Examining a major road-building program in India, researchers at Princeton University and the Paris School of Economics used an innovative technique to show that political corruption increased the chances that roads meant to connect isolated areas to the rest of the country would never be built, even though the government had paid for them.
The study, published in the Journal of Development Economics, found that almost 500 all-weather roads listed in the road program's monitoring data as having been completed and paid for were never built.
The researchers connected these "missing roads" to political corruption - specifically, to local politicians steering road contracts to favoured businesses in their own social networks.
"Our results indicate that corruption in this program directly harmed the 857,000 villagers whom the missing roads were meant to serve," said study lead author Jacob N. Shapiro, professor of politics and public affairs at Princeton's Woodrow Wilson School of Public and International Affairs.
Pradhan Mantri Gram Sadak Yojana (PMGSY) is a vast public works program begun in 2000, when the Indian government estimated that more than 300,000 villages across the country weren't connected to the outside world by all-weather roads.
The proposed new roads built under this program were intended to provide village residents with previously unavailable economic opportunities, and facilitate their access to government services such as education and health care. But if contractors took the money but failed to build the roads, those benefits would never materialise.
The detrimental effects of political corruption aren't easy to see or measure - by its very nature, corruption is usually hidden from view. To seek evidence of corruption, Shapiro and his colleagues looked at thousands of elections for local representatives to the state-level legislature, who are known as members of the legislative assembly, or MLAs.
After such selections, they analysed the degree to which road-building contracts shifted to contractors who shared the new MLA's surname. Because Indian surnames are closely linked to caste, religion and geographic provenance, surnames serve as a proxy for the politicians' social networks.
The researchers found that after a close election, the share of contractors whose surname matched that of the political winner rose more than 75 per cent, from about 4 percent to about 7 per cent. Next, examining data for almost 90,000 roads contracted by PMGSY, Shapiro and his colleagues sought roads that had been paid for but apparently didn't exist.
"We found that road contracts allocated to politically connected contractors were significantly more likely never to be constructed," Shapiro said. "While the program as a whole appears to have been extremely efficient, political influence clearly led to lower-quality execution."
The results were surprising because PMGSY had been designed with strong controls to prevent political corruption. Contractors were to submit separate technical and monetary bids; only if contractors could meet the technical specifications were their monetary bids opened, and the lowest bidder was to receive the contract.
Moreover, local politicians were given no role either in selecting contractors or in choosing where roads were to be built - decisions that were made by bureaucrats at the regional level.
How, then, were local politicians able to game the system and steer contracts to their preferred businesses? One possibility is that the winning politicians' social networks also include connections within the regional bureaucracy that awards the road contracts, and that they ask those connections to put a thumb on the scale to favour one contractor over another.
The researchers found evidence of this - when the highest-ranking district official overseeing PMGSY also shared the winning politician's surname, favoritism in the awarding of contracts was more likely. On the other hand, when bureaucrats were up for promotion and thus facing heightened scrutiny, corruption was less likely.
These patterns were exactly what the researchers expected to see if MLAs were influencing contracts through informal bureaucratic connections.
Strikingly, Shapiro and his colleagues suggest that giving local politicians no formal role in PMGSY, which was intended as a measure to fight corruption, may be having the opposite effect.
The researchers hypothesise that voters who believe members of the legislative assembly have no role in the road-building process won't punish these local politicians at the polls for missing roads, delays and cost overruns. "If voters held their politicians responsible for the services delivered under PMGSY," Shapiro said, "the MLAs would have an incentive to limit corruption.