If political power and motives to invest in goverment bonds are positively correlated across voting groups, then a self-selection equilibrimum can arise where the goverment's promise to repay its debt is credible. The paper contains a model to illustrate the working of such an equilibrium. There are two periods. At the start of the first period, individuals arrive with different endowments of wealth and intelligence. Each individual divides his wealth between holding government bonds and acquiring an education. The proceeds of the bonds are used to construct public infrastructure capital. Each person's education generates private human capital in proportion to his intelligence. During the second period, an election is held, where two parties compete via promises over the rates at which they will tax labor income and redeem bonds. This is done taking the first-period investment decisions as given. Also in this period, individuals supply physical labor which, together with their previously created human capital and the public infrastructure capital supplied by the government, produces an output. This accrues to the individuals as their gross pretax income. Here a model of investment in public infrastructure is presented in which a welfare maximizing social planner would encounter a time consistency problem, thwarting his efforts at implementing a socially optimal level of public investment. However, the decision how faithfully the government should honor its obligations comes as the result of electoral competition between two political parties, a level of commitment is restored. Political parties are credibly committed to pursue voters, and under some circumstances this means redeeming bonds in order to curry favor with electorally pivotal ``swing voters''. The model indicates that this political motive is sufficient to support investment in the public infrastructure when the ``natural'' bondholders, those with more capital than they can profitably plow back into their own productive activities, are also electorally pivotal. The model presented here took on a special form, emphasizing the investment choice between human capital and government bonds and the political anchor holding the government to its promise to redeem bonds. The basic point that political competition need not undermine the credibility of governments' promises to repay their debts, and may under some circumstances actually increase their credibility, is much more general, as the examples in the paper indicate.