Thursday, February 7, 2019

Upside Down CAPM: Part 9 - The mystery of long term returns

Timothy Taylor has a post up about long term returns.

There is this:
In real terms, the "safe" rate doesn't look all that safe.
Indeed, if you look at the "risky" assets like housing and corporate stock, but focus on moving averages over any given ten-year period rather than annual returns, the returns on the "risky" assets actually look rather stable.

May I suggest the upside down CAPM model?  "Risky" assets earn a relatively stable *expected* return, which is whipsawed by real shocks to cash flows.  Over longer time frames, the shocks tend to wash out, and the expected return approximates the realized return.  (Mainly here I'm talking about equities.) "Riskless" assets have an expected return that is more volatile, and reflects a discount from the stable expected return on at-risk capital, which shifts with sentiment and on-the-ground reality.  They have more stable short term cash flows, but long term returns that can fluctuate.

There is basically a risk arbitrage between volatile cash flows and the expected return on stable cash flows.

He discusses the r>g issue.  Upside down CAPM says that (at-risk) r is relatively stable.  When g is higher, then r and g tend to converge, and risk-free r rises with g and converges with at-risk r.  If we're worried about r>g, upside down CAPM says to increase g.  Mostly, that can be achieved with something like NGDP level targeting that minimizes nominal income volatility, reducing the discount that must be taken to avoid it.  I predict that under NGDP level targeting, debt levels would decline, real long term interest rates would rise, and average income growth would rise.


  1. If upside-CAPM is correct you would expect the level of aggregate investment to vary more than the returns on investment. In a slower-growth world (for whatever reason) you'd expect investment to fall. Interesting that this may be what we're seeing.

    As for NGDP targeting i have to big pushbacks:
    1) you cannot divorce monetary policy from politics. We have decades of credibility behind inflation targeting. As soon as you open up the mandate to change I suspect politicians will see that as a sign that the mandate is "up for grabs." I think the marginal benefits over inflation targeting (after all, the price level and GDP are pretty highly correlated over time) are too small to risk the "untouchable" status of the Fed.

    2) How do you pick the NGDP target level? The US labor force is projected to grow at something like 0.4% annually over the next decade. Productivity has been running at about 1.5%. So is the proper NGDP target 3.9% (0.4+1.5+2% inflation)?

    1. Yes. And, more importantly, it is a shift in the type of investment. I really hope I get a chance to look more closely at this: ( via )
      I think they are seeing upside down CAPM. Economists think low interest rates trigger investment, so this seems like a mystery to them, but low interest rates come from risk aversion, so it is more like a shift from investing in Tesla to investing in Ford.

      On NGDP targeting, the target doesn't matter. I like 5%-6% because that would approximate 2% inflation or less. I think it would be transformative, in part because of the upside down CAPM. Cyclical uncertainty comes from changing NGDP growth, not changing inflation. Reducing that uncertainty will lead to equity based investment, high growth, rising wages, and high returns to savers.

    2. I think you've just demonstrated the problem with NGDP targeting. I made the case for 4% NGDP based on demographics, trend productivity, and 2% inflation. If we picked 6% and my math is right you're looking at 4% inflation. My guess is that voters then force a change in mandate, which could easily precipitate another crisis. My view is if you have a very credible central bank, with a pretty good mandate, and politicians who generally stay out of their affairs - you don't mess with that.

    3. I'd say a mandate that had them debating in August and September 2008 about whether to raise rates or hold them steady is clearly not a good mandate. If you disagree with that, then you probably aren't as motivated to propose a change.

      Also, everyone is endlessly talking about Fed policy, including the President. Are rising wages inflationary, etc., etc. Nobody would talk about the Fed with NGDP level targeting. There would be nothing to talk about. There would be nothing to figure out. Nothing to debate about the causes of inflation or employment growth. I think that would naturally lead to a much less politicized Fed.

      I'm ok with 4%. I think 2-4% inflation is historically associated with the highest real growth, so I'm willing to go up a little. Also, I think real growth would be higher with an NGDP target regime, so I think even at 6%, inflation would be closer to 2%. Any target in those ranges would be fine. It's the volatility that gets people worked up.

    4. Doesn't matter too much what growth trajectory you pick for your nGDP level.

      The US has been on a 4% nGDP growth path since about 2009. That worked well enough.

      And George Selgin has good arguments for constant per capita nGDP. Both arguments from theory and from historical, practical experience.

    5. P.S. Since the US has been on a de-factor 4% nGDP path for about a decade now, there wouldn't be any big change. Not even a change in mandate, since you can re-interpret their dual mandate as a mandate for nGDP targeting.

      Otherwise, as a simple technical change that goes most of the way: switch from targeting the inflation rate to the trajectory of the price level. Ie don't let bygones be bygones, but make amends in the future for missing inflation targets in the past.

      If my reading is right, most market monetarists are not even too much into nGDP targeting as the ultimate instrument. It's just easy to communicate. Scott Sumner suggests that targeting nominal wages might be better, because that's where the stickiness is.

      But all seem to agree that level targeting is better than rate targeting, because it lets rational expectations do more work, so active monetary policy has to do less.

      Allowing private competitive issue of redeemable banknotes would go even further in letting market participants handle monetary policy themselves. Other financial deregulation like abolishing deposit insurance would also help.

  2. Replies
    1. Has this been shown to be effective?

      Also, I'm not saying that central banks can't stimulate, just that the stimulus comes through money creation, so that new at-risk investments are induced by higher NGDP expectations, not a lower hurdle rate on investments.

  3. I don't know if TLTRO's work. I don't even know if QE works. That's why I asked you.

    The above link is a bit wordy, but if you read down far enough there does appear to be a large flow of foreign capital into high-end housing.

    I am beginning to think that just as macroeconomists in the US need to think about housing a lot more, so they also need to think about foreign inflows of capital.

    Large inflows of capital may place the Fed into an untenable position. They can accommodate and thereby witness very high but probably unstable asset values, or they can tighten up and risk a Hyman Minsky moment---see 2008.

    1. Minsky? Minsky?! Oh, Ben. What are you doing? :-)

      One way we can tell QE worked is that interest rates went higher during QE.

      To the extent that foreign $ is drawn to high end real estate, it is probably because high end real estate tends to be in cities with low and regressive property taxes. Monetary policy has little power over the relative prices of those properties.

    2. And as Scott Sumner says (paraphrased): if QE wouldn't work, that would be an even greater gift from the heavens. Just have your central bank buy all the assets in the world, and live off the dividends.

    3. MG: well… Are you sure that QE worked in either Europe or Japan?

      To work in Japan, how large would QE have to become and would it be politically tolerable?

      Do not you think a smaller dose of money-financed fiscal programs make more sense? At the bottom, with QE you are hoping banks lend out the money. But what if real estate values are falling? Then, will banks lend out the money? Property lending is about one half of commercial bank lending in the US and an even higher fraction in Great Britain.

      Why must we choose the most tortured route possible to try to stimulate the economy?

    4. What do you mean by politically tolerable? The 'problem' you are describing is that the central banks make a huge seigniorage profit that they can hand over to their treasuries. Politically, that sounds like the opposite of a problem to me. (Even though, yes, economically we might prefer more austerity, ie the state to command less of the total GDP of the country.)

      What makes you think money-financed fiscal programs are more efficient than standard monetary policy?

      Why care about banks lending out money? Yes, banks can make extra money via lending and deposits. (And in an ideal system they could also issue their own notes with the same kind of capital already allowed against deposits..) But the money that the central banks make is adequate, too, so no need to rely on any bank lending.

      When people want to hold more money, we should give them more money. And if aggregate spending has fallen, give them some more money than they want to hold.

      > Why must we choose the most tortured route possible to try to stimulate the economy?

      I know, free banking would solve these problem easily. We are talking about more central bank action like QE here to solve problems we wouldn't have without central banks. But central banks are most likely here to stay.

      But at least QE avoids involving the fiscal part of the state, and allows private individuals to decide how to spend.

    5. See

  4. Hey, anytime a guy can quote a guy with a moniker like Hyman Minsky, one is obligated to do so. You may not be from the Pale, so you maybe this is lost on you.

    Yes, in the US, interest rates did go up with QE. But not so much in Europe and Japan, and QE has been tried in heavier doses there.

    In conclusion, the nice thing about macroeconomics is no one is ever wrong. There is always another study or national circumstance to cite, or barring that, one can take defense behind a bulwark of excellent-sounding but untestable theories.

    I believe in the NGDPLT. But monetary and fiscal authorities may need better tools to get there.

    I will keep flogging my capital-inflows horse.

  5. Kevin:

    "Since the fourth quarter of 2009, overall office prices have doubled (as have general CRE prices), yet national office rents have risen only about 15% (see Figure 1). The primary price driver for U.S. CRE assets instead has been capital flows."

    Well, Pimco could be wrong, the IMF could be wrong, the Fed could be wrong. And, it has been sai (but never proven) even I have been wrong.

    But it sure looks like heavy foreign capital inflows boost asset values.

    1. Never compare price ratios from trough to peak, Ben. Come on, my man. That's not telling you anything.