Saturday, September 1, 2012

The misbegotten birth of macro

As folks may have noted, I like graphs; they can be very useful illustrations, particularly of historical trends.  Consider this graph, taken from the 2012 US Federal Budget (via).

What is striking is the long-run stability of economic growth in the US, apart from one episode which stands out fairly dramatically. Very dramatically (pdf) given that:
industrial production fell nearly 50 percent from its prior peak. The unemployment rate reached 24 percent in 1933: About one in four people in the workforce was without a job. By 1933, the price level was more than 25 percent below its 1929 level.
Evan Soltas points to a particularly revealing indicator:
the median firm was operating at a loss: only 40 percent of firms were running a profit by 1933, as compared to roughly 90 percent under normal macroeconomic conditions.
Apart from that stand-out episode, it is a picture of an economy with a strong tendency towards an economic equilibrium around a steady rate of per capita economic growth. Sure, there are also business cycles but (again with an obvious exception) these are relatively mild, compared to the persistence of the underlying growth rate. Even that wildly abnormal episode also sees an acceleration in growth that "overshoots" and then a return to the long-run growth trend.

So, does one base study of macroeconomics--of the economy in aggregate--around the reality of that steady underlying growth or around the wildly abnormal episode? Unfortunately, what became known as macroeconomics was born during that abnormal episode and that origin has marked it ever since.

Long-run growth
For it really was a highly abnormal episode. To further see how abnormal, let's extend the time-frame backwards (using data from here).

Even without inserting a trend line, we can see that there is a strong, persistent, tendency to a steady rate of per capita economic growth, with that one stand-out episode.

What is not nearly as striking as the persistence of growth are the bumps in the road--the business cycle is comparatively minor in its effects on economic growth, except for that same stand-out episode.

The average rate of US per capita economic growth from 1790-2011 has been 1.8% per annum. But US economic growth did shift to a new trend line from about 1878.  The rate of per capita economic growth from 1790-1878 was 1.4%; from 1879-2011, 2.1%.

So, US economic growth can shift from one long-term trend to another, but such a trend can be very persistent. This upward shift in long-term growth around the beginning of the last quarter of the C19th took place in the US but not in the UK, which had a quite different experience. (The UK GDP data available from the above source only goes back to 1830.)

The 1830-1878 US per capita economic growth rate of 1.4%pa was not significantly different from the UK's rate of 1.3%pa over the same period. (Though, in the very long run of history up to that time, both rates were remarkably high, a result of shifting from an economy dominated by the land/labour constraint to one dominated by the capital/labour ratio.) Since the US could add inputs rather more easily than the UK (all that accessible land), the UK's greater rate of technological innovation was apparently significantly compensating for the US's addable inputs advantage.

Around 1878, things shifted. From 1879-1914, the US economic growth rate was 1.6%pa while the UK per capita economic growth rate slowed to 1.0%pa. US technology increasingly outstripped the UK's during this period (though not as much as is sometimes suggested: the British generally remained ahead in military technology).

Given the broadly similar institutional structure, likely the greater scale of the US economy--and greater access to cheap resources--assisted this surge in economic growth once the disruptions of the War Between The States had settled down. But part of the reason for the increase in US economic growth was probably the US's adoption of the gold standard in 1873--particularly after  the end of the Reconstruction Era meant a stable constitutional order--as said adoption eased the transfer of capital (human and financial) from the UK to the US. Since the UK had been on the gold standard since 1717--apart from the 1797-1821 suspension due to the Revolutionary and Napoleonic Wars--merely being on the gold standard was not the advantage, it was having a reliable Transatlantic payment system with the main exporter of capital (the UK).

Conversely, one can argue that the expansion of the gold standard in the 1870s (Germany, France and the US all adopted the gold standard in that decade, as did various other countries) significantly disadvantaged the UK.

From 1830-1873, the UK had a per capita growth rate of 1.5%pa; from 1874-1914 this dropped to 0.9%pa, a significant downward shift. The expansion in the gold standard in the 1870s meant that the output covered by the gold standard expanded significantly more than the stock of monetised gold, with clear deflationary effect (pdf). Highly urbanised, long-unified, at-the-technological-border UK had far less supply-side flexibility than a recently unified Germany reaping the transaction cost advantages of political unification with lots of peasants flocking to its industrialising cities or a US that was massively importing labour from Europe and incorporating a now enclosed frontier, having thoroughly settled its most divisive issue (slavery).

So the UK mostly got (bad) monetary-contraction deflation while the US and Germany achieved far more (good) output-growth deflation; hence the decline in the UK's economic growth rate, which is hard to explain otherwise. (The development of Germany and the US as centres of innovation should have made it easier for the UK to grow, not harder; especially as the gold standard made international payments highly reliable.) Since the Bank of England effectively managed the 1873-1914 gold standard (the appeal of which was precisely a "pound sterling for everyone"), there is considerable irony involved in this UK-centred monetary system operating to the UK's disadvantage.

If this is correct, it is yet another example of the besetting problem of metal-standard money--its profound vulnerability to the actions of other countries. Whether it is France abandoning bimetallism out of fear of (pdf) newly-unified Germany dumping silver onto international markets as it shifted to the gold standard or countries in the goldzone being dragged into the (ugly) deflation by the Bank of France and (pdf) the US Federal Reserve in 1929-32 or China being forced off the silver standard in the 1930s due to FDR's silver-buying program to placate silver-State US Senators driving up the price of silver, again and again being on metal-standard money has created grave vulnerability to the actions of other countries.

About business cycles
Looking at the 1870s to 1914 period, the figures indicate that the US had fairly dramatic business cycles. But, as a technologically innovative society (with the asset price uncertainty [pdf] that such entails) with considerable geographical indeterminacy about which areas would grow with which industries at what rate (generating further asset price uncertainty, especially given the importance of railroad investment and stocks), an intensified business cycle is hardly surprising. This US experience also helps make sense of (pdf) pre-Keynesian business-cycle theory (as does somewhat similar [pdf] British experience in the first part of the C19th). It was easy to build the wrong assets in the wrong places, leading to downturns while the malinvested capital was "liquidated". This clearly had little or nothing to do with central banks (the US did not have one until the creation of the US Federal Reserve in 1913; the UK, which had had a central bank since 1694, seems to have also had rather milder economic cycles during this later period) but a great deal to do with capital and uncertainty.

This is essentially a capital-based macroeconomics; the business cycle is analysed in terms of the dynamics of capital. A successor to such macroeconomics is Austrian business cycle theory. (Though real business cycle theory, with its notion of technology or other "real" shocks driving the business cycle, is in part a reprise of liquidationist theory.)

While technological and geographical uncertainty made the dynamics of capital--as processes of entrepreneurial discovery in conditions of uncertainty--a plausible generator of business cycles in the UK and the US during the C19th and early C20th, Austrian business cycle focuses on disconnects between the effect of credit expansion (typically blamed on central banks over-expanding the money supply) and underlying time preferences so that the production process becomes misaligned with consumer demands, leading to malinvestment and consequent economic downturns while such "capital overhang" is liquidated. Economic downturns--busts--are to be analysed as the consequences of the previous booms. (As Austrian economist Roger Garrison puts it in his Time and Money [pdf], Austrian business cycle theory is a theory of unsustainable booms.)

Apart from the Austrian business cycle theory requiring neither central banks nor entrepreneurs learning--they apparently make the same mistakes over and over again*, a quite different claim than the genuine technological and geographical uncertainty which plausibly generated the aforementioned C19th business cycles--unless depressions are specifically excluded, the theory requires, through its claim that downturns are to be explained by previous booms, even wildly abnormal events to be shoe-horned into a "normal" pattern.

Look back at the data presented graphically above. How plausible is it really that the wildly abnormal 1929-32 downturn is to be "explained" by the fairly unremarkable 1921-29 boom? Austrian business cycle theory, if it is treated as the standard pattern of business cycles, is also strangely narrow in its conception of how central banks can screw up. After all, a central bank over-contracting the money supply will also have unfortunate effects. So unfortunate that it can generate a serious economic downturn of almost any level all on its own

Though Austrian theory presents itself as a capital-based theory, it is actually a very specific form of monetary-based business cycle theory.  Bad monetary policy creating unbalanced credit expansion is the actual driver, capital misallocation is just the intermediate outcome.

In its way, the shift from technological and geographical uncertainty to central bank/credit dysfunction is a comforting notion. After all, precious little can be done about technological uncertainty (geographical uncertainty is probably something of a diminishing factor) but central bank/credit dysfunction is much more plausibly amenable to change--such as by abolishing central banks. Such greater comfort is purchased, however, via diminished plausibility and a dramatically narrowed role for uncertainty (which is necessary to generate faith that the business cycle can be entirely or largely abolished through stopping "unnatural"--that is, deviating from the natural rate of interest, the rate of interest which balances saving and investment--monetary and credit expansion). What is particularly sad about all this is that von Mises and Hayek were correct in that Fed was to blame for the wildly abnormal 1930s downturn (along with the Bank of France) but the Austrians clung to their confidence in "untainted" interest rates to calibrate saving and investment, and (by implication) lack of systematic uncertainty issues, so gravely misdiagnosed how the Fed was to blame and what the implications of that were.

The empirical evidence that busts are not correlated with previous booms but booms are correlated with (pdf) previous busts (Milton Friedman's so-called "plucking model") further undermines the analytical utility of the monetary/credit expansion Austrian business cycle theory.  A capital-based theory that incorporated some derivation of Schumpeterian "creative destruction"--that the more intense the bust, the more intense the selection pressures, the more robust the surviving capital allocation is--would much more plausibly deal with this bust-boom pattern than attempts to connect the bust back to the previous boom.

Considering labour
While Austrian business cycle theory developed from the "liquidationist" theory that had a considerable hold on mainstream economic thought prior to the 1930s Depression, the experience of the 1930s saw the rise of a quite different approach to analysing business cycles. This was Keynesian economics, which can quite plausibly (pdf) be viewed as labour-based macroeconomics. Not because it is not interested in capital (fluctuations in investment are central to Keynesian analysis) but because sticky prices, and especially sticky wages, are crucial to its analysis. As Canadian economist Nick Rowe puts it, in Keynesian macroeconomics, quantities adjust quicker than (key) prices.

Just as the pre-war experience of the US and UK gave plausibility to "liquidationist" approaches, so the UK's interwar experience provided plausibility to a labour-centred approach. As we can see above, the 1920-21 downturn was actually more severe in the UK than the 1929-31 downturn. The UK also did not fully recover from the downturn for many years. Not only did per capita GDP not recover to its 1919 level until 1927 but unemployment remained entrenched. While the introduction of unemployment insurance was blamed, this seemed a thin reed to explain why the British economy was persistently failing to employ available labour or recover to its previous rates of growth. John Maynard Keynes's analysis connected sticky wages, to dysfunctional investment levels, to inadequate demand, where prices could not be relied upon to solve the resultant coordination problem (on the contrary, an economy could become "stuck" at a below-capacity equilibrium) leaving a role for fiscal policy to manage aggregate demand.

Because Keynes and Friedrich Hayek, the young doyen of Austrian economics, were both powerful and charismatic intellects, their clash largely defined the new field of macroeconomics, with Keynes and Keynesianism being the clear winner among mainstream Anglosphere economics. But this new field arose out of the struggle to explain a wildly abnormal event, the Great Depression. The choice being between the Austrian approach--which treated the wildly abnormal as the product of a normal boom-and-bust process--and the Keynesian approach--which treated the wildly abnormal as the baseline for economic analysis. The latter error being more forgivable, given the interwar experience of Britain.
Yet, if we return to the long-run perspective, how plausible is that analysis of entrenched dysfunction?

Yes, the interwar period in the UK is clearly a period of sub-optimal economic performance. But it is bracketed by periods of persistent economic growth. From 1947-2010, the UK had a per capita economic growth rate of 2.0%pa, slightly higher than the US over the same period of 1.9%pa. It was nowhere near enough to make up for the US's higher growth rates over the 1878-1947 period, especially as US standards of living had been higher to start with. Still, we are back to an economy with a strong tendency to equilibrium around a persistent growth rate.

Money matters
For there is a third alternative to capital and labour-based macroeconomics; monetary macroeconomics. This was represented in the interwar period by the work of Gustav CassellR.G. HawtreyIrving Fisher (especially in his The Debt-Deflation Theory of Great Depressions [pdf]) and later, most famously, by Milton Friedman. (The capital-, labour-, monetary- triad is shamelessly taken from Garrison [pdf] via here.)

Cassell and Hawtrey both predicted (pdf) that a return to the gold standard after the Great War would carry with it grave deflationary dangers. This successful, before-the-event warning failed to hold the attention of mainstream economics because, however internationally prominent he was, Cassell was a Swede while Hawtrey was a Treasury official--handicaps in public advocacy and persuasion within the Anglosphere--and Fisher's reputation never recovered from his optimism about asset prices just before the 1929 crash.

The 1963 publication of Friedman and Schwarz's A Monetary History of the United States re-launched the monetary explanation of the Great Depression. Subsequent work by a range of scholars has extended and deepened the monetary causes analysis, so it is now a widely accepted view within mainstream economics. Friedman substantially agreed with Keynes's macroeconomic analysis while profoundly disagreeing with his underlying moral vision; thus, Friedman both completed and opposed Keynes. After all, why was there a contraction at the end of the 1921-29 boom is not a particularly interesting question--one was likely "due". Why was it so abnormally intense? Why did it last so abnormally long? These are the interesting questions.

Monetary macroeconomics has major analytical advantages over capital-based-yet-money-driven Austrian analysis. It is much more broadminded about how central banks can screw up; they can both over-expand the money supply and seriously contract the money supply. Friedman's view was that serious economic downturns always involved significant monetary contraction; something that continues to be true. Note, this is monetary contraction in the sense of money-in-circulation; contractionary monetary policy is perfectly possible with expanding monetary base. As Friedman notes in Monetary Mischief:
Base, or high-powered money, remained remarkably constant at about 10 percent of national income from the middle of the nineteenth century to the Great Depression. It the rose sharply, to a peak of about 25 percent in 1946 (p.255).
Monetary macroeconomics also has no problem coping with wildly abnormal economic downturns--all that takes is wildly abnormal monetary conditions. Money is so much more variable and pervasive than capital and labour, it is a much more plausible explanatory factor for major economic fluctuations across the entire economy.

Both Keynesianism and Monetarism have since been affected by the expectations revolution, particularly the Lucas critique. This led to New Keynesian economics and, more recently, Market Monetarism which, unlike old-style Monetarism, does not take the view that quantities speak for themselves.** New Keynesianism is also motivated by a search for microeconomic foundations for macroeconomics.

A long shadow
The shadow the 1930s cast over the emerging discipline of macroeconomics is nicely captured by Brad DeLong (pdf):
The Great Depression made it impossible--for a while--for almost anyone to believe that the business cycle was a fluctuation around rather than a shortfall below some sustainable level of production and employment (p.250).
With the memory of the Great Depression still fairly fresh, it was extremely difficult to argue that the normal workings of the business cycle led to fluctuations around any sort of equilibrium position (p.255).
The Great Depression had taught everyone the lesson that business cycles were shortfalls below, and not fluctuations around, sustainable levels of production and employment (p.256).
Then came the entrenched inflation of the 1970s and explanations based on some perennial tendency to demand shortfalls lost plausibility. Old-style Keynesianism failed the test of Stagflation. Moreover, if demand could both undershoot (the 1930s) and overshoot (the 1970s), then fluctuation around some equilibrium became much more plausible. Reinforcing that, the normality of persistent economic growth (at least in societies where production of capital is such that the capital/labour ratio is dominant rather than land/labour constraint) re-asserted itself--hence the notion that the economy is Keynesian in the short-run and classical in the long-run.

Now we are back in very abnormal economic conditions--though not nearly as abnormal as the interwar period--and the (macro)economics profession has not exactly covered itself in glory. The Austrians are mostly just re-running their 1930s mistake of trying to shoe-horn the deeply abnormal into a standard recurring pattern. To cling to the analysis of central banks who never learn and entrepreneurs who never learn as explaining current conditions involves being macroeconomists who never learn; there is a certain analytical unity there, I guess. The Austrians do have something going for them; at least they are willing to point the finger at central bankers.  Alas, the obsession about monetary/credit expansion diverts attention from the actual failures of key central banks and, in feeding the obsession about mythical inflationary dangers, does its bit to make things worse. (As was also true in the 1930s; then Austrian economists were determined to keep the golden fetters, now they either want to go back to such or, if not, through their "hard money" advocacy, they do their bit to keep the inflation-targeting fetters.)

After 1928-32, no one should have any illusions about how bureaucratically insular in disastrous ways central bankers can be. In 1928-32, the Fed and the Bank of France did not have some deep dark plan to help one side of politics, to enrich bankers or whatever. They just had disastrous policy beliefs that they followed to the bitter end. We are in a reprise of that. In both periods, key central banks failed to sensibly run the system they were supposed to be running. They were committed to it (or a certain conception of it) way beyond sense; running such amazingly badly due to a disastrously narrow conception of their policy aims (and, likely, because changing policy would admit fault).

And are doing so with the support of a depressingly large proportion of economists. For the Austrians are not alone in attempting to shoe-horn the abnormal into reassuring patterns of normality. Much of the rest of the economics profession is seeking, just like their interwar colleagues did, to cling to what used to work (simple, rather than over-the-business-cycle, inflation targeting in the current case; the gold standard in the previous one). The view that expectations about prices matter, and are a matter for monetary policy, but expectations about spending do not or are not still has a depressing hold on perceptions. (See this paper [pdf] {via}, for example, by a former member of the Bank of England's Monetary Policy Committee.)

The facts have changed but their thinking has not; Keynes would not have approved. (Neither would have Friedman.) Having reverted to the (macro)economics of normality, too many economists flounder in the face of significant abnormality. Macroeconomics, having been created in wildly abnormal times, and with this deep normality-abnormality misdiagnosis embedded in it, still suffers the legacy of its misbegotten birth.

* Either such information is deemed to be not available to entrepreneurs, or they have no demand for it, or the market cannot supply such demand yet it remains, in some sense, available to the theorist. The Austrian story also gains plausibility by theorising in terms of single unifying interest rate, abstracting away from the information provided by differences between interest rates for different time periods. Deeming central bank monetary expansion to be automatically inflationary (in either a price level or a credit expansion sense) makes the information directly available to the theorist but also, of course, to anyone else. Revealingly, in his Time and Money (pdf), Garrison treats technological change as a seamless change in the pattern of production yet labours mightily to explain why credit expansion will fool entrepreneurs--and will therefore seriously distort production--while the abstract point that this happens is nevertheless available to theorists.

** Expectations adds time to money in a natural way, as money cannot be effectively analysed without some consideration of expectations, since its use is entirely based on expectations.  Such expectations need be rational only in the weak sense that differences between agents' expectations and theory's implications require justification. (Merely making the expectations of agents be the same as the implications of one's theory--i.e. having difference=0--does not eliminate the need for such justification, as any level of difference has to be justified; without such justification, setting the difference at zero just avoids the issue under the pretence of having solved it.)
[Cross-posted at Skepticlawyer or at Critical Thinking Applied.]