Why can't Keynesianism work?
Keynesianism is the idea that during recessions/depressions markets can react to fear and over-correct. Government can make up for the fearful private sector over-cutting (and being less than efficient), by spending up to where levels should have been, and smoothing the reset. However, for Keynesianism to work you need 4 things to happen: (1) an information vacuum (2) you have to cut spending/programs as the economy gets better (3) you need to be replacing like jobs (4) you have to have near equal efficiency between government and private sector. Since none of those things happens in the real world, the Keynesian promises have never, ever, been realized.
Fear is the problem
To Keynes, the problem in a down economy was fear. Sometimes private sector would over-react and over-cut (e.g. markets weren't perfectly efficient). In those cases (depression/recession) unemployment would spike, the cost of labor would crash, and while things would eventually recover, there was more harm in that down-spike than there needed to be. So, his theory was that if in those down markets, government could spend up to where the economy should have been (the balance point), assuming we know what that was, and we could take up some of the slack (soften the fall from unemployment and create artificial consumption). While government still wasn't as efficient as private sector, it was less of a loss than doing nothing (and those jobs just not getting done). So in the short term (and micro-view), that's better than nothing. Once the economy went back up, you could CUT those government programs/jobs, the private sector would replace them with more efficient versions private sector jobs as things got better, and you changed a big dip into a smaller dip, which let you grow more overall.
(1) You have to exist in an information vacuum
If people SEE government spending (and borrowing), they're generally smart enough to react to it. Since they realize government spending means the government must either tax, print money or borrow (to fund that spending), thus they KNOW they’re going to get hit by either taxes or inflation: which hurts their purchasing power. So they react (sometimes in advance of the policies even kicking in), by saving more, spending less, or sheltering investements. That negative economic reaction swamps any of the theoretical benefits in Keynes algorithms, sometimes before the policies have a chance to be implemented.
Hayek believed that consumers were smart enough to see government spending and react. Keynes didn't factor that into his models. Guess which maps better in the real world?
Keynes positive multipliers (of spending) turn it into negative multipliers, if people are smart enough to know that government spending means inflation or taxes. Which is why spending didn't work in the information desert of the 1930's, but certainly can never work in today's world.
(2) You have to CUT government spending/programs when the economy gets better
If you don’t cut when you have the opportunity, all you succeeded in doing is replacing the more efficient private sector with the less efficient public sector, over time.
As Milton Friedman observed, "nothing is as permanent as a temporary tax/program".
Once you create a program it becomes a special interest and thus has impetus that not only keeps it around forever, but tries to make it grow. And in good times, everyone is more tolerant of that wasteful program, so there’s no leverage to implement a cutback. Which means that whole sector suffocates under the weight of a subsidized government competitor. (They reinvest, innovate and grow, less).
Thus if you didn't cut the program, you did soften the dip, a little... but you also soften the recovery by even more. The economy can never climb back to where the economy would have been without that less efficient replacement holding it back. Since any gain in economic efficiency at least partly translates to higher salaries, you lost a lot of earning/economic potential. Since it is just a missed opportunity and hidden cost, leftonomics pretends it doesn't exist. But we know better. We might not be able to quantify (or might disagree over) the exact value of a lost opportunity, but we should all agree that some potential was there.
(3) You really need to be replacing the same jobs you're losing
In the early 1900s, on the job training took hours. Whether working a hoe, hammer, or welder, a worker was a worker was a worker. There was lots of grunt work, and much less skilled labor. Assembling a car, digging a ditch, a monkey could do it. So creating productivity required a body. In the modern economy, it doesn't work that way.
Unions have made job mobility difficult in low-skill jobs, and most jobs aren't low-skilled union ones any more. More of our economy is built around high skill jobs (that have far less worker mobility). Replacing a computer programmer with a ditch digger doesn't work (or vice versa), nor does replacing a SEO Optimization person with an accountant work. So a government temporary bridge projects doesn't help out of work IT folks, or autoworkers, accountants and so on. Retraining programs can't solve recessions since, the training lasts longer than the recession itself. So all it can do is take potential workers out of the economic pool of contributor (even fractional contributor, if they're doing something below their skill level), and moves them to the pool of taker/burden for the duration of the recession (while they're in training). This magnifies the depth and length of the recession. (This was demonstrated with Obamanomics).
(4) Government has to be as efficient (or close to it) as the job it is displacing
If you replace 100 workers with 200, you lose 100 workers worth of money for the same output. That negative multiplier means the economy lost 100 workers worth of potential output. And it's the output (what we produce) that is real net value to the economy. That's where the famous Milton Freedman anecdote (that predates him) about replacing workers shovels with spoons comes from , it's not how many workers are working, it's how much they're producing that matters. And in the end, rarely, government programs start out as efficient, and they entropy quicker in a productivity sapping bureaucracy (there's no profit incentive for continual improvement in efficiency).
This is why Keynesian magic positive multipliers never actually worked in the real world.
The Keynesian version is that in recessions/depressions everything has to be perfectly efficient and instantaneous. Since it isn't, they see any lags, overreactions and inefficiencies as opportunities for government to step in and spend (stimulate) to where "things should have been", to smooth out the downturns.
Nice theory. FDR tried it in the great depression (believing in his "brain trust"), and the results were extending the depression by a decade, and we have many more examples of the failures of planners to be more effective than the free market. So while it's a great theory, it has never actually worked in the real world. Keynes was a brilliant micro-economist, with delusions of being a macroeconomist, but his religion of collectivism (authoritarianism) got in the way of understanding human nature (how people and thus economies would react) or the nature of governments.