Thursday, February 17, 2011

Some questions for proponents of NGDP targeting

There is a growing body of economists now trumpeting NGDP level targeting as the best way to get the US economy out of recession. This idea seems to be propounded mostly by Scott Sumner.

Now that rates are at the zero bound, the Fed can no longer stimulate the economy via its regular policy tool – decreasing interest rates. But this fact, as its proponents say, does not limit the Fed’s ability to stimulate the economy via unconventional tools, i.e., NGDP level targeting. The basic idea seems to involve the Fed involving in programs to loosen monetary base, QE being such a program. As the monetary base gets larger, this increases everybody’s expectations of inflation, thereby getting people to spend more now, instead of later. The increasing monetary base, along with the additional economic activity nudged into existence by the increasing inflationary expectation, will then lead to an increase in nominal GDP.

NGDP targeting proponents seem to recommend a 5% annual NGDP growth as a target.

I have attacked and berated additional monetary easing numerous times in this blog in the last few months, but in the interest of open-mindedness, I would like to learn more about this NGDP level targeting and its policy assumptions. If I can get some answers from anyone reading this who may know more about the idea, I could change my mind and my violent opposition to it.

Firstly, operationalizing NGDP targeting, as I understand it, will be via an increase in bank reserves (since this is the policy tool left for the Fed to loosen monetary policy). As such, the way for this new reserves to go out into the broader economy is via increased lending/investing activity by banks.

Since the objective of an effective NGDP level targeting central bank is to grow NGDP by 5% a year, I present the following chart which shows where NGDP should go, from now until 2020. Given the annual growth, we can now calculate what incremental nominal GDP is added for each succeeding year from now until 2020.

Now again, if my understanding is correct, all this additional NGDP will be due to additional lending. Am I wrong? Inaccurate? I know I’m oversimplifying things, since I’m assuming no organic growth, assuming instead that every $1 of incremental NGDP equals $1 of additional debt in the system. But I’m also assuming that there was no additional deflationary counter-pressure necessitating a greater than $1 debt for each $1 NGDP growth. In reality, if NGDP targeting were to be successful, there would probably more loosening than growth in the early years, while gradually less loosening to growth ratio would be needed, as the economy begins to grow back its legs.

But for simplicity sake, my chart shows me that to get to the policy objective of growing from $14.7 trillion GDP in 2010, to $23.9 trillion in 2020, we would be adding about $9.2 trillion of new money/new debt into the system in the coming 10 years. Is this accurate? To put this additional debt into perspective, given 310 million US population, this means $30,000 new debt per capita in the next 10 years. That’s 30 grand of additional debt per man, woman, and child now living in the US in the next 10 years.

Now the next column assumes that banks follow the Basel-approved leverage cap of about 10x capital. That means over the next 10 years, for banks to be able to lend $9.2 trillion, it has to set aside/raise almost a trillion in new capital. Where will this new capital come from?

Now I’m oversimplifying again, since not all incremental debt will need some form of capital to back it. If US banks were to simply buy new government debt in the next 10 years to attain each year’s NGDP growth target, then no additional capital need be raised. But then, this means, QE and /or NGDP targeting, is really just a mechanism to allow for more fiscal policy action, and/or deficit government spending. In which case, it is compatible to greater fiscal policy. Is this an accurate description?

Now again, we’ve also seen instances of banks being able to lend without setting aside capital. All banks need do is pass the loans on to another investor who intends to keep it as an investment. Now we know that there are millions of people from the rising investor class in the developing world who could be ‘prime’ candidates to pass these new debts along to. They are people who likely still cannot distinguish a leveraged mortgage security from a regular fixed income bond, and who cannot identify a toxic asset if they were staring at it with their very eyes. These are people who ‘theoretically’ can fund the capital needs of the new lending, by taking it away from the originating bank’s balance sheet. Multinational US banks can do this right now. Now, is this side effect a possibility that proponents have considered, and is the possible meltdown of the rest of the world worth it, for the sake of being able to conduct this NGDP targeting experiment?

Update: Many thanks to Andy Harless for engaging me in the comments, and for reminding me that QE2 involves buying bonds not just from banks, but from the entire bond-buying public. Do read the comments for more analysis.

6 comments:

Andy Harless said...

1. Why do you assume that the new investment would be financed with debt, and why with bank debt in particular? If the Fed were to pursue an NGDP target using its current methods, it would start by buying up as much of the national debt as necessary. This would happen to have the side effect of increasing bank reserves, but the primary effect woud be to make safe debt securities scarce. The former holders of government debt would have to find something else to do with their wealth. Some of them are indeed banks and might choose to make more loans. But others are non-banks and would choose to buy other securities, such as, perhaps, corporate bonds. Whoever sold them those bonds would then have to find something else to do with their wealth, and so on. All asset values would rise and thus encourage the production of all kinds of assets including various kinds of equity (in which I include home equity, single proprietor's equity, and so on).

2. Moreover, many of the former holders of government debt and other assets would choose to invest their wealth abroad, thus reducing the value of the dollar. (Your image of banks selling off loans to foreigners has things backwards.) Net exports would thus increase. And both the increases in investment and in net exports would have multiplier effects, thus increasing consumption (without requiring additional debt, since that consumption comes out of new income). Although bank lending (with the attendant capital adequacy issues) would likely be part of the mechanism for raising NGDP, it is not a necessary component, and the outcome could be accomplished even if banks refused to lend, or were not permitted to lend, any of the newly created reserves.

3. I think, in any case, that you mistake the overall mechanism. A gargantuan investment/export boom driven by a shortage of assets, as I outlined above, should be considered a worst-case scenario, a threat rather than an intention. The supply of assets would be like a hostage that the Fed threatens to kill if the economy refuses to grow "organically." But if the threat is credible, then it will not be necessary to kill the hostage. If agents believe that the Fed will achieve its NGDP target, that belief will tend to become a self-fulfilling prophecy even in the absence of major Fed asset purchases. Businesses will spend because they expect to sell more. Households will spend because they expect to earn more. Investors will demand more private sector assets not so much because the supply of public sector assets will be severely reduced but because the anticipation of NGDP growth will improve the risk/return characteristics of private sector assets. I think our experience thus far with the relatively modest QE2 tends to bear out this "self-fulfilling prophecy" scenario.

4. Note that most economists advocating NGDP targeting (at least those advocating "level" or "growth path" targeting as opposed to "growth rate" targeting) would be more ambitious than your assumptions suggest, becuase most would advocate making the 5% target retroactive (perhaps to 2007), thus requiring a catch-up period. So I guess if you still want to imagine that the growth is financed with bank debt, you can make it look even more outlandish.

Rogue Economist said...

Andy, thanks for your comment. Your explanation shows me that NGDP targeting is really no different than lowering interest rates, in that it encourages investors, not just banks, to look for other investment alternatives. A concern here though would be that, similar to what happened in the previous low interest rate environment, investors would be taking on risks they don't understand, or risks that they transfer to other people who don't know any better.

And commercial banks may not as easily invest in other securities substitutes as the non-banks (they need adequate capital), so substituting their treasuries with reserves just lowers their interest earnings further, and this will definitely be passed on to lower rates for deposits, even the longer dates ones, and to more fees being passed on to customers, to make up for lack of interest income.

Lastly, while increasing money supply and inflation causes some people to spend more now, it also causes other people (those who are already too stretched financially) to spend less. How does targeting account for them?

Andy Harless said...

"...investors would be taking on risks they don't understand, or risks that they transfer to other people who don't know any better"

I would say that is a valid concern, but I note that the same concern would be present in a world with perfect price adjustment, where monetary policy had no real effect. At low discount rates, assets values become sensitive to small differences in expected returns, and it becomes hard (even for experts but especially for non-experts) to determine which assets are worth buying. In a world with perfect goods price adjustment, high saving rates (or low rates of time preference) would still make asset prices unstable, so I see the high saving rates (particularly in places like China) as the underlying cause. I would argue that NGDP targeting is a "neutral" monetary policy that attempts to simulate the perfect price adjustment world, whereas a policy that attempts to avoid asset instability is an "active" one that takes advantage of price rigidity to keep interest rates artificially high. Of course it's still ultimately a matter of policy preference. Personally, I would put a higher priority on full employment than asset stability. Also, I would argue that foregoing NGDP targeting to avoid asset instability constitutes a sort of "giving in to extortion" in that it shields countries that pursue high saving rate policies from the international impact of those policies.

"commercial banks...substituting their treasuries with reserves just lowers their interest earnings further, and this will definitely be passed on to lower rates for deposits, even the longer dates ones, and to more fees being passed on to customers, to make up for lack of interest income."

I'm not sure what should be considered income. If banks hold treasuries instead of reserves, and the economy gradually recovers, they will have capital losses on the treasuries as interest rates move toward normal. To the extent that the expectations hypothesis applies, the capital losses should offset the difference in interest income (except that banks might be earning a duration risk premium, but this is presumably just enough to compensate them for taking the extra risk). In any case, QE (in inverse proportion to how effective it is) will increase banks' total assets and give them a larger base of reserves on which to earn (safe) interest, so it's not clear that it would make them less proftiable. (Note that, to the extent that the treasury substitution occurs outside the banking system, banks will simply have more fully reserved deposits and will be earning the spread between what they pay on deposits and what the Fed pays on reserves.)

"increasing money supply and inflation...causes other people (those who are already too stretched financially) to spend less"

To the extent that inflation shows up in the nominal incomes of financially stretched people, but not in their nominal debts, it will cause them to spend more. Granted, there will be some people -- those whose nominal incomes don't respond to inflation and who don't have large nominal debts, such as perhaps retired people scraping by on fixed incomes -- who will spend less. I would argue, however, that (1) these people are a minority in terms of their economic impact and (2) they have already received a windfall in that inflaton rates have been lower than expected, so it is not unfair to reverse this windfall.

Rogue Economist said...

"If banks hold treasuries instead of reserves, and the economy gradually recovers, they will have capital losses on the treasuries as interest rates move toward normal."

Isn't the pre-QE2 bond rate what we would already consider 'normal' rate? My understanding of QE's policy objective (unless I'm wrong) is to decrease long-term bond yield. Doesn't the Fed actually want yields to be abnormally low for the meantime, so that banks begin to invest instead in other long-term assets (such as loans)? So if QE makes banks start expecting long term rates to go higher than where they are now, due to QE-induced higher inflation, if ever they do extend long term loans right now, they will price in the higher long-term rate expectation, making it more expensive for anyone to borrow right now. In short, QE2 is making banks suffer the capital loss on their Treasury investments right now (where they wouldn't have without QE), and they will try to recover the loss from the next borrower, via even higher rate.

"In any case, QE (in inverse proportion to how effective it is) will increase banks' total assets and give them a larger base of reserves on which to earn (safe) interest, so it's not clear that it would make them less proftiable"

But how did their total assets increase when the added reserves just cancels out the decrease in their government bonds? And their pre-QE 10 year and 30 year bonds likely earned higher yield than what the Fed will pay in IOR.

Andy Harless said...

"QE2 is making banks suffer the capital loss on their Treasury investments right now (where they wouldn't have without QE), and they will try to recover the loss from the next borrower, via even higher rate."

The immediate impact of QE on yields is ambiguous, since the Fed is increasing inflation (and economic growth) expectations but at the same time adding its own demand for bonds to the market's demand. So it's not clear that banks end up with larger capital losses than they otherwise would. There's a carrot and a stick, and it's not clear which one predominates. In any case it's not clear how capital losses would affect the subsequent supply of loans. Banks face a certain demand curve for loans; they may find that they recover the losses more effectively by doing a greater volume of loans (which presumably have positive profit margins anyhow) than by charging higher rates.

"But how did their total assets increase when the added reserves just cancels out the decrease in their government bonds?"

Their total assets increase because many of the people will sell bonds to the Fed are outside the commercial banking system, and these people will deposit the proceeds in banks. Banks aren't obliged to sell any of their own bonds, but the reserves that the Fed creates to purchase bonds must end up in the banking system, and banks will be earning interest on those new reserves.

Rogue Economist said...

Andy, ok, understood. Your points were logical, and there's no point insisting anything otherwise. I guess how much you support monetary easing will really boil down to how concerned you are about what kind of risks some investors will take to retain the yield they lose when a QE-persistent Fed outflanks the private sector in the government bond market. You might say these investors deserve what's coming to them, if they take on risks they don't understand.

Still, if a Fed policy were to encourage the rise of various asset classes to effect an economic rebound, it should at least acknowledge and monitor the rise of unsustainable bubbles. A pricking could put the economy on worse footing then from where it started.