Comparing estimates from columns and suggests that the inclusion of time invariant parcel characteristics, such as parcel size and distance to the coast, as well as time-varying lagged crop choice and lagged recycled water deliveries is important.Our empirical approach is deliberate in conditioning on an array of factors that are likely correlated with both crop choice and salinity. The inclusion of the annual time trend and water prices in column provides an opportunity to test if our estimates of WTP are robust to their inclusion. The stability of our estimates between columns and lends credibility to our identifying assumption that, conditional on observables, salinity is uncorrelated with unobservables that may impact crop choice. A look at the marginal effects reveals how crop shares in the region change due to salinity. Using the estimates from column of Table 3.3, we hold the control variables constant at their average levels and show the predicted share of parcels in each crop type under a constant basin-wide TDS level of 2000 mg/L in Table 2.3. This is depicted relative to the average share of crops by type in the sample period. We see that given this increase in TDS of almost 1,360 mg/L, fallow ground would increase from 10.6% of parcels to 12.3% of parcels, holding all other variables at their averages. Likewise, vegetables and strawberries would increase slightly by 1.3 and 1.7 percentage points, respectively, and caneberries would decline by 4.5 percentage points. An increase in vegetable crops relative to fallowed land due to a change in salinity, holding all else constant, is intuitive due to their insensitivity to salt relative to the other regional crops, as shown in Figure 3.4. Interestingly,round plastic pots shares of strawberries also increase and the greatest declines are observed in caneberries, despite strawberries being relatively more salt sensitive. This may reflect the relative value per-unit of these crops or the fact that strawberries are often times rotated with vegetable crops such as broccoli, lettuce, and cauliflower for pest and soil management.

We next use these estimates to deduce growers’ willingness to pay for a reduction in groundwater salinity. We focus on crops whose estimates in Table 3.3 were significant at the 95% confidence level or greater, namely, vegetables, strawberries, and caneberries. The willingness to pay estimates for a reduction in groundwater salinity of 10 mg/L are reported in Table 4. While these willingness-to-pay estimates are wide-ranging, the high dollar values indicate that growers highly value water with low salinity levels. These WTP estimates are on the same order of magnitude as the expected revenue losses from a drop from 100% to 90% yield capacity as shown in Figure 3.4. To put these magnitudes into perspective, row cropland in Monterey and Santa Cruz Counties in 2021 ranged in value from $28,500 to $75,000 per acre, representing some of the most expensive cropland in the state. Contrast this to the range of values observed in 2021 for well-dependent cropland in Fresno county which ranged from $10,000 to $16,000 per acre . To assess the robustness of our results, we test the sensitivity of our results to two modeling choices. First, we take a closer look at our measure of salinity, which can be measured in different ways. We chose to focus on TDS in our main estimation, since it is widely used in the agronomic literature and is a measure of general salinity. This allows us to look at results that may be applicable across multiple types of salinity problems . One concern may be that growers actually respond to an alternative salinity metric when faced specifically with seawater intrusion. Chloride is a measure of salinity that specifically captures seawater intrusion, rather than salinity from other sources, such as soil, rock, or other natural materials. Table 2.5 reports marginal effects based on the estimation of our preferred specification from the last column of Table 3.3, except we replace the salinity variable with chloride measurements. On average, chloride levels are 15% of TDS values, so an increase in chlorides to 300 mg/L is similar to a shift in TDS to 2,000 mg/L. As shown, results are largely the same across these two highly correlated measures of salinity.

Larger marginal effects are estimated for vegetable row crops and caneberries when using chlorides, which may be due to the fact that the spatial distribution of chloride concentrations is different than that of TDS. Finally, we take into account the possibility that other water resources may be available to a subset of growers in the Pajaro Valley. Farmers located near the coast experience some of the highest salinity levels relative to the rest of the basin, and simultaneously impose the greatest externality on others when they pump groundwater. PV Water recognizes this, and in collaboration with the growers in the basin and the City of Watsonville, set up the Delivered Water Zone and a distribution system to deliver recycled water and other alternative supplies to the growers most impacted by seawater intrusion. The distribution system serves roughly 15%, or 5,000 of the 30,000 acres farmed in the Valley. The total quantity delivered has slowly increased from 667 AF in 2005 to 4,203 AF by 2016. This is a small fraction of irrigation water used within the DWZ, but it is likely still relevant for farm-level decisions. It is plausible that growers inside the DWZ have different projections of their future access to high quality water and are less responsive to groundwater salinity, compared to growers without access to delivered recycled water. In addition, anecdotal evidence from conversations with growers in the Valley suggests that growers inside the DWZ who had been unable to plant strawberries before they started receiving deliveries are now able to plant the salt-sensitive crop once again. While we conditioned explicitly on recycled water deliveries to farmers inside the DWZ in our initial estimation, we perform a robustness check by removing all parcels within the DWZ, to focus solely on the effects of salinity on crop choice when there are no other water sources available. The fraction of parcels planted in various crop categories differs for this subset so marginal effects in this case need to be compared to a different baseline. Marginal effects from the estimation of salinity impacts on this sub-sample are shown in Table 2.6.

Results are very similar in magnitude and significance to those reported in Table 2.3, suggesting that the existence of the DWZ is not biasing our estimation of the relationship between salinity and crop choice. Finally, we ask the question of what would happen to crop choices and to consumer welfare under a scenario in which the quality of groundwater deteriorates significantly across the basin. We simulate and compute the utility-maximizing crop choices for each parcel in each year under a high TDS scenario and plot the distribution. To do this, we keep the estimated marginal utilities for the attributes of crops, parcels, and climate variables the same as in the panel mixed logit model displayed in column of Table 3.3. For each parcel, we estimate the probability of choosing each crop type and use these to predict each parcel’s baseline crop choice. Then,hydroponic bucket we recalculate the probabilities of each crop being grown for each parcel after altering the vector of TDS values to reflect a higher salinity scenario. Ideally, we would simulate the change in TDS predicted by a climate change model of sea-level rise. This exercise is challenged by the fact that model predictions of sea-level rise, while necessary, are insufficient for deploying our model of crop choice. To use climate model output in our simulation, we would need to map sea levels to the salinity concentration in the groundwater wells on each parcel throughout the Pajaro Valley. Doing this would require the use of a hydrologic model of the groundwater basin, which is beyond the scope of this paper. Further, seawater intrusion is highly influenced by demand for groundwater resources, so the decomposition of changes in salinity to either sea-level rise or to groundwater overdraft is challenging. Climate change is predicted to cause higher temperatures and more variable precipitation, which may lead to increasing demand for groundwater, in addition to a rise in sea levels . Instead, we opt to model a realistic increase in TDS by looking at how much groundwater salinity has increased in the basin over our sample period. The average spring TDS during our sample period is 644 mg/L. From 1990-2020, four years had an average TDS greater than double this value, with an average increase of 10% annually. For a relatively straightforward simulation, we increase the current-period TDS by 100%, to see how crop choice would evolve with this plausible shift in salinity. All other variables, including weather, remain stable for this analysis, which allows us to focus on how salinity specifically impacts cropchoice distributions. The estimated change in the distribution of crop choices is plotted in Figure 2.8. The graph plots the difference between the original model estimates of crop choice and the simulated estimates of crop choice under an increase in TDS conditions of 100%. Caneberries experience the largest shift in parcels planted under the 100% increase, which coincides with their sensitivity to salinity, as well as their relatively lower profitability when compared to strawberries, another salt-sensitive crop. Vegetables experience the largest increase in the probability of being planted, as they are the least salt-sensitive. Fraction of land left idled also increases substantially under the high TDS scenario.Seawater intrusion, which occurs when saline water from the ocean enters a freshwater aquifer, can manifest from two primary drivers: groundwater extraction and sea-level rise. Pumping groundwater faster than the natural rate of recharge can move seawater to freshwater zones, and sea-level rise alters where saltwater sits relative to freshwater in the aquifer .

Salinity is a major concern for coastal agricultural production that is dependent on groundwater for its water supply , but it can also have significant impacts on inland irrigated agriculture as salts accumulate in the soil over time. Increased salinity levels in agricultural water lead to declines in agricultural productivity, and farmers are left with few mitigation strategies. In this paper, we empirically evaluate the likelihood that farmers switch crops in response to changing groundwater salinity, with an application to the Pajaro Valley, a coastal region in California. Unique spatial panel data on groundwater quality and land use spanning 11 years lends itself to a panel mixed logit model of crop choice. This revealed preference approach allows us to estimate the marginal willingness-to-pay for improvements in irrigation water salinity. We find evidence that growers are more likely to shift away from crops that are saltsensitive, such as strawberries and caneberries, relative to fallow ground, when facing an increase in groundwater salinity. The marginal WTP to pay for a 10 mg/L reduction in TDS varies by crop, and ranges from $1,613 for strawberry growers to $16,369 for caneberry growers. Our simulation of a 100% increase in TDS across the basin speaks to potential land use changes in the basin if salinity trends continue. We estimate that a change in TDS of this magnitude, which could realistically occur in a future defined by sea-level rise and continued groundwater overdraft, would result in a welfare reduction of $140 million. While our WTP and marginal damages estimates are restricted to an agricultural region defined by the jurisdiction of a single water management agency, most groundwater management decisions and investments are made at this scale. The paucity of robust, geospatial groundwater quality data that can be paired with accurate planting information precludes us from deriving estimates in other regions. While salt sensitivity is crop specific, which will drive differences in the WTP across regions, this methodology is generalizable and can be applied elsewhere to determine the benefits of reducing groundwater salinity. Salinity is becoming an increasingly common issue across the United States and the globe as sea levels rise and groundwater aquifers become more stressed under climate change. Estimating marginal damages from changing salinity can provide new context for the cost of climate change and the cost of groundwater overdraft, both of which are broadly important for groundwater management.The stability of water resources for agricultural production has always been an important topic, but the scale and urgency of the issue has dramatically increased in recent decades.