Is Bitcoin a Safe Asset?

You're probably thinking no, of course not. The dollar price of bitcoin can be quite volatile (see here). One can easily gain or lose 50% over a very short period of time. So if we're talking about an asset that offers a stable rate of return, Bitcoin ain't it.

Except that this is not what I mean by a safe asset. 

I'm not even sure how to precisely define what I mean by safe asset. Loosely speaking, I'm thinking about an asset that people flock to in bad or uncertain economic times. In normal times, it's an asset that is held despite having a relatively low rate of return, perhaps because of its use as a hedge, or because of its liquidity properties.

U.S. dollars (USD) and U.S. treasuries (UST) are examples of safe assets today. Now, you might think that they're safe because they're close to risk-free in terms of what they promise in the way of a nominal rate of return. A paper USD promises a zero nominal interest rate and you'll be sure to get that if you hold on to the note over time (USD in the form of central bank reserves presently earn 1/2%, but only depository institutions get this rate.). A UST bill also promises zero nominal interest and you can be sure to get that with full principal repayment upon maturity. The coupon payments associated with a UST bond are virtually risk-free.

But that's not a complete way to think about the risk associated with a security. First, economists (rightly) focus on the real rate of return on an asset. Investors don't care how many paper dollars are promised to them in the future. They (presumably) care about the purchasing power of those future dollars. If inflation turns out to be high, that future purchasing power will be low. The opposite holds true if inflation turns out to be low.

As for the "risk free" UST bill, its market price will generally fluctuate between the issue date and maturity date. This is sometimes called "interest rate risk." If you buy a bill that promises $100 a year from now for $99, you will make about 1% if you hold the bill to maturity. But if market interest rates spike up in the interim, and if you are forced to sell your bill to raise cash, you're likely to realize a substantial loss.

That's the thing about a safe asset. It's return can appear to be stable for long periods of time and then--bam--something happens. (Something always happens.) Interest rates may spike up--a sudden sell-off in bonds may occur. What might trigger such an event? All sorts of news. Foreign banks may need to liquidate their foreign reserves consisting of USTs for political or economic reasons. A sudden increase in inflation expectations would lower the expected real rate of return on nominal bonds, inducing a sell-off. A bond sell-off might even be triggered by a good news event. An increase in productivity growth increases the expected return on private capital investment, inducing portfolio substitution out of bonds, for example.

Another thing to keep in mind is that the asset classes that constitute safe assets can change over time. In my recent piece on secular stagnation, I noted that a "flight to safety" seems to occur near regime changes that imply productivity slowdowns. In 1974, investors flocked to gold and real estate--they ran away from USD (rapidly rising price-level) and UST (rapidly rising nominal interest rates). In 2008, the situation was quite a bit different--both USD and UST were highly sought after safe havens (with investors fleeing real estate).

The observations above suggest that the monetary policy regime matters a great deal for whether a fiat currency is perceived to be safe or not. When Nixon and his advisers chose to abandon the gold standard (against the recommendation of Fed chair Burns) in 1971, monetary policy appeared to lose its nominal anchor. So when the oil price shocks and productivity slowdown hit in the early 70s, investors ran away from cash. Gold is often credited as being a safe asset because of its supply "policy." But there must be more to it than this because, like gold, the supply of real estate is not very elastic. And yet real estate was not a safe asset in 2008.

Patience, Grasshopper. I will get to Bitcoin soon. Before I do, I want to ask "what makes an asset safe?" According to Gary Gorton, it has a lot (perhaps everything) to do with information asymmetry:

A "safe asset" is an asset that can be used to transact without fear of adverse selection; that is, there are no concerns that the counterparty privately knows more about the value of the asset. (Safe Assets, Working Paper, March 2016).

In other words, a safe asset is an object with attributes that traders can mutually agree on very quickly and at little cost. Objects with this property tend to become monetary instruments or, to use a more broad term -- exchange media (which includes objects commonly used as collateral to support lending arrangements). Safe assets tend to be "simple" assets. Historically, commodities such as salt, precious metals, or coined tokens. It's easy to verify your salary in salt (just taste it). It's a bit more difficult to assay gold. The whole purpose of coinage was to make objects easily recognizable without much effort. 

  

It goes without saying that most financial instruments are complicated objects. Consider your life insurance policy, which is relatively simple as far as financial products go. The reason you can't buy your morning latte with a slice of that asset is because it's simply too costly for the vendor to do the necessary due diligence. So you pay in cash. Everyone knows what cash is. Cash may be "junk" (i.e., unbacked), but at least everyone can agree that it's junk. There's nothing complicated about cash. (The same principle holds true for UST, which are used extensively as collateral in overnight lending arrangements.)  

  

Cash and gold are "simple" objects. The fact that they pay no interest makes them even simpler. In particular, there's no need to spend time investigating the reliability of a dividend paid by "barren" asset--everyone can agree right away that the dividend is zero. This type of informational symmetry appears to be in high demand in times of financial uncertainty (when nobody knows for sure what other people know about the securities they're selling.) Of course, the situation is somewhat more complicated when counterparties (intermediaries) are involved, but this is true of any asset.

This brings me to Bitcoin. I think that Bitcoin could be the world's next great safe asset. At least, it certainly seems to have all the properties that are desired in a safe asset.

Importantly, it is a "simple" asset. It's simple in the sense that it's a pure fiat object--the monetary objects (called bitcoin) constitute no legal claim against anything of intrinsic value. Bitcoin is simply a record-keeping technology (and economists have known for a long time that money is memory). It pays no interest. Possession corresponds to ownership (unless counterparties are involved). The ledger has proven itself secure (not a guarantee that is can never be compromised, of course). 

Now one might object that Bitcoin is not that simple, not to the average person on the street, at least. Bitcoin consists of 30MB of C++ code. And the algorithm that governs the accuracy and security of the ledger can be hard to understand. But I liken this to the way most people understand how their car engine works. We have a vague notion of how internal combustion works, how power is transmitted through the drive train, blah, blah, but all we really know for sure is that our collective experience with the technology has proven useful. We also know that there are mechanics out there that do know how a car engine works. Because the Bitcoin code is open source software, attempts to modify the code for personal gain at communal expense are easily detectable through expert eyes. And we trust that there are many expert eyes on the watch.

Finally, Bitcoin has a very simple monetary policy. Essentially, the policy is to keep the money supply fixed (actually, it will grow asymptotically to a fixed number, 21 million units). Although this money supply rule could potentially be modified by communal consent, there are reasons to believe that this is unlikely to happen. And even if it does happen, it can only happen if it somehow serves the community of Bitcoin users in some broad sense.

As is well known, there's been a bit of a civil disturbance in the Bitcoin community as of late. The issue, as I understand it, concerns a proposed amendment to the Bitcoin constitution (see blocksize controversy). People fear that if the amendment does not pass (and it does not look like it will), then Satoshi Nakamoto's original vision of a low-cost, high-speed, high-volume P2P payment system may fail to materialize. Others are confident that a solution, in some form, will eventually be found. (These people breathe optimism, remember. It's the fuel that powers entrepreneurship.)

But suppose that the original vision doesn't pan out. Suppose instead that Bitcoin hits a hard limit on the volume of transactions it can process (presently far below what Visa can accomplish). Suppose further that as the subsidy on block rewards (the seigniorage revenue used to finance book-keeping costs) becomes negligible. Then a fixed transaction fee (and possibly a substantial one at that) will have to be paid, since someone has to finance the book-keeping costs. If this were to happen, then it would only make sense to hold Bitcoin for large-value transactions (the fixed cost associated with each transaction would make small-value transactions uneconomical.) 

This "Bitcoin as a large-value transfer system" does not destroy my thesis: Bitcoin can remain a desirable safe asset. (Smaller players could presumably get involved by investing in Bitcoin ETFs, although doing so would introduce counterparty risk.)

I've argued before that Bitcoin makes for lousy money. I still believe this. If it isn't the unit of account, users are subject to extreme exchange rate volatility. In a world where it is the unit of account, a "flight to safety" event would cause an unexpected and severe deflation. We have the experience of the early 1930s to show us what a Bitcoin monetary policy can lead to. (And while a Bitcoin monetary system may free people from the inflation tax, it won't free them from more general forms of taxation.)

However, even if Bitcoin is not, in my opinion, a particularly ideal monetary instrument, this does not preclude it from serving as a safe asset or longer-term store of value. Once market penetration is complete, its return behavior is likely to mimic the return behavior of any other safe asset. Safe assets generally earn a low expected return (that is, they are priced dearly). Investors can expect to earn unusually high returns in a crisis event. But if you buy at the top, you can expect to realize unusually high losses when the crisis subsides. In short, it's a great investment -- assuming you can predict when a crisis will occur and when it will end!

There are a host of issues related to safe assets that I think deserve some attention. Let me offer a few that come to mind here. First, it's not even clear that safe assets are socially desirable. Bryant (2005) demonstrates that the existence of a safe asset can induce coordination failure. Is this an argument to be taken seriously? Second, I think that policymakers should be aware that the class of safe assets may change over time. Should policy be conditioned in any way on the existing set of safe assets? Third, how should we think about "close-to-safe-asset" substitutes that seem to proliferate in periods of prolonged economic tranquility? Barren assets like cash, gold and Bitcoin generate no income. It is evidently very tempting to construct "safe senior tranches" of private interest-bearing debt to compete with these low-return barren assets--a practice that sometimes gets out of hand--and with disastrous consequences. Should a central bank issue its own interest-bearing digital cash to discourage the practice?

Pasqua 2016

Does the Fed have a credibility problem?

Former Minneapolis Fed president Narayana Kocherlakota thinks that the Fed's recent communication strategy has resulted in an unforced "credibility dilemma." (The Fed's Credibility Dilemma.) What is the nature of this dilemma?

The message coming out of the Fed recently is one of "gradual normalization." Normalization refers to returning to the state of affairs that prevailed prior to 2008, when the Fed's balance sheet was much smaller and the policy interest rate (short term nominal interest rate) was much higher. "Gradual normalization" means that the Fed's policy rate is likely to move slowly and in an upward direction over the indefinite future.

But what "gradual normalization" means and how it is interpreted by market participants can be two different things. Kocherlakota is worried that the market is interpreting the phrase as meaning the Fed will raise its policy rate slowly and in an upward direction. While this may very well be the case, it should be clear that this is not what Fed officials mean to say. Consider, for example, this report on a recent speech delivered by Boston Fed president Eric Rosengreen.

He [Rosengreen] added that the Fed’s policy committee stated it “expects that economic conditions will evolve in a manner that will warrant only gradual increases in the federal funds rate.”

This is obviously not a commitment to raise interest rates gradually and in an upward direction. It is merely a forecast of the economic conditions that are likely to prevail in the near future, together with a statement of how the Fed is likely to react to these conditions should they be realized. Again, and from the same speech,

The future path of rates, he said, will depend on incoming economic data, and how that data affects policymakers’ outlook for the economy.

This is a statement about state-contingent policy, together with a gratuitous forecast of the conditioning factors. These are two conceptually distinct objects that, when combined, give financial market participants the information they seem to crave: a statement about the likely path of interest rates over the foreseeable future.

Kocherlakota's concern is that this type of communication is misinterpreted by market participants as a commitment to a deterministic policy path. I think he's correct that some (perhaps many) people do misinterpret in this manner. If so, then the Fed faces a "credibility dilemma" because, well, what happens if the conditioning factors turn out not exactly as forecast? If future inflation comes in high above target, then the Fed will be forced to raise rates sharply and prematurely--reneging on its "commitment" to raise slowly--thereby "losing credibility." Or, if the economy turns south, the Fed may be forced to lower its policy rate--again, reneging on its "commitment" and "losing credibility." The apparent dilemma is that commitment along some dimension must necessarily be sacrificed.

Kocherlakota concludes with this:

So what, if any, plans should the Fed communicate? For one, officials must recognize that their expectations for the economy, like all forecasts, are likely to prove wrong. As a result, they should be much clearer about their willingness to make large and rapid changes in monetary policy. Instead of talking about gradual normalization, they should stress that they are ready to do “whatever it takes” to keep employment up and inflation near target.

I mostly agree with this suggestion. The part I disagree with (slightly) is where he says "they should be much clearer about their willingness to make large and rapid changes in monetary policy." I think--although I could be wrong--that the public is already generally aware that the Fed will make large and rapid changes in monetary policy should economic conditions dictate. Look at how the Fed reacted in 2008. Ask yourself: if inflation was to spike up to 5% tomorrow, how do you think the Fed would react? We all know how the Fed would react!

That is, I do not think that the issue Kocherlokota raises has anything to do with credibility. In 1971, the executive branch temporarily hampered the Fed's low-inflation credibility by abandoning the gold standard. [Take note, you End-the-Fed types--Fed chair Arthur Burns argued strongly against abandoning the gold standard.] But that reputation was won back (at a terrible price) by Paul Volcker in the early 1980s. Since that time, inflation has been relatively low and stable. Maybe it's not perfect, but it's an inflation record that many countries would envy. Consider this, for example.

To the extent that we can count on anything, I think we can count on the Fed defending its record of low and stable inflation. It is arguably the only real credibility that a central bank has. On the plus side, it's the most important dimension of credibility. (There is also the second part of the dual mandate but a central bank's influence on long-run labor market conditions is more tenuous than its influence on long-run inflation.)

What about "forward guidance?" Yes, it works great in theory, if one assumes that a central bank can commit to take actions at future dates that are not in the best interest of the economy when that future time arrives (economists call this time-inconsistent policy). But let's just face it -- the reality is that commitment along this dimension does not likely exist. Does anyone really believe that the "commitment" made by FOMC voting members today will bind a future FOMCs comprised of completely different people with different points of view? Convince me otherwise.

The Fed is credible where it matters most: its commitment to keep inflation low and stable, and to do whatever it can, within its legislated powers, to help the economy function in the best manner possible for the good of the country at large. Is it a perfect institution? No. Can policies be improved along various dimensions? Almost surely. We're all in this together. Let's try to figure it out.

In the meantime, Kocherlakota is probably correct in suggesting that communications which imply (or are wrongly interpreted) as commitments that cannot possibly be met in most states of the world are not likely to be very useful. They may even prove counterproductive. But would taking an unexpected policy action necessarily whittle away at Fed credibility? I don't think so--not along the dimensions that matter. There is no credibility dilemma here. And it does little good to cast the issue in this light.  

***

Postscript. Friday, March 25, 2016. Tim Duy weighs in here claiming that there are two types of credibility (hard and soft) and that the Fed cares about both of them (when it should perhaps only care about one). 

Mabel Frances Timlin (1891-1976)

I had the honor and pleasure of delivering the 30th Timlin lecture in economics at the University of Saskatchewan last Wednesday. My talk was on Secular Stagnation, a topic I think Professor Timlin would have approved of. I enjoy giving public lectures. It's fun to connect economic theory to real world policy issues in a public forum. I think I often learn more from the interaction with the audience than they do. It was also an opportunity to learn more of Mabel Timlin and her remarkable story.

Mabel Frances Timlin was born in Wisconsin in 1891. It's not clear what motivated to move to Saskatoon in her youth. In 1921 she found employment as a secretary at the University of Saskatchewan. Evidently unimpressed with the papers she was typing for the economics faculty, she decided to study the subject in her spare time. She eventually completed her PhD at the University of Washington and become an assistant professor at the University of Saskatchewan in 1941, at the tender age of 50. 

Her PhD thesis Keynesian Economics was published in 1942. I had our library order a copy so that I might read it before my lecture. I have to say that I was thoroughly impressed with it. Her book did not consist of a simple regurgitation of Keynes (1936). Instead, it was a courageous attempt to distill some of his most important ideas and extend them using dynamic general equilibrium theory. According to David Laidler (in a personal correspondence):

I think her 1942 book was the very first "Keynesian" text and, among other things, included the first drawing of a demand for money-interest rate curve showing a liquidity trap. Modigliani cited her in 1943, but inadequately. [Correction: Modigliani "Liquidity preference and the theory of interest and money" Econometrica Jan 1944.]

She evidently transformed the Canadian macroeconomics profession in terms of its application of formal economic modeling. She became Canada's first female full professor of economics, the first woman to serve as president of the Canadian Political Science Association, the first woman outside the natural sciences elected a Fellow of the Royal Society of Canada, and one of the first ten women to serve on the executive committee of the American Economic Association. 

I include a piece below, sent to me by Robert Dimand, that provides a few more details of her career. I think it's quite an inspiring story.  

Timlin, Mabel Frances (1891-1976)

     The Keynesian economist Mabel Timlin was the first tenured woman among Canadian economists, first woman elected president of the Canadian Political Science Association (which then covered all social sciences, including economics), the first woman outside the natural sciences elected a Fellow of the Royal Society of Canada (1951), and one of the first ten women to serve on the executive committee of the American Economic Association (1958-60), despite becoming an assistant professor only in her fiftieth year, after a long career as an academic secretary. She was born in Forest Junction, Wisconsin, on 6 December 1891, and, after studying at the Milwaukee State Normal School, taught in Wisconsin and rural Saskatchewan. She became a secretary at the University of Saskatchewan in 1921, while studying for a BA there. At first Timlin intended to study economics there, but after seeing the Department of Economics and Political Science she decided (probably correctly) that she could learn more economics on her own. She took a BA with great distinction in English in 1929, and then directed the university’s correspondence courses in economics. Mabel Timlin became an instructor in economics at the University of Saskatchewan in 1935, after completing graduate course work in economics at the University of Washington during summers and a six-month leave. Her doctoral dissertation at the University of Washington, supervised by the much younger Raymond Mikesell, was accepted in 1940 and published as Keynesian Economics (1942). In 1941, Timlin became an assistant professor of economics at the University of Saskatchewan (associate professor 1946, full professor 1950) and a member of the executive committee of the Canadian Political Science Association (vice-president 1953-55, president 1959-60).

     Keynesian Economics did more than introduce Keynesian theory into Canadian academic life. Timlin offered one of the early general equilibrium interpretations of John Maynard Keynes’s General Theory, and was particularly noteworthy in treating it as a system of shifting equilibrium, presented with innovative diagrams on which she collaborated with the eminent geometer H. S. M. Coxeter. Timlin began work on Keynesian Economics in 1935, before Keynes published his General Theory: Benjamin Higgins had come to Saskatoon from the London School of Economics in 1935 for a one-year appointment, carrying a copy of the summary of Keynes’s Cambridge lectures that Robert Bryce had presented in Friedrich Hayek’s LSE seminar.

     Beyond her work on Keynes, Timlin also expounded international developments in welfare economics and general equilibrium analysis to a Canadian audience more used to historical and institutional economics than to formal theory (e.g. Timlin 1946). Timlin (1953) sharply criticized the Bank of Canada for failing to follow Keynesian countercyclical stabilization policies during the Korean War inflation. Much of her later work (e.g. Timlin 1951, 1958, 1960) concerned immigration policy, emphasizing the economic benefits of freer immigration.

     Mabel Timlin never married. Generations of former students were her extended family. She remained active as a scholar long after her official retirement in 1959, publishing a major report on the social sciences in Canada in 1968. She remained devoted to the University of Saskatchewan despite job offers from such institutions as the University of Toronto, and died in Saskatoon on 19 September 1976.

                                                                                                                 Robert W. Dimand

Selected works

1942. Keynesian Economics. Toronto: University of Toronto Press.

1946. General equilibrium analysis and public policy. Canadian Journal of Economics

   and Political Science 12, 483-495.

1947. John Maynard Keynes. Canadian Journal of Economics and Political Science 13,

   363-365.

1951. Does Canada Need More People? Toronto: Oxford University Press.

1953. Recent developments in Canadian monetary policy. American Economic Review:

   Papers and Proceedings 43, 42-53.

1955. Monetary stabilization policies and Keynesian theory. In K. R. Kurihara (Ed.),

   Post-Keynesian Economics, London: George Allen & Unwin, 59-88.

1958. Canadian immigration with special reference to the post-war period. In

   International Economic Association, International Migration, London: Macmillan.

1960. Presidential address: Canada’s immigration policy, 1896-1910. Canadian Journal

   of Economics and Political Science 26, 517-532.

1968. The social sciences in Canada: Retrospect and prospect. In M. F. Timlin and A.

   Faucher, The Social Sciences in Canada: Two Studies, Ottawa: Social Science Research

   Council of Canada, 25-136.

1977. Keynesian Economics, with biographical note by A. E. Safarian and introduction

   by L. Tarshis. Toronto: McClelland and Stewart, Carleton Library.

Bibliography

Ainley, M. G. 1999. Mabel F. Timlin, 1891-1976: A woman economist in the world of

   men. Atlantis: A Women’s Studies Journal 23, 28-38.

Spafford, S. 2000. No Ordinary Academics: Economics and Political Science at the

   University of Saskatchewan, 1910-1960. Toronto: University of Toronto Press.

Secular stagnation then and now

Secular stagnation refers to a prolonged and indefinite period of slow growth and high unemployment (or subnormal factor utilization). When was the last time this happened in the United States? Most people are likely to say the 1930s. In fact, it was the 1970s.

The 1970s were tumultuous years. There was the Vietnam war, oil supply shocks, and Watergate. The anchovies had disappeared off the coast of Peru. Clothing styles ranged from dreadful to appalling. Disco music was in. It was an awful time for those of us who lived through it.

The seventies are also known for a significant slowdown in measured productivity growth. (See Cullison 1989 for a useful review of issues related to measurement and interpretation). The most common measure of aggregate productivity is called Total Factor Productivity, or TFP for short (See Hulton 2000.)

Aside: What is TFP?  Let Y denote the value of what is produced in an economy over the course of a year. Let (K,L) denote measures of the capital and labor services used to produce Y. Let F(K,L) denote an "aggregator function" that specifies the manner in which capital and labor are combined to form output. Given an assumed F and measurements on (Y,K,L), the TFP is computed as the residual TFP = Y/F(K,L). That is, the TFP measures the value of output unaccounted for by K and L. (Alternatively, think of TFP as measuring the average product of a list of factor inputs aggregated in a particular way.)

The San Francisco Fed produces its own "utilization-adjusted TFP" series here. This is what  what their measure of TFP looks like since 1960.

The shaded episodes were constructed using my eyeball metric, but I think that most people are likely to identify similar regions. 

There is the matter of just how to interpret the productivity dynamic above. Personally, I find it hard to believe that productivity just grows in a straight line that the undulations we see above constitute measurement error. My own inclination is to interpret this pattern through a Schumpeterian lens (see here).  Productivity growth appears in the form of growth-regimes. A productivity slowdown occurs when the economy switches from a high-growth regime to a low-growth regime. The economic shock is most pronounced in the first few years following a growth slowdown. (Related to this, see Zeira 1997.) 

Economic theory suggests that the real rate of interest should (ceteris paribus) be low in a low-growth regime (and high in a high-growth regime). Let me compute a measure of the real rate of interest by taking the annual nominal yield on U.S. treasury debt and subtracting annual PCE inflation. Here is what the data looks like. 

Well, not a perfect fit (remember the ceteris paribus part) but close enough, I think, to be intriguing. In particular, note that both low-growth regimes identified above are associated with significantly negative real interest rates. The early 1980s look odd by this view but, of course, we know that this era was associated with another type of regime change. In particular, Fed policy moved from a high-inflation regime to a low-inflation regime, with this regime change occurring in 1980 under Fed chair Paul Volcker. 

Here is how the unemployment rate correlates with the real interest rate and growth regime.

Low-growth regimes beget low real interest rates and high unemployment rates. This seems consistent with Alvin Hansen's secular stagnation hypothesis (see my earlier post here). The pattern is evident in the most recent episode, as it is in the 1970s. Except nobody called it secular stagnation back then.  

The 1970s may not be viewed as an era of secular stagnation because nominal interest rates and inflation were rather elevated in that episode. Secular stagnation, with its Depression-era origin, is more naturally related with low nominal interest rates and low inflation.  But as the following diagram shows, secular stagnation can occur in high-inflation regimes as well as low-inflation regimes. 

The 1970s episode was called stagflation (an era of high inflation and high unemployment).

The two low-growth regimes above were different in an important way. When the economy transitions from a high to low-growth regime, the shock of regime change produces uncertainty. Investors will naturally move resources out of capital expenditure and into safer asset classes. Here is a critical question: What are the safe asset classes when a productivity slowdown occurs? The answer to this question seems to vary across episodes.

In the 1970s, the USD and UST securities were not among the set of safe assets. This was in large part due to the "unanchoring" of U.S. monetary policy following the breakdown of the Bretton Woods fixed exchange rate system. In August 1971, President Nixon announced that the USD would no longer be pegged to gold. More importantly than this, the public likely did not believe that monetary policy would keep inflation in check through other means (it took Volcker to convince the public of this several years later). The added fiscal pressures of the Vietnam war and the Great Society spending could not have helped this perception. As a result, the safe assets back then did not include government securities. Instead, investors flocked to assets outside the direct control of government, like gold and real estate.

In the most recent episode, real estate was most certainly not considered a safe asset. Investors began to walk away from real estate in 2006. They then ran way in 2008. Ironically, it was the USD and UST securities that proved to be among the most highly regarded assets this time around. This is no small part attributable to the fact that U.S. monetary and fiscal policies are presently perceived to be "anchored" (unsustainable paths for money and debt are viewed as temporary departures from a long-run stable anchor).

It's worth thinking about just how large the worldwide demand for U.S. money/debt must have grown since 2008. We can infer this enhanced demand from two observations. First, the supply of debt was increased substantially. Second, the price of that debt went up, not down (safe bond yields generally declined). What might have happened had the Fed/Treasury not intervened in the way they did?

To answer this latter question, we can look to the "hard money, tight fiscal" policy regimes in place when a low-growth regime hit the U.S. economy in 1929. In the early 1930s, short-term bond yields plummeted as today, and CPI inflation ran close to negative 10%.  The unexpected and dramatic deflation--produced by an elevated demand for money against a fixed money supply--almost surely exacerbated the depth of the contraction through well-known channels.

The lesson here is that responsible monetary and fiscal policy "anchors" a regime, rendering its money/debt a safe asset. But anchoring a policy regime does not require strict adherence to a fixed asset supply rule, like the gold standard, or year-over-year balanced budgets. A credible regime will permit the supply of safe assets to expand "elastically" when the demand for the product is enhanced. Doing so can help stabilize inflation around its expected value. Of course, it is important to let the elastic snap back should economic conditions dictate. The experience of the 1970s demonstrates what can happen when a policy regime becomes unanchored.

The optimal conduct of monetary and fiscal policy over the longer term when a productivity slowdown hits is much less clear. Alvin Hansen expressed skepticism that expansionary monetary and fiscal policy could do much of anything beyond the initial shock period. If anything, it might even do some harm if, for example, such policies led to a very large public debt. Instead, Hansen favored what today we would label "pro-growth policies." His conclusions stemmed from the fact that he viewed growth slowdowns as the byproduct of slowing innovation and population growth--phenomena that monetary and fiscal policies are ill-equipped to deal with.

The situation is slightly different today in that, unlike in Hansen's time, there is presently a huge worldwide demand for U.S. treasuries. This demand stems from three major sources. First, the UST is used widely in the shadow banking sector (in repo and credit derivatives markets) as collateral, a sector that has grown significantly since the 1980s. Second, many emerging market economies want to hold USTs as a safe store of value. And third, there has recently been an added regulatory demand for USTs stemming from financial reforms like Dodd-Frank and Basel III. Because of these factors, it is likely that the U.S. economy can sustain a much higher debt-to-GDP ratio than it has in past.

Much of what one might recommend in terms of optimal policy stems from what is assumed to drive productivity (and population growth). For economies operating below the technological frontier (e.g., EMEs), productivity slowdowns might be avoided, in principle at least, through some type of policy change.  However, the case is much less clear (to me) for economies operating near the technological frontier, where the Schumpetrian dynamic is more likely to govern the productivity dynamic. Some may point to the innovations produced during the second world war as example of how expansionary fiscal policy might enhance productivity growth. But surely, basic research and development can be better subsidized in a more targeted manner, without appealing to a massive and broad-based fiscal expenditure.

Alvin Hansen's Secular Stagnation Hypothesis

The phrase secular stagnation was first introduced by Alvin Hansen in a speech he prepared for the AEA meetings in 1938: Economic Progress and Declining Population Growth. If you haven't read the speech, I recommend that you do--it's an interesting and easy read. (Alternatively, Tim Taylor provides a concise and accurate summary here: Secular Stagnation: Back to Alvin Hansen.)

Let's think about what things must have looked like for Hansen in late 1938. Earlier in the decade, the United States suffered a major economic contraction. From 1929-1933, real GDP declined by over 25%. And despite a robust period of growth from 1933-1937, the economy slipped back into recession in 1938. 

1938 must have looked disappointing (to say the least) for those living at the time. After a decade of economic hardship (the unemployment rate remained elevated throughout the episode) and recovery, the economy's capacity to produce material wealth in 1938 was no greater than its capacity in 1928. Was this the end of growth?

What sort of growth did Hansen have in mind? Well, writing in 1938, one would probably have been most impressed with the rapid rate of economic expansion that occurred in the late 19th century (the new technologies introduced early in the 20th century would have impressed as well, of course). That was an era of rapid technological progress, high population growth rates, and the widespread development of new territory. Associated with these developments was a rapid growth in investment opportunities and capital spending. Everyone seemed to be working hard. And even if progress was at times interrupted by recession, these episodes were short-lived with rapid recovery. (Personally, I don't think things were quite as rosy as the narrative above suggests, but let's stick to the main story.)

So that's roughly the evidence. What about theory? Hansen notes that earlier economists were focused mainly on how economic growth (driven by technological change, population growth, etc.) affected material living standards (the level of real per capita income). Later economists began to notice that growth and economic stability might be related (the central tenant of modern real business cycle theory). But more recently, Hansen writes, "the role of economic progress in the maintenance of full employment of the productive resources has come under consideration." (It is notable that the economists he cites for initiating this line of inquiry includes Wicksell and not Keynes; see David Laidler).

The theory Hansen espouses seems to relate the level of employment (say, as measured by the employment-to-population ratio or the unemployment rate) to the rate of economic growth. The rate of economic growth is determined by technological progress and population growth. Both of these forces are secular in the sense they tend to operate over extended horizons (i.e., over decades and not from quarter-to-quarter or year-to-year). The effect of growth is to elevate the desire for capital expenditure. A larger population stimulates the construction of residential capital. New industries and technologies stimulate the demand for business fixed investment. Workers are needed to build the stuff. Ergo, a higher rate of growth over long periods of time begets a higher rate of resource utilization over the same period of time (higher average employment rate, lower average unemployment rate).

But the story above is incomplete. After all, what was just described is not inconsistent with (say) real-business-cycle (RBC) theory. The equilibrium level of employment in that class of models could vary with the parameter that describes the economy's long-term growth rate (whether employment is increasing or decreasing in the growth rate is likely to depend on, among other things, the relative strength of substitution and wealth effects). It is theoretically possible to generate secular stagnation in an RBC model, where a low-growth regime generates a low-employment regime. But in the RBC interpretation, low employment could be an efficient outcome, given a lower pace of economic growth. It seems clear that Hansen is suggesting that the low employment observed in a low-growth regime is inefficient. He writes:

For it is an indisputable fact that the prevailing economic system has never been able to reach reasonably full employment or the attainment of its currently realizable real income without making large investment expenditures. 

This is a bit of a mischievous statement in that it commingles an alleged fact with theory. (I don't want to make too much of this now, but consider my post here, in particular, the passage by Richard Rogerson related to this issue). Hansen does not get into theory very much at all, except to say:

I shall not attempt any summary statement of this analysis. Nor is this necessary; for I take it that it is accepted by all schools of current economic thought that full employment and the maximum current attainable income level cannot be reached in the modern free enterprise economy without a volume of investment expenditures adequate to fill the gap between consumption expenditures and that level of income which could be achieved were all the factors employed. 

So my best guess of what we might have here is a version of Friedman's "plucking model" (with full employment serving as a ceiling or capacity constraint) combined with some classical notion of the difficulty associated with matching the flow of national saving with the flow of national investment. This process evidently does not work well as it should unless the demand for investment is high--which, in turn, is not generally possible unless the economy (technology and/or population) is growing rapidly.

While the hypothesis seems similar to Keynes (1936) in that depressed investment is the proximate cause of persistently high unemployment, there is (I think) an important difference. Keynes emphasized the role of "animal spirits" in determining investment demand. Depressed expectations over the "prospective rate of profit on new investment" might become a self-fulfilling prophecy (see Farmer). Although I could be wrong, I don't ever recall Keynes suggesting that the cure for high unemployment was more rapid technological progress. I think of Keynes (1936) as an explanation for depressed levels, unrelated to growth phenomena.

Hansen, on the other hand, could be interpreted as holding the view that expectations are more firmly anchored on economic fundamentals ("...we are forced to regard the factors which underlie economic progress as the dominant determinants of investment and employment.") For this reason, Hansen seemed less enamored than Keynes on the use of expansionary fiscal policy to combat high unemployment. After all, if the fundamental problem is low growth, attempts to boost "aggregate demand" can at best confer only transitory benefits. Moreover, these benefits may over time be swamped by cost considerations, like a mounting public debt.

Insofar as Hansen (1939) provides policy advice, it sounds not so much like a pro-growth agenda as it does a set of anti-anti-growth recommendations. To paraphrase: "Population growth is fading. There are no new territories to settle and exploit. We can only hope for more technological advancement, so don't do anything to hamper this last great hope of ours. Except that it seems that we are: the growing power of trade unions, trade associations, and other monopolistic practices are restricting technological advance. This is a great folly."

Hansen's secular stagnation hypothesis was largely forgotten over time. I'm not entirely sure why this was the case, especially in light of the subsequent popularity of Keynesian theory. Maybe it had something to do with their respective policy recommendations. Active demand management sounds more appealing than dismantling trade unions, I suppose.

I sometimes hear people suggest that Hansen's hypothesis fell out of favor because it was proved wrong. Shortly after he wrote, the growth factors took off and the unemployment rate remained low on average. However, Hansen did not exactly offer a prognostication--his theory is better thought of (like any theory) as a conditional forecast.

While the settlement of new territory played a big role in the past, it was not likely to do so in the future (and in fact did not, as of 2016). And while medical technology played a big role in lowering mortality rates in the 19th and early 20th century, "no important further gains in this direction can possibly offset the prevailing low birth rate." To say that Hansen did not forecast the coming baby boom is correct only insofar as Hansen did not make any forecast--just a conditional statement that if the prevailing low birth rate was to persist into the future, then low population growth would contribute to depressed demand for capital formation. He was, of course, plainly concerned that the trend might continue, but that's not the same thing as asserting it would. His caution in making predictions is evident when he writes:

Of first-rate importance is the development of new industries. There is certainly no basis for the assumption that these are a thing of the past. But there is equally no basis for the assumption that we can take it for granted the rapid emergence of new industries rich in investment opportunities as the railroad, or more recently the automobile, together with all the related developments, including the construction of public roads to which it gave rise. 

So maybe the pace of technological advance will accelerate. Or maybe it won't. It's not something we can take for granted. Hansen seems concerned about the prospect for future growth. But he's not asserting, as Robert Gordon seems to, that our best days are necessarily behind us. He notes, quite properly in my view, the sporadic nature of growth:

Nor is there any basis, either in history or theory, for the assumption that the rise of new industries proceeds inevitably at a uniform pace. The growth of modern industry has not come in terms of millions of small increments of change giving rise to smooth and even development. Characteristically, it has come by gigantic leaps and bounds. 

He cites D. H. Roberston for this view, but it is also a recurring theme in Schumpeter's work (see my earlier post relating Schumpeterian growth to secular stagnation). 

So, where does this leave us? No, I don't think Hansen's theory was proved wrong by subsequent developments. That the conditioning factors turned out not to prevail should not be construed as a rejection of the theory. The theory should be evaluated on other grounds. 

The central proposition is that high growth (via technology and or population) is necessary to keep labor near "full employment" (the level of output near "potential.") In particular, high growth is necessary to stimulate the capital spending that will employ the labor input. I am curious to know what sort of empirical evidence one might bring to bear on this hypothesis. We have a lot more data at our disposal than Hansen. How should this data be organized? Should we estimate a Hamilton regime-switching model on growth rate regimes and correlate estimated growth regime against the average employment-to-population ratio? As well, there is the perennial question of how to identify theoretical objects like "full employment" or "potential GDP" in the data.

And even if we should find support for the hypothesis, there is the question of what sort of intervention, if any, might be desirable. On this issue, Hansen writes:

How far such a [stimulative] program, whether financed by taxation or borrowing, can be carried out without adversely affecting the system of free enterprise is a problem with which economists, I predict, will have to wrestle in the future far more intensely than in the past.  

 Well, he certainly got that prediction right!

The Great American Slump

Back in 2010, I asked whether the U.S. might be in for a prolonged economic slump, similar to what Canada experienced in the 1990s (The Great Canadian Slump). It looks like the answer should have been "yes."

I think this is an interesting comparison because the slump in Canada was not precipitated by a financial crisis. Moreover, there was no "zero lower bound" issue at the time. Unlike in the U.S. today, the yields on Canadian government debt were high, not low.

Let's take a look at some data. Consider first the employment-population ratio (population of adults), or EPOP for short.

In the early 1990s, employment in Canada dropped very sharply and dramatically. This decline was as severe as the decline in U.S. employment during the great recession. It took roughly a decade for the Canadian EPOP to attain its previous peak. In the Great White North, this episode is called the Great Canadian Slump.

The parallel between these two recovery dynamics, for two different countries, at two different times, is really quite remarkable. The following diagram normalizes the EPOP to 100 at its cyclical peak (1990Q1 for Canada and 2008Q1 for the U.S.) and then plots the recovery dynamic in each case.

We can repeat the same exercise for the labor-force participation rate (LFPR).

The declining U.S. labor force participation rate has been much talked about, of course. Some analysts believe that the decline is driven primarily by demographics. Canada's LFPR has also declined since 2008, but not nearly as dramatically as the U.S. To the extent that Canadian and American demographics are similar, the comparison above suggests a more prominent role for other factors (e.g., differences in national policies, cyclical conditions, etc.). In an earlier post comparing U.S. border states to Canada and the rest of the U.S., the evidence suggests that differences in national policies may be important:

Let me finish up here with one more comparison. Here I plot real per capita GDP against EPOP for Canada:

Back then people were talking about "jobless recoveries." As you can see from the diagram above, that's quite a jobless recovery. Here's what the corresponding data looks like for the United States today:

I don't hear too much talk about "jobless recovery" in relation to the U.S. today. The language has shifted to describing this recovery dynamic as "secular stagnation."

I'm still not entirely clear what people mean by term "secular stagnation." The phrase was originally coined by Hansen (1939). Hansen argued that rapid growth in technology and/or population was necessary to keep the economy at "full employment." The proposition sounds like a version of Okun's Law, except that the relationship posited seems to be between the level of employment (EPOP) as a function of the growth rate of the economy.

Hansen did not have much to say about interest rates, except to remark that they're mainly symptomatic and that economists tend to make too much of them. This view seems somewhat different from the one espoused by Larry Summers, who attaches greater significance to low interest rates--even suggesting that they're a defining characteristic of secular stagnation. Moreover, unlike Hansen, he seems to suggest at times that the direction of causality can be reversed; i.e., that a policy that increases the level EPOP can stimulate GDP growth (both in the short and long run, I presume). This latter proposition sounds a bit like Verdoorn's law. I'll have a bit more to say about all this in a future post.