Today's focus article looks at the future of the Fed's quantitative easing program (QE2)—or large-scale asset purchases (LSAPs) in the Fed's preferred terminology—after the recent backlash at home and abroad. In our view, the anxiety about the inflationary consequences of QE2 is misplaced. The reason is that the US economy and financial system is not well described by the simple "money multiplier" model, in which increased bank reserves automatically result in an inflationary lending boom. In particular, deposits subject to reserve requirements are much less central to the financing of US banks than assumed in this model.
Instead, it is better to think of QE2 as a swap of cash for Treasuries on the asset side of the private sector balance sheet. This swap should lower the bond term premium and thereby boost growth, at least to some degree, but it does not automatically lead to a large expansion in broader money and credit aggregates. Even if it did, the first effect in an economy operating with as much slack as the US economy currently has would probably be to boost output rather than prices.
Our central estimate is that QE2 is likely to boost real GDP by about ½% per $1 trillion (trn). While that is not a very large number, we believe that the benefits exceed the costs, at least so long as the political backlash does not restrict the Fed's operational independence.
However, the backlash probably does increase the hurdle for additional purchases beyond the $600 billion (bn) planned through June. Combined with the better data of recent weeks, it has increased the risk that Fed officials will ultimately stop short of our estimate of $2 trillion (trn) in cumulative QE2.
Where do we stand after all the criticism of the Fed's decision and the renewed tightening in financial conditions? Q: First of all, why do you keep calling it quantitative easing (QE2) when Ben Bernanke himself takes exception to the term?
A: Inertia, really. We fully agree with Bernanke that the impact of large-scale asset purchases (LSAPs) occurs almost exclusively via the asset side of the Fed's balance sheet—in particular, through the effects of its purchases on asset prices—rather than through the liability side (i.e. the quantity of reserves), as discussed below. But at this point it seems somewhat pointless to keep fighting this terminological battle.
Q: How does QE2 work?
A: The mechanics of QE2 are shown in Exhibit 1. Fed officials will buy $600 billion (bn) of longer-term Treasury securities and create $600bn in additional bank reserves in exchange. This increases the Fed's balance sheet as well as the so-called monetary base (currency plus bank reserves) by $600bn.
Q: So are Fed officials "printing money"?
A: If "money" is defined as the monetary base, the answer is yes. Each $1 in LSAPs will result in an additional $1 in bank reserves. However, if "money" is defined as one of the broader money or credit aggregates—which matter much more for economic activity and inflation—the answer is not necessarily. Unless banks decide to leverage the extra reserves by increasing lending, QE2 is simply a shift of assets held outside the Fed from longer-term Treasuries to cash and bank reserves.
Q: If that's all it is, why the worry that QE2 will cause much higher inflation?
A: The concern is that the increase in bank reserves will cause a sharp increase in bank lending, which in turn will boost demand and ultimately prices to an excessive degree. It is based on the traditional "money multiplier" theory, which holds that the availability of bank reserves is the key constraint on the amount of bank lending. When the Fed adds reserves to the system, the story goes, banks scramble to make more loans until these "excess" reserves have been turned into required reserves that are needed to support the higher level of bank loans. The supposed explosion in credit then leads to an unsustainable boom in demand and ultimately inflation.
However, we believe that the money multiplier model is a very poor guide in the US financial system. It stands and falls with the assumption that deposits subject to reserve requirement are the only source of bank funding, when in fact more than 90% of bank funding comes from other sources (including many deposits on which reserves are not required).
Thus, bank lending was not constrained by the availability of reserves even prior to the increase in bank reserves. Relieving a non-existent constraint cannot be expansionary, and it cannot be inflationary. Hence, the standard story for how QE will result in inflation is incorrect, in our view.
Q: Are you saying that QE2 has no effect on lending, output, and inflation? If so, why do it?
A: It does have an effect, but it works through a different channel. The shift in the composition of the private sector's asset holdings from longer-term Treasuries to cash should raise the price (lower the yield) of longer-term Treasuries. A lower Treasury yield should feed into a reduced discount rate and therefore a higher price of risky assets, and it may also weaken the US dollar. In turn, easier financial conditions—i.e. lower long-term interest rates, higher stock prices, and a weaker dollar—should boost economic activity.
Stronger economic activity will also deliver a (small) boost to inflation. In addition, there is likely to be—in fact, there has already been—a positive effect on inflation expectations. So we do think that there is a link between the Fed's actions and future inflation. It is just far less dramatic than in the typical "printing money" story, and much farther off in the future.
Q: If the effect is supposed to occur via lower long-term interest rates, why have Treasury yields in fact risen—and financial conditions tightened—since the QE2 announcement?
A: Probably mainly because the market has actually reduced its expectation of the cumulative amount of QE2. While the cumulative purchase announcement of $600bn itself was modestly ahead of market expectations, the somewhat slower-than-expected pace of purchases, the backlash against the decision at home and abroad, and the better economic data of recent weeks have created a lot more doubt whether Fed officials will buy more assets beyond June 2011.
That said, Treasury yields are still somewhat lower—and more importantly financial conditions are still significantly easier—than they were before the expectation that Fed officials would buy anew. As of Friday's close, our Goldman Sachs Financial Conditions Index (GSFCI) stood at 97.48, about 50bp easier than the level seen in August when the market began to shift to an expectation of QE2.
Q: How big is the boost to growth?
A: That's a hard question to answer, but our best estimate is about ½ percentage point of additional GDP growth per $1trn in LSAPs. Part of the reason why it is so uncertain is that the precise number depends on three separate estimates:
1. The sensitivity of the term premium to Fed purchases; 2. The sensitivity of financial conditions to the term premium; and 3. The sensitivity of GDP growth to financial conditions.
All of these variables are measured with considerable error, but our estimates are as follows:
1. A $1trn asset purchase lowers the term premium by around 30bp. 2. A 30bp drop in the term premium eases financial conditions as measured by our GSFCI by about 80bp. 3. An 80bp easing in the GSFCI normally raises real GDP growth by ¾ percentage point in the first year; however, given how clogged the housing channel is at present, we believe a more realistic figure may be ½ percentage point.
Q: Why do QE2 if its effect is so small?
A: For one thing, it's not that small in absolute terms. An extra ½% of GDP amounts to $75bn per year and perhaps corresponds to an extra 400,000 jobs. That's not a lot relative to nearly 15 million unemployed workers, but it's a start.
Moreover, monetary policy decisions always boil down to a cost-benefit calculation at the margin. The main perceived cost of QE2—or indeed any monetary easing decision—is a higher long-term risk of actual and expected inflation in excess of the Fed's 2% target. But while the risk of inflation does rise at least a bit with QE2, we need to set against this a reduced risk of deflation. Since deflation is probably the greater risk at present, the perceived cost is, on net, actually likely to be an additional benefit.
Q: Are there any other potential costs?
A: The most obvious cost is the political risk, including the possibility that Congress will curtail the Fed's independence and ability to deal with future crises. That is not our expectation, but it is not impossible either.
Another concern is that lower long-term interest rates may increase commodity prices, which would weigh on growth for a net commodity importer such as the United States. This could weaken—or in an extreme case even reverse—the positive impact of QE2 on growth. It is again difficult to assess the magnitude of this effect. However, we can glean some information from a 1999 Federal Reserve paper describing the impact of different shocks in the staff's model of the US economy, FRB/US. It suggests that a $10/barrel increase in oil prices lowers GDP growth over the next year by 0.2 percentage point. If so, it would take roughly a $25/barrel increase—a much bigger move than the $5-$10 increase seen in the runup to QE2—to offset the GDP effects of QE2.
Finally, there is some risk to the Fed's prestige if QE2 isn't followed by a visible improvement in the economy and the labor market. The best analogy is the stimulus package of early 2009. Although we (and most other economists) believe this helped the economy in 2009-2010, the effect was not large enough to overcome the underlying weakness. The package thus probably helped discredit activist fiscal policies in the eyes of many Americans. The problem is that it is difficult to distinguish between the "baseline"—the path of economic growth and employment in the absence of the stimulus—and the impact of the stimulus itself.
The Fed is running some risk of a similar outcome with QE2. If the recovery continues to disappoint, QE2 will undoubtedly be viewed as a failure, even if it in fact helped matters at the margin. However, the difficulty of distinguishing between the "baseline" performance of the economy and the effects of QE2 cuts both ways. If growth picks up, the Fed would likely receive part of the credit, even if the improvement occurs because the private-sector deleveraging process slows, or because of other non-monetary factors.
Q: So how close is the private sector to stronger, self-sustaining growth?
A: It is very difficult to have a confident view of timing, even though the qualitative economics of the situation is quite clear. Because the private sector is trying to pay down debt, its financial balance—i.e., the gap between its total income and total spending—has risen to a level that needs to be offset by larger government deficits than the public is willing to tolerate over anything but the very short term. This means that the economy as a whole suffers from insufficient aggregate demand during the private-sector deleveraging episode; in effect, the different sectors of the economy, when taken together, are "trying" to spend less than they earn, which is a recipe for economic weakness.
Once the private deleveraging has progressed sufficiently, the private-sector balance will fall and this will result in a boost to aggregate demand growth. At present, Exhibit 2 shows that the private sector balance currently stands at +7% of GDP, which is about 5 percentage points above the long-term average. A full return to the long-term average would be the equivalent of an exceptionally powerful fiscal stimulus program, but even a partial return would be quite helpful. The key condition for a drop in the private sector balance is a sufficient decline in the level of private-sector debt. Unfortunately, however, nobody has come up with a good model of what constitutes a "sufficient" decline. In the absence of such a model, we can perform simple "what if" calculations that assume a return of the debt/income ratio to a more or less arbitrary benchmark, or we can evaluate the experience of other countries that have gone through deleveraging cycles. In general, this type of analysis suggests that the process still has a ways to go, i.e. that the private sector balance will stay high for another couple of years. But given the lack of a clear benchmark for what constitutes an acceptable debt level as well as the significant differences between the deleveraging experiences across countries, it is quite a tentative conclusion.
Q: In the private sector deleveraging story you just told, there is not much room for monetary policy. Isn't that in itself a statement that QE2 is unlikely to have a big effect?
A: Yes, it is. We believe that in a balance sheet recession, monetary policy is inherently less powerful than fiscal policy because the private sector's attempt to reduce its leverage works at cross-purposes with lower interest rates. That doesn't mean that the effect is zero, because the pace of the deleveraging is likely to depend at least partly on interest rates and broader financial conditions.
In any case, however, it is better to think of QE2 as a type of "holding operation" that reduces the likelihood of deflation, which would make it yet more difficult for the private sector to reduce the real value of its nominal debt burden. QE2 is unlikely to turn the economy around, even though it probably does help at the margin.
Q: You have argued that the Fed may end up buying as much as $2trn. How confident are you in this estimate?
A: Frankly, not that confident. This is partly due to the inherent uncertainty in the economic outlook. Although we believe that the economy will still be fairly sluggish in coming quarters, the recent data have been a bit firmer than expected and faster growth is possible. This is important because our calculations for how much QE is "warranted" are quite sensitive to the economic inputs, i.e. the inflation and unemployment forecast one year ahead.
In addition, the backlash against the Fed's policy probably does raise the hurdle for substantial further purchases beyond June. In our analysis of the potential cumulative volume of QE2, we assumed that the Fed views purchases of assets as significantly more "costly" than an equivalent amount of short-term interest rate cuts. These costs include economic factors such as the tail risk of a significant increase in inflation expectations and/or the greater difficulty of exiting from the current highly accommodative policy stance. But they also include political factors such as an increased risk that Congress will curtail the Fed's independence in response to continued QE. At the margin, this is likely to reduce the committee's appetite for further QE beyond June.
Q: Is it possible that the committee will in fact decide to stop the purchases before they reach $600bn?
A: That is highly unlikely. Although the committee promised to review the policy regularly, we believe that the hurdle for actually stopping the purchases is very high. The only realistic scenario, in our view, would be a very substantial increase in inflation expectations to a level clearly above the recent norm. Otherwise, Fed officials would almost certainly complete the $600bn program.
Jan Hatzius