No society is devoid of institutions but many live with poor institutions. Institutions promote growth. This is a view now held firmly and widely. The task then is to ‘engineer’ growth-promoting institutions. Endogeneity characterises institutions; for example, groups enjoying political power influence economic institutions, but political power itself is a function of wealth. Given endogeneity, if the task is to design institutional reforms, the question then arises, as to what to reform first. We use the theories of institutional evolution put forth by Douglas North, Darron Acemoglu and Dani Rodrik and the historical experiences of different countries in the context of development (or non-development) of institutions, to determine the starting-point of institutional reforms, if the objective is to design institutional reforms. We argue that in Pakistan, neither large commercial interest nor fiscal constraints can force the de jure power to reform institutions. Typically, large commercial interests in Pakistan have thrived on favours from de jure power, and therefore do not have teeth. Given strategic interests of foreign powers, foreign aid will alleviate the fiscal constraint and the ruler-citizens bargain—though reforming institution in exchange for tax revenue will remain a dream. The country does not seem ready for a revolution either; the thought process that typically precedes revolutions seems to have barely begun. The alternative, that remains, then is the gradualist approach preferred by North, Acemoglu, and Rodrik. Institutional reforms in Pakistan should begin with reform of the educational system—the introduction of a common educational system for all and sundry up to a certain level. Two reasons make us chose the educational system as the candidate to start the process of institutional reform. First, a common educational system will produce a shared value system which, in turn, will reduce the heterogeneity in the society. Lesser heterogeneity in society will then facilitate an agreement over the minimal set of institutional reforms. Second, politicians being myopic, the de jure power is more likely to concede to the demand for reform of the educational system as compared to the demand to, say, put an end to rent-seeking. The former will affect the de jure power a generation hence, while the latter will affect them today.