20 December, 2017
Masayuki Kudamatsu
Wed 17 January | My last lecture |
Wed 24 January | Workshop |
Thu 15 February (9 am) | Submission deadline |
Read pages 24-108 of this book during the winter holidays
Recap of
Theory
II
Assumptions
Predictions
+
Hold in reality?
Consistent with reality?
The political agency model
Theory
II
Assumptions
Predictions
+
Citizens want a particular policy
Incumbent will be kicked out
if he/she fails to deliver the policy
Can the political agency model explain
Motivation for Today
Citizens always vote, never abstain
In the previous lectures we assumed:
In reality:
Not every voter turns out at national elections
Question for Today
Today's Road Map
Today's Road Map
We take an action if
Citizens go to poll if
Citizens abstain if
Citizens go to poll if
Citizens abstain if
weighted by
If I vote A, A wins.
But if I vote B, B wins.
weighted by
If I vote A, A wins.
But if I vote B, B wins.
In a large election (e.g. national elections)
An individual vote will almost never change the election outcome
100,000,000+ voters across Japan
Your single vote matters only when
those who vote are equally split between two top candidates
If we assume whether each citizen votes is random
we can calculate the probability of being pivotal
See Myerson (2000) for detail.
With 5 million citizens, it's only
0.00000081079%
Citizens go to poll if
0.00000081079%
1 yen
124 million yen
Only a small cost of voting
Bad weather
Schedule conflict
will then dissuade everyone from voting in national elections
In reality, many people, if not all, do vote in a large election
In poorer countries, people even form a long queue to vote
Image source: The Telegraph (2014)
South Africa in 2014
To solve this puzzle, we need to assume the cost of voting is
or
(i.e. voting per se makes people happier)
Today's Road Map
Cost of voting
Zero
Negative
Negative
If the cost of voting is zero
If the cost of voting is negative
Today's Road Map
The full-fledged model is highly mathematical
Here we see a simple version of the model
Voter 1: always prefer candidate 1
Consider an election with two candidates, 1 and 2
(Could also be a referendum on the proposed policy)
Suppose there are four voters
Voter 2, 3, 4: prefers the "correct" candidate
Voters 2, 3: do not know which candidate is "correct"
Voter 4: knows which candidate is "correct"
The majority rule decides which candidate wins
Flip a coin if it's a tie
Go to poll and vote candidate 1
Cost of voting is zero
A very small probability of being pivotal
is enough for citizen 1 to go to poll
Go to poll and vote the correct candidate
Cost of voting is zero
A very small probability of being pivotal
is enough for citizen 4 to go to poll
Voter 1 goes to poll and vote candidate 1 (by assumption)
Voter 4 goes to poll and vote candidate 1 if 1 is correct
candidate 2 if 2 is correct
Voter 1 goes to poll and vote candidate 1 (by assumption)
Voter 4 goes to poll and vote candidate 1 if 1 is correct
candidate 2 if 2 is correct
If 1 is correct
If 2 is correct
Voter 2 votes 2
Voter 3 abstains
We check if these actions are optimal
given the other voter's behavior
Given that voter 3 abstains...
If 1 is correct
If 2 is correct
If 1 is correct, 1 will win anyway
If 1 is correct
If 2 is correct, 2 will win for sure by voting 2
If 2 is correct
If 1 is correct, 1 will win anyway
Voting 2 is optimal
If 1 is correct
If 2 is correct
If 2 is correct, 2 will win for sure by voting 2
Voting 2 is optimal
If 1 is correct
If 2 is correct
If 1 is correct, 1 will win anyway
If 2 is correct, 2 will win for sure by voting 2
Given that voter 2 votes 2...
If 1 is correct
If 2 is correct
If voting 1...
If 1 is correct
If voting 1...
The outcome won't change if 1 is correct
If 2 is correct
If voting 1...
A wrong candidate can win if 2 is correct
If 2 is correct
If voting 1...
A wrong candidate can win if 2 is correct
The outcome won't change if 1 is correct
No reason to vote 1
If 1 is correct
If voting 2...
If 2 is correct
If voting 2...
The outcome won't change if 2 is correct
If 1 is correct
If voting 2...
A wrong candidate can win if 1 is correct
If 1 is correct
If 2 is correct
If voting 2...
The outcome won't change if 2 is correct
A wrong candidate can win if 1 is correct
No reason to vote 2
If abstaining...
If 1 is correct
If 2 is correct
If abstaining...
The correct candidate wins in both cases
Abstention is optimal
Voter 3 abstains to let the informed voter 4 be decisive
Given that voter 2 votes 2
Voter 2 votes 2 to let the informed voter 4 be be decisive
Voter 1 always votes 1 even if 1 is wrong
By voting 2, it's a tie so the informed voter 4 can decide
Given that voter 3 abstains
Some people advocate we must go to the polls
Don't listen to them
if you really don't know whom to vote
and no candidate is ideologically popular
Many studies show the correlation
But those who want to vote may try harder to obtain information
Jan 1997
Primary schools, daycare, elderly care, etc.
Experiment was evaluated by a consulting firm
4 out of 15 districts start the experiment
of decentralizing the city administration
Late 1999
Sep 2000
Referendum on whether to extend decentralization
to all districts or abolish it
The majority vote for abolishing
Jan 2002
Experiment was terminated
Citizens in the 4 experiment districts
More informed about the benefit of decentralization
Turnout should be higher than in the other 11 districts
Not by their own choice but by external forces (i.e. city govt)
Telephone survey of citizens in Copenhagen
conducted 2 months after the referendum
Ask their opinions on the decentralization experiment
Measuring informed-ness
Went well
Medium well
Bad
Don't know
Informed
Uniformed
Districts | % of informed citizens | Turnout (%) |
---|---|---|
Experiment | ||
Other |
Source: Table 2 of Lassen (2005)
Districts | % of informed citizens | Turnout (%) |
---|---|---|
Experiment | 61.9 | |
Other | 49.1 |
Source: Table 2 of Lassen (2005)
Districts | % of informed citizens | Turnout (%) |
---|---|---|
Experiment | 61.9 | 78.4 |
Other | 49.1 | 69.0 |
Source: Table 2 of Lassen (2005)
More voters abstain in response to a larger cost of voting such as
Bad weather
Registration requirement (not in Japan, though)
Time to think about whom to vote
Distance to the polling station
So the cost of voting is NOT zero...
Data shows
The cost of voting is assumed to be zero
We need to assume voting per se is valuable
Voters belong to groups of like-minded people
with the same preference over candidates
Cast a ballot if there is a consumption benefit from doing so
Two approaches
Some economists propose group-based voting models in which
Today's Road Map
The leader of like-minded people's group
determines the level of turnout among his/her group
To do so, the leader allocates costly resources to voters
e.g.
Trade Union (such as 連合 with 6m+ members)
Environmental groups
Religious groups (such as 創価学会 with 8m+ households)
Then it boils down to a model of costly voting with a few voters
With a few voters, the probability of being pivotal is not too small
Each group's leader
chooses # of votes for his/her favorite candidate
Unclear how leaders affect each voter's decision to vote
Social pressure?
Then why does each voter punish those abstaining?
How does each voter monitor the behavior of each other's?
Today's Road Map
Basic idea: "rule-utilitarian"
e.g.
Why don't some people throw away rubbish on street?
They choose the behavior that
if everyone follows the same behavior
society is in the best shape
This seems applicable to voting
Voters are motivated by a sense of civic duty
They cast a vote based on how well society is
(e.g. unemployment rate)
not just on how well they themselves are doing
Basic idea: "rule-utilitarian"
Again the full-fledged model is quite mathematical
We discuss a simple version of the model to show the main idea
Consider two voters in a referendum
Both voters prefer
a proposal to be approved
The proposal is approved
if at least one voter votes yes
Cost of voting on the day of the referendum: uncertain
Weather, How busy they are, Illness, etc.
Voters collectively decide
a cutoff voting cost below which they go to poll
By following this rule
they derive psychological benefits larger than the cost of voting
2's cost
1's cost
2's cost
1's cost
Cutoff
Cutoff
2's cost
1's cost
Cutoff
Cutoff
At least one voter casts a yes vote
so the proposal passes
2's cost
1's cost
Cutoff
Cutoff
Neither votes
so the proposal is rejected
2's cost
1's cost
Cutoff
Cutoff
A higher cut-off makes the adoption more likely
But the expected cost of voting increases
Extra
cost
Cutoff
The higher the cutoff, the larger the extra cost
Increase the cutoff from 10 to 11: extra cost is 11
11 to 12
12
Extra
benefit
Cutoff
The higher the cutoff, the smaller the extra benefit
2's cost
1's cost
Cutoff
Cutoff
The higher the cutoff, the smaller the extra benefit
Extra
benefit
or cost
Cutoff
Optimal cutoff maximizes the net benefit
Optimal
cutoff
Extra benefit
Extra cost
Extra
benefit
or cost
Cutoff
If the benefit from passing the proposal is larger
Optimal
cutoff
Extra benefit
Extra cost
Extra
benefit
or cost
Cutoff
If the benefit from passing the proposal is larger
Optimal
cutoff
Extra benefit
Extra cost
Turnout for national elections in Japan
Lower House
Upper House
Supporting evidence:
How can voters collectively decide the cost cutoff?
1
2
3
The standard logic of economics
cannot explain why some people go to poll and others don't
(If the cost of voting is zero)
Uninformed citizens may prefer abstaining
Some citizens must benefit from voting per se
but we don't exactly know why
Image source: www.thenation.com/article/shadow-lobbying-complex/
Merry Xmas & Happy New Year !
This lecture is based on the following academic articles:
Feddersen, Timothy J. 2004. “Rational Choice Theory and the Paradox of Not Voting.” Journal of Economic Perspectives 18(1): 99–112.
Lassen, David Dreyer. 2005. “The Effect of Information on Voter Turnout: Evidence from a Natural Experiment.” American Journal of Political Science 49(1): 103–18.