Impact of Campaign Spending on Electoral Contests

There was an interesting post on Freakonomics yesterday. The author Stephen Dubner also provides a link to a paper by Steve Levitt analyzing the impact of money spent on election campaigns. By analyzing the outcomes of several elections taking into account various factors, Levitt identifies to what extent money played a part in it.The analysis is fairly simple and might be amenable to adaptation to Indian conditions albeit with some changes. Apart from expenditure of the candidates, they include four variables: the partisan tendency of a voting district, incumbency factor, partisan shock or what we call a ‘wave’ in favor of a particular party and the inherent attractiveness of a particular candidate. Barring any scandals, the attractiveness of a candidate may be assumed to stay constant over successive elections and hence, even though not directly observable, is eliminated as a confounding variable by restricting the focus to the same candidates facing off in more than one election. Of course, the existence of only two major national parties simplifies calculations considerably.Not surprisingly, he finds the effect of national political events (‘waves’) very significant. So is the incumbency factor (note that the reference is to individual incumbency, not the party) and the effect of any scandal. As for spending, he concludes: “…Once district-specific factors and the quality of the competing candidates are controlled for, the impact of campaign spending on election outcomes, regardless of incumbency status, is small but positive”. How small? An extra $100,000 in campaign spending (with the opponent’s spending held constant) garners a challenger 0.3% of the vote while adding less than 0.1% to an incumbent.He also looked at whether spending had a greater impact in highly competitive elections than in non-competitive ones. His conclusion: “When only ‘competitive’ elections were included, the point estimates for the effects of candidate spending were actually slightly lower but were not significantly different from the results obtained using the overall sample”.Finally, he offers two explanations why politicians spend so much effort on fund-raising (also true in India though carried out away from the limelight) and spending money. One is that the opportunity cost of raising funds is very low as compared to the benefits of winning the election. The second is that politicians have confused correlation with causality when considering the relationship between spending and electoral success.Are these findings likely to be true for India as well? To be sure, the way in which money is spent in US elections varies significantly from India. I am not aware of prolific television ads in India which suggests that popular choices are relatively unencumbered by political advertising to begin with. The only way a candidate can hope to gain through the mass media is by influencing coverage in a popular TV channel (like Montesinos did in Peru), something that might benefit a party as a whole and secondarily its leader (like ‘Jaya’ TV, ‘Kalaignar’ TV, etc.) but would therefore not come within the ambit of individual campaign spending.A major expense incurred by candidates is on the liquor, sarees, cricket sets and other goodies gifted to lower income voters. Indeed the only study I have come across so far is one that was conducted by the Center for Media Research earlier this year documenting the extent of such bribes. Also, prominent interest groups/organizations/religious establishments need to be appeased in part through ‘donations’. This is apart from whatever is spent to keep party workers in high spirits (literally!) and for banners, pamphlets, rickshaws and other expenditure. The degree of effectiveness of such bribes to selected groups may be different from propaganda, voter registration, rallies and other legitimate efforts. Levitt’s findings may nevertheless still hold some value for upper middle class constituencies where these practices may not be widely prevalent. Lastly, multi-cornered contests may potentially render the marginal advantage more significant. The reasons he attributes to politicians spending such a fortune on election campaigns rings true in the Indian situation as well.The differences though make the comparison a difficult one. A similar study in India might well yield interesting results. There are however likely to be methodological problems. For one thing, finding successive elections where the same candidates face off may be difficult given the extremely competitive nature of the parties’ ticket selection process. Secondly, the multi-cornered nature of many contests adds to this problem and to the overall approach as well necessitating a more complex model. Thirdly, coalition politics makes it difficult to assess the incumbency factor. These difficulties are however not insurmountable and even a limited effort might be rewarding.Update 1: Thanks to Mr.Venkatesan for reminding me of a study done by the CSDS in India. I quote from his post on this blog:
“A study conducted by the Centre for the Study of Developing Societies (CSDS) in 24 constituencies spread across 17 States and Union Territories during the 1999 Lok Sabha elections found that almost all 122 candidates monitored by it had exceeded the expenditure ceilings imposed by the law (currently it is Rs.14 lakhs in a Lok Sabha election). The study found that money mattered to gain an entry into the electoral fray, and to remain visibly in the race, but you can’t hope to buy the votes and win elections. In other words, it is not true that the more you spend, the more likely you are to succeed.”Update 2: Mr.Venkatesan kindly provided me with a copy of the CSDS study. On the relevant questions concerning electoral spending, the report compiled a list of the percentages of votes earned by the winners and their main challengers as well as the amounts spent by these two groups. The groups were themselves divided into three categories depending on the margin of victory (under 5%, 5-10% and over 10%). Several inferences were drawn from the numbers:1. The winner did not always spend more than the runner-up. In some of the closely contested constituencies, the runner-ups spent more.
2. Those constituencies with the narrowest margin of victory did not record the highest spending. Rather, moderately competitive constituencies (margin of victory 5-10%) recorded the highest expenditures. Thus, spending did not correlate well with the degree of competitiveness (as judged from the outcome).
3. Polling and counting day expenditures are the highest as a fraction of the total expenditure on major items in the most closely contested seats. The inference is that by the polling day, candidates already have a keen sense of their chances and make a strategic choice to spend more if they believe the contest is close and the odds of victory are good.The conclusion from 1 and 2 is that ‘higher expenditure does not necessarily produce victory’, i.e. there are important factors besides money which make a real difference to the outcome. Note that this does not indicate whether money does or does not make a difference – that is not something that a study of this sort will tell us. It is quite possible that in terms of votes, the winner and the runner-up got different returns on their investment similar to what was found in Levitt’s paper. The data cannot however verify this nor can it be compared with the findings from that study as the terms incumbent and challenger used in that study indicate the candidates’ status before the election which is different from the labels winner and runner-up used here that are based on the outcome.

Join the discussion

This site uses Akismet to reduce spam. Learn how your comment data is processed.

2 comments
  • That’s quite reassuring. In the recent Delhi University Student’s Union(DUSU) elections, which I have always considered as a micro picture of how national elections are contested/won, the money spent seemed pretty much directly proportionate to the electoral victory. The oprative word here is ‘seemed’, it would be interesting to look into this.