![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
Abstract
Based on a unique data set of Indian legislators and their behaviour during Question Hour over a 30-year period (1980–2009), the paper establishes that there is a substantial gap in the volume of questions asked by legislators from national and subnational parties, even after accounting for party size and other covariates. Thus, despite increasing subnational party prominence in the electoral and executive arenas, national parties dominate activities to do with legislative oversight. The paper also explores mechanisms that may explain the difference in legislative activity between national and subnational party legislators.
Keywords:
Acknowledgements
The authors would like to thank Rohit Kumar at PRS Legislative Research for sharing data sources, Drs. Deshpande and K. S. James at the Institute for Social and Economic Change, Siddharth Swaminathan, and other colleagues at Azim Premji University for comments on earlier drafts of this paper. Also thanks go to Michael Clancy, Joseph Voelker, and the library staff at the University of Hartford for academic and institutional support.
Notes
1 We follow scholars who define subnational parties to include ethnic and territorial agenda parties (De Winter, Gómez-Reino, & Lynch, Citation2006; Müller-Rommel, Citation1998). Alternative characterisations can be found in Hepburn (Citation2009) and Jensen and Spoon (Citation2010). Another group – niche parties – pursues primarily non-economic interests, diverging from class politics and other traditional political fault lines (Wagner, Citation2011).
2 In several other institutions of the Lok Sabha, the rules are extremely biased against the opposition, forcing opposition legislators and parties to take to the streets and become deliberately disruptive (Wallack, Citation2008). However, we note that the literature that makes this claim, summarised by Wallack, is mostly anecdotal. To our knowledge, there has been no study of these issues. In general, there are very few data and little systematic empirical analysis to establish nuanced claims about the institutional performance of the Indian legislature.
3 We have a total list of 86 parties in the data set. The following paragraphs describe our classification of these parties into party types.
4 To be fair, the Left and Janata parties are also quite distinct from each other, as noted by the large literature on Indian parties (Chhibber & Petrocik, Citation1989; Fickett, Citation1993; Nikolenyi, Citation1998).
5 They could be politicians who could not get a nomination from any party and stand as ‘rebel’ candidates, or belong to a local elite class.
6 We conducted an inter-observer reliability test of our classification with the assistance of Dr Narayan Gatty. The key differences were among the three subtypes among subnational parties, a distinction that is unimportant in our analysis. We would like to thank Dr Gatty for his assistance.
7 It is given by the inverse of , where stateshares
is the fraction of the party's seats coming from state s. The Election Commission of India has a rule that categorises parties as ‘national’ or ‘state’, based on whether or not a party gains ‘recognition’ in at least four states. However, we feel that this categorisation does not adequately operationalise the concepts of national, mesonational and subnational parties. For instance, in the 2004 election, NCP (National Congress Party) was classified as national even though it had only nine MPs, all from the same state (Maharashtra). By contrast, RJD was classified as a state party even though it had more MPs (24), and had representation from two states. Or, in the 1999 election JD(S) (Janata Dal (Secular)) was classified as a national party even though it had one MP while TDP (Telugu Desam Party) with 29 MPs was classified as a state party. Because the Election Commission categorisation was found lacking, Ziegfeld (Citation2009) also uses a similar measure (based on vote share, not seat share), from which national and regional parties are classified based on a threshold.
8 The relatively large ENS for 1980 for the last party type is due to a short-lived splinter party (INC(U)) (Indian National Congress (URS)).
9 It was changed to 15 days in the 15th legislative term that began in 2009; our data are for the period prior to this term.
10 We assume that the balloting procedures for the selection of questions are random because there seems to be no apparent procedural bias to think otherwise. Further, even if there were bias, our analysis would still hold as long as the bias did not affect the national/subnational distinction. That said, however, further research is required to examine whether the Speaker or some other entity may bias admitting questions after vetting them according to admissibility procedures laid out in the House rules.
11 The ‘Zero Hour’, which begins right after the Question Hour in India, is perhaps a closer approximation to the Question Hour in other parliaments. The practice developed in India in the interval between 12 noon and 1 p.m. (that is, between the end of Question Hour and the beginning of the lunch hour). Legislators would informally request the Speaker to raise matters of immediate public controversy – comments or exposés made on the television or to newspapers, or other scandals – to force the government to respond formally or to seek a statement that lays out the official version of events. Initially, the Speaker would allow them because they were not listed in the legislative business of the day, the latter having been decided weeks in advance. This intervention soon became a popular practice, grabbing headlines, and was formalised into an ‘hour’ in 1962 (prior to the start of our data), although it was still based on spontaneous legislator participation. However, with increased participation, the Speaker's office soon had to deter legislators, in vain, by asking them to use other channels (special mention, calling attention notices) that could be used in normal business hours. In 1993, the Indian parliament formalised interpellation procedures for the Zero Hour. See http://rajyasabha.nic.in/rsnew/rsat_work/chapter-19.pdf. We chose to analyse the Question Hour rather than the Zero Hour for the following reasons. Both are autonomous of formal party strictures, but the Zero Hour is more spontaneous, making legislator behaviour in the latter driven more by media-based imperatives than by policy. The relatively more policy-focused behaviour in Question Hour is more useful in our longer term project of researching the governance impact of legislative participation. Further, even though Question Hour does not have the attention-grabbing drama of Zero Hour, it still retains considerable political salience in terms of policy feedback effects.
12 Question Hour time is sometimes cut short because taking quorum, affirming new members, obituary references and other announcements take precedence and are included within the schedule of the Question Hour. The Speaker can also cancel Question Hour because members who were listed for an oral answer were absent (Kaul, Shakdher, & Singh, Citation2009, p. 473).
13 The Lok Sabha is composed of 543 elected seats and two nominated seats. It must be dissolved every five years, although the president may also dissolve it earlier, typically when it fails to provide a stable federal government. On dissolution, a new set of members is elected in a general election based on universal adult franchise and the simple plurality rule in single-member districts. Occasionally by-elections are also held to fill individual seats that are rendered vacant owing to resignation, demise or dismissal of the member. In the 30-year period we consider (1980–2009), general elections took place in 1980, 1985, 1989, 1991, 1996, 1998, 1999 and 2004. We exclude the Lok Sabha term between 1989 and 1991 (two governments in the same term), because we only have data by Lok Sabha terms, not annually or by governments.
14 The secretariat publishes end-of-term reports detailing duration of sittings, enactments, budgetary legislation transacted, motions introduced and their status, discussions, and resolutions, referred to as ‘Statements'. Statement 28 of each report contains information on questions admitted from each MP during the term of that particular Lok Sabha. We used various issues of Statement 28 as the primary source to get information on the number of questions asked by each MP.
15 In our data set, on average a legislator asked approximately 10 times more unstarred questions than starred questions. Although we report results only for questions aggregated across these types, we also checked results separately for each type, and found that our conclusions regarding party effects hold for the two question types separately. Finally, note that there is also a third category besides starred and unstarred, namely ‘short notice’ questions. These are written submissions with shorter notice periods. In our data set, they represent less than 1 per cent of all questions asked.
16 As noted previously, not all questions submitted by legislators are scheduled to be asked, and a random ballot decides the latter. Further, it may be the case that not all scheduled questions are actually asked on the floor of the House, as Question Hour is occasionally curtailed or time runs out before the House can get to questions scheduled later in the hour. Data are not available to differentiate between questions that were admitted through the ballot and thus scheduled to be asked and those that were actually asked, because Statements 27 and 28 of the end-of-term Lok Sabha secretariat reports provide information only on questions admitted by type and there is no other official document summarising other information on the Question Hour. We are constrained for now to limit our analysis based on the volume of questions that were scheduled, and therefore put aside other related aspects such as motivations, outcomes (government response to particular questions), and question content.
17 While the quality of data provided by the Election Commission is generally very good, often the names of candidates were not standardised across elections. We thank Naveen Singhal for devising an algorithm to match names across elections in order to generate the variables for MPs' prior experience. The values generated by the algorithm were double-checked manually to eliminate false positives.
18 We constructed multiple categorical variables for education: no formal schooling, high school, bachelor's degree, higher degrees, and law degrees. Ministerial portfolios were coded by status.
19 See the articles in The Journal of Legislative Studies 2011 special issue devoted to parliamentary questions. NBR is a generalisation of the Poisson regression because it has the same mean structure but an extra parameter to model the over-dispersion. After model estimation, we conducted likelihood ratio tests for α = 0 to check whether NBR is more appropriate than the Poisson model. In all cases, we rejected the hypothesis that α = 0.
20 Here, we report estimation results for a pooled model. However, recognising that observations of multi-term legislators may be correlated, we also conducted a robustness check using the (unbalanced) panel structure of the data set. Specifically, we estimated the population-averaged GEE (generalised estimating equations) model on NBR specifying an error structure that is robust to correlations of observations of the same legislator over time. Our main results are robust to this alternative specification.
21 We also estimated models differentiating governing party legislators into those from single-party and coalition governments, and also adding a separate variable for ‘outside support’ (Arora, Citation2004). Our results showed that although coalition legislators asked more questions than single-party government legislators, the difference was not statistically significant, echoing a finding of Russo and Wiberg (Citation2010); and although opposition legislators asked more questions than ‘outside support’ legislators, the difference was not statistically significant. Hence, we use only a single dichotomous variable to represent government/opposition status.
22 It is interesting to compare the results in columns 1 and 2. The results for Left and Janata parties do not change much between the two columns because these parties were in opposition for most of the period under consideration. However, this is not the case with the BJP, and even less the case with Congress; the results for these two parties do vary substantially between columns 1 and 2.
23 We are, of course, happy to provide these results on request. For reasons of space, we also do not present estimation results for the year and state dummies. The year dummies show that there is a small decline in the number of questions asked by legislators. The state dummies show that legislators from some states (for instance, the north-eastern states) ask fewer questions.
24 Further, the more prominent such national executive experience – deputy minister versus minister of state versus cabinet minister – the fewer the questions asked in subsequent Lok Sabhas.
25 To explore this further, we adopted a non-parametric formulation using separate size categories (party size < 1 per cent, 1–2 per cent, 2–5 per cent, 5–10 per cent, 10–25 per cent and > 25 per cent). Estimation results (not reported here) show that legislators from both small-sized and large-sized parties ask more questions compared with legislators from medium-sized parties.
26 Here, we are conceptually motivated by the literature on issue salience of parties (for instance, Budge & Farlie, Citation1983; Budge, Klingemann, Volkens, Bara, & Tanenbaum, Citation2001). This literature emerged as a counterpoint to the median-voter conceptualisation of Downs (Citation1957), according to which parties compete by taking positions in a common issue-space. By contrast, the issue salience literature argues that parties compete by emphasising different sets of issues (that is, those salient to each party) rather than taking positions in a common issue-space.
27 One possibility would be to code the content of legislative transactions and activities to align with a comparative context, as suggested by scholars involved in the comparative agendas project (www.comparativeagendas.org). However, at present this is a daunting task for India, for reasons of data availability and comprehensiveness.
28 It also turns out that the national/subnational distinction is somewhat correlated with party age, with the subnational parties typically being more recent. Could it be that the reason national parties ask more questions is that they have a longer history? Several points suggest that this may not be the case. First, the BJP was founded in 1980 and has contested Lok Sabha elections since 1985, but from the beginning its MPs asked as many or more questions than the Congress or Left parties. Second, the Janata Dal (JD), a key Janata party, was formed only in the late 1980s and yet its 59 MPs asked a relatively high number of questions in 1991. Similarly, the other Janata parties in our classification (JNP and JNP(S)) were formed in the 1970s but asked a large number of questions in the 1980s. By contrast, DMK (Dravida Munnetra Kazhagam) and ADMK (All India Anna Dravida Munnetra Kazhagam) were state parties formed earlier that continued to ask relatively few questions in later decades as well.
29 Note that interest bundle size is distinct from other party attributes such as party size or sociological/ideological characteristics of party supporters (Wagner, Citation2011). We have also explored the idea that apart from interest bundle size, the coherence of interests articulated by a party – for instance, catch-all (Kirchheimer, Citation1966) as opposed to programmatic – could also matter, somewhat similar to the notion of congruence (Dandoy, Citation2011). However, we found that interest bundle coherence does not seem to be important for India, and that interest bundle size is the salient dimension.
Additional information
Srikrishna Ayyangar* is Assistant Professor of Comparative Politics at the School of Policy and Governance, Azim Premji University, Bangalore, India
Suraj Jacob is an Associate Professor at the School of Policy and Governance at Azim Premji University, Bangalore, India, email: [email protected]