ARTICLE IN PRESS

European Economic Review 51 (2007) 18961921 www.elsevier.com/locate/eer

Endogenous institutional change after independence

Heather Congdon Fors, Ola Olsson
Department of Economics, School of Business, Economics, and Law, Go teborg University, Box 640, 405 30 Go teborg, Sweden Received 10 June 2005; accepted 15 January 2007 Available online 4 May 2007

Abstract Independence from colonial rule was a key event for both political and economic reasons. We argue that newly independent countries often inherited sub-optimal institutional arrangements, which the new regimes reacted to in very different ways. We present a model of endogenous changes in property rights institutions where an autocratic post-colonial elite faces a basic trade-off between stronger property rights, which increases the dividends from the modern sector, and weaker property rights that increases the elites ability to appropriate resource rents. The model predicts that revenuemaximizing regimes in control of an abundance of resource rents and with insignicant interests in the modern sector will rationally install weak institutions of private property, a prediction which we argue is well in line with the experience of several developing countries. r 2007 Elsevier B.V. All rights reserved.
JEL classication: O17; O57; P14 Keywords: Institutions; Property rights; Independence; Resource rents; Rent seeking

1. Introduction The rules that societies live by have proven to be crucial for all kinds of economic development. A number of recent studies have established links between the general quality of countries economic institutions and, for instance, income per capita (Hall and Jones, 1999; Acemoglu et al. (AJR), 2001, 2002; Easterly and Levine, 2003; Acemoglu and
Corresponding author. Tel.: +46 31 7731341; fax: +46 31 773 1326.

E-mail addresses: heather.congdon@economics.gu.se (H. Congdon Fors), ola.olsson@economics.gu.se (O. Olsson). 0014-2921/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.euroecorev.2007.01.010

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1897

Johnson, 2005; Olsson and Hibbs, 2005). Perhaps more difcult to understand is what explains the wide international variation in measures of institutional quality. Empirical and theoretical efforts in this tradition have typically focused on deep historical explanations such as the various effects of colonialism (Acemoglu et al., 2001, 2002; Sokoloff and Engerman, 2000) or the role of the sovereign in the legal and economic systems of medieval Spain, France, and Britain (North, 1990; Glaeser and Shleifer, 2002). However, although institutions typically display a high degree of persistence, we argue that the institutional conguration of countries is also inuenced by more recent circumstances. The central issue that we address in this article is why more or less equally autocratic regimes installed widely different institutions for private property after independence. In countries like Singapore, the government pursued a strengthening of property rights and of the protection against government expropriation, whereas development was quite the reverse in some initially relatively developed countries like Ghana and Zambia. The major hypothesis advanced in this article is that the relative importance of natural resource rents played a central role for post-independence institutional development. The development literature contains numerous anecdotal accounts of how large inows of easily appropriable rents from oil and valuable minerals provided rulers in developing countries with strong incentives towards expropriation of private property (Bates, 1981; Bates and Collier, 1993; Ndikumana and Boyce, 1998; Bigsten, 2001, Sala-i-Martin and Subramanian, 2003). Several econometric studies have also conrmed the negative relationship between natural resource abundance and institutional quality (Leite and Weidmann, 1999; Sala-i-Martin and Subramanian, 2003); Dalgaard and Olsson, 2007). In order to understand this phenomenon, we present a model of endogenous change in property rights institutions and executive constraints. We take as given the deeper historical effects of for instance the disease environment and the identity of the colonizer, and model endogenous institutional choice for a more or less autocratic elite who maximize their own utility in a two-period setting. The rst period starts at independence and is typically characterized by sub-optimal property rights institutions from the perspective of the new regime.1 The ruling elite has economic interests in a modern, formal sector but can also appropriate rents from a natural resource sector. These circumstances imply that the ruling elite faces a basic trade-off between weak and strong institutions of private property. Strong property rights will make the modern sector prosper and raise the ruling elites incomes from that sector. Weaker property rights, on the other hand, means a poorly functioning modern economy but makes the ruling elites expropriation of resource rents easier. Our model predicts that ruling elites are more inclined to weaken property rights if easily appropriable natural resource rents abound, whereas the lack of an easy rents-sector coupled with substantial interests in a modern sector will motivate even an autocratic ruling elite to install stronger private property rights, which in turn results in higher growth. The costs of institutional change also play a central role in our model. In addition, a higher initial level of property rights protection will diminish, in some instances even negate, the negative impact of natural resource abundance on growth, as demonstrated
See Acemoglu et al. (2001, 2002) for possible explanations as to why colonial powers might establish extractive institutions in the colonies. Further, Djankov et al. (2003) argue that the institutions put in place by colonial powers were likely to be inefcient, even when the colonizers attempted to transplant their own institutional arrangements directly.
1

ARTICLE IN PRESS
1898 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

empirically by Mehlum et al. (2006).2 Therefore, the initial level of property rights protection plays an important role in both the incentive faced by the ruling elites to strengthen or weaken property rights as well as the cost constraint faced by the ruling elites. In this respect our model maintains a strong aspect of institutional persistence, as argued by authors such as Acemoglu et al. (2001, 2002, 2005) and Acemoglu and Robinson (2005), while at the same time allowing for more recent events to play a key role in the formation of property rights institutions. We argue that the main insight from our model is well in line with observed postcolonial experiences in many developing countries.3 Although independence from colonial rule is the main type of change that we have in mind, we believe that our model might also have relevance for understanding the institutional choices after other discontinuous regime shifts such as the transition from communism, or even the onset of colonization. Indeed, Beck and Laeven (2006) nd evidence that natural resources have been a major impediment to institutional development in transition economies. Our research extends a long tradition stretching back to Adam Smith that emphasizes the central role of property rights in development (Coase, 1960; Demsetz, 1967; Alchian and Demsetz, 1973; North, 1981, 1990; Firmin-Sellers, 1995; de Soto, 2000). Our argument about the forces behind the choice between weak and strong private property rights is to some extent inspired by Acemoglu et al.s (2001, 2002) colonial theory of how European settlers installed extractive institutions when the disease environment was hostile for permanent settlement and when there were easily exploitable human resources (proxied by population density and degree of urbanization). According to the same logic, strong institutions were created where permanent European settlement was feasible. The formal model in this article shares the basic prediction in Acemoglu et al. (2002, 2005) work that a regimes own interests in a progressive sector as well as the existence of easily appropriated rents are crucial for understanding institutional choice. Like North (1981), Acemoglu and Johnson (2005), and Djankov et al. (2003), we recognize that property rights institutions affect the economy along two dimensions. The rst dimension is the relationship between common men, for instance whether private property is secure against expropriation by a neighbor. The second dimension is the interaction between ruling elite and subject and to what extent the government can expropriate means from the people. In their empirical efforts to unbundle institutions to determine what kind of inuence different institutions have, Acemoglu and Johnson (2005) show that property rights constraining the ruling elite from expropriation appear to be the more important dimension for development.4 Furthermore, Glaeser et al. (2004) show in their empirical analysis that almost all former colonies in 1960 were dictatorships. The crucial question that arises from their analysis, and which we examine in this article, is why some autocratic ruling elites pursued growth-oriented policies (such as strengthening property rights institutions) while others did not.
See Sachs and Warner (2001), Gylfason (2001), and Woolcock et al. (2001) for discussions of the possible reasons for the observed curse of natural resources. 3 See the working paper version of this article (Congdon Fors and Olsson, 2005) for a more lengthy discussion of how post-colonial developments in Singapore, DR Congo, Botswana, Zambia, and Ghana are conducive to our model. 4 Throughout the article, we will use property rights institutions and constraints against the executive as synonymous terms, reecting the elites respect for private or state ownership. Respect for state ownership means that the elite does not regard state-owned property as their own.
2

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1899

This issue is also treated in a recent work by Rajan and Zingales (2006). Their basic argument is that it is the particular conguration of constituencies rather than political institutions that is the key for understanding why bad policies often prevail despite their social inefciency. In their model, comprehensive reforms (expansion of education and removal of barriers to entry in the modern sector) are often blocked because oligopolists and an educated class create a kind of unholy alliance that keep the masses out of education and everyone but the oligopolists out of business ownership. In a voting game between the three constituencies, it is shown that comprehensive reforms will only be carried out when the educated class is relatively large and when the oligopolists are very efcient. Rajan and Zingales model is different from ours in at least two crucial respects: Firstly, there is no state or government in the Rajan and Zingales framework, only three constituencies. In our model, a potentially predatory government plays a key role, as in many developing countries. Secondly, Rajan and Zingales type of pro-market reforms is essentially about removing barriers to entry in an initially oligopolistic market whereas the type of reform that we consider is a strengthening of protection against government expropriation for all. Finally, like us, Svensson (1998) analyzes the potential reasons why a government does not invest in stronger property rights. The basic reason in Svenssons model is that ruling elites are uncertain about staying in power due to political instability and hence do not internalize the full benet of institutional investment, a scenario which differs from ours where the potential for capturing rents is the key feature. Other models of property rights and growth include Tornell (1997) and Sonin (2003). The paper is organized as follows. Section two outlines the ruling elites basic trade-off in terms of investment in property rights institutions and proceeds by presenting a solution to the full model. Section three then makes labor supply endogenous and introduces variations such as trade liberalization. Section four concludes the paper. 2. The model In this section, we present and discuss the main assumptions in a two-period model of an autocratic ruling elites endogenous choice of costly property rights reforms. For simplicity, we assume that the ruling elite does not have to worry about being overthrown as long as they provide a minimum of public goods.5 The country has just emerged from independence and the ruling elite has inherited property rights institutions from a former colonial regime. While it is possible that these institutions are optimal, it is more likely that they are not. As will be shown, the model serves to explain how it even might be optimal for a ruling elite to weaken property rights institutions. 2.1. Basics A central element of our model is that property rights institutions and constraints against the executive should be weak in former colonies where the relative importance of natural resource rents is strong. The empirical case for this assumption is illustrated in
5 There is a growing literature studying the mechanisms through which non-benevolent rulers might stay in power for decades in developing countries despite the presence of potential rebels (Olsson and Congdon Fors, 2004; Acemoglu et al., 2003).

ARTICLE IN PRESS
1900 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

Figs. 1a and b. Fig. 1a shows the conditional correlation between a measure of rule of law and strength of property rights in 2004 on the vertical axis and the share of energy and mineral rents to GNI in 1999 on the horizontal axis for 67 former colonies (see A1 for a presentation of the variables and estimations involved). The underlying equation includes a number of geographical control variables plus Acemoglu et al.s (2001) variable Log Settler Mortality, which is widely interpreted as a proxy for the quality of property rights institutions during colonial times. In line with previous ndings mentioned in the introduction, both settler mortality and our natural resource measure have a strong and statistically signicant negative association with the rule of law. In Fig. 1b, we replace the previous resource variable with the log of country area. This variable is often seen as a proxy for natural resource wealth (Sachs and Warner, 1997) as well as a (negative) determinant of openness to trade (Frankel and Romer, 1999). In this specication, the negative estimate for our natural resource variable is even more signicant (t-value 5:27). Thus, even after controlling for a proxy for colonial history, natural resource rents appear to have a strong negative impact on property rights institutions.6 For the individual country, our theory builds on the notion that countries that had a large ow of resource rents at independence should experience a weakening of institutional strength. This should at least be true until the end of the Cold War around 1990, an event which gave rise to a trend break with a steadily increasing political freedom in most developing countries. Statistical assessments of our institutional assumptions are complicated by the fact that extensive time series for institutional variables are rare. In Fig. 2, we use a variable called Executive constraints from the Polity IV dataset where a high score (max +7) indicates that the executive is effectively constrained by law and is regularly held accountable by a legislature or in elections (Marshall and Jaggers, 2003), whereas a low score (1) implies unlimited authority, or that the state, for some reason, is not functioning 1.7 We focus here on development from independence until 1991 in three countries that are all heavily endowed with natural resource rentsNigeria, Republic of Congo, and Zambia. The rst two countries are indeed the two most resource-rich countries in Fig. 1a. Furthermore, Nigeria is maybe the most important case in Africa given its size and huge oil rents.8 As Fig. 2 shows, Nigerias score fell from 7 to 1 or 1 after independence with a short period of good years in the early 1980s. Republic of Congo and Zambia both experienced a steady decline throughout the period. Although these trends do not establish any proof of the validity of our assumption, they are well in line with anecdotal evidence cited in the Introduction. In a more extensive working paper, we provide further qualitative evidence in favor of our hypothesis from Singapore, DR Congo, Botswana, Zambia, and Ghana (Congdon Fors and Olsson, 2006). Let us now proceed with the model. Let us assume a scenario with an autocratic ruling elite who is primarily motivated by the possibility of making personal revenue with the least possible effort. The ruling elite has two potential sources of personal gain: Their private (non-state) interests in a modern sector that operates on the world market and a
For a more complete discussion of this type of results in the literature, see Dalgaard and Olsson (2007). We have given a score of 1 to what the dataset refers to as interregnum, transitional, or interruption episodes. 8 See Sala-i-Martin and Subramanian (2003) for an account of how oil rents have affected the Nigerian society.
7 6

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1901

a
2
SGP BR B BHS

1 Rule of law

ML I GM B A GH I CR

CH L HKG MU S TT O MRT EC U DZ A VEN AG O

MDG PAN MY S UR SE N Y LK A TU N BF UG TZ A BR AU Y A NZL TG O AJA M A S ZAF EG RWSL MA R A IN US D SLNERV SURVNM RCAN E DOM N ACM X ME GI IDN CIVKE N D BOL CO L GU Y HN PE R TCD CA F ARG BGD GT M ZAPRY R PA K HT I SDN

NG A COG

-1

-2 -0.1 0 0.1 Energy and mineral rents/GNI 0.2 0.3

b
2
SGP BRB BHS HK G MUS GM B TT O CR I GH A PA N Y MD G MY S UR SE N LK A TU N BFA MA R T TZ A EG Y ZA UGA MR AUS F IN D ME X NG A N SU R NE NIC VN MUS AR GI N CACMR KEN ID HNCIV COG ECU D COL GUY D BOLPE R N TC DZ A CA F AG O ARG BGD GT M VEN PR Y PA K ZA R SDN MM R CH L ML I

1 Rule of law

NZL TG O JA M RW A SL V SL E DOM

BRA

-1
HTI

-2 -6 -4 -2 0 2 4

Log of country area

Fig. 1. (a) Partial scatter plot, Rule of law in 2004 vs energy and mineral depletion as a share of GNI in 1999 among former colonies. Note: The gure shows the conditional correlation between the two variables in a multivariate regression. The estimated underlying equation in the gure is: Rule of law in 2004 0:9135 (constant) t 1:53 1:66 Nrg_Min1999/GNI t 2:34 0:283 Log Settler Mortality t 2:33 0:010 Latitude t 1:10 0:129 Sub-Saharan Africa dummy t 0:58 1:164 Neo-Europe dummy t 3:77. R2 0:54, N 67. OLS-estimator with robust standard errors was used. See Appendix A1 for variable descriptions. (b) Partial scatter plot, Rule of law in 2004 vs Log of country area among former colonies. Note: The gure shows the conditional correlation between the two variables in a multivariate regression. The estimated underlying equation in the gure is: Rule of law in 2004 2:71 (constant) t 4:61 0:183 Logarea t 5:27 0:237 Log Settler Mortality t 2:47 0:016 Latitude t 2:27 0:012 Sub-Saharan Africa dummy t 0:94 1:74 Neo-Europe dummy t 5:47. R2 0:65, N 69. OLS-estimator with robust standard errors was used. See Appendix A1 for variable descriptions.

ARTICLE IN PRESS
1902 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

8 7 6 Executive constraints 5 4 3 2 1 0 1960 -1 -2 1964 1968

Nigeria Congo, Rep. Zambia

1972

1976

1980

1984

1988

Year
Fig. 2. Development of Executive constraints since independence in three resource-rich countries. Note: The variable in question is referred to as XCONST in Marshall and Jaggers (2003).

more or less easily appropriable ow of rents from a state-controlled natural resource sector. These incomes are available in a current as well as in a future period. The ruling elite also controls (but cannot make personal gain of) the government budget with revenues from the natural resource sector and income taxes on labor. In order to avoid being overthrown, the ruling elite has to supply a xed (minimum) amount of public goods like infrastructure, defense, and police to uphold the most fundamental functions of the state. What remains in the public budget can be spent on costly reforms of the prevailing property rights institutions. It is assumed that all changes in property rights, regardless of whether they involve a weakening or a strengthening or rights, give rise to direct costs. The ruling elite gains utility only from personal enrichment. These basic assumptions are summarized in Eq. (1), showing the ruling elites utility function in general form: U r R0 M 0 ejZ 1 Z 0 j dR1 Z1 M 1 Z 1 . (1)

The utility of the ruling elite is a simple linear function of wealth in the form of current and future expropriated rents R0 and R1 Z 1 and of current and future prots from the ruling elites private interests in the modern sector M 0 and M 1 Z1 . Utility is also a negative function of the effort of changing property rights ejZ 1 Z 0 j. dp1 is a time discount factor. Z 1 is the quality of property rights institutions and constraints against the executive in the future period and is the ruling elites key choice variable. They have inherited a property rights regime of strength Z0 and consider increasing or decreasing this level. The personal costs of these changes are a function of the absolute level of Z 1 Z 0 DZ such that e0 jDZj40 and e0 0. In other words, investments over the prevailing level DZ40 give rise to an equally large disutility as disinvestments DZo0 and the least disutility is gained from status quo DZ 0 when no effort at all is exerted in either direction. A central assumption of the paper is that R01 Z 1 o0 since weak constraints against the executive makes expropriation of natural resource rents for personal gain easier. With strong constraints against the executive, the ruling elite will be unable to use the state

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1903

companies as a source of personal revenue. Equivalently, M 01 Z1 40 because stronger property rights has a positive effect on total output in the modern sector as well as a positive net effect on ruler rents. We assume that the specic mechanism through which stronger property rights translates into greater modern sector rents for the ruling elite is private rm ownership. If rms have a safe legal environment, transaction costs are reduced, rms produce more and even the elites prots increase. Direct expropriation of modern sector prots is assumed to be impossible since rms operate on the world market and could relatively easily shut down production and move elsewhere.9 The natural resource sector, on the other hand, is highly immobile and therefore an ideal object of expropriations. The price of expropriating resource rents is of course that modern sector output shrinks. The assumption that the same type of property rights institutions affect the natural resource sector and the modern sector is a cornerstone of our model. A potential objection might be that the institutions constraining the ruling elite from expropriating natural resource rents are different from the institutions affecting modern sector output. If that was the case, there would be no trade-off of the kind modelled here and a kleptocratic ruling elite should simply minimize the constraints against expropriation in the natural resource sector and maximize the strength of private property rights in the modern sector.10 However, as will be specied below, it will be assumed here that the ruling elites revenue from the modern sector comes from private minority share-holding. Since the ruling elite thus is directly involved both in the resource sector (through state-ownership) and in the modern sector (through private ownership), rm-owners in the modern sector who are not part of the elite will judge the quality of property rights by the standards of behavior that the ruling elite employs in the resource sector. If the ruling elite steals natural resource rents from the state-owned companies for their own enrichment, rational rm-owners will expect that the ruling elite sooner or later will do the same in the modern sector. Firm owners will therefore cancel some of their more risk-exposed activities and try to keep a low prole in general. If all rms undertake such extra precautions, transaction costs increase and efciency is hampered. Hence, both sectors will be affected by the same set of institutions, measured by Z t . If we disregard the government budget and all constraining factors for a moment, the optimal level of Z 1 from the ruling elites point of view is the solution to the rst-order condition: e0 jDZj qjDZj dR01 Z M 01 Z 0. 1 1 qZ 1 (2)

This condition relates the basic intuition behind the paper; investment in property rights institutions entails a trade-off for the ruling elite between the direct cost of effort ejDZj and the indirect cost of lower expropriated rents R1 Z on the one hand, and the benets 1 of greater dividends from the modern sector on the other.
One might of course imagine that some large rms in the modern sector are immobile in the short run and therefore suitable objects of predation. 10 Deliberate institutional differences between a formal and an informal sector are modelled in Olsson and Congdon Fors (2004).
9

ARTICLE IN PRESS
1904 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

2.2. Functional form With the general form above, however, we cannot derive an explicit solution for Z or 1 indeed say much else of interest. We will therefore specify a functional form of the utility function and of the government budget constraint that we believe capture the central aspects of the ruling elites investment decision. Starting with the appropriable rents Rt , we mentioned earlier that these can most easily be thought of as proceeds from a more or less state-controlled natural resource sector.11 Let us assume that there is a total ow of rents rt during period t 0; 1. For simplicity, we choose to model them as exogenous incomes that are generated without using any inputs and that rents are expected to be stable over time; r0 E 0 r1 r. Since the companies are state-owned, all rents are supposed to ow into the government budget. However, the predatory ruling elite will attempt to conscate rents for their personal enrichment. The amount that the ruling elite can lay their hands on Rt depends on the existing level of government constraints. We assume that   Z Z t r Rt max ;0 , (3) Z where Z is a critical level of institutions beyond which rent appropriation is impossible. The max sign above means that if Z t 4Z, then Rt 0.12 We will assume throughout that Zt oZ which arguably is the normal case for the developing countries that we have in mind. The amount of rents that are not expropriated by the ruling elite Z t r=Z ow into the government budget, as will be shown below. The ruling elites second source of personal revenue comes from share-holding in the private sector. To explain how such an ownership has been obtained is beyond the scope of this article.13 As a share-holder, the ruling elite gets their part of total dividends; M t ZPt where Zo0:5. The assumption of Zo0:5 means that the ruling elites holdings are relatively small, at least smaller than to give them a majority ownership in the private sector.14 We further assume that Z is uncorrelated with Z t . Prots in the modern sector are given by Pt pt Qt wLm;t pt Z t Lb wt Lm;t . m;t (4)

For simplicity, the quality of property rights institutions Zt increases modern sector output Qt linearly like a (Hicks neutral) total factor productivity variable. The channel through which a higher Z t affects output is a general increase in efciency and a reduction of transaction costs. Analogously, a weakening of property rights would induce rm owners
See Acemoglu et al. (2003) for a similar assumption. Real world examples of more or less state-controlled mining companies are Gecamines (copper) and MIBA (diamonds) in former Zaire, Debswana (diamonds, joint venture with De Beers) in Botswana, Endiama (diamonds) in Angola, and ZIMCO in Zambia. We also believe that marketing boards such as that for cocoa in Ghana might also serve as a source of easily appropriable rents (Bates, 1981). 12 This reects the idea that diversion of natural resource rents in developed countries like Canada, Australia, or the United States do not seem to enter the rulers objective function. 13 Evidence from Kenya and an analysis of the ruling elites private ownership patterns are discussed at length in for instance Bigsten and Moene (1996) where the phenomenon is referred to as straddling. 14 One could imagine that the ruling elite gains utility not through dividends, but through the general increase in the standard of living brought about by a thriving modern sector. In this case, Z would represent the utility derived from the modern sector.
11

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1905

to take extra precautions and force them to spend more time on improductive activities such as contract enforcement.15 Lm;t is labor supply to the modern sector at period t, and bo1 gives the (diminishing) returns to labor. Labor supply is treated as exogenous and xed in this section so that Lm;t Lm but will be made endogenous below. Workers are paid a wage rate wt which is equivalent to the value marginal product; wt pt Z t bLb1 . In other words, modern sector labor will be more m;t productive on the margin (and will be better compensated) if property rights institutions are strong. The price of modern sector output is pt . Below we will specify the price level to be p0 p1 1. In summary, the ruling elites personal revenue from the modern sector will be M t ZPt ZZ t Lb 1 b. m (5)

The ruling elites disutility of effort is an increasing function of the deviation from status quo: ejDZj yZ 1 Z 0 2 yDZ2 , 2 2

where y40 indicates the degree of effort required to carry out changes in property rights institutions. The assumption of a quadratic function ensures that investment and disinvestment are equally costly in terms of disutility and implies that the ruling elite has a kind of status quo bias. Whereas the rst derivative with respect to Z 1 can have any sign, the second derivative is unambiguously positive at y40. This means that there is an increasing marginal disutility of institutional change. What this is intended to show is that institutional changes will require more and more effort on the margin as jDZj increases and hence give rise to a greater and greater utility loss. All in all, these assumptions give us a more detailed functional form of the ruling elites objective function in (1):   Z0 yZ 1 Z 0 2 Ur 1 r ZZ 0 Lb 1 b m 2 Z !   Z1 d 1 6 r ZZ1 Lb 1 b . m Z In maximizing personal revenue, the ruling elite has to take into consideration not only the legal constraints but also the scal constraints as given by the government budget. We will assume that in order to stay in power for more than one period, the ruling elite has to supply a xed quantity of public goods G. This quantity includes the costs of police, defense, physical infrastructure, and basic government administration. The level of G might also be regarded as an indicator of the strength of one-man-rule. With a given revenue ow, a low G means that the ruling elite has a relatively great degree of freedom in deciding themselves what institutions to create and how much to spend on public goods. Conversely, a high G means that the ruling elite is highly constrained in the
We could also think of an increase in Zt as causing an entry of new rms, as in Rajan and Zingales (2006). In their model, however, entry makes prots shrink for the oligopolists already active in the sector, which causes them to vote against such a reform. A similar scenario is sketched in Do (2004) where entrepreneurs bribe a regulator to restrict entry. In our model, rm owners have no such power.
15

ARTICLE IN PRESS
1906 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

process of institutional reform and that they are obliged to satisfy rather ambitious goals in terms of public goods provision in order to avoid being thrown out of ofce.16 A more long-term type of government expenditure is institutional reform. As in much of the literature on adjustment costs in investment, the costs of institutional investment or disinvestment are a convex function of DZ: cZ 1 Z 0 2 =2 where c40 is a parameter. The second derivative is simply c40, implying an increasing marginal cost. In order to reap the benets of property rights investment, the ruling elite must be able to credibly enforce these rights. That entails investment in a number of supporting institutions, e.g. court systems, property registration ofces, etc. If the ruling elite chooses to disinvest, on the other hand, they will be faced with the costs of dismantling the existing structures and probably also with the cost of compensating the losers from such a reform in some way. It might be argued that disinvestment in property rights institutions is less costly than investment. A way to formalize this notion would be to include a c0 oc for disinvestment costs. In the extreme case, one could have a c0 0 which would imply that a worsening of institutions is costless. As will be shown below, such an assumption would mean that only positive investments were constrained by the government budget. This could easily be incorporated in the model.17 However, we do believe that it is reasonable to assume that rulers are always somewhat constrained from destroying property rights completely. Further, some positive level of second period property rights institutions is required in order to fulll the second period budget constraint (see below). Government revenue has two sources; the non-expropriated rents from the natural resource sector Z t r=Z and a proportional income tax on workers in the modern sector twLm tZ t bLb . The marginal tax rate t might be thought of as the (exogenously given) m revenue-maximizing tax rate as in a Laffer-curve. The governmental budget restriction in the initial period states that xed government outlay plus the costs of property rights investment or disinvestment must not exceed revenue, i.e.: G cZ1 Z 0 2 Z0 r ptZ 0 bLb . m 2 Z (7)

We assume for the remainder of the article that GotZ 0 bLb Z 0 r=Z so that the ruling m elite always has some scope for institutional reform. Further, the level of Z 1 must be such that second period revenue is sufcient to cover the second period xed government outlay, i.e.: GptZ 1 bLb m Z1 r . Z (8)

These budget restrictions imply that the optimal level of property rights institutions is ~ ^ ^ dened in the interval Z 2 maxZ ; Z ; Z where
1 1 1 1

~ Z1
16 17

tbLb m

G 40 r=Z

(9)

See Aghion et al. (2004) for a model of an endogenously determined political insulation from the people. A demonstration of the effects of asymmetric costs is provided upon request. The only effect would be that it could change a lower boundary solution if this turned out to be the equilibrium.

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1907

and ^ Z1 s 2tZ 0 bLb Z 0 r=Z G m Z0 v0, c (10)

while the upper boundary of property rights in period 1 is s 2tZ 0 bLb Z 0 r=Z G m ^ Z1 Z0 40. c

(11)

Therefore, even if property rights disinvestment was costless, the ruling elite would still ~ face the second period budget constraint, Z 1 . Thus, costless disinvestment will only make the weakening of property rights more likely in the case where (10) 4 (9). Note that the scope for reform in either direction increases with r; t and Lb and decreases with G. In m addition, the scope for reform in either direction increases with Z 0 and decreases with c with respect to the rst period budget constraint. The max-term dening the lower bound of Z reects the constraint that the actual Z is necessarily non-negative whereas a 1 1 negative institutional level might be nancially viable in terms of the rst period budget ^ constraint (i.e. if Z 1 o0. 2.3. Optimal institutional change The utility function and the governmental budget constraint form a maximization problem for the ruling elite max U r
Z1

subject to cZ1 Z 0 2 =2ptZ 0 bLb m and tZ 1 bLb m

Z0 r G Z

Z1 r XG. 12 Z By setting up a Lagrangian function G with multipliers l1 and l2 , we can derive the following KuhnTucker rst-order conditions:   qG dr r yZ n Z0 dZLb 1 b l1 cZ Z 0 l2 tbLb p0, 1 m 1 m qZ 1 Z Z qG Z0 r tZ 0 bLb G cZ Z 0 2 =2X0, m 1 ql1 Z qG Z1 r tZ 1 bLb GX0, m ql2 Z qG l1 0; ql1 qG l2 0; ql2 qG Z 1 0. qZ 1 (13) (14)

The third line shows the complementary slackness conditions. The second-order condition for maximum is fullled since the Lagrangian function is strictly concave in the relevant range Z1 40. The problem above implies that we can characterize the set of solutions in the following Lemma:

ARTICLE IN PRESS
1908 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

Lemma 1. The maximization problem in (12) 8 > > Z ~ ~ > if Z opt pZ 1 > 1 1 > > > > > > > > > Z 4Z if Z opt pZ ~ ^ ^ > > 1 1 1 1 > < Z1 : > > Z opt ^ ^ > if Z 1 oZ opt oZ 1 > 1 1 > > > > > > > > > ^ ^ > Z if Z opt XZ 1 > 1 1 > :

has four potential unique solutions: and l1 0; l2 40; ^ and l1 l 40; l2 0 and l1 0; l2 0; ^ and l1 l 40; l2 0;

i ii iii iv

~ ^ ^ where Z1 , Z1 and Z1 are given by (9), (10) and (11), respectively,   d r opt ZLb 1 b Z1 Z0 m y Z and ^ l

(15)

Z opt Z 1 ^ Z Z0
1

y ; c

^ l

Z opt Z 1 ^ Z0 Z
1

y ; c

l2

~ yZ 1 Z opt . r tbLb Z m

(16)

Proof. The results follow from straightforward manipulations of the rst-order conditions. & The rst two solutions (i)(ii) represent lower boundary extrema where the ruling elite disinvests in property rights institutions as much as they can afford, (iii) is an interior maximum with positive or negative investment, and (iv) is an upper boundary solution where the ruling elite uses all available government means for positive investment. The term Z opt is given by the unconstrained maximum qU r =qZ 1 jZ1 Zopt 0. This value 1 might or might not be attainable depending on the level of affordable reforms. The ^ ^ Lagrangian multipliers l , l and l2 reect the shadow value of net government revenue in case of constrained boundary solutions. The solutions derived in Lemma 1 might be used to express the following intuitive results: Proposition 1. (a) The ruling elite is more likely to strengthen (weaken) property rights institutions if Z and Lm are high (low) and if r and y are low (high). (b) In the case of positive boundary solutions, the change in the strength of property rights institutions in either direction increases with t, r, Lm and Z 0 and decreases with G and c. Proof. (a) Whether institutions are strengthened or weakened is determined by the sign of Zopt Z 0 d ZLb 1 b r=Z =y. A high Z and Lm and a low r imply that a positive m 1 sign is likely, and vice versa. (b) From (9)(11), we know that the greatest feasible institutional change in either q ^ ^ direction is Z 1 Z 0 Z 0 Z 1 2tZ 0 bLb Z 0 r=Z G=c40, while the minimum m ~ feasible level of Z 1 is Z 1 G=tbLb r=Z40. From these expressions, it is easily seen that m the result in (b ) applies. &

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1909

~ ^ In the case of Z Z 1 or Z 1 as in (i) and (ii) of Lemma 1, the ruling elite is totally 1 committed to destroying existing institutions. The proposition shows that this scenario might arise when the modern sector is small (low Lb ), when the ruling elites interests in the m modern Z sector are small, or when the rent ow r is large. The greater are r, t, Lm and Z 0 at this equilibrium, the stronger is the government budget and the ruling elite can afford an even greater weakening of institutions. This income effect might however be balanced by the fact that a weakening of property rights is less likely if Lm increases according to (a). A greater Gi.e. a smaller autonomy for the ruling elitewill in this equilibrium mean a smaller deterioration in institutions. When we have an interior solution such that Z Zopt , then the optimal level of 1 1 institutions in period 1 will depend positively on the inherited level of institutions in period 0 and negatively on resource rents r. Indeed, Figs. 1a and b provide an empirical counterpart of this theoretical result with settler mortality as a proxy for Z0 . However, if we look at institutional change as in Proposition 1, it is easily seen that a weakening of property rights is more likely if the rent ow r is large.18 In other words, the model predicts that countries with a relatively substantial rent ow at independence should face a worsening of property rights institutions. The logic is of course that high rents imply that a kleptocratic ruling elites opportunity costs of installing stronger property rights are high since a strong protection against expropriation will mean lower rents for themselves. However, apart from this substitution effect of an increase in r, rents also have an income effect via the budget constraint, as discussed above. In the interior solution, Z further increases with Z and Lm . The greater the ruling elites 1 interests in the modern sector and the greater the size of this sector, the greater is the likelihood of a positive change in second-period property rights institutions, as one might expect. ^ In the upper boundary solution where Z Z 1 , the ruling elite spends every available 1 penny in the government budget on improving property rights. The reason is simply that their marginal utility of Z 1 is positive in the whole feasible range. This good outcome is therefore more likely when the modern sector is relatively important for the ruling elites personal enrichment, i.e. when Z and Lm are high and r is low. From (11), we see that Z increases with r in the upper boundary solution due to the 1 income effect since an increase in r shifts the budget constraint further out. However, an increase in r also decreases the marginal utility of property rights investments, which is the force behind the substitution effect of r. At a certain level of r, an interior solution will arise in which case an increase in r will lower the optimal level of property rights.19 Thus, all else equal, the income effect of increases in r will dominate in the upper boundary solution if such increases start at very low levels of r, whereas beyond a certain level, the substitution effect of a higher r will dominate. It is also interesting to note that in this good equilibrium, a smaller scal autonomy for the ruling elite (a higher G) will imply a lower level of property rights institutions since the discretionary part of the budget shrinks. Similarly, high costs of institutional changereected by high levels of y and cwill inevitably result in small deviations from Z0 .20
More precisely, there will be a disinvestment in institutions if r4ZLb 1 bZ. m ^ This level of r is reached when l 0 which happens when Zopt r; Z1 r; . 1 20 ^ ^ The equations above show that both Zopt Z0 and Z1 Z0 Z0Z 1 will approach zero as y and c 1 approach innity.
19 18

ARTICLE IN PRESS
1910 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

If total output in the country is measured as modern production plus the ofcial or nondiverted ow of rents from the natural resource sector, then we can write Y t Qt Zt rt =Z. The growth rate of the economy in the case of an interior solution will therefore be   Y 1 Y 0 Z opt Z 0 d r ZLb 1 b v0. (17) m Z0 y Y0 Z0 Z The growth rate is thus negatively related to the ow of rents r. The negative marginal effect on growth of an increase in r decreases in absolute terms with Z 0 , implying that countries with relatively good initial institutions should experience smaller adverse effects of a high r. The corresponding growth rate for a country that optimally fully utilizes the government budget for positive investment is q 2tZ 0 bLb Z 0 r=Z G=c ^ Y 1 Y 0 Z1 Z0 m 40. (18) Z0 Y0 Z0 As mentioned above, for this category of countries in the good equilibrium, natural resources do not constitute a curse. On the contrary, growth will increase with an increase in r. These ndings perhaps contribute to explaining the results from the empirical growth literature that countries with good institutions seem to be able to escape the curse of natural resources (Mehlum et al., 2006). However, as shown above, beyond a certain level of r, an interior solution will arise and the (institutions-weakening) substitution effect will start to dominate. 3. Extensions In this section, we will extend our basic model to account for the effects of endogenous labor supply, natural resource booms and declines, foreign aid, trade liberalization, and impediments to institutional efcacy. 3.1. Endogenous labor supply In the derivations above, labor supply Lm was assumed to be exogenously given and xed throughout the two periods. This is not a totally satisfactory assumption since it can be easily imagined that labor supply should depend on for instance the tax rate and, perhaps more interestingly, on the quality of institutions in the modern sector. In this section, we will therefore derive labor supply endogenously and analyze how this alters the results from the previous section. We will assume that the decisions about institutional choice and labor supply are made as in a sequential game with the ruling elite acting as a leader whose supply of property rights institutions is taken as given by the workers. Let us assume that the ruling elites objective function is as in (6). Let us also postulate that there is only one group of workers with the linear utility function U l Lm;t 1 tZ t bLb gL Lm;t G for t 0; 1. m;t In each of the two periods, the working part of the population receives utility from aftertax labor income 1 tZ t bLb , from leisure or informal household production m;t gL Lm;t , and from the public goods G provided by the ruling elite. The only new

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1911

parameters here are L which is the xed total endowment of labor resources, and g40 that reects the marginal utility (or productivity) of household production. Note that property rights do not affect the output of informal household production.21 The workers maximize U l with labor supply Lm;t as the control variable, taking Z t as given. From the usual rst-order conditions, we nd that the equilibrium level will be  1=1b 1 tZ t b2 L . (19) m;t g Thus, labor supply in period t will be positively associated with the strength of property rights institutions in period t. This should make sense; the greater the levels of Z t , the greater the workers marginal product and the greater his or her labor supply to the modern sector. Conversely, the greater the utility from household production g, the lower the supply of modern sector labor. The ruling elite realizes by backward induction what labor effort that will be supplied to the modern sector and takes this level as given in their own optimization. This means that the ruling elites utility function becomes  b=1b !   1 X Zt yZ t Z 0 2 t 2b=1b 1=1b 1 t Ur . d 1 Zt r Z1 bb g 2 Z t0 The rst period government budget constraint is still that  b=1b cZ 1 Z 0 2 Z0 r 1=1b 1b=1b 1 t ptZ 0 b , G 2 g Z

(20)

which in turn implies that the highest attainable level of institutional change is s 1=1b 1b=1b 2tZ 0 b 1 t=gb=1b Z 0 r=Z G ^ ^ 40, Z1 Z0 Z0 Z1 c (21) while the second period budget constraint remains as  b=1b 1=1b 1b=1b 1 t GptZ 1 b Z 1 r=Z. (22) g When the optimization problem is changed in this way, the KuhnTucker rst-order conditions become more complicated. As visual inspection should make clear, the solution to this optimization problem will depend to a great extent on the levels of the labor elasticity parameter b as well as on the other parameters of the model. In order to simplify the analysis without losing focus on the essentials, we will make the following assumptions for the remainder of the analysis: b 2; 3 d y c g 1.
2 3 2 22 3.

(23)

The assumption that b is standard, i.e. that the share of labor in production is The assumption that the remaining parameters are equal to one is a simplication, which in
21 This could be the case if informal production does not require signicant investment (see, e.g., Besley, 1995). A similar assumption is made in Olsson and Congdon Fors (2004). 22 See for instance Krueger and Lindahl (2001) for a discussion of the most plausible world level.

ARTICLE IN PRESS
1912 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

many cases will be relaxed in the subsections below. The greatest attainable change in property rights is thus ^ Z1 Z0 Z0 ^ Z1 s   Z0 r 2 3 2 tZ 0 c1 t G , Z (24)

32 ~ where c 243, while Z 1 is the value of Z 1 that satises

~ tZ 1 3 c1 t2

~ Z1 r G 0. Z

(25)

Further, setting b 2 means that the ruling elites utility function above becomes a cubic 3 function of Z 1 . Unlike in the exogenous labor supply case, the second-order condition for maximum is not necessarily fullled in the endogenous labor supply case since the Lagrangian function is not strictly concave in the relevant range Z 1 40. Rather, as shown in Congdon Fors and Olsson (2006, Appendix 1), there are two extreme points, given by the expression 8 q > > 1 1 4Zd1 t2 Z0 r=Z > l;max > > Z1 ; > < 2Zd1 t2 q > 2 > > l;min 1 1 4Zd1 t Z 0 r=Z > >Z ; > 1 : 2Zd1 t2

Z l;opt i

(26)

l;min where d 16. Obviously, we can disregard Z 1 as a potential solution to the ruling elites 81 optimization problem. Upon inspection of (26), it is clear that we will have different types of solutions for Zl;max depending on the size of the expression under the square root sign. 1 Congdon Fors and Olsson (2006, Appendix 2) characterize in detail the solutions for Z l;max 1 and for Z . The analytically most interesting scenario arises when Z 0 r=Z40 and the 1 expression under the square root sign ranges between zero and one, which ensures that Zl;max 40. Figs. 3a and b show two types of solutions with this feature. In these cases, it is 1 l;max also evident that Z 1 oZ l;min . The local minimum also provides us with important 1 l;min information: At levels higher than Z 1 , investment in property rights once again starts to yield utility gains for the ruling elite. Now that we have established the local maximum and the three constrained cases, it remains to determine what the optimal solution to the ruling elites maximization problem will actually be. The optimal solution depends on the relationship between the constrained solutions and the unconstrained local maximum, and the utility derived by the ruling elite from each. In Congdon Fors and Olsson (2006, Appendix 2), all possible equilibria and the conditions associated with them are characterized formally. ~ On a more intuitive level, there are as before four potential solutions: (i) Z Z 1 , (ii) 1 l;max ^ ^ Z Z1 ; (iii) Z Z 1 ; or (iv) Z Z1 . In (i) and (ii), the ruling elite weakens property 1 1 1 rights as much as they can afford, whereas an interior maximum is optimally chosen in (iii). Whether the latter involves a strengthening or a weakening of institutions is not clear but depends on the parameters. This kind of equilibrium (with a worsening of property rights) is illustrated in Fig. 3a. (iv) shows the upper boundary solution, which might arise in four

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1913

Ur- utility of ruling elite

Z1-

Z1+

Zt -institutional strength
* Z1

Z0

Ur- utility of ruling elite Z1 + Z1

Zt- institutional strength Z0 + Z* = Z1 1

Fig. 3. (a) Optimal weakening of institutions with endogenous labor supply. Note: Interior maximum is optimal since Zn Zl;max oZ0 . (b) Optimal strengthening of institutions with endogenous labor supply. Note: Upper 1 1 ^ ^ boundary is optimal since Zn Z1 4Z0 4Zl;max ; U r Z1 4U r Zl;max . 1 1 1

^ different cases, for instance when Z l;max 4Z 1 . A noteworthy feature is further that Zl;max 1 1 ^ might not be the optimal choice even when it is affordable. As Fig. 3b shows, Z 1 might l;max give a higher utility than Z 1 since the utility function starts increasing again beyond the minimum. If we are to make a general characterization, the results in the endogenous labor supply case will be similar to those above: Proposition 2. (a) With endogenous labor supply, the ruling elite is more likely to strengthen (weaken) property rights institutions if Z and Z 0 are high (low) and if r and t are low (high). (b) In the case of positive boundary solutions, the change of property rights institutions in either direction increases with r and Z 0 and with t if to1, and decreases 3 with G. Proof. See Appendix A2.

ARTICLE IN PRESS
1914 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

As before, positive (negative) changes are more likely if Z is large (small) and r is small (large). If the imbalance between the incentives for strengthening property rights Zd1 t2 Z 2 and the incentives for weakening them r=Z is large in either direction, a boundary 0 solution is more likely where the results in (b) will apply. A difference from the previous section is that a strengthening of property rights is more likely if the level of inherited institutions Z 0 is high. This implies a kind of path dependence: Countries with strong property rights at independence will invest in even stronger rights whereas the reverse will be true for more weakly institutionalized colonies. The logic is that better initial institutions means that the modern sector is relatively more productive, so that the losses from weakening institutions will be greater, and there will be a greater incentive to strengthen property rights. Further, since r and Z0 affect both the unconstrained and the constrained solutions, they might have both income and substitution effects, as discussed above. The proposition shows that a rise in Z only has an impact on the maximum whereas G only affects the constrained solutions. An interesting case might occur if two otherwise identical countries have slightly different levels of required public goods, G. Let us assume that for some reason, one of the countries (country 1) gives a larger discretionary power to the ruling elite than in country 2 so that G 1 oG2 . Then, if the situation is as in Fig. 4, this means that country 1 will optimally be at the upper boundary solution whereas country 2 will be stuck at the lower boundary. This illustrates the notion that even small differences between countries might give drastically different outcomes regarding the optimal structure of institutions. 3.2. Natural resource booms and declines We have assumed that r0 E 0 r1 r, i.e. the rents from natural resources in the second period are expected to equal the rents in the rst period. In this section, we analyze the effects of r0 ar1 on ruling elites optimal choice of Z . Note that the rst period budget 1 constraint will be identical to that in (24) with r r0 , while the ruling elite will maximize
Country 1 Ur Z1 + Z1

Z0 Country 2

Zt

Fig. 4. A high and a low equilibrium level of property rights resulting from different country levels of minimum public spending G2 4G1 . Note: The gure illustrates how two countries with slightly different levels of mimimum public goods can have two widely different equilibrium levels of institutional strength with Country 2 disinvesting to its lower boundary and Country 1 (with a lower G) investing to its upper boundary.

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1915

second period utility taking E 0 r1 r1 into account, i.e. q 1 1 4Zd1 t2 Z 0 r1 =Z Z r;max 1 2Zd1 t2 and the second period budget constraint will now be ~ ~ 3 c1 t2 Z 1 r1 G 0. tZ 1 Z

(27)

(28)

If a natural resource boom is anticipated, we have r1 4r0 . The result is that Zr;max and 1 ~ Z 1 will occur at a lower level of property rights institutions while at the same time Zr;min 1 will occur at a higher level. The main result that Z r;max will fall with an anticipated resource 1 boom is intuitively straightforward. Because r0 does not change, there is no rst period income effect from the natural resource boom, while there is a second period income effect.23 3.3. Foreign aid So far we have assumed that the ruling elite can only nance changes in property rights institutions through domestic means. There is, however, a possibility that the ruling elite receives development aid from foreign donors. Ideally, this aid would be ear-marked for institutional development, yielding the following upper budget constraint : s   Z0 r ^ Z 1 Z 0 2 tZ 3 c1 t2 (29) AG , 0 Z ^ where A is foreign aid. Given that Z Z1 , the effect of aid would therefore be to increase 1 the optimal level of property rights institutions. If Z Z A;max , on the other hand, the 1 1 budget constraint is not binding and the ruling elite may choose to forgo the foreign aid in favor of a lower level of institutions. If it is the case that the ruling elite instead treats the aid as an additional source of rents in both periods, then the effect on the maximum will be the same as in the case of a natural resource boom as in (27), where r A enters in the place of r1 . This means that aid decreases the strength of property rights institutions in the case of an interior solution. Hence, foreign aid that is not used for investment in property rights institutions could result in worsening institutions and have a negative impact on economic development. This potential negative effect of aid appears in the theoretical model of Acemoglu et al. (2003), and seems to be somewhat supported by empirical evidence (Knack, 2000). Note that foreign aid also here increases scal revenue (by the amount Z0 A=Z and thus Z in an 1 upper boundary solution, but to a smaller extent than in (29). 3.4. Trade liberalization International trade is another factor that may inuence the ruling elites maximization problem. We assume that trade will inuence the price of the modern sector output and
~ If the shock to natural resource rents is unanticipated it will have not affect Zr;max or Z1 . As a result, neither 1 the local maximum nor the budget constraints change.
23

ARTICLE IN PRESS
1916 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

that a liberalization is usually associated with a fall in modern sector prices. This will be the case if the domestic price of manufactures before trade liberalization is higher than the world price. The effect of this price change depends on the timing of the liberalization. Start with the scenario where trade liberalization occurs in the second period, so that p0 1; p1 o1. Here, the rst period budget constraints will remain the same as in (24), while the second period budget constraint becomes ~ tZ 1 3 p3 c1 t2 1

~ Z1 r G 0 Z

(30)

and the equation for the interior maximum becomes q 1 1 2p3 Zd1 t2 Z 0 r=Z 1 Z p;max . 1 2 p3 Zd1 t 1

(31)

We can easily conrm that Z p;max increases with p1 . The intuition is that a lower price level 1 decreases labor supply and makes modern sector production less attractive in relative terms than rent seeking. If trade liberalization occurs in the rst period, and p0 Ep1 po1, then the rst period budget constraint becomes s   ^ ^ Z 0 Z 0 Z 2 tZ 3 p3 c1 t2 Z0 r G . (32) Z1 0 1 Z Clearly, a fall in p lowers the maximum feasible change in property rights institutions. However, Zp;max will also increase with p in the same way as in (31). In other words, trade 1 liberalization that is accompanied by a permanently lower price for modern sector goods will weaken property rights institutions if we initially have an interior or an upper boundary solution. The results are of course reversed if a trade liberalization causes an increase in modern sector prices. Another potential impact of trade liberalization is to increase the volume of natural resource extraction, hence increasing the rents from natural resources.24 In this case, the effect of trade liberalization will be the same as in the case of a natural resource boom as in (27). 3.5. Impediments to institutional efcacy Former colonies typically face many obstacles to institutional change, such as low population density, a complex geography, and great ethnic diversity (see for instance Herbst, 2000). These factors can inuence the ruling elites optimal choice of property rights institutions in two ways. First, it may be that y41 (rather than y 1 as assumed previously). This corresponds to a greater amount of effort being required on the ruling elites part to bring about institutional change, with the local
For example, Chichilnisky (1994) presents a model of international trade that demonstrates that countries with poorly dened property rights will appear to have a comparative advantage in resource-intensive production, even when technology, endowments and preferences are the same in all countries. This can, in turn, lead to overextraction of natural resources in the countries with low levels of property rights.
24

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1917

maximum dened by q y y2 2Zd1 t2 yZ0 r=Z y;max Z1 . Zd1 t2

(33)

The effect of an increase in y is to make any change, be it investment or disinvestment, less attractive.25 Another possibility is that these obstacles cause c41. This corresponds to an increase in the cost of institutional change via the rst period government budget. In this case, the budget constraint is as given in (21). It is directly evident that an increase in c will lower the amount of institutional change that is attainable.26 In the worst case, high costs of institutional change may compel a ruling elite to choose a low property rights equilibrium over a high property rights equilibrium. We have assumed that Z increases modern sector output at a 1:1 ratio. It is possible, however, that the effect of Z on modern sector output is less than unity. There is some theoretical support for the notion that the effectiveness of property rights on modern sector output is positively related to the state of technology (Demsetz, 1967; Alchian and Demsetz, 1973), so that low technology may decrease the effectiveness of property rights on modern sector output. Other potential factors include low levels of human capital accumulation, missing markets, limited access to credit, a high degree of ethnic fractionalization and geographical impediments. In this case, output in the modern sector would be e3 Z3 d1 t2 , where eo1. The effect on the second period budget constraint and 0 the optimum solution would then be identical to (30) and (31), while the rst period budget constraints become s   ^ ^ Z 0 Z0 Z 2 tZ 3 e3 c1 t2 Z 0 r G : Z1 (34) 0 1 Z The effect of eo1 is to lower the maximum feasible investment in property rights institutions, while at the same time lowering the level of property rights institutions at which the maximum occurs. Finally, an increase in g, the marginal productivity of household production, will have the opposite general effect of e on the budget constraints and the unconstrained optima.27 The above impediments to institutional efcacy all have in common that they restrict the scope of attainable institutional change compared to what would otherwise be the case. Further, in the case when output is adversely affected, the ruling elite may be more inclined to choose a low level of property rights institutions. 4. Conclusion In this article, we have attempted to model the decision process of property rights (dis)investment by an autocratic ruling elite. We have been motivated by the previous literature that has emphasized the crucial role institutions play in economic development, as well as by institutional changes that have taken place in former European colonies since
However, the sign of the partial derivative of (33) with respect to y can be either positive or negative. Whether this results in higher or lower level of Z depends on whether the ruler optimally strengthens or 1 weakens property rights institutions. 27 The only difference being that g will occur in quadratic form rather than cubic, as is the case with e.
26 25

ARTICLE IN PRESS
1918 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

independence. Further, the model has been motivated by the literature on the natural resource curse and the observation that many resource-poor countries in Asia have shown a considerably stronger economic performance than resource-rich African countries despite similar initial levels of institutional quality. The main insight of our model is the existence of a potential trade-off for an autocratic elite between strong institutions of private propertywhich increase revenue from the modern sectorand weaker property rights that facilitate the personal appropriation of rents from a natural resource sector. We show that even a completely self-interested ruling elite may have incentives to strengthen property rights if the modern sector is relatively protable. Additionally, the model retains a component of institutional persistence, which is present in much of the literature on the origins of institutions in former colonies. Although we have assumed that the ruling elite is only interested in personal enrichment, one could easily alter the model to t an altruistic ruling elite. In this case, the rent appropriation part of the utility function would disappear, and Z would measure the ruling elites willingness to expanding the modern sector. The cost of an investment would then set the limit to how much the elite could invest; variations in institutional investment would thus depend on initial levels of institutions. Although the article discusses a number of extensions to the basic model such as the impact of foreign aid and trade liberalization, we believe that several other aspects might be fruitfully analyzed within the framework outlined above. For instance, we have only briey touched upon the issue of why larger former colonies seem to have been disadvantaged in terms of institutional development. An econometric analysis is clearly needed in order to understand the exact channels of causation. A further line of inquiry might be the impact of stronger property rights on human capital accumulation, an issue that we have not dealt with at all here. Neither have we explicitly considered the possibility that the ruling elite initiates other types of institutional changes in the modern sector that are detrimental to the economy, such as the pursuit of import substitution strategies or entry restrictions into the modern sector. However, we conjecture that the opposing forces of a modern and a natural resource sector as modelled in this article may shed some light also on this issue. All else equal, it seems likely that a newly independent regime with a strong ow of mineral rents and a sense of self-sufciency is more inclined to adopt inward-oriented policies than a resource-poor country. Clearly, the curse of natural resources is a multi-faceted phenomenon that deserves further attention.

Acknowledgment We are grateful for comments from Arne Bigsten, Anne Boschini, EER editor Thorvaldur Gylfason, Anke Hoefer, Olof Johansson-Stenman, Johanna Jussila Hammes, Kalle Moene, Violeta Piculescu, two anonymous referees, and seminar participants at the Goteborg University, the Stockholm University, the EEA Meeting in Amsterdam, DET Workshop at the University of Copenhagen, the Malmsten Research Group, the Nordic Workshop in Development Economics, and the CSAE Conference at the Oxford University. Congdon Fors and Olsson both gratefully acknowledge nancial support from Vetenskapsradet and from the Wallander-Hedelius and Malmsten Foundations, respectively.

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1919

Appendix A A.1. Description of variables used in Figs. 1a, b, and 2

Variable Rule of law in 2004 Nrg_Min1999/GNI Logarea Log Settler Mortality Latitude Sub-Saharan Africa dummy Neo-Europe dummy Executive constraints

N 127 112 127 69 127 127 127

Max 1.93 0.36 16.12 7.99 60 1 1 7

Min 2.31 0 3.04 2.15 0 0 0 1

Mean 0.22 0.033 10.99 4.69 15.28 0.36 0.03

StDev 0.89 0.075 3.08 1.22 10.0 0.48 0.17

Description: Sample: 127 former colonies. Rule of law in 2004: Source: Kaufmann et al. (2005). Nrg_Min1999/GNI : Energy and mineral rents as a share of GNI in 1999. Source: World Bank (2004). Logarea: Log of total country area. Source: World Bank (2004). Log Settler Mortality: Log of annual mortality among soldiers and bishops per 1000 individuals during colonial times. Source: Acemoglu et al. (2001). Latitude: Distance from equator in absolute latitude degrees. Source: World Bank (2004). Sub-Saharan Africa dummy: Source: World Bank (2004). Neo-Europe dummy: Dummy for Australia, Canada, New Zealand, and the United States. Source: Acemoglu et al. (2001). Executive constraints: Degree of effective constraints (in terms of rule of law, constitutional sharing of power, etc.) on the executive. Note: All data are available from http://www.hgu.gu.se/item.aspx?id=2465 or upon request. A.2. Proof of Proposition 2 Proof. (a) Whether institutions are strengthened or weakened is determined by the sign of q l;max Z1 Z 0 1=2Zd1 t2 1 1 4Zd1 t2 Z0 r=Z Z 0 . A straightforward manipulation of this expression shows that Z l;max Z 0 40 if 1 2Zd1 t2 Z 0 4 1 q 2 . By squaring both sides, we receive 1 4Zd1 t2 Z0 1 4Zd1 t Z 0 r=Z which in turn implies that 2Zd1 t2 Z0 2 41 4Zd1 t2 Z 0 r=Z, l;max 2 2 Z0 40 if Zd1 t2 Z 2 4r=Z, which is the essence of (a). Zd1 t Z 0 4r=Z. Thus Z1 0 (b) From (25) and (24), we know that the greatest feasible institutional change in either q ^ ^ direction is Z1 Z 0 Z 0 Z 1 2tZ 3 c1 t2 Z 0 r=Z G40, while the minimum 0

ARTICLE IN PRESS
1920 H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921

~ ~ feasible level of Z 1 is dened by tZ1 3 c1 t2 Z 1 r=Z G 0. From these expressions, it is easily seen that the results in (b) regarding r; Z 0 ; and G apply. Since to1, the multiplicative term t1 t2 achieves its maximum at t 1. & 3

References
Acemoglu, D., Johnson, S., 2005. Unbundling institutions. Journal of Political Economy 113 (5), 949995. Acemoglu, D., Robinson, J., 2005. Persistence of Power, Elites, and Institutions. MIT, Cambridge, MA, Mimeo. Acemoglu, D., Johnson, S., Robinson, J.A., 2001. The colonial origins of comparative development: An empirical investigation. American Economic Review 91 (5), 13691401. Acemoglu, D., Johnson, S., Robinson, J.A., 2002. Reversal of fortune: Geography and institutions in the making of the modern world income distribution. Quarterly Journal of Economics 117 (4), 12311294. Acemoglu, D., Johnson, S., Robinson, J.A., 2005. Institutions as the Fundamental Cause of Long-run Growth. In: Aghion, P., Durlauf, S. (Eds.), Handbook of Economic Growth. Amsterdam, North-Holland. Acemoglu, D., Robinson, J., Verdier, T., 2003. Kleptocracy and divide-and-rule: A model of personal rule. NBER Working Papers No. 10136. Aghion, P., Alesina, A., Trebbi, F., 2004. Endogenous political institutions. Quarterly Journal of Economics 119 (2), 565611. Alchian, A.A., Demsetz, H., 1973. The property right paradigm. Journal of Economic History 33 (1), 1627. Bates, R.H., 1981. Markets and states in tropical Africa: The political basis of agricultural policies. University of California Press, Berkeley. Bates, R.H., Collier, P., 1993. The politics and economics of policy reform in Zambia. In: Bates, R.H., Krueger, A. (Eds.), Political and Economic Interactions in Economic Policy Reform. Blackwell, Oxford (Chapter 9). Beck, T., Laeven, L., 2006. Institution building and growth in transition economies. Journal of Economic Growth 11 (2), 157186. Besley, T., 1995. Property rights and investment incentives: Theory and evidence from Ghana. Journal of Political Economy 103 (5), 903937. Bigsten, A., 2001. Policy-making in resource rich countries: Lessons from Zambia. World Economics 2 (3), 139153. Bigsten, A., Moene, K., 1996. Growth and rent dissipation: The case of Kenya. Journal of African Economies 5 (2), 177198. Chichilnisky, G., 1994. North-South trade and the global environment. American Economic Review 84 (4), 851874. Coase, R.M., 1960. The problem of social cost. Journal of Law and Economics 3, 144. Congdon Fors, H., Olsson, O., 2006. Endogenous Institutional Change After Independence. Extended working paper version, available on personal website: hhttp://www.hgu.gu.se/item.aspx?id=2465i. Dalgaard, C.-J., Olsson, O., 2007. Windfall Gains, Political Economy, and Economic Development. Journal of African Economies, forthcoming. Demsetz, H., 1967. Toward a theory of property rights. American Economic Review 57 (2), 347359. de Soto, H., 2000. The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else. Black Swan Books, London. Djankov, S., Glaeser, E., La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2003. The new comparative economics. Journal of Comparative Economics 31, 596619. Do, Q.-T., 2004. Institutional Trap. Policy Research Working Paper 3291, World Bank. Easterly, W., Levine, R., 2003. Tropics, germs, and crops: How endowments inuence economic development. Journal of Monetary Economics 50, 339. Firmin-Sellers, K., 1995. The politics of property rights. American Political Science Review 89 (4), 867881. Frankel, J., Romer, D., 1999. Does trade cause growth? American Economic Review 89 (3), 379399. Glaeser, E., la Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2004. Do institutions cause growth? Journal of Economic Growth 9, 271303. Glaeser, E.L., Shleifer, A., 2002. Legal origins. Quarterly Journal of Economics 117 (November), 11931230. Gylfason, T., 2001. Natural resources, education and economic development. European Economic Review 45, 847859.

ARTICLE IN PRESS
H. Congdon Fors, O. Olsson / European Economic Review 51 (2007) 18961921 1921 Hall, R., Jones, C., 1999. Why do some countries produce so much more output than others? Quarterly Journal of Economics 114 (1), 83116. Herbst, J., 2000. States and Power in Africa. Princeton University Press, Princeton, NJ. Kaufmann, D., Kraay, A., Mastruzzi, M., 2005. Governance matters IV: Governance indicators for 19962004. World Bank Policy Research Working Paper 3630. Knack, S., 2000. Aid dependence and the quality of governance. World Bank, Policy Research Working Paper #2396, Washington, DC. Krueger, A.B., Lindahl, M., 2001. Education for growth: Why and for whom? Journal of Economic Literature 36 (December), 11011136. Leite, C., Weidmann, J., 1999. Does mother nature corrupt? Natural resources, corruption and economic growth. IMF Working Papers 99/85. Marshall, M.G., Jaggers, K., 2003. Polity IV project: Political regime characteristics and transitions, 18002003 hhttp://www.cidcm.umd.edu/inscr/polity/i. Mehlum, H., Moene, K., Torvik, R., 2006. Institutions and the resource curse. Economic Journal 116 (508), 120. Ndikumana, L., Boyce, J.K., 1998. Congos odious debt: External borrowing and capital ight in Zaire. Development and Change. 29, 195217. North, D., 1981. Structure and Change in Economic History. WW Norton and Company, New York. North, D., 1990. Institutions, Institutional Change and Economic Performance. Cambridge University Press, Cambridge. Olsson, O., Congdon Fors, H., 2004. Congo: The prize of predation. Journal of Peace Research 41 (3), 321336. Olsson, O., Hibbs, D.A., 2005. Biogeography and long-run economic development. European Economic Review 49 (4), 909938. Rajan, R., Zingales, L., 2006. The persistence of underdevelopment: Institutions, human capital, or constituencies? NBER Working Papers 12093, NBER. Sachs, J., Warner, A.M., 1997. Natural Resource Abundance and Economic Growth. Harvard University, Mimeo. Sachs, J., Warner, A.M., 2001. The curse of natural resources. European Economic Review 45, 827838. Sala-i-Martin, X., Subramanian, A., 2003. Addressing the natural resource curse: An illustration from Nigeria. NBER Working Papers, 9804, NBER. Sokoloff, K.L., Engerman, S.T., 2000. History lessons: Institutions, factor endowments, and paths of development in the new world. Journal of Economic Perspectives 14 (3), 217232. Sonin, K., 2003. Why the rich may favor poor protection of property rights. Journal of Comparative Economics 31, 715731. Svensson, J., 1998. Investment, property rights, and political instability: Theory and evidence. European Economic Review 42 (7), 13171341. Tornell, A., 1997. Economic growth and decline with endogenous property rights. Journal of Economic Growth 2, 219250. Woolcock, M., Pritchett, L., Isham, J., 2001. The social foundations of poor economic growth in resource-rich countries. In: Auty, R.M., (Ed.), Resource Abundance and Economic Development. Oxford University Press, Oxford (Chapter 5). World Bank, 2004. World Development Indicators CD-ROM, Washington.