Representative Democracy?

Some observations on the recently concluded parliamentary elections in India (for a closer look at the electoral data, see here and here): 

1. The BJP polled only 31% votes nationwide but won 52% seats in the parliament (282/543). Even more strikingly, the BJP polled only 42% votes in U.P. but won 89% seats (71/80).
2. BSP polled 4.2% votes nationwide (3rd highest after BJP and Congress) but won 0 seats. In U.P., the BSP polled 20% votes but won 0 seats.
3. Both AIADMK in Tamil Nadu and Trinamool Congress in W. Bengal polled fewer votes than BSP (and only a bit more than BSP in U.P.), but won 37 and 34 seats, respectively (vs. 0 for BSP).
4. SP polled 18 M votes in U.P. but won only 5 seats, whereas the Shiv Sena polled only 10 M votes in Maharashtra and won 18 seats.
5. About half of all Indians voted for regional parties, not national parties. 69% of the people voted for a party other than the BJP. 

These examples show how India’s parliamentary democracy, owing to its first-past-the-post voting system, fails to represent the political preferences of its citizens. This is in addition to the fact that elected politicians in India also do not represent its citizens sociologically (they’re far more likely to be upper caste, wealthier, Hindu, male, urban, etc.). According to the economist Jeffrey Sachs, "The first-past-the-post election tends to produce a small number of major parties, perhaps just two, a principle known in political science as Duverger’s Law. Smaller parties are trampled in first-past-the-post elections." Alternatives that reduce the drawbacks of FPTP exist. Isn’t it time to rekindle the debate on making Indian democracy more representative, both at the political and sociological levels?

Comments

[anonymous_indian]

I hope you agree that this is not a point that's relevant *only* when the BJP wins a majority, of course.

Besides, the BJP contested only 427 seats, and the Congress contested 467. I think that would skew the numbers somewhat. You'd need to count votes won as a percentage of votes contested for.

I'm not sure what the relative ratio of votes contested for by BJP vs Cong-I is in this case, but assuming electorate is uniformly divided amongst constituencies, BJP got 31.1/(100/543*427) = 39.55 % of the votes contested., and INC got 19.1/(100/543*467) = 22.21 % of the votes contested.

Many other complications in your analysis : BSP (for example) does not contest all or most seats in the country. The turn-out is not uniform across constituencies : the strong regional party states all have pretty high turnout (~75 % for Bengal, TN etc) and the Hindi belt has relatively low turnout (~55 % or so).

[Namit]

This is a well-known problem with democracies that use FPTP voting and I've written about it before in other contexts. By raising this again, I'm pointing out an old problem that can be mitigated and also noting that, in terms of votes, this wasn’t as much of a Modi sweep/landslide as the # of seats suggests. My goal is to help chill the triumphalism of BJP fans and amplify a corrective viewpoint. No matter how you slice it, fact is that 69% of the electorate voted for a party other than the BJP, yet the BJP won a clear majority of the seats in the Lok Sabha. I think this is a problem worth solving for any polity that claims to be a representative democracy.

[Rajesh]

Are you a political scientist? Do you have an alternate solution?

[anonymous_indian]

More comments coming, but a particular point can be noted in a hurry -

"The only way to get a minimum 50 percent mandate is to have a French style system of having one or more eliminating rounds before the final vote. This would knock out the parties with the lowest shares in progressive stages – leaving only the two finalists seeking 50-percent plus and victory."

This, too, forces people to make second choices, or worse.

"The other way is to have a two-party system, or a German-style proportional representation system where parties getting less than 5 percent are knocked out. In this election, a German-style system of proportional representation would have left only the BJP and Congress as worthy of parliamentary representation. No other party touched 5 percent at the national level."

http://www.firstpost.com/politics/why-gopal-gandhis-open-letter-to-modi-is-wrong-headed-and-odious-1531867.html

[Namit]

That article you quoted from was indeed written in a hurry; the author comes across as too busy defending Modi than arguing rationally. Runoff voting works best for electing an individual from a constituency. What's wrong with a second choice (if logistics and cost allow that; "instant runoff" is one option where people vote once but rank candidates)? We do that all the time in private life. Runoff voting penalizes people less if they don't vote strategically. A real world example where runoff voting would have changed the outcome is in Florida, 2000, if more Ralph Nader votes had gone to Gore in the second round. On the other hand, Clinton might not have won in 1992 with runoff voting if more Perot votes had gone to Bush in the second round. Look for other examples where it is practiced, esp. in Australia, Ireland, etc.

It’s true that outcomes often don't change going from FPTP to runoff voting, but it's also true that the outcomes of runoff voting tend to be more representative of people’s political leanings. This is most evident when a single party dominates one part of a political/ideological spectrum but many parties cluster in a different part of that spectrum. The latter undercut each others' votes. So the second voting round allows consolidation of the votes that are clustered closer.

For example, in Allahabad 2014, the BJP got 35%, SP 28%, BSP 18%, and Congress 11%. Under FPTP, BJP won. But under runoff voting, the BSP and Congress votes would almost certainly have meant a defeat for BJP and victory for SP (BSP came second in 33/80 U.P. constituencies, so they too might have won many seats by the same logic, instead of 0). Also, because Dalit and Muslim voters matter so greatly in the second round, none of the two leading parties can ignore them as much as they can in FPTP (BJP would have to care more about Muslim votes, so the runoff system discourages majoritarian attitudes and encourages bargaining and policy concessions). This is not to say that runoff voting is perfect. It has pros and cons, but it more often produces outcomes that better reflect the leanings of the electorate. Had the BJP won 282 seats using this method, that would have been a far greater achievement and a clearer mandate for its agenda. And yes, this applies also to non-BJP governments in the past.

A basic primer on runoff voting: http://en.wikipedia.org/wiki/Two-round_system

"Runoff voting encourages candidates to appeal to a broad cross-section of voters. This is because, in order to win an absolute majority in the second round, it is necessary for a candidate to win the support of voters whose favourite candidate has been eliminated. Under runoff voting, between rounds of voting eliminated candidates, and the factions who previously supported them, often issue recommendations to their supporters as to who to vote for in the second round of the contest. This means that eliminated candidates are still able to influence the result of the election. This influence leads to political bargaining between the two remaining candidates and the parties and candidates who have been eliminated, sometimes resulting in the two successful candidates making policy concessions to the less successful ones. Because it encourages concilliation and negotiation in these ways runoff voting is advocated, in various forms, by some supporters of deliberative democracy.”

Finally, I don’t enjoy conversing with “anonymous” people, so I urge you to identify yourself in subsequent comments.

[suresh]

The one advantage of the FPTP is its simplicity - it is easy to understand. A more complex system may produce "better" results (using what metric?) but will inevitably be more costly to run. There will be significant transition costs of moving to the new system including the cost of educating the electorate. While the costs are fairly clear, the benefits are much more nebuluous.

As you note, runoff voting is by no means the only alternative to FPTP - you have things like Borda count, approval voting, proportional voting, etc. A more recent innovation in this regard can be found in the book "Majority Judgement" by Y. Balinksi and R. Laraki, MIT Press, 2010. You might also want to see Bhaskar Dutta's article in Economic and Political Weekly, "The Fragmented Lok Sabha: A case for electoral engineering" in the April 15, 2009 issue.

Changing electoral systems is not easy in the best of times in any country. The British electorate threw out the Liberal Democrat proposal for changing the FPTP system. In our country, changing electoral system requires some sort of political consensus which, in a fractured polity like ours, is extremely difficult. It took considerable time to move even from paper ballots to EVMs. The technology was already there by the 1980s but it took a good few years for a political consensus to emerge. Perhaps your suggestion is in the nature of a start. TIf so, be prepared for a long wait.

Finally, your counterfactual analysis like all such analysis has to be taken cautiously. What you are doing is keeping the current party system and voting patterns but changing the voting system. This is fine as a start, But it is worth remembering that if you change the system, then the incentives for political mobilisation also change.

A reason for partitition was the British allowing political mobilisation along religious lines. It was not just that parties could restrict memberships to particular groups: you also had "Hindu" and "Muslim" seats. Did this make a difference? Well, Rajmohan Gandhi in his biography of Patel makes the claim that a reason for Patel and Nehru accepting partition was precisely their bitter experience of working with the Muslim League in the coalition government before independence. For instance, the League members did not accept Nehru as the head and met separately. I am sure the League has their own stories regarding Nehru and Patel...the point is that this type of conlict was exacerbated by the fact that mobilisation along religious lines was openly possible. (And not to forget, sports was not spared, in cricket we had the now defunct "Pentangular" trophy with Hindus, Muslims, Parsis, Christians and weirdly, "Europeans.")

Not surprisingly, independent India has a law whereby no political party can say that membership is restricted to particular groups. Okay, so parties can always get around this - if there is a non-Muslim member of the Indian Union Muslim League, the Majlis-e-ittehadul Muslimeen or a non-Hindu member of the All India Hindu Mahasabha or a non-Sikh member of the Akali Dal, I have yet to meet them. But still the fact that political mobilisation cannot be done openly on religious lines - you have no "Muslim" or "Hindu" seats - makes some difference and illustrates my point that counterfactual analysis has to be interpreted cautiously.

[Namit]

Suresh, it’s not clear to me that the cost will be much higher if, say, an instant runoff method is chosen. Voters still go into the booth once, but instead of choosing one candidate, they indicate their top two (or three) choices. The runoff is then performed electronically at the time of counting. Educating the voters to indicate their top 2-3 choices is not that hard I think. In any case, this can start with certain states and then get rolled out in others. The benefits seem clear enough to me, and I’ve indicated many in the previous comment. If not the UK, many other countries have implemented it. Of course, all costs and unintended consequences should be studied before changing any system. That vested interests will try to prevent change has fortunately not always stopped people from trying and sometimes succeeding. :-)

[suresh]

I am not an expert in this field (voting theory) so I won't get into a discussion. If you favour runoff voting, fine. I'll confine myself to saying that you can find any number of paradoxes with any voting system of your choice. For me, unless you are clear about the metric (proportion of "bad" outcomes or something like that), it is not clear that one system dominates another. Merely demonstrating that in a few situations, runoff voting does better than FPTP is not enough.

[Shanth]

Suresh, it is a bit disingenuous to say " you can find any number of paradoxes with any voting system of your choice", because instant-runoff and other forms of proportional voting are clearly more resistant to strategic manipulation and do better represent the popular mandate than first past the post systems. However, the degree of investment that the entrenched politicians and political strategists have in maintaining the current system make it very hard to introduce electoral reform, as such attempts in the US and Britain also show.

[suresh]

Shanth:

are clearly more resistant...

Hey, look, the Gibbard-Satterthwaite theorem asserts that (under some conditions) essentially the only rule which cannot be "manipulated" is the dictatorial rule. So I am not looking for a "perfect" rule. But unless you tell me the metric you use, I am sorry I don't find assertions like "clearly more resistant to strategic manipulation" very persuasive. It may be very clear to you but not to me.

If you want, matters are even more complex. Elections are also affected by who enters the elections. Dutta, Jackson and Le Breton show that essentially every non-dictatorial rule is affected by the incentives of those who choose to not enter the elction. As they note, Ross Perot's decision to enter the 1992 US Presidential elections probably cost Bush (senior) the election. Even in the current election, if some parties had joined hands and fielded a joint candidate, the BJP might have lost. In almost every election since independence, this has been a theoretical possibility.

Election theory is a fascinating area and I am aware of only a very small set of papers and results. If you know the literature and know that "proportional voting" is best, fine, and good for you. I don't find your arguments persuasive but I'll leave matters at that.

[Ashutosh]

It is really important to rekindle the debate. Here is another interesting article that I thought I should share. http://www.vikalp.ind.in/2014/06/questioning-first-past-post-system.html