I really enjoy reading, and listening to, Rob Arnott of Research Affiliates. He is one of those few people – Cliff Asness is another – who is both really smart, in a cutting-edge-research sense, and really connected to the real world of investing. There are only a handful of these sorts of guys, and you want to align yourself with them when you can.
Rob has written and spoken a number of times over the last few years about the investing implications of the toppling of the demographic pyramid in developed markets. He has made the rather compelling point that much of the strong growth of the last half-century in the US can be attributed to the fact that the population as a whole was moving through its peak production years. Thus, if “natural” real growth was something like 2%, then with the demographic dividend we were able to sustain a faster pace, say 3% (I am making up the numbers here for illustration). The unfortunate side of the story is that as the center of gravity of the population, age-wise, gets closer to retirement, this tailwind becomes a headwind. So, for example, he figures that Japan’s sustainable growth rate over the next few decades is probably about zero. And ours is probably considerably less than 2%.
He wrote a piece that appeared this spring in the first quarter’s Conference Proceedings of the CFA Institute, called “Whither Bonds, After the Demographic Dividend?” It is the first time I have seen him tackle the question from the standpoint of a fixed-income investor, as opposed to an equity investor. I find it a compelling read, and strongly recommend it.
Don’t miss the “Question and Answer Session” after the article itself. You would think that someone who sees a demographic time bomb would be in the ‘deflation’ camp, but as I said Rob is a very thoughtful person and he reaches reasonable conclusions that are drawn not from knee-jerk hunches but from analytical insights. So, when asked about whether he sees an inflation problem, or continued disinflation, or deflation over the next five years, he says:
“I am not at all concerned about deflation. Any determined central banker can defeat deflation. All that is needed is a printing press. Japan has proven that. Japan is mired in what could only be described as a near depression, and it still has 1.5% inflation. So, if a central bank prints enough money, it can create inflation in an economy that is near a depression.”
This, more than anything else, explains why keeping interest rates low to avert deflation is a silly policy. If deflation happens, it is a problem that can be solved. Inflation is a much more difficult problem to solve because collapsing the money supply growth rate runs counter to political realities. I don’t think this Fed is worried about inflation at all, and they’re probably not worried too much about deflation either any longer. But they believe they can force growth higher with accommodative monetary policy, when all available evidence suggests they cannot. Moreover, Arnott’s analysis suggests that we are probably already growing at something near to, or even above, the probable maximum sustainable growth rate in this demographic reality.
Maybe we can get Arnott on the Federal Reserve Board? Probably not – no one who is truly qualified for that job would actually want it.
**Note – If you would like to be on the notification list for my new book, What’s Wrong with Money?: The Biggest Bubble of All – and How to Invest with it in Mind to receive an email when the book is published, simply send an email to WWWM@enduringinvestments.com and I will put you on the list!
I just saw this interesting article in Econbrowser called “New estimates of the effects of the minimum wage.” It is both good news, and bad news.
It is good news because it clarifies a debate about the effect of the minimum wage which has been raging for a long time, but without much actual data. This article summarizes a clever approach by a couple of academic economists to examine the actual effects of increasing the minimum wage. The research produces solid numbers and confirms some theories about the effects of the minimum wage.
The bad news is that the effect of the minimum wage is just what theory says it should be, but liberal politicians have insisted isn’t true in practice. And that’s a net negative effect on overall welfare, albeit divided between winners and losers. However, even that ought to be good news, because this analysis also means that we can reverse the policy and reap immediate gains in consumer welfare.
First, the theory: microeconomics tells us that an increase in the minimum wage, if it is above the equilibrium wage for some types of labor, should decrease employment while increasing the wages of those who actually retain their jobs. (The usual argument for increasing the minimum wage is that the people who earn minimum wage aren’t making enough to live on, and supporters tend to forget that if people lose their jobs because the minimum wage is raised, then those people are making even less.) We often say things are “Econ 101,” but this really is Econ 101 in the sense that it is taught in every introductory economics class. There is no excuse not to know this:
In the chart above, the supply of labor is S and the demand for labor is D. In the absence of a floor (minimum wage), the clearing wage and quantity of jobs is at the intersection; at a minimum wage of a, however, there is a shortage of jobs equal to c-b. If the minimum wage is raised to a’, then the shortage of jobs increases to c’-b’. The question for society is whether the increase in joblessness is an acceptable cost to accept, in order to increase the minimum wage from a to a’. (Of course, the political calculation might also include the fact that people who become unemployed will be supported by the welfare state, and potentially vote to preserve and expand those public institutions that constitute it).
The problem for those who argue against the minimum wage, or for it being increased, is that they can point out this economic truism until they are blue in the face, while the other side simply says “nuh-uh” and denies it is true with the same fervor that they insist that Obamacare has actually lowered premiums and deductibles. The façade only cracks, maybe, when actual data is presented that shows the argument to be bankrupt.
This academic study does that cleverly, by examining changes in employment and wages in states where the federal minimum wage was binding (because the state minimum wage was lower, or non-existent) and states where it was not binding (because the state minimum wage was higher, so the federal minimum wage didn’t matter). Their conclusion:
“Over the late 2000s, the average effective minimum wage rose by 30 percent across the United States. We estimate that these minimum wage increases reduced the national employment-to-population ratio by 0.7 percentage point.”
That’s the sterile conclusion. Now let’s count the cost. Between July 2007 and December 2009, the national employment-to-population ratio (which is similar to, but not the same as, the labor force participation rate) declined from 62.7% to 58.3%; it has since risen to 59.2%. As the chart below (source: Bloomberg) shows, the labor force participation rate (in yellow) shows a more gradual decline but no recovery – as has been well-documented.
Now, some numbers. In November, the Civilian noninstitutional population (the denominator for the employment-to-population rate) was reported by the Bureau of Labor Statistics (BLS) to be 248,844,000. That means that if the authors are correct, the minimum wage has boosted the wages of unemployed workers at the bottom of the scale at the cost of about 1.74 million jobs (0.007 * 248,844,000).
Imagine what having another 1.74 million workers would do for GDP? Do you think it could make a difference for one of the worst recoveries on record?
It probably isn’t fair to assume that all of those 1.74 million workers is currently “unemployed” by the BLS definition. Many of them are likely not looking for work, in which case they would not be counted as unemployed. It is interesting to note, although surely spurious, that the series “Not in Labor Force, Want a Job Now” is about 1.7 million higher than would be expected given the unemployment rate (see chart, source BLS).
Alternatively, we could consider what it would mean to the Unemployment Rate if those 1.74 million workers were employed. This means they would also be in the Civilian Labor Force, so the participation rate (see above) would be 63.5% rather than 62.8%. If instead of coming from the “Not in Labor Force, Want a Job Now” group they came from the “Unemployed” group, the Unemployment Rate would be 4.7% instead of 5.8%. (Personally, I think that most of them are probably in the former category, as the Unemployment Rate has declined at approximately the rate we would expect from past recoveries, despite tepid GDP growth.) That is not inconsistent, of course, if GDP growth is lower because the labor force is simply smaller than it should be – and that is exactly the implication of this bit of research.
Again, the good news is that we can help the country and the downtrodden “structurally” unemployed with the same simple policy: reverse all increases in the Minimum Wage that have happened since 2007.
Back to school! It is the beginning of September, post-Labor Day, and students everywhere are back to school.
It is the time of year when investors, too, tend to be schooled – as bond markets tend to strengthen and equity markets to weaken (relative to the overall drift). It doesn’t happen every year, but the tendency in fixed income markets is strong enough that, as a rule, I demand much stronger reasons to sell bonds in September and October than during the rest of the year.
This year, we appear to be in for a special treat. We all get to learn new acronyms, like ISIS, and Americans are learning where Ukraine is on a map of the world. What fun.
Monetary policymakers tend to be resistant to further lessons, since after all they have had so many years of book learning that, darn it, they should know enough by now! And yet – there is so much about economics and monetary policy that we just don’t know; so much that isn’t knowable; and so much that we know with great confidence but just isn’t so.
However, I have been delighted to find that recently, the subject of money velocity has been appearing more frequently in policy circles. To a monetarist, velocity is one of a very small handful of things that matter, and its absence from discussions among the learned has been a terrible sign that monetarism was not merely in retreat, but almost extinct. And yet, the predictions of monetarism have been borne out time and time again (that is, the actual predictions, not the idea that printing money causes economic growth – a prediction that presupposes a high degree of money illusion is at work), while the predictions of Keynesian economists have only worked once the parameters are revised post-hoc to fit the crisis. Increased money supply growth got Japan out of its deflationary spiral – as predicted. None of the Keynesian solutions deployed over the last two decades have worked, but the first attempt at serious money-printing worked. (Although it remains to be seen if the BOJ will keep its pedal to the metal; it certainly hasn’t yet “doubled the money supply” as it had pledged to do).
High money growth – that is, transactional money and not inert reserves – always accompanies high inflation. For a time, money growth may be offset by declining money velocity, but we also know quite a bit about what causes money velocity to move. Last year I cited a rare paper by a central banker (Samuel Reynard at the Swiss National Bank) that really had insight on these almost-forgotten tenets of monetarism. And this year, I am delighted to note that some economists at the St. Louis Fed have published a brief note entitled “What Does Money Velocity Tell Us about Low Inflation in the U.S.?” While the authors, Yi Wen and Maria Arias, mistakenly focus on the velocity of base money, and thus reach an incorrect conclusion that individuals are “hoarding” money (when it in fact is sitting in bank reserves, untouched), it is nevertheless the right topic and the right question, and that’s most of the battle.
I have previously shown the chart of interest rates and money velocity, so let me show it again.
This is important, because it’s the single biggest risk to a significant inflation accident. While the low vacancy rate and the rapid growth in housing prices will continue to push rents higher, bringing median and/or core inflation above 3% by early next year, we can live with 3%. The risk for much worse inflation is all tied to a rebound in monetary velocity. It bears repeating. From 2008 to 2013, money growth was rapid but declining money velocity (tied to interest rate declines, mainly) restrained inflation. If money growth remains at the same level but money velocity merely stabilizes, it is consistent with inflation of 3%-4%. But if money velocity reverses even a part of its post-crisis decline, then inflation could move appreciably higher. Since Q2 of 2008, the velocity of M2 has fallen at a 3.76% annualized rate; were that to reverse, with the same money supply growth, then the 3-4% inflation becomes 6.75%-7.75% inflation, which I think we would all agree is a bad thing.
Now, the unfortunate thing is that models of velocity that incorporate interest rates and certain other factors already indicate that money velocity should be rising. The chart below shows our proprietary model of money velocity; as you can see, since mid-2013 there has been a large and growing gap between what the model implies and where money velocity has actually been recorded. This might well mean that the model is wrong. But we should also take it as indicating the risk of a rise in velocity is real, whether it is a 1% or 2% rise per year, or a 15% snap-back over a shorter period of time.
As I always admonish, that’s a big picture concern, and not something to trade tomorrow. I would be gradually accumulating positions in inflation swaps, caps, breakevens, and broad commodity indices. There is time before people start to get really concerned. But to my mind, what is interesting is that the central bankers are now at least starting to reconsider velocity.
With heavy travel over the last week and looming over the next couple of weeks, I figured that I really ought to get an article out before everyone forgets that I write a blog.
It isn’t that there is a dearth of topics. I have so much to talk about that I am brimming over; however, between the usual press of our Quarterly Inflation Outlook (which comes out after the CPI number this month) and the press of business-seeking activity, it has been difficult to put virtual pen to virtual paper.
Here is a great example. The New York Fed blog routinely gives me great material, both positive and negative. They’ve just published an article entitled “Inflation in the Great Recession and New Keynesian Models” with a followup article called “Why Didn’t Inflation Collapse in the Great Recession?” The pair of articles could just as easily be entitled, “When Your Model Doesn’t Work, Add a Parameter.”
I have said on a number of occasions that the credit crisis was a great test of the fundamental Keynesian hypothesis that inflation is caused by growth relative to potential output. And, in the event, that hypothesis was shown to be as bankrupt as Countrywide. I have always liked the way I summed up the state of the argument in 2012:
“The upshot is that we’ve just come off the biggest recession in 80 years, and inflation barely slowed. In fact, if you remove the effects of the bubble unwind in housing, it didn’t slow at all. If growth causes inflation, and if recessions are by definition deflationary, then we should have seen a decline in core prices.”
Here is the chart that accompanies that assertion:
Now, this doesn’t mean that the monetarists are right, but it assuredly means that the Keynesians are wrong. It is far too much, though, to ask for the peaceful surrender of this view. Instead, the Keynesians (or “New Keynesians” if you prefer) first recalibrated their models, like Goldman did in 2012. (Note, incidentally, that their re-calibrated model called for sharply declining core inflation starting from the moment they published that prediction, converging on 1.4% or so in 2013. In actuality, Median CPI basically went sideways from 2011 until recently. Core inflation declined, but only because of the one-off effect of the sequester, which I don’t imagine is what Goldman was forecasting).
What the NY Fed authors have done is to postulate that the real problem with New Keynesian models is that slack isn’t measured right, but rather that “the present value of expected future marginal costs is the more meaningful way of measuring slack.” It is a wonderful thing to be able to live in a world of models populated with unobservable variables that just happen to take on the right values to make the theory work. Even if, from time to time, one needs to re-calibrate when the model’s predictions don’t work out.
For the rest of us, the fact that monetarist models predicted that inflation would not plunge in the crisis, and have consistently given predictions wholly consistent with subsequent outcomes – without requiring re-parameterization – is a pretty strong argument that it’s likely to be closer to the right way to look at the world…even if it doesn’t give us as much to do.
On this site I almost never cross-reference posts that have been put up on the Enduring Investments blog, because access to that blog is only available to investors that we pre-screen while this blog is available to pretty much anyone. So, if I post something at the Enduring Investments site, it’s generally intended for a different audience than are the articles put here.
However, in this case I am making an exception because I think the article just posted on that blog, “Inflation and Insurers: How Inflation Resembles a Reinsurance Problem,” contains really important thoughts applicable to anyone in the insurance industry – and we’ve gotten feedback from a number of insurance companies that our presentation on this topic is timely and insightful. So, if you represent an insurance company or know of someone who ought to hear these thoughts, send them to the link above!
Since I wrote a blog post in early December on “The Effect of the Affordable Care Act on Medical Care Inflation,” in which I lamented that “I haven’t seen anything of note written about the probable effect of the implementation of the Affordable Care Act on Medical Care CPI,” several things have come to my attention. This is a great example of one reason that I write these articles: to scare up other viewpoints to compare and contrast with my own views.
In this case, the question is not a trivial one. Personally, I approach the issue from the perspective of an inflation wonk, but the ham-handed rollout of the ACA has recently spawned greater introspection on the question for purely political reasons. This is awkward territory, because articles like that by Administration hack Jason Furman in Monday’s Wall Street Journal do not further the search for actual truth about the topic. And this is a topic on which we should really care about a number of questions: how the ACA is affecting prices, how it is affecting health care utilization and availability, how it is affecting long-term economic growth, and so on. I will point out that none of these are questions that can be answered definitively today. My piece mentioned above speculated on possible effects, but we simply will not know for sure for a long time.
So, when Furman makes statements like “The 7.9 million private jobs added since the ACA became law are themselves enough to disprove claims that the ACA would cause the sky to fall,” we should immediately be skeptical. It should be considered laughably implausible to suggest that Obamacare had a huge and distinguishable effect before it was even implemented. Not to mention that it is very bad science to take a few near-term data points, stretching only for a couple of years in a huge and ponderous part of the economy, to extrapolate trends (this is the error that Greenspan made in the 1990s when he heralded the rise in productivity growth that was eventually all revised away when the real data was in). Furman also conflates declines in the rate of increase of spending with decelerating inflation – but changes in health care spending include price changes (inflation) as well as changes in utilization. I will talk more about that in a minute, but suffice to say that the Furman piece is pure politics. (A good analysis of similar logical fallacies made by a well-known health care economist that Furman cites is available here by Forbes.)
I want to point you to another piece (which also has flaws and biases but is much more subtle about it), but before I do let’s look at a long-term chart of medical care inflation and the spread of medical care inflation to headline inflation. One year is far too short a period to compare these two things, not least because one-time effects like pharmaceuticals losing patent protection or sequester-induced spending restraints can muddy the waters in the short run. The chart below (source: Enduring Investments) shows the rolling ten-year rise in medical care inflation and, in red, the difference between that and rolling ten-year headline inflation.
You can see from this picture that the decline in medical care inflation, and the tightening of the spread between medical care inflation and headline inflation, is nothing particularly new. Averaging through all of the year-to-year wiggles, the spread of medical care has been pretty stable since the turn of the century (which, since this is a 10-year average, means it has been pretty stable for a couple of decades). Maybe what we are seeing is actually the anticipation of HillaryCare? (Note: that is sarcasm.)
Now, the tightening relative to overall inflation is a little exaggerated in that picture, because for the last decade or so headline inflation has been somewhat above core inflation due to the persistent rise in energy prices throughout the ‘00s. So the chart below (source: Enduring Investments) shows the spread of medical care inflation over core inflation, which demonstrates even more stability and even less reason to think that something big and long-term has really changed. At least, not that we would already know about.
The other piece I mentioned, which is more worth reading (hat tip Dr. L) is “Health Care Spending – A Giant Slain or Sleeping?” in the New England Journal of Medicine. The authors here include David Cutler, whom Forbes suspected was tainting his views with politics (see link above), so we need to be somewhat cautious about the conclusions but in any event they are much more nuanced than in the Furman article and the article makes a number of good points. And, at the least, the authors distinguish between spending on health care and inflation in health care. A few snippets, and my remarks:
- “Estimates suggest that about half the annual increase in U.S. health care spending has resulted from new technology. The role of technology itself partly reflects other underlying forces, including income and insurance. Richer countries can afford to devote more money to expensive innovations.” This is an interesting observation that we ought to think carefully about when professing a desire to “bend the cost curve.” If we are reining in inflation, that’s a good thing. But is it a good thing to rein in innovation in health care? I don’t think so.
- The authors, though, clearly question the value of technological innovation. “The future of technological innovation is, of course, unknown. But most forecasts do not call for a large increase in the number of costly new treatments… some observers are concerned that a wave of costly new biologic agents (for which generic substitutes are scarce) will soon flood the market.” Heaven forbid that we get new treatments! “The use of cardiac procedures has slowed as well.” This is a good thing?
- “Health spending has clearly been associated with health improvements, but analysts differ on whether the benefits justify the cost.” Personally, it makes me uncomfortable to leave this question in the hands of the analysts. If the benefits don’t justify the cost, and the market was free, then no one will pay for those improvements. It’s only with a highly regulated market – replete with “analysts” doing their cost/benefit analysis on health care improvements – that this even comes up.
- Some of the statistical argument is a little weak. “The recent reduction in health care spending appears to have been correlated with slower employment growth in the health care field; this suggests that such changes may continue.” I’m not sure that the causality runs that way. Surely tighter limits on what health care workers can earn might cause slower employment growth? That’s at least as plausible as the direction they are arguing.
That sounds very critical, but I point these things out mainly to make them obvious. Overall, the paper does a very good job of discussing the possible causes of the recent slowdown in health care inflation (although they focus inordinately on “the first 9 months of 2013”, a period during which we know the sequester impacted health care prices), give plenty of credit to reforms instituted far before ACA implementation, correctly distinguish between utilization and prices, and highlight some of the promising trends in health care costs – and yes, there are some! The authors are clearly supportive of the ACA, which I am not, but by and large they raise the salient questions.
It matters less if we instantly agree on the solution than that we agree on the questions.
Note: The following blog post originally appeared on June 14, 2012 and is part of a continuing year-end ‘best of’ series, calling up old posts that some readers may have not seen before. I have removed some of the references to then-current market movements and otherwise cut the article down to the interesting bits. You can read the original post here.
That said, there could be some signs that core CPI is flattening out. Of the eight ‘major-groups’, only Medical Care, Education & Communication, and Other saw their rates of rise accelerate (and those groups only total 18.9% of the consumption basket) while Food & Beverages, Housing, Apparel, Transportation, and Recreation (81.1%) all accelerated. However, the deceleration in Housing was entirely due to “Fuels and Utilities,” which is energy again. The Shelter subcategory accelerated a bit, and if you put that to the “accelerating” side of the ledger we end up with a 50-50 split. So perhaps this is encouraging?
The problem is that there is, as yet, no sign of deceleration in core prices overall, while money growth continues to grow apace. I spend a lot of time in this space writing about how important money growth is, and how growth doesn’t drive inflation. I recently found a simple and elegant illustration of the point, in a 1999 article from the Federal Reserve Board of Atlanta’s Economic Review entitled “Are Money Growth and Inflation Still Related?” Their conclusion is pretty straightforward:
“…substantial changes in inflation in a country are associated with changes in the growth of money relative to real income…the evidence in the charts is inconsistent with any suggestion that inflation is unrelated to the growth of money relative to real income. On the contrary, there appears to be substantial support for a positive, proportional relationship between the price level and money relative to income.”
But the power of the argument was in the charts. Out of curiosity, I updated their chart of U.S. prices (the GDP deflator) versus M2 relative to income to include the last 14 years (see Chart, sources: for M2 Friedman & Schwartz, Rasche, and St. Louis Fed, and Measuring Worth for the GDP and price series). Note the chart is logarithmic on the y-axis, and the series are scaled in such a way that you can see how they parallel each other.
That’s a pretty impressive correlation over a long period of time starting from the year the Federal Reserve was founded. When the authors produced their version of this chart, they were addressing the question of why inflation had stayed above zero even though M2/GDP had flattened out, and they noted that after a brief transition of a couple of years the latter line had resumed growing at the same pace (because it’s a logarithmic chart, the slope tells you the percentage rate of change). Obviously, this is a question of why changes in velocity happen, since any difference in slopes implies that the assumption of unchanged velocity must not hold. We’ve talked about how leverage and velocity are related before, but an important point is that the wiggles in velocity only matter if the level of inflation is pretty low.
A related point I have made is that at low levels of inflation, it is hard to disentangle growth and money effects on inflation – an observation that Fama made about thirty years ago. But at high levels of inflation, there’s no confusion. Clearly, money is far and away the most important driver of inflation at the levels of inflation we actually care about (say, above 4%!). The article contained this chart, showing the same relationship for Brazil and Chile as in the chart updated above:
That was pretty instructive, but the authors also looked across countries to see whether 5-year changes in M2/GDP was correlated with 5-year changes in inflation (GDP deflator) for two windows. In the chart below, the cluster of points around a 45-degree line indicates that if X is the rate of increase in M2/GDP for a given 5-year period, then X is also the best guess of the rate of inflation over the same 5-year period. Moreover, the further out on the line you go, the better the fit is (they left off one point on each chart which was so far out it would have made the rest of the chart a smudge – but which in each case was right on the 45-degree line).
That’s pretty powerful evidence, apparently forgotten by the current Federal Reserve. But what does it mean for us? The chart below shows non-overlapping 5-year periods since 1951 in the U.S., ending with 2011. The arrow points to where we would be for the 5-year period ending 2012, assuming M2 continues to grow for the rest of this year at 9% and the economy is able to achieve a 2% growth rate for the year.
So the Fed, in short, has gotten very lucky to date that velocity really did respond as they expected – plunging in 2008-09. Had that not happened, then instead of prices rising about 10% over the last five years, they would have risen about 37%.
Are we willing to bet that this time is not only different, but permanently different, from all of the previous experience, across dozens of countries for decades, in all sorts of monetary regimes? Like it or not, that is the bet we currently have on. To be bullish on bonds over a medium-term horizon, to be bullish on equity valuations over a medium-term horizon, to be bearish on commodities over a medium-term horizon, you have to recognize that you are stacking your chips alongside Chairman Bernanke’s chips, and making a big side bet with long odds against you.
I do not expect core inflation to begin to fall any time soon. [Editor’s Note: While core inflation in fact began to decelerate in the months after this post, median inflation has basically been flat from 2.2% to just above 2.0% since then. The reason for the stark difference, I have noted in more-recent commentaries, involves large changes in some fairly small segments of CPI, most notably Medical Care, and so the median is a better measure of the central tendency of price changes. Or, put another way, a bet in June 2012 that core inflation was about to decline from 2.3% to 1.6% only won because Medical Care inflation unexpectedly plunged, while broader inflation did not. So, while I was wrong in suggesting that core inflation would not begin to fall any time soon, I wasn’t as wrong as it looks like if you focus only on core inflation!]
 The reference of “money relative to income” comes from manipulation of the monetary identity, MV≡PQ. If V is constant, then P≡M/Q, which is money relative to real output, and real output equals income.