To date, the most aggressive proposal being considered on Capitol Hill would eliminate pre-tax deferrals to 401(k) plans, and essentially shift the largest private sector employer-sponsored retirement savings vehicle to an all Roth 401(k) design.

Another proposal reportedly being considered would split contribution limits to half pre-tax and half after-tax.

The implications of such policies are potentially massive for the 401(k) market, which the Investment Company Institute values at $4.8 trillion, by far the largest swath of the $7.4 trillion defined contribution market.

Though the ICI doesn’t break out the split between traditional and Roth 401(k) assets, the vast majority of savings are clearly in traditional 401(k) plans.

More plans have adopted a Roth 401(k) option over the years. According to the Plan Sponsor Council of America, more than 62 percent of plans offer the Roth option. But only 19 percent of participants made contributions to Roth plans, which are often favored by smaller employers and younger workers with fewer investable assets.

The non-partisan Joint Committee on Taxation, which advises Congress with official research on the tax code, recently released data showing the tax-preferred treatment of defined contribution plans will cost $583.6 billion in so-called foregone revenue between 2016 and 2020. Traditional IRAs will cost $85.8 billion.

Those numbers will clearly be enticing to lawmakers that want to cut tax rates and justify them with revenue offsets.

But exactly how the JCT calculates foregone review is worthy of some scrutiny, say policy analysts at the Bipartisan Policy Center, a think tank that, as its name suggests, produces non-ideological research on legislative proposals.

The JCT uses a cash scoring method to determine what the tax incentives behind 401(k) plans cost the government in revenue. That means the actuaries count funds spent and revenue collected in the tax code, according to a BPC post published back in 2015 as part of the Center’s Commission on Retirement Security and Personal Savings.

Cash scoring presents a skewed understanding of foregone revenue, notes the BPC. The JCT considers the effect of 401(k) tax treatment over the course of several years. But savings in 401(k)s often occur over three and four decades. Withdrawals are ultimately taxed as regular income. Tax revenue on 401(k) contributions isn’t “foregone,” but rather it is delayed far outside the JCT’s near-term assessments.

A better way of understanding the tax benefits of traditional 401(k)s, for both savers and federal revenues, could be to use a net-present value method to score the cost of pre-tax deferrals.

“Under a five, or ten-year cash-based budget estimate, the cost to the federal budget of deferred taxation of a traditional retirement account looks deceptively large because it includes the entire current revenue loss from near-term contributions, but it ignores a significant portion of future revenue gains from withdrawals, much of which will occur more than five or ten years later,” BPC s says in its post.

A net-present value score, while not perfect according to the BPC, considers the immediate lost revenue on pre-tax contributions, as well as trying to account for the revenue captured when it is ultimately taxed down the road.

BPC offers a simple example of how the JCT’s cash scoring method may not be so sound of a practice for measuring the tax implications for traditional and Roth 401(k)s. Say a $50 contribution to a traditional 401(k) grows to $140 over 30 years. If taxed as ordinary income at 25 percent, the government earns $35, and the saver gets $105.

At the same tax rate, $50 would translate to a $37 contribution to a Roth 401(k). After 30 years, under the same assumptions, it would grow to $120, which the saver would not have to pay taxes on.

Under a cash score, the government captures more near-term revenue, but loses considerably more in the long-term.

Scoring the tax treatment on retirement accounts with a net-present value method is complicated, BPC says. It must factor assumptions on long-term interest rates, investment returns, the timing of account drawdowns, and the potential difference in marginal tax rates between when deferrals are made and when assets are drawn down in retirement.

A NPV score must also account for the cost of deficit spending on revenue that is not captured in the near-term.

Notwithstanding those complexities, BPC’s team is certain of one thing: “we know that the existing cash scores of retirement tax benefits misrepresent the true cost to the federal government.”

The JCT and Congressional Budget Office could benefit from more research on 401(k) savings vehicles, and their tax implications, in applying more reasonable assumptions to federal revenues, says BPC.

For those advocating to preserve the existing treatment of 401(k) plans, the question is whether that new research will be generated quickly enough to influence lawmakers.