The Myths about Forecasting, Business Cycles and Time Series, Which Prevent Economists to Forecast: With an Application to Shipping Industry ()
ABSTRACT
People did not forgive economists for not
forecasting the two great depressions. Especially the second one in end-2008—which followed the 1929-1933 “Great Depression” (starting on 29/10/1929:
the “Black
Tuesday”). Keynes
in 1936, wrote a book on “economics
of depression”, and
showed the way how full employment—through marginal efficiency of capital—can
be achieved. He was bypassed by: his death (1946) and his disciples’ efforts to “model economic growth” (Harrod, 1939 and Domar,
1946). Progress in capitalistic economies cannot be achieved…without Keynes’
animal spirits. The “wrong
beliefs” of my
fellow economists, which we called them “myths” for sensation, were showed: “myths” about
business cycles, time series and forecasting. Though in the long run we are all
dead, economic history…remembers. The “trade cycle” theory, which eventually became “business cycle”, was in
scientific focus from 1907 to 1941, and then disappeared. Cycles made the life
of shipowners difficult since 1741: one cycle every 10 years! Ships, however, are assets
of long life, living 3.2 times the typical shipping cycle—and we said—for the
first time—that the “duration of a shipping slump is related to the durability
of ships”...Moreover: cycles are influenced by the state of technology; this
stated also for the first time; cycles do not go up x years and exactly x years
down. Maritime economists by this made shipowners and shippers happy: but, as shown, most freight rate’s
peaks lased 1 - 2 years and troughs lasted up to 12 years
(1947-2016)…Cycles, early in history, attracted the attention1 of many writers (we counted 449 writers, including
Keynes with 12 papers). Most believe that “trade
cycle theory” started with Jevons (in 1909), but as Mandelbrot and Hudson (2004; 2008 preface [1]) wrote,
Bible described a cycle of 7 years up and down; cycles cannot be predicted and
their trends and turning points cannot be found. We have showed that these are
not true. Myths concerned also time series like: time series have no memory;
they move at the square root of time (H = 1/2)—as proved by Einstein (1905)—and they are
best represented by “bell” curve. Hurst (1951) showed that Einstein’s case is a
special one, and time series can move faster or slower than that. More
important was that if H > 0.50 ≤ 1, time
series have to produce cycles: this
is an important finding. The general formula shown here includes coefficients: alpha (fat tails; risk), beta (skewness), gamma (scale) and delta (location). Alpha indicates the height of a distribution and the longevity of
its tails—indicating what the real risk
is, when a variable falls beyond 3σ—“Dow” fell 22s
away on Black Monday 1987, and the freight rates index ~10σ in
end-2008 meltdown…. The myths about forecasting are connected with the fact
that econometricians prefer to predict nonlinear
time series using linear tools (e.g. GARCH2). We did the opposite: we
forecast the nonlinear (shipping)
time series index of dry cargoes 1741-2015 (7 years inside the sample and 5
years outside it) testing 5 nonlinear methods and eventually selecting the best one (i.e. the “Kernel
density estimation”). The
deviations obtained were from 2% to 10% (yearly) from actual—we also indicated
a falling trend. In fact there is no turning point up in shipping dry cargo
market…by 2020…