The emergence of the so-called “gig economy” – employment situations in which a worker may spread his or her labor among several part-time or freelance arrangements with several different employers – has engendered quite a bit of commentary from policy experts and, increasingly, from politicians.

A November 2015 policy brief from Ian Adams of the R Street Institute made the case that the existing system of employee benefits is ill-suited to these more flexible arrangements. Adams’ brief was part of a broader effort by a coalition of disparate groups – including the Aspen Institute, the American Action Forum, the Freelancers Union and numerous distinguished academics – who together signed an open letter calling for employee benefits to be made more flexible. Signatories called for a thorough restructuring of existing law, regulations and tax provisions to allow benefits to be portable between companies.

Our current system, the signatories argued, largely preserves the assumptions of 50 years ago: that people will work many years for the same company, and that it is that employer who generally will be responsible to see to their retirement needs. While pensions generally have already transitioned from defined benefit to defined contribution, it remains the case that most retirement saving is done through an employer retirement plan, even though there is nothing about saving money that necessitates an employer’s involvement. If this system ever were appropriate, it has become less so over time, as jobs have become much less secure and employees move rapidly from company to company.

In that vein, it is worth considering how a new retirement-savings system – one sufficiently flexible and portable for modern workers – might be constructed. But before we can hope to design that structure, we must first determine what features such a system would need.

Some high-level principles for reform are obvious. First, we should strive to build a system that is easy to use. Employees should be able to understand their benefits and manage them as smoothly as possible. At the moment, many employees do not take full advantage of their retirement benefits. This may be partly because they don’t understand their value, or because they are afraid of the complexity and potential for tax penalties if they make a mistake.

Second, compliance costs and fees should be as low as possible, for both employees and employers. Under the current system, employers who wish to maintain a full-featured benefits plan, such as a 401(k), incur such costs as per-participant fees, fidelity bonds and administrative fees. Less costly options are available, such as SIMPLE IRA plans, but these limit the ability of companies to offer 401(k) matching or profit-sharing bonuses. Far better would be if total costs were sufficiently low that neither companies nor employees had to compromise.

Third, we should remove unnecessary restrictions on the kinds of investments that can be made and the kinds of benefits that can be provided. For example, many companies provide temporary term-life insurance for their employees. But some employees might find permanent whole-life insurance more suitable for their needs, even if they have to pay the difference in cost themselves. Similarly, many 401(k) plans offer only a short list of mutual funds, with no way for participants to invest elsewhere. Even plans with the richest features offer few ways to invest in local businesses, for example. An ideal benefits system would be as flexible as possible. It would allow employees to design their financial lives the way they want and to take advantage of new financial opportunities that are impossible for policymakers to imagine today.

However, in designing a new system, we should be careful not to lose any of the key capabilities of the old one. Employee-benefits plans have developed other functions beyond simply providing insurance and encouraging retirement savings. Attempting to replace the current benefits system without fully understanding what it does could lead to tremendous disruption of people’s financial lives. We begin with a short inventory of some nonobvious uses of the current employee-benefits system.

First, 401(k) plans provide an important advantage over Individual Retirement Accounts: the ability to borrow from one’s own savings. One might argue that borrowing from your retirement savings defeats the point, and that IRAs already allow withdrawals for qualified purposes, such as buying a first home. But most people will face severe financial setbacks long before they reach retirement age, when they gain unrestricted access to their funds. If they have channeled much of their savings into tax-advantaged vehicles that cannot be accessed, people will be forced to borrow externally (often at high interest) to make up for any cash-flow crunch. Those with 401(k) accounts often find the ability to borrow for short periods to be a godsend.

Second, employers are encouraged by the current system to reward their lower-paid employees for saving for retirement. Employers often provide contribution matching, or even profit-sharing grants, to their employees. This is because so-called Actual Deferral Percentage and Actual Contribution Percentage (ADP/ACP) tests, as well as “top-heavy” tests, limit the amount of money that highly compensated employees can invest in their 401(k) accounts unless total participation is sufficiently high. The interests of highly compensated employees are thus served by providing additional retirement savings for lesser-compensated employees.

Third, employers can use vesting schedules as rewards and punishments for their employees, discouraging them from leaving the company by offering deferred compensation if they stay long enough. This benefits employers by reducing employee turnover and the considerable costs associated with it. Additionally, having a vesting schedule makes employers more willing to provide employee matching and profit sharing in the first place, since they can protect themselves from paying “extra” to a bad employee.

Fourth, Employee Stock-Option Plans (ESOPs) allow employees to purchase the stock of their company at lower cost. This can provide a nonobvious benefit to company founders — offering a way to sell one’s stake gradually, at favorable terms, without incurring a sudden tax liability and without disrupting company management by selling to an outsider. One can also exploit the ESOP structure to invest 401(k) savings in one’s own small business, by rolling assets into your new company’s retirement plan and using those assets to purchase company stock.

Once the inventory of uses is complete and we fully understand the requirements needed for the new portable-benefits system, we can make general recommendations for its design that are both compatible with the high-level principles noted above and that preserve key functions of the current system. Done correctly, a redesigned system can provide new benefits to both employees and employers without sacrificing any of the key features they enjoy today.