March 20, 2011
The Republican proposals to slash the budget seem to work from the premise that if we fire government employees that we will induce private employers to hire more workers. This runs directly opposite to the idea behind the stimulus, that if the government stimulated demand by spending money and/or cutting taxes it would create more jobs. Interestingly, there is new research that indicates that the stimulus did raise employment. In fact, it seems that its effect was even larger than the Obama administration had predicted.
But the Republicans seem uninterested in these research findings. They instead claim that the best way to create private sector jobs is by having the government fire workers and spend less.
It is difficult to follow the logic of this view. If we think of a cross section of employers – hospitals, construction companies, car factories, retail stores and restaurants – which ones on this list do we think will hire more workers after a big round of federal budget cuts and layoffs?
Do we think that hospitals will suddenly rush out and hire more nurses and doctors because because of the National Institutes of Health is cutting funding for cancer research? Will Wal-Mart expand its sales staff because the government is laying off people from Head Start? These stories don’t seem very plausible.
Undoubtedly some of the government employees losing their jobs will be experienced and highly educated workers who private employers will be anxious to hire, but this will generally be for positions that would have existed in any case. These former government employees will simply be displacing other workers who would have held these jobs; there will not be new jobs in the private sector created for them.
Of course there will be times in which government spending does impose a constraint on the private sector in the sense of pulling away resources from the private sector. This is measured by the interest rate. High interest rates will discourage private sector investment and consumption. Reducing spending in the government sector can reduce demand in the economy, which would lead to lower interest rates. Lower interest rates would then spur private sector spending and job creation.
However, it is very hard to tell this story right now. Interest rates have risen from the trough of the downturn, but with the 10-year Treasury bond rate at 3.4 percent, they are still at very low levels. It is difficult to believe that cutbacks in government spending will do much to lower the interest rate from its current level, nor that plausible declines in the interest rate will have much impact on demand. At a time when the economy still has enormous over-capacity in most sectors, firms are unlikely to increase their investment by much even if the interest rate were to decline somewhat. Heavily indebted consumers are also unlikely to boost their spending.
In short, if the Republicans really expect cutbacks in federal spending and employment to boost spending and employment in the private sector, they certainly have not explained how they think this will happen. If the Republicans really believe that lower public sector employment leads to higher private sector employment, it would be nice if they could produce some evidence to support this position.
We tried a simple test of the relationship between changes in government employment at the state and local level during the downturn and changes in private sector employment. The relationship seems to go the wrong way for the Republican’s story. Changes in public sector employment are highly correlated with changed in private sector employment as shown in the Table below.
| Variable | Model 1 | Model 2 | Model 3 | Model 4 |
|
Ch gov emp |
.54415 | .62617 | .71509 | 1.0645 |
| 1.65 | 2 | 2.31 | 3.18 | |
| 0.1073 | .0522 | .0257 | .0026 | |
| ch_HPI | .16201 | .15086 | .15145 | .085005 |
| 3.43 | 3.31 | 3.29 | 1.77 | |
| .0014 | .0019 | .002 | .084 | |
| Def 09 | 1.5308 | 1.9491 | ||
| 1.07 | 1.37 | |||
| .2899 | .1178 | |||
| shmanuf 08 | -.4313 | -.39588 | -.39985 | |
| -3.34 | -3.34 | -3.34 | ||
| .0018 | .0017 | .0017 | ||
| shconstr_08 | -.63671 | -1.2439 | -1.2417 | |
| -1.62 | -3.69 | -3.64 | ||
| .1134 | 6.2e-04 | 6.9e-04 | ||
| south | -.013242 | |||
| -1.27 | ||||
| .211 | ||||
| midwest | -.0008343 | |||
| -.0778 | ||||
| .9384 | ||||
| west | -.030792 | |||
| -2.66 | ||||
| .0112 | ||||
| _cons | .033758 | .052847 | .049224 | -.055985 |
| 1.25 | 1.94 | 1.8 | -14.1 | |
| .2199 | .0592 | .0793 | 1.8e-18 | |
| R2 | .637 | .565 | .547 | .396 |
The table shows four different tests of the relationship between public sector employment and private sector employment during the downturn. The dependent variable is the percentage change in private sector employment by state from the fourth quarter of 2007 to the fourth quarter of 2010 in each state. The first independent variable is the percentage change in government employment over the same period. If it were the case that public sector employment crowds out private sector employment, then we would want to see a negative coefficient for this variable. Instead we see a positive and mostly highly significant coefficient in these tests. This implies that more government employment is associated with more private sector employment.
Of course there are other factors that make it impossible to get a simple test of the relationship. It could be the case that when the economy is doing well we hire more government employees and when things are going badly we lay them off. The other variables are included to try to control for this effect.
The next variable on the list is the percentage change in the House Price Index from the second quarter of 2006 to the fourth quarter of 2008. Since this index is based on closings rather than contracts, the data for the fourth quarter of 2008 would have primarily reflected contracts signed between July and October of that year. This means it would be little influence by the steep drop in employment that began in October of 2008, minimizing the effect of reverse causation from job loss to house prices.
This coefficient is positive and highly significant in three of the four specifications shown here suggesting, unsurprisingly, that the collapse of the housing bubble was a major determinant of job loss in each state. Other research has shown that household debt was a major determinant of job loss. This skips the causal role of the housing bubble. People took on more debt in some states than others because the run-up in house prices allowed them to borrow more against their house.
The deficit variable for 2009 was an effect to control for the extent to which states’ economies were doing well or poorly. It was insignificant in the two regressions in which it is included. It was also insignificant in a number of other specifications not shown. This is likely in part due to inconsistencies in budget accounting across states.
The next variables are the share of employment in manufacturing and construction for each state in the fourth quarter of 2008. The reason for including these variables is that these sectors were proportionately the hardest hit by the downturn. The coefficients are negative and mostly highly significant, as would be expected. In other words, the states that had the largest share of their workers in construction and manufacturing at the start of the downturn saw the largest declines in private sector employment.
The last set of variables is regional dummies that are testing for differential impacts of the downturn by region. There is little clear difference between regions, so these variables are dropped in the other three regressions.
This is admittedly an imperfect test of the relationship between public and private sector employment, but it certainly does not support the view that if we lay off public sector workers the private sector will increase employment. In fact, it suggests that if we fire a lot of public sector workers then we should expect to see the private sector also laying off workers, magnifying the impact.
There are other efforts that more systematically examine the relationship between government spending at the state level and employment (e.g. here and here. These studies also find that more spending is associated with more employment, supporting the view that if the government at the federal or state level cuts back spending and/or public employment, we will see reductions in private sector employment, not increased private sector employment.
It would be good if the Congressional leadership could find some evidence to support their claim that reducing government employment/spending would somehow increase private sector employment before they embark on a deficit cutting extravaganza. The evidence that we have thus far indicates that eliminating government jobs will directly put people out of work and that the private sector will amplify the effect by eliminating even more jobs. That might make sense to the Washington policy crew, but that is not likely to sit well with the workers across the country who find themselves on the unemployment lines.
