You write, "because once you acknowledge real balance effects, you have to either keep the hoarding part and explain why the price mechanism doesn't work (sticky wages/prices) or discard the hoarding".
But the whole point is that wages and prices are sticky. Information doesn't disseminate and become ubiquitous and acted upon instantly. This is the big strength of Keynesian or New Keynesian economics; the assumptions that it depends on are far more realistic than the assumptions that real balance arguments and rational expectations arguments, and the like depend on.
It's far more realistic to assume, as Keynesian type theories do, that everyone in the world does not have every bit of information everywhere in the economy and cannot analyze all of it perfectly, with perfect expertise, instantly, when it comes to price and wage setting, and many other things.
It's much more realistic to assume that people severely lack information, and they acquire it relatively slowly, and don't analyze it perfectly with Ph.D. techniques, and that it takes time (which is money), and effort to analyze information, and there are costs and risks to changing prices and wages, and therefore prices and wages adjust relatively slowly and imperfectly.
These are much more realistic assumption to make about the economy, and that's why the empirical evidence is immensely stronger for Keynesian type theories, and for the effectiveness of Keynesian type prescriptions.
Real balances arguments, rational expectations arguments, and the like, only really work and are the whole story if it's true, or close to true, that that everyone in the economy has perfect information about everything, perfect expertise and education about everything, perfect rationality, and the ability to analyze and update all information constantly and instantly in their brains (although some of these make an exception for one or a few things like job search time). Obviously, we don't fit this bill with a majority almost picking W. in 2000, and actually picking him in 2004!
Here is Princeton Economist and John Bates Clarke Medal Winner Paul Krugman on this (from his book "Peddling Prosperity", page 47):
When Keynes published his theory of the business cycle, some conservative economists argued that there was no need for government policy to combat recessions because recessions would be self correcting. Their argument went as follows: In the face of high unemployment, wages and prices will tend to fall. This fall in wages and prices will increase the real money supply – that is the given stock of money in circulation will have steadily rising purchasing power. And this expansion in the real supply of money will in turn lead to an economic expansion.
Keynes did not deny the logic of this so-called classical argument; he was willing to concede that in the long run economic slumps would be self-correcting. But he regarded this self-correcting process as very slow, and as he pointed out in a widely quoted but rarely understood remark, "In the long run we are all dead." What he meant was: Recessions may eventually cure themselves. But that's no more a reason to ignore policies that can end them quickly than the fact of eventual mortality is a reason to give up on living.
On page 214, Krugman more specifically discusses what happens once you realistically take into account that it's far from the truth that all people are instantly calculating machines with perfect and constantly updated knowledge about every single bit of information in the economy:
Suppose now that in this imperfect world we once again play the Keynesian experiment of posting an increase in the desire of people to hold cash. Recall that in Chapter 1 we saw how an attempt by everyone at once to increase cash holdings leads to a general fall in employment and incomes. The same will be true here.
At this point, the monetarist says that wages and prices will fall, increasing the real value of the cash in circulation and curing the recession. But will wages and prices really adjust?
What if firms are reluctant to cut their prices, either because there are measurable costs to putting on new price tags (menu costs in the jargon of the new Keynesians) or because they just don't want to be bothered with rethinking their pricing schemes? (The difference between real costs of changing stickers and subjective costs of thinking is actually quite blurry – that's why near rationality may in a way be regarded as a higher form of true rationality.) If markets were perfect, a firm that charges too high a price would lose all its business. In a world of highly imperfect markets, a firm that charges a price a little too high gains almost as much from that higher price as it loses in sales; a firm that fails to cut wages has a gain in higher leverage over its workers that almost compensates for its higher cost; and so on. In other words, the costs to an individual firm of being a little less than totally rational and not cutting wages and prices may be quite small, small enough that reasonable people just don't do it.
And yet the individually reasonable decision not to cut prices in the face of a recession can have collectively disastrous results. If prices don't fall when people decide to hold more cash, then the slump in output and employment is not self-correcting. In an imperfect world, senseless things can happen to groups of people who behave sensibly as individuals.
The case for active monetary policy is now obvious. Suppose that we have slid into a recession, just as described. There is an easy way out: put more money into circulation, and spending, incomes, and employment will rise...
In addition to menu costs, information gathering and analysis costs and time, efficiency wages, and other non-behavioral (non-psychological) imperfections, there are also strong behavioral causes of sticky or relatively slow price adjustment. For example, many businesses are reluctant to cut prices until they can really be sure that they will be able to keep the price at that lower level for a while, because (especially for certain products) customers don't like price increases or jumpy prices. That kind of thing can hurt customer goodwill and loyalty. It is also well documented that managers facing substantial competition often put off raising prices because they don't want to be the first to do so.
One very interesting, important, and widespread behavioral phenomenon is that workers are much less resistant to taking a real paycut if it doesn't involve a nominal paycut. In other words, the empirical evidence shows overwhelmingly that most workers are far more resistant to a 3% paycut when inflation is 0, than they are to a pay freeze when inflation is 3%, even though they are basically the same thing.
Krugman notes this in his 1996 Economist article, "Fast Growth and Stable Prices: Just Say No":
Messrs Akerlof [Nobel Prize winner], Dickens, and Perry have produced compelling evidence that workers are indeed very reluctant to accept nominal wage cuts: the distribution of nominal wage changes shows very few declines but a large concentration at zero a clear indication that there are many workers whose real wages 'should' be falling more rapidly than the inflation rate but cannot because to do so would require unacceptable nominal wage cuts.
This nominal wage rigidity means that trying to get the inflation rate very low impairs real wage flexibility and therefore increases the unemployment rate even in the long run. Consider the case of Canada a nation whose central bank is intensely committed to the goal of price stability (the current inflation rate is less than 1%). In the 1960s Canada used to have about the same unemployment rate as the United States. When it started to run persistently higher rates in the 1970s and 1980s many economists attributed the differential to a more generous unemployment insurance system. But even as that system has become less generous the unemployment gap has continued to widen: Canada's current rate is 10%. Why? A Canadian economist Pierre Fortin points out that from 1992 to 1994 a startling 47% of his country's collective-bargaining agreements involved wage freezes. Most economists would agree that high-unemployment economies like Canada suffer from wage inflexibility; Mr. Fortin's evidence suggests however that the cause of that inflexibility lies not only in structural microeconomic problems but also in the Bank of Canada's anti-inflationary zeal.
It can also take a while for a person looking for work to understand and/or admit that the equilibrium wage in his field has dropped, and he has to accept less. It's not just bad luck that he's getting lower offers than were common last year, and instead the market has, in fact, changed. It can take a while for an individual to discover this, with all of the complication and random factors in the labor market. This is known as the Misperceptions theory.
To sum it up from Krugman's text "Macroeconomics", with Robin Wells (2006), "Wages do not fall quickly in the face of labor surpluses or rise quickly in the face of shortages. Almost all macroeconomists agree that wages adjust slowly to surpluses or shortages of labor." (page 381). And from the other Paul, Samuelson, you know the legendary Nobel prize winner, "Keynesian economists point to much evidence suggesting that prices and particularly wages move slowly in response to shocks, and few economists believe that labor markets are in constant supply-demand equilibrium. When the assumption of perfectly flexible wages and prices is abandoned, policy will regain its power to affect the real economy in the short run." (From Samuelson's text, "Macroeconomics", 18th edition, 2005, with William Nordhaus of Yale, page 364.)
The bottom line is that you can't take simple Austrian or classical models literally; that is you can't assume the actual economy is identical to the simple model. These models, and similarly simple ones, are just useful as a first step. An actual economic situation may have 10 major factors and many smaller, but still important, noisy ones. It's usually harder to understand such a situation if you just look at it as a whole. It can be useful instead to first look at a few of the factors in isolation, to understand better how those few factors work. But that's only the first step. You don't say from there, "Oh, that's how the economy with the many more factors works too". No, instead you next look at what happens when you add a few more factors. How much, and how, does this change things? Next you add more factors, and keep re-evaluating until you have considered all of the significant factors. You may find that some factors don't make a qualitative difference when you add them, or you may find that they do, or are likely to.
The next step, to help screen out mistakes you may have made in your analysis, is to look at the real world data, with lots of empirical studies, to see how well it fits your analysis.
A process like this is a lot more complicated than just acting like simple models are identical to reality, but this kind of in-depth and logical thinking has lead to great advances in our quality and quantity of life.
With regards to Keynes's views on wage decreases: Keynes' writings can be hard to interpret. The style of English is old, formal, artsy, and complicated, but there's no dispute among experts on Keynes that he thought in the long run prices and wages would adjust to an equilibrium (until the next disequilibriating shocks) where normal unemployment was restored, but as Keynes said, "In the long run, we're all dead" (so why wait, suffer, and lose wealth unnecessarily for a long time).
In the short run (without smart action by a central bank) funny things could happen like "The Widows Curse" or "The Paradox of Thrift", but these are not long run equilibriums, and some of them are very unlikely even in the short run. Most of all, they can all easily be stopped by smart action by a central bank (and/or fiscal action). For more on this, I recommend reading Krugman's 1997 Slate article, "Vulgar Keynesians".