A Stalemate-Proof Mechanism to Set the Budget

By Alex Bäcker, Ph.D.

After being more than 85 days late for the last budget, the California Legislature and the Gubernator are once again at a stalemate in discussions to approve a budget. This threatens to bring the state government to a screeching halt, and prompts the question, doesn't California deserve a mechanism to set the budget that isn't subject to stalemates?

In California, the magnitude of the budget is set by The California Balanced Budget Act’s requirement that the budget be balanced. All that’s needed, then, is to decide how to spend it. The candidates to receive money are all the programs and offices that have previously been voted into existence by the legislature. The question of how much to apportion to each program or office can best be thought of not as how much to fund any given program or office, but rather which program or office is the best to spend the next dollar on for each of the dollars in the budget. In other words, once a budget is subject to a cap, the process of devising a balanced budget can be broken into the process of figuring out the highest priority for each dollar available.

To decide how best to spend each dollar, the marginal benefit of a dollar invested in each program needs to be known. In the Federal Government, there is an office in charge of computing just that --the Office of Information and Regulatory Affairs. That office might estimate, for example, that the first million dollars spent on healthcare save a life for every $20,000, but that the next million save a life only every $50,000, whilst the first million dollars spent on firefighting programs saves a life every $40,000. Those estimates would dictate that, if legislators decide that saving lives is their first priority, among those two programs, the first million be spent on healthcare and the next on firefighting, and so on until the budget has been exhausted.

While critics ridicule assigning a value to a human life, the point of the process is not to value a life, but rather to decide how best to spend limited resources. Clearly, if your goal is to save lives, and you have limited resources, and one way of spending it saves a thousand lives and the others saves five hundred, the former will usually be preferrable. Not knowing the marginal efficacy of investing an additional dollar in each program is tantamount to making those important decisions blindly.

How do we get a large set of legislators with competing agendas to arrive at one budget? Again, a yes/no voting process is inappropriate, but there are voting mechanisms that fit the bill. The first constraint, as we saw above, is that the budget must add up to the total, so what is being voted for is what fraction of the budget to spend on each program or office. What we seek is a distribution that best matches the legislators' perceived priorities. To this end, each legislator (or a subset in a Budget Committee) could be asked to submit a distribution of the agreed-upon total budget into the programs and offices he sees fit to fund. These distributions could then be combined to arrive at the final outcome, not by consensus (which can be virtually impossible when you ask two thirds of a hundred and twenty legislators and a Governor to agree on each of hundreds of numbers, as we have seen repeatedly over the past year), but by finding out the most agreeable budget among all the ones proposed by legislators. This can be achieved in many ways, but a simple one is ask each to look for the budget whose most extreme allocation is least extreme. In other words, you take each line item in the budget one at a time and rank each legislator by how extreme their proposed allocation of the budget to that line item is. A useful concept here is the median of a distribution of numbers. The median is the number that has as many above it as below it: for example, the median of {2,4,9} is 4. The beauty of the median is that the actual number gets set by the most moderate legislators, limiting the influence of extremists on either side simply to defining what a moderate is. Except for the most moderate of all, each legislator’s influence is limited to whether they vote for more or less than the median. The exact value of their vote has no influence. A legislator who proposed the median allocation to a line item, i.e. the one who had as many legislators proposing more as legislators proposing less, would get an 'extremism' score of zero for that line item. Conversely, the legislators proposing the largest and smallest allocations to that line item would get the highest extremism score. This process is repeated for each line item, and each legislator receives a 'total extremism' score equal to the maximum extremism score he got across line items. The budget proposed by the legislator with the lowest total extremism score is adopted. This process will arrive at the "least objectionable" budget. Alternatively, the process above can be used to discover the two (or more) least objectionable budgets, which can then be put up for a vote where each legislator must select one of the two, with the budget receiving the greatest number of votes adopted.

The budget thus set would be much less likely to be objectionable to the Governor (or the President, if discussing the Federal budget). A similar process could be used to settle disputes between the Legislature and the Executive branches of government.

Perhaps it is time to complement the classic wisdom instilled in our legal system by the Founding Fathers, schooled in the liberal arts, with some mathematics. 

Dr. Bäcker is Founder and CEO of abInventio, the invention factory behind American Business Awards finalist QLess and Whozat.

Altadena, CA, February 2, 2009 (revised June 25)

References:

1. An alternative proposal with several good and complementary ideas to revise the state budget process, including multi-year planning, mid-year revisions, and accountability of existing programs can be found at californiaforward.com/CA_Forward_Revised.pdf

Read more of Alex Backer's Proposals for Government.