The slowdown in agricultural productivity growth and its causes

Agricultural productivity growth sounds rather boring and arcane, but it is perhaps the most important concept related to the health of the food and agricultural economy. At a basic level productivity growth occurs when the growth in all agricultural outputs exceeds the growth in use of all inputs. Crop yields are a partial measure of productivity growth - it measures one output (e.g., corn bushels) divided by one input (acres of land), but what economists really focus on is TOTAL factor productivity growth (all outputs and all inputs). Total factor productivity growth is intricately linked to sustainability, farm profitability, farm labor, consumer food prices, and more.

Phil Pardey and Julian Alston have written a new paper entitled “Unpacking the Agricultural Black Box: The Rise and Fall of American Farm Productivity Growth” that was just released by the Journal of Economic History. If you read one paper on trends and causes of agricultural productivity growth in the United States, this should probably be it. It is packed full of interesting data and discussion.

The key phenomenon they identify and attempt to explain is the following:

“... we present robust and compelling evidence of a structural slowing of productivity growth in U.S. agriculture, following a mid-century surge ... [R]ather than a constant rate of productivity growth the data are more consistent with a “big wave” surge in productivity growth peaking in the 1960s; a secular pattern in U.S. agricultural productivity similar to what others have found with reference to the economy as a whole, but with different timing”

Pardey and Alston provide a number of interesting figures related to these phenomena; here is one related to adoption of different technologies over time.

Source: Pardey and Alston, Journal of Economic History, 2021

Here is a bit from their conclusions:

At issue in many minds is whether anything like the rapid growth in measured farm productivity during the third quarter of the twentieth century could be recaptured in the coming decades. Was this productivity surge (and the subsequent slowdown) a one-time phenomenon, or something that can be repeated with new waves of innovation in genetics, informatics, and robotics, which can save on costs of labor (which remain stubbornly large as a share of total costs) or other increasingly scarce inputs—especially land and water? More concisely: What might have accounted for the surge and slowdown in American farm productivity?

To address these questions, we examined three alternative (albeit related and not entirely mutually exclusive) explanations for the surge slowdown phenomenon.

Those three explanations are:

• A decline in investments in agricultural research and development and the slowing growth in knowledge stocks;

• A “big wave of technological progress through the middle of the century contributed to a sustained burst of faster-than-normal productivity growth throughout the third quarter of that century” (see the figure above); and

• “the dynamics of the structural transformation of the U.S. farm economy and the role of asset fixity” … “This structural transformation involved a one-time shift, to reduce the number of farmers and the total farm labor force by two-thirds or more”

Do read the whole thing.

Where are people most sensitive to changes in the price of bacon?

Whether trying to understand the impact of taxes, animal welfare regulations, or meat packing plant shutdowns, we need an elasticity of demand for pork. The elasticity of demand tell us how the quantity of pork consumers want to buy changes with the price of pork. Given the importance of such questions, it probably isn’t surprising to learn that there are many studies aiming to measure elasticties of demand. These studies typically focus on THE elasticity of demand for pork - a single aggregate number. However, these aggregate assessments likely mask a great deal of heterogeneity across markets and different products.

In some new research with Glynn Tonsor, done for the National Pork Board, we utilized grocery store scanner data from 51 U.S. retail markets for 6 different pork products to estimate 51*6 = 306 market- and product-specific own-price elasticity estimates. Our data also enables us to observe differences in consumer purchasing and spending patterns across the country.

There are so many interesting results, it’s hard to succinctly summarize. Here are a few highlights.

First, consider variation in bacon purchases across four markets over time. Of the four locations in the figure below, per-capita bacon purchases tend to be highest in Phoenix and lowest in LA (it is worth noting that bacon prices tend to be much higher in LA than Phoenix). The impact of the initial COVID-19 disruptions is also apparent in the data.

There is wide variation in price sensitivity across location and pork product. The figure below summarizes the distribution of price elasticities over the 51 markets for the six pork products

Want to know how your locale ranks in terms of consumption, prices, or elasticity? Check out the full report.

Concentration and Resiliency in the U.S. Meat Supply Chains

That’s the title of a new working paper I’ve co-authored with my Purdue colleague, Meilin Ma. In the wake of the COVID-19 related disruptions to meat packing, I shared my thoughts about resiliency and ran crude simulations to try to understand how resiliency related to market concentration. In this new paper, we incorporate some of these ideas into a formal economic model that we can use to answer a variety of questions about the relationship between industry structure and resiliency. The model also helps us understand some of the price dynamics surrounding the packing plant shutdowns.

Here is the abstract:

“Supply chains for many agricultural products have an hour-glass shape; in between a sizable number of farmers and consumers is a smaller number of processors. The concentrated nature of the meat processing sectors in the United States implies that disruption of the processing capacity of any one plant, from accident, weather, or as recently witnessed – worker illnesses from a pandemic – has the potential to lead to system-wide disruptions. We explore the extent to which a less concentrated meat processing sector would be less vulnerable to the risks of plant shutdowns. We calibrate an economic model to match the actual horizontal structure of the U.S. beef packing sector and conduct counter-factual simulations. With Cournot competition among heterogeneous packing plants, the model determines how industry output and producer and consumer welfare vary with the odds of exogenous plant shutdowns under different horizontal structures of the sector. We find that increasing odds of shutdown results in a widening of the farm-to-retail price spread even as packer profits fall, regardless of the market structure. Results indicate that the extent to which a more diffuse packing performs better in ensuing a given level of output, and thus food security, depends on the exogenous risk of shutdown and the level of output desired; no market structure dominates. These results help illustrate the consequences of policies and industry efforts aimed at increasing the resiliency of the food supply chain, and highlights the fact that there are no easy solutions to improve resiliency by changing market structure. ”

Two biases - one solution

I was listening to a recent episode of Planet Money that discussed the sunk cost fallacy (or sunk cost bias). The episode reminded of something I’ve long thought: one bias, taken out of context, might in-fact help solve another bias (which itself seems to be problematic when viewed in isolation).

Let me start by describing the two biases. First, the sunk cost bias. I remember well a moment in college when I realized this economics thing might be for me. I was skiing with a group of friends, one of whom was worn out by lunchtime, announcing they were heading back to hotel. Another friend, encouraging the deserter to stay, said something along the lines of: “Common, these lift tickets are expensive. You’ve got to keep going to get your money’s worth.” I remarked we should stop pestering the deserter: the lift ticket was a sunk cost.

Nothing, at this point, would refund the cost of the lift-ticket. So, the decision was not whether to buy a lift ticket or not (that cost was sunk), but, rather, the decision was which course of action, now at lunch, would make the individual happier: A) continue skiing or B) rest in the hotel room. I was pleased that I seemed to convince my friends this was the right way to think about it. Lessons to avoid the sunk cost fallacy (or bias) are probably taught in virtually every ECON 101 class, and yet, it seems to be a bias to which we all routinely fall prey.

Consider a second, seemingly unrelated bias: present bias (or time-inconsistent preferences). It isn’t irrational to care about the present more than the future. But, it is problematic if the rate at which we discount the future changes depending on when we are asked. Consider a simple example. Which would you prefer: A) $100 today or B) 101 tomorrow? Now, a second question. Which would you prefer: C) $100 one year from now or D) $101 one year and one day from now?

It is common for people to choose A over B (“give me the quick $100 bucks now!”) and then D over C (“I’ve already waited a year, what’s one more day to get a dollar?”). There is a problem with that choice pattern. Choice of A over B implies a person is unwilling to wait a day for a dollar but choice of D over C implies the opposite: a willingness to wait a day for a dollar. When people exhibit these sorts of time inconsistent preferences, they tend to want to start a diet tomorrow. But, when tomorrow becomes today, they’re no longer willing to start the diet, and again plan to do it … tomorrow.

These two biases, the sunk cost fallacy and time-inconsistent preferences, are widely discussed in economic research, but rarely together. However, it strikes me that, at least in some circumstances, the sunk cost fallacy can help solve time-inconsistent preferences.

Consider a gym membership. If I exhibit time-inconsistent preferences, I won’t work out as much as I should. I will always imagine my future self being more disciplined and exercise-ready than my present-self ever will be. Yet, many of us pay large up-front gym membership fees. One economics study suggests people significantly over-pay for gym memberships and concludes we’d be financially better off choosing a “pay as you go” plan. But, what if paying a large-up front fee induces the sunk cost fallacy? “I’d better go to the gym to ‘get my money’s worth’”? If so, fretting over our sunk costs would lead us to exercise more than we might otherwise, helping offset the problem of time-inconsistent preferences, which, in isolation, would tend to lead us to exercise less than we otherwise might.

A commonly suggested solution for time-inconsistent preferences is to create commitment contracts. Commitment contracts occur when my present self undertakes actions (or commitments) to bind my future self, or at least makes it more costly for my future self to reverse course. An example is a Christmas Club savings account, a savings account where withdrawals are only allowed (without penalty) around the Holiday season. If people were perfectly rational, a Christmas Club account would be unnecessary; we’d just use our “regular” savings accounts that have more flexibility and, in all likelihood, pays higher interest rates. Yet, some people choose to use Christmas Club savings accounts as a type of commitment device (I’m binding my future self to not spend the money till the Holidays).

It strikes me that the psychological feelings we have around sunk costs act as a sort of commitment device. Although ECON 101 tells us to ignore sunk costs, the fact that we often fret over them suggests that, at least in certain circumstances, they may be binding us to a course of action our previous self wanted us to pursue.

What do farmers think about plant-based meat alternatives?

I’ve written several times over the past couple years about what consumers are thinking about plant- and lab-based meat alternatives. What are farmer’s thoughts? This is not an unreasonable question: all these meat alternatives rely on agricultural inputs, whether it be pea- or soy-protein, or starches for fermentation processes.

My colleagues Jim Mintert and Michael Langemeier, through the Center for Commercial Agriculture at Purdue and with support from the the CME, run a monthly survey of farmers and produce the Ag Economy Barometer, which tracks farmer sentiment about the direction of the farm economy.

They just released results from the February 2021 survey results. They were gracious enough to include a few ad hoc questions I suggested on what farmers are thinking of the emergence of plant-based meat alternatives.

From the release:

“Interest in alternative protein sources has increased markedly over the last year. The February survey included several questions designed to learn more about producers’ perspectives on the possible impact of alternative proteins on U.S. agriculture. Responses suggest ag producers think alternatives to animal protein will make inroads in the total protein marketplace over the next five years. For example, over half (55%) of producers said they expect alternative protein sources to capture up to 10 percent of the combined market for animal and plant-based protein while a much smaller percentage, approximately 15%, said they expect plant-based alternatives to capture 10 percent or more of the total protein market. In a follow-up question, producers were asked what impact they would expect to see on farm income if plant-based alternatives to animal protein capture a relatively large market share (25%) of the total protein market. A majority of producers said they think the impact on farm income arising from alternative protein capturing a 25 percent share of the total protein market will be negative, with approximately four out of ten producers saying they would expect to see farm income decline by 10 percent or more under this scenario.”

That a majority of farmers perceive negative effects of alt-meats on the agricultural economy is consistent with: 1) the fact that some respondents are likely livestock producers, and 2) a recognition that the amount of corn and soy needed to produce alt-meats is lower than the amount needed to produce an equivalent amount of beef, pork, or chicken.

Nonetheless, the emergence of alt-meat alternatives create opportunities for some farmers who may grow inputs for these new products. We added a final question on this topic to the survey, and the results are below. The results show 62% of producers indicating an unwillingness to grow a crop used in production of plant-based alternatives under contract. That strikes me as high and may include a bit of cheap talk. It may also be that the question was worded too vaguely. What are the conditions of the contract? What are the price premiums? Farmer would want to know answers to these questions (and more) before switching to a new crop, and the lack of specificity may explain the low stated unwillingness to crop used in plant-based alternatives under contract.