NEW HAVEN

DEMOCRACY
FUND
PROGRESS AND METRICS
ZOE BECKER
SEPTEMBER 13, 2023

1
TABLE OF
IN T RO D UC TI O N ................................ 3

CONTENTS M E TH O DOL OGY . .. . . .. . . .. . . .. . .. . . .. . .. . .. . .4

K EY FI N DI N GS ................................ .. 5

IN- T OW N & S M A L L D O NA TI O NS . .. . .. . . ..6

DO NA T IO N T IM I N G. . .. . .. . . .. . .. . . .. . .. . .. . . .8

RE GR ES SI O N A NA L Y SI S. . .. . .. . . .. . .. . .. . .1 1

C O NC L U SI O N . .. . .. . . .. . . .. . . .. . .. . . .. .1 2

2
INTRODUCTION
The New Haven Democracy Fund is a Connecticut pilot program that utilizes public financing to
distribute funding through matching funds/grants to mayoral candidates. The goal is to take the
focus of candidates away from fundraising and redirect it towards outreach, as well as increase
the percentage of funds coming from within the town. Candidates voluntarily participate and
must agree to all their contributions being limited to below $445 per individual/committee, and
limited to $445,000.00 total for the primary, and another $445,000.00 for the general election
(City of New Haven, n.d.).
Overall, since its inception in 2007, the New Haven Democracy Fund (NHDF) has met its goals
of increasing the percentage of mayoral campaign donations coming from within the town, as
well as dramatically increase the total number of donations from individuals, increasing voter
engagement and reducing the need for other sources like self-funding and committees.
Additionally, NHDF participating candidates received a significantly larger number of smaller
donations, compared to fewer donations of larger sums that non-NHDF participating candidates
received.
This paper uses public campaign filings to analyze and pinpoint where and who donations for
mayoral candidates are coming from, as well as compare the results of NHDF-participating
candidates compared to non-participating candidates.

3
METHODOLOGY
This study uses public SEEC data for the election years 2011, 2013, 2017, 2019, 2021, and
2023. The year of 2015 has been excluded due to it not being considered a “competitive
election”. All candidates with enough funding to fill out SEEC form 21 are accounted for,
except for some of Toni N. Harp’s 2017 campaign filings being inaccessible.
From the SEEC, we have utilized the addresses, donation amounts ($), donation types
(individual, committee, self, etc.), and donation dates of campaign filings for these
candidates. The NHDF kindly provided a list of candidates that did/did not participate, and
this paper also utilizes the New Haven Neighborhood Profile data on CTdatahaven
published in 2020, 2019, 2017, 2016, and 2012 concerning the years 2018, 2017, 2016, 2014,
and 2010 to be used in neighborhood-level linear regression analysis.
By using this data, we examine the trends of how participating and non-participating
candidates’ respective donation amounts have changed over time to gain a sense of how
the implementation of the NHDF has changed mayoral political engagement – have more
people donated? Which neighborhoods and towns? Has the composition of mayoral
funding changed over time? This study aims to answer these questions and more.
All statistics and graphs were created and calculated in Tableau, and the linear regression
analysis was computed through Stata. The regression tables are attached in the appendix.

4
KEY FINDINGS

OVER A 28% INCREASE OF

IN-TOWN DONATIONS

NHDF-participating candidates were

found to have over 28% more of their
total individual ($) donations from
donors within the town of New
Haven compared to nonparticipants.

$5,000,000+ OF SMALL
DONATIONS

Participating candidates, between

the years of 2011-2023, have raised
over 5 million dollars consisting of
donations under $445, while non-
participants rely heavily on large
donations

42% OF NEW HAVEN

DONATIONS ARE FROM TWO
NEIGHBORHOODS

42% of all individual donations from

within New Haven from 2011-2023
are from donors located in East Rock
and Westville

5
 VISUAL DATA
IN-TOWN DONATIONS and SMALL DONATIONS
From the election years 2011-2023, participating candidates have received over 47,000
donations (over $4 million) from within New Haven, in comparison to non-participants, who
have received only around 16,000 donations (around $2.6 million) from within the town. This is
over 28% more of the sum comparatively of their total individual donations ($) and over 18%
more of the total number of donations coming from in-town for participants.
This indicates that the program’s requirement of receiving “contributions [of no more
than] between $10 and $390 from 200 distinct New Haven registered voters” is working to not
only increase the percentage of mayoral funds coming from within the town but could also be
leading to more focused campaigns that increase engagement within the town.

6
 VISUAL DATA

98.41%, or over $5,132,124.36 of the funds raised for NHDF

participating candidates (excluding money received from the
fund) consists of donations from individuals of $445 and
under, almost completely compliant with the program’s
expectation.
On the other hand, over 63.76%, or $1,825,437.44 of non-
participating candidates’ funds were made of donations
consisting of over $445. Nearly half of the donations are
$1000 and over, which means that these donations are
deriving from people and committees who have the capital to
donate thousands of dollars at a time.
Yet, NHDF candidates still match and even exceed the total
amount of individual funds raised compared to non-
participants. Excluding Democracy Fund grants/matching
donations, but including self-funding and committees, NHDF
candidates have raised $186,761.50 more than non-
participating candidates since 2011. Additionally, NHDF
candidates have received over 28,800 more individual
donations compared to non-participants, suggesting that
more and more voters are getting involved in mayoral
fundraising, which could be an indication of the fund
contributing to more political engagement. NHDF candidates
also have a much smaller proportion of personal funds used.

7
DONATION TIMING
Participating candidates tend to receive an upswing of donations around the June prior to the
election, while non-participants receive far more towards the very beginning of the year (January-
March) and right before the election (late September and October), missing the large spike of donors
around June-July that participating candidates historically experience. Most of these donations within
these “spikes” are in-town, New Haven donations that disappear once filtered out (while the other
patterns remain consistent). One possible explanation for this is that NHDF candidates, being
incentivized to increase engagement within the town to get more in-town donors, are spending more
time campaigning within the town during “slower” times earlier in their campaigns, prior to the
September primaries where donations usually spike. The spikes of numbers of donations during this
period, as more and more candidates commit to the NHDF program, have steadily increased across
multiple elections.
Furthermore, NHDF-participating candidates disproportionately receive donations more from
certain neighborhoods compared to non-participants. Specifically, the neighborhoods of East Rock,
Prospect Hill, and Westville. From 2011-2023, NHDF candidates received over $1.4 million from East
Rock (non-participants received less than $300,000), over $850,000 from Prospect Hill (compared to
only $300,000 from non-participants), and over $800,000 from Westville (compared to only $600,000).
While this could simply be an indication of local neighborhood support for each candidate, these
neighborhoods have substantially grown in terms of numbers of donations across multiple campaigns,
nearly exclusively for NHDF-participants. See the figure below.

8
9
 POVERTY RATE AND DONATIONS PER CAPITA

Regression analysis of the number of donations per neighborhood from 2011-2023. The upper graph
measures the correlation between neighborhood poverty rate and the number (#) of donations per
capita multiplied by a factor of 100 for mayoral elections for non-NHDF participating New Haven
candidates between 2011-2023 (excluding 2021 and 2015), and the lower graph measures the
correlation between neighborhood poverty rate and the number (#) of donations per capita multiplied
by a factor of 100 for mayoral elections for NHDF participating New Haven candidates between 2011-
2023 (excluding 2021 and 2015).

10
 REGRESSION ANALYSIS
This paper tests the theory that these new NHDF votes are disproportionately in these three
neighborhoods due to their higher income, lower poverty status by running four separate linear regressions
using donation data from the SEEC from mayoral election years 2011, 2013, 2017, 2019, and 2023 compared to
neighborhood profile data from 2010, 2014, 2016, 2017, and 2018. The election year of 2015 was excluded due
to it being a non-competitive election, and 2021 was excluded due to not having enough non-NHDF
candidates to properly compare data between the two.
The first two regressions were calculated for NHDF-participants and non-participants respectively,
comparing the number of individual donations received per neighborhood capita (multiplied by a factor of 100
to make correlations clearer) to the percentage Black and poverty rate of that neighborhood for that time,
considering yearly and population fixed effects. The second two regressions consider those same above
variables, but instead measures the correlation of those demographic variables against the total sums ($) per
capita for each neighborhood for those election years. The conceptual model is below:
Count of Donations Per Capita / Donations ($) Per Capita = Neighborhood Poverty Rate + Percent of
Neighborhood Black +Election Year Fixed Effects + Total Neighborhood Population Fixed Effects
Overall, results showed that NHDF-participants had a slightly larger negative correlation for both
Poverty Rate/Donation Counts (coefficients of -4.652 and -2.803 respectively) as well as Poverty
Rate/Donation Sums (-3.84 and -3.828). Or, in other words:
FOR EVERY PERCENTAGE INCREASE OF THE POVERTY RATE IN A NEW HAVEN
NEIGHBORHOOD, PARTICIPATING CANDIDATES’ CAMPAIGNS ASSOCIATED A
DECREASE IN THE NUMBER OF DONATION COUNTS FROM THAT NEIGHBORHOOD BY -
0.04652 AND A DECREASE IN THE NUMBER OF FUNDS RAISED BY -$3.84. NON-
PARTICIPATING CANDIDATES EXPERIENCED A DECREASE OF -0.02803 FOR THE
NUMBER OF DONATION COUNTS AND -$3.83 FOR THE SUMS RAISED FOR A
NEIGHBORHOOD WITH A 1% INCREASE IN THE POVERTY RATE.

While both NHDF-participating and non-participating candidates also had a negatively correlated
result of percentage Black and donation counts/amounts, it was not statistically significant to consider on its
own for any of the regressions. These results certainly cannot prove for certain that this is the case, due to
having relatively low R-squared values (between 0.2 – 0.4 for all the below regressions, attached at the bottom
of this paper).
These results are significant enough to somewhat support the hypothesis that the NHDF is
incentivizing new people to donate in mayoral elections, but that these new donors reside in
disproportionately wealthier neighborhoods of New Haven. A possible explanation behind this could be that
the NHDF is succeeding in its goals to engage more voters/donors in-town, and the people who have the
ability to donate to these campaigns are wealthier individuals more likely to reside in the wealthy
neighborhoods of New Haven. This could also explain why these three neighborhoods make up such a large
portion of the NHDF June-July donation spikes, as these new donors might be actively picked up during the
middle of participants’ mayoral campaigns to maintain their status of having most of their donors/funding
come from within New Haven.

11
 CONCLUSION
Overall, the NHDF has met and exceeded certain metrics of improving in-town donor
engagement, likely amongst people who are new to donating/local political involvement.
However, to fully realize the positive effects of this program distributed across New Haven
neighborhoods in a more equitable way, more work must take place.

MORE IN-TOWN ENGAGEMENT

NHDF candidates receive a much larger portion of their funds from within the
town of New Haven than non-participating candidates. Additionally, they
recieve a much larger portion of these donations from individuals donating
smaller amounts of money.

GROWING NUMBERS

The number of newer New Haven donors is growing over time, especially for
participating candidates from the neighborhoods of East Rock, Prospect Hill,
and Westville, likely due to their lower poverty-rate and therefore larger
number of possible donors.

How could the NHDF mitigate this unequal impact of their program on the surrounding
neighborhoods? While it is inevitable that wealthier areas will always be able to donate more
to campaigns, some suggestions to mitigate this effect could be incentivizing candidates
through grants to campaign and gain more donations from lower-income neighborhoods,
making mayoral debates and other events physically closer to lower-income areas, or possibly
distributing money directly to low-income voters.
Regardless, the implementation of the NHDF has not only led to a large increase of the amount
of engagement in local New Haven politics but has illuminated the existing inequality that
exists in local political fundraising. The NHDF has become successful in meeting its metrics,
and with a larger scope, could be used to solve other issues of inequity and disengagement for
local politics.

12
Non-Participating Candidates’ Donation Counts (#) Per Capita Compared to Neighborhood Poverty Rate and
Percent Black

CountCap Coef. St.Err. t-value p-value [95% Conf Interval] Sig

Pctblack .637 .337 1.89 .062 -.032 1.306 *
Povertyrate -2.803 .728 -3.85 0 -4.251 -1.356 ***
TotalPop 0 0 1.12 .264 0 0
2011 0 . . . . .
2013 1.028 .188 5.46 0 .654 1.402 ***
2017 .037 .19 0.19 .846 -.341 .415
2019 .235 .188 1.25 .215 -.139 .61
2023 -.107 .203 -0.53 .6 -.511 .297
Constant .535 .235 2.27 .025 .067 1.002 **

Mean dependent var 0.470 SD dependent var 0.735

R-squared 0.414 Number of obs 93
F-test 8.582 Prob > F 0.000
Akaike crit. (AIC) 171.829 Bayesian crit. (BIC) 192.090
*** p<.01, ** p<.05, * p<.1

2. Non-Participating Candidates’ Donation Sums ($) Per Capita Compared to Neighborhood Poverty Rate and
Percent Black
DonCap Coef. St.Err. t-value p-value [95% Conf Interval] Sig
Pctblack -.452 .636 -0.71 .48 -1.716 .813
Povertyrate -3.828 1.375 -2.78 .007 -6.562 -1.093 ***
TotalPop 0 0 -0.29 .775 0 0
2011 0 . . . . .
2013 1.394 .356 3.92 0 .687 2.101 ***
2017 .016 .36 0.05 .964 -.698 .731
2019 .138 .356 0.39 .699 -.57 .846
2023 -.458 .384 -1.19 .236 -1.221 .305
Constant 1.739 .444 3.91 0 .855 2.622 ***

Mean dependent var 0.825 SD dependent var 1.298

R-squared 0.330 Number of obs 93
F-test 5.988 Prob > F 0.000
Akaike crit. (AIC) 290.202 Bayesian crit. (BIC) 310.463
*** p<.01, ** p<.05, * p<.1

13
3. Participating Candidates’ Donation Counts (#) Per Capita Compared to Neighborhood Poverty Rate and Percent
Black
CountCap Coef. St.Err. t-value p-value [95% Conf Interval] Sig
Pctblack -.931 .763 -1.22 .225 -2.447 .584
Povertyrate -4.652 1.617 -2.88 .005 -7.865 -1.438 ***
TotalPop 0 0 0.12 .904 0 0
2011 0 . . . . .
2013 .573 .427 1.34 .182 -.274 1.421
2017 -.112 .415 -0.27 .788 -.937 .713
2019 1.103 .416 2.65 .009 .277 1.928 ***
2023 .672 .421 1.60 .114 -.165 1.51
Constant 1.849 .512 3.61 0 .832 2.866 ***

Mean dependent var 0.856 SD dependent var 1.441

R-squared 0.236 Number of obs 97
F-test 3.935 Prob > F 0.001
Akaike crit. (AIC) 335.045 Bayesian crit. (BIC) 355.643
*** p<.01, ** p<.05, * p<.1

4. Participating Candidates’ Donation Sums ($) Per Capita Compared to Neighborhood Poverty Rate and Percent
Black
DonCap Coef. St.Err. t-value p-value [95% Conf Interval] Sig
Pctblack -1.861 .996 -1.87 .065 -3.839 .118 *
Povertyrate -3.84 2.112 -1.82 .072 -8.036 .357 *
TotalPop 0 0 -0.32 .749 0 0
2011 0 . . . . .
2013 .706 .557 1.27 .209 -.401 1.812
2017 .003 .542 0.01 .995 -1.075 1.081
2019 1.643 .543 3.03 .003 .564 2.721 ***
2023 1.125 .55 2.04 .044 .032 2.218 **
Constant 1.847 .668 2.76 .007 .519 3.174 ***

Mean dependent var 0.829 SD dependent var 1.872

R-squared 0.228 Number of obs 97
F-test 3.761 Prob > F 0.001
Akaike crit. (AIC) 386.813 Bayesian crit. (BIC) 407.411
*** p<.01, ** p<.05, * p<.1

14
BIBLIOGRAPHY
City of New Haven. (n.d.). Summary of Fund. Retrieved from City of New Haven:
https://www.newhavenct.gov/government/democracy-fund/summary-of-fund
CT Data Haven. (2012, 2016, 2017, 2019, 2020). New Haven Neighborhood Profile(s). Retrieved
from DataHaven.
https://www.ctdatahaven.org/find/data-resources?field_resource_type_tid=26
State of Connecticut Elections Enforcement Commission. (2011-2023). Documents/Filings.
Retrieved from https://seec.ct.gov/eCrisReporting/SearchingDoc.aspx

15