write summary one paragraph for each article

The Politics of Fertility and Economic Development

Yi Feng

Jacek Kugler

and

Paul J. Zak

Claremont Graduate University

This paper presents a formal model that characterizes the two faces of development — persistent povert y, and industrialization and rising incomes—and establishes that the interaction between politics and eco- nomics determines which path a nation travels. We demonstrate that political factors affect fertility decisions so that a one-time disturbance compounds across generations, impacting a country’s entire develop- ment trajectory. Modeling strategic multiobjective policy-setting by the government, we derive a new concept of political capacity and prove that a sufficient amount of political capacity is necessary to escape a poverty trap and develop the economy. Empirical tests for a sample of 100 countries from 1960 to 1990 provide strong support for the predic- tions of the formal model. In particular, we show that both political stability and political capacity significantly inf luence birth rates. We conclude that politics can be either a stimulant or a barrier to economic development.

This paper presents a formal model that characterizes the two faces of development—a poverty trap with persistent economic stagnation, and industri- alization and rising incomes—and establishes politics as a fundamental determi- nant of the trajectory taken by a nation. We demonstrate that policy choices at a single point affect a country’s development path by impacting fertility decisions across generations. The primary policy implication of our analysis is that sus- tained economic development has political prerequisites.

The model in this paper formalizes and extends a number of alternate approaches to demographic change and economic development. First, we cap- ture insights from the modernization literature that show that rising incomes lead to lower fertility rates ~Thompson, 1929; Notestein, 1945; with recent work by Bongaarts, Mauldin, and Phillips, 1990; Camp, 1993; Freedman, 1994; Sind- ing, Ross, and Rosenfield, 1994!. Additionally, our research is related to economic theories of development in which human capital is the engine of growth ~Lucas,

Authors’ note: This research is supported by a grant from the National Science Foundation ~SBR-9730474!. We thank Doris Fuchs, Monika Gruter Morhenn, participants at the 1999 International Studies Association conference, and five anonymous ISQ reviewers for comments, and Brian Buford-Efird, Chi Choi, and Natalia Maric for expert research assistance.

International Studies Quarterly ~2000! 44, 667–693.

© 2000 International Studies Association. Published by Blackwell Publishers, 350 Main Street, Malden, MA 02148, USA, and 108 Cowley Road, Oxford OX4 1JF, UK.

1988; Galor and Zeira, 1993; Stokey, 1996!, though we extend these studies by modeling the impact of fertility choices on the accumulation of human capital. As a result, our model is also related to the endogenous fertility literature ~Becker and Tomes, 1976; Becker and Barro, 1988; Becker, Murphy, and Tamura, 1990; Tamura, 1996; Azariadis, 1997; Galor and Tsiddon, 1997!. Furthermore, this research contributes to the literature examining the political economy of economic growth ~e.g., Barro, 1997! by focusing on the effects that political institutions have on fertility decisions. For example, the model provides an explanation for the baby booms that follow severe conf licts, a well-established empirical finding ~Organski et al., 1984!. Finally, we establish the mechanism of action through which increases in political capacity reduce fertility ~Arbetman and Kugler, 1997! by using a measure of political capacity based on strategic multiobjective policy-setting by the government.

In this paper, we present a formal equilibrium model that integrates political decision-making with population dynamics and economic development, reveal- ing the interdependence of each on the other. This permits a derivation of necessary conditions for development, which include sufficient political capacity and adequate political stability. The formal model shows that the equilibrium level of births depends nonlinearly on political stability, political capacity, income, and education. Analyzing annual data for a sample of 100 countries over the period of 1960 through 1990 shows that the formal model’s predictions for the interaction between fertility, politics, and economics are strongly supported. Economic factors, political capacity, and political stability all inf luence birth rates and therefore the prospects for economic development.

Equilibrium Dynamics and Predictions of the POFED Model

In this section, we present and analyze the equilibrium dynamics of a political- economic model in which decisions cumulate over generations. Figure 1 con- tains a schematic representation of the choices made by the three types of actors in the model: individuals, firms, and the government. By modeling all three sectors, we present a self-contained ~general equilibrium! political economy in which population and income endogenously grow ~or contract!, and both depend on evolving political structures. Each aspect of the model is presented and solved in Appendix 1, with all variables presented in Table A1. In this section we describe the model and characterize the derived equilibrium dynamics linking politics, fertility, and economic development, which we call the POFED model.

Consider a country with a large number of individuals who live three periods in overlapping generations. At each point in time, children, young adults, and older adults are alive. Each generation has a different level of human capital, h , while individuals within a generation are, for simplicity, identical. The economy has a single good that can be used for consumption or investment in physical capital, K .

Agents maximize lifetime utility during the two periods of adulthood, subject to a budget constraint in each period. During young adulthood, individuals work for firms paying a proportion of labor income t [ ~0,1! to the government as taxes, using the remaining income to fund their own and their children’s con- sumption, and to save for old age. During old age, agents are retired and consume the principal and interest on their savings.

Besides choosing how much to consume and save during young adulthood, individuals also choose how many children to have.1 In this model, children

1 To keep this rather complicated model as simple as possible, children are produced by parthenogenesis ~asexual reproduction!. This permits us to avoid the issue of marriage matching. For a model of the search for a marriage partner, see Burdett and Coles, 1997. We also ignore issues such as infertility and infant mortality, though these are ref lected in our empirics.

668 Politics of Fertility and Economic Development

acquire human capital from their parents and make up the labor force when they are adults. Human capital, along with physical capital ~plant and equip- ment!, accumulates or decumulates endogenously over time based on choices made by individuals, firms, and the government.

Two political factors impact individual choices: political instability and politi- cal capacity. Every government has as its goal self-perpetuation ~Magee, Brock, and Young, 1989; Arbetman and Kugler, 1995; Alesina, Roubini, and Cohen, 1997!. Since political instability increases the likelihood that the government will be overthrown, the maintenance of public order is the highest priority when setting government policy. The second policy goal is to use tax revenue to increase political support for the regime. This occurs when policies raise incomes for the general public, and when a proportion of tax revenue is transferred to political elites. Economic growth is a natural goal of politicians in democracies and autocracies, because raising individuals’ incomes sustains support for the government ~Tufte, 1978; Fiorina, 1981; Lewis-Beck, 1990! and, in addition, increases tax revenues and therefore the ability of the government to enact new policies.2 Combating political instability is reactive; that is, when the stochastic portion of political instability, e, is observed, the government reacts by varying police funding. On the other hand, enhancing growth is proactive. Policies that enhance growth require an expenditure plan prior to implementation.

Denote political instability as S , which is the proportion of a country’s physical capital destroyed in antigovernment violence ~Alesina, Ozler, Roubini, and Swagel, 1996; Chen and Feng, 1996; Zak, 1997!. Note that S is not the number of dem- onstrations, but the impact of political instability on the economy. Because our

2 McGuire and Olson ~1996! show that only predatory autocrats with short time-horizons will set policies that will cause the economy to contract rather than grow.

Fig. 1. Schematic presentation of the formal model.

Yi Feng, Jacek Kugler, and Paul J. Zak 669

purpose is to characterize how political instability affects fertility and economic development, we do not model the incentives for instability, but rather treat it as an aggregate phenomenon with two effects. First, political instability impacts individual decisions by reducing the physical capital stock, which attenuates equilibrium labor income ~as section A1.2 shows in Appendix 1!. Second, polit- ical instability impacts government policy-setting by pulling resources away from other programs as the government seeks to maintain itself in power ~as section A1.3 demonstrates!. Resources spent to maintain the regime cannot be spent on productive activities and constitute a deadweight economic loss.

The second political factor, political capacity x, is the effectiveness of the gov- ernment in implementing policy ~Arbetman and Kugler, 1997!. Capable govern- ments enhance the productivity of private firms by choosing policies that encourage economic efficiency. Arbetman and Kugler ~1997! show that politically capable governments improve a variety of economic activities such as attracting invest- ment, enhancing trade, and reducing inf lation. Because our concern is with development, we restrict political capacity to have a single effect, raising labor productivity.

In section A1.3 in Appendix 1, we derive a country’s maximal level of political capacity that a government can attain at any point in time, xt

*, with t denoting time, along with optimal spending on the police to maintain public order, pt

*, and the resulting tax rate, tt

*. The optimal policy set is a Nash equilibrium of the repeated game played by the government and citizens, after a proportion of tax revenue, s [ ~0,1!, is paid to political elites. Payments to elites are necessary for the government to remain in power and are therefore paid before other policies are funded. Next, the police are funded to keep the regime in power, with the residual tax revenue funding political capacity. In this way political capacity captures political constraints, and therefore policy discretion of the government. Once the government has chosen the optimal tax rate, police spending, and political capacity, individuals execute their optimal choices for births and savings.

The derivation of optimal political capacity reveals its positive dependence on tax revenue, and negative dependence on payments to political elites and the virulence of political instability. In practice, it is unlikely that governments can reach maximum political capacity because of additional political constraints that we have not modeled. When xt , xt

*, the specification of the production func- tion ~A1.6!, and of the law of motion for physical capital ~A1.12!, shows that output net of taxes is increasing in political capacity. Moreover, the ratio xt 0xt

*

can be viewed as a measure of the political constraints on policy-setting, which is inversely related to a government’s discretion.3 That is, a government with polit- ical capacity xt

* has maximal discretion in setting policy. We show in Appendix 1 that political instability both reduces tax revenue and

changes the optimal mix of government expenditures toward police funding and away from growth-enhancing policies ~by A1.14! and thus reduces political capac- ity. Governments can raise their capacity by increasing the tax rate t, but this causes a drag on the economy by reducing net-of-tax income, and therefore savings and capital formation. Thus, governments face intertemporal trade-offs between funding policies that stabilize current political and economic environ- ments and policies that can shape the future of these environments.

To understand policy-makers’ dilemma when setting policies that affect the economy, ref lect on the following thought experiment. Consider a country in which political instability is high. Because a government’s first priority is to stay in power and sufficient instability can threaten the ruling regime’s longevity ~Feng and Zak, 1999!, the first use of tax revenue is police funding to maintain

3 The derivation of x0x* provides a formal basis for the relative political capacity measure of Arbetman and Kugler ~1997!.

670 Politics of Fertility and Economic Development

public order. Secondarily, tax revenue is used to raise individual incomes, for example, by expanding infrastructure expenditures that raise productivity. Ris- ing incomes increase support for the ruling regime. Thus, the government’s objective when setting policy is to maintain itself in power by quelling rebellions and by raising popular support. We demonstrate in section A1.3 that govern- ments faced with highly unstable political environments allocate a large portion of expenditures to security and have little remaining revenue for programs to boost private incomes since government spending is constrained by tax receipts ~equation A1.10!. Thus, unstable governments produce less output than stable, capable governments. Because income is a fundamental determinant of birth rates ~by equation A1.5! political capacity and instability affect birth rates via their impacts on labor income. Furthermore, since in each period fertility deci- sions are made by those who were children in the prior period, political effects compound over generations as family size alters per child parental investments ~by equation ~4! below!. This changes a country’s development trajectory and the government’s stream of tax revenue, producing a spillover effect on future generations.

Next, we construct and analyze the sequence of equilibria of the POFED model. In Appendix 1, we derive the following utility maximizing time t optima for the desired number of children ~births!, bt

*, and for savings from middle to old age, a t11

*,

bt * 5 Max HE g~1 1 g!~1 2 a!~1 2 t!Dkt a ~1 2 St !a ~ xt ht !12a ,1J ~1!

a t 11 * 5 E $ b ~1 2 a!~1 2 tt !~K t ~1 2 St !!

a ~ xt ht ! 12a % ~2!

Equation ~1! shows that births decrease as human capital, h , and physical capital, K , rise since an individual’s labor income is increasing in both types of capital ~by equation A1.8!. Relation ~1! also shows that births increase when taxes, t, rise ~which reduces net labor income!, when the preference for children, g, becomes stronger, and when the proportional cost of children, D, falls.4 Political factors also affect births in ~1!. An increase in political instability, S , raises births since it reduces labor incomes and thus reduces the opportunity cost of raising children ~by equations A1.4 and A1.8, the opportunity cost of children is the income foregone while raising them!.5 Political capacity, x, decreases births in ~1! since government policy that raises economic efficiency increases individual incomes. Optimality condition ~1! is consistent with empirical evidence showing that as incomes rise, birth rates fall ~Birdsall, 1988!. To simplify the analysis, the mini- mum family size in relation ~1! is set to unity so that population is constant in the limit. Relaxing this assumption does not change the results.

Optimal old-age savings, equation ~2!, shows that an increase in political capac- ity ~holding the tax rate constant! raises labor incomes and therefore savings, while an increase in political violence reduces income and therefore savings. All these implications find support in empirical studies of savings behavior ~Blinder and Deaton, 1985; Venieris and Gupta, 1986; Arbetman and Kugler, 1997!.

We now turn to the demographic structure. Let Nt denote the working pop- ulation of young adults at time t . Recall that only young adults reproduce and that the old do not work. As a result, children in the current period make up the

4 The remaining parameter in ~1!, a [ ~0,1!, is the marginal productivity of physical capital ~see section A1.2!. Births are highly nonlinear in a.

5 Note that the expected value operator appears in ~1! and ~2! because political instability S has a stochastic element to it which corresponds to the partially unpredictable nature of antigovernment violence.

Yi Feng, Jacek Kugler, and Paul J. Zak 671

labor force of young adults in the subsequent period. Thus, the evolution of the working population is given by

Nt 11 5 Nt bt *, ~3!

where bt * is given by equation ~1!. Equation ~3! simply shows that the number of

young adults at time t 1 1 is the aggregate births at time t , Nt bt *.

Next, we specify the dynamics of human capital. Using the generational struc- ture of the model, we allow parents to transmit some of their human capital to their children. Since children’s inherited traits are more fully expressed when parental nurturing is high, family structure inf luences the intergenerational transmission of human capital. Hanushek ~1992! and Downey ~1995! show that as the number of siblings in a family increases, adult income and educational attainment of children falls. Thus, when family size is small, parental nurturing per child is higher and their adult productivity is enhanced. Combining the effects of household environment with the inheritability of human capital, a child’s human capital is, on average, increasing in his or her parent’s human capital and decreasing in the number of children in a family.6 Equation ~4! captures this structure where the human capital ht11 of each child is related to the parental human capital, ht , and the number of siblings in his or her family, bt ,

ht 11 5 vht bt

u . ~4!

The parameter v in ~4! is the maximum rate of intergenerational human capital transmission, while u is the dilution effect resulting from multiple siblings ~b . 1! vying for their parent’s attention.7

The final component of the equilibrium dynamics maps the saving decisions of individuals, equation ~2!, into the demand for investment by firms ~Appendix 1, section A1.2!. Because only young adults save, aggregate savings from time t to time t 1 1 is Nt a t11

*, which firms use to form the physical capital stock at time t 1 1. That is,

K t 11 5 Nt a t 11 * 5 E $ b ~1 2 a!~1 2 tt !~K t ~1 2 St !!

a ~ xt Nt ht ! 1 2 a %. ~5!

Observe in equation ~5! that next period’s physical capital stock, K t11 , is increas- ing in the current period’s physical and human capital stocks, is decreasing in taxes and political instability, and increases as political capacity rises. Thus, politics affects the rate of physical capital accumulation in equation ~5!, the number of births in equation ~1!, and the rate of human capital accumulation in

6 This argument is fully developed in Zak, 1999. Because Behrman and Taubman ~1989! show that 81% of educational attainment is attributable to one’s genetic endowment, we abstract from modeling the effect of formal education on human capital. Many studies have shown that IQ or “cognitive ability” predicts earnings ~Cawley et al., 1996; Ceci and Williams, 1997; Murray, 1997!. This is consistent with causation running from inherited ability, to human capital, to earnings. Note that only the transmission of workplace skills are modeled, not other personal traits.

7 The law of motion for human capital ~4! accords well with that used by Lucas ~1988!, and collapses to match Lucas’s exactly when the chosen number of children is one. A more general human capital accumulation function is contained in Bong, Wang, and Yip, 1996. Galor and Tsiddon ~1997! and Tamura ~1996! include nonconvexities in the accumulation of human capital, and several of the predictions that come from these models match those that come from ours, as the dilution effect in ~4! induces a nonconvexity. Zak ~1999! examines the stochastic transmis- sion of inherited human capital.

672 Politics of Fertility and Economic Development

equation ~4!. This demonstrates the interaction between politics, economics, and fertility in all parts of the model.

The equilibrium for the POFED model specifies a sequence of optima for all actors for each time t 5 1, 2, . . . along with a number of technical conditions ~see Appendix 1!. This equilibrium is completely specified by the dynamic sequence of the working population, equation ~3!; of human capital, equation ~4!; and of physical capital ~5!, all of which are constructed from optimal behaviors by individuals, firms, and the government. The three-dimensional dynamics can be reduced to two dimensions by writing the evolution of physical capital ~5! in per worker terms using the population evolution equation ~3! ~note the evolution of human capital ~4! is an individual, rather than aggregate, relation!. Defining per worker physical capital as kt 5 K t 0Nt , equation ~5! can be written in per worker terms,

kt 11 5 E $ b ~1 2 a!~1 2 tt !~K t ~1 2 St !! a ~ xt ht !

12a %0bt *. ~6!

Now, the equilibrium of this political-economic system is given by the evolu- tion of human capital equation ~3! and of per worker physical capital, equation ~6!, along with government choices for political capacity, x, police expenditures, p , and taxes, t. In the next several sections we characterize how politics affects fertility and economic development using the POFED model.

Development Trajectories

As indicated at the outset, the POFED model produces two types of development paths, a poverty trap and a balanced growth path. When physical or human capital is below threshold values derived below, the economy is caught in a low-income stationary equilibrium, a poverty trap. This occurs because a paucity of physical or human capital results in low labor incomes and birth rates are so high that human capital decumulates over generations due to large family size and low parental investments in children. Formally, using the law of motion for human capital ~4!, if bt . v

10u, then ht11 , ht and human capital decumulates over time. As a result, output contracts, rather than growing as a reduction in human capital, causes time t 1 1 physical capital and output to fall ~by produc- tion process ~A1.10! and the law of motion for physical capital ~6!!. A reduction in time t 1 1 labor income further raises births by ~1!, causing human capital to shrink at time t 1 2, initiating a self-reinforcing feedback loop. In the limit, as both types of capital and output decline, society is pushed toward a poverty trap.

Figure 2 plots the equilibrium dynamical equation for physical capital ~6! in kt 2 kt11 space.

8 It shows that if a country has little physical capital, then invest- ment is insufficient to sustain positive economic growth and the economy con- tracts into a poverty trap. Solving the equilibrium dynamical system ~3! and ~6! for the stationary point shown in Figure 2, the threshold for physical capital that determines whether the economy grows or contracts is

k* 5 bgv220u0D ~1 1 g!2. ~7!

The growth threshold for human capital can be derived using the same method. When a country’s physical capital is below k* labor income falls over time,

birth rates rise, and the economy falls into a poverty trap. When physical capital

8 For simplicity, human capital is not shown in Figure 2. The full dynamics of this model can be found in Zak, 1999.

Yi Feng, Jacek Kugler, and Paul J. Zak 673

exceeds k*, the economy grows endogenously, expanding rapidly during a period of transitional dynamics when birth rates fall and income and investment surge.

A second equilibrium trajectory, a balanced growth path, emerges when physical and human capitals are above thresholds that lead to a poverty trap. On this trajectory, portrayed in the far right of Figure 2, economic growth is endogenous— that is, it continues indefinitely without reaching a stationary equilibrium. The long-run rate of output growth is v, the maximum rate of transmission of human capital.9 On a balanced growth path, output is driven by new ideas that follow from the continuing accumulation of human capital.

The dynamics of the POFED model reveal that politics fundamentally impacts a country’s development trajectory. As depicted in Figure 2, low political capacity countries that are below the growth threshold contract rapidly into poverty; while those above this threshold grow more slowly in the transitional dynamics, taking longer to become fully developed ~i.e., to reach the balanced growth path!. Conversely, high political capacity countries exhibit surging growth in the transitional dynamics, rapidly becoming developed economies. Expectations of high instability have similar effects on economic growth by decreasing physical capital in production, which reduces incomes, causing births to rise, by equation ~1!, and reducing next period’s capital stock, by equation ~6!.

Political Capacity

A noteworthy implication of the POFED model is that the relationship between output growth and political capacity is nonmonotone. Recall that increasing

9 This long-run growth rate is derived by assuming the political capacity and political instability are constant in the long run, consistent with the evidence in Arbetman and Kugler, 1997.

Fig. 2. Growth paths with high and low political capacity.

674 Politics of Fertility and Economic Development

political capacity requires additional taxes which, by the savings function ~2! and law of motion for physical capital ~6!, reduce the savings by individuals and next period’s productive capital stock. In general, governments may under- or over- tax in an effort to raise their political capacity. This complex relationship between output and political capacity affects patterns of economic development.

For given levels of human and physical capital there is a threshold of political capacity that determines a country’s prospects for successful development. Using the law of motion for physical capital ~6!, it is clear that as political capacity becomes small ~ x r 0!, next period’s capital stock approaches zero, and is surely less than the growth threshold level, k*. Thus, an extended reduction in political capacity causes a country to fall into a poverty trap. Furthermore, a temporary change in political capacity for a country that is growing ~kt . k

*! alters a country’s growth path, and can cause it to contract into a poverty trap. This occurs if a reduction in political capacity from its base level xt , to the new level xt 2 h, satisfies,

h . xt 2 k *10~2 ~12a!!A210~2 ~12a!!kt

2a0~12a! ~1 2 St ! 2a0~12a!ht

21 . 0, ~8!

where A 5 b ~1 2 t!2D ~1 2 a!2. A one-time decrease in political capacity that satisfies inequality ~8! is more likely when physical and human capitals are low, and when political instability is high. A government that is politically incapable, therefore, causes the economy to falter, which by ~A1.13! further weakens the government, initiating a vicious circle of persistent political fragility and even- tual poverty. Moreover, equation ~A1.15! shows that rising birth rates reduce political capacity as public goods are shared among a larger population, further dimming the prospects for economic development.

When political capacity falls, but not enough to satisfy ~8!, the economy con- tinues to grow in the transitional dynamics, but converges to the balanced growth path ~i.e., becomes a developed economy! more slowly. As depicted by the dashed curve in Figure 2, a growth slow-down occurs in response to a one-time reduc- tion in political capacity since births increase and the accumulation of both human and physical capital slows. A country facing these circumstances will eventually develop, but without a politically capable government the develop- ment process is retarded. A similar derivation shows that high political instability slows a country’s pace of development and convergence to balanced growth.

Conversely, for a country in which the economy is contracting ~kt , k *! a

sufficient increase in political capacity, xt 1 h , xt *, changes its development

trajectory to one with positive and self-sustaining economic growth. The route that permits a country to escape poverty via political means requires a sufficient increase in political capacity, with the increasing being larger when physical and human capital are low, and political instability is high. Further, the new level of political capacity must not exceed the growth-maximizing level, xt

*. If xt 1 h . xt *, an increase in political capacity reduces growth as the economic drag from

higher taxes exceeds the benefit of increased production efficiency. Thus, there is a limited range of actions that governments can take to stimulate economic development.

Figure 2 suggests, and equation ~8! demonstrates, that changes in the direc- tion of a country’s development trajectory due to a change in political capacity are most likely for “middle-income” countries—those with kt near k

*. In these countries, political capacity is moderate and a change in capacity can radically alter economic performance. Poor countries have very little latitude to increase political capacity as they are constrained by low tax revenues. For this reason poor countries are more likely to remain trapped in poverty ~see section A1.3!. By contrast, high-income countries have large stocks of physical and human

Yi Feng, Jacek Kugler, and Paul J. Zak 675

capital and therefore high political capacity so that a change in capacity has little impact on the economy.

The most important implication of these findings is that politics can provide a means to alter the path of economic development. Unlike economic factors that evolve slowly, political factors often change rapidly, and these short-term f luctuations affect a country’s development trajectory. Political change is not the only trigger to development. As equation ~8! shows, high levels of per worker human or physical capital that provide a sufficient “push” to sustain develop- ment can, over a range, offset the effects of low political capacity. Unfortunately, as the experience of aid donors to developing societies attests, the stocks of human and physical capital are typically low in developing countries, and their augmentation is difficult. Thus, adequate political capacity is among the neces- sary conditions for economic development, and the one over which policy- makers have the most control.

The POFED model’s political dynamics produce both a poverty trap and a transition to balanced growth. This deduction is consistent with empirical find- ings showing that there are “convergence clubs” in cross-country data ~Pritchett, 1997; Quah, 1997!. As predicted by the POFED model, the data show that poor countries tend to stay poor. What we provide is a new political explanation for this finding, and a means for an exit. Consistent with the transitional dynamics of the POFED model, empirical studies show that middle-income countries either contract into poverty or grow rapidly and join wealthy countries. Our analysis provides an explanation for this pattern and demonstrates that political capacity determines whether a country is above or below the growth threshold. Finally, the empirical evidence indicates that developed countries grow at a roughly constant rate over time ~e.g., Razin and Yuen, 1993; Barro and Sala-i-Martin, 1997!. These results match the balanced growth dynamics in the model in which political factors have little effect on economic growth. Thus, including politics and fertility into a model of economic growth consistently accounts for the set of development trajectories within a single model. It is worthwhile to reemphasize that the dramatic impact of politics on economic development in the POFED model is contingent on politics affecting fertility decisions. Because of this depen- dence, even a one-time change in the political environment affects economic factors over many generations.

Fertility

In the POFED model, fertility is the linchpin that connects politics to long-run economic performance, with the implications of the model derived from rational decision-making by individuals. As a result, our empirical analyses will center on the model’s implications for births, bt

*. The main predictions for fertility from equation ~1! are that births decrease when ~i! political capacity, x, rises; ~ii! polit- ical stability, 1 2 S , rises; ~iii! human capital, h , increases; and ~iv! income per worker, y 5 Y0N , grows.10 We discuss each of these in turn.

The first, and most important, prediction of the POFED model is that births decrease when political capacity rises. Empirical confirmation of these effects was first reported over a decade ago and extended since ~Organski et al., 1984; Arbetman and Kugler, 1997!. The POFED model provides an explanation for these empirical findings by disclosing the joint dynamics of fertility, politics, and economic growth. We show that politically capable governments seeking political survival enhance human capital accumulation by spending on public investment, raising productivity, and thus labor incomes. As a result, capable governments

10 By the production function ~A1.6!, physical capital per worker and income per worker are proportional to each other so the births equation is equivalently specified in terms of physical capital or income.

676 Politics of Fertility and Economic Development

unwittingly initiate demographic transitions, leading to low birth rates, a stable population, and rapid economic development.11

The second implication of the model is that political instability has a positive impact on birth rates. Demographers have documented the baby booms that follow conf licts, causing population to rebound after even great wars ~Coale, 1975; Organski and Kugler, 1980!. The formal explanation we provide for this phenomenon is new. The POFED model shows that as political instability rises, labor income falls, reducing the opportunity cost of children. Since children provide utility, a parents’ best response when incomes fall is to increase the desired number of children. Aggregating such responses results in the baby booms that have echoes across generations.

Because political capacity raises output growth and therefore tax revenues, it partially determines the economic impact of instability. Figure 3 illustrates this by plotting equation ~1! for different levels of political capacity for an identically sized instability shock e. The figure shows that when political capacity is low, instability causes a large increase in births, which stalls economic development. This occurs because low political capacity governments allocate little revenue to output-raising expenditures when instability f lares up ~by equation ~A1.14!!, result- ing in low labor productivity and concomitantly, low wages. By equations ~1! and ~4!, such actions result in an increase in the number of births, which decreases the human capital transmitted to children. Therefore, the patterns depicted in Figures 2 and 3 for societies caught in poverty traps are associated with govern- ments that are politically unstable and have low political capacity. Conversely, as Figure 3 shows, high political capacity governments stimulate income growth— even with episodic instability—causing a demographic transition in which birth

11 Political capacity has a complex relation to births. Section A1.3 in Appendix 1 establishes that a country’s political capacity is constrained by both political and economic conditions, and as we showed in section 3.2, the level of political capacity, especially in middle-income countries, determines if nations grow into developed econ- omies or contract into poverty. Thus, politicians have some ability to inf luence fertility and the pace of economic development through their policy choices.

Fig. 3. Fertility, income, political capacity, and instability.

Yi Feng, Jacek Kugler, and Paul J. Zak 677

rates fall, stimulating rapid economic development. Therefore, the effect of political instability on fertility is countervailed in a high political capacity coun- try. The third implication is that high parental human capital reduces birth rates. With smaller families, each child receives greater parental investment, and as a result, has more human capital when he or she becomes an adult. The dynamic link to growth is that high human capital workers are more productive, receive higher wages, and themselves have fewer children, each of whom will, on aver- age, have high human capital.12

The final implication of the POFED model is that births decrease when income grows. This deduction provides formal support for a now standard finding in the modernization literature. Scores of empirical studies support this general rela- tionship ~e.g., Coale, 1975; Demeny, 1989!. The POFED model discloses the mechanism through which this result obtains. As incomes rise, the opportunity cost of time spent raising children increases so those parents optimally choose to have fewer children and invest more in each child. This is the primary reason for small families in developed economies, and, as the model shows, leads to a rapid accumulation of human capital across generations. On a balanced growth path, countries are fully developed “information economies” where low birth rates maximize the accumulation of human capital which is the engine of continued growth.

Statistical Specification and Measurement

Taking natural logs of the equilibrium births equation ~1! produces the following equation, which is used to test the propositions derived from the POFED model:

ln~bit ! 5 b0 1 b1 ln~ xit ! 1 b2 ln~1 2 Sit ! 1 b3 ln~ yit ! 1 b4 ln~hit ! 1 ei , ~9!

where subscript i indicates country, and t indicates year. Recall that the depen- dent variable, b , is birth rates, and the independent variables are political capac- ity, x; political stability, 1 2 S ; per worker output, y ; and human capital, h . Because of well-known measurement problems with physical capital k which appears in ~1!, we proxy physical capital with per worker output, y . This preserves the model’s structure since the production function ~A1.6! shows that physical capital and output are directly proportional to each other. Lastly, the parameter b0 is a constant that comprises the preference, cost, and production parameters from the model, and e is a white noise error term. Below we discuss the opera- tionalization of the variables in equation ~9!.

Births

A commonly used measure for births is the crude birth rate, defined as the number of children born per thousand of the population. The use of birth rates as the dependent variable helps to alleviate reverse causation. The source of these data is the United Nations.

Political Capacity

Government political capacity ref lects the ability of political elites to tap into human and material resources. Government elites mobilize this pool of resources to promote their objectives within the limits imposed by competing domestic political actors and by competitive pressures from the external environment ~Organski and Kugler, 1980:69!. Political capacity is proxied by a government’s

12 On the trade-off of child quality vs. quantity see Becker and Tomes, 1976 and Hanushek, 1992.

678 Politics of Fertility and Economic Development

ability to collect revenues. Governmental operations depend upon resources extracted from the population, as governments cannot survive without revenue to fund their programs. For this reason, taxes are indirect indicators of govern- mental presence. Failure to impose and extract taxes is one of the essential indicators of governmental inability to obtain and maintain support ~Organski and Kugler, 1980!.

In this study, we utilize the relative political extraction ~RPE! measure devel- oped by Arbetman and Kugler ~1995!, based on the difference between the observed taxation in a country and the expected taxation level determined by economic factors. A country of high political extraction presupposes a strong and capable government, which is likely to implement its policy effectively ~See Appendix 2 for operational details!.13

Political Stability

Political stability is derived from a measure of political instability. Political insta- bility is latent in the social and political system of a country. As measures of political instability, Feng ~1997! distinguishes between unconstitutional govern- ment change, major constitutional government change, and minor constitu- tional government change. He finds that unconstitutional government change ~such as a military coup d’état! has a pronounced negative consequence for economic growth. It is this type of government change that is utilized in the empirical tests.

Similar to Cukierman, Edwards, and Tabellini, 1992, and Feng, 1997, which extend early work by Barro ~1991!, we first measure political instability using a limited dependent variable model. The probability of unconstitutional govern- ment change is a function of ~i! economic variables measuring the recent eco- nomic performance of the government ~e.g., previous levels of inf lation, consumption, and income!; ~ii! political events accounting for significant polit- ical incidents that may signal an imminent government change ~e.g., riots, assas- sinations, general strikes, and revolutions!; ~iii! political structures indicating systemic stability ~e.g., the selection of the effective executive of the state, par- liamentary responsibility, and the effectiveness and selection of the legislature!; and ~iv! dummy variables grouping countries according to their continents to control for the systemic effects not explained by the model. From the fitted values of the logit model using pooled time-series cross-national data, the prob- ability of unconstitutional government change for each country in any given year in the data set is estimated ~see Appendix 3 for details!. Political stability is then calculated by subtracting the estimated probability of unconstitutional govern- ment change from one. Alternative measures of political instability in this paper include riots, assassinations, general strikes, and antigovernment demonstra- tions. While we believe that political instability in our theoretical model is best measured through a probability function, we shall also test the effects of these violent political acts on birth rates. The data on them are from Banks, 1996.

Income per Capita and Human Capital

The source for real GDP per capita is the Penn World Tables ~Summers and Heston, 1995!. This data set adjusts national income levels for purchasing power parity, so that the cost basis of expenditures is comparable across countries and

13 RPE is an imperfect measure of political capacity as derived in section A1.3 in which the government evaluates both the cost ~t! and benefit ~ x, p ! of programs given the political and economic environments. Higher taxes raise output if xt , xt

*, ∀t which we assume holds for all countries.

Yi Feng, Jacek Kugler, and Paul J. Zak 679

over years. The level of real GDP per capita is obtained from an equation based upon a country’s real domestic consumption relative to that of the United States.

Human capital is proxied by data on literacy. Literacy provides a measure of the quality of education, as opposed to, for example, years in school. The source of the literacy data is the Cross-National Time-Series Archive ~Banks, 1996!.

Empirical Evidence

Table 1 reports the result of a cross-country time-series regression analysis of equation ~9!.14 The data span from 1960 through 1990 for 100 countries. The theoretical model identifies variations that are examined in subsequent estima- tions. In the base statistical model, we use real GDP per capita as a control variable for the level of income, though it is also a policy variable in our theo- retical model. In Table 2 and Table 3, we use dummy variables to isolate differ- ential effects of politics on fertility regarding high-, middle-, and low-income countries. While Table 1 provides a general statistical framework to examine our theoretical position, Table 2 and Table 3 uncover additional nuances about and insights into our theory.

In Table 1, all independent variables are lagged to instrument the variables the theory identifies as jointly endogenous with births, as well as to capture dynamic changes in the underlying structure. The variation of the lag structure is provided in the first row of the table. While the first column tests the “con- temporaneous” effect of political institutions on birth rates, the remaining col- umns use the first through fifth order lags of the independent variables.15

The unit of lags is annual. As all variables are simultaneously determined ~i.e.,

14 Beck and Katz ~1995! show that some specially designed time-series methods ~e.g., the Parks method! can lead to a serious over-fitting problem in panel data, making it more likely to obtain a “statistically significant” parameter estimate. They also demonstrate that the statistical results from ordinary least squares ~OLS! estimation is robust, and therefore recommend the use of OLS estimation for time-series cross-section data.

15 Contemporaneous political variables may capture two consequences. The first is that parents may foresee the changes in the political environment in the near future and react by varying family size as the POFED model shows. This is the ex ante effect. The second possibility is that political institutions are semi-permanent and path- dependent ~Cukierman, Edwards, and Tabellini, 1992!. Therefore, the contemporaneous effect may ref lect a continuation of past effects of political institutions on fertility decisions. We are grateful to a referee for this insight.

Table 1. Politics, Economics, and Births: A Regression Analysis, 1960–90

Lag 5 0 Lag 5 1 Lag 5 2 Lag 5 3 Lag 5 4 Lag 5 5

Constant 6.446*** 6.435*** 6.420*** 6.406*** 6.384*** 6.366*** ~0.056! ~0.057! ~0.058! ~0.058! ~0.057! ~0.059!

Political stability 20.219*** 20.224*** 20.276*** 20.294*** 20.330*** 20.341*** ~0.093! ~0.094! ~0.094! ~0.094! ~0.094! ~0.095!

Political capacity 20.028* 20.030* 20.035** 20.036** 20.040*** 20.048*** ~0.016! ~0.016! ~0.016! ~0.017! ~0.017! ~0.018!

Real GDP per capita 20.280*** 20.279*** 20.278*** 20.277*** 20.276*** 20.276*** ~0.010! ~0.010! ~0.010! ~0.010! ~0.010! ~0.010!

Literacy rate 20.221*** 20.223*** 20.223*** 20.223*** 20.222*** 20.222*** ~0.014! ~0.014! ~0.014! ~0.013! ~0.013! ~0.013!

OR 2 0.774 0.771 0.770 0.768 0.766 0.765 s 0.223 0.225 0.225 0.226 0.227 0.228

Numbers in parentheses are standard errors. ***significant at the 1% error level, one-tail; ** significant at the 2.5% error level, one-tail; *significant at the 5% error level, one-tail

680 Politics of Fertility and Economic Development

the formal derivations are part of a general equilibrium model!, we use lagged values of the variables identified in the model as instruments. We also vary the lags to demonstrate the robustness of the instrumental variables estimation. The joint simultaneity of variables in general equilibrium models always results in

Table 2. Politics, Economics, and Births: A Regression Analysis, 1960–90

Lag 5 0 Lag 5 1 Lag 5 2 Lag 5 3 Lag 5 4 Lag 5 5

Constant 5.544*** 5.531*** 5.520*** 5.509*** 5.494*** 5.480*** ~0.058! ~0.059! ~0.059! ~0.059! ~0.059! ~0.060!

Political stability 20.104# 20.103# 20.137* 20.144* 20.160** 20.163** ~0.075! ~0.076! ~0.076! ~0.076! ~0.076! ~0.077!

Political capacity 20.040*** 20.042*** 20.044*** 20.045*** 20.047*** 20.052*** ~0.132! ~0.013! ~0.013! ~0.013! ~0.014! ~0.015!

Real GDP per capita 20.157*** 20.157*** 20.158*** 20.159*** 20.160*** 0.160*** ~0.009! ~0.009! ~0.009! ~0.009! ~0.009! ~0.009!

Literacy rate 20.210*** 20.208*** 20.204*** 20.201*** 20.198*** 20.195*** ~0.011! ~0.011! ~0.011! ~0.011! ~0.011! ~0.010!

OECD 20.431*** 20.433*** 20.432*** 20.432*** 20.431*** 20.430*** ~0.017! ~0.018! ~0.017! ~0.018! ~0.018! ~0.017!

OECD 3 Stability 23.277*** 22.928*** 22.645*** 20.214*** 20.171*** 21.502*** ~0.801! ~0.797! ~0.797! ~0.766! ~0.768! ~0.766!

OR 2 0.853 0.882 0.850 0.850 0.849 0.848 s 0.180 0.181 0.181 0.182 0.183 0.183

Numbers in parentheses are standard errors. ***significant at the 1% error level, one-tail; **significant at the 2.5% error level, one-tail; *significant at the 5% error level, one-tail; #significant at the 10% error level, one-tail

Table 3. Births and Poverty Traps, 1960–90

Lag 5 1 Lag 5 2 Lag 5 3

Constant 6.418*** 6.396*** 6.374*** ~0.053! ~0.054! ~0.055!

RPE 3 Low 0.108** 0.078 0.044 ~0.046! ~0.093! ~0.046!

RPE 3 Middle 20.065*** 20.064*** 20.059*** ~0.018! ~0.019! ~0.019!

RPE 3 High 20.064** 20.073* 20.080* ~0.031! ~0.031! ~0.032!

Political stability 20.244*** 20.274*** 20.308*** ~0.093! ~0.093! ~0.094!

Real GDP per capita 20.273*** 20.270*** 20.268*** ~0.009! ~0.009! ~0.009!

Literacy rate 20.231*** 20.233*** 20.235*** ~0.013! ~0.013! ~0.013!

OR 2 0.770 0.767 0.765 s 0.181 0.181 0.182

Numbers in parentheses are standard errors. ***significant at the 1% error level, one-tail; **significant at the 2.5% error level, one-tail

Yi Feng, Jacek Kugler, and Paul J. Zak 681

estimation difficulties, but using lagged variables as instruments is the standard way to address this problem while continuing to estimate the derived equilibrium relationships. This method works in dynamic models because of the time-series persistence of such variables as income and political capacity.

Across all specifications in Table 1, every estimated coefficient has the correct sign and is statistically significant at conventional levels. The baseline estimation shows that the most powerful effect on birth rates is from income, consistent with the modernization literature. Indeed, a 1% increase in income results in an annual reduction of birth rates of 0.28%. Thus, consistent with our model, increases in income are strongly associated with declines in births. The second strongest effect on birth rates comes from political stability. If political stability rises by 1%, the annual birth rate falls by 0.22% and its value increases to 0.34% when the independent variables have five lags. Similarly, a 1% increase in the literacy rate reduces births by 0.22%. Lastly, political capacity has a significant but smaller effect on birth rates, with a 1% increase in capacity reducing birth rates by 0.03%. The most important of these results is that political variables— political stability and government capacity—have statistically significant effects on birth rates, even when income and literacy are controlled for at all lags, indicating robust support for the POFED model. Nevertheless, the effects of politics on births are diluted in the entire sample since the theory demonstrates that capacity and stability are most important for middle-income countries, which leads to a test to be conducted later in this section. Below we examine the effects of politics on births in more detail.

The lag structure, designed to test the dynamic effects of the POFED model, reveals several interesting phenomena. Political stability has an enduring nega- tive effect on birth rates. Across all lags political stability is statistically significant, and highly so. Moreover, the parameter estimates increase as the lags rise, show- ing that political stability has long-term effects. This suggests that when the political system is unstable, it tends to remain unstable, exhibiting strong path dependence ~Cukierman, Edwards, and Tabellini, 1992; Feng, 1997, 2000; Ben- son and Kugler, 1998!.16 Compared to political stability, the statistical signifi- cance of political capacity varies over time, with the lagged political capacity producing a more pronounced effect on births. Its contemporaneous effect and the effects of lower order lags are relatively weak.

Controlling for OECD countries provides additional insights about the differ- ences between less developed and developed nations.17 In most OECD countries, political systems are mature and an individual’s retirement is funded through a combination of savings, pensions, and social security schemes. To test our model in this context, we create an intercept dummy variable OECD that takes the value of one for OECD countries and zero otherwise. We also create a slope dummy variable OECD3STABILITY, which is OECD times Political Stability. This variable is used to determine whether the effect of political stability on birth rates differs between developed and developing societies. As shown in Table 2, the statistical evidence supports the POFED model. First, the intercept dummy variable is negative and statistically significant. Developed nations have lower baseline birth rates. The interaction variable is also negative and significant. Being an OECD country is negatively related to the birth rate, even after income levels are controlled. This indicates the possibility that the pension, social security, and welfare systems characteristic of OECD countries may have some inhibiting effects

16 We have also experimented with political violence variables such as riots, assassinations, general strikes, and antigovernment demonstrations. Using the base model, we find the following parameter estimates and their standard errors. RIOTS, 0.001 ~0.0008!; ASSASSINATIONS, 0.002 ~0.001!; GENER AL STRIKES, 20.001 ~0.00097!; and DEMONSTR ATIONS, 0.002 ~0.008!. The results on other policy variables remain robust.

17 We thank a referee for suggesting this analysis.

682 Politics of Fertility and Economic Development

on births. Alternatively, the OECD variable may represent low levels of political instability. The supporting evidence of this inference is that when OECD and the interaction term of stability and OECD are included, the explanatory power of STABILITY decreases, suggesting that the OECD may be associated with the lack of major political instability that would otherwise be ref lected in birth rates. Observe that GDP, literacy, and political capacity continue to carry the correct signs and remain statistically significant for lags zero through five, as do OECD and OECD3STABILITY, demonstrating the robustness of the POFED model.

The POFED model demonstrates that the largest impact of political capacity on births occurs for middle-income countries. To test this implication of the model, we use World Bank values to classify each country as having low, middle, or high income.18 Next, we create three binary dummy variables for the three levels of income: LOW, MIDDLE, and HIGH. Finally, we reestimate the param- eters with three new variables in the equation, LOW 3RPE, MIDDLE 3RPE, and HIGH3RPE. These results are contained in Table 3 for one, two, and three lags of the independent variables.

Table 3 shows that RPE has a positive but generally statistically insignificant effect on births in low-income countries, while it exhibits a negative and signif- icant effect on births in middle-income and high-income countries. These results reveal some nuances in the theoretical model. It should be noted that the level of significance of the interaction term for the middle-income group is the stron- gest of all groups while that for the low-income group is the weakest, judging by t-statistics. Particularly, the lack of statistical significance at the higher orders of lags for the low-income group implies that the pattern of the effects of govern- ment capacity on births is not strong for these countries. The positive estimated coefficient is nevertheless consistent with the derivation of political capacity. As section A1.3 shows, poor countries have relatively low political capacity. The empirics suggest that low-income, high political capacity countries are over- taxing citizens to fund government programs ~i.e., x . x*!, so that an increase in political capacity has a net-of-tax negative impact on individuals’ incomes, causing births to rise.19 In contrast, governments in middle- and high-income countries stimulate their economies by raising incomes and human capital with- out over-taxing, resulting in a demographic transition to low birth rates. These results strongly support one of the novel predictions of the POFED model—that political capacity is necessary to reduce fertility and stimulate economic devel- opment in middle- and high-income countries.

Our findings on the effects of political institutions on fertility change provide important insights into the mechanism of economic growth and development. So far, though, we have only examined the direct effect of politics on births, while the theory shows a more complicated interaction. Specifically, section A1.3 establishes that political capacity is increasing in GDP per capita and in human capital. Thus, the POFED model shows that an increase in income or human capital reduces births both via a direct effect and indirectly by raising political capacity.

Consider an increase in literacy. Referring to the parameter estimates in Table 1 with one lag in the independent variables and using a value of 103 for the share of output paid to capital ~a! common in developed nations, a 1% increase in literacy results in a direct reduction in births of 0.22% and an increase in

18 We use World Bank criteria which classifies countries with 1997 per capita income below $785 as low income, sets a range of $785–$9655 for middle income countries, and calls countries with incomes of $9655 and above high income.

19 A negative correlation between RPE and income is evidence that countries over-tax their populations. This value, for the poor country subsample using logged data, is 20.06 which causes the LOW3RPE variable to have a positive effect on births.

Yi Feng, Jacek Kugler, and Paul J. Zak 683

political capacity of 2% which causes births to fall by 0.06%.20 The sum of the direct and indirect effects of a 1% increase in literacy is an annual decline in births of 0.28%. As the POFED model shows, this decrease compounds across generations leading to a significant reduction in fertility over time. By a similar calculation, a 1% increase in income has a direct effect of a 0.28% reduction in birth rates and also raises political capacity by 1%. The sum of the direct and indirect effects of increasing income by 1% is therefore an annual reduction in birth rates of 0.31%. If both income and education increase by 1%, the direct and indirect effects lead to a reduction in birth rates of over one-half percent per year ~0.59%!.

Conclusions

The POFED model links political structures to long-run economic performance, showing how both politics and economics affect fertility choices and therefore human capital accumulation over generations. A unique feature of the model is that we derive growth trajectories as a function of the underlying politics. In doing this, we demonstrate why nations fall into poverty traps, and why they develop into advanced economies. The formal structure shows that a demo- graphic transition prompted by political factors provides a means to escape poverty.

The POFED model also reveals that the period of rapid transitional growth is fragile. Politically weak governments engender relatively high rates of population growth, which reduces productivity and incomes, and diverts resources from production to child rearing. Such increases in births reduce the human capital of future generations, imperiling prospects for economic growth and causing political capacity to decline. In this situation, we show that a vicious circle of economic contraction and political impotence emerges. Conversely, developing nations, particularly those in the middle-income group, that are able to improve their political structures can jump-start their economies by inducing a demo- graphic transition, which lays the foundation for rapid transitional growth.

The barrier to development in the least developed societies is politics. As it is difficult to build political structures when participation and resources are lim- ited, poor nations face the daunting challenge to improve their political capacity and escape from the poverty trap at the same time. External forces such as international aid may help lead societies out of a poverty trap, but the funda- mental resources required to erase poverty primarily reside within the nation, particularly with the government leadership, in whose hands rest the means to elevate the society from poverty.

Appendix 1. POFED Model Formal Theory

This appendix presents and solves the individual’s utility maximization problem, the firm’s profit maximization problem, and the government’s optimal policy problem which are the components of the POFED model. Thereafter, a political- economic equilibrium is defined. POFED is a dynamical general equilibrium model, which means that given a utility function, a specified production process, and initial conditions, all aspects of the model evolve endogenously and are jointly dependent on each other. For example, we show that political capacity affects production decisions by firms, the utility maximization problem for indi- viduals, and thus the process of economic development. The model itself has an

20 Using the derivation of political capacity ~A1.13!, and the production function ~A1.6!, one can show that ln~ xt ! 5 ln@~1 2 s! ~1 2 a!

2 # 1 ln~Yt !, so that d ln~ xt !0d ln~Yt ! 5 1. Similarly, d ln~ xt !0d ln~ht ! 5 ~1 2 a!0a 5 2 when a 5 103.

684 Politics of Fertility and Economic Development

overlapping generations structure, as discussed in the text, in which individuals live for three periods: childhood, young adulthood, and old age.21 We append to this basic structure production decisions by firms, fertility choices by individuals, the transmission of human capital from parents to children, and a government that has multiple policy objectives. Table A1 defines all variables used in the model.

A1.1. Individual Decisions

Consider a country in which, at each point in time, a large number of children, young adults, and older adults are alive. Each generation has a different level of human capital, h , while individuals within a generation are identical. The econ- omy begins at time t 5 0 and continues indefinitely. The economy has a single good that can be used for consumption or investment in physical capital, K . Physical and human capital accumulate or decumulate over time based on the optimizing behavior by individuals and the government.

Agents maximize lifetime utility during the two periods of adulthood, subject to a budget constraint in each period. Since a child’s consumption is funded by his or her parent, no utility accrues during childhood. The budget constraint for a young adult, equation ~A1.2! below, equates consumption c1 to after-tax labor income wh ~1 2 t! @wage, w , times human capital, h , less taxes, t [ ~0,1!# , after paying eb for children’s consumption @b children ~births! which each cost e to

21 The overlapping generations model was developed by Samuelson ~1958! and Diamond ~1965!. The best reference to the overlapping generations model is Azariadis, 1993.

Table A1. Definitions of Variables in the POFED Model

Definition Range

Parameters a Capital’s share of output ~0,1! b Patience parameter in utility ~0,1! g Preference for children .0 d Depreciation rate of physical capital @0,1! s Transfer of tax revenue to political cronies @0,1! v The rate of transfer of human capital from parents to children .0 u Dilution effect on human capital transmission from multiple children .0 D Proportional constant in the cost of child-raising ~0,10wt ht ~1 2 tt !!

Variables c0 Consumption of young adults .0 c1 Consumption of old adults .0 b Births in a family ≥1 w Economy-wide average wage .0 h Human capital .0 a Assets saved for old age ≥0 t Tax rate on labor income @0,1# e Per child cost of children, e 5 D ~wh !2 .0 Y Aggregate output .0 K Physical capital stock .0 S Political instability @0,1# x Political capacity .0 N Number of young adults ~labor force! .0 r Cost of loans to firms .0 R Yield on savings: 1 1 r 2 d .0 p Police expenditures .0

Yi Feng, Jacek Kugler, and Paul J. Zak 685

raise# , and save a for old age. The budget constraint for an old agent, equation ~A1.3!, shows that consumption c2 is funded by the principal and interest on savings from young adulthood R a , where R is one plus the net interest rate. All agents have identical logarithmic and temporally separable utility functions.

Combining the elements above, the expected lifetime utility maximization problem for an individual born at time t 2 1 is

Maxc1, t ; c 2, t 11; bt E $~1 2 b! ln~c1, t ! 1 b ln~c2, t 11 ! 1 g ln~bt !% ~A1.1!

s.t.

c1, t 5 wt ht ~1 2 tt ! 2 et bt 2 a t 11 ~A1.2!

c2, t 11 5 R t 11 a t 11 ~A1.3!

where b [ ~0,1! denotes the preference for consuming when middle-aged versus old-aged, g . 0 is the preference for children, and E is the expectations operator.22 Agents maximize expected utility because, due to political effects described below, income and the return on savings are stochastic. Further, because there is a large number of atomistic agents in the model, the actions of a single individ- ual have no effect on aggregates. As a result, individuals take political instability, S , and political capacity, x, as given in solving their utility maximization problem ~A1.1!–~A1.3!.

We choose a parameterization of the cost of children, e , that ref lects the income foregone for the time spent with children, which is the primary cost of child-rearing ~Birdsall, 1988!. To wit, let the cost of children be quadratic in net labor income, et 5 D ~wt ht ~1 2 tt !!

2, with the constant 0 , D , 10wt ht ~1 2 tt !. In addition, we impose the condition that the minimum number of children in a family is one so that in the limit population is constant at its replacement rate. Note that with logarithmic preferences, individual optima from problem ~A1.1!– ~A1.3! are unique and strictly positive. As a result, the Largrange multiplier method is not necessary, and the optimization problem is solved by substituting out middle-age and old-age consumption using each period’s budget constraint and maximizing over a t11 and bt .

Solving for the optima from the utility maximization problem ~A1.1!–~A1.3! under these conditions, the solutions for the number of children bt

*, and sav- ings, a t11

*, are

bt * 5 Max HE g~1 1 g!Dwt ht ~1 2 tt ! , 1J ~A1.4!

and

a 1, t 11 * 5 E

bwt ht ~1 2 tt !

~1 1 g! . ~A1.5!

Equation ~A1.4! shows that an individual’s optimal number of children is positively related to the preference for children, g, and negatively related to net-of-tax labor income wh ~1 2 t!. Optimal old-age savings ~A1.5! is a constant proportion of net-of-tax labor income, with this proportion increasing as agents

22 Note that as is standard, we ignore the integer constraint on children. In addition, problem ~A1!–~A3! is written in per effective worker terms, i.e., economic variables are written relative to working agents’ human capital, h . For this reason, labor income is given by wh , that is, the wage times human capital, h , rather than simply w .

686 Politics of Fertility and Economic Development

become more patient ~ b increases!, and falling as the preference for children ~g! becomes stronger since children have a cost. Observe that government policy affects individual decisions. As the tax rate t increases in ~A1.4!, births increase as net-of-tax labor income, and therefore the opportunity cost of children, falls. Similarly, taxes reduce income and therefore savings in ~A1.5!.

A1.2 Politics and Production

As discussed in the text, political instability is a mapping St 5 S ~Yt21 , pt , et !: R 3 r

@0,1# which is the proportion of the physical capital stock that is destroyed during antigovernment uprisings, where Y is aggregate income, p is government fund- ing for the police which reduces the ability of demonstrators to destroy capital, and e;G is a random variable that denotes the level of discontent with the political milieu, where G is a CDF with finite mean and variance. Following the evidence in Feng, 1997, and Zak, 1997, we assume that S is decreasing in lagged output, Y, is decreasing police spending, p , and increasing in e.23 If instability is so large the entire physical capital stock is destroyed ~S 5 1!, then the govern- ment is considered overthrown. Note that S is not the number of demonstrations, but the impact of demonstrations on the economy.

Firms produce output, Y, with a modified Cobb-Douglas production function,

E $Yt % 5 E $~K t ~1 2 St !! a ~ xt Nt ht !

12a %, ~A1.6!

with the marginal productivity parameter a [ ~0,1!. Equation ~A1.6! shows that political instability reduces stock of productive capital and therefore the output that is produced. Net capital in production is K t ~1 2 St ! since proportion S of the capital stock is destroyed in antigovernment demonstrations. Note that the expected value operator appears because St has a random element to it at time t . The second political factor in the model is that political capacity, x, affects the productivity of private firms by implementing policies that enhance economic efficiency. Political capacity, x, enters in the production function ~A1.6! by raising the productivity of labor.

A representative firm chooses physical capital per effective worker, uK t [ K t ~1 2 St !0Lt , to maximize profits, where effective labor supply is Lt [ Nt ht xt , by solving

Max uKt , E $Yt % 2 rt uK t , ~A1.7!

where rt is the cost of financing capital investments which is taken as given by firms and markets are perfectly competitive. Solving ~A1.7! after substituting in the production function ~A1.6! produces the firm’s demand function for per effective worker physical capital, uK . Noting that the production function is homo- geneous of degree one, permits one to solve for the demand for effective labor L . Given these demand schedules and supply decisions made by consumers, the market clearing wage for labor, w , and return on savings, R 5 1 1 r 2 d, are found. The return to savings takes into account the rate of depreciation of

23 Gupta, Singh, and Sprague ~1993! show that the number of demonstrations and demonstrators follows an inverted U pattern relative to police coercion in a cross-country sample, and Francisco ~1996! finds some evidence of backlash against the police in German demonstration data. If such a backlash occurred in the economic impact of demonstrations, S , a rational policy-setter, who we model choosing optimal police expenditures, would never generate such an outcome. As a result, our assumption of a ~locally! monotone relation between S and police spending is justified. We thank a referee for pointing this out. See Zak and Feng, 1998, for a model in which the equilibrium dynamics include optimal government policy to maintain public order, and Feng and Zak, 1999, for a dynamic model of regime change.

Yi Feng, Jacek Kugler, and Paul J. Zak 687

physical capital in production, d [ @0,1# , with r 2 d the net interest rate. Using ~A1.6! and ~A1.7!, factor prices wt and R t11 are the marginal product of effective labor and the marginal product of capital plus one minus depreciation,

wt 5 ~1 2 a!E $~K t ~1 2 St !! a xt

12a ~Nt ht ! 2a %, ~A1.8!

R t 11 5 1 1 aE $~K t 11~1 2 St 11 !! a21 ~ xt 11 Nt 11 ht 11 !

12a % 2 d. ~A1.9!

Conditions ~A1.8! and ~A1.9! show that political instability, S , reduces both wages and the return to savings, while political capacity raises both w and R . Note that ~A1.6! and ~A1.8! indicate that the aggregate wages paid to labor are a fixed proportion of output, wt ht Nt 5 ~1 2 a!E $Yt %.

Substituting wage ~A1.8! into the first order condition for births ~A1.4! pro- duces the equilibrium births equation ~1! in the text. Similarly, putting ~A1.8! into the savings relation ~A1.5!, we obtain the equilibrium savings function ~2!.

A1.3. The Politics of Policy-Setting

Aggregating the taxes paid by working agents shows that the government receives tax revenue, twhN . Using the production function ~A6! and equilibrium wage ~A8!, tax revenue can be written as t ~1 2 a!Y, which shows it is a proportion of aggregate output. The reactive portion of government policy, police expendi- tures, is a fixed proportion s [ @0,1! of tax revenue. The remaining tax revenue is spent on proactive policies that enhance income growth. Tax revenues and expenditures balance at each time t , producing the government budget constraint24

~1 2 s!~1 2 a! tt Yt 5 xt , ~A1.10!

where Yt is given by ~A1.6!. Since output is increasing in physical capital, politicians choose the tax rate t

and spending x to maximize expected capital deepening,25

Maxt E K t 11 K t

. ~A1.11!

This maximization is subject to two constraints, the government budget balance relation ~A1.10! and the equilibrium law of motion for physical capital which is found by summing the savings of working agents, ~A1.5!, using the equilibrium wage ~A1.8!. The capital market clearing condition shows that aggregate savings at time t funds the capital stock at time t 11,

K t 11 5 E $ b ~1 2 a!~1 2 tt !~K t ~1 2 St !! a ~ xt Nt ht !

12a %. ~A1.12!

The solution to ~A1.11!, which holds in expected value, generates a spending plan for political capacity, Ext

*,

Ext * 5 @E ~1 2 s!~1 2 a!2K t

a ~1 2 St ! a ~Nt ht !

12a # 10a. ~A1.13!

After this plan and the tax rate are determined, the stochastic portion of political instability e is observed, establishing the potential impact of antigovernment

24 For simplicity, government borrowing is ignored. 25 A full discussion of this method of setting government policy can be found in Zak, 1997, and Ghate and Zak,

1999.

688 Politics of Fertility and Economic Development

demonstrations on the economy as well as tax revenue. Spending on the police, which is nonstochastic, is given implicitly by

pt * 5 s ~1 2 s!~12a!0a ~1 2 a!20aK t ~1 2 S ~Yt 21 , pt

*, et !!~Nt ht ! ~12a!0. ~A1.14!

Police expenditures—the highest priority spending—follow rule ~A1.14!, while spending on growth-enhancing policies is the residual of actual tax revenue after funding the police, which, on average, follows proactive spending rule ~A1.13!. Under a technical condition, ~A1.14! indicates that governments optimally increase spending on the police when political discontent e increases, and decrease police spending when the economy grows such that physical and0or human capital increase.26

Because maintenance of public order is the government’s top priority, the proactive spending plan ~A1.13! holds on average, but will not hold at every point in time. The actual amount of political capacity is xt

* 5 Ext * 1 Dt , where

the difference between actual and expected tax revenues is Dt 5 ~1 2 s! 10a

~1 2 a!20aK t ~Nt ht ! ~12a!0a @1 2 St 2 $E ~1 2 St !

a %10a # .27 Loosely speaking, the term D is positive or negative if the effect of political instability on the economy exceeds or falls below its expected value.28

Equation ~A1.13! is the maximal value of government programs that stimulate capital accumulation, which takes into account both the increase in output and the impact on the future level of the capital stock from the taxes that fund government programs. The derivation shows that political capacity is increasing in physical capital, K , and in aggregate human capital, Nh . Observe that when the production function parameter a is less than half, political capacity increases faster than linearity in aggregate human capital.29 Thus, the accumulation of human capital, typically measured by education or literacy, is an important contributor to a nation’s political strength. Relation ~A1.13! also reveals two effects through which political instability, S , affects political capacity, x*. First, political instability directly reduces a country’s productive capacity by destroying part of the physical capital stock. By this income effect, an increase in political instability reduces output and thus tax revenue by ~A1.10! which reduces the government’s ability to fund productivity-enhancing projects. Second, by the substitution effect, when political instability rises the government’s first priority is to raise expenditures on the police by ~A1.14!. For a given tax rate t, such preemptive reallocation decreases policy discretion of the government. Thus, unstable governments are less politically capable because of the combined impact of the income and substitution effects.

We now relate political capacity to demography. Using the condition for the desired number of children ~A1.4! ~assuming bt

* . 1! and equilibrium wages ~A1.8!, we can write political capacity as a function of births,

Ext * 5 @Eg0~11g! ~1 2 a!2D ~1 2 St !

aK t a ~Nt ht !

12abt * # ~10~12a!!. ~A1.15!

26 The technical condition is 2dS0dP . 10@s ~1 2 s!~12a!0a ~1 2 a!20aK t ~Nt ht ! ~12a!0a # which we assume holds

throughout the analysis. 27 Because the government budget constraint holds with certainty, the tax rate can be shown to be a constant,

t* 5 1 2 a. This follows from the timing of the game: choose the expected value for proactive policy Ex*, then observe the shock e as the period begins and allocate police spending p* to combat instability S , collect tax revenue, and then use the tax revenue less police spending for political capacity.

28 Due to the nonlinearity of the expression for D, the condition for D . 0 requires that 1 2 St . @E ~1 2 St !

a # 10a. Interestingly, this derivation shows that the variance of government spending is proportional to the variance in political violence.

29 The parameter a is the share of output paid to physical capital which is typically measured at one third; see Cooley, 1995: ch. 1.

Yi Feng, Jacek Kugler, and Paul J. Zak 689

Equation ~A1.15! shows that the optimal political capacity falls as births increase because, as the population grows, the demand for public goods supplied by the government rises. Meeting this demand stretches the resources of the govern- ment, reducing political capacity.

The derivation of political capacity via ~A1.11! shows that government policy is strategic. Because the derivation of optimal government policy uses the equilib- rium law of motion for physical capital that is based on optimal individual behavior, the resulting policy takes into account how individuals will react to changes in government actions t, p , and x. That is, the solution to ~A1.11! is a Nash equilibrium of the Stackelberg game played by the government and citi- zens, with the government being the first mover. Once the government has chosen the optimal tax rate, t*, police spending, p*, and political capacity, x*, individuals execute their optimal choices for births and savings as given by ~A1.4! and ~A1.5!.

A political-economic equilibrium for this model is a set of prices $wt , R t11 % for t 5 0, 1, 2, . . . , such that given these prices, the law of motion for human capital ~4!, a sequence of government policies $tt , xt , pt % for t 5 0, 1, 2, . . . , initial values of physical capital K 0 . 0, human capital h0 . 0, and population N0 . 0, consum- ers maximize utility by solving ~A1.1!–~A1.3!, firms maximize profits by solving ~A1.7!, the government sets policy to maximize capital deepening and security using ~A1.13! and ~A1.14!, and all markets clear. At each time t , an equilibrium exists and is unique because the objective functions of consumers and firms are strictly concave.

Appendix 2. Operationalization of Relative Political Extraction ~RPE!

The following concept advanced by Organski and Kugler ~1980:74! lays the foundation for the measurement of relative political capacity. As government operations depend upon resources extracted from the population, governments cannot survive—let alone govern—without such resources.

Taxes are exact indicators of governmental presence. Few operations of govern- ments depend so heavily on popular support—or on fear of punishment. Rev- enues affect so directly the lives of most individuals in society, and few are avoided so vigorously. Without some form of tax revenue, there is no national unity, and no control. Failure to impose and extract taxes is one of the essential indicators of governmental incapacity to obtain and maintain support.

Guided by this theoretical principle, Arbetman and Kugler ~1995! create a measure of relative political extraction ~RPE! obtained in three steps. First, an ordinary least squares regression is run on the following model:30

Tax

GDP 5 b0 1 b1~time ! 1 b2S Mining

GDP D 2 b3S Agriculture

GDP D 1 b4S Exports

GDP D 1 e.

In the second step, the predicted value for the tax ratio is obtained using the parameter estimates derived from the first step. In the third step, the following ratio is calculated:

Relative Political Extraction 5 Actual Government Revenue

Predicted Government Revenue .

30 The adjusted R2 for this regression is 0.3904.

690 Politics of Fertility and Economic Development

Third, if the above ratio is larger than one, then the government is defined as “strong” since it collects more taxes than otherwise predicted, based upon eco- nomic factors. Such a government is also regarded as politically capable and efficient. If the ratio is less than one, then the government fails to collect the taxes it is expected to obtain on economic grounds, and it is regarded as polit- ically incapable.

Appendix 3. Operationalization of Political Instability: Maximum Likelihood Estimates for the Probability

of Unconstitutional Government Change

Para. Est. Std. Err. Wald x2 P . x 2 Std. Est.

Intercept 23.398 1.072 10.051 0.002 Consumption 24.140 0.996 17.271 0.000 20.203 Inf lation 20.039 0.503 0.006 0.938 20.003 GDP per capita 20.001 0.000 10.676 0.001 20.569 Assassinations 0.025 0.073 0.115 0.734 0.013 General strikes 0.156 0.128 1.496 0.221 0.050 Guerrilla warfare 0.066 0.090 0.536 0.464 0.029 Crises 0.413 0.123 11.281 0.001 0.136 Purges 20.119 0.119 1.013 0.314 20.057 Riots 0.054 0.057 0.905 0.341 0.067 Revolutions 0.013 0.171 0.005 0.942 0.003 Demonstrations 20.153 0.097 2.504 0.114 20.182 Regime type 20.015 0.221 0.005 0.945 20.005 Constitutional change 20.020 0.255 0.006 0.937 20.003 Monarch 20.272 0.829 0.108 0.743 20.036 President 0.556 0.648 0.736 0.391 0.153 Premier 1.390 0.697 3.977 0.046 0.364 Military 0.423 0.680 0.387 0.534 0.066 Executive selection 0.057 0.161 0.125 0.724 0.024 Parl. responsibility 20.066 0.189 0.122 0.727 20.046 Cabinet size 20.019 0.016 1.390 0.238 20.077 Cabinet changes 0.332 0.147 5.067 0.024 0.108 Legis. effectiveness 20.757 0.212 12.709 0.000 20.454 Legis. selection 0.185 0.197 0.885 0.347 0.074 Asia 0.971 0.483 4.045 0.044 0.207 Africa 0.734 0.508 2.081 0.149 0.185 Latin America 1.509 0.514 8.633 0.003 0.327

Criteria for assessing model fit:

Criterion Intercept Only Intercept and Covariates (Chi-square for covariates) AIC 1183.198 1041.449 — SC 1189.447 1210.173 — 22 LOG L 1181.198 987.449 193.749 with 26 DF ~p 5 0.0001! Score — — 192.565 with 26 DF ~p 5 0.0001!

Association of predicted probabilities and observed responses:

Concordant 5 80.4% Somers’ D 5 0.619 Discordant 5 18.6% Gamma 5 0.625

Tied 5 1.0% Tau-a 5 0.043 ~505,119 pairs! c 5 0.809

References

Alesina, A., S. Ozler, N. Roubini, and P. Swagel ~1996! Political Instability and Economic Growth. Journal of Economic Growth 1:189 –212.

Alesina, A., N. Roubini, and G. D. Cohen ~1997! Political Cycles and the Macroeconomy. Cambridge: MIT Press.

Yi Feng, Jacek Kugler, and Paul J. Zak 691

Arbetman, M., and J. Kugler ~1995! “The Politics of Inf lation: An Empirical Assessment of the Emerging Market Economies.” In Establishing Monetary Stability in Emerging Market Economies, edited by T. D. Willett, R. C. K. Burdekin, R. J. Sweeney, and C. Whilborg. Boulder, CO: Westview Press.

Arbetman, M., and J. Kugler, eds. ~1997! Political Capacity and Economic Behavior. Boulder, CO: Westview Press.

Azariadis, C. ~1993! Intertemporal Macroeconomics. New York: Basil Blackwell. Azariadis, C. ~1997! The Economics of Poverty Traps, Part One: Complete Markets. Journal of

Economic Growth 1:449 –486. Banks, A. S. ~1996! Cross-National Time-Series Data. Binghamton, N Y: SUN Y Binghamton. Barro, R. J. ~1991! Economic Growth in a Cross-Section of Countries. Quarterly Journal of Economics

106:407–444. Barro, R. J. ~1997! Determinants of Economic Growth. Cambridge: MIT Press. Barro, R. J., and X. Sala-i-Martin ~1997! Technological Diffusion, Convergence and Growth. Jour-

nal of Economic Growth 2:1–26. Beck, N., and J. N. Katz ~1995! What to Do ~And Not to Do! with Time-Series and Cross-Section

Data. American Political Science Review 89:634 –647. Becker, G. S., and R. J. Barro ~1988! A Reformulation of the Economic Theory of Fertility. Quarterly

Journal of Economics 53:1–25. Becker, Gary S., K. M. Murphy, and R. Tamura ~1990! Human Capital, Fertility, and Economic

Growth. Journal of Political Economy 98:S12–S37. Becker, G. S., and N. Tomes ~1976! Child Endowments and the Quantity and Quality of Children.

Journal of Political Economy 84:S143–S162. Behrman, J., and P. Taubman ~1989! Is Schooling “Mostly in the Genes”? Nature-Nurture, Decom-

position Using Data on Relatives. Journal of Political Economy 97:1425–1446. Benson, M., and J. Kugler ~1998! Power Parity, Democracy, and the Severity of Internal Violence.

Journal of Conflict Resolution 42:196 –209. Birdsall, N. ~1988! “Economic Approaches to Population Growth.” In Handbook of Development

Economics, Vol. 1, edited by H. Chenery and T. N. Srivivasan. New York: Elsevier Science. Blinder, A., and A. Deaton ~1985! “The Time-Series Consumption Function Revisited.” In Brookings

Papers on Economic Activity 2, edited by W. C. Brainard and G. L. Perry. Washington, DC: Brookings Institution.

Bong, E. W., P. Wang, and C. K. Yip ~1996! A General Two-Sector Model of Endogenous Growth with Human Capital: Balanced Growth and Transitional Dynamics. Journal of Economic Theory 32:1667–1717.

Bongaarts, J., W. P. Mauldin, and J. F. Phillips ~1990! The Demographic Impact of Family Planning Programs. Studies in Family Planning 21:299 –310.

Burdett, K., and M. G. Coles ~1997! Marriage and Class. Quarterly Journal of Economics 112:141–168. Camp, S. L. ~1993! Population: The Critical Debate. Foreign Policy 90:126 –144. Cawley, J., K. Conneely, J. Heckman, and E. Vytlacil ~1996! Measuring the Effects of Cognitive

Ability. NBER Working Paper No. 5645. Ceci, S. J., and W. M. Williams ~1997! Schooling, Intelligence and Income. American Psychologist

52:1051–1058. Chen, B., and Y. Feng ~1996! Some Political Determinants of Economic Growth: Theory and Empir-

ical Implications. European Journal of Political Economy 12:609 –627. Coale, A. J. ~1975! “The Demographic Transition.” In The Population Debate: Dimensions and Perspec-

tives, Vol. 1. New York: The United Nations. Cooley, T. F., ed. ~1995! Frontiers of Business Cycle Research. Princeton, NJ: Princeton University Press. Cukierman, A., S. Edwards, and G. Tabellini ~1992! Seigniorage and Political Instability. American

Economic Review 82:537–555. Demeny, P. ~1989! World Population Growth and Prospects. Working Paper No. 4, The Research

Division of the Population Council. Diamond, P. A. ~1965! National Debt in a Neoclassical Growth Model. American Economic Review

55:1026 –1050. Downey, D. B. ~1995! When Bigger Is Not Better: Family Size, Parental Resources, and Children’s

Educational Performance. American Sociological Review 60:746 –761. Feng, Y. ~1997! Democracy, Political Stability and Economic Growth. British Journal of Political Science

27:391–418. Feng, Y. ~2000! Democracy, Governance and Economic Performance: Theory, Statistical Analysis and

Case Studies. Manuscript. Claremont Graduate University.

692 Politics of Fertility and Economic Development

Feng, Y., and P. J. Zak ~1999! The Determinants of Democratic Transitions. Journal of Conflict Resolu- tion 43:162–177.

Fiorina, M. P. ~1981! Retrospective Voting in American National Elections. New Haven, CT: Yale University Press.

Francisco, R. A. ~1996! Coercion and Protest: An Empirical Test in Two Democratic States. American Journal of Political Science 40:1179 –1204.

Freedman, L. P. ~1994! Family Planning as an Instrument of Empowerment. International Family Planning Perspectives 20:31–33.

Galor, O., and D. Tsiddon ~1997! The Distribution of Human Capital and Economic Growth. Journal of Economic Growth 2:93–124.

Galor, O., and J. Zeira ~1993! Income Distribution and Macroeconomics. Review of Economic Studies 60:35–52.

Ghate, C., and P. J. Zak ~1999! Cycles, Thresholds and Growth: The Politics of Fiscal Policy. Working Paper. Claremont Graduate University.

Gupta, D. K., H. Singh, and T. Sprague ~1993! Government Coercion of Dissidents: Deterrence or Provocation? Journal of Conflict Resolution 37:301–339.

Hanushek, E. A. ~1992! The Trade-off between Child Quantity and Quality. Journal of Political Econ- omy 100:84 –117.

Lewis-Beck, M. S. ~1990! Economics and Elections: The Major Western Democracies. Ann Arbor: University of Michigan Press.

Lucas, R. E., Jr. ~1988! On the Mechanics of Economic Development. Journal of Monetary Economics 22:1–42.

Magee, S., W. A. Brock, and L. Young ~1989! Black Hole Tariffs and Endogenous Policy Theory. Cam- bridge: Cambridge University Press.

McGuire, M. C., and M. Olson ~1996! The Economics of Autocracy and Majority Rule: The Invisible Hand and the Use of Force. Journal of Economic Literature 34:72–96.

Murray, C. ~1997! IQ and Economic Success. The Public Interest 128:21–35. Notestein, F. ~1945! “Population, The Long View.” In Food for the World, edited by T. W. Schultz.

Chicago: University of Chicago Press. Organski, A. F. K., and J. Kugler ~1980! The War Ledger. Ann Arbor: University of Michigan Press. Organski, A. F. K., J. Kugler, T. Johnson, and Y. Cohen ~1984! Birth, Death and Taxes: Political and

Economic Transition. Chicago: University of Chicago Press. Pritchett, L. ~1997! Divergence, Big Time. Journal of Economic Perspectives 11:3–17. Quah, D. T. ~1997! Empirics for Growth and Distribution: Stratification, Polarization and Conver-

gence Clubs. Journal of Economic Growth 2:27–60. Razin, A., and C. W. Yuen ~1993! Convergence in Growth Rates: A Quantitative Assessment of the

Role of Capital Mobility and International Taxation. NBER Working Paper No. 4336. Samuelson, P. ~1958! An Exact Consumption-Loan Model of Interest with or without the Social

Contrivance of Money. Journal of Political Economy 66:467–482. Sinding, S., J. Ross, and A. Rosenfeld ~1994! Seeking Common Ground: Unmet Need and the

Demographic Goals. International Family Planning Perspectives 20:23–28. Stokey, N. L. ~1996! Free Trade, Factor Returns, and Factor Accumulation. Journal of Economic Growth

1:421–447. Summers, R., and A. Heston ~1995! The Penn World Table (Mark 5.6). National Bureau of Economic

Research. Tamura, R. ~1996! From Decay to Growth: A Demographic Transition to Economic Growth. Journal

of Economic Dynamics and Control 20:1237–1261. Thompson, W. S. ~1929! Population. American Journal of Sociology 34:959 –975. Tufte, E. ~1978! Political Control of the Economy. Princeton, NJ: Princeton University Press. Venieris, Y., and D. Gupta ~1986! Income Distribution and Socio-Political Instability as Determinants

of Savings: A Cross-Sectional Model. Journal of Political Economy 94:873–883. Zak, P. J. ~1997! Institutions, Property Rights and Growth. Working Paper. Claremont Graduate

University. Zak, P. J. ~1999! Genetics, Family Structure and Economic Growth. Working Paper. Claremont Grad-

uate University. Zak, P. J., and Y. Feng ~1998! A Dynamic Theory of the Transition to Democracy. Working Paper.

Claremont Graduate University.

Yi Feng, Jacek Kugler, and Paul J. Zak 693