This is pure speculation (also called theory). I have no respect for economic theory definitely including my own efforts, so the post will all be after the jump.
I will write about, sketch and definitely not write out a model in which Fed purchases of 7 year Treasury notes causes low investment. It happens to be a fact that following the Fed’s purchases of 7 year Treasury notes investment has been lower than forecast when the purchases began. I don’t consider this anything along the lines of evidence, not even weak evidence, and am theorizing for the fun of it.
7 year notes are not completely safe assets. Real returns over the full 7 years depend on inflation. Returns over briefer periods depend on inflation and future shorter term rates. This means that QE2 might have an effect on the economy by removing risky assets changing the risk born by private agents. This is a plausible explanation for the apparent effectiveness of QE1 (the Fed bought mortgage backed securities, commerical paper and made loans to banks). It is one rational for quantitative easing in general.
The problem is that removal of a stochastic asset does not necessarily reduce risk. Insurance makes stochastic payments. It generally reduces risk (except for CDSs it turns out). It seems plausible to me that medium and long term Treasury securities provide a useful hedge for firms considering whether to invest in fixed capital.
If I wrote a model (which I won’t) uncertainty would be uncertainty about future real GDP growth. It is possible that GDP will grow robustly so that returns on fixed capital are high or that it will grow slowly (or there will be a second dip) so returns are low. Firms are assumed to be risk averse. Partly this is due to the administrative costs of bankruptcy. Partly it reflects a principal agent model as the CEO of a firm can’t hedge his or her exposure to the risk of bankruptcy by being CEO of a diversified portfolio of firms.
This means that uncertainty about the returns on fixed capital causes lower investment. Fortunately, managers can hedge this risk by buying medium and long term Treasury securities. It is clear (here assumptions about monetary policy and Taylor rules and such) that poor GDP growth will cause low short term interest rates in the future. Also the non model would have an expectations augmented Phillips curve if I bothered to write it down. Poor GDP growth causes low inflation.
For both reasons a firm with a huge pile of cash might prefer to invest some in fixed capital and the rest in medium and long term Treasuries in order to avoid risk and possible bankruptcy.
If the Fed removes long term Treasuries, this becomes less attractive. It becomes more attractive to keep the financial wealth of the firm in cash or t-bills. This reduces the hedge on the risk in the returns on fixed capital and reduces the optimum investment in fixed capital.
For this argument to work, it is necessary that firms be sitting on huge amounts of financial assets. As they are.
I am quite sure that a model along the lines sketched here could be written and would give the desired result that QE2 is contractionary. Importantly, the model relies on the assumption that quantitative easing consists of purchases of medium or long term Treasury securities and not, say, high yielding corporate bonds.
My personal sincere view is that QE2 was a total flop and that it was a total flop because of problems with quality not quantity. I really think that, for quantitative easing to work, the Fed must purchase a large quantity of low quality assets. I think that the FOMC had a heated debate and compromised agreeing to buy assets which they don’t normally buy, but agreeing on assets which were as similar to their usual T-bills as possible. This makes sense if it was required by law or if their aim was to minimize the bang for the buck. Otherwise it is incomprehensible.