2012 Uttar Pradesh Elections

The following entry is by Devesh Tiwari, a Ph.D. candidate in political science at UCSD.

2012 Uttar Pradesh Elections: Sweeping mandate, humiliating defeat or none of the above?

Over a three week period, approximately 60 percent of Uttar Pradesh’s 126 million eligible voters participated in state level elections that took place in seven stages, making this election larger (and logistically more complex) than the national elections of many countries. By way of background, Indian states are parliamentary democracies where the majority of the legislature must support the executive. On March 6th, the results were announced and the Samajwadi Party (SP) easily gained a majority of seats, placing the incumbent Bahujan Samaj Party (BSP) out of power . The results were striking. The BSP, which entered the race holding 206 seats (51 percent of the 403 seat assembly) lost nearly 60 percent of its seats, winning only 80 seats (20 percent of seats). The SP nearly doubled the number of seats it held, increasing it from 97 (24 percent of seats) to 224 seats (56 percent of seats). Such a drastic swing could characterize this election as giving the SP a massive mandate, and could be equally characterized as a humiliating repudiation of the BSP.

A closer look at the election results reveals that both characterizations are overblown. Uttar Pradesh uses a “First Past the Post” electoral rule whereby the party with the most votes in a district wins the seat. This system is known to produce disproportional results where a party may win a higher proportion of seats than votes. For example, in the 2005 Parliamentary elections in the United Kingdom, the Labour Party won 55 percent of districts while only obtaining 35 percent of the vote. Thus while disproportionality is not something new, its magnitude in Uttar Pradesh is impressive.

The SP won 56 percent of districts by only winning 29 percent of the state vote. The BSP’s vote share of 26 percent translated to a seat share of only 20 percent. In other words, the 3 percentage point difference in vote shares resulted in a 36 percentage point difference in seat share.

While the BSP was on the short end of the disproportionally stick in 2012, they benefited from it in 2007. In that election, the BSP won 51 percent of seats by only winning 30 percent of the vote. The SP, on the other hand, won 24 percent of the seats while winning 26 percent of votes state wide. Taken together, the SP more than doubled the number of seats between elections by increasing their vote share by a mere 3 percentage points.

So what accounts for these startling results? Why is disproportionality so high in Uttar Pradesh? The answer is that political competition in Uttar Pradesh is highly fragmented. In 2012, each district had, on average, 13 parties contesting the election; in 2007 the average was about 9. Partisan fragmentation in Uttar Pradesh is further evidenced by the fact that in 2012, the average district vote share of winning candidates was only 35 percent (and was 36 percent in 2007). Thus the combination of party system fragmentation, and the disproportionality caused by the FPTP electoral rule, exacerbates vote share volatility to even higher levels of seat share volatility, producing high levels of political uncertainty.

Such a political environment changes the incentives of both legislators and political parties on two fronts. First, it creates incentives for parties to place a higher premium on fielding candidates that can win elections, even at the expense of party reputation and governance. This may partially explain why parties in Uttar Pradesh (and India as a whole) would be willing to field candidates with ties to criminality; winning seats is more important than long term party reputations and governance. Second, it shortens the time horizon of parties. Since a governing party would have little faith that their tenure in office would last long, they have no incentive to invest in programmatic policies whose political benefits would be realized in the future. Instead, parties would focus on short term political exchanges whereby they trade government benefits for political support.

Uttar Pradesh’s political environment undermines the production of programmatic policies and thus reinforces the political logic of clientelism. These are precisely the incentives that Uttar Pradesh does not need. As one of the poorest, and most corrupt, state in India, Uttar Pradesh would greatly benefit from reforms that reduce corruption, increase bureaucratic quality and increase the investments of public goods and services. Thus while Uttar Pradesh is a democracy in the sense that there is alternation of power, it has not produced a system whereby parties act for the benefit of the people as a whole.

10 thoughts on “2012 Uttar Pradesh Elections

  1. There must be few rivals to Uttar Pradesh for most disproportionality ever: leading party’s advantage ratio (A) of 1.93 (%seats/%votes).

    In the dataset I have on FPTP elections, which consists of 210 elections (none from Indian states)–the only case with A>1.7 for a leading party under 40% of the vote is British Columbia 1972 (NDP 69% of seats on 39.4% of votes, A=1.75).

    The one that Devesh mentions, UK 2005, is a “mere” A=1.56. The only other cases over 1.5 (again, where the leading party is under 40%) are BC in 1960 and Ontario in 1990.

  2. Thanks for highlighting an election involving a place with 126 million people, that otherwise gets no coverage at least in the U.S.

    That said, what were the statewide popular vote percentages of the parties? I was going to criticize the article for not giving this information, but I just went to the Indian Election Commission website (http://eciresults.nic.in/statewiseS24.htm) and it doesn’t appear to have statewide vote totals. That probably explains the circuitous “the Samajwadi Party won 206 seats, but they won most of them by narrow margins and with less than a third of the vote” writing of the first half of the article.

    Though I do think the author would have been better advised to take some time, enter the vote totals from the election site onto a spreadsheet, and then write something up more informative (eg that had the statewide vote totals) a week later, assuming someone at the Indian Election Commission doesn’t do this. Was the piece written on some sort of deadline.

    I also have a more substantive criticism. The system of government in Uttar Pradesh appears to be identical to that used, for example, in Alberta. None of the criticisms of single member plurality parliamentary government in the article applies to Alberta, which has been governed by the same political party for decades, and to my knowledge convicted criminals run for office. The convicted criminal thing appears to be probably unique to India and probably has nothing to do with the electoral system. In the U.S., single member plurality is usually criticized on the grounds that it restricts the number of parties effectively competing in elections, not that it encourages fragmentation and rapid turnover!

    In fact the federal Indian Parliament, though elected by single member plurality, has behaved like a chamber selected by proportional representation for some time now. Though the voting system helps shapes the party system, there are other factors in a political system that may be more important. And a bad party system or political culture will make any electoral system in turn look bad.

  3. Incidentally, I’m not a big fan of single member plurality, and the scheme has a specific problem in regards to Indian elections, in that the size of the populations translates to oversized constituencies even at the state level.

    For me, the big advantage of single member plurality is the “single member” feature, that each voter has a locally elected go-to member that they know will either reliably back or oppose the government, according to how his neighbors voted in the last election, and will raise local issues. If the constituencies or electoral districts are too populated, the connection is lost and you might as well go to multi-member districts and PR. I’m not sure what the upper threshold is, but India has almost certainly passed it, as would China if it were more democratic, and this may also be true at the federal level for the U.S.

    Likewise, for small countries, multimember districts will often be so small in population and area that breaking them up into single member districts wouldn’t significantly improve voters’ connections with their representatives.

  4. I tend to agree with Ed about SMDs. The question is at what level of population are they appropriate. How many people can one person reasonably represent? Also, how does MMP fit into all of this?

  5. Ed,
    Thanks for the comments. I did include vote share information in the post, but you are right. A table would have been much clearer:

    ————————————–
    Party | Seat Share Vote Share
    ———-+—————————
    BJP | 0.12 0.15
    BSP| 0.20 0.26
    INC | 0.07 0.12
    Other| 0.03 0.16
    RLD| 0.02 0.02
    SP| 0.55 0.29
    ————————————–

    Note that the INC and RLD had a pre-electoral coalition, their coalitional seat share is 9 percent while its coalitional vote share is 14 percent.

    I do not believe that SMD alone is responsible for the fragmentation found in Uttar Pradesh, but given that the fragmentation exists, SMD is responsible for these extremely disproportional results and high “seat” volatility.

    I should also mention, that there is wide variation in party systems across India, and over time.

    Thank you for your comments though, I appreciate it.

  6. Much of this (except, strictly speaking, disproportionality) could easily be solved with the application of PNG’s solution: the alternative vote.

  7. Ed, you must not have read very closely. The statewide vote totals of the SP and BSP for the two most recent elections are in the original post! (And now I have cleared from moderation Devesh’s earlier posting of further data.)

    As for the PNG solution, in practice, it has not made much difference. It is not full AV, and districts are still being won with far less than half the votes. Besides it is not as if AV is less disproportional with a large number of parties than FTP.

  8. Full AV would indeed be necessary. The first advantage would be majorities for seats. The second would be a 2-party preferred that (usually) justifies an overall majority better than say, a 30% of the vote as achieved under FPTP in some of the above examples.

  9. India can’t use full AV as most of the population is illiterate or has low numeracy. FPTP is the easiest system in the world when it comes to voting and understanding.

    Has India (Federal Level or any state) ever had a reverse plurality election? Where the 2nd place party formed a majority government, but the 1st place party was in opposition.

    Full AV in India (Uttar Pradesh) produce perverse election results as the state has a multi-party system? I heard that Full AV in a multi-party system produces perverse results.

    India (Uttar Pradesh) would be better off with some form of PR, Open List PR with moderate sized multi-member districts with an at large adjustment layer to ensure proportionality across the state or a one vote MMP system.

    With PR, there is more compromise, and stability of policy, and more long term political planning that India needs. India may have alternation of parties in office, but the way FPTP works is that it promotes patronage spending for the party winners seats.

    If India ever used some form of MMP, a two vote system would be abused as it was in Lesotho and Italy, a one vote system would be best.

    Papua New Guinea uses Limited Preferential Voting for that same very reasons low literacy and numeracy. The question about Papua New Guinea is that why does it use Limited Preferential Voting when it would be better off using some form of Proportional Representation; Why not STV?

  10. I did seem to have reading comprehension issues with that one. I do remember reading the line “The SP won 56 percent of districts by only winning 29 percent of the state vote”, but my brain registered that as “The SP got only 29 percent of the votes in 56 percent of the seats it won”. My brain then decided “so what” and skimmed through the rest of the paragraph. My wife was urging me to go to bed when I read that.

    Still, it would be good if listing the popular vote percentages each party won was a standard feature of writing about elections, eg:

    SP 29%
    BSP 26%
    BJP 15%
    INC/ RLD 14%
    Other 16%

    SP 206 seats (+ 127)
    BSP 80 seats (- 126)
    BJP 47 (-4)
    INC/ RLD 37 (+5)
    Others (19) (-2)

    Some of the above information has been taken from the Wikipedia article on the election.

    A few things jump out at me from these totals. First, the “other” vote seems a bit high and that probably accounts for alot of distortion in the seats to vote ratio. It also means that the four main parties are fighting over 84% of the electorate.

    Second, there are four major parties, presumably all campaigning across the state, which is unusual in any party system. The only parallel I can think of for single member plurality off hand was the weak one of the Liberal-PC-NDP-Social Credit contests in Canada in the early 1960s.

    Check out the last two UK elections:

    2010 Conservatives 36% Labour 29% Lib Dems 23%

    2005 Conservatives 32% Labour 35% Lib Dems 22%

    Now add a fourth party or coalition taking 14% of the vote to these totals and the major party vote percentages go down to Uttar Pradesh levels. And Labour won a majority of the seats in 2005.

    Now in Uttar Pradesh in 2012, the SP got a 3% margin over its closest rival, exactly the margin Labour had in 2005 in the UK. The closest third party was 14% behind, again not unusual for third parties with substantial support like the Lib Dems.

    Parties win majorities in FPTP style legislatures all the time with 3% margins over their closest rivals. The statewide vote percentages for all the major parties are low due to the unusual factors of a fourth major party or coalition and the high “other” vote.

    I gather there are no distinct regions within the state, so if a party gets a statewide lead it will be repeated across the state (no high deviation in votes)? This would also turn a small popular vote margin into a big lead in seats.

    This shows a major weakness of single member plurality, in that parties can turn popular vote percentages registering in the 30s (granted 29% is unusually low) into legislative majorities by gaining small margins over their closest rivals. I think something like open list proportional representation would likely be appropriate to India, but more because the large population of the constituencies negate much of the local representation advantage of single member districts.

    But note that in these cases, though the aftermath of individual anomalous elections can be brutal on government policies and the party system, the fix is often administered by the voters themselves, who start to vote more tactically.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s