Stacy Dickert-Conlin, Katie Fitzpatrick, and Laura Tiehen
The Role of Advertising in the Growth of the Supplemental Nutrition Assistance (SNAP) Caseload
The views and opinions expressed here do not necessarily reflect the views of the Economic Research
Service, USDA.
Outline
Questions and Background Data Ad Placement Methodology Results Future Extensions
Questions
Is an advertising campaign an effective tool to increase SNAP participation? Existing evidence on Food Stamps –
small scale experiments Small to Moderate Effects in other
means-tested programs Examine the effect of a multi-year
advertising campaign
Preview of Results
Overall Radio advertisement is associated with 2 to 3% increase in caseloads. Applications also increase Approved applications do not
Spanish language TV and Radio ads are sometimes negatively correlated with caseloads and applications
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20
20
5,000,000
10,000,000
15,000,000
20,000,000
25,000,000
30,000,000
35,000,000
40,000,000
45,000,000
50,000,000
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
11.00
12.00
Individuals Receiving SNAP
Unemployment Rate
Households Receiving SNAP
Figure 1: National Monthly SNAP Caseload & Unemployment Rate, 1979-2011
Take-Up Rate (Individuals/Eligibles)
0
10
20
30
40
50
60
70
80
90
100Explanations for Take-Up Rates < 100%
* Stigma & Transaction Costs
* Information
Background on SNAP
Enormous Change in Food Stamp Program 1996 Welfare Reform 2002 Farm Bill 2008 Name Change
Variation in Eligibility Across States
SNAP Advertising Campaigns
USDA-funded campaign started in 2004 to increase awareness of SNAP Emphasize the potential eligibility
of working households Focus on the ability to purchase
healthy food Introduce the new program name
that occurred in 2008
SNAP Advertising Campaigns
Year: 2004 2005 2006 2007 2008 2009 2010
Number of Radio
Ads
13,750
10,892 16,888 22,124 23,032 20,467 26,838
Months Ads
Aired
March AprilJuly
August
AprilMay
August Septem
ber
AprilMayJuly
August
March April
August Septem
ber
January
February MayJune
September
October
March AprilMayJune
Spanish-language TV ads aired in September and October 2006
SNAP Advertising Campaigns
SNAP Advertising Areas
2004 2005 2006 2007 2008 20090
500
1,000
1,500
2,000
2,500
3,000
Target County Spill CountyBorder County County Without An Ad
Comparison of Phone Calls Placed to a SNAP Informational Hotline in Months with and without Advertising Campaigns
2004 2005 2006 2007 2008 2009 20100
5,000
10,000
15,000
20,000
25,000
30,000
35,000
40,000
Average Monthly Calls AnsweredYear
Source:USDA
Comparison of Phone Calls Placed to a SNAP Informational Hotline in TV Advertising States, by Referral Source
0
50
100
150
200
250
300
350
400
450
500
550
TV
Radio
Non-Advertis-ing
Dependent Variables
Caseload Monthly, county-level (total
individuals/population)▪ Unbalanced panel of monthly-level data
from ~2,600 counties in 46 states ▪ Data from 2000 to 2010
Applications, Approvals, Denials Currently monthly, county-level for 14 states
▪ Data from 2000 to 2010
Caseload Data – 46 States
All States except Connecticut, Illinois, Mississippi, New Hampshire, Rhode IslandMonthly Data from 2000 - 2010
Application Data – 14 States
Data Legend
Applications, Approvals, & Denials
Applications & Approvals Only
Applications Only
No Data
Table 2: Growth in County-Level SNAP Caseloads, Applications, Approval, by Presence of a Radio Ad
Caseload Applications Approvals Denials
Year
Without Ad
With Ad
Without Ad
With Ad
Without Ad
With Ad
Without Ad
With Ad
2004
4.38*** 8.52 6.90** 15.75 5.20*** 14.94 6.23* 14.61
2005
4.56*** 6.04 -17.00**
*
2.59 -15.60**
*
4.63 1.18* 14.42
2006
-0.004**
*
2.56 -9.61* 15.64 -8.01 -9.10 5.29 6.24
2007
3.70*** 5.08 5.70 7.75 11.49 7.81 16.76 18.84
2008
6.65 7.27 3.89 3.77 32.23 30.41 41.15 34.95
2009
9.34 8.99 -4.85***
3.67 -1.18***
11.75 -15.76*
**
-0.57
Ad Placement
Advertising location decisions made in the fall Typically, 4-6 months prior to airing of advertisements
Limited funding Many fewer counties chosen than
recommended
If chosen, location received advertisements for all months of the campaign in that year
Ad Placement – Year by Year
TARGET COUNTIES
County proposed to receive advertisements by a Regional Administrator (+ and large)
State food security rates (+)
State Very Low Food Security Rates (-)
State SNAP participation Rates (-)
State adoption of the SNAP name (+)
SPILL COUNTIES
Democratic state government (+)
Counties with larger urban populations
Counties with higher shares of Hispanic persons + small
Still + and large when dependent variable is year of first ad.
Specification (1) - Panel
ctsmc
cl
ltclct
lltcl
lltcl
ct
t
tDemogUnempFSPolicies
STVAdRadioAd
capeSNAPOutcom
*
*
__
)/ln(
12
1)(
6
1)(
6
0)(
Specification (2) - DinD
ctsmccl
ltcl
ctl
ltcl
ctl
ltl
ctl
ltl
ctl
ltcl
ct
ttDemoEcon
FSPSTVAd
spillRadioAd
RadioAd
noadRadioAdNo
capeSNAPOutcom
**
_
*_
target*_
*_
)/ln(
12
1)(
6
1)(
6
0)(
6
0)(
6
0)(
Specification: Control Variables
Food Stamp Policies
Demographic Characteristics Over Age 60 Female Headed
Households Black, Hispanic
Below Poverty Line
Urban
Call Centers
E-Applications E-Signatures Broad-based Categorical Eligibility Immigrant Ineligibility Rules Name Change to SNAP
Table 4: Effect of Advertising Policies on the SNAP Caseload – FE Dependent Variable: ln(SNAP individual caseload/population)ct
Main Specification
Counties Ever Proposed for an Advertisement
Counties Never Proposed for an Advertisement
(1) (4) (5)
Long Term Effects:
6 months of Radio Ads 0.0327 *** 0.0171*** 0.0550***
6 months of TV Ads 0.0066*** -0.0225*** 0.0966***
12 months of county unemployment 0.0392*** 0.0389*** 0.0401***
Month Fixed Effects YES YES YES
Year Fixed Effects YES YES YES
State Time Trends YES YES YES
Observations 246,574 34,325 212,249
R-squared Within 0.8534 0.8643 0.8689
Number of Counties 2,637 359 2,278
Table 5: Effect of Radio Advertising Policies on the SNAP Caseload – DinD -Dependent Variable: ln(SNAP individual caseload/population)ct
Independent Variables Coefficient Estimates
6 months of radio ads:
Target Counties 0.0319***
Spill Counties 0.0191***
No Ad Counties -0.0044***
Border Counties (omitted)
6 months of television ads: -0.0258***
12 Months of unemployment 0.0307***
Month Fixed Effects Yes
Year Fixed Effects Yes
State Time Trends Yes
Observations 256,474
R-squared within 0.7524
Number of Counties 2637
Table 6: Effect of Radio Advertising on Per Capita SNAP applications and enrollments – FE Panel
CaseloadsFull
Sample
CaseloadsONLY states
with Application
data
Applications
Approvals
(1) (2) (3) (4)
6 Months of Radio Ads
0.0327 *** 0.0097***0.01869*
**
-0.0229**
*
6 Months of TV Ads
0.0066*** -0.0662*** 0.0346***-
0.0229***
12 Months Unemployment
0.0392*** 0.0308*** 0.0417*** 0.05922***
Observations
246,574 72,450 51,510 50,393
R-squared Within
0.8534 0.8484 0.6073 0.5859
Counties 2,637 799 863 795
Table 7. Effect of Radio Advertising Policies on the SNAP applications and enrollments
CaseloadsFull Sample
CaseloadsONLY states w/ App data
Applications Approvals
(1) (2) (3) (4)6 months of Radio Ads:
Target Counties 0.0319*** 0.00014*** -0.0402*** -0.05998***
Spill Counties 0.0191*** 0.004355 0.0145 -0.003612
No Ad Counties -0.0044*** -0.00947 -0.02234*** -0.02612** Border Counties (omit)
6 months of TV ads: -0.0258*** 0.012459** 0.028437** -0.014651**12 months unemployment
0.0307*** 0.023107*** 0.01407*** 0.017655***
Month & Yr Fixed Effect & State trends
YES YES YES YES
Observations 256474 72,450 31,713 50,393
R-squared within 0.7524 0.771 0.4113 0.3058
Number of Counties 2637 799 863 795
Table 8: Effect of Radio Advertising Policies on the SNAP applications and enrollments – FE Panel – Hispanic locations
CaseloadsFull Sample
Caseloadsstates w/ high
shares of Hispanic
Applications Approvals
(1) (2) (3) (4)6 Mo of English Radio Ads 0.0338*** 0.0434*** 0.11809 0.0513**6 Months of Spanish Radio Ads 0.0035*** 0.0014*** -0.06235*** -0.0783***
6 Mo of TV Ads 0.0077*** 0.0363*** 0.00465*** 0.01808***
12 Months Unemployment 0.0392*** 0.0437*** 0.0563435*** 0.0745***Month & Yr Fixed Effect & State trends YES YES YES YESObservations 246,574 66,374 16,322 16,170Number of Counties 2637 710 213 205
Conclusions
Overall Radio advertisement is associated with 2 to 3% increase in caseloads. Applications also increase with exposure to radio ads Approved applications do not Why?
▪ Ads encourage recertification?▪ Caseworker behavior changes?
Spanish language TV and Radio ads are sometimes negatively correlated with caseloads and applications Why?
Policy Implications
Outreach has long been an important component of policy EITC, Medicaid, Veteran’s Benefits
Results, thus, far indicate its most effective for those already participating
Next Steps
Gather and include additional application data
Consider the effect of repeated exposure to advertising
Examine observable characteristics of the caseload: average benefit amount, age, household size, employment, etc
Advertisements
Radio Adshttp://www.fns.usda.gov/snap/outreach/radio/default.htm
TV Adshttp://www.fns.usda.gov/snap/outreach/psas.htm
Distribution of Pattern of Advertising Timing 2004- 2010
0000001
0000011
0000110
0001010
0001110
0010100
0011010
0011110
0100010
0101000
0101100
0101111
0110010
0111000
1000000
1000010
1000100
1001000
1001110
1010001
1010110
1011001
1011110
1100010
1100100
1100111
1101010
1101111
1110010
1111000
1111100
11111110
20
40
60
80
100
120
140
Follows pattern: 1 = advertising in that year, 0 = no advertising in that year
*~54% of the sample received no advertising in the 7 years
Year Pattern 2004-2010