Non-agricultural land use types include residential, industrial, natural habitat, and other. For the analysis, I join artichokes with vegetable row crops, blueberries and vine crops with caneberries, and cover crops and unknown agricultural land with fallow ground, given the limited number of parcels in each of these categories. These detailed land survey data are coupled with tax assessor ownership and parcel boundary data from the County Assessor offices to form appropriate decision units. These data delineate property boundaries and enable assignment of land use and groundwater quality to each farm at the land parcel level, designated by the Assessor’s Parcel Number in the tax assessor data. The average size of a parcel is 33.94 acres. I also use ownership data from the County Tax Assessor to aggregate parcels to the ownership level . This helps both with crop decision-making , and with water attribution. Since not every parcel has a well or recycled water turnout, but almost all parcels use groundwater or recycled water, I want to assign that parcel water from the most likely source,danish trolley which would be a well or turnout on another parcel owned by the same person. The definition of “agricultural land” is therefore a parcel with documented ownership information that has been designated by PVWMA at some point to have produced a crop, and has known access to water.
To be classified as having known access to water, the parcel must contain a well or a recycled water turnout, or the owner of the parcel has a different parcel with well/turnout access. In order to provide a comparison point on how restrictive this definition is, we look at the tax assessor land use classifications. After filtering out land uses that involve residences, businesses, and industry, across Santa Cruz and Monterey counties there are 1653 parcels that could plausibly be in agriculture. After restricting the dataset to parcels that also have a clearly linked source of water, I’m left with 1048 parcels. Therefore, this is a relatively restrictive set of qualifications. While this could lead to an overestimation of water applied per acre for the parcels classified as being in agriculture, as described below, there is no evidence that this is biased in a particular direction for growers receiving delivered water. Moreover, the difference-in-differences analysis and event studies do not use quantities of recycled water delivered in their analysis. There are 978 documented wells across Pajaro Valley and 102 recycled water turnouts in the DWZ, all with quarterly measurements of how much water was pumped or delivered. For all of the wells owned by the same owner, I pool the total water pumped and area-weight the water across all parcels planted in a crop during that year. I follow the same procedure for all turnouts owned by the same owner.
This assumes that water needs across all crops are similar, which is largely the case for crops grown in the Pajaro Valley, which require between 2-3 acre-feet a year. I classify a parcel as using water if they are listed for an agricultural purpose that is not fallow. This includes “unknown agriculture”. This does not include parcels that have no information on their agricultural status, making the assumption that PVWMA is capturing all agricultural fields. I gather outside data to estimate crop prices and weather patterns. Data on temperature and precipitation are from PRISM Climate Data, which incorporates coastal weather patterns and land elevation into its projections. The data projections have approximately an 800 m resolution. I use the gridded data cell that lines up with the centroid of each of the parcels. I use averages of monthly mean temperatures and total precipitation in the spring of each year, at the parcel level. Since several of the land use types include multiple crops, I take area-weighted averages of crop prices and revenue for each land use category, and average these across Santa Cruz and Monterey counties.The following section evaluates the impact that recycled water has on growers that receive water deliveries using a discrete crop choice model. Growers receiving recycled water may see benefits of two forms: their yields may improve with higher quality water, and they may be able to grow more salt-sensitive crops.
To capture these benefits in tandem, I use a revealed preference, panel mixed logit model that evaluates the impacts of salinity and recycled water on crop choice, and calculates the willingness to pay for improvements in salinity. This framework also allows for a simulation to find the counterfactual crop choices without recycled water deliveries. The modeling framework, empirical strategy, and results of the model are detailed below. Spatially and time-varying data on groundwater quality, recycled water deliveries, and land use allow for the estimation of a panel mixed logit model to understand the impacts of salinity and recycled water on crop choices. The research design relies on observable changes in groundwater quality and recycled water deliveries, along with controls for observable and unobservable factors that may be correlated with both salinity, recycled water, and crop choice. I take a similar approach to the working paper by Sears, Bruno, and Hanemann 2022 that estimates damages associated with groundwater salinity. Building on these methods, this approach differs in a few key ways. Most importantly, I consider only parcels that can be clearly linked to a source of water, since the analysis specifically accounts for recycled water deliveries. Additionally, part of the water assignment structure involves adding in tax assessor data to the model, and standard error clustering at the ownership level. This helps account for decision-making on crop rotations, allocation of water across parcels, and crop diversification that may occur. Finally, I allow for an unbalanced panel of parcels in agriculture: a parcel is able to leave agriculture and is still counted in the analysis up until the point of departure. where the parameters can be estimated by maximum likelihood. The panel mixed logit structure model avoids invoking the Independence of Irrelevant Alternatives assumption that troubles multinomial and conditional logit models , which is particularly important in a setting with perennial crops and repeated observations over time. The mixed logit model provides the framework for the direct welfare estimates of the delivered program. Under this structure, a simulation is run that estimates growers’ crop choices if they did not have access to recycled water, and estimates welfare with and without the water deliveries. In addition, the estimates from the mixed logit model provide willingness-to-pay estimates for improved water quality by crop, by dividing the marginal utility of reducing salinity by the absolute value of the parameter estimate on crop price α. Identification of the WTP hinges on the assumption that, conditional on the suite of relevant spatial and time-varying parcel-level observables and an annual time trend,vertical aeroponic tower garden unobservable factors are not correlated with both salinity levels and crop prices.One potential concern with this analysis is that groundwater salinity may be endogenous to crop choice, if salinity is controllable by an individual farmer.
While soil salinity problems can often be managed by an individual grower, provided that they have enough freshwater to leach salts out of the root zone, groundwater salinity is harder to influence. Groundwater salinity is largely based on unobserved hydrological and transmissivity properties of the underlying aquifer, along with recharge rates from precipitation and runoff, distance to the coast, and aggregate basin-wide groundwater pumping. In the crop choice model, I explicitly control for groundwater pumping, depth to groundwater, temperature, and precipitation. It is also possible that basin-wide groundwater pumping may be influenced by other unobserved economic factors that also impact crop choice. I control for the distance to the coastline, since this is an important determining factor of soil texture and the spatial distribution of salinity in Pajaro Valley, and because coastal micro-climates determine how well crops grow in certain areas. Features of the parcel, such as its size and crop history, are also likely to influence planting decisions and could be correlated with salinity. The inclusion of a linear time trend accounts for basin-wide unobservables that may trend linearly with both salinity and crop choice over time. Finally, the inclusion of county fixed effects and time and parcel random effects can combat regional effects that may be correlated with salinity. To identify the impacts of recycled water on crop choice, there may also be concerns about how recycled water is allocated. Conditional on being in the delivered water zone, assignment of recycled water is somewhat random: parcels that lie on the currently constructed portion of the Coastal Delivery System are the only parcels able to receive water deliveries. Growers apply for their recycled water turnout to be turned on. Application acceptance is conditional on whether there is excess recycled water available to serve the needs of all current users, and on underlying groundwater salinity. Recycled water availability depends on facility capacity, which has slowly increased over time. Finally, while underlying groundwater salinity does impact application acceptance, a parcel’s groundwater salinity is not highly correlated with its impact on the aquifer salinity: i.e. the positive or negative externality from recycled water deliveries or groundwater pumping to neighboring parcels is not well-linked to a parcel’s own salinity levels. This is especially important for the identification of the DiD and event studies, since salinity is explicitly controlled for in the crop choice model.
Finally, it’s important to note that the outside option in the choice model is fallow land, rather than land that leaves agriculture. I drop parcels from the sample after they leave agriculture permanently. This is standard in the crop choice modeling literature, largely because it is difficult to think about comparing annual profits to a lump sum payment received when exiting agriculture. However, it does limit the model to thinking about relatively short-run effects of salinity damages. It is plausible that a grower experiencing dramatic increases in groundwater salinity may not believe that water quality will improve or that they will receive recycled water on a fast-enough timeline to remain in business. In the appendix, I use linear probability models to test this hypothesis, finding that increases to salinity are not significantly linked to parcels leaving agriculture.Results from the crop choice model are presented in Table 3.3. Column 1 shows the basic conditional logit estimates and Column 2 presents the mixed logit results, which is the preferred specification and has a lower AIC. The table reports crop-specific estimates of the effect of groundwater salinity, where salinity is measured as the total dissolved solids from March-May of the growing season. In the mixed logit, the coefficients on the crop indicator variables are treated as random variables and are allowed to vary across parcels. Their coefficients and estimated standard deviations are also reported in the table, while coefficients on additional independent variables are suppressed. Standard errors are clustered by owner. I control for parcel-specific factors that directly affect salinity, including the depth to the groundwater table, groundwater pumping, the cumulative precipitation from March-May of the growing year, and the average mean daily temperature from March May of the growing year. Also included are factors related to the parcel that may affect crop choice, including the parcel size, recycled water deliveries, distance to the coastline, and additional aggregate controls for agency’s water pumping fee and a linear annual trend. All reported coefficients are relative to fallow ground, which increases in response to an increase in salinity. Results demonstrate that, compared to fallow ground, an increase in groundwater salinity decreases the probability that a farmer will grow a high-value, salt sensitive crop. We see the largest effects among vegetables, strawberries, and caneberries, the most profitable and some of the most salt-sensitive crops grown in the region. Since not all crops in our choice set are grown in the delivered water zone, the clearest way to evaluate the impact that recycled water has on crop choice is to look at the marginal effects of changes in salinity on crop choice for parcels that do and do not receive recycled water deliveries. These results are presented in Figure 3.8. Results are presented for each of the major crops that are grown both inside and outside of the delivered water zone, for three levels of TDS: 500, 1000, and 1500 mg/L. These are all relatively moderate levels of salinity: enough to impact yields, but not enough to completely destroy a crop.