Why Some California Tax Measures Win and Others Fail

Every election cycle, numerous cities, school districts, and other local government entities propose ballot measures seeking to increase taxes, introduce new taxes, or extend existing taxes.

Using The Ballot Book's comprehensive election database, we've analyzed the performance of local tax measures across California to reveal key patterns that influence voter approval. Our detailed data allows us to uncover the nuanced elements that truly drive tax measure success or failure.

Several factors influence the success of these tax proposals. One of the most significant factors is often the phrasing of the ballot question, along with the specific type of tax being proposed.

The California Elections Data Archive (CEDA) is a collaborative initiative involving the Center for California Studies, the Institute for Social Research (ISR) at California State University, Sacramento, and the California Secretary of State. It offers comprehensive data on local tax measures spanning nearly three decades.

We conducted a thorough analysis of the data to enhance our understanding of the performance of various tax measures. This assessment considered factors such as the rhetoric employed in the ballot question, the type of tax proposal (e.g., business tax, transient occupancy tax, sales tax), and the nature of the tax (whether it is a new tax, an extension, or a tax initially set to expire but made permanent).

Of the 1,746 local tax measures analyzed between 2012 and 2023, 73.1% were approved (1,276 measures) and 26.9% failed (470 measures). However, this does not necessarily mean that all passed measures received more than 50% of the vote—different types of taxes require different thresholds for passage, such as 55% or two-thirds, depending on the jurisdiction and purpose of the measure.

This type of nuanced analysis exemplifies how The Ballot Book helps political professionals move beyond surface-level assumptions to uncover the actionable insights that drive more effective campaign strategies. By combining historical voting patterns with detailed ballot language analysis, we can identify the specific factors that resonate with California voters.

For the analysis of ballot language, we leveraged advanced natural language processing via the OpenAI API to analyze and categorize the tax measures. This approach allowed us to systematically evaluate hundreds of differently worded ballot questions using consistent criteria—something that would be prohibitively time-consuming and potentially inconsistent if done manually. This technology helped us objectively classify each measure into the following categories:

• "new tax": A tax that has never existed before in this jurisdiction
• "rate increase": An explicit increase in the rate/amount of an existing tax (e.g., raising from 5% to 7%)
• "extension": Continuing an existing tax that would otherwise expire, for a limited additional time period
• "make permanent": Removing the expiration date from a previously temporary tax
•  "scope expansion": Broadening what's covered by an existing tax without changing the rate (e.g., updating definitions to include new technologies)
• "decrease": Explicitly lowering a tax rate
• "repeal": Eliminating an existing tax

We also used the API to categorize the rhetoric used in each tax proposal into one of the following categories:

• "prevent cuts": Text explicitly mentions preventing cuts, losses, or reductions to services (e.g., "prevent the reduction in maintenance")
• "maintain": Text emphasizes keeping or continuing current service levels (e.g., "maintain neighborhood crime patrols", "continue funding", "preserve")
• "improve": Text emphasizes enhancing, expanding, or improving services (e.g., "improve earthquake preparedness", "enhance library services")
• "neutral": Tax purpose is stated without emotional language or enhancement/maintenance framing

Here is the breakdown in how each tax type performed (note that in order to provide a more accurate understanding, this data removed "repeal" and "decrease" as categories):

Here is the breakdown in how each tax classification performed:

And here is the breakdown of the framing within the ballot language (similar to the first table, this table also filters out measures that either repealed or reduced a tax):

Based on these figures, the type of tax most likely to pass would be a property tax extension framed with language that talked about maintaining services. (This combination averaged a yes vote of 73.4% across 172 measures)

Conversely, the type of tax with the least chance of passing would be a utility tax extension that was either a tax increase or new tax with a neutral framing. 

For political professionals developing ballot measure campaigns, these insights provide a data-driven foundation for crafting more effective strategies. The Ballot Book's comprehensive database enables this level of detailed analysis across every jurisdiction in California, helping campaigns make smarter decisions about everything from ballot language to messaging strategy.