Let's imagine we are in a world where central banks are given key roles in the macroeconomy, and have been for decades or even centuries in almost every country. In this imaginary world, studies into the relative efficacy of free banking regimes have been undeservedly overlooked, and the orthodoxy among major economists, even ones otherwise sympathetic to free markets is that they are a bad idea. Major policymakers, let's imagine, are completely unaware of the free banking alternative, and most even use the term to mean something completely different. Proposals to enact free banking have not been mentioned in law making chambers for decades or centuries, if at all. It has not been in any party's policy platform for a similar period of time, in this imaginary world.
What's interesting about this imaginary world is that it is in fact our world. Economists like George Selgin, Larry White, Kevin Dowd (among many others) have done very convincing research about the benefits of free banking. And free banking may one day become a real prospect, perhaps in a new state or a charter city. But free banking has lost the battle for the time being, and abolishing the central bank and government intervention in money is as unlikely as abolishing the welfare state. Now one might say that if free banking is a desirable policy, it is worth continuing to wage the intellectual war for the benefit of future generations, who could benefit from the scholarship. Work done now could end up influencing and improving future monetary policy.
I do not discount the possibility this is true. At the same time, free banking is a meta-policy, not a policy—a way of choosing what monetary regime to enact, rather than a specific monetary regime. After all, it is at least possible that free banks could together target consumer prices, the GDP deflator, the money base, the money supply measured by M2, nominal income/NGDP. And for each of these different measures there are an infinite number of theoretical growth paths, and a large number of realistically plausible growth paths they could aim for. Now, free bankers say that the market will make a good decision, and I can buy that. But let's say we're constrained to choose a policy without the aid of the market mechanism: can we say there are better or worse central plans?
The answer is: of course we can! Old-school monetarism, targeting money supply aggregates, was a failure even according to Milton Friedman, whereas CPI targeting, for all its flaws, delivered 66 quarters of unbroken growth and a period so decent they named it the Great Moderation. The interwar gold standard brought us the stagnation of the 1920s (in the UK) and coming off us brought us our relatively pleasant experience of the Great Depression. Literally the order in which countries came off the gold standard is the order they got out of the Great Depression. And even though the classical gold standard worked pretty well, few of its benefits would obtain if we went back. Some central plans (the interwar gold standard, M2 targeting) don't work, some work a bit (the classical gold standard, CPI) and arguably some work pretty well (NGDP targeting is one in this category, according to Friedman, Hayek and I). If we are stuck with central planning, then why not have a good central plan?
And just because I'm allowing the term "central planning" to describe NGDP targeting, we needn't describe it as "government intervention in money". I don't think they are really the same thing. "Government intervention in money" brings to mind rapid inflation, wild swings in the macroeconomic environment; in short the exact circumstances that NGDP-targeting aims to avoid. Targeting aggregate demand keeps the overall macro environment stable—a truly neutral monetary policy—allowing firms and households to make long-term plans, and preventing recessions like the last one, caused as it almost certainly was by drastic monetary tightening. Indeed, as monetary policy determines the overall path of aggregate demand, we might easily call "sound money" policies aiming for zero inflation or a frozen base as dangerous government meddling—they allow the actually important measures like nominal income to fluctuate drastically.
Consider an analogy: school vouchers. Many libertarians may favour a system where parents can spend as little or as much as they want on schooling (considering distributional concerns separately), rather than having central planners decide on the voucher-set minimum. But we usually see a voucher system as an improvement on the status quo—parents may not be able to fully control how much is spent on their children's education but at least they can pick their school. Popular and successful schools grow to accommodate demand, while unpopular and unsuccessful schools can be wound down more quickly. Libertarians may see this as a way from the ideal situation, but none would therefore denounce the policy. The analogy isn't perfect, but I like to see NGDP targeting as similar to school vouchers, versus status quo schooling as the CPI target. Libertarians shouldn't make the perfect the enemy of the good.