How Government Shutdowns Grow Government

Following Marc’s great post on congressional dysfunction, I’d like to point how political science tells us that the availability of government shutdowns actually causes the growth of government spending. The analysis follows the 1979 spatial analysis of zero-based budgeting by Thomas Romer and Howard Rosenthal.

Suppose that there is one dimension of politics: the size of the federal budget. There are a fiscally conservative party and a fiscally liberal party. For simplicity, assume the median, electorally decisive American voter is somewhere between the two. We could plot the parties’ and median voters’ positions on this dimension like this, where “C” is the conservative party, “M” is the median voter, and “L” is the liberal party:

Federal budget

 0|----------------------------------|--------|-----------|------------------------------------------| 100% of GDP
                                    C        M           L

Now suppose that there is a need to pass a budget. If the budget doesn’t pass, the government partially shuts down (S). Once the government shuts down, the median voter M perceives the outcome as being more favored by the conservative party, with ideal point C. The liberal party with ideal point L can make a budget proposal that must get approval from both parties, so conservatives have the opportunity to accept or reject it – in the latter case, the government stays shut down. After the budget is approved or rejected, there is an election, and the median voter M votes for the party with the closer budget position. Parties care most about winning election, then secondarily obtaining their preferred budget.

In this example below, once the conservative party gets associated with S, causing the shutdown, then L is able to propose its ideal point (L). Conservatives accept the budget, because otherwise they would remain associated with S, and the median voter prefers L to S, so would turn conservatives out at the next election.

 0|----------|-----------------------|--------|-----------|------------------------------------------| 100% of GDP
            S                       C        M           L

The median voter will only be willing to vote for conservatives who reject a liberal budget proposal if S is closer to their ideal point than the liberal budget proposal. Knowing this, L will propose something close enough to the median voter to prevent that outcome – and conservatives will accept it. Take the following example, where P is the proposal liberals make:

 0|----------|----|-----|----------|----------------------|------------------------------------------| 100% of GDP
            S    C     M          P                      L

P is infinitesimally closer to M than S is, so M votes for the liberal party, unless the conservatives also vote for the budget.

So once a shutdown happens, a bigger budget than the median voter prefer (let alone the conservative party) looks inevitable. Knowing this, conservatives won’t want the government to shut down to begin with. But that still means liberals have a lot of bargaining power, and the budget will tend to grow.

In real life, of course, shutdowns happen very occasionally. Why? Well, we haven’t modeled primary politics. The conservative party may favor a shutdown when primary voters are conservative and non-strategic, and so will punish conservative legislators who do not hold the line on the budget.

Now, imagine that instead of the government’s partially shutting down when there’s no budget, the reversion outcome is simply that the budget from last year remains in place – something like “A” on the graph below. Then the liberals will need to propose something like “P/” instead: the budget is smaller!

 0|----------|----|----|-|-|---------|----------------------|------------------------------------------| 100% of GDP
            S    C    A M P/        P                      L

The concept from bargaining theory doing the work here is “reversion outcome.” Power in a bargaining situation goes to the actor with the better reversion outcome. Government shutdowns are a worse reversion outcome for conservatives electorally than are maintenance budgets. Even though they’re a worse reversion outcome ideologically for liberals, electoral benefit usually trumps ideological benefit for elected professional politicians.

All this implies that if congressional Republicans were smart, they would push to pass a law requiring a maintenance budget rather than a government shutdown in the event of a budget impasse, as is the case in almost every other country. Or maybe I should say: “if Republican primary voters were smart, Republican congresspeople would do this.”