Authors: Dr. James Robinson and Dr. Pablo Querubin
In this paper we investigate the institutional determinants of regional inequality in the Philippines. We emphasize how political institutions shape regional inequality through various channels: the dominance of political dynasties which leads to a lack of accountability and a non-pluralistic society; the weakness and absence of the national state which mani- fests itself in violence and warlordism. This weakness creates and is sustained by problems of illegitimacy in parts of the country, such as Mindanao, which are distinct culturally and historically have only been incorporated relatively recently into the polity. We also show that land inequality plays a relatively separate role in determining poverty and public policy with higher land inequality be strongly associated with better development and public goods out- comes. We argue however, that this likely captures the indirect impact of land inequality working via the political system in a situation with political institutions which are extractive in many ways.
Authors: Dr. James Robinson and Dr. Pablo Querubin
In this paper we investigate the institutional determinants of regional inequality in the Philippines. We emphasize how political institutions shape regional inequality through various channels: the dominance of political dynasties which leads to a lack of accountability and a non-pluralistic society; the weakness and absence of the national state which mani- fests itself in violence and warlordism. This weakness creates and is sustained by problems of illegitimacy in parts of the country, such as Mindanao, which are distinct culturally and historically have only been incorporated relatively recently into the polity. We also show that land inequality plays a relatively separate role in determining poverty and public policy with higher land inequality be strongly associated with better development and public goods out- comes. We argue however, that this likely captures the indirect impact of land inequality working via the political system in a situation with political institutions which are extractive in many ways.
Authors: Dr. James Robinson and Dr. Pablo Querubin
In this paper we investigate the institutional determinants of regional inequality in the Philippines. We emphasize how political institutions shape regional inequality through various channels: the dominance of political dynasties which leads to a lack of accountability and a non-pluralistic society; the weakness and absence of the national state which mani- fests itself in violence and warlordism. This weakness creates and is sustained by problems of illegitimacy in parts of the country, such as Mindanao, which are distinct culturally and historically have only been incorporated relatively recently into the polity. We also show that land inequality plays a relatively separate role in determining poverty and public policy with higher land inequality be strongly associated with better development and public goods out- comes. We argue however, that this likely captures the indirect impact of land inequality working via the political system in a situation with political institutions which are extractive in many ways.
Angara Center Lecture Series, University of the Philippines
Pablo Querubn James Robinson NYU and Harvard August 20, 2014 Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 1 / 20 Why Nations Fail Poor economic outcomes are the result of extractive economic institutions that fail to create incentives and opportunities for the vast mass of people. Extractive economic institutions are a consequence of extractive political institutions which have two dimensions 1 Narrow distribution of political power (lack of pluralism). 2 Weak and ineective state (lack of political centralization). Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 2 / 20 Why Regions Fail Though the focus in our book is very much on national outcomes there is a lot of discussion of regional variation US South versus the rest of the US: South historically had more extractive economic institutions Slavery, far less public good provision, with the obvious results: 50% of national GDP per-capita in 1860, 40% in 1940, higher inequality, less manufacturing industry, urbanization, less innovative even in the economic activities in which it specialized. A consequence of more extractive political institutions White males got the vote far later in the South in 19th century, after Civil War the solid south absence of political competition, black disenfranchisement until the 1960s. Breakdown of extractive political and economic institutions in the South coincided with it converging to the US mean. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 3 / 20 Spatial Distribution of Extractive and Inclusive The US South example makes the point that even within a country with basically inclusive institutions there can be extractive institutions as well and they may be regionally concentrated creating spatial/regional inequality. In this case where the extractive institutions are all bad things go together creating underdevelopment. Is the Philippines like this? Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 4 / 20 Regional Development Patterns in the Philippines Large regional variation in levels of income per-capita and poverty. But one basic fact is that Mindanao is substantially poorer than the rest of the country. The poverty rate is signicantly lower in Luzon than in Mindanao. The poverty rate goes from 70% in Lanao del Sur (in Mindanao) to 4% in Metro Manilla. Income per-capita in 2006 dollars ranges from US$942 Tawi-Tawi (in Mindanao) to US$5,101. Poverty and low income clustered in Mindanao, lack of poverty and high income in Luzon. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 5 / 20 LuZCn MlnuAnAC vlSA?AS overLy lncldence, 2009 23.38 44.90 37.83 lncome per CaplLa (uS$), 2006 2430.82 1710.37 1964.30 uynasLy Perflndahl, 1988-2010 0.46 0.33 0.47 Share 8lggesL uynasLy, 1988-2010 0.36 0.62 0.37 ConfllcL Lplsodes, 2006 1.64 9.40 2.44 Share Wlld opulaLlon, 1903 0.18 0.36 0.02 lracLlon Musllm opulaLlon, 2000 0.02 0.23 0.01 lracLlon of Mayors who are "uaLus" 0.00 0.02 0.00 Access Lo LlecLrlclLy 0.63 0.41 0.49 Access Lo lped WaLer 0.39 0.36 0.39 Carbage CollecLlon by 1ruck 0.13 0.08 0.08 LlemenLary School LnrollmenL 0.84 0.73 0.79 8oad uenslLy 0.38 0.20 0.26 Cood Covernance lndex, 2008 187.00 148.47 147.90 Cverall Land Clnl, 2002 0.70 0.64 0.69 1able 1 Summary SLaLlsLlcs 0 2 0 4 0 6 0 8 0 P o v e r t y
I n c i d e n c e ,
2 0 1 2 0 20 40 60 80 100 Poverty Ranking LUZON VISAYAS MINDANAO Poverty Ranking (2012) Figure 1 1 0 0 0 2 0 0 0 3 0 0 0 4 0 0 0 5 0 0 0 I n c o m e
p e r
C a p i t a
( U S ) ,
2 0 0 6 0 50 100 Income per Capita Ranking LUZON VISAYAS MINDANAO Income per Capita Ranking (in USD), 2006 Figure 2 What Explains Variation in Poverty? Argument 1: There is Something Special about Mindanao (or Luzon) We present evidence that poverty is signicantly negatively correlated with 1 with the presence of political dynasties at the municipal level (coecient estimates for congress and governor similar but noisier and insignicant). 2 with the presence of violent incidents by non-state armed actors (NPA and MILF) (state absence - lack of monopoly of violence) 3 with the historical incidence of wild tribes from the 1903 US Census (which we also interpret as colonial state absence) and fraction of population muslim in 2000. But adding Mindanao and Luzon dummies reduces the size of the coecient of (1) by half, switches the sign of (2) and makes (3) insignicant. Thus many bad things go together in Mindanao (and good things go together in Luzon) and the sum of their impacts on poverty is quantitatively large. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 6 / 20 What Explains Variation in Poverty? Argument 2: The Puzzling Role of Land Inequality We present evidence that poverty is signicantly negatively correlated with land inequality. The quantitative eect is very large. We show that higher land inequality is associated with several important channels, higher literacy and educational attainment, and higher provision of public goods measured by access to potable water, road intensity. This is true inside and outside Mindanao and land inequality is not signicantly dierent on average in Mindanao. This within-country ndings mirrors something we found in joint work with Daron Acemoglu and Mara Angelica Bautista 5 years ago for Colombia (Economic and Political Inequality in Development: The Case of Cundinamarca, Colombia) (and has been since discovered by others - e.g. Claudio Ferraz and Fred Finan in Brazil). Very dierent from the US case. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 7 / 20 Whats Going on with Land Inequality? We propose three interpretations of this partially building on our earlier work 1 Economic and political elites are distinct. High land inequality indicates the presence of strong economic elites who can place de facto constraints on the activities of political elites which is good for public good provision (hypothesis, corruption should be lower in places with high land inequality). 2 The relationship is picking up an confounding variable, state presence. Big landed estates are only viable where the state can enforce property rights and build infrastructure etc. So extractive elements (coercive labor relations in sugarcane) are dominated by the positive eects of state presence (being neglected is worse than being exploited?). 3 In such a patrimonial society, economic elites are heavily involved in service provision. In the Philippines (like Cundinamarca) land inequality and political concentration are negatively correlated. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 8 / 20 Correlates of Poverty Reduced Form Evidence Political dynasties measured between 1988 and 2010 by 1 A Herndahl index of concentration of seats at dierent levels of government in the hands of a single family (e.g. 8 elections for mayor in this period, if all mayors are from the same family H = 1, if all mayors were from a dierent family H = 0.125). 2 Fraction of seats held by the family with the largest number of seats (also ranges from 0.125 to 1). Conict - number of violent episodes coded from newspaper reports between November 2005 and October 2006. Wild tribes in 1903 proportion of the total population classied as Wild (only at the provincial level). US made a distinction between civilized and barbarian with civilized being basically those Christianized, wild was everyone else. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 9 / 20 (1) (2) (3) (4) (5) (6) Dynasty Herfindahl, 1988-2010 9.586 4.294 4.454 (1.877) (1.545) (1.105) Share Biggest Dynasty, 1988-2010 10.193 4.829 4.821 (1.950) (1.603) (1.138) Mindanao Dummy 6.863 6.871 (0.884) (0.883) Luzon Dummy -14.201 -14.182 (0.791) (0.791) Province Fixed Effects NO NO NO NO YES YES Observations 1,611 1,611 1,611 1,611 1,611 1,611 R-squared 0.016 0.017 0.347 0.348 0.724 0.725 Dependent Variable is Poverty Incidence, 2009 Table 2 Poverty Incidence and Political Dynasties (Municipal-Level Regressions) (1) (2) (3) Conflict Episodes, 2006 0.391 -0.275 -0.309 (0.063) (0.061) (0.072) Mindanao Dummy 9.273 (1.007) Luzon Dummy -13.756 (0.838) Province Fixed Effects NO NO YES Observations 1,427 1,427 1,427 R-squared 0.026 0.334 0.715 Dependent Variable is Poverty Incidence, 2009 Poverty Incidence and Conflict (Municipal-Level Regressions) Table 3 (1) (2) (3) (4) (5) Share Wild Population, 1903 9.117 4.807 (3.848) (4.436) Fraction of Population Muslim, 2000 16.046 -0.732 23.068 (1.991) (1.881) (3.219) Mindanao Dummy 1.894 6.965 (4.445) (0.989) Luzon Dummy -9.050 -14.366 (3.440) (0.794) Observations 77 77 1,560 1,560 1,560 R-squared 0.070 0.223 0.040 0.340 0.737 Dependent Variable is Poverty Incidence, 2012 Dependent Variable is Poverty Incidence, 2009 Table 4 Poverty Incidence, Indigenous Population and Fraction Muslim (Provincial-Level Regressions) (1) (2) Fraction of Population Muslim, 2000 0.068 0.091 (0.010) (0.023) Province Fixed Effects NO YES Observations 413 413 R-squared 0.105 0.411 Table 5 Share of Datu Mayors and Percent Population Muslim (Municipal-Level Regressions) Dependent Variable is Share Datu Mayors, 1988-2010 Only for Mindanao (1) (2) (3) Dynasty Herfindahl, 1988-2010 4.477 3.236 3.383 (2.050) (1.710) (1.198) Conflict Episodes, 2006 0.241 -0.285 -0.322 (0.067) (0.061) (0.070) Fraction of Population Muslim, 2000 12.485 -0.238 22.339 (2.178) (1.957) (3.436) Luzon Dummy -13.922 (0.837) Mindanao Dummy 9.329 (1.073) Province Fixed Effects NO NO YES Observations 1,406 1,406 1,406 R-squared 0.052 0.342 0.732 Table 6 Reduced Form Determinants of Poverty (Municipal-Level Regressions) Dependent Variable is Poverty Incidence, 2009 Correlates of Poverty Potential Channels We now present evidence through which these variables may be inuencing poverty. We use public goods outcomes from the 2000 Philippine Census. Fraction of households that use electricity as a source of lighting. Fraction of households with faucet piped water. Fraction of households with garbage collected by a truck. Primary school enrollment (fraction of children aged between 6 and 12 attending school). Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 10 / 20 Electricity Piped Water Garb. Truck School Atten. Good Governance Index (1) (2) (3) (4) (5) Dynasty Herfindahl, 1988-2010 -0.077 -0.174 -0.052 -0.048 -25.358 (0.031) (0.031) (0.021) (0.011) (11.606) Observations 1,580 1,580 1,580 1,580 1,464 R-squared 0.004 0.019 0.004 0.011 0.003 Dynasty Herfindahl, 1988-2010 -0.030 -0.167 -0.037 -0.030 -17.205 (0.028) (0.031) (0.021) (0.010) (11.438) Mindanao Dummy -0.086 -0.034 -0.009 -0.043 1.345 (0.016) (0.017) (0.012) (0.006) (6.390) Luzon Dummy 0.167 -0.004 0.078 0.048 38.139 (0.014) (0.015) (0.010) (0.005) (5.636) Observations 1,580 1,580 1,580 1,580 1,464 R-squared 0.194 0.022 0.061 0.177 0.046 Table 7 Political Dynasties and Public Goods Provision (Municipal-Level Regressions) Dependent Variable is: Panel A: Simple Bivariate Regression Panel B: With Mindanao, Luzon Fixed Effects Electricity Piped Water Garb. Truck School Atten. Good Governance Index (1) (2) (3) (4) (5) Conflict Episodes, 2006 -0.007 -0.004 -0.001 -0.003 -1.550 (0.001) (0.001) (0.001) (0.000) (0.355) Observations 1,414 1,414 1,414 1,414 1,309 R-squared 0.039 0.011 0.002 0.057 0.014 Conflict Episodes, 2006 0.000 -0.003 0.002 -0.001 -0.959 (0.001) (0.001) (0.001) (0.000) (0.407) Mindanao Dummy -0.094 -0.016 -0.026 -0.040 7.009 (0.017) (0.019) (0.013) (0.006) (6.913) Luzon Dummy 0.188 0.020 0.095 0.051 35.581 (0.014) (0.016) (0.011) (0.005) (5.691) Observations 1,414 1,414 1,414 1,414 1,309 R-squared 0.233 0.014 0.084 0.206 0.047 Table 8 Conflict and Public Goods Provision (Municipal-Level Regressions) Dependent Variable is: Panel A: Simple Bivariate Regression Panel B: With Mindanao, Luzon Fixed Effects Electricity Piped Water Garb. Truck School Atten. Good Governance Index (1) (2) (3) (4) (5) Fraction of Population Muslim, 2000 -0.360 -0.317 -0.133 -0.173 -58.705 (0.031) (0.032) (0.022) (0.011) (11.578) Observations 1,550 1,550 1,550 1,550 1,435 R-squared 0.079 0.060 0.023 0.136 0.018 Fraction of Population Muslim, 2000 -0.207 -0.365 -0.093 -0.125 -44.099 (0.033) (0.036) (0.024) (0.012) (13.096) Mindanao Dummy -0.040 0.043 0.010 -0.015 10.119 (0.017) (0.019) (0.013) (0.006) (7.125) Luzon Dummy 0.171 -0.004 0.079 0.050 38.489 (0.014) (0.015) (0.010) (0.005) (5.636) Observations 1,550 1,550 1,550 1,550 1,435 R-squared 0.213 0.065 0.067 0.227 0.053 Table 9 Fraction Muslim and Public Goods Provision (Municipal-Level Regressions) Dependent Variable is: Panel A: Simple Bivariate Regression Panel B: With Mindanao, Luzon Fixed Effects Land Inequality and Poverty Reduced Form Evidence Now we present the reduced form results with land inequality and poverty. Land inequality measured from 2002 Agricultural Census at the provincial level imputing fraction of landless households. This is not a Mindanao versus Luzon phenomenon (same relationship is signicant using just Luzon or just Mindanao data). Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 11 / 20 Land Inequality and Poverty Potential Channels Same as before from the 2002 Census plus: Road density (Department of Public Works and Highways). Functional illiteracy (Human Development Report 2006). Note the connection between land inequality and public good provision - consistent with both the above stories: economic elites constrain politicians, land inequality correlated with state presence. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 12 / 20 0 2 0 4 0 6 0 8 0 .4 .5 .6 .7 .8 .9 Land Gini LUZON VISAYAS MINDANAO Poverty Incidence (2012) and Land Inequality Figure 3 (1) (2) (3) (4) (5) Overall Land Gini, 2002 -48.714 -35.493 -28.995 -6.310 -60.588 (14.596) (13.771) (19.619) (38.207) (22.017) Mindanao Dummy 4.435 (3.745) Luzon Dummy -8.032 (3.400) Sample All Provinces All Provinces Luzon Visayas Mindanao Observations 78 78 38 16 24 R-squared 0.128 0.295 0.057 0.002 0.256 Table 10 Poverty Incidence and Land Inequality (Provincial-Level Regressions) Dependent Variable is Poverty Incidence, 2012 Road Density Electricity Piped Water Garb. Truck School Atten. Func. Illiteracy (1) (2) (3) (4) (5) (6) Overall Land Gini, 2002 0.595 0.158 0.402 0.592 0.137 -20.448 (0.242) (0.191) (0.182) (0.163) (0.068) (10.194) Observations 72 72 78 78 78 77 R-squared 0.079 0.166 0.060 0.148 0.051 0.051 Overall Land Gini, 2002 0.399 0.158 0.287 0.575 0.055 -12.181 (0.239) (0.191) (0.183) (0.166) (0.059) (9.628) Luzon Dummy 0.119 0.069 0.053 0.088 0.040 -6.422 (0.059) (0.047) (0.045) (0.041) (0.015) (2.377) Mindanao Dummy -0.038 -0.065 -0.052 0.055 -0.036 1.889 (0.065) (0.052) (0.050) (0.045) (0.016) (2.639) Observations 72 72 78 78 78 77 R-squared 0.198 0.166 0.138 0.198 0.347 0.230 Panel A: Simple Bivariate Regression Panel B: With Mindanao, Luzon Fixed Effects Table 11 Land Inequality and Public Goods Provision (Provincial-Level Regressions) Dependent Variable is: Putting the Results Together: The Ampatuans in Maguindanao On November 23, 2009 a convoy of female relatives of Esmail Toto Mangudadatu and journalists was intercepted by around 100 armed men in the municipality of Datu Saudi-Ampatuan. The relatives were on their way to the capital of Maguindanao, Shari Agauk, to le papers for Mangudadatus candidacy to contest the election for the governorship of Maguindanao in 2010. 21 women and 36 men were killed. That the massacre happened in the municipality of Datu Saudi-Ampatuan on the way to Shari Aguak was signicant. Both names reect the rise to power and political dominance in the province of the Ampatuan family. The family claims to be descended from a preacher Shari Aguak who brought Islam to the area, probably in the 14th century. The capital of Maguindanao was renamed in his honor in 2000. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 13 / 20 History of the Ampatuans Andal Ampatuan Sr started the dynasty. He started as vice-mayor then mayor of Shari Aguak (then Maganoy) during the dictatorship of Ferdinand Marcos, was back as mayor in 1988. After 10 years in this job he stood down and was elected governor of Maguindanao in 2001. He was re-elected in 2004 and then again in 2007 when he was unopposed. As Toto Mangudadatu discovered, this was his preferred type of election contest and he anticipated something similar when his son Andal Jr planned to replace him as governor in 2010 (since Andal Sr was term limited). In 1995, Maguindanao had eighteen municipalities, by 2009 it had 36, most of them run by Andal Srs sons, nephews, in-laws and other members of his extended family. The municipality of Datu Unsay was created in 2003. Its rst mayor, still in power at the time of the massacre, was Andal Ampatuan Jr, (nickname is Datu Unsay). The municipality of Datu Saudi-Ampatuan, also created in 2003, was named after his eldest son. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 14 / 20 The Politics of Persistent Poverty How is the absence of the state and the rule by the Ampatuans an equilibrium? Andal Srs consolidation of power was greatly helped by his ability to forge links with national politicians such as Gloria Macapagal Arroyo, elected President of the country in 2004. Arroyo received statistically improbable numbers in Maguindanao. In at least two towns in the province, Arroyo won all the votes cast in the election, while her hugely popular opponent, ex lm star Fernando Poe Jr. got zero votes. Andal Sr. had also promised a 12-0 sweep for Arroyos senatorial lineup, and he delivered. This is similar to what drives regional inequality in other parts of the world, the Hayes-Tilden agreement of 1877 where the North agreed to stop trying to reform the US South (ending Reconstruction) in exchange for votes in the electoral college. Santa Fe de Ralito in Colombia (Acemoglu, Robinson and Santos, 2013). How does a northern Italian politician like Berlosconi manage to win big in Sicily? Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 15 / 20 Why in Maguindanao? All the structural factors we have identied point towards high poverty in Maguindanao. During the colonial period the state was absent (98% of the people were wild in 1903). Since independence the state has also been weak and clearly lacks a monopoly of violence. Political dynasties emerged. Very low levels of land inequality (Gini of 0.68) so no landed elites to oppose the rise of the Ampatuans (interestingly the more functional parts of Mindanao with far better governance and low politics, such as Davao City in the province of Davao del Sur (Land Gini 0.84) or General Santos in South Cotabato (Land Gini 0.83) far higher levels of land inequality. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 16 / 20 Is Mindanao Historically Condemned to Poverty? Note that Mindanao is closer to the equator than Luzon. Is it condemned by its geography to poverty and under-development? In fact in the early modern period, Mindanao, even Maguindanao where the Ampatuans now hold sway was the most developed part of the Philippines, producing spices like pepper for export and integrated into the mercantile trade routes of Southeast Asia. In fact, as we discuss in Why Nations Fail, the relative development of the Philippines was reversed (A Reversal of Fortune) by the Dutch East India Company which systematically attacked the Sultanates of Maguindanao, Sulu and elsewhere. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 17 / 20 Reality Check: Leyte and Samar Obviously hard to compare Mindanao and Luzon, many omitted variables and hard to distinguish between the direct and indirect eects of factors that lead to extractive political institutions. An interesting natural experiment is Leyte and Samar. Two islands, right next to each other, connected by a bridge since 1973. No ethnic, religious or signicant geographical dierences Yet huge gap in terms of poverty and income per-capita. Why? Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 18 / 20 Leyte and Samar: Some Ideas The developmental dierences between Leyte and Samar are very consistent with our emphasis on extractive political institutions. Long history of relative absence of the state in Samar (wild tribes) which continues today (road density, irrigation, presence of NPA and non-state armed actors) Relative governance failure (Leyte has ourishing participatory budgeting, Samar had one attempt - Santa Rita - which collapsed) and large dierences in accountability (Leyte has a radio station and independent media, Samar has neither). Suggests that the dierences in political institutions between two similar places may cause large dierences in poverty (and not the other way round: modernization?) Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 19 / 20 !"#$%&'( )#$("*+ -&'%.(&'( /001 -&'#2( )(" 34!%*4 56789 /000: ;%<. ="%>(?9 @10A B C%*D E4.%# E#4. F(&?%*+ G''(?? *# -""%H4*%#& I#J G%"!#"*? IKE=LMEI 8GNGE OPJ1/1@Q @@P/ RPP 0JQS/ 0J@/Q 0J@1Q @J0 MG8=MEI 8GNGE OSJOS10A @@O0 RPP 0JQQR 0J01Q 0JO11 @J0 8GNGE O@JP01/S @/A@ RPP 0JR01 0J0PO 0JA0P @J0 TMU=M AAJARSO1 @S@Q 0 0JR1Q 0JAQ/ 0JSO0 AJ0 8K7=LMEI TMU=M ASJAS0SS @QRO 0 0JSOS 0JAOS 0J101 /J0 V-T-EGI A/J@1//R @ARS 0 0JRRA J J J Conclusions: Back to Why Regions Fail In this talk we presented some simple facts about the correlates of poverty in the Phillippines. There are two big stories 1 Bad things go together in Mindanao, Good things go together in Luzon, accounting for the large dierences in poverty rates (and income per-capita) between them. 2 Land Inequality is strongly associated with lower levels of poverty. Both of these ndings can be related to the underlying nexus of extractive and inclusive institutions in the Philippines. For example, Mindanao suers from a long-running absence of the state which leads to violence, the undersupply of public goods, low educational attainment, high poverty and low incomes. This is a key part of extractive political institutions This spills over into political dynasties, which adversely inuences the other margin of extractive political institutions - a pluralistic distribution of power - which further increases poverty. Querubn Robinson (NYU and Harvard) Why Regions Fail August 20, 2014 20 / 20