This dataset produces annual codes through expert analysis levels of de/centralization in Pakistan's federation between 1956 and 2020 (also covering East Pakistan until 1971). It includes three politico-institutional measures -'constitutional autonomy’, the ‘provinces’ institutional autonomy’ and the ‘central manipulation of provincial elections’. It also separately codes the legislative and administrative autonomy of 22 policy areas - Agriculture, Citizenship and immigration, Culture, Currency and money supply, Defence, Economic activity, Education - pre-tertiary, Education - tertiary, Elections and voting, Employment relations, Environmental protection, external affairs, Finance and securities, Health care, Language, Law - civil, Law - criminal, Law enforcement, Media, Natural resources, Social welfare and Transport. Finally, it codes five fiscal areas ‘proportion of own-source revenues out of total provincial revenues’, ‘restrictions on own-source revenues’, ‘proportion of federal conditional transfers out of total provincial revenues’, ‘degree of conditionality (for conditional grants only)’ and ‘provincial public sector borrowing autonomy’.

Expert coding. The main sources for the coding of the policy areas were: 1) the texts of Pakistan’s 1956, 1962, 1972 and 1973 constitutions; 2) the texts of the various constitutional amendments; 3) acts passed by the federal and provincial legislatures and ordinances passed by the president or provincial governors 4) judicial rulings 5) the operation of inter-governmental institutions such as the Council of Common Interests (CCI) and the National Economic Council (NEC). In addition, we consulted contemporary academic literature concerning the operation of the political system, particularly under military regimes.The fiscal data were compiled using data from Pakistan’s annual Economic Surveys and Budgetary White Papers at the national and provincial levels, as well as the National Finance Commission of 2010. Codes were assigned by one of the two authors, and then checked by the other author to ensure inter-coder reliability. When either author was unsure, we checked specific scores and shared the code book with relevant experts.