Wednesday, January 15, 2014

Per Capita GDP growth and Vote Share


    In 2004 the BJP lost the general election, despite their belief that the Indian economy's performance was outstanding and the voters would reward it for this performance.   Many analysts and most opposition parties accepted the contention that the economy was performing very well, but the fruits of this growth had not reached the “Aam Admi / Aurat” who voted against the BJP.  As a contrarian I argued that the economic performance was in fact worse than during the previous regime and the voters had in fact punished the government for poor economic  outcomes (Virmani(2004))[i].
     Last week the Prime Minister gave a press conference to assert that the economic performance of his government had been outstanding and the Congress party was therefore likely to win the election again. Unfortunately he is making a mistake very similar to the one made by the BJP.  The actual performance that matters to the “Aam Admi / Aurat” voter, is much worse than the performance benchmark (as per the model developed by the author). The outcome is therefore likely to be equally disappointing for the Congress party.

The Model

    Under normal circumstances, elections are held every five years. At the end of the five years, the simplest way for the “Aam Admi / Aurat” to judge a Government’s performance is to see how (s)he and his(her) family, friends, neighbors and acquaintances have done during those five years and compare this with what happened during the previous five years (earlier government).  If the performance is better/worse than in the benchmark years there would be an increased/reduced likelihood (probability) of him(her) voting for the party in power.
   Reduced to the barest essentials, there are three economic indicators that in our view matter to the independent/swing/marginal voter.  These are Economic Growth, Public Goods and Services (popularly referred to as “Bijli, Pani, Sadak”) and transfers/subsidies net of taxes (“populist giveaways”).  The growth of per capita income determines not only the change in living standards of the average voter but also the employment opportunities (job or self-employment) created and available to the average person.  Public goods include a range of local & municipal services such as roads, drinking water, sewage, sanitation, drainage, public health and basic education. An improvement (worsening) in the quality and/or volume of such services would tend to have a positive (negative) impact on the voter and thus tend to increase (reduce) the ruling party’s vote share.  A change in net transfers (subsidy - taxes) would be similarly associated with a change in vote shares. It must be kept in mind that it is not the paper laws, verbal announcements or budgetary allocations (expenditures or subsidies) that matter to the Aam Aurat, but the actual delivery of services at the ground levels and the delivery experience. The quality of governance and the degree of corruption & harassment are a very important element of the Aam Aurat’s judgment about improvement or deterioration in delivery . 
   In this note we focus on the growth of per capita GDP as an indicator of the improvement or worsening of economic conditions for the average of marginal voter: Thus, if the rate of growth of per capita income has increased (decreased) the vote share of the ruling party is likely to increase (decline).

Past Elections

       Given this hypothesis, let us look briefly at the past two elections. The NDA government was in power from 1999-2000 to 2004-5 when the general elections were held. During this tenure the Per Capita GDP at factor cost grew at an average of 4.1% per year compared to an average of 4.5% per year in the previous five years. Thus average per capita growth during the NDA government's tenure was 0.4 per cent points per year (9%) lower than during the previous five years making a reduction in BJP-NDAs vote share likely.  It appears that the “India Shinning” campaign may have jarred with the reality as seen by the more aware/better informed urban voters and thus exaggerated the relatively small decline in per capita growth. Another potential factor may have been that new highway programs and Gram Sadak Yojna (very important "public goods") had not yet produced results that the Aam Admi could feel in the small towns and villages of India. The result was a reduction in the BJP’s vote share by 6.7 per cent points from 23.8% to 22.2%
      The Cong party consequently came into power in 2004-5 and ran the Government as part of the UPA till 2008-9.  During this period the growth of per capita GDP accelerated by an incredible 2.7 per cent points (69%) to 6.8 per cent per year.  As predicted by our model this 69% acceleration in the rate of growth of per capita income, resulted in an increase in the vote share of the Congress party, leading to a sweeping victory in the 2009 general election. This growth acceleration had such a large impact on the "Aam Aadmi/ Aurat" that the effect of other infrastructure & social programs was completely overshadowed.  The vote share increased by only 7.9% to 28.6% (from 26.5% in 2004).
       As is apparent from the varying results of different pols, the translation of votes shares into seat shares is a very difficult exercise. It depends on the distribution of victory margins, the number of serious parties (two or more) in the States as well as on alliances.  Consequently in 2009, the increase in the proportion of seats won by the Congress was much higher (49%) than the increase in the vote share..   

2014 General Election

   What does the model say about the forthcoming general election?  During the tenure of the current government (UPA2) from 2009-10 to 2013-14, per capita GDP growth is likely to average about 5.4 per cent per annum.  This is a decline of 1.4 per cent points or 21% relative to the growth rate during the previous five years. Thus (ipso facto) the fall in Cong vote share in the 2014 general election is likely to be greater than that of the BJP in the 2004-5 election but less than the gains of the Cong in the 2004 election. This suggests a decline in vote share between 6.5% and 7.5%.
   Two questions have been raised with respect to this model's predictions for the next election. One is whether the fact that part of the decline in growth is due to global factors matter?  In my view, what matters to the "Aam Admi/Aurat" is what they feel and see on the ground, not the data and analysis that financially sophisticated people spend so much time reading and discussing. The Second question is how important is economic performance during the most recent period, say the last 6 to 18 months, to the voters judgement?  We know that in a sophisticated economy and highly educated polity like the USA, the recent period is overwhelmingly important.  It is therefore likely that economic performance in the last few years (2-3) has a greater weight in the decisions of  highly educated, well informed, urban Indian voter.  However, what happens during the past 4-5 years is still much more relevant to the "Aam Aadmi."


     The slowdown in per capita income growth during 2009-10 to 2013-14, relative to the average per capita growth during 2004-5 to 2008-9, will have a substantial negative effect on the Congress party’s vote share in the general election.  In addition, the issue of Central government corruption and governance will also affect voter behavior as it is directly related to the quality and volume of public goods and services supplied by the government, as well as to the effectiveness & efficiency of welfare transfers.  Both these will have a further adverse effect on Congress party’s vote share, thus offsetting any positive effect of programs like NAREGA.  The "rights laws" and increased petroleum, fertilizer and electricity subsidies will have little or no effect on vote shares in the general election.

A version of this article appeared on the editorial page of The Indian Express, on January 16, 2014, under the Banner, "The GDP Poll.";postID=7024682515623021753 

[i] Arvind Virmani, “Economic Growth, Governance And Voting Behaviour: An Application to Indian Elections,” Working Paper No. 138, ICRIER, July 2004. .

No comments: