My new paper (open access; meaning free to read) is out in the new edition of Public Choice (published online first in February this year). This was my second PhD paper at Oxford, and one I am particularly fond of given the importance of the topic and one of the cornerstone arguments of my upcoming book Elite Networks: The Political Economy of Inequality (more on that below).
What’s the main finding?
In short, I looked at the effect of political connections on the
allocation of TARP funds to US banks, and found that TARP recipients that lobbied the government,
donated to campaigns, or whose top execs had direct connections to politics
received better bailout deals.
Let’s unpack this.
In 2008, as the crisis unfolded in the US, the banking industry
elevated its lobbying and campaign spending activities. You might remember the
panic days in Sep & Oct ’08 where it seemed like the financial world is
collapsing. Getting bailed out was a priority for many banks, especially the biggest ones which had regular daily meetings with top government officials (Treasury Secretary Paulson and Geithner of the New York Fed).
By the end of 2008 and during 2009, the banks that spent the
most on political campaigns and lobbying, on average, received the largest
bailout packages (relative to size of assets). Is that relationship causal?
 |
|
|
 |
| Total number of financial services firms lobbying, peaking in 2009. The straight vertical line denotes the start of the bailout allocation process (October 28, 2008) |
Based on the graphs above, no. The problem is that
riskier banks lobbied and donated more because they are more prone to risk, implying
they also will have a greater chance of getting bailout funds. Risk exposure, not political connections, could thus be the main explanation for bailout
allocations.
To prove causality we must address this problem by using some
randomization in outcomes between treated and controlled units. Recall my previous blog on the econ Nobel for causal inference to understand why.
It just so happens that during the 2008 panic, the US was
in the midst of an election campaign for the President and the new Congress. That context allows me to exploit very close electoral races (decided within a 1% margin) between
Congressmen as sources of as-good-as-random assignment of politicians connected
to TARP recipients.
Why close races? Because in such
close races a random element like luck (e.g. rain on election day) is most likely to have
tipped the outcome, meaning that bare winners and losers—and firms attached to
them—are supposed to be statistically interchangeable.
I apply two methodological approaches based on natural
experiment design: regression discontinuity design and instrumental variables. Basically, I compare connected firms whose politicians won
narrow elections to connected firms whose politicians lost narrow elections.
The RD estimation finds that a connected politician’s
close electoral victory increased the bailout allocation for that politician’s
connected firm by around 13%, on average.
 |
| The effect of a connected political victory on the share of bailouts to total assets (y-axis) for close races |
The IV approach finds that firms with more connected
politicians who won closely contested elections received more TARP funds (the
effect is a 19% increase in TARP funds for a one-standard-deviation increase, which is about 12
more connected politicians winning close elections). It is easy to see how the effect grows stronger for
firms which had more connected politicians, meaning that those at the upper
extremes of the distribution (the biggest 8 banks e.g.) will yield higher
benefits from their political connections.
These are two different effects, but they both suggest a
similar conclusion: being politically connected clearly made a difference in
the allocation of TARP funds among those financial institutions who received
them. The results do not imply that some banks were deliberately
favored over others, just that favored banks benefited because of their
proximity to the right people in power. If being politically connected matters
in general, in times of crisis it matters even more.
Hopefully the study will provide a blueprint for further
research on political influence and the distribution of government funding,
particularly during times of high-stress events or serious economic hardships (e.g.
public procurement & bailouts during COVID).
