I haven't read all the readings yet, but wanted to make a comment regarding the Voigt paper. While I found his arguments compelling in several different ways, I found his "we should measure institutions as a package but not forget the role that each institutions plays" argument a bit muddled, and of course it plays well into the topics of this week's readings, endogeneity. While I see both sides of the coin here, that institutions influence other institutions, it is difficult to discern what we should be measuring if we try to account for every possible institution's influence on individual behavior as well as other, outside institutional behavior. It seems to me, at this stage in the game, that providing simplified models with good measurements would be a more appropriate way to examine institutions, and then perhaps more complex models can be built to later infer the effect different institutions have on each other.
Baby steps, essentially.
Also, the Kitschelt chapter did a wonderful job of tearing apart Fish's argument about superpresidentialism. As someone who has admired Fish's work, I enjoyed the critique of his argument. Additionaly, Kitschelt also has given me a lot to consider in terms of choosing data and variables...
To account for the variation in institutional development, many historical institutionalists, who embrace the notion of "path dependence," have argued that we need to reach the distant past to dig out the "deeper" factors. However, if the answers we need to resolve the variations that puzzle us lie in the distant past, how far should we go back into the history? Where can we find a proper "zero point?" The empirical studies we have this week more or less suggest that we trace back to a fixed variable beyond the concern of political scientists. For example, AJR utilized the mortalities of European settlers to explain the different paths of institutional evolution in former colonies, BL traced back to the implementation of land tenure systems to elucidate why some districts are more productive, and Kitschelt referred to the precommunist conditions to understand why the diverse patterns of regime changes in all post-Soviet states. When they a fixed point, they have no intention to explain what caused the variations in the mortality, land tenure system (which BL even argued had nothing to do with the characters of each district), and precommunist conditions, which should be the tasks of historians. For me, following their rationale, to explain why the Chinese Communist Party has successfully domesticated some ethnic minorities into the Chinese state while others remain rebellious, it may be possible to investigate their past cultural contact (e.g. national education) with the dominant Han in the pre-communist period.
However, this practice is surely contestable. Except for that the selection of the zero point can be arbitrary, we usually lack complete data for previous centuries, which means we may not be able to confirm our results in other contexts (this explains why European colonies are normally chosen as the cases). Second, the causal reasoning can be very problematic if we are overly obsessed with the so-called "deep" causes without clearly delineating the causal "path." Due to the lack of reliable data and measurement, it may be very hard for us to transform our "conjecture" into "theories." How can we avoid these problems? Finally, if we would like to reach the far past to explain the variation in institutional evolution, there may be a conflict between institutions being endogenized and institutions being the primary causes as what matters may not be institutions anymore but something lying behind it in the distant history. In this sense, institutions may not matter too much as they simply act as the proxy variables.
Another question I have in mind regards AJR's article. While they surely gave a compelling explanation about the causal relations between the European colonizers and current political and economic institutions, they seem to assume that all people under the imperial rule play no role during the process of institutional development and state-building. For example, compared with South Korea and Taiwan, who used to be under the control of Japan before the WWII, the domestic community in both countries actually play a important role in prompting the Japanese colonizers to take favorable policies and institutional framework (which is more significant in Taiwan as during the 1920s there was a strong political movement against the colonial rule with the request of more autonomy organized by local people). More crucially, when accounting for the democratic transition in Taiwan and South Korea during the 1980s, it seems like no scholar will attribute to the colonial period to explain the democratic development in both countries, which bring us back to the question that if it is really necessary to look back into the distant past.
In Acemoglu et al's paper on "The Colonial Origins of Comparative Development", their theory of how current institutions could affect economic performance is based on 3 premises, the last of which is that early institutions persisted even after independence and have a significant impact on current institutions. Although this premise seems intuitive and logical at first sight, I feel that it deserves closer scrutiny. The authors argue that there is "a close association between early institutions and institutions today". Yet their regression results show that "past institutions alone [only] explain 20% of the variation in the index of current institutions", as opposed to the much higher explanatory power of settlement patterns with respect to variation in early institutions (50%). Despite the statistical significance of their regression results, do you think they have shown substantive evidence that past institutions have had a large impact on current institutions? To what extent do you think these results support the authors' premise that path dependence and historical legacies are crucial for understanding institutional development and in turn, the impact of institutions on economic performance?
In addition, the authors provided 3 economic reasons to account for the persistence of colonial institutions after independence. In fact, there is a broader argument within the institutionalist literature which contends that “once institutions are in place, they reproduce themselves [and the conditions that gave rise to them], and block any change.” The logic is that “to develop new activities, political, educational and legal institutions would have to have changed so as to allow access of poor people to productive assets. Yet the established elite needed inegalitarian institutions to defend its privileges. Hence, this elite [would be] unwilling to change the institutions and without the institutional change the new development path could not be pursued.” (Przeworski) To what extent do you agree with the argument that institutional reform is dictated by the elites? Although the poor and uneducated might not be empowered through formal institutions, what of the de facto power they hold vis-à-vis the threat of revolutions, e.g. 1789 French Revolution? (Acemoglu and Robinson have also discussed this in their article “Why did the West extend the franchise? Democracy, Inequality and Growth in Historical Perspective”) Do institutions then persist until there is threat of a revolution, i.e. until the whole system collapses and is replaced by a new system?
Last but not least, within the spectrum of institutions (political, legal, economic etc), to what extent do you think educational reform is the most important for perpetuating a good set of institutions in a country? In other words, although different types of institutions are inextricably linked, are there any that are more important, more “primary” than others in the process of establishing and sustaining a good institutional environment?
Although I think education is an important institution for disseminating norms (and for encouraging critical thought, obviously), I think that Acemoglu's focus on property rights, particularly the measure of "fear of expropriation", may be the right focus for measuring institutions. If a government can ensure that it's self-constraint, as well as constrain other actors, one could assume that at least this one institution is functional. Moreover, if one's property is secure, then he would be free to act towards other pursuits.
If a hypothetical country had a great education system, but a not so great property rights protection system, then the education system would be worthless, because few would utilize the system to its full advantage. I think it was Pzeworski that pointed out behavioral outcomes need to be predictable in order for institutions to be "good." Property rights protection serves as a proxy for several other institutions, because it implies certain institutions function in order for property rights to be fully protected.
First, I feel the measurement flaw is one of the most legitimate critiques to existing econometric research on institution. But if one tries to follow the notion that the indicator of institution should capture lasting features of political environment, I wonder how this variable could be used to explain, or be explained by, volatile economic trajectory.
Secondly, for establishing a solid causal relationship, it is fine to date back to 100 (or 400) years to find a reasonable instrumental variable. However, the best outcome is just to show that institution is a “black box”, thus such approach, as Glasear et al (2004) points out, “does not tell us what causes growth”. Perhaps the more credible way to provide feasible advice on development is to deeply explore some concrete cases, including both successful and failing ones.
My understanding is a lot of studies have done to analyze why the elites extended franchise to the poor. By the looking at some European countries (e.g. Germany and the UK) during the 19th century, one very conventional argument is that the newly emerged middle class people voiced their need to have the chance to participate in the decision-making process. Acemoglu and his colleagues proposed another perspective by contending that democracy, with the broadened basis of constituencies, will help the rich secure their commitment and refrain the request of redistribution demanded by the poor or middle-class. Also, while revolution indeed plays some role in prompting the extension of franchise, some studies argue that elections can actually help the central government monitor the performance of local officials (so they can assure their accountability) and control the rampant patronage networks that used to prevail in the English parliament. Some also say elections can act as a useful mechanism of cooptation to seek support for economic reform (which is the case of Bismarck's Germany in the sense that he needs the support from the peasants, if I am not mistaken). You can even consider elections as a way for dictators to collect information from the subjects (G. Cox) or distribute patronage between the leaders and their alliances. Many similar arguments have been applied to investigate different cases in the 20th century.
Anyway, what I mention is just a simple overview about what has been said regarding this topic, even though I think it may not be very relevant to what we will discuss this week (maybe we can have more discussions on this during the week on autocracy and authoritarian regimes).
Kitschelt’s piece critiquing most proposed explanations of post-communist regime change as either too deep (lacking causal mechanisms) or too shallow (tautologies) does, as Kim mentions above, do a good job in presenting methodological problems that social scientists face when attempting to explore causality. In my opinion, however, Kitschelt’s critique of “shallow” explanations is flawed. Though his preference for “deep” explanations is clear, what is unclear is that the mechanisms he describes are actually “deep” or are in fact correlated with “shallower” mechanisms as well. In his critique of Roeder, Kitschelt claims that the critical variables are the “long-term practice of state building an the mobilization of secondary associations (interest groups, parties) before incorporation in the Soviet Union as well as the mode of incorporation into the Soviet Union” (69). Kitschelt classifies the states in question into four categories. To illustrate his theory, he claims that in fully democratic countries (the Baltics) there was a history of interwar democracy. Soviet response to this institutional legacy was a mixed strategy of accommodation and repression. As a result, when the Soviet Union collapsed (hardly an inconsequential exogenous shock!) citizens mobilized and “made indigenous and non-indigenous elites seek a regime transition through negotiation” (70). Though Kitschelt critiques Fish and others for lacking causal mechanisms I don’t see how this explanation provides a causal mechanism more plausible than those they provide. If geography, economic proximity/incentives, religion and other variables don’t provide a clear path of causality, then how does the variable that captures history of interwar democracy? Clearly, one can posit that the institutional legacy of democracy and foundation of rule of law based society impacted the cultural “tool-kit” available to elites seeking mobilization strategies, but how is this strategy deep if other institutional legacies are not? Additionally, several of Kitschelt’s comments in this section are puzzling. When describing the Baltics Kitschelt claims, “eventually even the nonindigenous population finds it advantageous to assimilate into the new independent, Western-oriented democracies” (70). While this phrase in itself is misleading because it is a gross understatement (the non-indigenous population could either assimilate or be kicked out!) it also undermines Kitschelt’s point. The fact that populations “preferred” to assimilate into Western-oriented democracies gives credence to the neighborhood and economic incentive variables that Kitschelt critiques earlier. Further, Kitschelt’s characterization of the Group 2/3 (he mixes up the typologies here-he claims in the summary that group 2 consists of semi-democracies but then describes that section as what he calls group 3 above) countries is as “odd” as he claims Roeder’s is. I do not understand his claim Belarus is a country where there is an “unstable oscillation between semi-democratic and oligarchic regime forms in an environment of volatile parties and interest associations” or his contention that “even elite fragments bent on authoritarianism, such as those that asserted themselves in Belarus in the late 1990s, are likely to encounter increasingly stiff resistance to the realization of their preferences” (70).
After reading the article by Grosjean and Senik, several questions come to mind. In a nutshell, the authors say that "building democratic institutions can play as an ingredient in favor of market liberalization, whereas early market development is no guarantee of a later popular support to democracy." (33) This preposition, stands in stark contrast with the now widely accepted notion that market development will lead to democratization due to the emergence of groups that will demand political spaces to advance their interests. Although the authors argue that the data does not support this idea, we should ask the following questions: 1) Modernization theory says that market liberalization will lead to democratization after an ever growing middle class emerges, however for the authors, it was difficult to find data supporting the existance of a consolidated middle class, which will in turn result in the question of how long should we wait to observe demands of democratization before we say the data does not support it? 2) There has to be a substantial gap in political power between the rich and the middle class for the latter to demand democratization however, there is also the need for political parties to be different in order for elections to matter. Therefore, are the authors jumping the gun by stating that liberalization does not appear to reinforce democratic values? 3)when they say that demcoratic institutions favor market liberalization, shouldn't we consider that for (functional) democratic institutions to develop in the first place it is necessary a minimum standard of living that usually happens after some degree of market liberalization took place? So in that way, the market (in a distributional sense) will precede some degree of democratization?
I want to get back to Emily's comment, which I think very good because she indicates what concerns me about the work of historical institutionalists: We may get the history wrong when we are trying to establish the causal path from a distant path as there may simply be too many interpretations available and we need to judge which part of history we would like to highlight as the key (independent) variables as the foundations of contemporary institutions. Even if we have a specific factor in our mind, establishing solid and credible path of causality is still a tricky issue and subject to further contestation.
However, even if his own explanations, by taking the "deeper" factors into account, may be flawed, his point that tracing back into history still holds. The question is, if clinging to the idea that "institutions matter" we would like to treat institutions as the primary independent variables that lead to the variation in the patterns of economic development, we still have to go beyond this point to explain where lies of the sources for different institutions coming into being.
By the way, if I am not mistaken, I believe Kitschelt deals with the classification of political transition of post-Soviet States in a paper. Maybe it will be useful to read that paper to see how Kitschelt established his arguments more.
I question the validity of Acemoglu et al.'s 2SLS analysis in their paper on "The Colonial Origins of Comparative Development." In some sense this is a technical criticism, but it also reflects a serious substantive problem regarding teh validity of their results.
Acemoglu et al. lay out the following model (pg 1370): settler mortality (M) affected colonial settlements (S), which in turn impacted early institutions (E). The impact of those early institutions are prevalent in current institutions (C) which affect current GDP (G). M -> S -> E -> C -> G. This is quite a complicated model with substantial areas of potential endogeneity. For their 2SLS model to be valid, their instrument (M) must only affect GDP (G) through current institutions (C) after controlling for additional covariates. They recognize this condition for the validity of 2SLS and try to deal with it, but in my view they fail.
Analyzing their data closely, a few things are apparent: (1) all of the aforementioned theoretical variables are correlated; (2) their actual 2SLS model in truth only estimates a model of M -> C -> G and ignores the mediating theoretical variables (S and E); and (3) their justification for the validity of the 2SLS (i.e. the exclusion restriction) centers on controlling for variables like disease, geography, and climate, but it ignores many other ways in which the mediating variables might affect current GDP.
Let me move away from technical jargon. It is unsurprising that current GDP is related to early settlements and institutions, but the authors try to make a causal argument based on the fact that early settler mortality will **only affect GDP through current institutions**. It is equally plausible that the true driver here is the fact that these colonies were actually settled (S), and everything that has happened since that time -- technological development, relationships with business in the colonizing countries, and any other economic factors that are related to settlement independent of institutions -- have a substantial impact on current GDP.
In fact, this interpretation is supported by their overidentification results (pg 1394). Every single instrument used in place of settler mortality yields significant coefficients in the second stage. Actually, I now notice that footnote 31 which describes these results mentions that, "There are in fact good reasons to expect institutions circa 1900 to have a direct effect on income today (and hence the overidentifying tests to reject our restrictions): these institutions should affect physical and human capital investments at the beginning of the century, and have some effect on current income levels through this channel." Why is this secluded to a footnote??
Perhaps I'm being too harsh here, after all I know how influential this paper has been. What do you guys think? I assume other people have made similar arguments given the paper's prestige (sorry, I am not as familiar with this literature as other people in the class).
Acemoglu et al make a cogent argument that the types of institutions put in place by the colonial powers have affected the types of institutions in place today. The article touches upon the larger debate of colonial legacy and whether or not a certain legacy is more conducive to the establishment of strong institutions: Coatsworth blames the Iberian legacy, which put emphasis on extracting resources, not institution building, for Latin America’s economic stagnation; Zakaria sees the British colonial legacy as providing a better foundation for the development of democracy because of its emphasis on institutional development (constitutionalism, the rule of law); etc.
What is consistently neglected—and what is mentioned by one entry above—are the domestic institutions that are already in place when colonialism enters the scene and how the endogenous interacts with the exogenous. A qualitative look around the Middle East provides a basic understanding of this: the exogenous institutions implemented by the Ottoman Empire blended with the already existing institutions (a consultative council in Mt. Lebanon; a structure of municipal and provincial government still in place today in much of the Arab world but with varying nuances due to local pre-existing institutions) and how these hybrid institutions were then affected and reshaped by colonialism. Hence, I’m not sure any model or argument that neglects pre-existing institutions is sound. I think Glaeser, La Porta et al say it best by writing, “Institutions…are points on this opportunity set determined by efficiency, history, and politics”.
For me, the theme that stand out of the readings is the relation between growth/development and institutions, and the methodological aspects of it. I would like to make three short comments.
1) I have not been convinced of the necessity and faisability of determining which of the economic growth/ development or institutions depend on the other one. I very much agree with the quotation of Mariscal and Sokoloff (2000: 198) by Przeworski: “when variables are mutually reinforcing or simultaneously determined, discerning what is exogenous and what is endogenous is not transparent.”
2) I see the Acemoglu, Johnson and Robinson’s article as innovative, but their chain-of-variables argument did not convince me. (In a certain way, I see it as an example of a lack of specification of the mechanisms that lead from the antecedents to the political consequence, that Kitschelt attributes to “excessively deep comparative-historical explanation”). The criticisms of Glaeser et al. (emphasizing the role of human capital) seem to make a lot of sense to me. As for my reading of the AJR article, on an empirical basis, I remain stunned by the mortality rate variable and the lack of context in order to sustain its use. Given that the data starts in 1815, and that at this time the colonization in the different colonies were not at the stage, how can we compare them? For example, the first winter French peoples have spent in Canada has been devastating, with a high mortality rate caused by the scurvy, but this was in the 16th century and Indigenous peoples quickly teached the Europeans how to survive in winter. How to account for similar learning and adaptative processes in other colonies if the analysis does not take into account at which stage of colonization/ adaptation the colonies compared were?
3) Kitschelt’s proposal is unclear to me… In which ways is its causal analysis more convincing than the ones he is criticizing?
I would like to comment on two papers that are linked because they are neoinstitutionalist in approach (AJR and Benerjee Iyer) and on the critique by Przeworski. AJR is an admirable paper. That being said, I want to raise two concerns regarding the key issue in their paper (settler mortality [SM] as an instrumental variable [IV]) and the operationalization of institutions. The latter is very dealt with by Glaeser et al. The former is as interesting as controversial. Does SM satisfy the exclusion restriction? What about other prior factor to SM such as disease environment, prior wealth or human capital. For example, could disease environment not only affect settler mortality but aslo GDP through population density and technology? Acknowledging that it is a tremendous paper, endogeneity problems between growth and institutions can exist, which makes the causal claim more complicated than just using a good IV.
In Benerjee and Iyer, the endogeneity problem can exist in the following way: can other factors like ease of conquest affect the timing and thus the taxation system in turn? Also, regarding the mechanisms that connect the taxation system (X) and product yields (Y) than just class conflict and investment? Probably. The degree of inequality and bureaucracy can be other important mechanisms.
Finally, just a couple of lines on Przeworski. Very lucid and clear. Main remark: "Neither perspective [new institutionalism and geography/factor endowments] offers a theory of institutional change". The answer: "Institutions and development are mutually endogenous and the most we can hope for is to identify their reciprocal impacts." Unfortunately, even if he is right, the main remark is more satisfying than the answer.
*Disclaimer: some of the ideas above are the result of a session of the Political Economy of Development class by Macartan Humphreys.
Przeworski -- to be quite honest, I was surprised at how much I liked it, because I'm usually not a huge fan of his work (possibly personal preference, but a lot of important truths are elided in the global models of development and modernization theory.)
The things I enjoyed are twofold:
1. I really admire and think it is deeply important to recognize and place one's work in the context of intellectual traditions that are deeply relevant but not au curant in political science. I think explaining the Marxist search for primacy as a failure that should inform thinking about neo-institutionalism was enlightening and merits discussion in class. I was also gratified to see Parsonian theory mentioned very briefly, as the efforts of modernization theory in the mid-20th century are less well-known in the field as I feel they should be. (Talcott Parsons is a seminal figure in the intellectual history of American political science and was specifically concerned with developing an overarching theory of social action that became the foundation of modernization theory -- I do not believe we have read him or discussed his impact in any of the department's 3 seminars that address political development.)
2. Life seems pretty endogenous to me. The search for primary causes and causal chains in social action can be fruitful, but may also be to some extent an artifice of the decision to turn from political "philosophy" to political "science". Obviously, I'm studying at an American university, so I think there's a benefit to be had from the rigors of a positivist hermeneutic, but an over-commitment to that line of thinking can lead to some rather silly ideas about the world; accepting endogeniety and its complexity seems to be a viable alternative in which the political "studier" attempts to accurately describe a complex phenomenon and give indications of its applicability to other problems, but does not try to make authoritative, hegemonic, or decisive claims about the nature of political dynamics in all of history (see North, Weingast, and the other guy if you think I'm being hyperbolic in my last statement).
Przeworski brings up a salient methodological point: There is nothing in the nature of institutions that make their causal efficacy “primary”. However, Przeworski, in my view, misrepresents North’s argument. First, he uses a different definition of institutions: Stating that “if different institutions generate different outomes, then one can stick any institutions into any historical conditions and expect that they would function in the same way as the have functioned elsewhere. Install independent judiciary, establish clear property rights, create independent central banks, and manna will fall from heaven” he writes in direct contradiction of North’s understanding of institutions as a set of both formal and informal norms and rules. In North’s view, enforcement characteristics of a set of formal rules vary greatly among countries; for instance, if property rights are not seen as legitimate by the population, their enforcement will be much more costly. For North, norms concerning property rights are part and parcel of the institutional set-up.
Second, Przeworski seems to conclude from the fact that institutions could be endogenous, i.e. that they could be caused by the same economic outcomes that they are then meant to explain, that they always in fact are endogenously determined by the forces of production. However, this question is not only a philosophical question of how we understand causality; it is also an empirical question. Przeworski quotes North and Thomas (1973), arguing that “new institutional arrangements will not be set up unless the private benefits of their creation promise to exceed the cost.” But North later came to realize that norms and ideology might play a role. Although norms and ideology might also be affected by the forces of production – as they indeed are in the classical Marxist understanding – there is no reason why this must be the case. A case in point could be the Protestant reformation in Europe. It would be hard to argue that the forces of production impacted greatly on Martin Luther or Philip Melanchthon. The many coincidences of fate that shaped the religious wars, alliances and battle outcomes that finally led to the Peace of Augsburg with its Cuius regio, eius religio, are also not easily seen as a consequence of the forces of production.
As seen by the Benerjee and Iyer paper on land rights in India, there are also cases where formal institutions indeed could be taken as exogenous. Much of the later literature on institutions seek out such cases to demonstrate the causal efficacy of institutions, though it might not be clear that they are in this case more “primary” than other causal factors.
And granted, in many other circumstances, institutions might have been causally shaped by forces of production. However, there is one other sense in which institutions could be seen as more “primary”. In North’s understanding of institutions as they develop in the 1980s and 1990s, an important point about informal institutions is that they change slowly – sometimes exceedingly slowly. That is, even if they have risen as a response to a certain set of factor endowments, they outlast those configurations and take on a life of their own. Slavery, segregation and attitudes towards the black population in the US could be one example.
I agree with Przeworski that there is nothing in the nature of institutions that in and of itself makes them causally primary. However, the fact that informal institutions, in particular, once thoroughly established, often seem to be resistant to abrupt change. It is the empirical robustness of such institutions that would make them seem “primary”: Factor endowments might change, formal institutions might change, but if informal institutions stay in place, economic outcomes might remain the same. This – sometimes too – pessimistic view on the possibility for economic change in poor countries underlies North’s answer went asked about whether he had any advice for Russia, immediately after the break-up of the Soviet Union: “Get a new history.”
I really like Yari's discussion regarding the issues of data and measurement for AJR's article, many of which are also addresses in depth by Glaeser et al. It is sharp but it is surely legitimate. But I think there might be way to think about these issues. First of all, there may be not a very good causal connection between the data and their theoretical claims. . But is there any way for us to improve it? It is problematic because their theory is flawed or because the data and the measurement are not good enough, which may be probably the best they could get when they were working on the project? If their theory is indeed arguable, as Glaeser et al. contended, then is there any better or more convincing alternative? At least for me I do not think the development argument is much better than the institutionalist view. Also, regarding the problem of over-identification, I think it might be because many political scientists tend to have the theory (or hypotheses) first and then look for the data that will support their claims in their own opinion. So to avoid this problem, maybe we should reverse the practical procedures?
I haven't read all the readings yet, but wanted to make a comment regarding the Voigt paper. While I found his arguments compelling in several different ways, I found his "we should measure institutions as a package but not forget the role that each institutions plays" argument a bit muddled, and of course it plays well into the topics of this week's readings, endogeneity. While I see both sides of the coin here, that institutions influence other institutions, it is difficult to discern what we should be measuring if we try to account for every possible institution's influence on individual behavior as well as other, outside institutional behavior. It seems to me, at this stage in the game, that providing simplified models with good measurements would be a more appropriate way to examine institutions, and then perhaps more complex models can be built to later infer the effect different institutions have on each other.
ReplyDeleteBaby steps, essentially.
Also, the Kitschelt chapter did a wonderful job of tearing apart Fish's argument about superpresidentialism. As someone who has admired Fish's work, I enjoyed the critique of his argument. Additionaly, Kitschelt also has given me a lot to consider in terms of choosing data and variables...
This comment has been removed by the author.
ReplyDeleteTo account for the variation in institutional development, many historical institutionalists, who embrace the notion of "path dependence," have argued that we need to reach the distant past to dig out the "deeper" factors. However, if the answers we need to resolve the variations that puzzle us lie in the distant past, how far should we go back into the history? Where can we find a proper "zero point?" The empirical studies we have this week more or less suggest that we trace back to a fixed variable beyond the concern of political scientists. For example, AJR utilized the mortalities of European settlers to explain the different paths of institutional evolution in former colonies, BL traced back to the implementation of land tenure systems to elucidate why some districts are more productive, and Kitschelt referred to the precommunist conditions to understand why the diverse patterns of regime changes in all post-Soviet states. When they a fixed point, they have no intention to explain what caused the variations in the mortality, land tenure system (which BL even argued had nothing to do with the characters of each district), and precommunist conditions, which should be the tasks of historians. For me, following their rationale, to explain why the Chinese Communist Party has successfully domesticated some ethnic minorities into the Chinese state while others remain rebellious, it may be possible to investigate their past cultural contact (e.g. national education) with the dominant Han in the pre-communist period.
ReplyDeleteHowever, this practice is surely contestable. Except for that the selection of the zero point can be arbitrary, we usually lack complete data for previous centuries, which means we may not be able to confirm our results in other contexts (this explains why European colonies are normally chosen as the cases). Second, the causal reasoning can be very problematic if we are overly obsessed with the so-called "deep" causes without clearly delineating the causal "path." Due to the lack of reliable data and measurement, it may be very hard for us to transform our "conjecture" into "theories." How can we avoid these problems? Finally, if we would like to reach the far past to explain the variation in institutional evolution, there may be a conflict between institutions being endogenized and institutions being the primary causes as what matters may not be institutions anymore but something lying behind it in the distant history. In this sense, institutions may not matter too much as they simply act as the proxy variables.
Another question I have in mind regards AJR's article. While they surely gave a compelling explanation about the causal relations between the European colonizers and current political and economic institutions, they seem to assume that all people under the imperial rule play no role during the process of institutional development and state-building. For example, compared with South Korea and Taiwan, who used to be under the control of Japan before the WWII, the domestic community in both countries actually play a important role in prompting the Japanese colonizers to take favorable policies and institutional framework (which is more significant in Taiwan as during the 1920s there was a strong political movement against the colonial rule with the request of more autonomy organized by local people). More crucially, when accounting for the democratic transition in Taiwan and South Korea during the 1980s, it seems like no scholar will attribute to the colonial period to explain the democratic development in both countries, which bring us back to the question that if it is really necessary to look back into the distant past.
ReplyDeleteIn Acemoglu et al's paper on "The Colonial Origins of Comparative Development", their theory of how current institutions could affect economic performance is based on 3 premises, the last of which is that early institutions persisted even after independence and have a significant impact on current institutions. Although this premise seems intuitive and logical at first sight, I feel that it deserves closer scrutiny. The authors argue that there is "a close association between early institutions and institutions today". Yet their regression results show that "past institutions alone [only] explain 20% of the variation in the index of current institutions", as opposed to the much higher explanatory power of settlement patterns with respect to variation in early institutions (50%). Despite the statistical significance of their regression results, do you think they have shown substantive evidence that past institutions have had a large impact on current institutions? To what extent do you think these results support the authors' premise that path dependence and historical legacies are crucial for understanding institutional development and in turn, the impact of institutions on economic performance?
ReplyDeleteIn addition, the authors provided 3 economic reasons to account for the persistence of colonial institutions after independence. In fact, there is a broader argument within the institutionalist literature which contends that “once institutions are in place, they reproduce themselves [and the conditions that gave rise to them], and block any change.” The logic is that “to develop new activities, political, educational and legal institutions would have to have changed so as to allow access of poor people to productive assets. Yet the established elite needed inegalitarian institutions to defend its privileges. Hence, this elite [would be] unwilling to change the institutions and without the institutional change the new development path could not be pursued.” (Przeworski) To what extent do you agree with the argument that institutional reform is dictated by the elites? Although the poor and uneducated might not be empowered through formal institutions, what of the de facto power they hold vis-à-vis the threat of revolutions, e.g. 1789 French Revolution? (Acemoglu and Robinson have also discussed this in their article “Why did the West extend the franchise? Democracy, Inequality and Growth in Historical Perspective”) Do institutions then persist until there is threat of a revolution, i.e. until the whole system collapses and is replaced by a new system?
Last but not least, within the spectrum of institutions (political, legal, economic etc), to what extent do you think educational reform is the most important for perpetuating a good set of institutions in a country? In other words, although different types of institutions are inextricably linked, are there any that are more important, more “primary” than others in the process of establishing and sustaining a good institutional environment?
This comment has been removed by the author.
ReplyDeleteAlthough I think education is an important institution for disseminating norms (and for encouraging critical thought, obviously), I think that Acemoglu's focus on property rights, particularly the measure of "fear of expropriation", may be the right focus for measuring institutions. If a government can ensure that it's self-constraint, as well as constrain other actors, one could assume that at least this one institution is functional. Moreover, if one's property is secure, then he would be free to act towards other pursuits.
ReplyDeleteIf a hypothetical country had a great education system, but a not so great property rights protection system, then the education system would be worthless, because few would utilize the system to its full advantage. I think it was Pzeworski that pointed out behavioral outcomes need to be predictable in order for institutions to be "good." Property rights protection serves as a proxy for several other institutions, because it implies certain institutions function in order for property rights to be fully protected.
First, I feel the measurement flaw is one of the most legitimate critiques to existing econometric research on institution. But if one tries to follow the notion that the indicator of institution should capture lasting features of political environment, I wonder how this variable could be used to explain, or be explained by, volatile economic trajectory.
ReplyDeleteSecondly, for establishing a solid causal relationship, it is fine to date back to 100 (or 400) years to find a reasonable instrumental variable. However, the best outcome is just to show that institution is a “black box”, thus such approach, as Glasear et al (2004) points out, “does not tell us what causes growth”. Perhaps the more credible way to provide feasible advice on development is to deeply explore some concrete cases, including both successful and failing ones.
My understanding is a lot of studies have done to analyze why the elites extended franchise to the poor. By the looking at some European countries (e.g. Germany and the UK) during the 19th century, one very conventional argument is that the newly emerged middle class people voiced their need to have the chance to participate in the decision-making process. Acemoglu and his colleagues proposed another perspective by contending that democracy, with the broadened basis of constituencies, will help the rich secure their commitment and refrain the request of redistribution demanded by the poor or middle-class. Also, while revolution indeed plays some role in prompting the extension of franchise, some studies argue that elections can actually help the central government monitor the performance of local officials (so they can assure their accountability) and control the rampant patronage networks that used to prevail in the English parliament. Some also say elections can act as a useful mechanism of cooptation to seek support for economic reform (which is the case of Bismarck's Germany in the sense that he needs the support from the peasants, if I am not mistaken). You can even consider elections as a way for dictators to collect information from the subjects (G. Cox) or distribute patronage between the leaders and their alliances. Many similar arguments have been applied to investigate different cases in the 20th century.
ReplyDeleteAnyway, what I mention is just a simple overview about what has been said regarding this topic, even though I think it may not be very relevant to what we will discuss this week (maybe we can have more discussions on this during the week on autocracy and authoritarian regimes).
Kitschelt’s piece critiquing most proposed explanations of post-communist regime change as either too deep (lacking causal mechanisms) or too shallow (tautologies) does, as Kim mentions above, do a good job in presenting methodological problems that social scientists face when attempting to explore causality. In my opinion, however, Kitschelt’s critique of “shallow” explanations is flawed. Though his preference for “deep” explanations is clear, what is unclear is that the mechanisms he describes are actually “deep” or are in fact correlated with “shallower” mechanisms as well. In his critique of Roeder, Kitschelt claims that the critical variables are the “long-term practice of state building an the mobilization of secondary associations (interest groups, parties) before incorporation in the Soviet Union as well as the mode of incorporation into the Soviet Union” (69). Kitschelt classifies the states in question into four categories. To illustrate his theory, he claims that in fully democratic countries (the Baltics) there was a history of interwar democracy. Soviet response to this institutional legacy was a mixed strategy of accommodation and repression. As a result, when the Soviet Union collapsed (hardly an inconsequential exogenous shock!) citizens mobilized and “made indigenous and non-indigenous elites seek a regime transition through negotiation” (70). Though Kitschelt critiques Fish and others for lacking causal mechanisms I don’t see how this explanation provides a causal mechanism more plausible than those they provide. If geography, economic proximity/incentives, religion and other variables don’t provide a clear path of causality, then how does the variable that captures history of interwar democracy? Clearly, one can posit that the institutional legacy of democracy and foundation of rule of law based society impacted the cultural “tool-kit” available to elites seeking mobilization strategies, but how is this strategy deep if other institutional legacies are not? Additionally, several of Kitschelt’s comments in this section are puzzling. When describing the Baltics Kitschelt claims, “eventually even the nonindigenous population finds it advantageous to assimilate into the new independent, Western-oriented democracies” (70). While this phrase in itself is misleading because it is a gross understatement (the non-indigenous population could either assimilate or be kicked out!) it also undermines Kitschelt’s point. The fact that populations “preferred” to assimilate into Western-oriented democracies gives credence to the neighborhood and economic incentive variables that Kitschelt critiques earlier. Further, Kitschelt’s characterization of the Group 2/3 (he mixes up the typologies here-he claims in the summary that group 2 consists of semi-democracies but then describes that section as what he calls group 3 above) countries is as “odd” as he claims Roeder’s is. I do not understand his claim Belarus is a country where there is an “unstable oscillation between semi-democratic and oligarchic regime forms in an environment of volatile parties and interest associations” or his contention that “even elite fragments bent on authoritarianism, such as those that asserted themselves in Belarus in the late 1990s, are likely to encounter increasingly stiff resistance to the realization of their preferences” (70).
ReplyDeleteAfter reading the article by Grosjean and Senik, several questions come to mind. In a nutshell, the authors say that "building democratic institutions can play as an ingredient in favor of market liberalization, whereas early market development is no guarantee of a later popular support to democracy." (33)
ReplyDeleteThis preposition, stands in stark contrast with the now widely accepted notion that market development will lead to democratization due to the emergence of groups that will demand political spaces to advance their interests. Although the authors argue that the data does not support this idea, we should ask the following questions:
1) Modernization theory says that market liberalization will lead to democratization after an ever growing middle class emerges, however for the authors, it was difficult to find data supporting the existance of a consolidated middle class, which will in turn result in the question of how long should we wait to observe demands of democratization before we say the data does not support it?
2) There has to be a substantial gap in political power between the rich and the middle class for the latter to demand democratization however, there is also the need for political parties to be different in order for elections to matter. Therefore, are the authors jumping the gun by stating that liberalization does not appear to reinforce democratic values?
3)when they say that demcoratic institutions favor market liberalization, shouldn't we consider that for (functional) democratic institutions to develop in the first place it is necessary a minimum standard of living that usually happens after some degree of market liberalization took place? So in that way, the market (in a distributional sense) will precede some degree of democratization?
This comment has been removed by the author.
ReplyDeleteI want to get back to Emily's comment, which I think very good because she indicates what concerns me about the work of historical institutionalists: We may get the history wrong when we are trying to establish the causal path from a distant path as there may simply be too many interpretations available and we need to judge which part of history we would like to highlight as the key (independent) variables as the foundations of contemporary institutions. Even if we have a specific factor in our mind, establishing solid and credible path of causality is still a tricky issue and subject to further contestation.
ReplyDeleteHowever, even if his own explanations, by taking the "deeper" factors into account, may be flawed, his point that tracing back into history still holds. The question is, if clinging to the idea that "institutions matter" we would like to treat institutions as the primary independent variables that lead to the variation in the patterns of economic development, we still have to go beyond this point to explain where lies of the sources for different institutions coming into being.
By the way, if I am not mistaken, I believe Kitschelt deals with the classification of political transition of post-Soviet States in a paper. Maybe it will be useful to read that paper to see how Kitschelt established his arguments more.
I question the validity of Acemoglu et al.'s 2SLS analysis in their paper on "The Colonial Origins of Comparative Development." In some sense this is a technical criticism, but it also reflects a serious substantive problem regarding teh validity of their results.
ReplyDeleteAcemoglu et al. lay out the following model (pg 1370): settler mortality (M) affected colonial settlements (S), which in turn impacted early institutions (E). The impact of those early institutions are prevalent in current institutions (C) which affect current GDP (G). M -> S -> E -> C -> G. This is quite a complicated model with substantial areas of potential endogeneity. For their 2SLS model to be valid, their instrument (M) must only affect GDP (G) through current institutions (C) after controlling for additional covariates. They recognize this condition for the validity of 2SLS and try to deal with it, but in my view they fail.
Analyzing their data closely, a few things are apparent: (1) all of the aforementioned theoretical variables are correlated; (2) their actual 2SLS model in truth only estimates a model of M -> C -> G and ignores the mediating theoretical variables (S and E); and (3) their justification for the validity of the 2SLS (i.e. the exclusion restriction) centers on controlling for variables like disease, geography, and climate, but it ignores many other ways in which the mediating variables might affect current GDP.
Let me move away from technical jargon. It is unsurprising that current GDP is related to early settlements and institutions, but the authors try to make a causal argument based on the fact that early settler mortality will **only affect GDP through current institutions**. It is equally plausible that the true driver here is the fact that these colonies were actually settled (S), and everything that has happened since that time -- technological development, relationships with business in the colonizing countries, and any other economic factors that are related to settlement independent of institutions -- have a substantial impact on current GDP.
In fact, this interpretation is supported by their overidentification results (pg 1394). Every single instrument used in place of settler mortality yields significant coefficients in the second stage. Actually, I now notice that footnote 31 which describes these results mentions that, "There are in fact good reasons to expect institutions circa 1900 to have a direct effect on income today (and hence the overidentifying tests to reject our restrictions): these institutions should affect physical and human capital investments at the beginning of the century, and have some effect on current income levels through this channel." Why is this secluded to a footnote??
Perhaps I'm being too harsh here, after all I know how influential this paper has been. What do you guys think? I assume other people have made similar arguments given the paper's prestige (sorry, I am not as familiar with this literature as other people in the class).
Acemoglu et al make a cogent argument that the types of institutions put in place by the colonial powers have affected the types of institutions in place today. The article touches upon the larger debate of colonial legacy and whether or not a certain legacy is more conducive to the establishment of strong institutions: Coatsworth blames the Iberian legacy, which put emphasis on extracting resources, not institution building, for Latin America’s economic stagnation; Zakaria sees the British colonial legacy as providing a better foundation for the development of democracy because of its emphasis on institutional development (constitutionalism, the rule of law); etc.
ReplyDeleteWhat is consistently neglected—and what is mentioned by one entry above—are the domestic institutions that are already in place when colonialism enters the scene and how the endogenous interacts with the exogenous. A qualitative look around the Middle East provides a basic understanding of this: the exogenous institutions implemented by the Ottoman Empire blended with the already existing institutions (a consultative council in Mt. Lebanon; a structure of municipal and provincial government still in place today in much of the Arab world but with varying nuances due to local pre-existing institutions) and how these hybrid institutions were then affected and reshaped by colonialism. Hence, I’m not sure any model or argument that neglects pre-existing institutions is sound. I think Glaeser, La Porta et al say it best by writing, “Institutions…are points on this opportunity set determined by efficiency, history, and politics”.
For me, the theme that stand out of the readings is the relation between growth/development and institutions, and the methodological aspects of it. I would like to make three short comments.
ReplyDelete1) I have not been convinced of the necessity and faisability of determining which of the economic growth/ development or institutions depend on the other one. I very much agree with the quotation of Mariscal and Sokoloff (2000: 198) by Przeworski: “when variables are mutually reinforcing or simultaneously determined, discerning what is exogenous and what is endogenous is not transparent.”
2) I see the Acemoglu, Johnson and Robinson’s article as innovative, but their chain-of-variables argument did not convince me. (In a certain way, I see it as an example of a lack of specification of the mechanisms that lead from the antecedents to the political consequence, that Kitschelt attributes to “excessively deep comparative-historical explanation”). The criticisms of Glaeser et al. (emphasizing the role of human capital) seem to make a lot of sense to me. As for my reading of the AJR article, on an empirical basis, I remain stunned by the mortality rate variable and the lack of context in order to sustain its use. Given that the data starts in 1815, and that at this time the colonization in the different colonies were not at the stage, how can we compare them? For example, the first winter French peoples have spent in Canada has been devastating, with a high mortality rate caused by the scurvy, but this was in the 16th century and Indigenous peoples quickly teached the Europeans how to survive in winter. How to account for similar learning and adaptative processes in other colonies if the analysis does not take into account at which stage of colonization/ adaptation the colonies compared were?
3) Kitschelt’s proposal is unclear to me… In which ways is its causal analysis more convincing than the ones he is criticizing?
I would like to comment on two papers that are linked because they are neoinstitutionalist in approach (AJR and Benerjee Iyer) and on the critique by Przeworski. AJR is an admirable paper. That being said, I want to raise two concerns regarding the key issue in their paper (settler mortality [SM] as an instrumental variable [IV]) and the operationalization of institutions. The latter is very dealt with by Glaeser et al. The former is as interesting as controversial. Does SM satisfy the exclusion restriction? What about other prior factor to SM such as disease environment, prior wealth or human capital. For example, could disease environment not only affect settler mortality but aslo GDP through population density and technology? Acknowledging that it is a tremendous paper, endogeneity problems between growth and institutions can exist, which makes the causal claim more complicated than just using a good IV.
ReplyDeleteIn Benerjee and Iyer, the endogeneity problem can exist in the following way: can other factors like ease of conquest affect the timing and thus the taxation system in turn? Also, regarding the mechanisms that connect the taxation system (X) and product yields (Y) than just class conflict and investment? Probably. The degree of inequality and bureaucracy can be other important mechanisms.
Finally, just a couple of lines on Przeworski. Very lucid and clear. Main remark: "Neither perspective [new institutionalism and geography/factor endowments] offers a theory of institutional change". The answer: "Institutions and development are mutually endogenous and the most we can hope for is to identify their reciprocal impacts." Unfortunately, even if he is right, the main remark is more satisfying than the answer.
*Disclaimer: some of the ideas above are the result of a session of the Political Economy of Development class by Macartan Humphreys.
Przeworski -- to be quite honest, I was surprised at how much I liked it, because I'm usually not a huge fan of his work (possibly personal preference, but a lot of important truths are elided in the global models of development and modernization theory.)
ReplyDeleteThe things I enjoyed are twofold:
1. I really admire and think it is deeply important to recognize and place one's work in the context of intellectual traditions that are deeply relevant but not au curant in political science. I think explaining the Marxist search for primacy as a failure that should inform thinking about neo-institutionalism was enlightening and merits discussion in class. I was also gratified to see Parsonian theory mentioned very briefly, as the efforts of modernization theory in the mid-20th century are less well-known in the field as I feel they should be. (Talcott Parsons is a seminal figure in the intellectual history of American political science and was specifically concerned with developing an overarching theory of social action that became the foundation of modernization theory -- I do not believe we have read him or discussed his impact in any of the department's 3 seminars that address political development.)
2. Life seems pretty endogenous to me. The search for primary causes and causal chains in social action can be fruitful, but may also be to some extent an artifice of the decision to turn from political "philosophy" to political "science". Obviously, I'm studying at an American university, so I think there's a benefit to be had from the rigors of a positivist hermeneutic, but an over-commitment to that line of thinking can lead to some rather silly ideas about the world; accepting endogeniety and its complexity seems to be a viable alternative in which the political "studier" attempts to accurately describe a complex phenomenon and give indications of its applicability to other problems, but does not try to make authoritative, hegemonic, or decisive claims about the nature of political dynamics in all of history (see North, Weingast, and the other guy if you think I'm being hyperbolic in my last statement).
Przeworski brings up a salient methodological point: There is nothing in the nature of institutions that make their causal efficacy “primary”. However, Przeworski, in my view, misrepresents North’s argument. First, he uses a different definition of institutions: Stating that “if different institutions generate different outomes, then one can stick any institutions into any historical conditions and expect that they would function in the same way as the have functioned elsewhere. Install independent judiciary, establish clear property rights, create independent central banks, and manna will fall from heaven” he writes in direct contradiction of North’s understanding of institutions as a set of both formal and informal norms and rules. In North’s view, enforcement characteristics of a set of formal rules vary greatly among countries; for instance, if property rights are not seen as legitimate by the population, their enforcement will be much more costly. For North, norms concerning property rights are part and parcel of the institutional set-up.
ReplyDeleteSecond, Przeworski seems to conclude from the fact that institutions could be endogenous, i.e. that they could be caused by the same economic outcomes that they are then meant to explain, that they always in fact are endogenously determined by the forces of production. However, this question is not only a philosophical question of how we understand causality; it is also an empirical question. Przeworski quotes North and Thomas (1973), arguing that “new institutional arrangements will not be set up unless the private benefits of their creation promise to exceed the cost.” But North later came to realize that norms and ideology might play a role. Although norms and ideology might also be affected by the forces of production – as they indeed are in the classical Marxist understanding – there is no reason why this must be the case. A case in point could be the Protestant reformation in Europe. It would be hard to argue that the forces of production impacted greatly on Martin Luther or Philip Melanchthon. The many coincidences of fate that shaped the religious wars, alliances and battle outcomes that finally led to the Peace of Augsburg with its Cuius regio, eius religio, are also not easily seen as a consequence of the forces of production.
As seen by the Benerjee and Iyer paper on land rights in India, there are also cases where formal institutions indeed could be taken as exogenous. Much of the later literature on institutions seek out such cases to demonstrate the causal efficacy of institutions, though it might not be clear that they are in this case more “primary” than other causal factors.
And granted, in many other circumstances, institutions might have been causally shaped by forces of production. However, there is one other sense in which institutions could be seen as more “primary”. In North’s understanding of institutions as they develop in the 1980s and 1990s, an important point about informal institutions is that they change slowly – sometimes exceedingly slowly. That is, even if they have risen as a response to a certain set of factor endowments, they outlast those configurations and take on a life of their own. Slavery, segregation and attitudes towards the black population in the US could be one example.
I agree with Przeworski that there is nothing in the nature of institutions that in and of itself makes them causally primary. However, the fact that informal institutions, in particular, once thoroughly established, often seem to be resistant to abrupt change. It is the empirical robustness of such institutions that would make them seem “primary”: Factor endowments might change, formal institutions might change, but if informal institutions stay in place, economic outcomes might remain the same. This – sometimes too – pessimistic view on the possibility for economic change in poor countries underlies North’s answer went asked about whether he had any advice for Russia, immediately after the break-up of the Soviet Union: “Get a new history.”
I really like Yari's discussion regarding the issues of data and measurement for AJR's article, many of which are also addresses in depth by Glaeser et al. It is sharp but it is surely legitimate. But I think there might be way to think about these issues. First of all, there may be not a very good causal connection between the data and their theoretical claims. . But is there any way for us to improve it? It is problematic because their theory is flawed or because the data and the measurement are not good enough, which may be probably the best they could get when they were working on the project? If their theory is indeed arguable, as Glaeser et al. contended, then is there any better or more convincing alternative? At least for me I do not think the development argument is much better than the institutionalist view. Also, regarding the problem of over-identification, I think it might be because many political scientists tend to have the theory (or hypotheses) first and then look for the data that will support their claims in their own opinion. So to avoid this problem, maybe we should reverse the practical procedures?
ReplyDelete