42
Rule by ordinances: Minority governments and decree power in India July 23, 2019 Abstract Unilateral decree powers are common to presidential systems; they are rarely found in parliamentary ones. We analyse decree powers in one such rare setting: India. The Indian Constitution authorises the executive to promulgate ordinances - legislate unilaterally - if some conditions are met. We show that minority governments in India have systematically used this power to legislate in the absence of sufficient numbers in the parliament, in a way that is inconsistent with the constitutional mandate. We find that minority governments promulgate more ordinances, enact fewer legislations, and often re-promulgate failed ordinances. The difference in patterns of ordinances and parliamentary legislations between majority and minority governments suggests that lacking numbers, minority governments use ordinances to substitute out parliament’s law making role. Such conduct suggests that the unilateral action theory, usually summoned to illuminate conflicts between the executive and legislative branches in presidential systems, has explanatory purchase in parliamentary systems too. Keywords: decree power; minority governments; unilateral action theory; parliamentary systems 1

Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Rule by ordinances:

Minority governments and decree power in India

July 23, 2019

Abstract

Unilateral decree powers are common to presidential systems; they are rarely found in

parliamentary ones. We analyse decree powers in one such rare setting: India. The Indian

Constitution authorises the executive to promulgate ordinances - legislate unilaterally - if some

conditions are met. We show that minority governments in India have systematically used

this power to legislate in the absence of sufficient numbers in the parliament, in a way that is

inconsistent with the constitutional mandate. We find that minority governments promulgate

more ordinances, enact fewer legislations, and often re-promulgate failed ordinances. The

difference in patterns of ordinances and parliamentary legislations between majority and

minority governments suggests that lacking numbers, minority governments use ordinances

to substitute out parliament’s law making role. Such conduct suggests that the unilateral

action theory, usually summoned to illuminate conflicts between the executive and legislative

branches in presidential systems, has explanatory purchase in parliamentary systems too.

Keywords: decree power; minority governments; unilateral action theory; parliamentary

systems

1

Page 2: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

1 Introduction

Narendra Modi became India’s seventeenth prime minister on May 26, 2014. The nation

handed his party and its alliance partners a stunning majority. But the government’s leg-

islative agenda never quite took off. It had a majority only in the lower house of India’s

bicameral parliament; the opposition still controlled the upper house. Unable to shepherd

legislation through the chambers, the Modi government resorted to decree powers to legislate

– in particular, the power to promulgate ordinances.

On December 29, 2014, just days after the winter session of parliament concluded, the

Modi government promulgated the Right to Fair Compensation and Transparency in Land

Acquisition, Resettlement and Rehabilitation (Amendment) Ordinance 2014 (LARRO). The

amendment introduced contentious changes to land law in India. Key rights granted to land

owners by the original legislation (enacted in 2013) stood curtailed.1 The opposition decried

the move. When the houses reassembled in February 2015, the government did not present the

ordinance in the upper house. So, LARRO lapsed.2 But it was re-promulgated on April 03,

2015 when the chambers broke for recess, and still again on May 30, 2015. The opposition, in

the meanwhile, launched a strident campaign.3 It worked. Before long, public opinion turned

sharply against the ordinance, and the government gave in. LARRO lapsed for a third time

on August 29, 2015; it was not reissued.4 Although the Modi government’s rush to muscle

in the law failed, it had permanent consequences – land acquisitions under the revised rules

made while the ordinance was in effect became forever valid.

This duel for supremacy over law-making powers is grounded in a provision of the Indian

constitution: Article 123. This provision was incorporated to provide the executive with

latitude to get through moments of legislative urgency when parliament is not at hand to

enact legislation. The provision is not meant to function as a parallel mechanism that gov-

ernments may invoke when they fail to enact legislations through the usual parliamentary

route. However, we argue in this paper that it has become just that; minority governments,

1Anon, Cabinet approves ordinance on amendments to Land Acquisition Act, Economic Times, December 29,2014.

2An ordinance is meant to be temporary – it can only be promulgated when the parliament is not is session andit expires six weeks after the parliament reconvenes unless the parliament legislates it into a law.

3Anon, Rahul attacks Modi government on farmers’ issue, The Hindu, April 20, 2015.4Nistula Hebbar, Land ordinance allowed to lapse, The Hindu, August 29, 2015.

2

Page 3: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

in particular, routinely rely on ordinances to prosecute their legislative agenda in a way that

is unconstitutional. Contrary to the founding mandate, we show how Article 123, has become

a parallel legislative setup in India.

Although India has a parliamentary system, the constitution endows the executive with

original legislative powers.5 Ordinarily, decree powers - powers to legislate unilaterally - are

found in presidential systems, especially in Latin America.6 But even in systems that do

not directly confer presidents with law-making powers, they are known to usurp a legislative

role. The United States is the most obvious example. In taking care that “the Laws be

faithfully executed”, American presidents have introduced executive orders with far reaching

and permanent consequences.7

Why do presidents legislate? Do they do so in order to circumvent legislatures and write

their policy preferences into law? A large body of literature has coalesced around two com-

peting explanations: the “unilateral action” and “delegation” theories respectively.8 The

idea of conflict between presidents and legislatures is central to the unilateral action theory.

Presidents and legislatures may not share legislative or policy preferences. This is especially

true when divided governments are in action. The theory posits that a president shall turn

to decree powers to bypass an unfriendly legislature, and impose his or her preference into

law. This theory predicts that the share of decrees increases as support for a president’s poli-

cies ebbs in the legislature (Neto, Cox, and McCubbins 2003). Conversely, delegation theory

interprets the use of decree power as a power sharing mechanism. For example, presidential

decrees may help individual legislators avoid controversial issues. By allowing a decree to

occupy the field, legislators may avoid addressing a matter in the legislative chambers by

delegating the legislative role to the president (Carey and Shugart 1998). Hence delegation

theory predicts the opposite: decrees rise when a president cohabits with a friendly legislature.

The empirical record, however is mixed (Cheibub and Limongi 2010).

We locate our analysis of ordinances in India within this literature. But there are key

5India is not alone among parliamentary systems. Italy, too, grants the executive original legislative powers.6For an overview of decree powers and their functioning, see Cheibub and Limongi (2010), Reich (2002), and

Negretto (2004).7Howell (2003). See also Mayer (2001), Rudalevige (2005), Christenson and Kriner (2017), and Lowande (2018)

for additional accounts of executive orders in the US.8See for example, Pereira, Power, and Renno (2008), Pereira, Power, and Renno (2005), Power (1998), and

Negretto (2004).

3

Page 4: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

institutional differences. India, like we mentioned, has a parliamentary system of sorts: the

executive and legislative branches are fused. The executive comes from and is accountable to

the legislature (Cheibub 2007). Yet, bicameral parliamentary systems, too, may experience

divided governments. This happens when different parties or coalitions hold majorities in

the two houses. The Modi government is a neat example. For its entire term (May 26, 2014

to May 23, 2019), it had a majority only in the lower house (Lok Sabha); the opposition

controlled the upper house (Rajya Sabha). But there are other examples, too. Vishwanath

Pratap Singh, India’s eight prime minister (December 02, 1989 to November 9, 1990), for

example, had a minority in both houses of parliament. When faced with a lack of majority in

the legislature, the tendency to legislate through executive powers, we argue, appears to be

strong even in India’s parliamentary system. In other words, confronted with (numerically)

unfriendly legislatures, executives in parliamentary systems, too, reach for decree powers. As

such, we find empirical support for “unilateral action” over “delegation” regarding the use of

decree power in parliamentary systems.

Our paper also speaks to the literature on coalition governments in parliamentary sys-

tems. In particular, there has been an increasing focus on the interaction between coalition

governments and parliamentary activity in shaping legislative outcomes.9 Martin and Van-

berg (2005) argue that the relevance of the parliament to the legislative process increases

when coalition governments are in power. This is because parliament plays a central role in

allowing different parties in the coalition to come up with a compromise position (Martin

and Vanberg 2004). Moreover, parliamentary debates provide an opportunity for coalition

parties to communicate with their constituents on policy issues (Martin and Vanberg 2008).

In this paper we document the flip side of the effect of minority governments on parliamen-

tary activity. In particular, we show that minority governments are more likely to fail at

creating a coalition in support of their legislative agenda and may use ordinances to escape

parliamentary scrutiny. These governments therefore sideline parliament and weaken its role

in the law making process.

We present a typology of majority and minority governments in bicameral parliamentary

systems and model their behaviour to derive five hypotheses: One, minority governments

9Gallagher, Laver, and Mair (2005) note that about three-fourths of governments in Western Europe since the1950s have been coalitional.

4

Page 5: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

promulgate more ordinances than majority ones, whereas majority governments enact more

legislations than minority ones. Two, for majority governments, there is no correlation be-

tween bills passed in a session and ordinances promulgated in the subsequent break, whereas

this correlation is negative for minority governments. Three, minority governments are less

successful in converting ordinances into acts of parliament. Four, only minority governments

re-promulgate ordinances. And finally, minority governments promulgate ordinances earlier

in the break relative to majority governments. Read cumulatively, these findings compel

us to the conclusion that minority governments systematically abuse Article 123 to enact

their legislative agenda. This article of the constitution has blossomed into a separate, but

unaccountable, legislative process. In law-making matters, then, minority governments in

India mirror presidential executives: they are armed with distinct legislative powers and are

(effectively) unaccountable to the legislatures for their use.

Overall, the paper unfolds in three core sections. In section 2, we introduce in greater detail

the law and practice of ordinances, and how Supreme Court interpretations have reconfigured

Article 123. Later, we present a typology of majority and minority governments, outlining

how the reconfigured Article 123 incentivises different types of governments. In section 3, we

construct a model to predict differences in ordinance related patterns between majority and

minority governments, as we define them. Finally, in section 4, we take the predictions of the

model to the data.

2 Article 123: Constitutional interpretations

Article 123 of the Indian constitution authorises the president to promulgate ordinances

under certain conditions. (Article 213 confers a similar power on governors in the states.)

These ordinances are the equivalent of parliamentary, not delegated, legislations. In other

words, the president may legislate using ordinances laws that only parliament can otherwise

enact. However, consistent with India’s parliamentary system, the power to promulgate

ordinances, in practice, is exercised by the council of ministers. The latter decides all aspects

of ordinances: their necessity, timing, and content. Presidents command limited discretion

in these matters. Usually, they only sign ordinances into legal effect.10

10See On-line Appendix A for the full text of Article 123.

5

Page 6: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

India’s ordinance mechanism is an extraordinary one. It endows a parliamentary exec-

utive the power to legislate independently. To prevent it from transforming into a parallel

legislature, Article 123 enumerates certain limitations. A council of ministers must satisfy

a suite of procedural and substantive conditions before and after promulgating ordinances.

First, India has a bicameral system, and ordinances are impermissible if both houses of par-

liament are in session. They may be promulgated only if at least one house of parliament is

not in session. In other words, a government cannot resort to ordinances if the usual legisla-

tive route is available. Second, a council of ministers must be “satisfied that circumstances

exist which render it necessary for [it] to take immediate action”. This is a substantive re-

quirement. Ministers cannot invoke Article 123 merely because it is convenient to do so. Or,

merely because a government lacks the necessary majority to legislate through the chambers.

Once houses reconvene, ordinances must be placed before them. This is the third condition.

The requirement gives both houses an opportunity to scrutinise them. The parliamentary

Rules of Procedure also treat this requirement seriously. The Rules, for example, mandate

that a statement explaining the necessity for an ordinance must be laid before both houses

of parliament once they reconvene.11 Fourth, Article 123 grants the council of ministers six

weeks from the day both houses reconvene to convert ordinances into acts through the usual

parliamentary procedures. Otherwise, ordinances lapse; they “cease to operate”. Ordinances,

in other words, are temporary legislation. Once promulgated, they remain in effect for a lim-

ited duration. To become permanent, the executive must have them converted into acts of

parliament within a specified duration.

These conditions, taken together, suggests that the ordinance mechanism in the Indian

Constitution is a reasonable arrangement. The executive may make parliamentary laws in

moments of legislative urgency. But they are temporary. The laws become permanent only

if parliament properly enacts them in due course. And this need for parliamentary approval

means that the executive, ultimately, cannot defy the wishes of the two houses.

This, however, is not the case in practice. In a series of decisions, the Indian Supreme

Court has, effectively, rewritten Article 123. The conditions, both ex-ante and ex-post,

11See Dam (2014b) for more institutional details on the process of promulgating ordinances, and a discussion onthe constituent assembly debates. Also see Dam (2014a) and Dam (2016) for the deeper legal arguments againstviewing ordinances as a substitute for lawmaking in parliament.

6

Page 7: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

010

2030

40Nu

mbe

r of O

rdin

ance

s

1940 1960 1980 2000 2020year

Figure 1: Number of ordinances by year from May 1952 to May 2019

no longer matter; they do not constrain the executive in meaningful ways. Consider, for

example, the first condition: At least one house of parliament must not be in session. But

who decides if a house is, or should be, in session? The council of ministers. What if the

council of ministers prorogues or dissolves a house in session to get around the control? Is that

constitutional? Decisions regarding prorogations and dissolutions are not subject to judicial

review, the Supreme Court decided in 1969. The council of ministers, therefore, are at liberty

to summon or prorogue houses. They alone decide when houses may be in session. Getting

around the first control, therefore, is easy. Similarly, consider the second one. Governments

may promulgate ordinances if they are immediately necessary. Who decides if ordinances are

in fact necessary? The Supreme Court in R. C. Cooper v Union of India 1970 held that the

council of ministers are the sole judge of necessity; the court will not second guess ministerial

assessments.12 This has important implications for the ordinance mechanism. A government

unsure of support in parliament may resort to an ordinance and, thereby, circumvent the

legislative process. Similarly, a government wishing to avoid public debate on a bill at least

initially may promulgate an ordinance at a point when the chambers are not in session.

What of the ex-post conditions? Once the houses resume, governments must place the

ordinances before them. The houses may then convert them into acts of parliament. This

is the most desired outcome for the government. The houses, however, may reject the or-

12This view, which operates throughout our sample period, was recently reversed in Krishna Kumar v State ofBihar 2017.

7

Page 8: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

dinances. Or, aware of impending legislative defeats, a government may avoid trying to get

the ordinances converted into acts. In these latter instances, the ordinances shall lapse; they

shall “cease to operate”. But what happens to all actions taken when the ordinances were

in effect? Do those actions get wiped out as well? In other words, do failed ordinances lapse

retrospectively? Not according to the Supreme Court. The court has consistently held that

ordinances only lapse prospectively. In other words, all actions taken or initiated during the

time ordinances were in effect shall remain valid. This suggests that ordinances are no longer

temporary legislation. This is especially true of ordinances that seek to introduce one-off

changes in the law: establish a university, establish a national commission, demonetise cur-

rency etc. In such instances, a government’s failure to convert ordinances into acts carries no

constitutional significance. Because of the Supreme Court’s interpretation, all actions taken

while the ordinances were in effect automatically become permanent.

Finally, imagine that an ordinance lapses. But a government is keen to keep it in effect.

Can it re-promulgate the lapsed ordinance once the houses are no longer in session? In other

words, can a government prolong the term of a defeated or lapsed ordinance by resurrecting

it – by re-promulgating it? In 1986, the Supreme Court held that re-promulgation violates

separation of power in the Indian Constitution and, therefore, was unconstitutional. How-

ever, the court also carved out some exceptions. Under certain circumstances, governments,

the court held, may re-promulgate ordinances. Like before, this decision too weakens par-

liamentary control over ordinances. Governments may promulgate ordinances to circumvent

parliament, and endlessly re-promulgate them.

Taken together, these Supreme Court decisions have coloured an innocuous constitutional

provision with a troubling hue. The re-interpreted Article 123 is an invitation to indulge in

ordinances. It incentivises governments to circumvent parliamentary constraints and legislate

through decrees.

In theory, it is possible to study each ordinance and the circumstances under which it was

promulgated, to analyse whether conditions existed that required immediate action. This

would tell us whether the promulgation of the ordinance was constitutional. However, in

practice this is infeasible since there is no objective threshold in the constitution that can

be verified empirically by the judiciary. Notice, however, that not all governments are alike.

8

Page 9: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Governments in parliamentary systems, broadly speaking, are of two kinds: majority and

minority. Minority governments have an additional incentive, relative to majority govern-

ments, for misusing Article 123 as they lack sufficient numbers in parliament. We analyse

whether this difference in incentives leads to a difference in legislative behaviour. The dif-

ference in legislative behaviour, if it exists, suggests a particular misuse of Article 123 by

minority governments.

To identify this misuse, we take a statistical approach by showing a difference in the

pattern of ordinance promulgation between majority and minority governments. We argue

that these patterns suggest that minority governments are using ordinances to push legislative

agendas that do not enjoy parliamentary support. They are therefore effectively invoking

Article 123 to circumvent the two houses of parliament. However, it is worth emphasising

that our findings point to constitutional infractions at a statistical level. In other words, we

find that the misuse of Article 123 by minority governments is the most parsimonious and

compelling explanation that fits the data at an aggregate level. Given the statistical nature

of our study we make no claims about such misuse with respect to any particular ordinance.

We follow the “known cause” test proposed in Dam (2014b). Under this test, a necessary

condition for the constitutional validity of an ordinance is that the government promulgating

the ordinance should not know of the circumstances that necessitate it before the parliament

goes out of session. In other words, if the necessity arises when parliament is in session,

but the government waits and uses the end of a parliamentary session to promulgate an

ordinance, this necessary condition is violated. We do not claim to observe the arrival of

the circumstances that necessitated each ordinance. Instead, to operationalise this test, we

begin by assuming that the underlying distribution that generates the circumstances that

necessitate lawmaking is the same for minority and majority governments. However, in the

absence of numbers in parliament, we expect minority governments to systematically violate

this condition. In particular, we expect that when a necessity arises during a parliamentary

session, a minority government will wait for the parliamentary session to end and thereafter

promulgate an ordinance. This leads to differences in the observed patterns of lawmaking

through ordinances and acts in parliament between majority and majority governments. In

section 3, we propose a statistical model to derive these patterns and test them in section 4.

9

Page 10: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Session t

Xt

Break t

Yt

Session t + 1

Xt+1

Break t + 1

Yt+1

Figure 2: Legislative necessities in sessions and breaks

3 A statistical model: Five hypotheses

In this section we present a statistical model of legislative activity to derive hypotheses that

we take to the data in section 4. In particular we derive five hypotheses and discuss how

they support the claim that minority governments use ordinances to overcome their lack of

numbers in parliament rather than in a way that is consistent with the constitutional mandate

described in section 2.

Consider a setting with t ∈ {1, 2, . . . T} time periods. In practice T is 15 for each gov-

ernment that serves its full term since there are usually three sessions every year. Each time

period is further divided into a parliamentary session followed by a break. On each day in a

session or break there is a probability p with which a legislative necessity arises. A legislative

necessity is the convergence of some circumstances that make it attractive for the government

to pass a particular law. For example, these circumstances could be political in nature such

that some issue becomes salient and the political payoff from legislating on it increases.

Let Xt and Yt be the number of legislative necessities that arise in a session and the

subsequent break, respectively, in period t. Furthermore let the length of the session and the

break be St and It. As a result, the number of legislative necessities that arise in the session

and the break are random variables that are binomially distributed and we have

Xt v B(p, St) and Yt v B(p, It). (1)

The number of actual acts and ordinances13, denoted by X∗t and Y ∗t , will be legislated from

within the set of legislative necessities that have already arisen till that point. Figure 2

illustrates a 2 period slice of the timeline we have just described.

As discussed earlier, the constitution requires that for an ordinance to become a permanent

law, it must be legislated in parliament once a session begins. Failure to do so will lead to

13We use the term “acts” to mean laws that are passed through parliament and distinguish these from “ordinances”which are laws promulgated by the executive.

10

Page 11: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

the ordinance expiring six weeks after a session commences. To model the government’s

inclination to convert an ordinance into an act, we assume that at the end of the first time

period, with probability φ the government desires to convert an ordinance into an act in the

next period. φ = 1 implies that the government would like to convert all ordinances it passed

in a break, into acts in the following session. Conversely, with probability 1−φ, the need for

the ordinance ceases and the government makes no attempt to convert it into an act in the

subsequent session. Let Y ∗t be the number of ordinances from the break in period t for which

the necessity of an act in period t+ 1 continues to exist.

Note that this set up is agnostic about whether a legislative necessity is such that there

is need for “immediate action” as required by the constitution for the promulgation of an

ordinance. This is because whether such a threshold is met for a particular ordinance is

not something we can directly observe in our data. We instead focus on the differences

in the pattern of law making between majority and minority governments to document a

substitution of parliament’s role in case of the latter.

Having modeled the arrival of legislative necessities we now turn to modelling the actual

number of acts and ordinances that majority and minority governments pass.

Majority Government (B) Consider a government that has a majority in parliament.

If the necessity arises in a session the government passes an act; if it arises in a break it

promulgates an ordinance. Let X∗Bt and Y ∗Bt be the number of acts and ordinances passed in

the session and break in period t by a majority government.14 Hence, in period 1 we have

X∗B1 = XB1, Y ∗B1 = YB1. (2)

At the end of period 1 the need for Y ∗B1 ordinances survives where

Y ∗B1 v B(φ, YB1). (3)

14We use subscript A for minority governments as an abbreviation for the Hindi alpa sankhyak or “minority” andB for bahumat or “majority”.

11

Page 12: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

As a result, in period t > 1 we have

X∗Bt = XBt + Y ∗Bt−1 and Y ∗Bt = YBt. (4)

Simply put, a majority government faces no constraint on lawmaking in parliament. It passes

an act in parliament when the necessity arises during a session, promulgates an ordinance

when the necessity arises during a break, and if necessary, converts the ordinances into an

act in the subsequent session.

Minority Government (A) Consider a minority government, being a government that

does not have a simple majority that is required to pass acts in parliament. If the necessity

arises in the session, the government attempts to create a coalition in parliament that will

support the act. For each necessity negotiation fails with probability q ∈ (0, 1). In case nego-

tiation succeeds, the law is passed with support from other parties. In case negotiation fails,

the government passes an ordinance in the subsequent break.15 XAt and YAt, as described

in (1), are the number of legislative necessities that arise in session and break in period t,

respectively. Let X∗At and Y ∗At be the number of acts and ordinances passed in the session and

break in period t by a minority government. Following this discussion in period 1 we have

X∗A1 v B(1− q,XA1), Y ∗A1 = YA1 +XA1 −X∗A1, and Y ∗A1 v B(φ, Y ∗A1). (5)

X∗A1 is the number of legislations that the government manages to successfully pass in parlia-

ment of the XA1 legislative necessities. At the end of the session the government promulgates

ordinances Y ∗A1 which are composed of the remaining necessities along with the new necessi-

ties that arise in the break. At the end of the break, the necessity for Y ∗A1 of Y ∗A1 ordinances

survives, and the government attempts to convert these into acts in period 2, and we have

X∗A2 v B(1− q,XA2 + Y ∗A1), Y ∗A2 = YA2 +XA2 + Y ∗A1 −X∗A2 (6)

15If negotiations fail, the government usually prefers to withdraw the bill rather than see it fail in parliament.Thereafter it may use the break to promulgate an ordinance instead. For our theoretical results, it does not matterwhether the bill fails in parliament or whether it is withdrawn before a vote since in both cases it will be countedin XAt and not counted in X∗

At. Empirically, we only observe X∗At, that is successful bills (acts), and therefore for

our purposes unsuccessful bills are the same as bills that are never brought to the parliament out of an anticipationof legislative failure.

12

Page 13: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Similarly, the number of acts and ordinances in legislative session t ≥ 2 is

X∗At v B(1− q,XAt + Y ∗At−1), Y ∗At = YAt +XAt + Y ∗At−1 −X∗At (7)

3.1 Implications

Using this model, we derive five hypotheses that we take to the data in section 4.

3.1.1 Number of acts and ordinances

We compare the expected number of acts and ordinances for minority and majority govern-

ments. The expected number of acts in session t for majority and minority governments are

E(X∗Bt) and E(X∗At). Assuming the same length of sessions, we see that

E(X∗A1) < E(X∗B1)

⇔ (1− q)X1 < X1 (8)

which is always true. For time period t when T ≥ 2, using (4) and (7), we compare the acts

in session t passed by majority and minority governments and find

E(X∗At) < E(X∗Bt)

(1− q)E(XAt + Y ∗At−1) < E(XBt + Y ∗Bt−1) (9)

Since the underlying distribution of X and Y is assumed to be identical regardless of govern-

ment type, the expression above simplifies to

(1− q)(E(X) +

1− (φq)t−1

1− φqφ(E(Y ) + qE(X))

)< E(X) + φE(Y ). (10)

Since the left hand side is increasing in t, taking the limit when t→∞ the inequality simplifies

to the sufficient condition

0 < q(1− φ)(E(X) + φE(Y )). (11)

The inequality in this sufficient condition always holds since φ and q are between 0 and 1.

13

Page 14: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Similarly, for ordinances, we observe that for t ≥ 1

E(Y ∗At) > E(Y ∗Bt)

⇔ 1− (φq)t

1− φq(E(Y ) + qE(X)) > E(Y ) (12)

Since the left hand side is increasing in t, setting t = 1 the inequality simplifies to the sufficient

condition

E(Y ) + qE(X) > E(Y ), (13)

which is always satisfied.

Hypothesis 1. Minority governments are expected to pass fewer acts in each session and

more ordinances in each break, relative to majority governments.

The intuition for this result is simply that minority governments aren’t always successful in

addressing legislative necessities in parliament and resort to using ordinances as an alternative.

Consequently, during the tenure of a minority government, we expect to see fewer acts during

parliamentary sessions and more ordinances during breaks. Note that since Xt and Yt are

random variables, Hypothesis 1 is about the expected number of acts and ordinances and

not the actual number. In other words, empirically, we expect that on average minority

governments pass fewer acts per session and more ordinances per break relative to majority

governments.

3.1.2 Relationship between acts and ordinances within a session

Hypothesis 2. Acts passed in a session and ordinances promulgated in the subsequent break

are negatively related during tenures of minority governments, that isCov(X∗

At,Y∗At)

Var(X∗At)

< 0,

whereas they are unrelated during the tenures of majority governments, that isCov(X∗

Bt,Y∗Bt)

Var(X∗Bt)

=

0.

14

Page 15: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Proof. Consider a minority government in period 1 that passes X∗A1 acts and Y ∗A1 ordinances.

Cov(X∗A1, Y∗A1)

Var(X∗A1)=

EXA1,YA1(Cov(X∗A1, Y

∗A1|XA1, YA1))

Var(X∗A1)+

Cov(E(X∗A|XA1, YA1),E(Y ∗A|XA1, YA1))

Var(X∗A1)

= −EXA1,YA1

(Var(X∗A1|XA1, YA1))

Var(X∗A1)+

Cov((1− q)XA1, YA1 + qXA1)

Var(X∗A1)

= −q(1− q)(E(XA1)−Var(XA1)

Var(X∗A1)

)< 0. (14)

To see this is true, note that XA1 has a binomial distribution B(p, S1). Hence E(XA1) =

pS1 > p(1− p)S1 = Var(XA1). Similarly we observe that in period t ≥ 2

Cov(X∗At, Y∗At)

Var(X∗At)=

EXAt,YAt,Y∗At−1

(Cov(X∗At, Y∗At|XAt, YAt, Y

∗At−1))

Var(X∗At)

+Cov(E(X∗At|XAt, YAt, Y

∗At−1),E(Y ∗At|XAt, YAt, Y

∗At−1))

Var(X∗At)

= −EXAt,YAt,Y

∗At−1

(Var(X∗At|XAt, YAt, Y∗At−1))

Var(X∗At)+q(1− q)Cov(XAt, Y

∗At−1)

Var(X∗At)

= −q(1− q)

(E(XAt)−Var(XAt) + E(Y ∗At)−Var(Y ∗At)

Var(X∗At)

)< 0. (15)

This is true since both XAt and Y ∗At−1 are independent random variables with binomial dis-

tributions and consequently the mean is greater than the variance. Finally, the denominator

Var(X∗At) is strictly positive.

For majority governments in period 1, we observe that Cov(X∗B1, Y∗B1) = Cov(XB1, YB1) =

0. Similarly in period t ≥ 2, we observe Cov(X∗Bt, Y∗Bt) = Cov(XBt + Y ∗Bt−1, YBt) = 0, and

the denominator is strictly positive.

Minority governments are expected to promulgate more ordinances in precisely those

breaks where the preceding session sees a large number of failures of negotiation, and con-

sequently a fewer number of acts being passed. On the other hand, when most negotiations

succeed and the government is able to pass acts in parliament, there is on average less of

a need for ordinances in the subsequent break. Consequently, we observe a negative rela-

tionship between the number of acts passed in a session and the number of ordinances in the

subsequent break. This pattern does not exist for majority governments as ordinances are not

used as a substitute for law making in parliament. This prediction captures the substitution

15

Page 16: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

of parliament by the executive in case of minority governments. Note that this prediction

can be tested within the tenure of a government since it is a prediction about the relationship

between acts and ordinances across contiguous sessions and breaks.

3.1.3 Converting ordinances into acts

In section 2 we discussed how ordinances have a short life span as they expire at the end of

six weeks after the next session commences, unless the parliament votes to make it perma-

nent. Our model allows us to predict how majority and minority governments may differ in

converting ordinances into acts.

Hypothesis 3. The number of unconverted ordinances, and the proportion of ordinances

that remain unconverted are both greater under minority governments.

Proof. The number of ordinances for a minority government is Y ∗At. Of these (1 − q)E(Y ∗At)

get converted into acts in the next period in expectation. Hence the expected number of

unconverted ordinances is given by

E(Y ∗At)− (1− q)E(Y ∗At)

=(1− φ(1− q))E(Y ∗At)

=(1− φ(1− q))1− (φq)t−1

1− φq(E(Y ) + qE(X)) (16)

and the proportion of ordinances that remain unconverted is 1 − φ(1 − q). For a majority

government the number of unconverted ordinances is

E(Y ∗Bt)− φE(Y ∗Bt) = E(Y ∗Bt)(1− φ) = E(Y )(1− φ), (17)

and the proportion is 1− φ.

The proportion of unconverted ordinances is greater under minority governments since

1 − φ(1 − q) > 1 − φ. This, along with the inequality in (12), which shows that minority

governments pass more ordinances, implies that the number of unconverted ordinances must

also be greater.

16

Page 17: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

3.1.4 Re-promulgation of ordinances

If an ordinance is not converted into an act through parliament, it lapses and becomes inactive.

A government can choose to then re-promulgate it in the subsequent break.

Hypothesis 4. Only minority governments re-promulgate ordinances.

To see this is true, note that Y ∗At depends on Y ∗At−1, since a minority government is not

always successful in converting an ordinance into an act. On the other hand, for a majority

government, Y ∗Bt does not depend on Y ∗Bt−1. More simply, a majority government converts

all ordinances that need to be extended into acts in parliament. Consequently, it does not

need to re-promulgate ordinances. This is not the case for a minority government.

3.1.5 Timing of ordinances within a break

Minority governments use ordinances as a fall back when they are unable to pass an act

in parliament. Consequently, minority governments have a stock of unsuccessful acts that

need to be promulgated as ordinances at the start of each break. Majority governments

on the other hand, pass ordinances only when a necessity arises after the commencement

of the break, and this happens with probability p on each day. Consequently we expect to

see ordinances being randomly placed within a break for majority governments, whereas for

minority governments we expect to see more ordinances at the start of the break.

Hypothesis 5. Minority governments pass more ordinances at the start of a break whereas

no such pattern exists for majority governments.

We have attempted to show that the five hypotheses that will be tested in section 4 can

be derived organically from the simple premise that minority governments turn to ordinances

in the presence of failure in getting their legislative agenda through parliament. Hypotheses

1, 3, 4, and 5 are fairly obvious from the premise itself. Hypothesis 2 is less trivial and

requires analysis that is presented in section 3.1.2. This model can be seen as a test of the

internal validity of the five hypotheses in that we see they are mutually consistent, and they

arise naturally in a simple unified framework, without the need for additional parametric

assumptions.

17

Page 18: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

4 Data and Findings

In this section we describe the data sources, explain how we classify governments, and test

the five hypotheses derived in section 3.1.

4.1 Data description and classifying governments

The data we use comes from three sources. First, we extract data on acts from the Parlia-

mentary Handbook which records the number of acts per session of the Lok Sabha. Second,

we use updated data from Dam (2014b) which records the date of promulgation of each or-

dinance. These two sources give us data on the number of acts passed in each session and

ordinances promulgated in each break, from 1952 to 2019. Finally, we collected data from

the website of the Indian parliament on the composition of the two houses of parliament for

our sample period. The summary statistics are presented in Table 1.

We wish to examine whether there is a systematic difference in the pattern of ordinance

making by majority and minority governments. To do so we need to first define what we

mean by majority and minority governments. Since India has a bicameral legislature, we

need to be careful in classifying governments where the ruling party may have a majority in

one house but not the other.

A majority government is when the party (in case of single party governments) or parties

in government together16 (in case of coalition governments) have more than 50% seats in both

houses of parliament. This definition of a parliamentary majority is the one that is most

consistent with the anatomy of lawmaking in India as laid out in the Indian constitution –

to pass a law in parliament it has to separately receive majority support in both houses of

parliament. Hence the government having a majority in both houses of parliament is sufficient

to ensure its legislative agenda goes through parliament without the need for support from

parties outside the government.

Note that based on this definition it is possible for a coalition government, common in

16In India it is common for parties to offer “outside support” to governments. These parties support the govern-ment in parliament but do not accept cabinet positions. Parties supplying such support are not bound by norms ofcollective responsibility. It is a weaker form of support. Hence, we do not add their seats to those of the ruling par-ties when calculating the number of government seats. For example, the Communist Party of India (M) and otheraffiliated parties offered outside support to the Manmohan Singh government between 2004–2009. In calculatingthe status of the Manmohan Singh government we leave out the seats held by this group of parties.

18

Page 19: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

India since the 1990s, to have a majority. This may be problematic in theory. Multiple

coalition partners may find it hard to forge a consensus among themselves when a necessity

for a law arises.17 This may lead to behaviour that is similar to that of minority governments

described in section 3 even when the coalition together has a majority in both houses. As

it happens, in the Indian context this has never been the case: No coalition government has

held a majority in both houses in parliament. The difference between majority and minority

governments can therefore be seen, in the Indian context, as the difference between single

party majority governments and minority governments, where the latter include both single

party minority and coalition minority governments. Table 6 in On-line Appendix A describes

the statuses of all governments in our sample.

A potential objection to this definition is that a government with a large Lok Sabha (lower

house) majority may get around a Rajya Sabha (upper house) minority if it calls for a joint

sitting of the parliament. Summoning a joint sitting of the parliament is difficult in India.

Unless a bill is voted down by the Rajya Sabha, Article 108 of the Constitution requires that

the government wait for six months (not including breaks of four days or more), before a joint

sitting can be held. Consequently, the government may need to wait for a significant length

of time before this condition is met. In practice, there have only been three joint sittings of

the parliament as of 2019.

Nonetheless, we also test whether our results are robust to an alternative definition of

majority. Our second definition of a majority government is when the party (in case of single

party governments) or parties in government together (in case of coalition governments) have

more than 50% seats in in the Lok Sabha, the lower house of parliament.18 This definition of

a parliamentary majority is based on a necessary condition for a government to be considered

as having a majority. The definition is consistent with the fact that although acts have to

pass through both houses of parliament, the Lok Sabha has more de facto and de jure powers.

First, for a party or parties to form a government, it must only prove it has support of the

Lok Sabha. Second, the Rajya Sabha has fewer lawmaking powers than the Lok Sabha as

finance bills need only the support of Lok Sabha to become laws. Third, the Lok Sabha has

17For a detailed description and analysis of politics and governance under coalition governments in India seeRuparelia (2015) and Sridharan (2014).

18This is how a minority government is commonly defined in the literature. See for example p6 in Strøm (1990).

19

Page 20: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 1: Summary statistics

Variable Obs Mean Std. Dev. Min MaxBy day in breakNumber of ordinances 15,013 .046 .279 0 8At least 1 ordinance (Binary) 15,013 .036 .185 0 1By break in at least 1 houseOrdinances by break 247 2.798 3.365 0 24Session length in days 247 38.304 19.957 3 123Break length in days 247 60.781 28.91 3 177By Lok Sabha sessionNumber of acts 209 18.105 9.4 0 47Majority in both houses (Binary) 209 .488 .501 0 1Majority in Lok Sabha (Binary) 209 .608 .489 0 1Unconverted ordinances 193 .927 1.807 0 13Proportion of ordinances unconverted 193 .236 .359 0 .999

See the discussion of equation (22) for the definition of proportion of ordinances unconverted.

twice as many members as the Rajya Sabha. As a result, perhaps anticipating failure in a

joint sitting, parties in the Rajya Sabha may be inclined to negotiate with the government.

We find that the results are qualitatively similar regardless of which of the two definitions

of majority we use. In the Indian context this is unsurprising. There are only two governments

in our sample that change type (from minority to majority) when we go from the first to the

second definition: The Morarji Desai government from 1977 – 1979 and the Modi government

from 2014 – 2019 (see Table 6). Both these governments had majority in the Lok Sabha but

minority in the Rajya Sabha. Consequently, the distinction between the two definitions is

empirically irrelevant since the classification of all other governments, comprising most of our

sample, remains unchanged. The results with majority being defined as a majority in Lok

Sabha are presented in On-line Appendix C.

In the summary statistics in Table 1 we observe that about 48.8% of the sessions in our

sample period have majority governments based on our baseline definition of having majority

in both houses, whereas 60.8% of sessions have majority governments based on at least having

majority in the Lok Sabha.

20

Page 21: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

4.2 Number of acts and ordinances

Hypothesis 1 predicts that minority governments pass more ordinances and fewer acts relative

to majority governments. In the first four columns we regress

X∗t = α+ βMajorityt + Z′tΓ + εt. (18)

Recall that X∗t is the total number of acts passed in session t. Our main coefficient of interest

is β, which measures the average difference between majority and minority governments in

the number of acts passed per session.

The results are presented in the first four columns of Table 2. We observe that majority

governments pass more acts per session and this effect is significant at the 1% level. To

ensure that the observed difference between majority and minority governments is not driven

by unobservables, we include twelve month dummies, that is a dummy for the month in

which the session started. This attempts to absorb any seasonality in law making. Since the

theoretical results are conditional on the length of the session and the break, we also include

these in the session controls as it is possible that these may affect the number of acts passed.

We also include session within Lok Sabha dummies (typically there are 15 sessions if a Lok

Sabha lasts the full 5 years) in the set of session controls. This is an attempt to capture any

common trends in the legislative arcs of governments over their tenure, such as governments

being more legislatively active at the start of their term. Finally, we also include a linear time

trend to absorb any long run political trends that affect law making but are also correlated

with the type of governments that get elected.

In the columns (5) to (8) we regress

Y ∗t = α+ βMajorityt + Z′tΓ + εt, (19)

where Y ∗t is the total ordinances passed in the break of session t. The right hand side of the

equation remains the same. We see that the estimates for β, our coefficient of interest, are

similar across columns (5) and (8). Quantitatively, our results indicate that minority gov-

ernments pass one more ordinance per break relative to majority governments. Furthermore,

majority governments pass five to seven more acts per session relative to minority govern-

21

Page 22: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 2: Acts and ordinances by majority and minority governments

Acts in a session Ordinances in a break(1) (2) (3) (4) (5) (6) (7) (8)

Majority 7.148∗∗∗ 7.048∗∗∗ 5.771∗∗∗ 5.442∗∗∗ -1.011∗∗ -1.182∗∗ -0.921∗ -1.068∗

(1.203) (1.134) (1.117) (1.690) (0.430) (0.462) (0.482) (0.628)

Constant 14.62∗∗∗ 10.31∗∗∗ -0.637 0.532 3.342∗∗∗ 4.924∗∗∗ 3.307∗∗∗ 3.856∗∗

(0.883) (3.825) (4.687) (6.920) (0.335) (1.177) (1.108) (1.708)

Month dummies No Yes Yes Yes No Yes Yes Yes

Session controls No No Yes Yes No No Yes Yes

Linear time trend No No No Yes No No No YesObservations 209 209 209 209 247 247 247 247R2 0.145 0.308 0.466 0.466 0.023 0.143 0.306 0.306

Dependent variable in the first four columns is the number of acts passed in parliament during a session and in thelast four columns is the number of ordinances promulgated in a break. Session controls include dummies for sessionwithin Lok Sabha, the length of the session in days and the length of the break in days. Robust standard errorsreported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

ments. Note that although in theory the number of sessions and breaks should be the same, in

our data there are more breaks (247) than sessions (209). This is because the acts data comes

from Lok Sabha sessions whereas we have disaggregated data for ordinances. As ordinances

are promulgated when at least one of the two houses of parliament is not is session, there is

a larger number of breaks relative to the number of Lok Sabha sessions.

An alternative way of testing Hypothesis 1 is to use maximum likelihood estimation

(MLE). To do so, we could use the prediction of the model that Y ∗t and X∗t are binomi-

ally distributed, and compare the means of the observed distributions of Y ∗t and X∗t for

majority and minority governments. We use ordinary least squares (OLS) for the sake of

simplicity. Moreover, the binomial distribution is the discrete approximation of the normal

distribution, and the MLE estimator for β for a normally distributed error term is the same

as the OLS estimator.

One concern with these results is that they do not convincingly control for time varying

trends. The Indian party system has fragmented over time: Majority governments used to

be the norm but over time they have become the exception (see Table 6). It is possible that

omitted factors that change the Indian polity over time lead to more minority governments

being elected, and are also independently responsible for the promulgation of more ordinances

and fewer acts. The linear time trend that is included in columns (4) and (8) may not be

22

Page 23: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

adequate to absorb these changes if these changes were non-linear in nature. Since most of

the variation in the type of government is caused by Lok Sabha elections, it is not possible

to estimate equations (18) and (19) with Lok Sabha dummies.

The challenge here is that Hypothesis 1 makes a claim about differences in the means of

acts and ordinances between majority and minority government. In contrast, Hypothesis 2

makes a claim that can be tested separately within the tenures of government, and this allows

us to control for year dummies that absorb unobservable, potentially non-linear, time varying

factors. This is what we turn to next.

4.3 Substitution of acts by ordinances

Hypothesis 2, which predicts a negative relationship between ordinances and acts for minority

governments and no correlation for the same for majority governments. Note that the sample

analogue ofCov(X∗

At,Y∗At)

Var(X∗At)

in Hypothesis 2 is also the OLS estimator for βA in the regression

Y ∗t = α0 + α1 Majorityt + βAX∗t + λX∗t ×Majorityt + ε. (20)

Similarly, the sample analogue ofCov(X∗

Bt,Y∗Bt)

Var(X∗Bt)

is also the OLS estimator for βA + λ. To see

this clearly, note that the sample analogue for the expressions in Hypothesis 2 are the OLS

estimators for the slope coefficient if we regress ordinances on acts separately for the majority

and minority government subsamples. This allows us to conveniently test Hypothesis 2 with

standard regression tools by testing whether βA < 0 and βA + λ = 0.

The results are reported in Table 3. The estimates for βA are consistently negative and

significant and this confirms that during the tenures of minority governments, more bills are

followed by fewer ordinances and vice versa. On the other hand, the estimates for λ are

consistently positive. The last row of the table presents the p-value for the null hypothesis

that βA + λ = 0. Our failure to reject this null hypothesis across all specifications confirms

that the number of bills and ordinances across contiguous session-break periods is unrelated

in the tenure of majority governments.

As mentioned earlier, these results are a stronger confirmation of the differences in the

patterns of acts and ordinances for majority and minority governments than the ones pre-

23

Page 24: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 3: Substitution of acts with ordinances

(1) (2) (3) (4) (5) (6) (7)Majority -0.482 -0.624 -0.405

(0.454) (0.524) (0.558)

Number of Acts -0.0673∗ -0.0754∗ -0.0894∗ -0.110∗∗∗ -0.133∗∗∗ -0.127∗∗∗ -0.143∗∗∗

(0.0378) (0.0415) (0.0459) (0.0389) (0.0432) (0.0409) (0.0466)

Number of Acts × 0.121∗∗ 0.130∗∗ 0.105∗ 0.0732 0.0811 0.0852 0.0594Majority (0.0524) (0.0568) (0.0567) (0.0515) (0.0604) (0.0547) (0.0628)

Constant 3.326∗∗∗ 4.265∗∗∗ 0.352 -0.484 -1.390 -0.912 -3.583(0.299) (1.096) (1.223) (1.060) (2.002) (1.428) (2.543)

Month dummies No Yes Yes No No Yes Yes

Session controls No No Yes Yes Yes Yes Yes

Lok Sabha dummies No No No Yes No Yes No

Year dummies No No No No Yes No YesObservations 209 209 209 209 209 209 209R2 0.029 0.070 0.236 0.428 0.638 0.464 0.665p-value H0 : βA + λ = 0 0.142 0.173 0.688 0.356 0.314 0.326 0.124

Dependent variable is total ordinances in a break. Session controls include dummies for session within Lok Sabha,the length of the session in days and the length of the break in days. Robust standard errors reported in theparentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

24

Page 25: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

sented in section 4.2. This is because, while comparing the number of acts passed in a session

to the number of ordinances in the subsequent break, we can control for more omitted vari-

able concerns. In particular, in addition to the controls already described in section 4.2, we

also control for Lok Sabha dummies. This absorbs any variation across governments in how

prolific they are in using ordinances. Alternatively, we also control for year dummies which

controls for temporal variation in ordinances at a finer year-on-year level. These control for

any non-linear temporal changes that simultaneously affect the number of acts and ordinances

such as the fragmentation of the Indian polity over time. Note that year dummies and Lok

Sabha dummies cannot be simultaneously included since the variation in Lok Sabha is purely

temporal.

Although it is constitutionally permissible to promulgate an ordinance when at least one of

the two houses in parliament in not in session, we never observe ordinances being promulgated

when the Lok Sabha is in session and the Rajya Sabha is not. Therefore, to map Y ∗t to X∗t ,

we sum the ordinances passed between two Lok Sabha sessions and this leaves us with 209

sessions and corresponding breaks.19

4.4 Converting ordinances into acts

The mechanism to ensure that an ordinance becomes permanent is to convert an ordinance

into an act of parliament by presenting it before parliament and calling for a vote. If minority

governments use ordinances to systematically bypass the parliament, we expect that fewer

ordinances will be successfully converted to acts by the parliament during their tenure, as

predicted by Hypothesis 3.

Recall that Y ∗t is the number of ordinances in period t. Let X∗t+1 ∈ {0, 1, . . . Y ∗t } be the

number of acts in period t+ 1 that were ordinances in period t. We regress

Y ∗t − X∗t+1 = α+ βMajorityt + Z′tΓ + εt. (21)

Results are reported in Table 4. In columns (1) to (4), we observe that majority governments

19There is one break at the start of the first parliament between the time the constitution of India came intoforce but before the first Lok Sabha session commenced, which cannot be mapped to a prior session and is thereforedropped from the analysis.

25

Page 26: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

have fewer unconverted ordinances than minority governments. This is true after including

month dummies and session level controls including session within Lok Sabha dummies, the

length of session t+ 1, and the length of the break in t when the ordinances in question were

promulgated.20

One concern with these results is that they are driven by the fact that majority govern-

ments pass fewer ordinances, and consequently have fewer ordinances to convert into acts in

the subsequent interval. However, Hypothesis 3 also predicts that the fraction of ordinances

that remain unconverted is greater under minority governments. We test this by using the

fraction of ordinances that remain unconverted as the dependent variable. We regress

Y ∗t − X∗t+1

Y ∗t + 0.01= α+ βMajorityt + Z′tΓ + εt. (22)

A positive constant 0.01 is added to the denominator so that the fraction is always well

defined, since the number of ordinances in the previous break can be zero.21 The estimates

reported in columns (5) to (8) suggest that majority governments have about 25% fewer

unconverted ordinances relative to minority governments.

4.5 Re-promulgation of ordinances

Hypothesis 4 predicts that only minority governments will re-promulgate ordinances. It is

striking that all 76 instances of re-promulgation of ordinances, as of May 2019, have occurred

during the tenures of minority governments. If instead we define a majority government as one

that has majority in the Lok Sabha (but not necessarily in the Rajya Sabha), then we observe

12 instances of re-promulgation by majority governments as the Modi government (2014 –

2019) is now classified as a majority government. An example of this is the re-promulgation

of LARRO mentioned in the introduction. In that case the numbers then change to 64

re-promulgations by minority governments relative to 12 by majority governments.

20Note that the number of observations drops since we exclude the contiguous break-session observations wherethe government has changed, since a new government may not convert the ordinances of the previous governmentinto acts for reasons unrelated to the lack of parliamentary majority. Note also that in this regression, like the onediscussed in section 4.2, we cannot control for year or Lok Sabha dummies as these are almost perfectly collinearwith the Majority variable.

21Results with the sub sample of breaks where the number of ordinances is strictly positive are very similar inmagnitude and statistical significance.

26

Page 27: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 4: Conversion of ordinances into acts

Unconverted ordinances Fraction of ordinances unconverted(1) (2) (3) (4) (5) (6) (7) (8)

Majority -1.142∗∗∗ -1.133∗∗∗ -1.118∗∗∗ -1.145∗∗∗ -0.260∗∗∗ -0.251∗∗∗ -0.259∗∗∗ -0.271∗∗∗

(0.244) (0.275) (0.249) (0.264) (0.0479) (0.0509) (0.0489) (0.0530)

Constant 1.490∗∗∗ 1.533∗∗∗ 0.556 0.776 0.364∗∗∗ 0.478∗∗ 0.250∗ 0.237(0.230) (0.579) (0.655) (1.234) (0.0411) (0.199) (0.140) (0.303)

Month dummies No Yes No Yes No Yes No Yes

Session controls No No Yes Yes No No Yes YesObservations 193 193 193 193 193 193 193 193R2 0.100 0.119 0.222 0.256 0.132 0.155 0.234 0.252

Dependent variable in the first four columns is the number of ordinances from the previous break that are notconverted to an act in the session. The dependent variable in the last four columns is the fraction of ordinancesfrom the previous break that are not converted into a legislation in the session as described in equation (22). Sessioncontrols include dummies for session within Lok Sabha, the length of the session in days and the length of the breakin days. Robust standard errors reported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

By its very nature re-promulgation implies that the original legislative necessity arose be-

fore the session of parliament, and the government had a full session to convert the ordinance

into an act. Re-promulgation implies that the ordinance died during the session but the

government resurrected it in the next break. The fact that all instances of re-promulgation

have happened under minority governments strongly suggests that ordinances are being used

in a way that is inconsistent with the constitutional mandate in Article 123 of only using the

mechanism for immediate necessities that arise when parliament is on a break.

4.6 Timing of ordinances within a break

In this section we examine the timing of ordinances within breaks for majority and minority

governments. Based on Hypothesis 5, we expect that minority governments will pass more

ordinances at the start of a break whereas ordinances will be randomly dispersed in a break

for majority governments.

Figure 3 shows the empirical cumulative distribution of ordinances over the proportion

of time elapsed from the start of the break for majority and minority governments.22 The

distribution for majority governments first order stochastically dominates that of minority

22To compute the cumulative distribution, we have aggregated the number of ordinances across all breaks sepa-rately for majority and minority governments.

27

Page 28: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

0.2

.4.6

.81

0 .2 .4 .6 .8 1Proportion of time to the end of the break

Majority governments Minority governments

Figure 3: Ordinance cdf within a break

governments indicating that at any point of time in a break, minority governments have

promulgated a greater share of their ordinances earlier relative to majority governments.

To examine this more closely we run the following regression:

Yst =αt + λDays since start of the breaks ×Majorityt

+ βA Days since start of the breaks + Z′stΓ + εst (23)

where Ys is number of ordinances passed on day s in break t. We regress this on the number

of days that have elapsed since the start of the break on day s and the interaction of this

variable with the indicator for majority government. Note that the variable Majorityt is

subsumed whenever Lok Sabha dummies, or year dummies, or overall session dummies are

included since there is almost no variation in the majority status of the government within

the tenure of a Lok Sabha. Based on Hypothesis 5, we expect βA < 0 since the likelihood

of an ordinance in the tenure of a minority government should be higher at the start of a

break. We also expect βA + λ = 0, since no such relationship ought to exist for majority

governments.

Hypothesis 5 makes a claim about differences in behaviour between majority and minor-

ity governments within each break. Since the unit of observation is a day in a break, this

28

Page 29: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 5: Timing of ordinances within a break

(1) (2) (3) (4) (5) (6) (7)Majority -0.0461∗∗∗ -0.0464∗∗∗ -0.0461∗∗∗ -0.0474∗∗∗

(0.00810) (0.00801) (0.00799) (0.00816)

Days from start of break / 100 -0.0625∗∗∗ -0.0548∗∗∗ -0.0544∗∗∗ -0.0617∗∗∗ -0.0559∗∗∗ -0.0553∗∗∗ -0.0594∗∗∗

(0.0109) (0.0109) (0.0109) (0.0131) (0.0131) (0.0129) (0.0122)

Days from start of break / 100 0.0971∗∗∗ 0.0995∗∗∗ 0.0989∗∗∗ 0.0972∗∗∗ 0.0847∗∗∗ 0.0821∗∗∗ 0.0708∗∗∗

× Majority (0.0158) (0.0162) (0.0162) (0.0162) (0.0153) (0.0165) (0.0165)

Constant 0.0756∗∗∗ 0.0974∗∗∗ 0.0943∗∗∗ 0.104∗∗∗ 0.0612∗∗∗ 0.0800∗∗∗ 0.0336(0.00710) (0.0114) (0.0121) (0.0146) (0.0154) (0.0309) (0.0260)

Month dummies No Yes Yes Yes Yes Yes No

Day of week dummies No No Yes Yes Yes Yes Yes

Session controls No No No Yes Yes Yes No

Lok Sabha dummies No No No No Yes No No

Year dummies No No No No No Yes No

Overall session dummies No No No No No No YesObservations 15013 15013 15013 15013 15013 15013 15013R2 0.003 0.006 0.009 0.010 0.017 0.023 0.035p-value H0 : βA + λ = 0 0.003 0.001 0.001 0.015 0.039 0.081 0.348

Dependent variable is the number of ordinances promulgated on a day in the break. The number of days thathave elapsed since the start of the break on day s is divided by 100 to reduce the number of decimal places in thecoefficient estimates. Session controls include dummies for session within Lok Sabha, the length of the session indays and the length of the break in days. Overall session dummies are dummies for each time at least one of thehouses of parliament was not in session. Robust standard errors reported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05,∗∗∗ p < 0.01

29

Page 30: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

specification allows us to control for overall session dummies αt, that is, one dummy for each

break in our sample. These account for things such as differences in the total number of

ordinances passed in each break, the differences in particular trajectory of a government that

could induces greater or fewer ordinances in a particular break, or any other unobservable

time-varying trends. The results in column (7) therefore present evidence of difference be-

tween majority and minority governments that is least likely to be contaminated by omitted

variables.

The results presented in Table 5 strongly support the hypothesis that minority govern-

ments pass ordinances early in the break. We observe that the estimates for βA remain

negative and significant. The last row in Table 5 reports the p-value for the null hypothesis

that βA +λ = 0. We find that in all cases the estimates of βA +λ are positive which suggests

that unlike minority governments who promulgate ordinances early in a break, majority gov-

ernments do so later in a break. However, in our most robust specification with overall session

dummies we fail to reject the null hypothesis which suggests that the timing of ordinances

within a break is random for majority governments.

A potential problem with the results in Table 5 is that the dependent variable has a large

number of zeroes as most days in an average break see no ordinances. To handle this concern

we re-estimate (23) using negative binomial regression. The results are reported in Table 7

in On-line Appendix B. We observe that the inferences on βA and βA +λ remain unchanged.

Finally, we change the dependent variable to a dummy that indicates whether at least one

ordinance was promulgated on the day and re-estimate (23). The results reported in Table 8

in On-line Appendix B are very similar to the ones in Table 5.

5 Conclusion

The Indian constitution authorises the executive to promulgate primary legislation when

immediate action is required and the parliament is not in session. We observe that this

provision has been abused by governments to enact their legislative agenda by bypassing

parliamentary scrutiny when they lack support in parliament. Based on this simple idea, we

have constructed a statistical model that makes five predictions that are borne out in the

30

Page 31: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

empirical results.

Our results indicate that the strategic use of decree power to circumvent the legislature

is not limited to presidential systems. Therefore, the ability of parliamentary systems to

deliver congruence between the executive and the legislature may be overstated – minority

governments compensate for their weakness in the legislature by attempting to substitute out

legislative functioning through decrees. Although this behaviour may be welfare enhancing

as it allows minority governments to find a way out of potential legislative dysfunction, it

also allows them to exert less effort to forge a consensus or at least a working majority in

parliament. This may create long run institutional costs as executive power is strengthened

at the expense of the legislature. We document a consistent pattern of substitution of the

parliament by the executive when it comes to law making by minority governments. To the

extent that this behaviour is undesirable, our results suggest that it may be worthwhile having

a constitutional mandate that more clearly circumscribes the boundary on the executive’s

ability to legislate based on criteria that are judicially verifiable.

This analysis, however raises an interesting question: Why does parliament not censure

minority governments for their systematic abuse of the ordinance making power? After

all, other parties in parliament can summon a no-confidence motion to oust a government

that lacks a majority in the Lok Sabha. Tentatively, there may be three reasons. First,

coordination on a no-confidence motion is harder than opposing individual acts in parliament.

Second, even if coordination is possible, opposition parties may not find it politically attractive

to trigger elections. Third, political parties may not view such use of ordinances as a deep

constitutional infraction sufficient to bring down governments. The abuse of Article 123 is a

common constitutional occurrence; all parties when in power have invoked it. This coexistence

of abuse and the absence of parliamentary censure remains an avenue for further research.

References

Carey, John M and Mathew Shugart (1998). “Calling out the Tanks or Just Filling

Out the Forms?” In: Executive Decree Authority. New York: Cambridge University

Press, pp. 1–29.

31

Page 32: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Cheibub, Jose Antonio (2007). Presidentialism, Parliamentarism and Democracy. New

York: Cambridge University Press.

Cheibub, Jose Antonio and Fernando Limongi (2010). “From Conflict to Coordina-

tion: Perspectives on the Study of Executive-Legislative Relations”. Revista Ibero-

americana de Estudos Legislativos 1 (1), pp. 38–53.

Christenson, Dino P. and Douglas L. Kriner (2017). “Constitutional qualms or poli-

tics as usual? The factors shaping public support for unilateral action”. American

Journal of Political Science 61 (2), pp. 335–349.

Dam, Shubhankar (2014a). “An institutional alchemy: India’s two parliaments in com-

parative context”. Brooklyn Journal of International Law 39 (2), pp. 613–655.

— (2014b). Presidential Legislation in India. New York: Cambridge University Press.

— (2016). “Making parliament irrelevant: A postcard from India”. The Theory and

Practice of Legislation 4 (1), pp. 65–78.

Gallagher, Michael, Michael Laver, and Peter Mair (2005). Representative government

in modern Europe. 4th ed. Boston: McGraw-Hill.

Howell, William (2003). Power Without Persuasion: The Politics of Direct Presidential

Action. Princeton: Princeton University Press.

Krishna Kumar v State of Bihar (2017). Vol. SC 213. All India Reporter.

Lowande, Kenneth (2018). “Delegation or unilateral action?” Journal of Law Eco-

nomics & Organization 34 (1), pp. 54–78.

Martin, Lanny W and Georg Vanberg (2004). “Policing the Bargain: Coalition Govern-

ment and Parliamentary Scrutiny”. American Journal of Political Science 48 (3),

pp. 13–27.

— (2005). “Coalition policymaking and legislative review”. American Political Science

Review 99 (1), pp. 93–106.

— (2008). “Coalition government and political communication”. Political Research

Quarterly 61 (3), pp. 502–516.

32

Page 33: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Mayer, Kenneth R. (2001). With the stroke of a pen: Executive orders and presidential

power. Princeton: Princeton University Press.

Negretto, Gabriel L. (2004). “Government Capacities and Policy Making by Decree in

Latin America: The Cases of Brazil and Argentina”. Comparative Political Studies

37 (5), pp. 531–562.

Neto, Octavio Amorim, Gary Cox, and Mathew D. McCubbins (2003). “Agenda Power

in Brazil’s Camara dos Deputados, 1989-98”. World Politics 55 (4), pp. 550–578.

Pereira, Carlos, Timothy J. Power, and Lucio Renno (2005). “Under What Conditions

Do Presidents Resort to Decree Power? Theory and Evidence from the Brazilian

Case”. Journal of Politics 67 (1), pp. 178–200.

— (2008). “Agenda Power, Executive Decree Authority, and the Mixed Results of

Reform in the Brazilian Congress”. Legislative Studies Quarterly 33 (1), pp. 5–33.

Power, Timothy (1998). “The Pen is Mightier than the Congress: Presidential Decree

Power in Brazil”. In: Executive Decree Authority. New York: Cambridge University

Press.

R. C. Cooper v Union of India (1970). Vol. SC 564. All India Reporter.

Reich, Gary (2002). “Executive Decree Authority in Brazil: How Reactive Legislators

Influence Policy”. Legislative Studies Quarterly 27 (1), pp. 5–31.

Rudalevige, Andrew (2005). The new imperial presidency: Renewing presidential power

after Watergate. Ann Arbor: University of Michigan Press.

Ruparelia, Sanjay (2015). Divided we govern: Coalition politics in modern India. C.

Hurst & Co. Ltd.

Sridharan, E., ed. (2014). Coalition politics in India: Selected issues at the Centre and

the States. New Delhi: Academic Foundation.

Strøm, Kaare (1990). Minority government and majority rule. New York: Cambridge

University Press.

33

Page 34: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

A Data description

Text of Article 123

123. Power of President to promulgate Ordinances during recess of Parliament

1. If at any time, except when both Houses of Parliament are in session, the President is

satisfied that circumstances exist which render it necessary for him to take immediate

action, he may promulgate such Ordinance as the circumstances appear to him to require

2. An Ordinance promulgated under this article shall have the same force and effect as an

Act of Parliament, but every such Ordinance

(a) shall be laid before both Houses of Parliament and shall cease to operate at the

expiration of six weeks from the reassemble of Parliament, or, if before the expi-

ration of that period resolutions disapproving it are passed by both Houses, upon

the passing of the second of those resolutions; and

(b) may be withdrawn at any time by the President.

Explanation: Where the Houses of Parliament are summoned to reassemble on

different dates, the period of six weeks shall be reckoned from the later of those

dates for the purposes of this clause.

3. If and so far as an Ordinance under this article makes any provision which Parliament

would not under this Constitution be competent to enact, it shall be void.

34

Page 35: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 6: Government classification by majorities

Prime Minister Start date End date Lok Sabha Rajya Sabha Acts Ordinancesmajority majority

Jawaharlal Nehru 13.05.52 10.05.57 yes yes 348 40Jawaharlal Nehru 11.05.57 31.03.62 yes yes 317 18Jawaharlal Nehru 01.04.62 26.05.64 yes yes 129 7Lalbahadur Shashtri 09.06.64 11.01.66 yes yes 81 9Indira Gandhi 24.01.66 03.03.67 yes yes 57 15Indira Gandhi 04.03.67 16.11.69 yes yes 143 29Indira Gandhi 17.11.69 14.03.71 no yes 73 9Indira Gandhi 15.03.71 23.03.77 yes yes 482 99Morarji Desai 24.03.77 27.07.79 yes no 130 21Charan Singh 28.07.79 13.01.80 no no 0 7Indira Gandhi 14.01.80 30.10.84 yes yes 329 56Rajiv Gandhi 31.10.84 30.12.84 yes yes 0 2Rajiv Gandhi 31.12.84 01.12.89 yes yes 334 35VP Singh 02.12.89 09.11.90 no no 33 10Chandra Shekhar 10.11.90 20.06.91 no no 30 6PV Narasimha Rao 21.06.91 15.05.96 no no 277 108AB Vajpayee 16.05.96 31.05.96 no no 0 0HD Deve Gowda 01.06.96 20.04.97 no no 56 23Inder Gujral 21.04.97 18.03.98 no no 5 23AB Vajpayee 19.03.98 12.10.99 no no 56 25AB Vajpayee 13.10.99 21.05.04 no no 297 33Manmohan Singh 22.05.04 21.05.09 no no 248 36Manmohan Singh 22.05.09 25.05.14 no no 179 25Narendra Modi 26.05.14 15.05.19 yes no 180 55

35

Page 36: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

B Additional results

Table 7: Timing of ordinances within a break (Negative binomial)

(1) (2) (3) (4) (5) (6) (7)Number of OrdinancesMajority -1.050∗∗∗ -1.047∗∗∗ -1.053∗∗∗ -1.071∗∗∗

(0.163) (0.163) (0.163) (0.165)

Days from start of break / 100 -1.553∗∗∗ -1.381∗∗∗ -1.374∗∗∗ -1.596∗∗∗ -1.368∗∗∗ -1.352∗∗∗ -1.315∗∗∗

(0.279) (0.289) (0.286) (0.323) (0.316) (0.297) (0.273)

Days from start of break / 100 2.420∗∗∗ 2.472∗∗∗ 2.519∗∗∗ 2.460∗∗∗ 1.923∗∗∗ 1.751∗∗∗ 1.535∗∗∗

× Majority (0.381) (0.387) (0.392) (0.392) (0.359) (0.348) (0.356)

Month dummies No Yes Yes Yes Yes Yes No

Day of week dummies No No Yes Yes Yes Yes Yes

Session controls No No No Yes Yes Yes No

Lok Sabha dummies No No No No Yes No No

Year dummies No No No No No Yes No

Overall session dummies No No No No No No YesObservations 15013 15013 15013 15013 15013 15013 15013R2

p-value H 0 : β A+ λ = 0 0.001 0.000 0.000 0.007 0.058 0.162 0.364

Dependent variable is the number of ordinances promulgated on a day in the break. The number of days thathave elapsed since the start of the break on day s is divided by 100 to reduce the number of decimal places in thecoefficient estimates. Session controls include dummies for session within Lok Sabha, the length of the session indays and the length of the break in days. Overall session dummies are dummies for each time at least one of thehouses of parliament was not in session. Robust standard errors reported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05,∗∗∗ p < 0.01

36

Page 37: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 8: Timing of ordinances within a break

(1) (2) (3) (4) (5) (6) (7)Majority -0.0236∗∗∗ -0.0241∗∗∗ -0.0238∗∗∗ -0.0244∗∗∗

(0.00487) (0.00491) (0.00490) (0.00495)

Days from start of break / 100 -0.0359∗∗∗ -0.0312∗∗∗ -0.0309∗∗∗ -0.0352∗∗∗ -0.0320∗∗∗ -0.0312∗∗∗ -0.0366∗∗∗

(0.00628) (0.00634) (0.00634) (0.00751) (0.00756) (0.00744) (0.00733)

Days from start of break / 100 0.0547∗∗∗ 0.0575∗∗∗ 0.0572∗∗∗ 0.0564∗∗∗ 0.0495∗∗∗ 0.0467∗∗∗ 0.0406∗∗∗

× Majority (0.00979) (0.0102) (0.0102) (0.0102) (0.00979) (0.0105) (0.0110)

Constant 0.0513∗∗∗ 0.0648∗∗∗ 0.0636∗∗∗ 0.0684∗∗∗ 0.0476∗∗∗ 0.0527∗∗ 0.0285(0.00380) (0.00614) (0.00718) (0.00881) (0.0111) (0.0219) (0.0187)

Month dummies No Yes Yes Yes Yes Yes No

Day of week dummies No No Yes Yes Yes Yes Yes

Session controls No No No Yes Yes Yes No

Lok Sabha dummies No No No No Yes No No

Year dummies No No No No No Yes No

Overall session dummies No No No No No No YesObservations 15013 15013 15013 15013 15013 15013 15013R2 0.002 0.006 0.010 0.010 0.016 0.021 0.031p-value H 0 : β A+ λ = 0 0.012 0.001 0.001 0.022 0.051 0.106 0.640

Dependent variable is an indicator for whether an ordinance was promulgated on the day. The number of days thathave elapsed since the start of the break on day s is divided by 100 to reduce the number of decimal places in thecoefficient estimates. Session controls include dummies for session within Lok Sabha, the length of the session indays and the length of the break in days. Overall session dummies are dummies for each time at least one of thehouses of parliament was not in session. Robust standard errors reported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05,∗∗∗ p < 0.01

C Alternative definition of majority

In this section we present the results where the definition of majority is changed to majority

in the Lok Sabha.

37

Page 38: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 9: Acts and ordinances by majority and minority governments

Acts in a session Ordinances in a break(1) (2) (3) (4) (5) (6) (7) (8)

LS Majority 4.629∗∗∗ 4.914∗∗∗ 3.731∗∗∗ 1.503 -1.241∗∗ -1.555∗∗∗ -1.370∗∗ -1.470∗∗

(1.314) (1.229) (1.168) (1.367) (0.479) (0.487) (0.540) (0.573)

Constant 15.29∗∗∗ 8.919∗ -1.652 9.579 3.612∗∗∗ 5.889∗∗∗ 4.313∗∗∗ 4.818∗∗∗

(1.049) (4.527) (5.646) (6.656) (0.417) (1.427) (1.367) (1.701)

Month dummies No Yes Yes Yes No Yes Yes Yes

Session controls No No Yes Yes No No Yes Yes

Linear time trend No No No Yes No No No YesObservations 209 209 209 209 247 247 247 247R2 0.058 0.241 0.420 0.442 0.031 0.160 0.321 0.321

This table is the analogue of Table 2 with LS majority. LS Majority is an indicator for whether the government hasmore than 50% seats in the Lok Sabha. Dependent variable in the first four columns is the number of acts passedin parliament during a session and in the last four columns is the number of ordinances promulgated in a break.Session controls include dummies for session within Lok Sabha, the length of the session in days and the length ofthe break in days. Robust standard errors reported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

Table 10: Substitution of acts with ordinances

(1) (2) (3) (4) (5) (6) (7)LS Majority -0.556 -0.717 -0.656

(0.465) (0.491) (0.547)

Number of Acts -0.0799∗ -0.0893∗ -0.107∗∗ -0.115∗∗ -0.153∗∗∗ -0.132∗∗∗ -0.160∗∗∗

(0.0445) (0.0466) (0.0534) (0.0448) (0.0518) (0.0468) (0.0548)

Number of Acts × 0.113∗∗ 0.126∗∗ 0.114∗∗ 0.0722 0.102∗ 0.0813 0.0795LS Majority (0.0545) (0.0555) (0.0563) (0.0537) (0.0600) (0.0549) (0.0607)

Constant 3.519∗∗∗ 5.205∗∗∗ 1.089 -0.428 -1.402 -0.814 -3.496(0.369) (1.511) (1.251) (1.047) (2.041) (1.433) (2.538)

Month dummies No Yes Yes No No Yes Yes

Session controls No No Yes Yes Yes Yes Yes

Lok Sabha dummies No No No Yes No Yes No

Year dummies No No No No Yes No YesObservations 209 209 209 209 209 209 209R2 0.033 0.077 0.248 0.428 0.642 0.464 0.668p-value H0 : βA + λ = 0 0.297 0.275 0.801 0.228 0.244 0.174 0.077

This table is the analogue of Table 3 with LS majority. LS Majority is an indicator for whether the governmenthas more than 50% seats in the Lok Sabha. Dependent variable is total ordinances in a break. Session controlsinclude dummies for session within Lok Sabha, the length of the session in days and the length of the break in days.Robust standard errors reported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

38

Page 39: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Figure 4: Ordinance cdf within a break

0.2

.4.6

.81

0 .2 .4 .6 .8 1Proportion of time to the end of the break

LS Majority governments LS Minority governments

This figure is the analogue of Figure 3 with Lok Sabha majority.

Table 11: Conversion of ordinances into acts

Unconverted ordinances Fraction of ordinances unconverted(1) (2) (3) (4) (5) (6) (7) (8)

LS Majority -0.904∗∗∗ -0.890∗∗∗ -0.907∗∗∗ -0.898∗∗∗ -0.175∗∗∗ -0.176∗∗∗ -0.175∗∗∗ -0.182∗∗∗

(0.299) (0.314) (0.307) (0.319) (0.0547) (0.0576) (0.0560) (0.0605)

Constant 1.480∗∗∗ 2.140∗∗∗ 0.636 1.661 0.343∗∗∗ 0.591∗∗ 0.245 0.421(0.280) (0.742) (0.690) (1.475) (0.0466) (0.231) (0.149) (0.348)

Month dummies No Yes No Yes No Yes No Yes

Session controls No No Yes Yes No No Yes YesObservations 193 193 193 193 193 193 193 193R2 0.060 0.085 0.189 0.225 0.057 0.099 0.165 0.188

This table is the analogue of Table 4 with LS majority. LS Majority is an indicator for whether the government hasmore than 50% seats in the Lok Sabha. Dependent variable in the first four columns is the number of ordinancesfrom the previous break that are not converted to an act in the session. The dependent variable in the last fourcolumns is the fraction of ordinances from the previous break that are not converted into a legislation in the sessionas described in (22). Session controls include dummies for session within Lok Sabha, the length of the session indays and the length of the break in days. Robust standard errors reported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05,∗∗∗ p < 0.01

39

Page 40: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 12: Timing of ordinances within a break

(1) (2) (3) (4) (5) (6) (7)LS Majority -0.0443∗∗∗ -0.0421∗∗∗ -0.0418∗∗∗ -0.0448∗∗∗

(0.00920) (0.00918) (0.00916) (0.00955)

Days from start of break / 100 -0.0688∗∗∗ -0.0583∗∗∗ -0.0579∗∗∗ -0.0656∗∗∗ -0.0595∗∗∗ -0.0589∗∗∗ -0.0644∗∗∗

(0.0122) (0.0119) (0.0119) (0.0135) (0.0129) (0.0129) (0.0131)

Days from start of break / 100 0.0861∗∗∗ 0.0829∗∗∗ 0.0824∗∗∗ 0.0820∗∗∗ 0.0697∗∗∗ 0.0680∗∗∗ 0.0629∗∗∗

× LS Majority (0.0159) (0.0160) (0.0160) (0.0162) (0.0147) (0.0154) (0.0169)

Constant 0.0801∗∗∗ 0.101∗∗∗ 0.0982∗∗∗ 0.110∗∗∗ 0.0680∗∗∗ 0.0871∗∗∗ 0.0398(0.00826) (0.0124) (0.0131) (0.0164) (0.0152) (0.0307) (0.0261)

Month dummies No Yes Yes Yes Yes Yes No

Day of week dummies No No Yes Yes Yes Yes Yes

Session controls No No No Yes Yes Yes No

Lok Sabha dummies No No No No Yes No No

Year dummies No No No No No Yes No

Overall session dummies No No No No No No YesObservations 15013 15013 15013 15013 15013 15013 15013R2 0.003 0.006 0.009 0.010 0.017 0.022 0.035p-value H0 : βA + λ = 0 0.089 0.035 0.035 0.239 0.457 0.531 0.895

This table is the analogue of Table 5 with LS majority. LS Majority is an indicator for whether the government hasmore than 50% seats in the Lok Sabha. Dependent variable is the number of ordinances promulgated on a day inthe break. The number of days that have elapsed since the start of the break on day s is divided by 100 to reducethe number of decimal places in the coefficient estimates. Session controls include dummies for session within LokSabha, the length of the session in days and the length of the break in days. Overall session dummies are dummiesfor each time at least one of the houses of parliament was not in session. Robust standard errors reported in theparentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

40

Page 41: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 13: Timing of ordinances within a break (Negative binomial)

(1) (2) (3) (4) (5) (6) (7)Number of OrdinancesLS Majority -0.971∗∗∗ -0.902∗∗∗ -0.914∗∗∗ -0.960∗∗∗

(0.171) (0.171) (0.170) (0.172)

Days from start of break / 100 -1.753∗∗∗ -1.503∗∗∗ -1.500∗∗∗ -1.741∗∗∗ -1.497∗∗∗ -1.516∗∗∗ -1.499∗∗∗

(0.318) (0.317) (0.317) (0.351) (0.326) (0.318) (0.294)

Days from start of break / 100 2.180∗∗∗ 2.086∗∗∗ 2.125∗∗∗ 2.102∗∗∗ 1.603∗∗∗ 1.588∗∗∗ 1.483∗∗∗

× LS Majority (0.399) (0.396) (0.397) (0.397) (0.351) (0.349) (0.357)

Month dummies No Yes Yes Yes Yes Yes No

Day of week dummies No No Yes Yes Yes Yes Yes

Session controls No No No Yes Yes Yes No

Lok Sabha dummies No No No No Yes No No

Year dummies No No No No No Yes No

Overall session dummies No No No No No No YesObservations 15013 15013 15013 15013 15013 15013 15013R2

p-value H0 : βA + λ = 0 0.076 0.033 0.022 0.263 0.718 0.793 0.942

This table is the analogue of Table 7 with LS majority. LS Majority is an indicator for whether the government hasmore than 50% seats in the Lok Sabha. Dependent variable is the number of ordinances promulgated on a day inthe break. The number of days that have elapsed since the start of the break on day s is divided by 100 to reducethe number of decimal places in the coefficient estimates. Session controls include dummies for session within LokSabha, the length of the session in days and the length of the break in days. Overall session dummies are dummiesfor each time at least one of the houses of parliament was not in session. Robust standard errors reported in theparentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

41

Page 42: Rule by ordinances: Minority governments and decree power ...mysmu.edu/faculty/madhavsa/research/ADDraft1.pdf · di erence in patterns of ordinances and parliamentary legislations

Table 14: Timing of ordinances within a break

(1) (2) (3) (4) (5) (6) (7)LS Majority -0.0236∗∗∗ -0.0221∗∗∗ -0.0219∗∗∗ -0.0233∗∗∗

(0.00522) (0.00525) (0.00524) (0.00535)

Days from start of break / 100 -0.0399∗∗∗ -0.0332∗∗∗ -0.0329∗∗∗ -0.0373∗∗∗ -0.0348∗∗∗ -0.0343∗∗∗ -0.0424∗∗∗

(0.00689) (0.00685) (0.00685) (0.00789) (0.00768) (0.00766) (0.00814)

Days from start of break / 100 0.0492∗∗∗ 0.0474∗∗∗ 0.0471∗∗∗ 0.0471∗∗∗ 0.0422∗∗∗ 0.0407∗∗∗ 0.0414∗∗∗

× LS Majority (0.00957) (0.00977) (0.00976) (0.00982) (0.00945) (0.00968) (0.0110)

Constant 0.0543∗∗∗ 0.0672∗∗∗ 0.0660∗∗∗ 0.0723∗∗∗ 0.0513∗∗∗ 0.0561∗∗ 0.0310∗

(0.00437) (0.00657) (0.00755) (0.00934) (0.0111) (0.0219) (0.0187)

Month dummies No Yes Yes Yes Yes Yes No

Day of week dummies No No Yes Yes Yes Yes Yes

Session controls No No No Yes Yes Yes No

Lok Sabha dummies No No No No Yes No No

Year dummies No No No No No Yes No

Overall session dummies No No No No No No YesObservations 15013 15013 15013 15013 15013 15013 15013R2 0.002 0.005 0.009 0.010 0.016 0.021 0.031p-value H0 : βA + λ = 0 0.161 0.049 0.049 0.245 0.385 0.461 0.900

This table is the analogue of Table 8 with LS majority. LS Majority is an indicator for whether the governmenthas more than 50% seats in the Lok Sabha. Dependent variable is an indicator for whether an ordinance waspromulgated on the day. The number of days that have elapsed since the start of the break on day s is dividedby 100 to reduce the number of decimal places in the coefficient estimates. Session controls include dummies forsession within Lok Sabha, the length of the session in days and the length of the break in days. Overall sessiondummies are dummies for each time at least one of the houses of parliament was not in session. Robust standarderrors reported in the parentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

42