Saturday, September 26, 2015

Zero intolerance

In an earlier post, I summarized the main arguments people have used for and against a September lift-off.  There are, of course, other arguments one can make and bond guru Bill Gross isn't shy about offering his view on the matter in his September 23 2015 investment outlook.

According to Gross, the Fed's low interest rate policy constitutes a form of "financial repression." His argument, as far as I can tell, goes as follows. Long-term prosperity depends on the stock of productive capital. The stock of productive capital is augmented by investment (the flow of newly produced capital goods). Investment is financed out of saving. Low interest rates discourage saving. Therefore, low interest rates are ultimately a prescription for secular stagnation.

Gross claims that "no model will lead to this conclusion." I'm not exactly sure what he means by that. I think what he means "forget about theory, let's just look at the facts." Unfortunately, facts do not always speak for themselves.

So what sort of evidence does he select to support his conclusion? He begins by noting that inflation-adjusted interest rates (on high-grade bond instruments, I presume) were on average negative over the period 1930-1979 and on average positive since then (thanks to Volcker) until recently. Here's what the data looks like since the end of the Korean war (FRED only gives me the interest rate series since then).


The blue line plots the nominal yield on a one-year treasury, the red line plots the one-year CPI inflation rate. Blue minus red gives us a measure of the realized inflation-adjusted return on a nominally risk-free security. Returns were relatively high in the 1960s, 1980s, 1990s, and relatively low in the 1970s, 2000s, and 2010s (so far).

In relation to this observation, Gross writes approvingly of Fed policy decisions in the early 1980s:
But then Paul Volcker turned the bond market upside down and ever since (until 2009), financial markets enjoyed positive real yields and a kick in the pants boost to other asset prices, as those yields gradually came down and increased the present value of bonds, stocks and real estate.
This is a bizarre statement in some respects. First, as the data above makes clear, it was nominal yields that gradually came down--real yields remained elevated for two decades after the event. Second, he evidently does have a model of how a policy-induced increase in the nominal interest rate leads to prosperity: as yields march downward from an elevated level, capital gains are realized in a broad range of asset classes. This is a bizarre argument both in its own right and because it ignores the initial capital losses realized on wealth portfolios when the policy rate is suddenly increased.

So, no, I don't think that model makes much sense. But if so, then how do we make sense of the data? The fact is that the U.S. economy generally did prosper after the 1981-82 recession. Most economists attribute this subsequent era of prosperity in part to the fact that Volcker's policies ushered in an era of low and stable inflation. Jacking up the interest rate was just a temporary measure to bring inflation down. And once inflation began to drift down, nominal yields declined because of the Fisher effect. The fact that real yields remained elevated was just the by-product of an accelerated growth in productivity (after the 1970s productivity slowdown) that likely had little to do with monetary policy.

But maybe this conventional interpretation is incorrect. Could Gross be on to something? Maybe there is a model that justifies his conclusion. I've been thinking about this lately, wondering just what such a model would look like. Here is what I came up with.

Consider a textbook macro model. Let S(r,y) denote the supply of saving, assumed to be increasing in real income y (GDP) and the real interest rate r. Let I(r,x) denote the demand for investment, assumed to decreasing in the real interest rate r and increasing in the expected productivity of capital investment, x.

In a closed economy, domestic saving must equal domestic investment, so S(r,y) = I(r,x). This equation gives us the famous IS curve: the locus of (y,r) combinations consistent with S=I. This relation exists, in one form or another, in virtually every macro model I'm aware of.

The neoclassical view is that the market, left to its own devices, will determine a "full employment" level of income, y*. With y* so determined, the equilibrium rate of interest r* is determined by market-clearing in the loanable fund market, S(r*,y*) = I(r*,x).

Business cycles are generated by fluctuations in the x. High x is associated with optimism, low x with pessimism (over the expected return to capital spending). The diagram below demonstrates what happens when the economy switches from an optimistic outlook to a pessimistic outlook. Let point A denote the initial equilibrium position. A decline in x to x' shifts the investment demand schedule downward. Lower investment demand puts downward pressure on the interest rate--the economy moves along the saving schedule from point A to B. If depressed expectations persist, then the lower level of investment leads to a lower stock of productive capital. This has the effect of depressing GDP. As income declines from y* to y', the saving schedule shifts down and the economy moves from point B to C.



Point C is characterized by lower income, lower investment, lower saving and a lower real interest rate. This is the neoclassical explanation for why periods of why real interest rates are procyclical. Low interest rates are not the product of "financial repression." They are symptomatic of a depressed economic outlook. And any attempt to artificially increase the real interest rate is going to make things worse, not better. One cannot legislate prosperity by increasing the interest rate.

Now, if I understand Gross correctly, he seems to be saying that present circumstances are not the byproduct of depressed expectations. The problem is that the Fed is keeping the real interest rate artificially low. Let's try to interpret this view in terms of the following diagram. The economy naturally wants to be at point A, where the interest rate is higher, along with saving, investment and income. But the Fed is keeping the interest rate artificially low--at zero, in the diagram below.



The effect of the zero interest rate policy is to discourage saving. While the demand for investment is high (point D), there's not enough saving to finance it (point B). As such, the level of investment falls from A to B. The lower investment eventually translates into lower GDP. As income declines, the saving schedule shifts down and the economy eventually settles at point C. This is secular stagnation brought about by the Fed's financial repression.

So is this a sensible argument? There is a problem with it that Gross touches on in his piece when explaining how low interest rates are harmful:
How so? Because zero bound interest rates destroy the savings function of capitalism, which is a necessary and in fact synchronous component of investment. Why that is true is not immediately apparent. If companies can borrow close to zero, why wouldn’t they invest the proceeds in the real economy? The evidence of recent years is that they have not.
Indeed, the logic of the argument is not apparent at all. With interest rates so low, the business sector should be screaming for funds to finance huge new capital expenditures (point D in the diagram above). But they are not. Why not? At this stage, he simply abandons the logic and refers to the evidence. As if the evidence alone somehow supports his illogical argument.

There are, in fact, some logical arguments that one can use to interpret the facts. One is given by the neoclassical interpretation in the first diagram above. Expectations are depressed because the investment climate is poor (feel free to make a list of reasons for why this is the case). The demand for investment is low. Low investment demand is keeping the real interest rate low. The Fed is just delivering what the market "wants" in present circumstances. Raising the interest rate in the present climate would be counterproductive.

There is another argument one could make. Suppose that the investment climate is not depressed. There are loads of positive NPV projects out there just waiting to be financed. Unfortunately, financial conditions are such that many firms find it difficult to find low-cost financing to fund potentially profitable investment projects. In the wake of the financial crisis, creditors still do not fully trust debtors to make good on their promises. As well, regulatory reforms like the Dodd-Frank Act may make it more difficult to supply credit to worthy ventures. In the lingo used by macroeconomists, firms may be debt-constrained. The situation here is depicted in the following diagram.


The debt-constraints that afflict the business sector's investment plans caps the total amount of investment that will be financed by creditors--the investment demand schedule effectively becomes flat at this capped amount. The financial crisis moved the economy from point A to B. As before, lower investment ultimately reduces the productive capacity of the economy, so that income declines. The decline in income shifts the saving schedule down--the economy moves from B to C. The equilibrium interest rate is low--not because of Fed policy, but because investment is constrained. Savers would love to extend more credit if investors could be trusted and if regulatory hurdles were removed. But alas, present circumstances do not permit this saving flow to be released (except, potentially, to finance government expenditures or tax cuts). The effect of a policy-induced increase in the interest rate in this case would be to lower income even further. (The saving schedule would have to shift down even further to ensure that S=I.)

If the analysis above is correct, then the recommendation to increase interest rates in the present climate is off base. Low interest rates are not the cause of our ills--they are symptomatic of deeper problems. The way to get interest rates higher is to adopt policies that would stimulate investment demand (the neoclassical view) and/or adopt measures that would remove financial market frictions (the debt-constraint view). A deficit-financed tax cut (or subsidy) on investment spending would constitute one such measure.

There are, of course, other models that one could use to justify a policy-induced increase in the interest rate in present circumstances. Some members of the FOMC, for example, view the economy as having largely recovered and are now worried about the effect of very low interest rates on the prospect of future inflation. These types of arguments, however, are quite a bit different from the Gross hypothesis. But if he wants higher interest rates, maybe he should use them! A bit of a warning though: I don't think his bond portfolio is going to like the consequences.

14 comments:

  1. David
    I agree with you.
    I think that investing in an open, dynamic system is not restrained by notions like “marginal productivity theory”.
    The conditions do not remain the same.
    The utilization of production capacity is not predetermined, meaning that the capital stock is not given.

    ReplyDelete
    Replies
    1. I do have a two-country model that touches on some of these issues:
      https://research.stlouisfed.org/publications/review/12/05/187-196Andolfatto.pdf

      Delete
  2. David, I think you have this exactly right. Have you encountered Kevin Erdmann's blog? He has a long series of posts exploring the proposition that our stagnation is due to constraints in residential mortgage credit specifically. Here is one recent post:

    http://idiosyncraticwhisk.blogspot.com/2015/09/housing-series-part-63-more-evidence.html

    In my mind, this dovetails what you've written here beautifully. You have the academic-macro perspective on what is causing low interest rates --- "savers would love to extend more credit if investors could be trusted and if regulatory hurdles were removed." Kevin has a big pile of data-driven analysis that suggests that this barrier is firmly in place in residential mortgages, leading to massive distortion in home values, underinvestment in construction, and excess returns to land owners. Everything fits.

    -Ken

    Kenneth Duda
    Menlo Park, CA

    ReplyDelete
    Replies
    1. Ken, I haven't seen Erdmann's blog but will take a look now. I'd love to see the evidence he has supporting the credit-constraint view. Thanks for the link!

      Delete
  3. My first comment is a very general one about your blog posts. I appreciate them all (even though I may not agree) because of the high clarity of thinking and communication. So thanks for that. It's rare, though it shouldn't be rare.

    Second, I do happen to agree with you in this case. I come from a neoclassical finance perspective, so maybe that's why. At any rate, I've been perusing data recently that squares up nicely with the "low demand for investment" hypothesis. For example, total lending from private markets in the U.S. is up every quarter (according to BIS data). Banks are not increasing as much as money dealers, but to a big public corporation that makes almost no difference. And while lending to the Treasury has grown abnormally (relative to recent history) quickly, it doesn't appear to be detracting from growth in lending to non-gov't entities.

    Second, cash accounts at public companies continue to be abnormally large, as do s-t and l-t investment accounts. This indicates that they have plenty of financial resources, and few good projects to fund.

    Finally, I think too much attention is paid to levels. Spreads are much more important to money dealers and investors. For example, a firm won't invest in a 20% NPV project if the WACC is 22%. Similarly, it will invest in a 6% project if the WACC is 4%. And moving the levels around just changes the top and bottom numbers, but may not change the spread.

    Now, all that said I think that there might be some legitimate concerns over what low rates do to the investing household. For example, we are seeing more credit plays in bond funds than ever before. Arguably, this is a function of abnormally low risk-free rates. Maybe, maybe not. Too few good projects could lead all this extra money to go shopping for any yield it can pick up - thus resulting in observed low rates. Is the Fed pushing or pulling?

    One bit of data that might indicate pushing by the Fed is on the balance sheet of the Fed. The Fed holds 0$ of bills, and upwards of $4 trillion in notes and bonds. Such buying could (we'd need empirical research to see if it did) hold down yields in long-term debt markets.

    ReplyDelete
    Replies
    1. Thanks Prof J, much appreciated.

      On the surface, the large build up of cash by firms may indicate that what's wanting are opportunities and not financing. However, it could be they're building up cash hoards as precautionary saving--something they do because they *expect* financing to be costly or unavailable, should a good opportunity arise. In reality, probably both the "low investment demand" and "credit constraints" forces are at work--they're not mutually exclusive.

      You make an excellent point regarding spreads vs. levels.

      On the Fed's assets, yes I think the average duration is about 10 years, last time I checked (far longer than normal). It's not clear to me what the quantitative impact of the Fed's purchases have been on long bond yields. Maybe 10-20bp? Remember that the Fed "only" holds about 20% of marketable treasury debt (again, last time I checked). There's still a huge domestic and foreign demand for US government debt, apart from the Fed.

      Delete
    2. A third option on the cash hoarding is the tax differential between U.S. and foreign countries. I think it's an empirical issue. Another issue is that credit itself may not be constrained locally, but moving money abroad might be because of reporting requirements and so forth. As you suggested, I think there are quite a few frictions at work here, not the least of which is Dodd-Frank, which my banking friends are still trying to sort out. Lots of frictions in banking and lending right now.

      Coming back on the Fed's assets, you are quite right that the Fed's absolute position in Treasuries isn't enormous. But here are two thoughts: 1. it is very unusual for the Fed to be that active in long-term markets, at least since they were suggested to stop it in 1952 (or '51?). So it is "small," but the effects are not clear.

      2. Insofar as the Fed stands as a dealer of last resort, their position size isn't nearly as relevant as their indication that they are willing to take positions to keep yields in the current bandwidth.

      Again, I don't think the levels matter near as much as spreads, but it is possible that too low of levels compresses spreads.

      Delete
  4. What "accelerated growth in productivity" in the 80s as compared to the long period of negative safe real interest rates ? Productivity growth (labor and total factor) was higher in the 50s and 60s.

    Also there is essentially no evidence that interest rates have any effect on saving or, as they should in theory, the rate of growth of consumpion.

    You seem confident that the effect of real interest rates on consumption isn't negligible. Can you point to any empirical evidence ? Notably consumption growth was normal over the long period of extremely high real interest rates after having been extremely high over the 60s -- a period with negative safe real interest rates.

    ReplyDelete
    Replies
    1. Rob,

      By the acceleration in productivity growth in the 1980s, I simply meant that it is well-known that productivity growth slowed in the 1970s. This piece by the SF Fed supports the claim I made in my piece (see their figure 2): http://www.frbsf.org/economic-research/publications/economic-letter/2015/february/economic-growth-information-technology-factor-productivity/

      In terms of the relationship between consumption (or GDP) growth and real interest rates, the only claim I made is that real returns are procyclical. I did not state how strongly procyclical. But even just eyeballing the data above I think one would conclude that high real interest rates tend to be associated with periods of rapid growth. The 1960s were a period of high real returns. Or are you disputing the data in my first diagram?

      I'm not sure where you're coming from with these comments. Are you defending Gross' thesis (i.e., attacking my critique of it?). Or are you just calling into question the other possible interpretations I offer? I'm not wedded to any particular interpretation myself.

      Delete
  5. Im not sure if anyone has thought about the lack of demand for funds for investment being caused by low rates as we approach the lower bound. If you price an asset off its yield (or cost of capital), it is non linear to the yield (cost of capital) and becomes exponential as that yield approaches zero. it may be that people, can afford to service the interest cost and have a positive NPV, but the high price of the asset and the volatility of it discourages its purchase.

    ReplyDelete
  6. Well, poor Bill lost his moorings when the great Paul McCulley left, and, ick, El Arian took more responsibility. In a negative sense, the transition and denouement was a victory for Keynes, rarely noted.

    ReplyDelete
  7. I thought your plot was the same as the one featured here at first, but it's not. I don't know if there's any connection.

    ReplyDelete
  8. Nice post. I use the phrase "reasoning from a price change" a lot in my blog. Gross's comments are a perfect example.

    ReplyDelete
  9. Reading Gross' outlook, I see he starts to get close to an understanding of what's going on, and then he swerves away into incoherence. "...zero destroys existing business models such as life insurance company balance sheets and pension funds, which in turn are expected to use the proceeds to pay benefits for an aging boomer society." Indeed, there is a shortage of assets (and safe assets in particular), which is making it very difficult for households and their agents to provide for the future. He doesn't seem to appreciate that just raising interest rates -- effectively swapping one kind of asset (interest-bearing assets such as Treasury securities or excess reserves) for another (currency and required reserves) -- wouldn't address the problem. There's no reason to think that raising interest rates would make capital more productive, so if it were to increase the overall rate of asset returns, it would have have to do so by reducing the quantity of capital to a point where the marginal product is higher. But in that case the total quantity of assets would contract, making the asset shortage worse. Maybe pension funds' business model would be fixed, but then their assets under management would have to go way down.

    ReplyDelete