Why didn’t the stimulus create more jobs?

There are many studies of the stimulus, but finally there is one which goes behind the numbers to see what really happened.  And it’s not an entirely pretty story.  My colleagues Garett Jones and Daniel Rothschild conducted extensive field research (interviewing 85 organizations receiving stimulus funds, in five regions), asking simple questions such as whether the hired project workers already had had jobs.  There are lots of relevant details in the paper but here is one punchline:

…hiring people from unemployment was more the exception than the rule in our interviews.

In a related paper by the same authors (read them both), here is more:

Hiring isn’t the same as net job creation. In our survey, just 42.1 percent of the workers hired at ARRA-receiving organizations after January 31, 2009, were unemployed at the time they were hired (Appendix C). More were hired directly from other organizations (47.3 percent of post-ARRA workers), while a handful came from school (6.5%) or from outside the labor force (4.1%)(Figure 2).

One major problem with ARRA was not the crowding out of financial capital but rather the crowding out of labor.  In the first paper there is also a discussion of how the stimulus job numbers were generated, how unreliable they are, and how stimulus recipients sometimes had an incentive to claim job creation where none was present.  Many of the created jobs involved hiring people back from retirement.  You can tell a story about how hiring the already employed opened up other jobs for the unemployed, but it’s just that — a story.  I don’t think it is what happened in most cases, rather firms ended up getting by with fewer workers.

There’s also evidence of government funds chasing after the same set of skilled and already busy firms.  For at least a third of the surveyed firms receiving stimulus funds, their experience failed to fit important aspects of the Keynesian model.

This paper goes a long way toward explaining why fiscal stimulus usually doesn’t have such a great “bang for the buck.”  It raises the question of whether as “twice as big” stimulus really would have been enough.  Must it now be four times as big?  The paper also sets a new standard for disaggregated data on this macro question, the data are in a zip file here.


"hiring people from unemployment was more the exception than the rule in our interviews."

If workers were randomly drawn from the labor force (both employed and unemployed to fill these positions, you would get the same result. Yet every worker who leaves a job to take on a stimulus-created position might open up an opportunity for an unemployed worker (they might not, but a first-order analysis would suggest they would).

Exactly. Stimulus money moves employee from Company A to new job B. Company A hires replacement. That's a net employment gain.

That assumes that that worker was not a ZMP. Perhaps Company A did not need that worker, or could produce the same amount of output without that worker.

In order for the stimulus to work the way everyone wanted it to work, you have to assume that all firms were running as lean as possible to start.

Firms should have fired any ZMP worker. In any event, not all workers are ZMP and not all workers hired away are ZMP workers.

I think trimming the firms employee's and demanding more from who's left is bad for employee morale (which would lead to decreased output), But in the recession, morale was offset by the sour economic mood, so it was possible that these ZMP workers were created in this way.

In the good times, there were options and the firm runs the risk of losing more then who they cut (employee-employee loyalty). In the bad times, the options are fewer as everyone is cutting, people will fight a little harder for what they have.

In which case, we should assume that fiscal stimulus has the effect of improving the efficiency of the private sector.

I disagree, I think the recession had the effect of improving the efficiency of the private sector. Firms were forced to cut back, and employees, fearing their jobs will be cut too, picked up the slack thereby improving efficiency and creating these ZMP workers. I don't see how the stimulus could improve private sector efficiency.

A ZMP worker who is hired away by the government (and not replaced) should both contribute to private sector efficiency (because the private sector is no longer paying their wages for nothing) and increases total output (assuming the ARRA job has any positive effect). Alternatively, you could say that the ARRA hiring accelerated the efficiency gains, because the ZMP worker would have been laid off eventually.

It's the ARRA hires who were replaced where you have to look for negative impacts. If the replacement worker is less productive, or (as Bernard suggests below) is not hired because they would be ZMP, then there is a potential crowding out story. A replaced worker, however, indicates an net increase in employment from ARRA. It's the not-ZMP-not-replaced worker where the scenario is entirely negative. I'm not convinced that at 9% employment, the applicant pool is weak enough for that effect to dominate, but that would be the story.

Actually, I believe that the point was that the workers being hired for ARRA jobs were _not_ ZMP, quite the opposite. The problem was, rather, that the potential replacements from the unemployed pool _were_. Net result: some folks get very busy (and bid up) and some folks keep sitting.

But the point of Tyler's post is that apparently didn't happen. Company A got by on fewer workers or hired someone who was recently retired (or job B was filled by someone who recently retired in the first place). At least in my experience, government contract work frequently requires workers with a very high level of specific expertise and an employer can't easily substitute someone, no matter how eager, into the position. I'm sure the Recovery Act was a gold mine for people with 20+ years of government contracting or NEPA experience, but not so much for your out of work construction worker or insurance agent.

That's Tyler's opinion, but the study did not look into whether Company A got by on fewer workers or hired unemployed workers.

I think it's pretty obvious that not all new hires are not going to be net jobs created, because of David O's point about the numbers and because employers would rather hire from the people currently working at Company A, rather than the people that Company A laid off.

42% seems pretty impressive to me.

It depends. Impressive compared to what? Compared to the desire of 100% it's not that great. But what should our expectation be? What is the % for non-receiving firms? What is the % for non-recession firms? How much $ per unemployed hired? UI benefits go directly the the unemployed, spent on things the unemployed consume.

You're right that numbers need context. The right context would be something like how many of these hired people would be in unemployment if the money was not spent, which would be rather difficult to figure out. I don't see the point in comparing to the % for non-recession firms or non-receiving firms.

$ per unemployed hired would also be good to know. I'll do a little digging into the papers and try to find this out.

$ per unemployed hire usually ignores the value of whatever they are doing (if they build a widget, then the added value should be netted), ignores alternatives (if they would have received unemployment, that should be netted) and any multiplier effects. Also note that the government can currently borrow at negative real rates.

The reason I wonder about $ per unemployed hire is for comparison because you could just give these people money, or pay them to dig holes, or pay them to go to school, or pay them to simply stay out of prison.

The other comparisons I mention are also not to derive an absolute pass/fail, but a relative indicator. Did we have to spend more to employ people in this recession? And if so, does that imply that this is not a typical Keynesian template.

Considering that the archetypal Keynesian would say that there was too little stimulus, wouldn't that imply that the small amount would have the greatest marginal benefit if there are diminishing returns? So, assuming linearity (which we don't except for this hypothetical), aren't these are the best results we could hope to achieve?

"The right context would be something like how many of these hired people would be in unemployment if the money was not spent"

I think the right context is how many people are hired at some point after the stimulus ends.

I believe we all agree that the papers Tyler cited are misleading because they ignore the secondary hiring by firms that lost employees to the stimulus funded firms.

The questions are then magnitude, bang for the buck, etc.

I don't agree that they are misleading for that reason. Researchers report the results they got from the method they used. If the first comment about a paper brings up that question, isn't it a fairly obvious one? Isn't a claim to be misled by it kind of silly?

You are assuming that the downstream companies hired people to replace the poached employees. Just as plausible is that they did without. The authors could investigate this.

That is why they call them working papers.

That’s Tyler’s opinion,

Exactly, and it's presumptuous to express that opinion and then dismiss the opposing view as "just a story."

This is especially so when Tyler's opinion implies that the firms from whom the new hires came were keeping these employees despite the fact that they were unneeded.

One more point: if the stimulus enabls a worker to move to a better-paid (i.e. more productive) job, isn't that some benefit? Looks like it puts some idle resources to work. So the gain is that increase plus the previously unemployed replacements. Looks to me like that could be a double benefit.

"One more point: if the stimulus enabls a worker to move to a better-paid (i.e. more productive) job, isn’t that some benefit? Looks like it puts some idle resources to work. So the gain is that increase plus the previously unemployed replacements. Looks to me like that could be a double benefit."

And what of the economic cost to the business that lost that employee? It seems to me that the comments here are assuming that if a company loses an employee, then rehires someone else to fill that position, no harm is done, and therefore you have +1 jobs created.

In the real world, employees that are poached from other companies cost a lot more than just their salary. Hiring new people is expensive - so expensive that our company hires headhunting firms to find them and pays upwards of 20-30% of their first year salary as a premium to the headhunting firm. But even if the employee walks in off the street, there is cost to the company in terms of training, HR work, and risk.

Furthermore, the employee who is enticed away may be carrying with him significant local knowledge and domain experience built up over years which made his company more efficient. That knowledge is lost.

Finally, that employee may be filling a critical role, and the loss of the employee could have effects that trickle down through hundreds of other people. Lose a key project manager and waste a month finding a replacement, and the cost to the company could be orders of magnitude greater than the person's salary.

But getting back to risk: You know that when a company hires someone, it knows that there is about a 10% chance that this person will be a major asset to the company, an 80% chance that this person will be of average productivity and a decent hire, and a 10% chance that the person will cost the company a lot of money before he or she can be let go or moved to a new job more suited to that person's skill set. Hiring new people is always a risk - and it's one of the reasons why companies don't lay off every ZMP worker in a recession, so long as the worker is productive when the economy recovers. The transaction is expensive.

So even if every company that loses a worker eventually replaces that worker, there is still a large cost to the economy. How big? I don't know. Do you? How can you be so sure that those costs don't swamp the benefits of the stimulus? Does Paul Krugman know? If not, how can he be so sure that it's a factor he can ignore? And doesn't it vary widely based on the kinds of people hired? A stimulus project to build a high-speed train is going to have to poach the top people in their fields in the early days. What does that do to the industries those people were poached from? And what if what they were doing before is of much higher value to the economy, and the subsidy of the stimulus moved him to lower-value work?

NO, the stimulated company A hires someone from company B who DOES NOT replace the worker because they already have too much labor relative to their current demand. If they do replace the worker, it will be at a lower wage, a temp, or someone out of retirement.

In any event, employment receives no boost.

If hey already have too much labor, why haven't they let the excess worker go? And how do you know the replacement will earn a lower wage, etc. You're making things up.

Indeed, and if their previous employers didn't hire anyone new because of the economy, it's likely they were about to lay somebody off soon anyway. Also, the phrasing "more the exception than the rule" is misleading - I would expect the unemployed to make up less than a quarter. If those recent graduates don't get hired right away, they tend to become unemployed plus there are long term repercussions on their success.

At least in my experience, there's a *huge* gap between "not doing so well, non-critical workers are not replaced" and "doing disastrously, firing non-critical workers". I've seen organizations operate for years in the former, and only a handful of times have they gotten rid of workers.

Employers don't like firing people either.

Tyler, thanks for the links to these new papers. Let me pay you with a link to Larry Kotlikoff's Purple Tax Plan


I look forward to your comments. Indeed, for accelerating the recovery, political agreement on a tax reform is much more urgent and important than "stimulus" plans or inflationary Fed policies.

Let me give you something else on fiscal policy to think about:


"To make the Purple Tax Plan's consumption tax truly progressive, the plan includes a monthly payment (demogrant), which ensures that those living at or below the poverty line pay no tax, on net, on their consumption. "

I have a problem with this. The underlying problem is that the two ideological dump buckets can't come to terms with what being poor means. There are two types of 'losers,' those who fail because of the system and those who fail despite the system (just as there are two types of 'winners'). Thus there are also two ways to categorize parts of the system. A consumption tax opens the door to user fees associated with consumption and if I were Larry I wouldn't waste that opportunity.

This doesn't sound a whole lot different from the much older, and widely studied, FairTax plan. I don't really see how this is any better, in fact I find the FairTax preferable on the margin. Why keep FICA taxes, even if you make them more progressive? In my mind payroll taxes like FICA, regardless of the slope of their rates, are one of the most undesirable forms of taxation: when it comes down to it, this is just an extra cost on hiring workers (or a reduction in their effective wages). Consumption taxes are wholly voluntary and automatically scaling to a taxpayers wealth/income, while simultaneously enhancing incentives for investment in an unbiased way. Conceptually the only form of taxation I find to be more attractive is a land value tax, like those advocated by Georgists; an LVT dissuades land speculation, enhances incentives to use land efficiently, and causes no dead weight loss (due to the nominally fixed supply of land).

Assuming, for the sake of argument, the Keynsian theory, how does one target spending in a 21st century economy. A 1930s economy presumably had a high percentage of low skill labor for the New Deal alphabet soup to employ. Our modern economy is more heterogenious and reliant on specialized human capital.

This criticism was made at the time of the stimulus. How, for example, would a bridge project employ newly unemployed finishing carpenters and plumbers? Even if projects were "shovel ready", would the stimulus effect still be moot?

The question becomes, even if Keynsian theory is sound, is it a workable engineering solution? Or, rather is it as useful as a "worm hole".

I guess that leads to the question of whether Irene is more job stimulating that $15B of stimulus (http://thinkprogress.org/yglesias/2011/08/31/308468/overhyped-hurricane-irene-likely-to-be-one-of-the-ten-costliest-disasters-ever).

My guess? It probably is - simply because of the wealth in the region that it hit.

This is an interesting set of final results. Perhaps instead of a stimulus the money should have been more directed in stuff like a revival of the Civilian Conservation Corps perhaps? We would have at least gotten some good parks out of the deal.

A random commenter somewhere noted that this isn't the labor market of the 1930s: you can't just give people shovels and put them to work digging, which I think is a large part of the problem. And the modern unemployed aren't seeing their utility fall as much as the unemployed in the 1930s (the modern poor in general aren't missing a lot of meals and tend to have most major appliances and conveniences, few of which were in common use in TGD), so a fair proportion of them wouldn't deign to dig ditches for money anyway -- as evidenced by failed attempts to give unemployed ex-convicts (the bottom of the labor market) jobs normally done by immigrants, such as tomato picking.

I think Bill N has it right -- the economy of today is too granular and specialized for broad central planning solutions to work, or at least they will be much less effective, especially given the falling utility of work generally.

This is exactly what brought me over here to comment.

How many of the workers employed by ARRA were just swinging hammers? My guess is few. Surveyors, heavy machinery operators, etc are skilled labor who had to have those skills beforehand. I'm sure there were some positions that were just menial, but I doubt many were.

This is another good reason to plan out any fiscal stimulus in good times, the other side of the coin of actually having surpluses is to have ACTUAL shovel-ready projects.

I have decided that I will always have a shovel-ready project for when I lose a job, a garden.

along the same vein, why would you spend all that effort swinging those hammers when you can get the same amount of income on unemployment benefits?

Because very few people *like* being unemployed, pay differential or not.

I have to say, I've never quite understood why people who are not lazy, stupid or evil, and are surrounded by friends who are not lazy, stupid or evil, who work with people who are not lazy, stupid or evil, are often convinced that the world is mostly populated by people who are lazy stupid or evil.

(Okay, in this post, just lazy, but still...)

It's not about being lazy. Unemployment has a specific purpose (right or wrong) - to bridge the gap between jobs. People were expecting their jobs to come back when Obama was supposed to turn the economy around. The idea was: "let someone else more hungry than me take that Hammer swinging job, they'll stimulate the economy and my job will come back, I'll just wait it out".

This is all my personal conjecture.

I don't know, I don't think being rational is "evil."

I know a handful of long-term unemployed. They aren't jumping at every job offer because they have UI, a working family member supporting them, and children which would add a significant expense to returning to work. They don't especially like being unemployed, but they aren't suckers either.

Exactly -- they can face a ridiculous MTR.

There is some truth in this.

Anecdote time: in 1937, with WPA money, Chicago began to dig subways downtown. They were dug by men working with shovels. Today New York City is digging the Second Avenue Subway line. Not by men with shovels but by a giant boring machine. So the work of the 1930s, building subways, is being done today by specialized men running specialized machines rather than low skill men with shovels.

I'm not wholly convinced that there are a lot of empty jobs that the unemployed are refusing to do, though, and I think we've argued over this before.

You can't stimulate your way out of a real supply shock. We are now trying on the fiscal side what Fed Chairs Burns and Miller tried on the monetary side in the 1970s. The solution which actually worked was Volcker's tough love (plus a lot of Gulf of Mexico and North Sea oil).

2008 and 2011 were / have been both years of oil shocks. Oil, as we know, is subject to demand inelasticity (price) during periods of economic growth, even in the face of high and rising oil prices. Now why is that? Because oil is an enabling commodity--other activities depend upon it; therefore, the loss of oil consumption must be greater (by definition) than the incremental cost of the oil It should be possible to quantify this.

Find us the paper with that analysis!


You are touching on some of the things I have been ruminating over for some time. There is not a whole lot of economic research that is devling into questions of oil consumption, demand elasticity, and recessions. I have seen very few economic analysis that ties the dot com boom with 90 cents a gallon gas. Perhaps I am off base and correlation is not causation. Why are we not seeing the inflation that we had in the 70s? Slow growth, low interest rates...something doesn't add up.

Why didn’t the stimulus create more jobs?

there was only $48 billion for transport infrastructure & $72 for other infrastructure in ARRA; most was aid to the states for medicaid, education, etc; and 37% was tax cuts, including the sneaky $5 a paycheck less in withholding....most of that went to walmart, & hence employed the chinese...

wish this was acknowledged much more often--expecting Keynes-supporters to treat $70bn AMT tax cut that already occurs every non-depression year as "stimulus government spending" is all you need to know to understand the intellectual depravity of this debate

The backwards-looking analyses by Zandi, Goldman, etc that I've seen all estimate that the combined impact of TARP and the ARRA was several million additional jobs at peak unemployment. These estimates all came well before the recent downwards revisions of economic performance in 2008 and 2009; I don't know what the same folks would say today.

To what extent was the projected impact of the stimulus supposed to come from direct hiring?

It may be an absolutely isolated case study, but a few months ago I noted that at my school we tore up the steps to a building across the street from me and then rebuilt them. The new steps look and function just like the old steps, which were fine. I've no idea if the workers previously had jobs or not. And fwiw, even if they did have jobs, I've no idea if they were documented jobs in America, jobs in Mexico, etc, meaning I don't know how their direct employment affects employment figures.

Anyway, our school recently decided to do the same thing w/ a parking lot. We destroyed the old one, which was fine, and are now building a new one, which looks remarkably like the old one, save for a few extra bushes (that make it dangerous to turn out of the parking lot because it is now harder to see).

As far as I can tell, we're paying people to dig holes and fill them back up, while being inconvenienced via the loss of use of steps and parking while they do their jobs. I suppose the argument is that these workers, while they may or may not count in the employment statistics, spend some of their money here (rather than sending it all back home), and some of that trickles out to businesses that will either hire or not fire? Should we consider this to be a success?

Could they have been fixing pipes or wiring under the steps?

Yeah, just saying "the old steps looked just like the new steps so there was no point in replacing them" is a pretty shallow analysis. If the state replaces an 80 year old bridge at the end of its functional life with a new bridge, it'll look more or less the same. That doesn't mean there hasn't been a capital improvement.

Maybe so. But in my case I was made to move out of a lab that was producing (modest) results (but results nonetheless which is more than could be said for many) so that they could redo the floor so as not to lose funds at the end of the fiscal year. Now the lab sits empty. On net, I'd go with the incompetence assumption.

Many ex-Soviet people will recall here Strugacki brothers' gem about "a crew of Danaides* in telogreikas with jackhammers" — the authors' note being to the effect that "when the Danaides' case was unsealed for judicial review, the court found that they were married off by force and commuted their sentence, putting them to a less meaningless labour. Now they are doomed to eternally breaking up the asphalt they themselves have recently laid."

The stimulus projects could have drawn in more unemployed workers if they had offered lower wages. If the goal is to employ otherwise idle workers, it makes no sense to offer wages high enough to attract workers already employed elsewhere. Davis-Bacon is horribly counterproductive for fiscal stimulus.

You raise the problem of discriminating against the company that hires at prevailing wages to support the company that does not. If you use prevailing wages, you hold that issue constant, unless you want to expand the base of companies hiring below the prevailing wages.

Managers of stimulus projects face the same risk as managers in private industry.

Candidate A appears qualified but is unemployed. Candidate B appears qualified and is employed. Who do I hire? More importantly, what is "wrong" with candidate A. He has an excellent resume, interviews well but he has was laid off nine months ago. Why hasn't anyone hired him already? Am I missing something?

I've been hiring people (exempt and non-exempt) for over 20 years and I have to admit that in nearly all those cases, I've gone with the candidate B, because B is the less risky choice. The only factor that changes this calculation is a personal recommendation by someone I trust.

Would it be stretching this analysis too far to apply the same principles to hiring companies with stimulus dollars? In one respect, the stakes are higher. As unpleasant as a bad hire is, choosing the wrong company borders on a professional disaster for the manager making the decision. One of the default rules I've learned over the years is, "You can never go wrong choosing IBM!" Selecting the brand leader, 800 pound gorilla, etc is always the first line of defense when a project misses deadlines, budgets or goals.

I was involved with a project to extend broadband internet access to the un- or underserved in Ohio called "ConnectOhio." The people who run the thing were already employed and the state was cutting funding so they applied for and received a stimulus-funded grant to continue and expand their operation. I believe the award was $10M. The planning was still not complete and implementation only beginning this summer. The money was going to: community colleges and libraries to provide instructional sessions, providing vouchers for discounted equipment, and of course, another couple of years of salary for the ConnectOhio folks. I did the math and I think the total number of participants they were shooting for was like 10,000, so that is a very high cost per participant. In other words, it was terribly inefficient and did not create a single job.

It seems there is a class of worker we can call 'would have been unemployed if not for the stimulus'.

Why do we not talk about them? I don't think it is crazy to believe they exist.

I suppose I have to read the paper... 8^0

Re: would have been employed if not for the stimulus. My firm received stimulus funds as a component of a university-led grant. As part of this non-profit award we retained 2 employees that we would have laid off. To be clear these would have been our lowest MP employees, but far from ZMP. We added clients after the worst of the recession and not only retain those employees, but have added several more.

While I realize that this is just one data point, my opinions of the stimulus are colored by my direct experience in which it not only saved jobs, but saved subsequent search & training costs.

If you used Tyler's methodology, if your research project involved two segments--administration and the actual segment that conducted the research--and you could not do the project without the administration segment, Tyler's group would interview the administration and find they did not need to hire any new administrator, and in fact, the ARRA took an administrator of his/her otherwise productive activities--and that it did create two jobs at the research level.

You can see this clearly in the paper where it talks about 6 engineering firms which supposedly support the effort of construction companies that do hire people. 6 engineering companies reported they did not hire anyone, or that it supplanted existing work.

So, in Tyler's report, it would show 50% of the firms reported that they did not hire new employees or save a job.

what bugs me is that the same people who say that we had shortages of skilled, trained labor (during a serious recession!) also seriously doubt the value of education, or of making sure people are working if not just to keep their skills up, ie of building skilled labor capital.

in a serious recession it seems to me the least we can be doing is building / maintaining capital, and if capital in our economy is mostly human capital, why not push for development of that?

You can tell where this paper is going:

1. "How did the $787 billion American Reinvestment and Recovery Act (ARRA) affect the
U.S. economy?" -- Sorry, it wasn't $787 in stimulus spending. Remember the tax cut.

2. "we
conducted a first-of-its-kind study,..." Well, actually no. You conducted interviews, but not a first of its kind study, unless you mean taking interviews.

3. "Keynesian economists claim that government spending in a recession can have a large
―multiplier effect,‖ where a dollar of government spending quickly increases the size of
the overall economy by a dollar or more: the government hires unemployed workers, who
then ―multiply‖ the effect when they shop for consumer goods." Not true either, based on the comments above. Trick is to mistate what the other side says so you can knock it down.

4. Their counterthesis is that the government hired away workers who would have otherwise been in the workforce doing private jobs: "In the alternative neoclassical model of the economy, when the government hires
workers, it hires them away from other private-sector firms, so government spending creates zero additional employment. Of course, both of these views are caricatures, simplifications; the question for policy makers is which simplification best matches the facts. And based on our interviews, the neoclassical simplification fits the facts better than the Keynesian one." Based on the interviews, smashed on the interviews: ask yourself from your own experience whether you think that the part of the ARRA devoted to public spending "hired away" persons who would otherwise been used in the private sector, and if you think so, remember part of the ARRA was money sent to state governments--persons who were retained by state governments, not private employers.

5. "This is not to say that ARRA-funded projects created no value nor that ARRA projects failed to create any temporary jobs. We do not conclude whether the stimulus ―worked‖ or ―failed." Tell this to Tyler.

6. Table 1
Table 1: Quick Facts about Job Creation
Total Interviews 85
Interviewees who discussed job creation 59
Interviewees who reported saving/creating jobs 41
Interviewees who reported no job creation/savings 18

Hmm. So, of the 59 interviews you had, 41 reported saving or creating jobs to the Mercatus interviewers.

7. There is some technical stuff on how ARRA job counts are conducted. There is also some examples of how a job was hired away from the private sector to fill an ARRA database job. Supposedly the private sector filled in the lost employee but that is not discussed or checked.

8. They do not like Davis-Bacon, but offer no evidence that there was a wage increase from a move: "Job shifting into an ARRA-funded job would be particularly attractive to many workers because the terms of the statute often required ARRA-funded jobs to pay so-called―prevailing wages,‖ or ―Davis-Bacon wages,‖ which are typically equivalent to unionlevel wages. This is great news for the worker who gets the job, but it’s usually bad news for the national economy."

9. They do not consider multipliers from purchases other than labor, ie, if a construction contract uses cement that would not have otherwise been purchased, Mercatus does not include the employment at the cement factory. "As mentioned above, we also found evidence that ARRA often created work without creating jobs: we heard many versions of, ―Things were slow until the stimulus money
came along; ARRA gave our employees something to do.‖ In Keynesian terms, these organizations were labor hoarding, holding onto workers through the slowdown even though they didn’t have much work to do. In these cases, ARRA funds boosted profits by plugging a hole in the company’s revenue stream; whether or not ARRA actually saved jobs at labor-hoarding firms is still a matter of speculation." No discussion of multiplier for purchases.

10. One of the major errors of the study is to presume that if there is, for example, a construction project, that ALL parts of the project have to create jobs. For example, a construction project may require an engineering firm that isn't going to lay anyone off, or hire anyone either, because of the ARRA project. At the same time, the construction company IS going to lay off or has layed off employees AND will not lay off and will rehire employees if the ARRA project goes forward. 6 of the 18 examples of the where there were no job increases involved interviews with engineering firms. "When they took ARRA work, they were turning down other work." Yet these were some of the examples, and interviews, of why they claimed the ARRA didn't necessarily create jobs.

11. I don't know what to think of this criticism, other than to say I'm glad the government collected statistics: "Many of the firms our teams interviewed complained about ARRA’s detailed reporting
requirements, especially the time and energy required to learn the reporting system.
While there’s room for improvement, such complaints seem to stem from the nature of
bureaucracy, not failures of Keynesian theory." Sure glad someone did collect data on the government side, though. At the same time, though, I am sure that how you quantify a job or report is debatable.

re 8; correction, later they do show wage increase.

If Keynesians do not claim a 'large' multiplier then what do they claim? If they aren't claiming that their plan is better than the opportunity cost, then what are they claiming?

Did the stimulus create as many jobs as the people crafting the stimulus projected? In that sense, you don't have to say that the stimulus failed, even though it may have failed relative to expectations.

Your 11. is funny to me. "At least we required wasteful efforts at wasteful effort accounting!"

Andrew, They didn't measure a multiplier, in case you missed. A multiplier includes revenue and job hires down the line, not just the immediate purchase. So, if there is a construction job, they measured the construction hire, not the effects of the cement purchase, not the effects of the hired employee buying something, and so forth. Don't give them credit for what they didn't do.

I may not have explained point number 10 to clearly, so let me explain by an example.

Let's say that there is a construction project to build a new highway. To build the highway, you need an engineering firm to design it, and a construction firm to build it. You need both firms.

Tyler interviews the engineering firm, and they say: no new hires from the ARRA, and in fact, we just used our existing personnel whom would would not have laid off.

Tyler interviews construction company: Yeah, we saved or created 30 new jobs.

Tyler reports: 50% of the firms reported that they did not save or create new jobs.

Luv it.

Bill, that's a fine point, but I don't sense the partisanship you seem to be accusing Tyler and the authors of. We don't need to downgrade Obama and Keynesians when they downgrade themselves.

"As President Obama conceded in an October 2010 New York Times interview, ―There’s no such thing as shovel-ready projects." Our interviews provide a thorough, reality-based vindication of the president’s statement."

The President's quote is a worse indictment than anyone from George Mason could issue because of course there are shovel-ready projects, but they aren't shovel-ready when you need them if you did not prepare. We did not prepare the surplus funds, and we did not prepare the project plans. Can the government be counter-cyclical at all? If not, that calls into question the Keynesian underpinnings. Maybe it didn't work because we did it wrong, and you can blame Bush as far as that goes, but I don't remember a lot of Democrats calling for surpluses. Partly it's understandable because we haven't felt like we weren't in recession or entering recession for 10 or 15 years even when unemployment was "impossibly" low.


I'm sorry, that was not a "fine" point to show that when in one part of the report the claim that "6 engineering firms" reported they did not hire or save jobs, and then point out that there are 18 respondents who claim they did not hire or retain new people. It is not a fine point to quote their own report and tie it into another part of the report, showing that FULLY 1/3 of the respondents who did not report a save or hire were engineering firms, and that engineering firms are complements to other firms who probably did hire. What's more, I do not know the characteristics of the other firms--were they complements too?

I'm an antitrust lawyer and deal with economists all the time. When someone starts doing number counts of firms, rather than a statement of the relevant number--the number of jobs--you have to be an idiot to overlook what is going on. (I know someone will call me an idiot with this remark, but so what).

Speaking of idiots, I know or I'll bet the press will report this as 18 of 51 AARA recipients reported they did not create or save new jobs, based on interviews by the Mercatus Center.

Wanna bet.

Should be 18 of 59, not 51.

just 42.1 percent of the workers hired at ARRA-receiving organizations after January 31, 2009, were unemployed at the time they were hired (Appendix C). More were hired directly from other organizations (47.3 percent of post-ARRA workers), while a handful came from school (6.5%) or from outside the labor force (4.1%)(Figure 2).

42% hardly counts as 'the exception'. More importantly, though, should not those hired from school or, even more so, from outside of the labor force be counted?

What definition of 'unemployment' are we using here? If it's the official one, we all know that it does not come close to encompassing everyone who is in need of a job and cannot find one. Could 'outside the labor force' mean those who never began taking unemployment or those whose benefits have run out? Even if we're talking about a retiree who has to start working again because his 401k dried up or a housewife who has taken work to cover for an unemployed spouse, it is still a win for society. An expensive and inefficient one, yes, but still a win.

How many of those 47.3% hired from other organizations would have been unemployed, if they weren't hired for ARRA projects? Not long ago it was: "The effectiveness of a stimulus will be measured by how many workers it bridges over until most of the deleveraging is over." That sounds like "bridging" to me.

Our conclusion that hiring from unemployment was "the exception" came from our first paper, based on 85 in-person interviews; the 42% estimate comes from the separate survey paper with 1300 respondents.

It was important for us to convey the findings separately--conversations are different from forms---and in the interviews, hiring from unemployment was the exception.

In both cases, the federal government and other researchers can create larger samples, and ask workers and organizations even more questions about how ARRA changed their work environment.

Fungible money meets non-fungible labor.

No real recovery until the prior malinvestments are flushed out. The government is doing everything in its power to prevent this from happening.

Will you publish the names of the firms that were interviewed so that someone else can interview them and check the data?

We provided the firms with anonymity--something they surely value. We discuss anonymity and human subjects research guidelines in the paper.

We would welcome other researchers (especially in the federal government) conducting their own interviews. The federal government could surely interview thousands of organizations while only spending a tiny fraction of the stimulus.

And the data for our separate survey paper is all online, at Tyler's link.


6 of the 18 firms that claimed they did not add or save jobs were engineering firms.

Could you please list the industries/businesses for the other 12, and also indicate for each whether they were complementary to another business (ie, construction firm) that did add workers.

My point regarding engineering firms is that they are complements to construction firms in a building project. I wouldn't be surprised that an engineering firm didn't hire/save a job for an ARRA construction project, but the person actually doing the construction on the project DID hire people.

You can't ignore complements in a project. Take one project, building a road: 0 extra engineers, 25 construction workers. Yet, you can say 50% of the firms did not hire/save in this example.

It seems to me Obama -- or Congress -- is given little credit for the actual stimulus that was implemented. There were people at the time who argued forcefully for a stimulus plan much more focused on rebuilding infrastructure and other new spending. The actual plan that emerged devoted a subdominant fraction of resources to this, so that comparable shares could to go toward sustaining state governments and tax cuts. It was a balanced plan, and the value in balance is it covers lots of bases.

So now we can debate whether a sub-portion of the plan created jobs, or maintained jobs that would have been cut (I'm not sure really what is the formal difference between these, though the paper is evidently distinguishing them), or doing something else. Maybe intelligent people disagree about these results, maybe not, but intelligent people disagreed back when the issue was being debated, so it was not a stupid call.

The smart call was not to put all eggs in one basket, and take a balanced approach. And no one seems willing to give credit for that.

It seems that you would be better off paying the unemployed to pick up litter for $7.00/hour.

You want us to create the Department of Homeland Clean-up Patrol?

Too many companies refuse to hire unemployed people. It only became moderately noticeable enough that a few stories appeared in the major news outlets this time, but it has gone on for years. Even the NY Times finally noticed this year.

My previous employer was a large white-collar company with lots of experience at wallowing at the public trough, and they were refusing to hire folks in the US (and only hiring offshore because of lower labor costs). I got tired of getting assigned the work that folks who quit had been doing, so I ended up leaving for greener pastures.

In the city I live in, the only "shovel ready" projects were road repairs that the current economic climate could not afford to fund. Other projects could not get ARRA funding because they were too big, or to far away from having environmental impact statements completed.

This is the most ludicrous argument I have ever read:
"Many of the created jobs involved hiring people back from retirement. You can tell a story about how hiring the already employed opened up other jobs for the unemployed, but it’s just that — a story. I don’t think it is what happened in most cases, rather firms ended up getting by with fewer workers."

So apparently the argument is that firms had excessive workers that they were employing, but once the ARRA was passed they decided to employ fewer workers. ARRA greased the wheel to make workers work more. There is some fixed stock of labor but flexible amount of work they can do. Firms cannot shed these workers, but if demand rises they can make them work harder.

This is pretty bizarre. Like the paper and the whole post.

Keynesian theory says companies engage in "labor hoarding" during recessions: They'd like to fire a couple of people, but they're afraid to hurt morale, for instance. Labor hoarding a normal part of the Keynesian toolkit. We argue that if Keynesians are right about labor hoarding, then workers who leave voluntarily are doing the firm a favor.

Of course, in the neoclassical world, it's even simpler: There's a fixed number of workers, some frictionally (or overoptimistically) unemployed, and so on average you're playing musical chairs with jobs. I think they're something to both stories---and both stories tell me that job-switching is unlikely to lead to job creating.

More details in the paper, like this one: Rates of job switching in our sample are close to rates of job switching over the last few decades in the US.

Why isn't labor hoarding also a part of the classical toolkit as well--it costs money to let someone go, you're at risk if you can't get back the person you trained or if you have to train someone later to replace the person you lost.

Because of this, some companies don't mind UI insurance--because they can let go of an employee faster than they would otherwise and be able to reinstate them later. They pay the premiums for this over a business cycle. That would also explain why rates of job switching are relatively constant across all periods.

That only follows if the people who are being hired from other jobs were the ones that the other firms wanted to layoff but were "afraid" to. But that would go against Tyler's new argument about labor polarization, since now he is arguing that the people hired are the better off people, not the ones the firms wanted to let go.

Exactly. And isn't it the skilled, productive workers who are more likely to switch - the ones firms aren't hoarding? So are they doing their employers a favor by leaving? And, as I mentioned above, there could easily be a gain to the economy just from the job switching, as workers move to more productive jobs. The obvious case is the part-timer becoming full-time, or the engineer moving from driving a cab to doing engineering.

If your last name is Rothschild, you are not to be trusted. 'Stimulus' is just another political word-play intended to give a postive spin on something comepletely criminal, i.e. creating more debt and hence profit for the ruling bankster elite (Rothschild, Morgan, et al). Until Americans wake up, end the FED and put the bankster criminals on trial for treason and crimes against humanity...the shit will just cointinue to slide toward a world government and monetary system.

Building Rome so that it could fall seems reasonable to me. A single world government we will have!

You realize that Rothschild is a fairly common Jewish name, right, not the exclusive province of a rather aged banking clan?

Wow, so many problems with this summary. Many of the immediate issues have already been mentioned. But what about stimulus money that went to state governments? Many advocates of a larger stimulus wanted to see more money going to the states to prevent government layoffs. If you're mindset is that teachers are lazy and can be fired at no cost, these kinds of layoffs are not a problem. But with a growing population in most states, laying off teachers to hire new ones in 3-5 years is asinine and counter-productive. These are certainly jobs that could have been saved.

And I have to say that the claim that hiring an already employed worker will lead to his spot being filled by someone else who is unemployed as 'just a story' is crap. Either come up with a way of investigating this or just shut-up. The requirement that any new hire be currently unemployed to count as real job creation is a completely unreasonable way to evaluate the situation. Very rarely are you going to find firms opening up new offices and hiring boatloads of unemployed workers in any economy. But that doesn't mean we don't talk about factories or offices creating x numbers of jobs where x is exactly the number of people employed.

"Too many companies refuse to hire unemployed people"

Yeah, I've noticed this too. And I'm unemployed.

But it just means that to put the unemployed back to work, the economy has to pick up strongly enough that employers are forced to hire even from the unemployed to meet demand. And if the economy worsens, conditions worsen for the employed too as they either become unemployed in turn or have to deal with higher workloads. So in the end it still comes down to the overall economy.

But I also agree with the poster who said that you can't stimulate your way out of supply shock. We have a long ways to go before a dent is made in unemployment.

I've seen supply shock used elsewhere.

How is this a supply shock rather than a demand shock?

Seems to me that we had a bunch of debt-financed spending, and that was brought to an abrupt halt which led to unemployment, and the two linger together and continue to keep demand low.

"And I’m unemployed."

Then get the fuck off the computer and go look for a job.

How about you get into the 21st century. The computer is where you find jobs nowadays.

And he's right. It is well established that the best way to get a job is to have a job. Nobody wants to hire those of us who are unemployed. It's ridiculous, but true.

Some day, Troy, you're going to look back at all the ridiculous things you've posted online (I love it when the Keynesian lawyer Daniel Kuehn bitch-slaps you over at the GMU blog) and be deeply embarrased. Will you be sufficiently embarrassed to stop posting? We can only hope so.

There was some stimulus that was effect, only it wasn't in the stimulus bill. The census was economical and effective, hiring people works.

Go fuck yourself.

Here is what Gary Becker wrote in January 2009. It seems like he anticipated a study like this

"Some of this infrastructure spending may be very worthwhile-I return to this issue a bit later- but however merited, it is difficult to believe that they would provide much of a stimulus to the economy. Expansion of the health sector, for example, will add jobs to this sector, but it will do this mainly by drawing people into the health care sector who are presently employed in jobs outside this sector. This is because unemployment rates among health care workers are quite low, and most of the unemployed who had worked in construction, finance, or manufacturing are unlikely to qualify as health care workers without considerable additional training. This same conclusion applies to spending on expanding broadband, to make the energy used greener, to encourage new technologies and more research, and to improve teaching. An analysis by Forbes publications of where most jobs will be created singles out engineering, accounting, nursing, and information technology, along with construction managers, computer-aided drafting specialists, and project managers. Unemployment rates among most of these specialists are not high."


The response to Becker is "where most jobs are created" is not the place where job vacancies currently are, and where there would be jobs, if the economy were at full employment. Stimulus doesn't "take away" jobs from, say, accounting or IT, the area where most "jobs created" during the recession are. They add jobs to sectors where jobs were not, but would be, if there is a turnaround.

Sorry, but I am not sure I follow that. I think there can be a time issue here. If the government or the firms that are selling something to the government are going to hire someone for the new job, it will take time since there are not alot of people looking since their UE rate is low. So it will take some time for the new position to be filled.

Then even if it does get filled by Bill from company A, then company A has to decide if they will fill Bill's position. Then after that lag, they start taking applications and interviewing. This can be a slow process since it will be hard to find applicants since these people already have low UE rates.

Cyril, My point was that the hiring was in business sectors where there is high unemployment, not low unemployment. Take road building, construction for example, and apply that towards renovating school buildings to make them more energy efficient. Becker's point was that stimulus draws some jobs to areas that are currently growing; my point is to the contrary. The real question should be: are these jobs sustainable, or should we also, and in conjunction with this, look at retraining at the same time.

Those of us who actually work in industry and are involved in large engineering projects of the type the stimulus was designed to 'stimulate' could have told you this without waiting for a study. We tried to. No one was listening.

It is maddening to hear 'workers' talked about as if they are interchangeable - Oh, a whole bunch of home construction workers have been layed off? Don't worry, we'll build a road or a bridge and employ them!" The only problem being that the type of construction home builders are trained for has nothing to do with bridges. Perhaps people like those in the Obama administration lack the appreciation for the real complexity of these jobs and assume that any blue collar work is trivial and interchangeable with a little retraining, but it's not the case.

Not only that, but the people who can *start a project are very different than the people needed to bring it to completion, and in general the people needed at the beginning of a project are the least likely to be unemployed. In fact, even in a recession there is a shortage of such people. So it was predictable as rain that new stimulus projects would have to scavenge project leaders, architects, managers, and senior engineers from other existing projects that may have more value. It was absolutely, 100% unavoidable.

Not only that, but it is incredibly destructive: Pulling a manager from an existing project can cause damages far exceeding the salary of that person. If a project manager or architect is enticed away from a project that has a $100,000 per day development cost, and his leaving causes a month of delays while a new manager is found and brought up to speed, that's a cost that will never show up in the stimulus accounting - but his $150,000 job will be counted as a 'job created'. No one will know that in addition to the stimulus money used to hire him, the real cost of that job was an additional $3 million dollars. I've never seen a single Keynesian model take that kind of destruction of existing projects into account or try to quantify the effect. You'd think this might be important to consider - especially in an era where specialization is so important, where even low-level positions require specialized training.

It was also inevitable that some of the people hired would be pulled out of retirement - when you need top guys, you're not going to find them in the unemployment line. You'll find them sitting on their sailboats in the Carribbean.

No one but the Austrians seems to care about the actual fine structure of the economy, and the effect a barrage of government money might have on it. None of the people driving policy seem to have an understanding of just how long and complex supply chains are, how fragile they can be, and what it does to a company to pull key people out of existing projects because they were enticed away by stimulus programs.

My company has hundreds of job openings. OUr inability to fill them has nothing to do with aggregate demand - we simply can't find the kinds of people we need. The bad incentives baked into this economy for the past two decades have diverted the workforce into non-productive areas. It's pushed people out of science and engineering and into finance and law. The bloated housing sector attracted people away from computers and electronics into carpentry and plumbing. We've distorted our labor pool, over-built in numerous industries while leaving chronic shortages in others. Easy access to student loans and easy, long-term repayment has disconnected the choices of students for school and faculty from the economic needs of the country. All of these problems are completely masked by focusing on aggregates.

Let me ask a simple question: If your productive capacity is totally focused on making tables, but your population has plenty of tables but has a real shortage of chairs, and no one is trained to make chairs, how much fiscal stimulus would you need to pick up aggregate demand? Answer: Infinite. People don't want more tables at any price. As long as that's all you're offering, aggregate demand will remain low, and all your table makers will remain unemployed. In fact, the stimulus is just masking the problem and delaying the necessary adjustment to a 'chair economy'.

Aggregate demand may be way down, but Apple has no problem selling every iPad it makes.

I guess maybe in the past, in a simpler time with simpler labor requirements, you could get away with assuming that the law of large numbers would apply and labor could be treated the same. Who says that's still the case?

"Let me ask a simple question: If your productive capacity is totally focused on making tables, but your population has plenty of tables but has a real shortage of chairs, and no one is trained to make chairs, how much fiscal stimulus would you need to pick up aggregate demand? Answer: Infinite. People don’t want more tables at any price. As long as that’s all you’re offering, aggregate demand will remain low, and all your table makers will remain unemployed. In fact, the stimulus is just masking the problem and delaying the necessary adjustment to a ‘chair economy’."

1) we have a huge hole in the economy, which you might call a liquidity trap, but whatever you call it, it exists because interest rates on treasuries are often negative at this point.

2) when you are in something that might be called a liquidity trap, hiring people to make tables - and using the excess tables to bomb your enemies or anchor ocean reefs or to build giant table sculptures or simply burning them for home heating - any of these will stimulate the economy. ('bombing' is a reference to the effect of WWII on the Great Depression, which was another example of something you might call a liquidity trap.)

How about instead of hiring people to make tables we just give them money instead and let them spend on whatever they want? It might be more optimal or efficient. The resources that would have made the tables we did not need might make something more valuable.

That would be my preferred approach.

I just wanted to respond to Dhanson's specific hypothetical.

(1) We did that in ARRA already--it was the hundreds of billions of $ (multiple times the direct government spending) in tax breaks. As predicted for a balance sheet recession, this was primarily used to pay down debt. While that's important, it would have been better achieved as per Rogoff, Krugman (and Tyler!) etc through temporary inflation. By contrast, the point of direct spending was to employ idle resources, but of course Olympia Snowe prevented that from happening in any significant amount. So failure on all fronts.

(2) Infrastructure is a public good, and we have $2bn of missing maintenance. This is not optional, and is only being deferred until it's catastrophic (more falling bridges), at which point (presumably) government borrowing will be more expensive, and there will be fewer idle resources, making crowding out an actual concern rather than Republican bogeyman.

But don't worry, I'm sure Tyler has another post in the works to explicitly take on doctrinal conservative dogma that tax breaks are everywhere and always the most effective government policy by looking at their (much larger!) component of ARRA.

Cyril, Giving people money with a high MPS directly gets you nowhere. I'm wealthy: give me a tax cut, I save it and with part of it buy a Korean TV, which is what I did v. build or repair a highway in the US.

I said give the money to the people who would have been making the tables. If they are currently unemployed, they don't necessarily have a high MPS

Also, and I wish I could find this, I thought I read that sometime last summer, less than half the money that the stimulus allocated to infrastructure still had not been spent, due to the well-known policy lag problems (as mentioned in many textbooks).

I could be wrong about the piece of data I mentioned. I thought I had saved the article that said that buy I cannot find it. So if anyone knows what % of the infrastructure money has been spend up til now (or at some time last year), please post it.


So your proposal is to give money to the unemployed rather than employ them in a project like building or repairing a road? Pretty radical.

There's a liquidity trap because demand is down, and interest rates can't go any lower to prop up demand. But doesn't that beg the question? It is really a liquidity trap if the real problem is that people don't want to buy what suppliers are producing, and suppliers aren't investing because they can't find the people they need to build the things people really want, and people are unemployed because they don't have the skills needed to supply the demand for labor? If I've already got a good table and no chairs, you can run your interest rate negative and you're still not going to entice me to buy another table.

For question 2, since you've expanded my simple analogy, let's take that a little further - Let's say there's a chair industry that's fully employed, and a table industry that isn't. The signal the economy is sending is that we want fewer table manufacturers and more chair manufacturers. The Keynesian answer, on the other hand, is that the table manufacturers are idle, so therefore we should borrow money and use it to stimulate table production. In the meantime, given the size of the debt and deficit, the chair manufacturers see the borrowing as a tax. So now your actual effects are countering the changes the economy is trying to make. Rather than getting more chair manufacturers and fewer table manufacturers, you're taxing chair makers to prop up table makers, which will constrict the chair industry and prop up the table industry. You may see some slight rise in table purchases, but once the stimulus stops you're left in exactly the same situation and the economy crashes again.

It seems to me that this is *exactly what the stimulus has done. The real job creators are entrepreneurs and small businessmen. In the meantime, we have a large over-supply of housing and possibly too many car makers. We have a lot of people without training in things that matter to the economy, and with useless degrees no one wants to pay for. We have a bloated public sector. So what does the government do? It borrows money, and uses it to bail out car makers, the housing market, and to prop up salaries and jobs for government workers. That may get you some small, temporary bump in the economy, but once the stimulus is gone, you're left with the same problem - except now you have additional debt dragging down everyone.

In addition, your centrally-planned projects pull productive resources out of very key positions throughout the private economy, killing projects (or forestalling projects that would have happened except for the loss of key people) and possibly putting other people out of work. You've injected uncertainty into the economy by destroying price information. You've mucked up supply chains with fiat orders and caused companies to restructure to a new 'reality' that is temporary - only they don't know it. Every intervention, every company bailed out, every new program causes business plans to go in the garbage.

This is the 'unseen man' - the myriad small businessmen, venture capitalists, entrepreneurs, and private investors who find their plans thwarted, their risks raised, their estimation of the meaning of price movements damaged by the noise of fiscal injection. They're the ones who have stopped hiring. But the loss of jobs in a private contracting company that had to shut down because it lost key people to a large company's stimulus project is never seen, never recorded, and doesn't show up in the 'jobs created or saved' statistic.

Don't get me wrong - I don't think this is the entire story of the economy. My point was that I believe this is a significant factor that cannot be ignored if you really want to understand what a fiscal stimulus does and whether it's a good idea. Aggregate numbers and IS/LM curves have their uses, but we must never, ever lose sight of the fact that they are at best a gross simplification of a very complex, probabaly unknowable economy and attempts to manipulate them through policy should be approached with great caution.

Wouldn't knowing more about who was poached and why tell us if the potential problem you describe is really a problem?

Also, why is the problem you describe so much better if it happens purely because of actions in the private sector, which of course seems entirely possible if not absolutely

This is a good post, but one thing.

You say you have many job openings, well, that may be true for you. But the aggregate amount of jobs openings in the economy is far, far below the number of unemployed workers, not only in the U.S. but everywhere else. This suggests there is a demand side problem.

There are certainly structural factors, don't get me wrong. But it's not all structural.

So, DHanson, you're in favor of job retraining.

Good for you.

We can do both stimulus and job retraining.

If you think this is simply a matter of 'job retraining', you're fooling yourself.

First, you have to understand that what's lost isn't just the formal knowledge of a subject in a user's head that was learned in a classroom. What's lost is years of experience, domain knowledge, knowledge of the company's assets, which employees can be trusted to carry out a task and which ones need supervision, which ideas look good on paper but don't work in that particular industry, etc.

If you lose an IT guy and replace him with an identically-trained IT guy, you've still lost a great deal. You've lost the first guy's knowledge of which systems were troublesome, which users needed a little more handholding, which managers were unlikely to approve an upgrade because they don't know what they're doing, and on and on. People are not robots - they adapt. They build on local knowledge to make their jobs more efficient. That's one reason wages are sticky and people don't get laid off in recession even when they are currently not productive - there's a real cost to the company of losing what the employee is carrying around in his head.

The other thing that confuses me when people suggest that 'job retraining' is the answer - just how long do you think this retraining should take? Do you think an ex-financial guy can become a productive bricklayer after a 12 week training course or something? Hell, Obama's 'green' program partially failed because it couldn't get people trained in the application of weatherstripping in a timely fashion.

Today, to be productive in a technical job (including blue collar jobs like plumbing and electrical work) requires not just training, but years of apprenticeship. Even jobs that used to be simple now require extensive knowledge.

Based on what you said about your firm, then, they can't find people and they can't train people.

Can I short your firm?

We train people all the time. That's another reason why people aren't easily interchangeable. We not only need people with the right resume, but once we hire them we have to train them in numerous standards, procedures, techniques... They have to learn how to use our computer systems, how to log all their data, how to read industry-specific documentation, yada yada. Again, if a stimulus project pulls them away we have to go through all the training with a new hire. Replacing an existing employee with a new one is a very expensive thing to do.

What we can't do and don't do is hire people with a degree in political science and train them to be computer engineers. And we don't hire people trained in database management and put them in jobs requiring knowledge of real-time systems and machine SCADA. Those skills are just too far apart.

But I'm guessing from your flippant answer that you don't have any real, substantive comments to make about the points raised by myself and others, including the authors of the linked papers.

DHanson, So now you are also saying that on the one hand Arra didn't create jobs, and on the other hand was responsible for hiring away employees. Interesting.

Hi Bill. Have you been following this thread at all? Have you glanced at the paper? You're last comment strongly suggests you haven't. This is a paper that suggests stimulus dollars are indeed hiring people...just people with jobs mostly. And we shouldn't really care to much about them. Your comment seems to miss the general topic of conversation entirely. "So you have no frame of reference here, Donny. You're like a child who wanders into the middle of a movie and wants to know..." How about reading through it, thinking about your last comment, and then jumping back into the fray. You're trolling Dhanson, who has offered some substantive thoughts and ideas, with mind vomit.

AlsoBill, I absolutely read both papers. I have defninitely been following Look at my comments on the prior post where I point out some of its errors. this is the post earlier in the day The error here is that, as others have pointed out in the prior post earlier today, that if A loses an employee to B, A hires, so on net there is an increase. See the post on the earlier today for economists criticisms of the report.

Also, AlsoBill, my comments have been directed at the report which said that some respondents did not hire or that they did not save a job. In the report, they identified 18 firms in this category. In a different part of the report, and for a different reason, they disclosed that 6 engineering firms reported that they did not hire/save jobs as a result of ARRA. Well, engineering firms are complementary to construction companies--those construction firms are the folks on the ARRA project that created the job; they are the complements to the engineers on a project that includes both engineers and construction companies. So, if the universe is construction projects, then half of the number of firms will respond saying they didn't create or save jobs. I have challenged the authors to disclose the nature of the firms who did not report that they save/hired under ARRA--I suspect they are suppliers or complementors to those who did hire and save jobs.

As to Dhanson, he says one moment that he is looking for 100 employees. Then I say, you should support retraining. Then he says his firm can't retrain or hire people. Then I say, can I sell his business short. Then he says jobs are specialized. Does he say that he lost employees to an ARRA project. No. But, that doesn't stop him. Who is trolling.

Oh, and AlsoBill, you should also note that these reports also claimed that technical engineering firms did not hire as a result of ARRA (so much for the technical employee loss argument), and that employee switching rates were no different before, during or after ARRA. So much for the ARRA poaching inducement argument. And, as others pointed out, a person who might not have been able to stay on at a firm, and was a week from being let go, is reported as being hired away. But, I don't believe this argument. This is primarily construction and you can believe what you want, but I don't believe they are yet back on their heals.

Someone with economic knowledge + actual experience in a firm vs. someone with (to be completely generous) economic knowledge only. I'm going with the first guy. Dhanson knows what he's talking about, and Bill doesn't even begin to understand what Dhanson is talking about.

Some economist should write a paper with Dhanson. That's how valuable his information is. Bill is a waste of pixels.

Troy, My clients are Fortune 500 firms. I don't have to own one to watch what they do. And, in my line of work, antitrust, I get to read a whole bunch of marketing plans and 5 year plans whenever a merger is investigated or challenged.

DHanson spoke on AARA as if his company lost employees to companies that won AARA bids. My bet is that DHanson's company has not relation whatsoever to any business that won an AARA contract, and therefore could hire away any of his employees.

You'd be wrong. My company is one of the larger recipients of ARRA funds. A lot of the job losses I mentioned are actually internal - people pulled away from one division at high cost because they were needed for a stimulus project. The company only accepted the high cost of transfer because the government stimulus money was worth even more.

So for example, let's say we received a $50 million contract to build something under ARRA. We could only successfully bid in time by pulling a project manager from a $5 million dollar project and three senior engineers from other projects. The cost to the company of that restructuring might be $2 million dollars, but it's worth it to chase the $50 million, which would result in a profit of maybe $5 million.

What never shows up in the government statistics, however, was the $2 million cost, which might represent low-level developers being laid off, or other projects canceled that would have hired people but now won't because we don't have a project manager to spare to start it, or projects that will have their schedules impacted because some of the expertise driving them was lost. So we might hire 50 people for the new stimulus project, but lay off 30 people in other projects that had to be sacrificed to go after the bigger one. Those 30 people layed off will not be counted against the 50 jobs 'created'.

DHanson, Now you have something to say and are credible in saying it.

But, I am a bit confused.

Did you compete for the ARRA project? If you did, and won, then you would necessarily be taking resources (employees) away from a competitor, or even internally, to a more valued use. The cost you identified from moving employees from one project to another, you clearly stated, was covered by the ARRA contract you won. So, if you allocate those costs to the ARRA project, your other projects were as they were before. Now, what you seem to be saying is that to acquire more people, you pay a cost--but that cost too should have been allocated to the ARRA contract when you first bid on it, because otherwise you are not optimizing your overall profitability when you took the ARRA contract, or you should have bid higher to reflect these costs. In other words, if it wouldn't have been profitable for your enterprise as a whole to bid on it, you wouldn't have bid on it, or your would have bid higher to cover these internal costs. Your firm has fixed costs and marginal costs, and as long as you are covering fixed, you should be able to do both projects and when you bid, you should be considering what changes in marginal costs you will incur when you make the bid, otherwise it would make more sense not to bid at all. What you described, to put it kindly, made no sense at all because you were not considering marginal costs to the entire enterprise when you made your bid.

One further point you ignore in your analysis is the following, which can be best explained by posing a hypothetical: Assume a valley in which there are only two construction contractors, each employing 10 people. Suddenly, aggregate demand for construction in the valley drops because of a financial crisis and an unwillingness by banks to lend because consumers are skittish. If left without any intervention, total demand for each firm would decline by 50%, with each firm laying off 5 employees. Total construction employment will decline in the valley to 10 employees from previous 20. The government offers an ARRA contract that will employ 6 people. Both firms bid, and the lowest cost firm wins. It keeps 10 employees and hires 1 employee, which either comes from the other employer, or from the surplus. Total number of workers hired is now 16, whereas it would have been 10. Someone says: you took the other firms employee: whether or not you did, that employee went to the highest use and the most efficient firm, because both firms bid on the project. So, if one employee leaves the other firm, he is going to a higher use, and the other firm then replaces that employee.

You should have more faith in competitive markets, even if demand patterns are temporarily changed to foster employment..


There is a difference between what's good for the company and what's good for the economy as a whole.

First, assume that we have a finite supply of top level managers and senior engineers. Within the organization, there is scarcity for their services. This is absolutely true - most of these guys are overworked, and some of our projects are in the weeds because our managers can find enough cycles to focus on them because they're stretched so thinly. Such people are also rare, and generally fully employed. That's why they get paid the big bucks.

Second, understand that when bidding on contracts, or in the initial phases of setting up a new project if a bid is won, the people involved are almost ALL high level people. To cost out a job, you need architects, project managers, business leaders, forecast people, etc. There are no ZMP workers here - not even close. 80% of these guys will be in the top 10% of contributors to productivity in the company. If the bid is won, The initial phase of the project will consist of a small number of high-value people, and can go on for anywhere from a couple of months to a couple of years, depending on the size and complexity of the project. Only then will lower level people even be brought onto the team.

These are also not people who can be easily replaced. Each one of them has been in the company for years or decades, and has intricate knowledge of operations. They're also very high performing people. You can put an ad out for a new manager, or take a chance and start a chain of promotion, moving people up into the vacated roles, but then you have a lot of people who have to retrain, take management courses, try to get up to speed on projects and all the rest. It's very hard to do.

So now along comes the government and says, "Hey! If you successfully bid on this project, we'll give you $50 million dollars!" What are you going to do? Your choices:

A) look for people who are currently running projects that are employing people but ultimately won't return that much, kill the project, and poach them for the new one, or

B) Pull a person here and there from existing projects, and hope the gaps can be filled in or schedules won't be affected too harshly.

B) say "No thanks" and don't bother going after the government money.

If the money is big enough, and you think your chances of winning the contract are high enough, you go for option A or B. BOTH options have significant cost, but in this case, the company has determined that the potential profit warrants the risk of bidding. So the company restructures, bids on the stimulus project, and gets it. Now it has to pull even more people from other projects to work on the new one. In some cases, they may be able to re-hire those people, but in some cases there may be too many shortages of key personnel and they just shut lesser projects down, or stretch out their schedules, or scale back the scope.

The point is that all of these options have costs associated with them. If you manage to rehire people, you have added HR costs, you have losses while you get the employee trained and productive, etc. If you promote people to the management positions, that leaves a gap below. Lots of reshuffling, a loss of efficiency (at least temporarily), and lots of costs. We have projects that have hundreds of people on them - people that earn $50/hr or more. If chasing a stimulus job costs even one lost day in the schedule, that's tens of thousands of dollars. But if things really go bad because key guys are gone, and the schedule slips by a long period, the cost could be in the hundreds of thousands or the millions.

Now, that's fine for the company - If it costs $1 million to chase a $50 million contract, and the company has a 50/50 chance of getting it and if it gets it it will make a profit of $5 million, that's a good decision from the company's standpoint.

However, that $1 million cost is internalized and doesn't show up in the statistics - it may not even be discoverable. For example, sometimes when a schedule slips it causes the cancellation of a project that was scheduled to start next year, because the resources won't be free in time. So the hiring that was going to occur next year for the new project never happens, but it's impossible to connect it to the stimulus from outside analysis. Only the guys inside the company know why that project was cancelled, or even that plans for it existed at all.

Then consider the ripple effects. If we blow a schedule, the customer who was counting on delivery doesn't get it, and that customer in turn loses revenue they were counting on from leveraging our product and perhaps lays off people. If what we're building is an intermediate product in the supply chain of someone other company's product, that product is now in trouble. And so it goes.

Now, I can't quantify these effects across the economy. My gut tells me that secondary costs like this are not insignificant, and in fact may be so large that they totally destroy the value of the stimulus. At the very least, I know that estimates of jobs created that don't take these effects into account will be higher than they should be.

Papers like the ones linked in these recent posts are a good start at trying to shake out some of these effects and get a handle on the unseen costs of fiscal stimulus. They're not going to be perfect, because they're tackling a very difficult problem. But they're better than just assuming that the effects of the stimulus can be perfectly described by models that use simplified aggregate values.


All of what you say may be true: but, no one forced the company to bid on the ARRA contract, and when it bid, it must have (unless it is stupid) considered how the additional work would affect the efficiency of the overall enterprise, how it would affect costs (both real and opportunity) when it made the decision to bid.

You haven't disputed that, to your credit.

Their bid is a volitional, competitive act, which has to include or anticipate those costs. It can either include those costs or refrain from bidding.

Now, you can say that management is stupid, didn't consider all costs, etc.: but think of this : perhaps they are just trying to get more work out of you and make the place more efficient.

Those are all legitimate problems, but hardly ones that can't be overcome.

I should point out that this kind of restructuring, schedule slip, and canceled projects due to lack of human resources happens all the time, stimulus or no stimulus. It's part of that 'creative destruction' that goes on in the economy. If a new demand props up that's worth more than what we're currently investing in, we may abandon a project and go after a new one. Sometimes a large customer can make a demand that causes a schedule to slip, but the customer is valuable enough or waving enough money at us that we'll agree to make the change.

The difference is that when these things happen as a result of real market forces, we're actually moving resources for lower value use to higher value use. In the end, the economy gains more than it otherwise would have had we not restructured. That has to happen because if it were not profitable, we wouldn't do it.

But when government throws money around by fiat and doesn't care whether it's going into production of things that people actually need (or that they need more than the thing our company was already building), those costs are not recouped by the economy. There's no higher-value output to compensate for the losses required to change direction. So the economy loses value. That's the loss that is not captured in stimulus estimates.

You seem to be making a political assumption here: that government never produces things that the economy actually needs. Sometimes a project funded by stimulus dollars is actually really needed. In these cases, the costs are recouped by the economy, usually over a long period of time.

If a new project causes gaps and delays but forces people to learn new skills and try new roles to make up for gaps, doesn't this increase overall human capital? It also seems like there could be some (at least) temporary hiring to replace some of the lower level work previously done by people who are now doing higher level work to fill gaps left by experienced managers. This might be an ugly process, but it increases both employment in the short run and human capital in the long run if done right, and these seem like useful goals of government policy in a down economy that has structural weaknesses.

The assumption that there are a finite number of senior people/project managers is highly questionable. Are there really that few mid-level people willing to work very hard to quickly learn the skills needed for a higher level role? I thought moving up to a senior role led to a large increase in pay, which would be a strong motivating factor? I know somebody who moved up the ladder quickly into a high paying senior engineering job, really didn't enjoy the stress but saved the money he earned, and is able to finish his career with very low stress/lower paying work or part time work. Maybe this kind of career path needs to be advertised as an option to attract more people into senior roles...

Based Upon this: http://www.recovery.gov/Pages/default.aspx : It looks to me like you studied the $223 Billion in Contracts, Grants, & Loans. If so, did you study all three?

It looks to me like the money was well spent: Here's why: It is not clear that Govt Funded Projects will work the same in every situation. Following Debt-Deflation, where workers have been laid off preemptively, Fisher's model would lead some of us to believe that overcoming the shell shock of the downturn would be harder in the current case. But, in any case, we didn't know how well it would work, nor are we certain as yet. Therefore, it's still a mixed bag, & think it was worth a shot. After all, we might well find ways to better target such funds next time, even if we don't spend as much.

Also, in my view, these funds act as a Placebo Effect. But, to study that, you need to broaden the study to include the fact that the Placebo might affect various groups of people differently.

So, finally, given the amount spent as a percentage of the ARRA, if I understand the figures ( & I can easily be wrong ), it looks like a worthwhile attempt.

Tyler, you write a decent amount about how fiscal stimulus will not help, and obviously monetary policy isn't helping. I'd love to hear more explicitly about what your actual policy suggestions are: so far, what comes to mind is "workers should accept lower wages", and I can't think of anything else. But that itself isn't even policy.

There seems to be a lot of silly, unwarranted assumptions being tossed around here.

For example, the assumption that the worker hired is a ZMP one therefore he would not have been replaced when he left his job with Company A to take a stimulus job with company B.

1. There's no good reason to think a worker who took a stimulus job would be any worse than the typical worker. Therefore unless you beleive *all* workers are producing zero value (then who is creating GDP? robots? invisible angels?), they should be no worse than the average worker's product.

2. If the worker is ZMP, his job with Company A is likely to be very unsecure. Just because he hasn't appeared on the unemployment rolls, the fact is the stimulus job he took is likely counts as a job created for him.

3. The worker was more likely IMO to be above ZMP. After all in an unsecure economy where many people cling to their jobs he choose to leave an existing position to take a stimulus job....he may be over confident but if we assume rationality on the part of economic actors he is likely a valuable employee and knows it. Therefore Company A would have probably had to fill his empty postion with another worker. Would this have imposed some search costs on Company A? Possibly but think about it, we have an environment where all a company has to do is put up an ad that they have one job and they will get 1000 applications, half from people who are overqualified. Companies feel free to advertise "don't bother applying if you aren't already working". Many stimulus projects were ideal work for blue collar males, a demographic that's been hit especially hard by this recession. The search costs on Company A to replace the worker who jumped ship is likely minimal.

4. To the degree that 'wage stickiness' is an issue holding up unemployment, churning in the labor market is good to 'unstick' prices. The ability for firms to renegotiate wages with new workers (which is what company A would be doing when it fills its empty position), is exactly what someone concerned with 'wage stickiness' would want to see. Likewise the employed worker who opts to leave his established job for a different company, different project, might be more willing to accept a smaller wage increase in exchange for a different set of opportunities a new employer may be able to offer.

5. Only 42% of jobs went to the unemployed? The unemployment rate is/was about 10%! Assuming a frictionless market where employed and unemployed workers look more or less the same one would think out of 100 jobs created 10 would go to the unemployed and 90 would go to employed workers who leave their jobs. This figure actually cuts against the ZMP concept! If unemployed workers were about equal to employed ones in their marginal product, one would expect the new jobs to be filled by a 90-10 split between already employed and unemployed. If unemployed workers were really lower then average in terms of their marginal product one would expect even more of the positions to be filled by employed workers. OK the fact that an unemployed worker is more likely to be searching and applying for jobs than an employed one cuts the other way, but still 42% of newly created jobs worked at cutting down unemployment. Of the 58% of newly created jobs that were filled by employed workers, if those workers were on the verge of getting laid off (say because they were ZMP), then again those new jobs worked on cutting down unemployment. If those 58% of newly creatd jobs were filled by workers who were needed by their original employers, then a good portion of them were likely filled by unemplooyed workers still lowering unemployment.

You can tell a story about how hiring the already employed opened up other jobs for the unemployed, but it’s just that — a story. I don’t think it is what happened in most cases, rather firms ended up getting by with fewer workers.

In other words, your model here is that the firms were already highly inefficient keeping people on the payroll for no reason. When those people left to take stimulus jobs, the firms opted not to fill those positions because it was possible to make do with less. If that's true then that's actually a positive thing. If the economy is able to produce more output with less cost then stimulus spending will not cause either inflation or crowding out. So what's the problem?

Yea it would be nice if pushing the economy to full employment also solved the 'hard core unemployment' problem but isn't this making the good the enemy of the perfect? To the degree we have a problem with ZMP workers or hard core unemployment, wouldn't it be better to look at those problems minus a borderline depression?

Also let me get the hypothesis being presented here straight: Firms already have employees on the books. Stimulus bill releases funds for a host of various projects. *Other* firms bid on those projects, win and put out help wanted ads in order to do the work. People in the original firms quite their jobs and take jobs in the *other* firms... Most importantly, according to Tyler the original firms simply don't bother filling up the positions that are emptied because those firms discover they can get their work done with fewer people.

Errr, why wouldn't those original firms simply bid on the projects and just use those 'ZMP' workers to do them? They would earn additional income from the contracts and make their current labor force more productive.

Look a more obvious hypothesis here is that the stimulus worked, we should be doing much more of it for an extended period of time to drive down the unemployment rate unless inflation or interest rates start to go up. At that time, and only at that time, should we start studying how extensive the 'structural problem' is with hard core unemployed. In the meantime this seems like a lot of pointless nitpicking by people who either don't take the recession very seriously or who think the recession is somehow a good thing but don't want to actually come out and say that (maybe divine punishment for overconsuming....or perhaps political punishment for the nerve of the 'little people' for electing, horror, a democrate as President). If the house is on fire you turn the hose on it. You don't sit around trying to find the Home Depot receipt for the drywall to make sure its water resistant.

Comments for this post are closed