How can government policy make us better off? To bring about a welfare improvement, there must be some collective action we can take through our government that cannot be replicated by the private sector. The government must have some particular advantage in the activities it chooses in order to be doing anything useful. If the government is bad at running coal mines, it should let private firms run coal mines, and if the government is a bad banker, it should stay out of the banking business. However, we know that the government has an advantage in doing some things. For example, I think we can all agree that the government has an advantage in running the army.
If the government is no better or worse than the private sector in some activity, then if the government engages in more of that activity this is irrelevant. The government's activity simply displaces the same private activity one-for-one. If the government and the private sector have exactly the same technology for producing coffee cups, the government cannot increase the supply of coffee cups by producing more, unless it drives the private sector producers out of business. This is essentially the basis for all the government neutrality theorems we know about. For example, the Ricardian equivalence theorem states that, if the government and the private sector are equivalent in terms of their ability to collect on their debts, then the timing of taxation does not matter. Less government saving is undone by more private saving. The Modigliani-Miller theorem in corporate finance works in the same way. A firm's financing decision is irrelevant because it is undone by asset-holders, under certain conditions.
Central banks were established because there was some consensus that the government (or quasi-government) has an advantage in supplying currency, and in running intraday payments systems. Economists sometimes question whether monetary systems could be designed where one or both of these functions could be carried out efficiently by the private sector, but those ideas have never gathered much steam in public policy debate. In the United States, the Federal Reserve System has an essential monopoly (with the odd insignificant exception) on the issue of circulating small-denomination securities (currency), either through an implicit prohibition on private note issue, or because the issue of private currency is unprofitable. The Fed also dominates intraday payments arrangements among financial institutions. The clearing and settlement of large-value payments is accomplished mainly through the exchange of reserve account balances.
Ignoring payments systems issues for simplicity, the Fed's typical actions matter because of its monopoly on currency issue. In normal times, excess reserves are essentially zero, and a swap of reserves for Treasury bills by the Fed effectively increases the stock of currency (a Fed liability) while increasing the stock of T-bills on the asset side of the Fed's balance sheet. The Fed, being a large intermediary that can do something the private sector cannot, can thus move market asset prices, in particular the overnight market interest rate - the fed funds rate.
Now, since November, the Fed has been engaged in something unusual - QE2 - which essentially involves swaps of interest bearing reserves for long-maturity Treasury bonds. In spite of what Ben Bernanke might lead you to believe, this is not business as usual. This is not about issuing currency to finance the purchase of T-bills. What is going on then? The Fed is financing a portfolio of T-bonds by essentially rolling over overnight debt. That's what the reserves are under the current circumstances. The marginal unit of reserves does not serve any transaction role in the payments system. It just sits overnight.
Now, this type of intermediation bears a striking resemblance to what Gary Gorton's "shadow banks" do. Shadow banking is about holding long-maturity assets (could be asset backed securities, but long Treasuries certainly work), financing these asset holdings with overnight repos (with the assets used as collateral), and rolling over the repos. The Fed does not put up any collateral to the holders of reserve accounts, as apparently these financial institutions think that the Fed will always be good for it. The Fed has never suspended withdrawal privileges (conversion to currency) on its reserve accounts, for example.
Thus, QE2 is essentially shadow banking and, as such, it is an activity replicated in the private sector. Thus, QE2 is irrelevant. But, you might argue that shadow banking is a risky activity. This intermediation activity involves borrowing short and lending long, so maybe if the Fed does less of this, displacing an equal quantity of private intermediation activity, then this will transfer risk from the private sector to the Fed. Not so fast. The Fed cannot take risk off the private sector's hands in this manner. Should short rates increase (under the Fed's control of course), then the Fed will earn less profits, and pass on less to the Treasury, which then has to deal with it. The Treasury is no better equipped to share this loss among private economic agents than is the private sector.
What does this imply for current Fed policy? The Fed essentially has the same tool it always has for implementing monetary policy. While Fed policy is usually characterized in terms of the fed funds target rate, now the relevant policy instrument is the interest rate on reserves (IROR). The Fed has all the control over policy that it needs by manipulating the IROR.
Normally, as the economy recovers, the Fed needs to tighten, by increasing the fed funds rate target so as to control inflation. The Fed could always keep the fed funds target rate low during the recovery, but this would necessitate open market purchases to support the low target, which would be inflationary. Under current circumstances, if the Fed remains passive while the economy recovers, by keeping the IROR at 0.25% for an "extended period," then as the private sector creates more assets that can be intermediated and transformed into liquid tradeable assets, this will displace reserves, and ultimately lead to increases in the price level and an increase in the stock of currency. Just as in normal times, the Fed's policy rate must increase to choke off the inflation.
The good news here is that, since QE2 is irrelevant, the Fed can reverse it without cost. Ideally, the Fed should sell enough assets to reduce excess reserves to zero, so as to be back in a regime that it understands better. Given the state of the Fed's balance sheet, this cannot be done by just selling Treasury bonds, as T-bond holdings are currently at about $1.3 trillion, while the quantity of reserves outstanding is about $1.5 trillion. Selling all of its agency securities (essentially identical to Treasuries) still will not quite do it (that gives another $130 billion), so the Fed would have to dispose of a relatively small quantity of mortgage-backed securities.
Now, some people will suggest that empirical evidence contradicts what I have just stated. If QE2 was irrelevant, what about those asset price movements that coincide with the QE2 announcements? Those effects are consistent with my story, in that the QE2 announcements carried news about the future path for the policy rate. Essentially, QE2 suggested something to financial market participants about the length of the extended period with the IROR at 0.25%. Clearly it could not have been a commitment device (except perhaps because the Fed does not like to admit mistakes), as reversing QE2 is irrelevant.
The implications of this for current monetary policy are fairly dramatic. Bernanke would have to admit that previous policy decisions were wrongheaded (though not necessarily disastrous), but that certainly beats living a lie. He would also have to accept that all his talk about the array of policy instruments at the Fed's disposal is smoke and mirrors. Quantitative easing is a sham. Term deposits can at best serve to make reserves less liquid, and therefore more costly for the Fed, as they will need to command a higher interest rate than the IROR. Reverse repos are irrelevant. Do you think the Fed will admit to its errors? I doubt it.