P[ 2 µ], that is, ferred by the incumbent ruler will now depend on the when innovation reduces the likelihood that the ruler value of A. will remain in power. Then, in this relevant area of the

123 Economic Backwardness in Political Perspective February 2006

parameter space where blocking can occur, the coeffi- Here, because At tends to increase over time (even 1 + − − when there is blocking, because incumbents are being cient of R on the right-hand side, P[ 2 γµ (α 1)], is greater than the corresponding coefficient on the left- replaced and newcomers innovate), eventually At will 1 + reach the threshold A∗(R/h) as long as A∗(R/h) < ∞, hand side, P[ 2 µ], so an increase in R makes blocking more likely (i.e., it makes it less likely that (15) holds). and from this point onwards, incumbents would no Conversely, an increase in h, the human capital of the longer block innovations. Therefore, a possible devel- labor force, makes blocking less likely. Intuitively, a opment path implied by this analysis is as follows: higher level of R implies a greater loss of rents from first, incumbents block change, but as many of these relinquishing office, increasing the strength of the polit- incumbent rulers are replaced and some economic and ical replacement effect. In contrast, the higher level of political change takes place, the society eventually un- h increases the gains from technology adoption relative dergoes a political transition—–it reaches the threshold to R, making technology adoption more likely. where even incumbent rulers are no longer opposed More explicitly, condition (15) implies that, as long to change. Of course, the arrival of such a political 1 + − 1 + − − transition may be very slow. as P[ 2 µ]α P[ 2 γµ (α 1)] > 0, the ruler will innovate at time t if Summarizing the main implication of this analysis: ∗ A ≥ A (R/h) t      Proposition 3. Political elites are more likely to block P 1 + γµ − (α − 1) − P 1 + µ economic change when political rents, R, are high and ≡   2   2  human capital of the workforce, h, is low. P 1 + µ α − P 1 + γµ − (α − 1) 2   2 1 − βαP 1 + µ R This proposition is important for our discussion be- ·  2  · . (16) low because it implies that elites are more likely to − β 1 + µ τh 1 P 2 block change when political stakes, as captured by R, are greater. As a result, in this model we can think of The inequality P[ 1 + µ]α − P[ 1 + γµ − (α − 1)] > 0is 2 2 two distinct roles of preexisting political institutions: imposed because if it did not hold, the incumbent would first, they determine µ, the degree of political compe- never innovate irrespective of the value of A,sointhis ∗ =∞ tition, and affect the likelihood of economic change case we set A (R/h) . Condition (16) states that via this channel; and second, they affect the political the incumbent will innovate if the current state of tech- ∗ stakes, R, and determine the gains to the elites from nology is greater than a threshold level A . The greater staying in power and their willingness to block deve- is R/h, the higher is this threshold level, implying that lopment. innovation is less likely (or will arrive later). Finally, it is interesting to note that although we have Next suppose, instead, that the incumbent would like so far treated h as given, the same forces that determine to block innovation. Then as long as he remains in = whether incumbent rulers want to block change will power At+1 At. This enables a simple characteriza- also determine whether they want to invest in the hu- tion of the value functions, again using the natural lin- man capital of the population. A greater human capital earity conjecture (see the Appendix). We then obtain of the labor force is likely to increase output, but may that the incumbent will block if   make it easier for the masses to organize against the 1 ruler, and hence may erode the incumbency advan- P + γµ − (α − 1) (
w (x = 0) Ah +
r (x = 0) R) 2 tage of the ruler (see Bourguignon and Verdier 2000).   Therefore, the political replacement effect may also 1 ≥ P + µ (
w (x = 1) αAh +
r (x = 1) R), (17) serve to discourage rulers from investing in human 2 capital or even block initiatives to increase the human capital of the masses. where the right-hand side features
w(x = 1) and
r(x = 1), because if the ruler finds it profitable to innovate this period, he will also do so in the future. This condition EXTERNAL THREATS AND DEVELOPMENT will be satisfied if only if A ≤ A∗(R/h) as given by (16); that is, only if A is lower than the critical threshold Political elites’ attitudes toward industrialization characterized above. On reflection, this result is not changed dramatically in Russia after the defeat in the surprising, because given the best responses of the citi- Crimean war and in Austria-Hungary after the 1848 zens, the decision problem of the ruler is characterized Revolution. Similarly, Japanese elites started the pro- by a standard dynamic programming problem, and has cess of rapid industrialization after they felt threat- a unique solution. ened by the European powers and the United States. ∗ Overall, the ruler will innovate when At ≥ A (R/h) One can also argue that the potential threat of com- ∗ and block whenever At < A (R/h). This result has an- munism was an important influence on the elites’ atti- other interesting implication. Before, the innovation tudes toward development in South Korea and Taiwan. decision was independent of A, so an economy that had It therefore appears that external threats may be “adverse” parameters would always experience block- an important determinant of whether elites want to ing. Although there would still be some improvements block technical and institutional change. In this sec- in technology as incumbents were replaced, incum- tion, we extend our model to incorporate this possi- bents would always block change whenever they could. bility.

124 American Political Science Review Vol. 100, No. 1

To model these issues in the simplest possible way, The intuition for both comparative statics is straight- suppose that at time t, rulers find out that there is a one- forward. With a more powerful external threat or a period external threat at t + 1, which was unanticipated more developed perpetrator, the ruler will be “forced” before. In particular, another country (the perpetrator) to allow innovation so as to reduce the likelihood of an with technology Bt may invade. If an invasion takes invasion. place, the ruler is kicked out of power and gets zero utility from that point on. Whether this invasion will HISTORICAL EVIDENCE AND take place or not depends on the state of technol- INTERPRETATION ogy in two countries, and on a stochastic shock, qt. If qtφBt > At, the perpetrator will successfully invade We now use our analysis so far to provide an “in- and if qtφBt ≤ At, there will be no invasion. Hence, terpretation” of the different industrialization expe- φ ≥ 0 parameterizes the extent of the external threat: riences during the nineteenth century. As the data when φ is low, there will only be a limited threat. This on output and industrial production from Maddison formulation also captures the notion that an economy (1995) and Bairoch (1982) show (Table 1 Panels A and that produces more output will have an advantage in a B), whereas some countries, including Britain and the conflict with a less productive economy. United States, industrialized rapidly, others, such as For simplicity, suppose that there will never be an Japan, Russia, and Austria-Hungary, lagged behind. invasion threat again in the future, and assume that Even Germany, though a famous example of rapid qt is uniformly distributed on [0, 1]. Suppose also that industrial catch-up, did not really take off until after Bt = δAt−1. This implies that there will be an invasion 1850. We provide historical evidence suggesting that if a proximate cause of this divergence is that in some countries political elites did not want to introduce eco- 1 + xt(α − 1) qt > , nomic institutions encouraging entrepreneurs to adopt φδ new technology and innovate, and in fact, actively tried to block industrialization (see Acemoglu, Johnson, and where recall that x is the decision of the incumbent to t Robinson 2002 for econometric evidence). Moreover, innovate. Using the fact that q is uniform over [0, 1], t the evidence suggests that the primary reason why and the same definition of the function P[·], we have elites blocked such change was because they feared the probability that the ruler will not be invaded at time losing their political power. In addition our theory also t, conditional on xt,as   provides an interpretation for why there were such adverse attitudes toward change in some countries, 1 + xt (α − 1) P . and not in others. It suggests that the political elites in φδ Britain and subsequently Germany did not strongly op- The important point here is that the probability of in- pose industrialization and the associated institutional changes because they were sufficiently entrenched and vasion is higher when xt = 0 because output is lower. Let us now return to the basic model presented political rents were relatively low. In Britain this was above with R = 0. The same reasoning as before imme- true throughout the eighteenth and nineteenth cen- diately establishes that at time t the ruler will innovate turies, whereas in Germany it was not until the sec- if ond half of the century that political elites forged the         coalition that entrenched their political power. We also 1 α 1 1 suggest that external threats from the European pow- αP + µ P ≥ P + γµ − (α − 1) P . 2 φδ 2 φδ ers and the United States were important in changing the attitudes of the Japanese elites toward industrial- (18) ization and modernization, spurring rapid industrial- ization there. This condition differs from the basic condition we In contrast, in Russia and Austria-Hungary, the po- previously derived, (10), because of the probability of litical elites were powerful enough to block develop- 1 ∈ invasion. In particular, when P[ φδ ] (0, 1), external ment, and because their political power was not to- threats make blocking less attractive, because a rela- tally secure, they feared technological and institutional tively backward technology increases the probability change. We also suggest that because of the more feudal of foreign invasion. Therefore, the emergence of an land/labor relations and the persistence of the abso- external threat might induce innovation in a society lutist monarchy into the nineteenth century, political that would otherwise block change. In fact, an increase stakes in Russia and Austria-Hungary were substantial in δ or φ will typically make blocking less likely. For for the landed aristocracy, encouraging them to oppose → → 1 → example, when δ 0orφ 0, we have P[ φδ ] 1, so industrialization to protect their rents. the threat of invasion disappears and we are back to Needless to say, such interpretations are necessar- condition (10). For future reference, we state this result ily speculative and more conclusive evidence requires as a proposition: proper statistical testing of the ideas we develop here. We regard the main contribution of this section as Proposition 4. Political elites are less likely to block showing that there is direct historical evidence that the development when there is a severe external threat (high mechanism which is the central focus of this paper was φ) and when the perpetrator is more developed (high δ). important to nineteenth-century industrialization.

125 Economic Backwardness in Political Perspective February 2006

Britain fore the British political elite, by adopting a strategy of gradual concessions, were able to control the po- At the end of the eighteenth century, the political litical equilibrium and maintain power for at least a system in Britain was controlled by a rich, primarily century following the onset of the political impact of landowning aristocracy. The British elites at the start of industrialization. In terms of our model, both of these the next century faced an ongoing process of industrial- factors correspond to a high value of µ, reducing the ization which, by creating new groups of wealthy busi- opposition of existing elites toward institutional and nessmen and finally a powerful working class, forced technical change. In addition, the long history of Britain the aristocracy to concede political power. as a trading nation and mercantile power meant that This process happened incrementally through the many aristocrats had relatively diversified wealth, and three Reform Acts, the first of which was in 1832, the many of the feudal land relations had long disappeared others in 1867 and 1884. A crucial policy change was (see Cain and Hopkins 2000; Saville 1994). There- the 1846 repeal of the Corn Laws. Although the exten- fore, the elites could benefit from industrialization, and sion of the franchise that was also introduced by these had less to lose in terms of political rents. Moreover, reforms was certainly fought for many years (e.g., in the political institutions which emerged following the the 1840s the Chartist movement was successfully un- Glorious Revolution restricted predation not only by dermined), the political elites in Britain did not block the monarchy against parliament but also by parlia- the processes set in force by the first industrial revolu- ment against commoners (Thompson 1975). These in- tion. Indeed, as Mokyr (1990, 243) argues about Britain stitutional characteristics of the British society further “the landowning elite, which controlled political power reduced political rents, R in our model, making the before 1850, contributed little to the industrial revolu- elites less likely to be oppose change. tion in terms of technology or entrepreneurship. It did not, however, resist it.” It seems that the battle over the extension of the franchise was more relevant for Germany determining how much of the gains from industrializa- tion would be redistributed to the poorer segments of Industrialization occurred in Germany in the context the society, rather than over whether industrialization to the rise of the Prussian state within the German should proceed or not (see Acemoglu and Robinson federation of states and the creation of the Empire 2000b, 2005 on the extension of the franchise). in 1871. The resulting political institutions ensured the British political elites were not always in support of entrenchment of the elites, in particular the landed aris- industrialization. For example, Christopher Hill (1981, tocracy, the Junkers. For example, the Junkers forged 22) describes the hostile attitude of the landed aris- the coalition of ‘Iron and Rye’ with the rising industrial tocracy to modernization during the early seventeenth class to secure their economic interests. Gerschenkron century. However, the rise of parliament and the com- (1943, 49) describes this coalition as “a compromise mercialization of British society during the following between modern industry and the feudal aristocratic centuries overcame these forces and paved the way groups in the country.” And Eley (1984, 153–54) notes for industrialization. The Glorious Revolution of 1688, that “the option of the German bourgeoisie’s lead- often emphasized as the final step in the rise of parlia- ing fractions for a politics of accommodation with the ment over the monarchy (North and Weingast 1989), landowning interests ...was fully compatible with the also represented the final step in a much longer process pursuit of bourgeois interests ... The bourgeoisie en- of evolution. Not only were the political powers of the tered the agrarian alliance ...as the best means of se- monarchy and traditional aristocracy relatively muted curing certain political goals.” in Britain compared to other European countries, but However, before the 1848 revolution and the emer- also economic institutions were much more modern- gence of this coalition, the landed elites did not uni- ized. Feudalism and servile labor were almost com- formly support industrialization. For example, Tracey pletely gone and most of the landed elite had become (1989, 106) writes: “there was relatively little industrial commercial farmers (see Moore 1966). development in the east, because of lack of natural Why were British elites, on balance, not opposed to resources and also as a result of deliberate opposition economic and institutional development? Three rea- to industrialization by the Junkers and the Bund der sons, highlighted by our model, may have been impor- Landwirte, who feared the spread of socialism and also tant. First, the political settlement which had emerged did not want the Polish proletariat to participate in from the Glorious Revolution meant that the landed industrialization.” aristocracy in Britain, though holding only relatively Tipton (1974, 962) argues that in the first half of the limited powers compared to some other European nineteenth century the political elites “adamantly op- aristocracies, did not face severe threats to its posi- posed plans for eastern industrialization on the grounds tion and power in society. In particular, the House of that the danger of socialism would increase with the Lords guaranteed the security of landed interests until expansion of employment in factories” (see also his the Liberal government of Asquith after 1906. Second, 1976 book). Trebilcock (1981, p. 76) makes similar the transition from aristocratic to popular rule was a arguments noting “the Prussian state ...was attracted prolonged one. Although relatively universal male suf- only to a particular type of industrial development, frage came after 1867, aristocratic power was strong in that with military utility. Outside this area, the au- government until well into the present century. There- thorities contributed little in the way of investment

126 American Political Science Review Vol. 100, No. 1 or encouragement to private entrepreneurs.” Indeed, of Iron and Rye had been formed, allowed them to “they expressed hostility to major innovations in in- compensate for the adverse direct economic effects of dustry or transport, and they were notably suspicious industrialization. So in terms of our model, we can see of the railways at the outset.” He continues that as the German case as one where, after the 1850s, the late as the midcentury “the political preferences of political elites were relatively entrenched, that is, high the Junker groups definitely restricted the scope for µ and the political stakes, R, were limited. This com- economic advance.” bination of elite entrenchment and low political rents Yet things began to change in the 1850s “a new al- made it unprofitable for the elites to block change. liance formed, combining authoritarianism with bour- Rapid industrialization among its main rivals may have geois elements, against the menace of peasant and also contributed to the favorable attitudes towards in- proletariat. By modifying the vested interests of the dustrialization in Germany. previous four of five decades, it created a climate more favorable to industrial advance” (Trebilcock 1981, 76). The threat from the rapidly industrializing Japan Britain and France and from the 1848 wave of revo- lutions may have also been important for the change Japan industrialized and modernized rapidly in the in the attitudes toward industrialization, adding some second half of the nineteenth century. Although the element of defensive modernization to the German seeds of industrialization could be traced back to the case. Tokugawa era (e.g., Macpherson 1987), rapid indus- Another factor facilitating the emergence of this trialization appears to have been related to the emer- more positive attitude toward change may have been gence of a serious external threat. The overwhelming that, despite the important role of the Junker elite, British victory against the Chinese in the Opium War the political stakes were also relatively limited for the of 1842, the imposition of an “open door” to world landed aristocracy. This was mostly because of the re- trade on Japan by European powers and the United forms induced by the Napoleonic wars. Blackbourn States, and perhaps most importantly, the arrival of (1997, 71) notes that in the parts of Germany that they the American fleet in Tokyo Bay in 1853 made it clear occupied, the French rule amounted to a “crash course to the Japanese elites that they were facing a serious in modernization that removed the institutions of the external threat. old regime, separated church and state, rebuilt the Before the emergence of the external threat from administrative bureaucracy on a new basis, and made the West, Japan was under the control of the Tokugawa possible the relatively untrammeled accumulation and shogunate which was a coalition of large landowning disposition of property that is one hallmark of a mod- interests. Although Japan had been a relatively pros- ern civil society.” Where the French did not rule, as in perous country since at least the fifteenth century,it was Prussia, their impact was to induce defensive reform not always open to commerce, innovation, and foreign and modernization. However, there is a consensus that trade. For example, the elites did not welcome foreign these reforms did not have the revolutionary effects missionaries, and foreign trade was seriously limited that they were once claimed to have had. Although the and controlled; or in the words of Macpherson (1987, serfs were freed in Prussia in 1807, Trebilcock (1981, 34) p. 38), “...[they] ...virtually prohibited overseas notes, “the small peasantry, immensely more numerous trade.” In terms of our model, this makes sense as but holding only a fraction of the peasant land, the rural a calculated conservatism to maximize the longevity masses between the Elbe and the Vistula, had to wait of the regime. In fact, the main aim of the Tokugawa until the 1850s and 1860s before the miraculous cloak of shogunate was to bring peace and stability to Japan fol- emancipation swept over them.” Labor relations were lowing an era of prolonged infighting and civil war, and certainly less feudal than in Russia or Austria-Hungary, many of the institutional changes needed to promote though they were probably less modernized than in industry would have created turbulence and strength- Britain. ened competing groups. Political reforms after the 1848 Revolution and the The Tokugawa equilibrium was shattered by the emergence of parliamentary institutions in Germany emergence of the external threat from the West, cul- appear to have been strategically designed, as in minating in the arrival of the American fleet in Tokyo Britain, to give a sufficient degree of representation Bay. These changes eventually led to the Meiji Restora- to the Junker elites. For example, the legislature was tion of 1868. Although the Restoration installed the bicameral with a federal council composed of delegates Meiji Emperor, this was purely symbolic, and the main from the states (the Bundesrat) and a national elected drive to industrialization had already started before parliament (the Reichstag). Prussia had 17 of the the Restoration as an explicit strategy of defensive Bundesrat’s 58 votes and “the conservative Prussian modernization in response to these external threats. elite could essentially block proposed national legisla- Curtin (2000, 163) describes this as follows: “The two tion that was contrary to its interests (Berman, 2001, sides were so similar that the brief but crucial fighting 439). After the 1870s, the Junkers were able to gain pro- that ended the Tokugawa era was a struggle between tection for their output, insulating themselves econom- competing military oligarchies seeking to control a new ically from the worst effects of industrialization—–such centralizing government, which would probably have as falling land rents. Therefore, in Germany the contin- sought to carry out similar policies, no matter which ued political power of the Junkers, once the coalition side won.”

127 Economic Backwardness in Political Perspective February 2006

It is interesting that the drive for modernization in threat. Alexsandr’s reforms led to rapid industrializa- Japan took a special form, strengthening the central- tion (Portal 1965), supporting the notion that state ized government and increasing the entrenchment of policies before the Crimean War were important in bureaucratic elites. In terms of our model, this can be blocking development. The sudden change of atti- viewed as a strategy to industrialize while also mini- tudes in the face of foreign threat is also consistent mizing the probability of replacement, somewhat rem- with our emphasis on external threats inducing inno- iniscent to the industrialization experience in Britain vation. and Germany where the nonindustrial elites main- This period of industrialization also witnessed tained their political power despite the process of in- heightened political tensions, consistent with the view dustrialization. In Japan, the Restoration, despite its that times of rapid change destabilize the system emphasis on Japanese traditions, quickly wiped out (McDaniel 1988; Mosse 1958, 1992). Mosse (146) the powers of the daimyo (the great lords), the main argues: “early industrialization ...resulted in an ac- threat to the power of the centralized state. At the celeration of change and an increased dynamism in same time, much of industrial activity was delegated a previously static society. Russia ...was transformed to a core group of wealthy families, known as the from a backward (or underdeveloped) into a develop- Zaibatsu, which was in close contact with the gov- ing country. She had entered the railway age. Society ernment, so the threat from economic change to the experienced a gradual transformation. Although the existing political regime was minimized. The choice of landed gentry was in a state of terminal decline, the Japanese constitution also reflected the same desires. bureaucracy recruited from various social strata, had They adopted the Prussian constitution, which gave begun to replace it as the ruling class. The liberal profes- the appearance of representative government, but re- sions and the entrepreneurial bourgeoisie were gaining tained the oligarchy in control via the bicameral system in numbers and importance. Early industrialization had with the upper house reserved largely for the ruling created a small but growing proletariat. By the turn of elite. The constitution also gave the military sweeping the century, the social changes set in motion by per- powers. estroika were beginning to have an impact on public Overall, the Japanese experience can be interpreted affairs.” as an example of defensive modernization and industri- The history of industrialization in nineteenth- alization in response to an external threat, reminiscent century Russia therefore illustrates the contradictions to the results we derived earlier, and the pattern of that political elites faced in promoting innovation. Un- industrialization following the change of attitudes can til the impetus of the Crimean War, industrialization be interpreted as an example of adopting technolog- was blocked. The defeat of the Crimean war is there- ical and institutional change, while strengthening the fore the turning point in the attitudes of the Russian control of the oligarchy on centralized power. state to economic development. Why was the Tzar so opposed to institutional and technological change in the period before the Crimean Russia war? One possibility suggested by our analysis is the high level of the political rents, R, generated by the Russia provides stark contrast to the cases we have state of land/labor relations in Tzarist Russia. The feu- examined so far. During the reign of Nikolai I between dal social structure in nineteenth-century Russia gen- 1825 and 1855 (in the wake of the Decembrist putsch) erated high rents for political elites and implied that only one railway line was built in Russia. Economic de- the landed aristocracy had few commercial interests, velopment was opposed since, as Mosse (1992, 19) puts and was not associated with industrial groups. So the it, “Economic development—–elsewhere in Europe primary beneficiaries of industrialization would have this was the age of early industrialization and rail- been groups outside the landed aristocracy. Moreover, way construction—–was deliberately held back,” since because land is relatively easy to expropriate, the elites, “it was understood that industrial development might especially the Tzar, had a lot to lose from political lead to social and political change.” In a similar vein, changes. These factors increased the political stakes Gregory (1991, 74) argues: “Prior to the about face in for the elites, making them fear industrialization more the 1850s, the Russian state feared that industrializa- than in Britain or Germany. Furthermore, the poor tion and modernization would concentrate revolution level of human capital of most of the workforce also minded workers in cities, railways would give them implied that the gains from industrialization were lower mobility, and education would create opposition to the relative to land rents that the elites obtained under monarchy.” the ancien regime. Another important factor was the It was only after the defeat in the Crimean War tight control of the state by the Tzar, who had perhaps that Nikolai’s successor, Alexsandr II, initiated a large- even more to lose from political transition. No policy scale project of railway building and an attempt to could be made without the Tzar’s agreement and all modernize the economy (among other things, by in- government ministers were personally appointed by troducing a Western legal system, decentralizing gov- him. Here Russia was very different from the Euro- ernment, and freeing the serfs). The reasons underly- pean countries, such as Britain, where the monarchy ing this industrial drive appear defensive: Alexsandr staged a long, carefully orchestrated withdrawal from II, most probably correctly, perceived that Russia’s political power. In line with this McDaniel (1991, 32) technological inferiority left it vulnerable to foreign notes that “The struggle between kings and burghers

128 American Political Science Review Vol. 100, No. 1 so characteristic of much of western European history In addition, as in Russia, Austria-Hungarian elites had few parallels in Russia.” received substantial rents from unreformed feudal land/labor relations. Furthermore, the fact that they The Habsburg Empire were landowners meant that they could be easily expro- priated in the event of political transition. Another pos- In the nineteenth century, the Habsburg Empire was sibly important factor in the case of Austria-Hungary known as “Europe’s China” or the “sick man on the was the heterogeneity of the empire. The Austrian aris- Danube” (see Good 1984, 1991) and was generally re- tocracy thought that modernization would destroy the garded as having fallen far behind the other industri- social glue which kept the empire together. Poor human alizing countries of Europe. Historians typically argue capital also again lowered the incentive to industrialize that this was what led to the weakness and disintegra- by increasing the importance of land rents relative to tion of the empire during and after the First World War. income from industrial activity. The consensus view among historians now appears to be that the main explanation for the slow growth of Austria-Hungary in the nineteenth century is the op- CONCLUSION position of the state. For instance Gross (1973) argues: “In domestic as well as foreign policy the Vormarz¨ In this paper, we constructed a simple model where regime, from 1815 to 1848, was determined to pre- political elites may block technological and institu- vent another French Revolution anywhere in Europe. tional development, because of a “political replace- From this principle Francis I derived not only his op- ment effect.” Innovations often erode political elites’ position to the growth of industry (and with it the incumbency advantage, increasing the likelihood that Proletariat) ...but his general reluctance to permit any they will be replaced. Fearing replacement, political change whatsoever.” elites are unwilling to initiate economic and institu- The analysis of Freudenberger (1967) is similar. He tional change. We show that elites are unlikely to block notes (498–99): “In the 1790s, fear of the French Rev- development when there is a high degree of political olution added a new dimension to government pol- competition, or when they are highly entrenched. It icy. High officials in the government, with the young is only when political competition is limited and also Emperor Francis I in strong sympathy, felt that indus- the elites’ power is threatened that they will block de- trialization would create an industrial proletariat which velopment. We also show that such blocking is more to them was the carrier of revolutionary ideas and for likely to arise when political stakes are higher, and that reason ...not only opposed large-scale enterprises in the absence of external threats. We argue that this but also favored a policy that kept the population model provides an interpretation for why Britain in- agrarian.” dustrialized first and Germany followed relatively soon As with the Tzar, the Habsburg emperors opposed during the nineteenth century, while the monarchy and the building of railways and infrastructure and there the landed aristocracy in Russia and Austria-Hungary was no attempt to develop an effective educational blocked development. system. Blum (1943, 26) pointed to the pre-modern There are many interesting areas for future research, institutional inheritance as the major blockage to in- both empirical and theoretical. At the empirical level, dustrialization arguing that “these living forces of the we need to develop quantitative measures of the degree traditional economic system were the greatest barrier of political competition and elites’ attitudes toward to development. Their chief supporter was ...Emperor technological change both for today and in the past, Francis. He knew that the advances in the techniques and investigate the relationship between competition of production threatened the life of the old order of and the likelihood of blocking. At the theoretical level, which he was so determined a protector. Because of it may be informative to derive different (potentially his unique position as final arbiter of all proposals testable) implications from models where the fear of for change he could stem the flood for a time. Thus replacement comes from different sources (challenges when plans for the construction of a steam railroad from new groups, fear of revolution, or threats from were put before him, he refused to give consent to other subgroups within the elite). More important their execution ‘lest revolution might come into the would be to model political competition more care- country’.” fully, using a setup that nests both insurrections and The creation of the first railway line had to wait until party competition (unfortunately, such a setup does Francis’ death in 1835, yet even after that the govern- not currently exist). ment under Metternich kept to the same policies. It Finally, our account suggests that the relative se- was the revolution of 1848 that perturbed this stasis. curity of the British and German elites was impor- Eduard Marz places “the beginning of the industrial tant in the development trajectories of these countries. age in the 1850s ...when the Ancien regime ceased But the security was, at least in the case of Germany, to exist” (quoted in Blum 1948; Good 1984, 40–41). and also in part in the British case, the outcome of As in Russia after the Crimean, the response of the a coalition that the elites formed. Therefore, the ap- domestic elites was peasant emancipation and agrar- proach here needs to be complemented with a theory ian reform followed by a switch in economic policy of coalition-formation between different groups. We towards a rather vigorous promotion of industry (see view this as a very difficult, but exciting area for future Eddie 1989). research.

129 Economic Backwardness in Political Perspective February 2006

APPENDIX which defines A∗(R/h). This solves for the critical threshold of the state of technology as Derivation of the Value Functions Leading      P 1 + γµ − (α − 1) − P 1 + µ to Equation (10) A∗(R/h) =   2   2  1 + − 1 + − − P 2 µ α P 2 γµ (α 1) Here for completeness, we solve for the value functions of the   citizens and incumbent ruler. Using the conjecture, V(A) = 1 − βαP 1 + µ R v(x)A and
x = 1, ·  2  · , 1 − βP 1 + µ τh   2 0 v (x) A = A(1 − τ) + βv (x) A x α − z dz which is the expression in (16) in the text. Note that in 1 1 1 −µ + − − + 2 (21) we must have that P[ 2 γµ (α 1)] > P[ 2 µ]and  1 + µ α − 1 + γµ − α − >  1 +  − P[ 2 ] P[ 2 ( 1)] 0—–if the first inequality 2 γµ α 1 + (1 − x) dz + (α − z) dz . (19) did not hold, the incumbent would strictly prefer to innovate, − 1 − as discussed in the text, and if the second inequality did not α 1 2 γµ hold, the incumbent would strictly prefer to block. These ∗ The undetermined constant is found to be inequalities immediately imply that A (R/h) is increasing in R/h as claimed in the text. (1 − τ) v(x) =     Next, consider the situation where a ruler does not inno- 1 +γµ − − − 0 − − 2 + α 1 − vate. Then the recursion for the value function can be written 1 βx α 1 −µ z dz β(1 x) α−1 dz 1 −γµ(α z) dz 2 2 as For the ruler, the same type of argument leads to a value
= +
= = + +
= function, w(x 0)Ah r(x 0)R τAh R β(w(x 0)Ah    1 +   2 γµ 1 +µ 1 +γµ 2 2 +
= = + + − r(x 0)R) dz, w (x) A τA βw (x) A xα dz (1 x) dz . α−1 0 α−1 which is (20) τ
w (x = 0) =   Hence, − 1 + − − 1 βP 2 γµ (α 1) τ w(x) =      . and 1 − β xP 1 + µ α + (1 − x)P 1 + γµ − (α − 1) 1 2 2
r (x = 0) =   . − 1 + − − 1 βP 2 γµ (α 1) Value Functions Leading to Equation (15) Now using these expressions and condition (17), we have   First consider the situation where there is always innovation 1 ταAhN R in the future. Then, we have the following recursion defining P + µ   +   2 − βα 1 + µ − β 1 + µ
w(x = 1) and
r(x = 1): 1 P 2 1 P 2  
w(x = 1)Ah +
r(x = 1)R = τAh + R + β(
w(x = 1)αAh 1 τAhN = P + γµ − (α − 1)   2 − 1 + − −  1 + 1 βP 2 γµ (α 1) 2 µ +
r(x = 1)R) dz. 0 R +   , (22) 1 − βP 1 + γµ − (α − 1) The undetermined coefficients are obtained as 2

∗
= = τ  which solves for AN = A (R/h), establishing the claim in w (x 1) 1 1 − βαP + µ the text that the incumbent ruler will block whenever At < 2 ∗ ∗ A (R/h), and innovate whenever At ≥ A (R/h). and 1
r (x = 1) =  . − 1 + REFERENCES 1 βP 2 µ Now using these expressions, condition (15) can be rewritten Acemoglu, Daron. 2003. “Why Not a Political Coase Theorem?” as Journal of Comparative Economics 31 (4): 620–52. Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2002. “Reversal of Fortune: Geography and Institutions in the Making   of the Modern World Income Distribution.” Quarterly Journal of 1 ταAh∗ (R/h) R P + µ   +   Economics 117 (November): 1231–94. 2 − 1 + − 1 + Acemoglu, Daron, and James A. Robinson. 2000a. “Political Losers 1 βαP 2 µ 1 βP 2 µ as a Barrier to Economic Development.” American Economic   Review Papers and Proceedings 90 (May): 126–30. 1 τAh∗ (R/h) = P + γµ − (α − 1)   Acemoglu, Daron, and James A. Robinson. 2000b. “Why Did the 2 − 1 + West Extend the Franchise? Democracy, Inequality and Growth 1 βαP 2 µ in Historical Perspective.” Quarterly Journal of Economics 115 R (November): 1167–99. +   , (21) Acemoglu, Daron, and James A. Robinson. 2002. “Economic Back- − 1 + 1 βP 2 µ wardness in Political Perspective.” NBER Working Paper #8831.

130 American Political Science Review Vol. 100, No. 1

Acemoglu, Daron, and James A. Robinson. 2005. Economic Origins cations.” In Patterns of European Industrialization: The Nine- of Dictatorship and Democracy. New York: Cambridge University teenth Century. eds. Richard Sylla and Gianni Toniolo. New York: Press. Routledge. Alesina, Alberto, and Allan M. Drazen. 1991. “Why are Stabi- Gross, Nachaum. 1973. “The Industrial Revolution in the Habs- lizations Delayed?” American Economic Review 81 (December): burg Monarchy, 1750–1914.” In The Fontana Economic History of 1170–88. Europe. Vol. 4, ed. Carlo M. Cipolla. London: Fontana Arrow, Kenneth J. 1962. “The Economic Implications of Learning- Books. by-Doing.” Review of Economic Studies 29 (June): 155–73. Hellman, Joel S. 1998. “Winner Takes All: The Politics of Partial Re- Bairoch, Paul. 1982. “International Industrialization Levels from form in Postcommunist Transitions.” World Politics 50 (January): 1750 to 1980.” Journal of Economic History 11 (Fall): 269–333. 203–34. Berman, Sheri E. 2001. “Modernization in Historical Perspective: Hill, Christopher. 1981. The Century of Revolution,NewYork: The Case of Imperial Germany.” World Politics 53 (April): 431– Norton. 62. Krusell, Per, and Jose-V´ ıctor´ Rıos-Rull.´ 1996. “Vested Interests in a Blackbourn, David. 1997. The Long Nineteenth Century. London: Theory of Growth and Stagnation.” Review of Economic Studies Fontana. 63 (April): 301–29. Blum, Jerome. 1943. “Transportation and Industry in Austria, 1815– Kuznets, Simon. 1968. Towards a Theory of Economic Growth.New 1848.” Journal of Modern History 15 (March): 24–38. Haven: Yale University Press. Blum, Jerome. 1948. Noble Landowners and Agriculture in Austria, Macpherson, W. J. 1987. The Economic Development of Japan 1868– 1815–1848; A Study of the Origins of the Peasant Emancipation of 1941. New York: Cambridge University Press. 1848. Baltimore: Johns Hopkins University Press. Madison, James., Alexander Hamilton, and John Jay. 1987. The Fed- Bourguignon, Francois, and Thierry Verdier. 2000. “Oligarchy, eralist: A Collection of Essays Written in Favor of the New Consti- Democracy, Inequality and Growth.” Journal of Development tution. New York: Penguin. Economics 62 (August): 285–313. Maddison, Angus. 1995. Monitoring the World Economy 1820–1991. Bueno de Mesquita, Bruce, and Alastair Smith. 2004. “The Politi- Paris: OECD. cal Economy of Corporate Fraud.” Unpublished, Department of McDaniel, Timothy. 1988. Autocracy and Modernization in Russia. Politics, New York University. Princeton: Princeton University Press. Cain, Philip J., and Anthony G. Hopkins. 2000. British Imperialism. McDaniel, Timothy. 1991. Autocracy, Modernization and Revolution 2nd Edition. New York: Longman. in Russia and Iran. Princeton: Princeton University Press. Curtin Philip D. 2000. The World and the West: The European Chal- Mokyr, Joel. 1990. The Lever of Riches.NewYork:OxfordUniversity lenge and Overseas Response in the Age of Empire.NewYork: Press. Cambridge University Press. Moore, Barrington Jr. 1966. The Social Origins of Dictatorship and Eddie, Scott. 1989. “Economic Policy and Economic Development in Democracy. Boston: Beacon Press. Austria-Hungary, 1867–1913.” in Peter Mathias and Sidney Pollard Mosse, W. E. 1992. An Economic History of Russia, 1856–1914. eds. The Cambridge Economic History of Europe, Volume 8, New London: I.B. Taurus Press. York: Cambridge University Press. North, Douglass C., and Barry R. Weingast. 1989. “Constitutions Eley, Geoffrey. 1984. “The British Model and the German Road: and Commitment: The Evolution of Institutions Governing Public Rethinking the course of German History before 1914.” In The Choice in Seventeenth-Century England.” Journal of Economic Peculiarities of German History. eds. David Blackbourn and History 49 (December): 803–32. Geoffrey Eley. Oxford: Oxford University Press. Olson, Mancur C. 1982. The Rise and Decline of Nations. New Haven: Fernandez,´ Raquel, and Dani Rodrik. 1991. “Resistance to Reform: Yale University Press. Status Quo Bias in the Presence of Individual-Specific Uncer- Parente, Stephen L., and Edward C. Prescott. 1999. “Monopoly tainty.” American Economic Review 81 (December): 1146–55. Rights: A Barrier to Riches.” American Economic Review 89 Freudenberger, Herman. 1967. “State Intervention as an Obstacle (December): 1216—–33. to Economic Growth in the Habsburg Monarchy.” Journal of Eco- Portal, Roger. 1965. “The Industrialization of Russia.” In The Cam- nomic History 27 (December): 493–509. bridge Economic History of Europe.Vol.7,Part2,eds.Peter Gerschenkron, Alexander. 1943. Bread and Democracy in Germany. Mathias and M. M. Postan. New York: Cambridge University Berkeley: University of California Press. Press. Gerschenkron, Alexander. 1962. Economic Backwardness in Histor- Robinson, James A. 1997. “When is a State Predatory?” Unpub- ical Perspective. Cambridge: Harvard University Press. lished, Department of Government, Harvard University. Gerschenkron, Alexander. 1970. Europe in the Russian Mirror: Four Saville, John. 1994. The Consolidation of the Capitalist State: 1800– Lectures in Economic History. New York: Cambridge University 1850. London: Pluto Press. Press. Thompson, Edward P. 1975. Whigs and Hunters: The Origin of the Good, David F. 1984. The Economic Rise of the Habsburg Empire, Black Act. London: Allen Lane. 1750–1914. Berkeley: University of California Press. Tipton, Francis B. 1974. “Farm Labour and Power Politics: Germany Good, David F. 1991. “Austria-Hungary.” In Patterns of European 1850–1914.” Journal of Economic History 34 (December): 951–79. Industrialization: The Nineteenth Century. eds. Richard Sylla and Tracey, Michael. 1989. Government and Agriculture in Western Gianni Toniolo. New York: Routledge. Europe, 1880–1988. New York: New York University Press. Gregory, Paul R. 1991. “The Role of the State in Promoting Eco- Trebilcock, Clive. 1981. The Industrialization of the Continental nomic Development: The Russian Case and its General Impli- Powers. New York: Longman.

131