Last year the produce industry had a major discussion on the possibility of mandatory assessment to support a generic promotion program. We tried to access the most knowledgeable people in the field, having discussions, for example, here and here with Harry Kaiser, the Gellert Family Professor of Applied Economics and Management, Department of Applied Economics and Management, College of Agriculture and Life Sciences at Cornell University and the preeminent scholar in the field.
Yet, in the end, what we found was a paucity of knowledge on broad-based generic produce promotion in the United States. Everyone was left arguing by inference from milk or beef or by making assumptions regarding the relevancy of programs in Australia or elsewhere to the American environment.
The proposal for a mandatory assessment and the generic promotion that went along with it was put aside. Yet the debate over the proposal is beginning to bear fruit for the industry and the world in the form of better understanding of the way generic marketing can influence sales of fresh produce.
Harry Kaiser from Cornell, Tim Richards from ASU and Brad Rickard and Jura Liaukonyte from Cornell have been working on a project that uses experimental economics to examine consumer response to commodity-specific media advertising (e.g., apple ads) and broad-based media advertising for fruits and vegetables (e.g., the UK 5-a-day campaign); their objective has been to understand the relative economic implications of using these two types of advertising, as well as a third approach that combines the two types.
The approach of the scholars is very rigorous and quite innovative. There is little question that the results will be of interest to both academics and the industry.
As part of our efforts at The New York Produce Show and Conference to gather the intellectual resources of the region and bring them to a broader audience, we invited Brad Rickard to unveil this important research at the event. We asked Pundit Investigator and Special Projects Editor Mira Slott to get us a sneak preview:
Brad Rickard |
Q: What sparked your interest in analyzing consumer response to commodity-specific and broad-based promotion programs for fruits and vegetables using experimental economics?
A: I was following this dialogue on generic promotions in the Perishable Pundit. An editorial in PRODUCE BUSINESS in 2009 did a good job questioning Got Produce?, the term referencing a proposed generic promotion expanding the five-a-day effort. There was a lot of uncertainty that needed to be resolved before asking producers to vote on something like this.
Q: Do you have a particular connection to the produce industry?
A: I’m from a fruit and vegetable farm in Canada, Ceresmore Farms. We’re a family farm up in Ontario, which over the years has grown processing vegetables, apples, pears and some field crops. It lends me a bit of credibility in studying this subject.
There was a pretty big push in Australia on a generic produce campaign a few years ago. A study written by some R&D folks showed the program was a success, but I have colleagues knowledgeable on that campaign that conducted their own analysis. A lot of money and effort was put into that Go for 2&5 promotion. This was a motivation for me, as generic promotions are an area of work economists haven’t weighed in on. We do things differently than R&D folks and psychologists.
Q: In what ways? How could research based on models of applied economics offer new perspective?
A: Jim Prevor laid out key questions needing more research to be done before producers and handlers could really vote on a generic promotion. We honed in on issues we felt were critical: First, what would be the difference between spending more money on commodity-specific promotions relative to a generic five-a-day program? And another very real question, if you’re doing something one way and you switch, what would be the impact, or what would happen if you did a combination of both efforts?
Q: Did you also consider money thresholds, number of times consumers need to be exposed to the ads, the scope of the promotion, etc., in weighing the impact of a generic campaign? Produce industry executives often lament that their promotional marketing budgets pale in comparison to big junk food manufacturers and other processed food conglomerates.
A: Out of English-speaking countries — Australia, the UK, Canada, and the U.S. — the U.S. is least funded on a per capita basis in this area.
Q: The Got Milk? campaign inundated consumers nationwide with clever ads repeatedly, promoting the message through famous people touting the attributes of drinking milk. Was that considered a success?
A: There are different ways to measure these things. Got Milk? was successful in getting a lot of advertising awards, but producers were concerned it wasn’t selling more milk. In a Supreme Court case, producers argued it wasn’t increasing their bottom line.
The difficult thing in some cases is the counter-factual effect: what would have happened otherwise? In the case of dairy, there was a declining trend in how much milk people were drinking. A lot of people argue the Got Milk? promotion didn’t increase demand for milk, but slowed the decrease in demand. If you just look at broad brush strokes, you’d think Got Milk? didn’t do anything.
Q: It reminds me of the economic argument related to the recent recession; that if we didn’t do the bailouts, the economic picture would have been significantly worse…
A: It’s a close analogy to Got Milk? and hard to isolate what would happen otherwise. That is a good segue to a generic produce promotion. We wanted to examine this question as economists. How can we account for these changing times and add an element of control into our analysis so we can say something meaningful?
At Cornell, we have a Lab for Experimental Economics and Decision Research — LEEDR is the acronym. It’s a sophisticated state-of-the-art lab with 24 stations, and several researchers on campus use it, those curious how consumers, producers or policy-makers respond to economic incentives or changes. We thought this would be one way to shed some light on Got Produce? — by looking at the consumer side, the producer side, and the general economy and policy stages, so we decided to approach this in three stages. The first stage was to get inside the consumer’s head, when he or she sees promotion efforts, how does it register?
We wanted to investigate one of Jim Prevor’s points: How would a five-a-day generic promotion compare to the status quo commodity promotions? And following that, the counter-factual question, in the absence of these programs, what would you see?
Q: How did you go about that in a lab setting?
A: We set up an experiment comprised of six sub-experiments, where we brought a bunch of people into the lab, and every case had auctions for 8 or 10 different fruits and vegetables. We’d talk about those items, a pound of potatoes, tomatoes, carrots, and red peppers, counting tomatoes as a vegetable, and apples, bananas, oranges and grapes.
The auctions were conducted with 24 “consumers” at a time; done as a private affair, with housing around the cubicles. We ran an English style of auction, where the price started at 0, and as the price crept up, the participant when ready would click on their computer and submit their bid.
We wanted to try and make this feel somewhat real. We showed some short T.V. clips, and inserted promotional pieces. For three batches of people, we only showed the T.V. clips and no promotional pieces. For the other treatments, we’d introduce the participants in one instance to commodity-specific apple ads from New York and Washington State. And then there were another three batches of people that received an apple treatment and potato treatment with Idaho potato ads appearing. A third group had broad-based ads.
Q: Did you create a mock Got Produce? promotion for the purpose of the experiment?
A: Since we didn’t have U.S. generic produce ads, we took some from the Australia campaign Go for 2&5, and some from the UK campaign. We borrowed T.V. ads.
Q: By using UK and Australian promotions, wouldn’t your participants be struck by the different accents and cultural nuances? Were the campaigns similar in content and messaging?
A: The generic ads emphasized variety and health benefits rather than taste, and were colorful. The dialects were different but we don’t think this caused any problems. We had some treatments where we interacted with the groups and did a little broad-based generic, and a little commodity, which we called the hybrid affect.
Overall we did six separate treatments that ran through auctions during the different scenarios. There were different people in the different batches, which was an issue that we tried to account for. Say there were 72 different people, so 72 times six separate treatments, so roughly 400 people in total, and these are all different people in different batches, all exposed to media, some with promotion, some mixed.
Q: How did you account for the wide variety of people in your analysis?
A: After the experiment, we asked everyone to complete a survey to tell us about themselves, 30 different questions, their age, height and weight so we could come up with their BMI, income level, whether they were a vegetarian, female, male, how did you like the ads, etc. We tried to statistically control details about these people, and isolate all those effects to say something meaningful about the bids they were submitting into the computers.
Q: What if someone would never buy apples, regardless of how clever the promotion?
A: That’s a good question. One of the survey questions was, tell us your preferences. Then if a consumer made a particularly low bid for apples when he or she saw lots of apple ads, we could account for that.
Q: How did you recruit participants? Did you get a demographically diverse population?
A: Sometimes these experiments on university campuses are criticized because they are all students. We went out of our way to recruit staff, not students or faculty. They were all from Cornell, but could be an administrative assistant, a bunch worked as janitors, the hockey coach, we had a nice demographic. I have a chart of New York demographics and how our 400 people sample compared to New York at large. It’s impossible to get a cross-section that perfectly represents the population at large, or the New York population.
Q: What were participants told when they came into the room?
A: We paid participants $25 to come. To create an incentive-compatible experiment, we told them that at the end of the auction, people would randomly be chosen to have to buy their products. So they would either leave with $25 or a combination of money and produce.
We told them that if they see apple ads, they don’t know another group is not seeing any apple ads, or potato ads, etc.
Q: Did Phase 1 of your experiment lead to any significant findings?
A: We’re trying to understand what consumers are willing to pay for fruits and vegetables. We tried to simulate the noise we’re all subjected to.
In Phase 1, we wanted to explore whether consumers respond to these different kinds of ads. The punch line is we saw significant results for the way people filled out the survey and significant results on how they responded to the various ad approaches. These broad-based programs, or ones mixed in with some apple or potato ads, always seemed to increase their bids.
Q: How considerable were the differences?
A: We had a base line of a whole bunch of bids from our control group that didn’t see any ads. A phenomenon occurred where, for example, people bid $1.10 for apples, and when they saw the apple ad didn’t increase their bid, but if they were shown broad-based and commodity ads or just broad-based ads, their bid would grow significantly.
If you lump information together, there was willingness for people to increase their bid by 15 percent to 18 percent when watching just broad-based ads relative to the control group that saw no ads, which translates to a seven percent to eight percent increase in quantity purchased.
Q: Since participants only had the choice of 8 or 10 produce items to bid on, wouldn’t that seriously distort results? What would have happened if they had the option to use their $25 to buy a candy bar or a roast chicken? Also, couldn’t the controlled setting and limited time-period impact preferences as well?
A: This represents an upper limit on the effects of advertising, understanding we were doing this in a lab, where there wasn’t someone also selling cookies or shoes. This is a good point and one that has haunted WTP [Willingness to Pay] studies for economists, especially those that interview consumers or perform choice experiments without compensating the subjects or forcing subjects to follow up on their bids.
The best answer here is that in our experiment we randomly ask subjects to follow up on their bids; half of the auctions become binding at the end and the highest bidder for these items makes the purchase and their bid is subtracted from the $25 they receive from participating. We hope that this feature encourages subjects to pay more attention to their bids and makes our experiment more in line with reality.
Since we paid our subjects $25 to participate in the experiment, and most of them left with $25, this money is thought to be used for other food products like steak and Twinkies. This isn’t perfect, but it is better than many other approaches out there. Furthermore, we thought that things might get out of control if we included more than 8 products in the experiment, or if the experiment lasted more than one hour.
Q: What did your experiment reveal about commodity-centered ads?
A: The more important part of Phase 1 is that we see the impact for the broad-based ads, but we don’t see it for commodity-specific ads.
Q: Is it correct that the commodity ads focused on specific state products, and if so, did the purchase alternatives include state identifiers?
A: The apple and potato ads were state-specific. We first decided to focus on apples and potatoes given that they represent important fruit and vegetable crops in the U.S. Then we went looking for ads and these were the best we could find. Three of the four apple ads were for New York apples. Although we didn’t explicitly auction “New York” apples or “Idaho” potatoes, we did introduce the products in a way that indicated they were similar to the ones shown in the ads, auctioning “Empire” apples and “Russet” potatoes.
Q: Why did you elect to use commercials from state-based programs rather than national programs such as those existing for watermelon and mango?
A: A national campaign would have been much better, but we were pretty set on apples and potatoes given their market share, and the fact that they are produced in many states in the U.S. Plus, we wanted fruits and vegetables that were often sold in a one-pound size in the range of one to three dollars. The regression analysis does control for the “quality” of the ads, so the WTP estimates should be free of any distortions concerning consumer response to the particular ad.
Q: Do you think the actual content of the ads played a role?
A: The broad-based ads emphasized health. What I think happens is that a broad-based ad is for a category, all fruits and vegetables. When advertising apples, we might see a little jump for apples, but a decline in another commodity. I think of a pie chart; if the apple group puts out an ad, it might be successful in increasing apple sales, consumers will buy less bananas or oranges. So by targeting a niche, the increase is at the expense of close substitutes.
The results from this study indicate that the nature of the advertising matters. If just advertising one thing, it substitutes away from something else in the category. If advertising a category, it is now competing against another category. It’s not necessarily a revolutionary discovery.
When the French put an ad in Wine Spectator, you won’t see a one page ad for a particular brand of champagne… you see a generic ad for French champagne; the higher tide lifts all the boats kind of thing. A clever broad-based ad has the ability to increase the category. Several commodity-specific ads end up competing against each other. It’s a destructive game of cross advertising where everyone is trying to grab a slice of the pie. Instead, the industry could be competing against junk food, or meat, or dairy. Generic ads, getting people to think of fruits and vegetables, could be a better strategy for the industry.
Q: How do you compensate for the limitations of a prefabricated lab setting to address the effects of real-world factors?
A: At least in the auction, we included money incentives compatible with purchasing something, mitigating that problem of consumers saying one thing and doing something else. We do ask participants things like how many vegetables they eat per week. Since they might be concerned about revealing certain information about themselves, we don’t take their names when accumulating survey results.
In Phase 2, we want to examine exposure issues; how much exposure to the ads do consumers need in the real world to get the results in the lab? How much of a budget is required, is it $3 million or $30 million? Those are things we can address in our lab using economics. Also, there are other financial questions we need to address. The produce category is made up of many different players. In a broad-based ad campaign, should strawberry people pay as much as broccoli people? Are strawberry people more likely to benefit? How do you determine relative costs across commodities? What are the different ways to think about this? We think we could simulate scenarios, a spectrum of possibilities. Is the consumer response to advertising strawberries more than to spinach or broccoli?
And then if you put ads together what attributes do you emphasize? If you’re doing an apple ad, it’s easy to pinpoint certain attributes, such as sweet and crunchy…for blueberries it could be antioxidants, and for baby carrots it might be kid-friendly or convenience. Imagine a broad-based program. Look at ones being used in the UK or Australia and what they emphasized, breaking down all the components… Do apples and strawberries chip into this program, or increase banana sales?
First we start with consumers and then we go to the producers. The third step is the market linkage; questions involving the retailers, producers, the consumer connection, different supply chains and how these would change results. Public policy comes into this if the scheme eventually ends up being funded by growers, handlers and packers with a matching program by government. We need to run through these scenarios and the distribution impacts.
Q: What you describe seems like a big leap from your initial lab research results…
A: Our research involves phases. We started with big broad brush strokes. We’ve done the first phase, written it up and it is under review by academic journals. As we explore answers to these questions, our research raises new questions.
Professor Rickard is part a generation of bright young economists who are transforming the field of what used to be called “agricultural economics.” Like his Cornell colleague Miguel Gomez — who, as we mentioned here, is also presenting at The New York Produce Show and Conference — these economists are approaching issues without fear or favor, looking to use innovative techniques in a quest for real knowledge and understanding.
It is not all about produce; in fact, we first read a paper by Professor Rickard when he made a slam dunk case that New York State ought to allow supermarkets to sell wine. And, as the authors acknowledge, the questions regarding generic promotion won’t be resolved by one simulation. But it is an important start.
We are especially thrilled that this research was undertaken because it is really what the Perishable Pundit and PRODUCE BUSINESS and the conference portion of The New York Produce Show and Conference are all about: Raising issues, identifying questions, creating an intellectual format, provoking better thinking, finding the path that will initiate industry improvement.
Obviously we look forward to additional phases. We are particularly interested in the rent-dissipation hypothesis, which we wrote about here and the differential impact on long-lead crops such as pears as opposed to row crops that can quickly be increased in volume.
We also have a suggestion for another application of this methodology. All over the country, many states have run programs such as Jersey Fresh to promote their own state’s product. The research we have seen over the years indicates that these programs, a variant of generic promotion, are quite successful, mostly within a given state’s population. So if only one state does this type of program, it will likely be very profitable because it will switch in-state buyers to that state’s produce.
What if, however, as has happened, other states launch similar promotions? If all these promotions were equally effective, we might, as a country, spend a lot of money on promotions that leave total consumption fixed.
On the specific study Professor Rickard is discussing at The New York Produce Show and Conference, we would also like to see an exploration of the degree to which the marketing creates short term purchasing changes as opposed to long term consumption changes.
This strikes us as a limitation on this methodology. Effective marketing may get one to stock up on apples or potatoes. This may, however, reduce purchases next week. Long term, consumers are unlikely to buy more produce unless they also eat more produce. It is not clear if the advertising will produce that kind of behavioral change.
This is important work, and it represents new ways of exploring for answers to important questions. We can’t wait to find out more.