ERS Studies Using USDA Food Consumption Survey Data Biing-Hwan Lin, Lisa Mancino, Francis Tuan, and Travis Smith Economic Research Service, USDA May 2009
Jan 04, 2016
ERS Studies Using USDA Food Consumption Survey
Data
Biing-Hwan Lin, Lisa Mancino, Francis Tuan, and Travis SmithEconomic Research Service, USDA
May 2009
What We Eat in America (WWEIA)
Part of the National Health and Nutrition Examination Survey (NHANES)
Includes one or two days of dietary recall— what was eaten, how much, where, and when
Can be linked to: Socio-demographic characteristicsHealth indicators
Knowledge and attitudes about diet and health
Food and Commodity Economic Database (FCED)
Created by USDA to use with food survey data
Used to translate foods all the 7,000+ foods reported consumed into a limited number of commodities
Needed to bridge food consumption data with commodity consumption analysis
Four main areas of ERS research with these data
Who eats what, when and where?
What are the economic and behavioral determinants theses choices?
How might these choices change in the future?
How do these choices affect health?
Who eats what and where?
0
20
40
60
80
100
Per
cent
of t
otal
Home Fast food Other
Source: USDA’s Continuing Survey of Food Intakes by Individuals, 1994-96.
Dry bean consumption by food source
Who eats what and where?
0
2
4
6
8
10
12
14
16
Ground Stew Steak Beef dish Other cut Processed
At home Away from home Restaurant Other
Ground beef is consumed more in outlets away from home that at home
Source: USDA, ERS, Agriculture Research Service, 2000: 1994-96 and 1998 Continuing Survey of Food Intakes by Individuals (CSFII).
Pou
nd
s
Additional ERS research on who eats what, when and where
Vegetables dry beans, spinach, tomatoes, frozen potatoes, onion, mushroom, garlic, cucumbers, celery, cabbage, sweet pepper, sweet potatoes, snap beans, sweet corn, carrot
Fruits oranges, apples, watermelon
Nuts tree nuts, peanuts
Animal products
beef, pork
Others sweeteners
Determinants of food choice—income
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
$- $10 $20 $30 $40 $50 $60 $70 $80 $90 $100
Income (in Thousands)
Daily
Veg
etab
le S
ervi
ngs
Dark Green/ Deep Yellow Tomatoes Fried Potatoes Other
Determinants of food choice—dietary knowledge
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0 2 4 6 8 10 12
Dietary Knowledge Score
Daily
Veg
etable
Ser
vin
gs
Dark Green/ Deep Yellow Tomatoes Fried Potatoes Other
How might choices change in the future?
0
5
10
15
20
25
30
35
40
45
1977-78 1987-88 1989-91 1994-96 2003-06
Per
cent
of
Tot
al C
alor
ies/
Foo
d E
xpen
ditu
res
Food expenditures on food away from homeCalories from food away from home
Calories from fast food
Consumption projections
Regression analyses are conducted to examine the effects of income, social, and demographic factors on commodity consumption
Regression results are used to project commodity consumption
Analysis of potato consumption indicates lower intake per person
Index (2000=100)
Potato product 2000 2005 2010 2015 2020
French fries 100 100 100 99 98
Potato chips 100 99 97 96 94
Baked potatoes 100 101 101 104 106
Other potatoes 100 99 98 96 95
Projections of per capita potato consumption, 2000-2020
Lin and Yen, “U.S. Potato Consumption: Looking Ahead to 2020.” Journal of FoodProducts Marketing, 2004, 10(2).
But total US consumption will rise
Index (2000=100)
Potato product 2000 2005 2010 2015 2020
French fries 100 104 108 112 115
Potato chips 100 103 106 109 111
Baked potatoes 100 105 110 117 125
Other potatoes 100 103 107 109 112
Projections of total US potato consumption, 2000-2020
Lin and Yen, “U.S. Potato Consumption: Looking Ahead to 2020.” Journal of FoodProducts Marketing, 2004, 10(2).
A comprehensive projection
Economic and demographic factors
Lin, Variyam, Allshouse & Cromartie. “Food and Agricultural Commodity Consumption in the United States: Looking Ahead to 2020.” ERS 2003
Possible changes—background on our analysis
Foods are separated into 25 groups, consumed at home and away from home
Food consumption is affected by social, demographic, and economic characteristics
Forecast future food consumption by using forecasted social, demographic, and economic conditions
Food consumption is converted to commodity (22 groups) using two technical databases—Pyramid Servings Database and Food and Commodity Intake Database
Changes in demographic makeup indicates more fruit
Lin, Variyam, Allshouse & Cromartie. “Food and Agricultural Commodity Consumption in the United States: Looking Ahead to 2020.” ERS 2003
Changes in dietary patterns and awareness have additional impact
Lin, Variyam, Allshouse & Cromartie. “Food and Agricultural Commodity Consumption in the United States: Looking Ahead to 2020.” ERS 2003
Associations between diet and health
Correlations between women’s BMI and age, race, dietary patterns, TV watching, and smoking for both low- and high-income
Beverage consumption, eating out, importance of maintaining healthy weight, and exercise correlated with BMI only among women from high-income household
Among children, age, race, income, and mother’s BMI were significantly correlated with child BMI
Lin, Huang and French, International Journal of Obesity (2004), 28
Food choices and health—few Americans eat a healthy diet
Percent change from 2001-2002 consumption needed to meet 2005 Guidelines
Source: National Health and Nutrition Examination Survey 2001-2002.
Why might that be a problem?
Majority of American adults are either overweight or obese
Rates are increasing among children as well Obesity is believed to cause a number of
health problems Certain dietary patterns are associated
increased risk of obesity But do these dietary patterns cause poor diets
Why it can be hard to show causality—example of food away
from home What to eat is jointly determined with where to
eat
Not accounting for relevant unobservables will bias estimates If choosing FAFH is driven by fondness for certain
(less nutritious) foods → ↓bias FAFH’s impact on diet quality
Our approach to this issue—fixed effects analysis
Requires two or more days of dietary intake
DQit=Diet Quality on day t for individual i
FAFHit=Number of FAFH meals for i on day t
Xi=Additional explanatory variables for i that affect DQ
μi=Unobservables for i that also affect DQ
εit=Stochastic error term
itiitiit FAFHXDQ
Our approach to this issue—fixed effects analysis
With two days of dietary intake, we find within individual differences over both days
Or more simply,
)()(
)()()(
1212
12
iiii
iiiiii
FAFHFAFH
XXDQDQ
iii FAFHDQ )(
Our data
Two days of dietary recall data
As dependent variables, we focus on calories and specific components of diet quality
Control for meal patterns and whether intake day was a weekend
Our findings
After controlling for self-selection issues, each additional meal away from home adds about 130 daily calories significantly lowers intake of fruit, whole-grains
and dairy and increases intake of certain fats and added sugars
Eating one meal away from home each week translates to almost one extra kilogram a year
Other applications
This could be easily extended to specific commodities or food groups
It would be simple to use this sort of fixed effects estimator with more days of intake data
謝謝 Our contact information
Biing-Hwan Lin ([email protected])Lisa Mancino ([email protected])
Francis Tuan ([email protected])Travis Smith ([email protected])
Economic Research Service, USDA1800 M St NW
Washington DC 20036-5831www.ers.usda.gov