In a recent Economix post, Casey Mulligan asserts that aid to the states and localities is unwarranted given that state and local government employment is doing just fine. His graph highlighting cumulative gains/losses ends in January 2009, to show what had transpired by the time the stimulus bill was being debated. How do things look if one extends the sample to August 2009? And what about spending as opposed to employment?
Figure 1 depicts the level state and local employment. The two dashed lines indicate the reference dates in Professor Mulligan’s comparison.
Figure 1: State and local government employment, in thousands, seasonally adjusted. Dashed lines refer to begin-end dates used in Mulligan’s graph. Gray shading denotes NBER defined recession dates assuming trough at June 2009. Source: BLS, Employment Situation August release and NBER.
Using those two reference dates, state and local employment has indeed increased. And in fact, the level of employment in August is still above that in December 2007 (but only by 43,000 higher, instead of the 110,000 cited by Professor Mulligan for July — I think I see a trend). However, I think gradients are of even more interest, and looking at y/y and q/q annualized growth rates, one sees that currently employment is falling. This is highlighted in Figure 2.
Figure 2: Year-on-year growth rate in state and local government employment (blue), and quarter-on-quarter annualized growth rate (red), calculated as log differences. Dashed lines refer to begin-end dates used in Mulligan’s graph. Gray shading denotes NBER defined recession dates assuming trough at June 2009. Source: BLS, Employment Situation August release, NBER, and author’s calculations.
So, I’m willing to say public policy types were able to see the handwriting on the wall, note that in the future that state and local tax revenues would decline, and in the absence of countervailing action, would have to cut employment deeply.
Moving from the statistics, there are some odd arguments in the post. After observing the relative losses in the private vs. state/local sectors, he concludes:
For these reasons, an effective stimulus law would have allocated state and local government something from 4 percent (its share of layoffs) to 14 percent (its share of employment) of its funds.
I’ll just make two observations: (1) we have government employment partly because of externalities/public goods arguments (the private sector underprovides police officers, fire fighters, etc.), and (2) not all the funds given to the states is used to pay government workers. In fact, as noted by a recent GAO report some had been used to undertake road repairs, which entailed hiring private construction contractors (Professor Mulligan has stressed underutilized resources in construction in the past).
As opposed to looking at employment, one could look at what government spending at the state and local level was doing just before the stimulus bill kicked in.
Figure 3: Quarter on quarter annualized growth rate of state and local government spending (blue, left axis), and level, Ch.2005$ SAAR (red, right axis). NBER defined recession dates shaded gray, assuming trough at 2009Q2. Source: BEA, 2009Q2 second release, and author’s calculations.
It’s clear that state and local real spending on goods and services was tanking in 2008Q4 and 2009Q1. It picked up in 2009Q2; most reasonable people agree that in the absence of the transfers to the states, spending would have been lower.