Food Affordability

Introduction

Much attention has been focused on ensuring physical proximity to healthy food, often defined as having a full service supermarket within walking distance from home. Federal, state, and municipal policies have provided incentives to encourage supermarkets to open or expand in neighborhoods deemed to have insufficient food retail. Cities and not-for-profit organizations also have programs to support other forms of food distribution, such as mobile fruit and vegetable vendors, farmers markets, grocery delivery services, and emergency food providers such as food pantries and soup kitchens. Far less attention has been focused on another set of important barriers to healthy food: food affordability and food value (quality and price), two dimensions that determine what quantities and types of food households can afford to buy and how far people must travel to purchase food that they can afford and believe it is a good value. Food affordability, in turn, is a function of household income and the overall cost of living, including particularly significant competing household costs such as rent, utilities, childcare, and healthcare. Because time is valuable, food affordability is also a function of the time required to travel to and from supermarkets, and the tradeoffs that consumers make between travel and shopping time versus food costs. The cost of food and other household necessities varies significantly from city to city and within cities, in part due to differing market demand, variable commercial rents, and different costs of doing business, levels of competition, and the structure of the food retail sector within communities. Often regional or global food retailers have cost advantages compared to independently owned grocers that are reflected in retail prices.  In this dashboard we describe two dimensions of neighborhood food affordability (food prices and distance traveled) using several different methods.

Food Price Variability (Method 1)

To assess price variability by neighborhood, we collected, cleaned, and visualized the in-store prices of a selection of products sold in a citywide chain of 133 independently managed grocers by scraping the price information of identical products from the the chain's website.

The dashboard shows a map of all the stores in this chain, color coded by price for a basket of six specific items that were sold in each store. The market basket price ranges from a low of $14.53 to a high of $27.69. To put these differing prices into context, using the filters on the right, you can restrict results based on the percentage of households within a quarter mile of each store facing severe rent burden (spending 50% or more of their income on rent) and with incomes below the federal poverty level.  You can also use the drop-down menus to filter the data by borough, community district, or city council district.   

The charts below the map represent a scatter plot of the stores showing price and level of severe rent burden (on the lower left), and the variation in prices of the individual grocery items (on the lower right).

This visualization presents only a snapshot of prices from one grocery chain at a single point in time.  It is meant to illustrate the potential for this method to enable more comprehensive, real-time monitoring of food prices by community.

Food Price Variability (Method 2)

We visualized food price data collected manually by researchers from the New York City Department of Health and Mental Hygiene (DOHMH) who visited 163 supermarkets in 55 of NYC’s 59 Community Districts. The surveyor recorded lowest in-store prices for 10 perishable food items.  

The visualizations below show the spatial distribution of stores surveyed shaded according to the cost of the food basket (calculated as the summed price of the 10 items). A basket of ten food items ranged in price from $14.98 to $35.11. You can also use the drop-down menus to filter the data by borough or community district  

Below the map, box-and-whiskers plots visualize the distribution of the prices for each pre-selected food item. For example, the largest variability in price was recorded for a pound of ground beef ($1.99-$10.99) and lowest for a pound of bananas ($0.39-$0.99). 

The visualization represents a narrow look of the vast number and variety of items that are sold. It is meant to illustrate the importance of capturing the variability in food prices (even for a subset of goods) within NYC. 

How were stores selected?  

Individual supermarkets were chosen through purposeful sampling based on accessibility by public transport. DOHMH prioritized sampling in gentrifying neighborhoods that may be experiencing rapid changes in the food environment. The supermarkets sampled included global brand grocery stores as well as local store brands that are independently managed. It is important to note that the sampling strategy introduces potential biases in how representative this data may be of the general food landscape in NYC as supermarkets that are accessible via public transit may not be representative of the universe of supermarkets in NYC. 

How was the pricing data collected? 

To reduce the variability in type and quantity of each item in the food basket, DOHMH defined item-level parameters to guide the data collection once in the store. First, we defined a “preferred item presentation” that the data collector should identify first. This preferred item presentation was determined based on amount per selling unit (e.g., pound or gallon), variety (e.g., “Vine tomatoes”), and other characteristics (e.g., leanness in beef). If the exact preferred item presentation was not identified in the store, then an alternative item was chosen, based on predetermined considerations. For four items (eggs, bananas, whole wheat bread and strawberries) no alternative item was defined based on in-store observations done prior to data collection. Finally, having identified the item (preferred presentation or alternative item), the variety sold at the lowest price was recorded in the database. 

Food Price Variability (Method 3)

Prices for individual grocery items are key determinants of food affordability, but perceptions of a combination of prices and other variables, including food quality (e.g., freshness, brand, presentation), product selection, store ambience, and promotion and marketing all play an important role in determining which stores people value most and where they choose to shop for food. To measure this composite indicator of value, we aggregated and visualized crowd-sourced assessments of high and low prices of approximately 800 supermarkets across the five boroughs by analyzing the texts of supermarket reviews on Google Maps.  The two dashboards that follow show how price perceptions and views of quality vary by grocery chain and neighborhood (filter by borough, community district, or city council district). 

Distance Traveled to Supermarkets

Consumers travel to the supermarkets they deem to be good value (based on factors like price, quality, variety, cleanliness, acceptability), often skipping nearby stores. We analyzed a dataset of anonymized mobile device location data to visualize how far people from different neighborhoods travel to the supermarket in which they shop.  For neighborhoods throughout NYC, we mapped the percentage of trips from home to supermarkets that are located within the home community district and the percentage of trips to stores outside the community district to indicate which neighborhoods have stores that appear to meet the shopping needs of the neighborhood and which neighborhoods are under-served, in that shoppers choose to travel to other communities to buy groceries.  One can use the sliders on the right to filter areas with high rent-burden and high food insecurity. 

Potential Use Cases

The data presented above show that there is broad variability in prices for identical or similar grocery items, even within the same grocery chain and even among stores located relatively close to each other.  For households on fixed incomes, these variations in prices can greatly affect whether a sufficient amount of healthy and nutritious food can be purchased in a given week or month.  Low-income households may face high grocery prices in their immediate neighborhoods (or perceive the prices to be high), and as a result may travel to other neighborhoods where the stores provide greater value for their food dollars.  These findings illustrate that programs to improve access to food need to consider cost, perceived value, and other variables in addition to the physical proximity of supermarkets to consumers. Collecting real-time data on prices can help identify economic barriers to healthy food and enable city officials and non-profit organizations to target potential areas of need and tailor their programs to increase economic as well as physical access to healthy food in communities facing high prices. Using crowd-sourced data to understand how consumers rate grocery stores based on perceived costs, value, and other variables can inform programs to encourage supermarket development and expansion in neighborhoods underserved by food retailers tailored to meet local need. Location data can illustrate consumer decisions about where to shop, providing another perspective on which neighborhoods would benefit from additional food retail.