California recently passed the New Motor Voter Act, a law designed to register eligible residents to vote by default when they use the Department of Motor Vehicles (DMV), unless they decline. Other states have or are considering similar laws. But because of California’s diversity and size — the 2016 population was 39.2 million and climbing — the Golden State’s law garnered special interest when it passed last fall.

In a new report, we look at the law’s likely effect on the demographics of California’s electorate, and at the number of new potential voters it might register in its first year. We find that supporters are right to see great promise in the law, but how the law is implemented will be far more important than many have suggested.

The new law could dramatically change California’s electorate. Emphasis on “could.”

First, some background. California’s New Motor Voter program isn’t universal. It applies only to those who use the DMV to get a new driver’s license or ID card, or renew or change an address on an existing driver’s license or ID card. And it applies only to customers who use the DMV after the law goes into effect — currently slated for July 2017.

But it’s pretty close to universal. Almost every California resident who is eligible to vote eventually gets either a driver’s license or an ID card. And many, many Californians use the DMV for one of those tasks each year. Between July 2014 and July 2015, that included about 10 million people, including roughly 2.6 million unregistered but eligible potential voters.

If most unregistered Californians eventually end up on the rolls, it would significantly change the electorate’s demographics.

To gauge how much, we used the current enrollment rates from a similar program in Oregon — where roughly 93 percent of eligible DMV customers have been registered since the beginning of this year — and registration estimates from the Current Population Survey of the U.S. Census Bureau. That’s very optimistic. California’s uptake rate is unlikely to be as high as Oregon’s for reasons we will get into below. But it’s useful for setting high-end expectations.

The chart below shows that currently underrepresented groups would see a jump in their share of the electorate. (Note that these groups are not mutually exclusive — for instance, many non-college graduates also make less than $20,000 per year.)

For instance, the share of Latinos in the electorate would grow by four percentage points. For Asian Americans, the increase would be 1.7 points. Those increases are almost as large as the growth of these two groups in the California electorate between 2006 and 2014.

As a result, California’s electorate would better represent the state’s population. It would dramatically reduce or even eliminate the gap between each group’s share of registered voters and their share of the entire adult population (including noncitizens, though of course those noncitizens are not eligible to register or vote).

But exactly how the law is put into effect will make a big difference.

Despite these promising numbers, they’re just a possibility. How much the electorate actually changes will depend on how the law is implemented.

Many assume California’s New Motor Voter system is “opt out” (customers are registered unless they actively decline) as opposed to “opt in” (customers must actively agree to be registered). But, in fact, customers are automatically registered only after they attest they are eligible to vote. If they don’t answer when asked whether they are eligible, they won’t be registered.

It’s quite different in Oregon. There, the state determines whether a DMV user is eligible to vote and then sends those customers an opt-out card in the mail. Unless those who receive the card fill it out and return it within 21 days, they are added to the rolls by default.

That’s critical. A large academic literature — presented succinctly in Richard Thaler’s and Cass Sunstein’s popular book Nudge — shows the simple change from “opt in” to “opt out” can dramatically boost a program’s participation. The design of California’s program, which is opt in until the eligibility question is answered, could dampen its effect.

Again, the key word is “could.” To help ensure success, California should require customers to answer the eligibility question to complete their DMV transaction. (There’s still time to do this before the new system goes live next summer.) This would make California’s program effectively opt out, though maybe not as much as in Oregon.

Of course, being registered doesn’t mean someone will vote — and demography doesn’t determine results.

To be sure, the law doesn’t guarantee that new voters will actually cast ballots. It just removes one barrier. Many newly registered voters will be disengaged from the political system, having never voted or been contacted by any candidate or campaign. To get these new members of the electorate — especially those from historically underrepresented groups — to pull the lever will require aggressive outreach and education.

Many observers wonder whether this program will shift the balance of power between Democrats and Republicans. We think this question misses the point. A more inclusive electorate is by definition more democratic, regardless of outcome. But for what it’s worth, predicting partisan results has misfired pretty badly in the past. Research (gated) suggests that even universal turnout would leave the vast majority of partisan outcomes unchanged.

In sum, the law’s potential is great but highly dependent on the way it is implemented. While the new motor voter system is an exciting development in political reform, more work and important decisions remain before we can fully determine its effect.

Eric McGhee is a research fellow at the Public Policy Institute of California, where he focuses on elections and political reform. 

Mindy Romero is founder and director of the California Civic Engagement Project (CCEP) at the UC Davis Center for Regional Change.