Dollars and Jens
Tuesday, August 25, 2015
finance: prices and fluctuations
The stock market has been doing something, and if you care you can go read about that elsewhere, but it has led to some discussion of why prices of financial assets bob around at times, and even the relatively informed discussion seems a bit narrow-minded. My own research, such as it is, is based on value that assets gain from liquidity or other effects related to heterogeneous agents, but this post is going to eschew even that, working simply from the basic Lucas asset pricing model that everyone knows and loves:
Price = total expected discounted dividendsRobert Shiller somewhat famously (in certain circles) observed several decades ago that dividends are much less volatile than price, and concluded that markets are irrational. (This summary is only slightly unfair to him, and, to be clear, I agree with him that markets aren't perfectly rational, but that's quite a leap from the evidence he provided.) When I first saw that, my thought was, "it's not the dividends that are fluctuating; it's the expectations". It turns out that many — I perhaps overly associate this with John Cochrane, and have been convinced myself now to join their camp — believe that fluctuations in price are driven by that term in between: discounted.
From this last point of view, then, a drop in the stock market is a reflection primarily of investors demanding a higher return than they were demanding before. One of the reasons for demanding higher returns is higher perceived risk, which can get recursive for agents with short time horizons: if that risk isn't a risk that dividends will be weaker than expected in the distant future, but a risk that other people will be demanding higher returns when you're looking to cash out your position, then it's hard not to imagine that there are likely to be multiple equilibria. Another reason for demanding a higher return, though, is that it is expected that new investment options will become more attractive in the near future; this especially might mean that the Federal Reserve is expected to raise interest rates, such that short-term bond investments will be more attractive than they have been in the last seven years.
I will note that long-term bonds tend to go up (their demanded yield down) on days when stocks go down and vice versa — expected risk-free returns in general can't be driving day-to-day moves in both bond markets and stock markets, so your very short term fluctuations, insofar as they're "rational", must involve some changes in expectations, risk premia, etc. In general the arguments for market efficiency work better the longer the time-frame, and daily movements are going to be driven almost entirely by entities that respond to market fluctuations on a daily basis. If you're wondering why the stock market has been expensive, compared to its recent earnings and historical price to earnings ratios, you can't ignore the yields available in the bond market, and if you think bond prices are going to come down in the next couple of years, you should probably be expecting stock prices to at least stop their climb.
Tuesday, August 04, 2015
economic statistics in the news
Last year on another blog I briefly recorded why labor compensation series oranges seem to lag productivity series apples, and this sort of statistic has been picked up on by one of the major presidential candidates, and so is worth noting again. I've seen a couple of other statistics (one related to that one) that are construed to mean something they probably don't, and would like to briefly note them here.
The one that's related to the wages/productivity statistic is "labor share of income", which classically was strikingly regular at 2/3; the official NIPA figure has, over the past several years in particular, come down from about 62.5% to about 58.5%. This figure is the ratio between labor expenses, as measured by NIPA, and GDP. Suppose a company produces a textile item that it sells for $4; it pays $1 for the raw input (cloth, etc.), $1 per item for machinery (which wears down and occasionally has to be refurbished or replaced), $1 per item for labor expenses, and the owner receives $1 per item in profit. This shows up as $3 of GDP—the cost of machinery is not (I think this would surprise many educated people) deducted, and on an economy-wide scale is basically double-counted—and $1 of income to labor, where a better measure might record $2 on the GDP side of the ledger and $1 to labor income. If the company is small enough, though, that the owner does all of the work—hiring no employees—and is unincorporated, then this counts as $3 to GDP and nothing to labor income—"proprietor's income" is a different category that, in order to get to 58.5%, must be excluded. It can be hard to tell which portion of "proprietor's income" should be counted as income to labor and which as income to capital (which is why it's given its own category in the first place); if you employ the crude solution of excluding proprietor's income and capital depreciation from GDP, it turns out that the ratio of labor income to Net Domestic Product Excluding Proprietor's Income has continued to bounce around between 67% and 69% for a long time. From an economist's standpoint, it is interesting that self-employment has increased, and that the capital used in production has shifted toward things like software that depreciate more quickly, but it should be noted that the changes in the NIPA labor-income to GDP ratio do not really reflect a change in the amount of actual economic income from production that is going to workers.
Another ratio I've seen is the debt to GDP ratio of, in particular, Greece; it has increased from about 130% to about 170% since the first default a few years ago. This has been suggested to indicate that even the creditors would be better off if they had allowed less austerity on the part of Greece. The only way to make even logical sense of that is to suppose that maintaining a constant debt to GDP ratio would have been a reasonable counterfactual; it does not seem plausible to me that any amount of "stimulus" would have led Greek GDP, measured in any moderately consistent (credible or not) way, to be 17/13 times what it actually is. Even if you trust the official GDP figure as much as you do its American counterpart, it's not clear that even that late starting point for the GDP figure was "sustainable", and of course the problems with Greek official statistics are well-known. What I want to call more attention to, though, is that numerator, and the form of default that is most popular among sovereign debtors, especially those trying to pretend that they aren't really defaulting, which is called "terming out"; the nominal size of the debt (in whatever its unit of account) is held fixed, but the payment dates are pushed into the future, with no interest accruing for the extra time the debt is owed. This is, of course, nonsense; the change of terms associated with the first default was widely viewed as effectively being about a 30% haircut to bond holders, who mostly sold at those reduced prices to European governmental institutions. The way debt is accounted, though, this 30% actual haircut made no change to the official amount of Greek debt outstanding; the numerator in that debt-to-GDP figure did not change from an accounting standpoint (because it was specifically engineered not to change from an accounting standpoint), but it did drop from a true economic standpoint.
Puerto Rico seems to have about a 90% ratio; I'd be interested in seeing ratios computed for both Greece and Puerto Rico in terms of cashflows discounted by a risk-free interest rate, viz. what would the present value of their bonds be if we were sure they wouldn't default? If you're trying to figure out whether a debtor could possibly avoid default, that seems like a useful measuring stick, and it has the virtue of changing when you change the real economics of the bonds. (It would surely raise all of the debt-to-GDP figures, but in particular for Greece would raise the 130% by a larger factor than the 170%.)