The Stock Market Crash Really Did Cause the Great Recession

Date01 October 2015
Published date01 October 2015
AuthorRoger E. A. Farmer
DOIhttp://doi.org/10.1111/obes.12100
617
©2015 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 77, 5 (2015) 0305–9049
doi: 10.1111/obes.12100
The Stock Market Crash Really Did Cause the Great
Recession*
Roger E. A. Farmer
Department of Economics, UCLA, 8283 Bunche Hall Box 951477, Los Angeles, CA
90095-1477, USA (e-mail: rfarmer@econ.ucla.edu)
Abstract
This paper studies the connection between the stock market and the unemployment rate.
I establish three facts. First, the log of the real value of the S&P 500 and the log of a
logistic transformation of the unemployment rate are non-stationary cointegrated series.
Second, the stock market Granger causes the unemployment rate. Third, the connection
between changes in the real value of the stock market and changes in the unemployment
rate has remained structurally stable over seventy years. My results establish that the fall
in the stock market in the autumn of 2008 provides a plausible causal explanation for the
magnitude of the Great Recession.
I. Introduction
In a recent paper in the Journal of Economic Dynamics and Control, (Farmer, 2012b),
Roger Farmer pointed out that there are transformations of the U.S. unemployment rate
and the real value of the S&P 500 that are non-stationary but cointegrated. Farmer provided
a Vector Error Correction Model (VECM) where changes in stock market wealth cause
changes in the unemployment rate. He estimated this model, using data on unemployment
and the real value of the S&P 500 from 1953q1 through 1979q3, and showedthat the model
provides an excellent fit to data from 1979q4 through 2011q1.
Rosnick (2013) has argued that a univariate model provides a better prediction of the
unemployment rate than Farmer’s published model. I show here, that although the uni-
variate model provides more accurate out-of-sample forecasts than theVECM, a bivariate
model that includes information from the stock market outperforms both alternatives. My
results establish that the stock market contains significant information that helps to predict
the future unemployment rate. A big stock market crash, in the absence of central bank
intervention, will be followed by a major recession one to four quarters later. Further, the
connection between changes in the stock market and changes in the unemployment rate
has remained structurally stable for seventy years.
JEL Classification numbers: E30, C10.
*I would like to thank an anonymous referee of this journal for suggestions that considerably improved the final
version of the paper. I wouldalso like to thank C. Roxanne Farmer for her editorial assistance.
618 Bulletin
The exchange between Farmer and Rosnick raises two questions, both of which I take
up in this paper. The first question is philosophical.What does it mean for one time series
to cause another? I establish in section V. that the stock market causes the unemployment
rate in the sense of Granger (1969, 1980) and I discuss the implications of that finding for
economic policy.
The second question is more narrowly defined. If the stock market Granger causes the
unemployment rate, how can a model that ignores stock market information provide a
more accurate forecast than one that exploits this information to inform its prediction? I
answer that question in section VI. where I draw on the results of Clements and Hendry
(1988, 1999) and Hendry (2004) who show that, in the presence of a structural break, a
misspecified VECM can provide misleading forecasts.
II. Related literature
The correct way to model a pair of non-stationary cointegrated time series is with a VECM
(Granger, 1981; Engle and Granger, 1987). Given the causal link from the stock market to
unemployment it should be possible to predict the future history of the unemployment rate
using its own past and the past history of the stock market. But in the presence of struc-
tural breaks, VECMs are not robust to shifts in the underlying equilibria. Models that are
overdifferenced, and therefore mispecified, are knownto outperform well specified models
that have undergone a structural break (Clements and Hendry, 1988, 1999; Hendry, 2006;
Castle, Fawcett and Hendry, 2010). This paper illustrates the result that overdifferencing
improvesforecasting ability in the context of the unemployment–stock market relationship,
previously studied in Farmer (2012b).
I am not the first to investigatethe connection between wealth and subsequent movements
in economic activity.Lettau and Ludvigson (2004, 2011) provide a statistical model of con-
sumption, wealth and labour earnings as non-stationary, but cointegrated time series, using
the methods surveyedin Hendr y (2004) and Hendry and Juselius (2000, 2001). I look instead
at the relationship betweenthe real value of the stock market and the unemployment rate.
The connection between the stock market and unemploymentwas recognized by Phelps
(1999) who pointed out that the stock market boom of the 1990s was accompanied by a
reduction in the unemployment rate and Fitoussi et al. (2000), who found a similar corre-
lation between the stock market and unemployment for a variety of European countries.
Following Phelps (1999) and Hoon and Phelps (1992), these authors explained this con-
nection using Phelps’ (1994) structuralist model of the natural rate of unemployment. My
explanation for persistent unemployment (Farmer, 2010a, 2012a,b, 2013a) is closer to the
models of hysteresis described by Blanchard and Summers (1986, 1987) and Ball (1999)
than the structuralist model of Phelps. Nevertheless, the theoretical foundation for per-
sistent unemployment in Farmer (2010a, 2012a,b, 2013a) is very different from the one
provided in their work.
III. Properties of the data
The data I use in this analysis include the S&P 500 from Shiller (2014) and the unem-
ployment rate from the Bureau of Labour Statistics. The stock market data are quarterly
©2015 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT