This is a post by David Doyle and Robert Elgie
Since the beginning of the 1990s there has been ongoing debate about the relative effect of different regime types, specifically presidentialism, semi-presidentialism and parliamentarism. From the very beginning of this debate, there has been an acknowledgement of variation in presidential power not just across these three regime types, but also within each regime type. This within-type variation causes a problem for cross-national studies. If the within-type variation is large, then estimating the effect of regime types themselves may lead to spurious results. For this reason, Crisp, Desposato and Kantha (p. 447) have stated that there is a “need for more explicit theoretical depictions of the institutional variation in the class of regimes referred to as presidential as well as the need for a systematic empirical exploration of the impact of that diversity on regime performance” (1).
In this context, some observers have preferred to estimate the effect of presidential power on outcomes rather than the effect of regime type more broadly. For example, Hicken and Stoll (p. 1114) note that the power of the Colombian president has varied over time as a result of constitutional amendments, even though Colombia has maintained a presidential regime throughout. As a result, they prefer to measure variation in presidential power over time and estimate the effect of such variation. They note: “our overall index of presidential powers reveals variation within each type of regime that the simple trichotomy [of presidentialism, parliamentarism, and semi-presidentialism] obscures. It is this greater level of precision that leads us to prefer the index”.
However, while there is now a well-established literature demonstrating that variation in presidential power has consequences across a range of political and economic outcomes, there are many different measures of presidential power. In fact, we have identified 19 separate and original measures of presidential power, plus a further 16 studies that used one of these measures with different countries, time periods, and/or scores from the original study.
This range of measures raises a number of issues. Individual measures are sometimes poorly correlated with each other, meaning that findings are sensitive to the particular measure of presidential power that is used. There is also a considerable loss of information across the set of measures as a whole as countries are included in some measures but not others and then for only certain time periods. More generally, as Jessica Fortin has recently shown, there are no theoretical priors to tell us which indicators of presidential power we should choose or how the scores for the individual indicators should be aggregated (3).
We agree with Fortin’s analysis. However, we draw a different conclusion from her. She concludes very skeptically, effectively questioning whether any measure of presidential power is likely to be valid. By contrast, we assume that most social science concepts, such as voter turnout, social equality, corruption, and so on, suffer from equivalent problems of construct validity. Therefore, we should not give up on the effort to generate a dataset of presidential power scores. Instead, we should focus on the reliability of the data that underpins the concept we are trying to capture.
We wish to generate a time-series cross-sectional dataset of presidential power scores with country years as the units of observation. To do so, we choose not to construct a new measure of presidential power from scratch. Instead, we draw upon the comparative and local knowledge already embedded in the existing measures of presidential power that we identified. To maximize the reliability of the scores we derive them solely from measures that are based on institutional indicators of presidential power and on the basis of a method that accounts for potential idiosyncrasies of country scores in the existing measures. In addition, we report the standard errors and the confidence intervals for all the country years in our measures, providing information with which scholars can make an informed choice about whether or not a particular country should be included in an estimation and which of our measures might best be used in comparative studies.
The paper outlining our full methodology will soon appear in British Journal of Political Science and is available in advance here. In the meantime, we are making available the full set of presidential power scores, including standard errors and confidence intervals for each country time period, in a separate page at the header of this blog. We also provide more detail about the scores.
Overall, we encourage people to keep developing new measures of presidential power and to update existing measures for as many countries and as long a time period as possible. One of the advantages of our approach is that new country scores can be easily incorporated into the method we have used, creating the potential for country coverage to be further extended, for existing country scores to be updated, and for cross-national measures to become even more reliable.
(1) Crisp, Brian F., Scott W. Desposato, and Kristin Kanthak. 2011. Legislative Pivots, Presidential Powers, and Policy Stability.” Journal of Law, Economics and Organization 27 (2): 426-452.
(2) Hicken, Allen, and Heather Stoll. 2008. “Electoral Rules and the Size of the Prize: How Political Institutions Shape Presidential Party Systems.” Journal of Politics 70 (4): 1109-1127.
(3) Fortin, Jessica. 2013. “Measuring presidential powers: Some pitfalls of aggregate measurement.” International Political Science Review 34 (1): 91-112.