Given that about 7.1% of pork with be produced Prop 12 rules , this implies about 0.54 million of 7.6 million sows in North America will be confined under California’s housing standards. California pork consumers will pay about $188 million annually to provide four square feet more per sow on average for about 540 thousand sows in North America. Thus, California buyers of covered pork will pay about $87 per square foot of additional housing space. The passage of Prop 12 in California by a significant majority indicates citizens’ interests in improving animal welfare. Regulations such as Prop 12 and its counterparts in other states such Local jurisdictions have increasingly imposed regulations on agricultural production processes within the jurisdiction to address issues associated with pollution, animal welfare, and farm worker health and well-being. Several papers have studied the impacts of such regulations, with the work summarized by Sumner . These regulations differ considerably in their impacts from those that restrict farm production practices for products sold within a local jurisdiction. The first type creates heterogeneous production costs and alters the comparative advantage of different production regions but generally does not affect downstream operations. This paper explores the economic implications of the latter group of regulations, with specific application to the impact of California Proposition 12 on the North American pork supply chain. Key innovations of the model are allowing heterogeneity in the costs of farms to meet the mandate and incorporating that a mandate in many cases will only apply to a portion of the output of the live animal. The model incorporates capital conversion for compliance at farms and variable production proportions between covered and non-covered pork in processing farm raw products into finished consumer products.

The model shows how these aspects interact and drive substantial price and quantity adjustments along vertically linked markets and across geographically different markets. Simulations show that, rolling benches for growing despite significant industry opposition to Prop 12, its mandates do not impose much negative total effect on hog producers in North America. Most firms that elect to comply with Prop 12’s mandates will increase profits, and losses to non-compliers are slight. Prop 12 causes moderately higher prices in California for covered pork products and generates a consumer welfare loss of about $188 million annually.Prop 12 will not make stall housing operations adopt California’s standards. These pork products will be diverted for the California market under Prop 12. Because California’s standards are stricter than typical group housing, breeding pigs confined in converting operations will have slightly more space than before. Prop 12 and, more generally, the regulations on products sold in local jurisdictions represent only one policy instrument to improve welfare for farm animals. To illustrate this point, I considered a simple alternative policy under which the California government would raise a general fund to directly subsidize farms that convert their housing practices. I showed that, for the same cost to California, this alternative policy could incentivize conversion of about three times as much sow housing to compliance with Prop 12 regulations as Prop 12 itself will achieve if it becomes fully implemented. This example illustrates that Prop 12 and, more broadly, regulations imposed at the point of purchase are likely not efficient ways to influence conventional farming practices.Google Surveys provides inferred respondent characteristics based on internet use rather than reported demographics by respondents to represent the general population of internet users. Google collects demographic information of their users when they make an account or while they use Google’s service. Google identifies general demographic information about websites when sufficient Google users visit those websites.

Given this demographic information specific to websites, Google infers visitors’ demographic information. Google Surveys cannot collect data from non-internet users, but the share of internet users in the U.S. population was about 91% in 2020 . Google Surveys cannot obtain responses from non-internet users, but it is not rare that responses from part of the population are not collected in survey studies. For example, survey studies often collected responses from only college students , people in a local region , and customers in a store . Several papers have found evidence that the inferred demographics of Google Surveys provided representative samples of the U.S. population and reliable estimation results . Table 5.3 reports respondent shares by demographics. The numbers in parentheses are the corresponding 95% confidence intervals. Table 5.3 reports the results by three different samples: a subsample including only respondents who selected “I don’t buy carrots,” a subsample including only respondents who did not select that option, and the full sample. Table 5.3 shows important patterns in the data about inferring demographics. Google inferred gender, age category and region based on search patterns and other information about the URL of the respondent computer. The shares of respondents without inferred demographic information was about 20% for gender and age, but negligible for region. The gender was not inferred for about 18% of respondents in the full sample. The age group was not inferred for about 20% of respondents. The geographical location was inferred for more than 99% of respondents. The shares of “not inferred” are slightly higher in the respondents selecting “I don’t buy carrots” than the other respondents. Table 5.4 excludes respondents for whom Google Surveys did not infer the full set of demographics, presumably because they did not have sufficient information on that respondent. The overall sample shares are similar to the U.S. population shares. The share selecting “I don’t buy carrots” is more male and younger than those who responded to the carrots purchase choices and the U.S. population. The gender pattern in the total sample is very similar to the U.S. population. The age range is slightly more middle aged with fewer 25-34 and fewer over 65. In addition, a smaller share of respondents is in the Northeast and South and more are in Midwest relative to the U.S. population. Given the findings of this section, I consider the following points in the model specification of Chapters 6 and 7. First, I include demographic variables as explanatory variables. Second, I use sampling weights based on demographic groups to make the sample represent the population.

Google Surveys also provide information about how long individual respondents elapsed between when the survey was opened and when it was completed. I explore the response time because several prior studies using online responses found that the inclusion of response times as a control in statistical estimation reduced random responses and standard errors of estimated parameters . The concern is that respondents that are too quick may not be actually reading the questions, and respondents that take too long were likely interrupted in their responses. Table 5.5 reports descriptive statistics on the response time by the WTP question types, the choice of “I don’t buy carrots,” and whether inferred demographics were provided. The table includes ten categories. Three features are common across the categories. First, the average response time is slightly less than 30 seconds for most categories. Second, the standard deviation within each category is high relative to the average for all the categories in the table. Third, the min and max values are substantially different from the average in each category. Three points are noticeable in comparison with categories. First, on average, respondents took about the same time for the Yes-No questions as the Multiple-Choice questions . Second, respondents choosing “I don’t buy carrots” tended to spend less than the other respondents. Third, on average, respondents without inferred demographics spent more than those with demographics. Based on the findings of this section, I consider the following model specification in Chapters 6 and 7. First, I include the response time as an explanatory variable in regressions. Second, I compare the models with and without outliers in response time. The outliers include both those with a very short response time and a very long response time because the relationship between response time and response reliability is possibly not linear . A very long response time possibly indicates insufficient attention to the survey because respondents often do multiple activities simultaneously on the internet.Specifically, corporate control refers to control of political and economic systems by corporations in order to influence trade regulations, tax rates, and wealth distribution, among other measures, vertical air solutions and to produce favorable environments for further corporate growth. Structural racialization refers to the set of practices, cultural norms, and institutional arrangements that are reflective of, and help to create and maintain, racialized outcomes in society, with communities of color faring worse than others in most situations. In this light, the production of racial/ethnic, gender, and economic inequity in the United States is more so a product of cumulative and structural forces than of individual actions or malicious intent on behalf of private or public actors. In order to challenge and eliminate corporate control and structural racialization in the United States, therefore, it is necessary to analyze the ways that public and private institutions are structured. It is also necessary to analyze how government programs are administered and operate in ways that reproduce outcomes that marginalize low-income communities, women, and communities of color in terms of health, wealth, land access, power, and degree of democratic influence. Additionally, as this report aims to do, it is crucial to analyze the genesis and formation of critical institutions and structures themselves.Therefore, the US Farm Bill—the flagship piece of food and agricultural legislation since its inception in 1933, which informs the heart of public and private policies that make up much of the US food system—is the subject of this report. This report is of particular importance now for two reasons. First, the Farm Bill will be under consideration again in 2019, yet there is no comprehensive critique of the Farm Bill that addresses its underlying contradictions, particularly with regard to racial/ethnic, gender, and economic inequity. Second, it is imperative that campaigns by grassroots, community, and advocacy organizations—generally most active during the period of Farm Bill negotiations in Congress—have enough time to gather adequate information and conduct in-depth analysis for targeted yet comprehensive policy change. As such, the timing of this report is also imperative for coalition-building efforts and the growth of an effective broad-based food sovereignty movement.Corporate consolidation and control have become central features of the US food system, and of the Farm Bill in particular. As of 2014, large-scale family-owned and non-family-owned operations account for 49.7% of the total value of production despite making up only 4.7% of all US farms. As of 2013, only 12 companies now account for almost 53% of ethanol production capacity and own 38% of all ethanol production plants. As of 2007, four corporations own 85% of the soybean processing industry, 82% of the beef packing industry, 63% of the pork packing industry, and manufacture about 50% of the milk. Only four corporations control 53% of US grocery retail, and roughly 500 companies control 70% of food choice globally. At every level of the food chain, from food production to food service, workers of color typically earn less than white workers. For example, a majority of farm workers who receive “piece rate” earnings , and many of whom are migrants from Mexico, frequently earn far less than minimum wage—an exploitative practice deeply tied to immigration policy, as elaborated upon below. On average, white food workers earn $25,024 a year while workers of color make $19,349 a year, with women of color, in particular, suffering the most. Furthermore, few people of color hold management positions in the food system, while white people hold almost three out of every four managerial positions. One result of this racial disparity in food system labor is that non-white workers experience a far greater degree of food insecurity than their white counterparts.Food insecurity in the US disproportionately affects low-income communities and communities of color, and these communities are over represented in the lowest-paying sectors of the labor market. For example, as of 2013, 14.3% of US households—17.5 million households, roughly 50 million persons—were food insecure. The report also found that the rates of food insecurity were substantially higher than the national average among Black and Latino/a households, households with incomes near or below the federal poverty line, and single parent households.