This is speculative. I don't know whether it works empirically. Or, I should say, I don't know how much it works empirically. The effect I am talking about here might be large, and explain almost everything. Or it might be small, and explain almost nothing. I'm just throwing it out there.
Legally required reserves, if those reserves pay low or no interest, are a tax on banks. For a 10% required reserve ratio, where required reserves pay 0% interest, for every $100 in deposits a bank issues, it must give an interest-free loan of $10 to the government. If market rates of interest are (say) 5%, that is equivalent to a 0.5% annual tax on deposits held at banks.
The main reason that Canada got rid of required reserves 20 years ago was precisely to eliminate this tax on banks, and to 'level the playing field' between banks and other financial intermediaries. (And because the Bank of Canada figured it did not need required reserves to help it control monetary policy.) The Bank of Canada also pays interest on the very small amounts of reserves the banks do choose to hold, and that interest rate is only 0.25% below the target for the overnight rate of interest.
If you put a tax on apples, and not on pears, and if apples and pears are substitutes in both supply and demand, the market will respond by producing fewer apples and more pears. The pear industry will expand to avoid the tax on apples. We will see more pear trees, and fewer apple trees.
I do not pretend to understand the shadow banking industry, but for my purposes here all I need to assume is that "shadow banks" are a close substitute for banks, that are not regulated like banks. In particular, "shadow banks" are not legally required to hold reserves, so are not legally required to make an interest-free loan to the government, and so do not pay the reserve-tax that regular banks are required to pay. We would therefore expect that required reserve ratios would make the shadow banking industry larger than it otherwise would be. (And the regular banking industry smaller than it otherwise would be.)
You can think of this as just another application of Milton Friedman's Optimum Quantity of Money. If currency pays 0% interest, while market rates of interest are positive, so the central bank earns monopoly profits by issuing currency, that is like a tax on holding currency. The equilibrium (real) quantity of currency is smaller than the efficient quantity. Regular banks, with less than 100% reserves, are a way to avoid the tax on currency. But if regular banks are required to hold positive reserves that pay low or no interest, then bank deposits pay some of that tax too, and shadow banks are a way to avoid the tax on regular banks' deposits.
Again, the question is not so much whether this effect exists empirically but how big it is. And that will depend on the degree of substitutability between banks and shadow banks, both on the supply-side and on the demand-side. A tax of (say) 0.5% on bank deposits might sound small, but if it is a large fraction of the total costs of using one form of banking vs another, and if both demand and supply can easily switch between banking and shadow banking in response to small changes in relative prices, a small tax could have large effects.
Do countries (like Canada) that have no required reserves have larger ratios of regular banking to shadow banking, other things equal?
Do countries that pay interest on reserves have larger ratios of regular banking to shadow banking, other things equal?
Dunno. Maybe other things matter more.
Just some speculative musings on reading Thomas Palley's post (HT Mark Thoma).
I own bank shares.