Too many discussions of changing government start with cutting jobs or adding jobs, eliminating departments or creating new programs--it's always "more or less." Here's the question we should be asking: What problems are we trying to solve and what tools of government will solve them most effectively and efficiently and with the fewest unintended consequences?

"Tools of government," you ask? Let me introduce you to Lester Salamon's transformative 688-page volume "The Tools of Government: a Guide to the New Governance," which remains as freshly illuminating as it was when it was published a decade ago. Once you begin to see government activities from this perspective, you'll never again read or hear about government without thinking about the tools of government. (By way of disclosure, in 2011 and 2012 I taught a section of Prof. Salamon's "Tools of Government" course at Johns Hopkins University.)

Salamon lists 14 types of tools: Direct government (the military, security guards at airports); government corporations and government-sponsored enterprises (the Postal Service); economic regulation; social regulation; government insurance; public information; corrective taxes, charges and tradable permits; contracting for things (from pencils and computers to highways and bombers); contracting for services; grants; loans and loan guarantees; tax expenditures; vouchers; and tort liability.

To take just one example, in fiscal year 1999 the federal government "spent" $332.8 billion on housing, using an array of these tools: subsidies ($21.1 billion), direct loans ($1.1 billion), mortgage credits ($400 million), public housing ($6 billion), rural housing ($600 million), veterans' housing ($1.6 billion), loan guarantees ($187.6 billion) and tax expenditures ($114.4 billion). In other words, over 90 percent of the federal government's intervention in the housing market that year came in the form of loan guarantees and tax expenditures, not in directly providing subsidized housing. We can talk about the unintended consequences of the loan guarantees, but we can't blame the Great Recession on public housing.

Salamon "grades" each tool in terms of its effectiveness, efficiency, equity, manageability and political legitimacy. Contracting, for example, is only effective where a competitive market exists. Regulation may be efficient from government's perspective, but it imposes substantial compliance costs on industry. Many activities simply are not manageable by government: I lived on Capitol Hill during President Nixon's Phase Four wage and price controls. A local Greek restaurant couldn't raise the price of its souvlaki, but the portions got substantially smaller. No government has the resources to monitor portion sizes.

And political legitimacy has become a larger issue in past decades. Before the attacks of 9/11, for example, the federal government never seriously considered consolidating the many existing airport-security operations--there was no perceived need for a federal presence. But with the external threat posed by al-Qaida, the Transportation Security Administration (TSA) was created in just two months.

Salamon also delineates four "key tool dimensions": degree of coerciveness, directness, "automaticity" and visibility. Coerciveness? Just try to walk by the TSA guards at an airport or buy fireworks in New York City. School and housing vouchers are low in directness, while (again) the TSA is very direct. "Automaticity" refers to the degree of government involvement in the operation of the tool. Tort liability is highly automatic: the job is handed over to the courts. Direct education loans and regulation are hardly automatic at all: Someone has to issue the loans, and it takes lots of "someones" to write, edit, review, defend and implement the regulations. Visibility is in the eye of the beholder. As noted above, visible housing aid goes for the poor and veterans, while the mortgage deduction for the rest of us is a tax expenditure and therefore almost invisible.

I'd like to see legislative bodies look at a variety of tools--and how they have been used in other jurisdictions--before matching a problem with a tool. I'd like to see those deliberations made more public. And as I've argued elsewhere, I'd like to see "legislative impact statements" made available for public comment and crowdsourcing before enactment. That's where open government should start.

Look at one example of legislation from the tools framework: tobacco vs. illegal drugs. We'd probably agree that both are harmful. But which is the greater problem? Tobacco accounts for one in five American deaths each year (more than 440,000) and the health care costs are about $96 billion a year. The health related costs of illegal drugs were $15.8 billion in 2002. But we use the strongest tool, direct government (police, border agents, even military action) against illegal drugs. At the federal level, we use the weakest tool, public information, against tobacco, and there are even court fights about how graphic to make the warning labels. (State and local governments do impose some taxes on tobacco, but hardly enough to equal the $200 billion in costs.) From the tools perspective, this is exactly backwards.

No one would try to fix a leaky faucet or change a flat tire without picking up the right tools. So why do we try to solve our common problems without using the best tools of government?