Available via license: CC BY-NC-ND
Content may be subject to copyright.
International Journal of Technology (2012) 1: 77-84
ISSN 2086-9614 © IJTech 2012
DO SALES OF MEN’S UNDERWEAR REALLY PREDICT THE STATE OF THE
ECONOMY?
Phil Smith
1,2*
1
Hoshin, Data Hoshin, Studio Hoshin, Wilmslow Road, Manchester, M146HZ
2
PT Multi-Interdana, Jl. Gonseng Raya No. 43, Jakarta 13770, Indonesia
(Received: November 2011 / Revised: November 2011 / Accepted: December 2011)
ABSTRACT
Alan Greenspan, the then Chairman of the US Federal Reserve, was reported in 1997 to have
studied men’s underwear sales, to provide an early warning for recession in the US economy.
Since then many economics editors have written about men’s underwear sales and a Men’s
Underwear Index, to support their views on the future direction of the economy; and the OECD
use a basket of indicators to forecast the world economy (OECD, 2008). This paper explores
the theory and its empirical robustness for 57 countries. There is some limited statistical
evidence that sales of men’s underwear might be an indicator of the US economy, or more
precisely of looming recession. But the relationship is far from clear; therefore as an indicator
it should be approached with extreme caution. Certainly more detailed and robust investigation
is required. Looking across a further 56 counties, men’s underwear sales appear to be unrelated
to the economy as a whole, seeming to behave as a basic commodity. However, for Armenia,
Ecuador and Kuwait, there may be some validity in understanding the relationship further.
Keywords: Apparel consumption; Economic forecasting; Men’s underwear index; Recession;
World economy
1. INTRODUCTION
Given that sales data is often released prior to official economic data, a number of economists
and policymakers have attempted to construct an index which will forecast some element of the
economy. The OECD, for example, uses a composite of leading indicators to forecast
movements in the World economy (OECD, 2008). The academic antecedents of these indices
can be traced back to 1942 and Roy Harrod’s ‘Towards a Dynamic Economics’; whilst more
recent empirical debate is contained in Yerex (2011), Bai & Ng (2008), Davidson & Hinkley
(2003), McGuckin et al. (2001), Rogers (1998), Emerson & Hendry (1994), Stock & Watson
(1992), and Diebold & Rudebusch (1991). Perhaps the most persistent of these indices is The
Economists ‘Big Mac Index’; indeed The Economist (2011) also lists the lipstick, cranes and
men’s underwear indices.
Commentators on apparel sales have long suggested that sales of men’s underwear are a leading
indicator (it predicts rather than follows other indicators) for the economy (The Jakarta Globe,
2009; Smith, 2011). The received wisdom is that as soon as a recession hits, the first item of
discretionary spending to be cut are men’s underwear (in other words they have a high level of
economic elasticity).
*
Corresponding author’s email: phil@hoshin.co.uk, Tel: 44 161 256 0349, Mobile: +62858 90637307
78 Do Sales of Men’s Underwear Really Predict the State of the Economy?
Since 1997, it has been widely reported that Alan Greenspan said that men’s underwear sales
‘… is almost always a prescient, forward impression that here comes trouble' (National Public
Radio, 1998; the majority of the reports were actually in 1998 and 1999, resulting from
publicity to Alan Greenspan’s book published in 1997). Since this revelation, economics editors
in many countries have discussed men’s underwear sales and described a Men’s Underwear
Index (MUI). This MUI is often attributed to Mintel (a global publisher of market intelligence;
see 2008) and editors normally refer to Mintel’s forecast data in describing this MUI
(Washington Post, 2009); but no serious attempt at constructing a MUI is evident.
2. METHODOLOGY
This paper only seeks to establish if there is any validity in the widely held belief that men’s
underwear sales decline prior to an economic downturn, by using a simple correlation. It does
not attempt to explain the relationship, or construct a more general theory based on the
elasticity of demand of men’s underwear sales. It is limited to this unambitious analysis because
the comparative international data does not support a more sophisticated, or more advanced,
statistical analysis. Indeed, it can at best only indicate that there could be a relationship not fully
to prove or disprove the theory, as it does not guarantee the existence of causality. As the theory
appears to have originated in the US and that there has been more commentary on the MUI in
the US, the initial investigation explores the relationship between underwear sales and the US
economy. Nevertheless, as with any investigation of this type it is only as good as the data
available. Unfortunately, relevant US Government statistics are not released quickly enough to
be of value in predicting the near future (although, Alan Greenspan would have had access to
unpublished statistics).
More encouragingly, US trade statistics are published each month and normally available just
two months after the end of the relevant month (Census.gov., 2011). In addition, Gross
Domestic Product (GDP) data is available fairly quickly, although only on a quarterly basis.
Therefore, Consumption Imports of Men's and Boys' Underwear and Nightwear can be used as
a proxy for men’s underwear sales. Whilst these are more a reflection of retailer’s predicted
sales than actual sales, their movements should provide some insight into actual sales (albeit
reacting slower than actual sales). Taking the period from January 2007 until May 2011,
consumption imports are tracked and plotted against quarterly GDP. Because consumption and
GDP use different measurements, the data has been statistically transformed (using a Z Score),
so that they can be more readily compared. Z scores are an artificial construct so their scales
have been omitted from the linear axis of the line graphs; furthermore it is important to
recognize that a negative figure is simply a point on the artificial scale.
The United Nations collate relevant statistics from a number of countries (not including the US)
that facilitate this sort of global comparison (Unstats.un.org., 2011). Unfortunately they are not
collated and published quickly enough to be used for forecasting future movements of global
economies. In addition, they are only published on an annual basis (data is considered from
1995 to 2008). But they can be investigated to suggest if men’s underwear is a useful predictor
in countries other than the US. Z Scores is even more useful comparing these disparate
datasets. Furthermore the data relates solely to men’s underwear sales.
In addition, to plotting the data on simple line graphs, a simple linear correlation has been
constructed for the period tracked and just for recessions. Given the seasonality of the US data
it may have been wise to also use a statistical method to iron out these effects (such as moving
average, or even a forecasting technique such as the Box Jenkins Method, see Box & Jenkins,
1970), although there isn’t really enough data for this to be effective in tracking the sorts of
movements and relationships we are interested in.
Smith 79
3. RESULTS AND DISCUSSION
Looking initially at data from the US, the picture is complicated due to seasonal fluctuations
and the problems of comparing monthly with quarterly data. At first sight monthly changes in
consumption imports do not appear to be a leading indicator of the economy.
But using trend lines to iron out the fluctuations reveals a potential relationship between
consumption imports and the economy. Although this does not provide any evidence on the
nature of this relationship.
(a) (b)
(c) (d)
Figure 1 US men’s and boy’s underwear and nightwear
In November 2008 the trend lines cross, suggesting that the relationship is far from
straightforward. A simple linear correlation of the two datasets does establish a positive
relationship i.e. sales fall when the economy slows and increase when it revives. But at 0.511
(on a scale of 1 to -1), this is hardly a convincing result.
In July 2008 the US went into recession, but consumption imports for this period do not decline
until November 2008 (the month the trend lines cross). This may be an artefact of using
consumption imports as a proxy, but it would seem to imply that sales follow the economy
rather than being a leading indicator, as Alan Greenspan’s theory suggests. Nevertheless, as the
correlation for this period rises to 0.839, it would seem to support at least a part of his theory
and the overall theory being tested in this paper.
Looking at the UN data, immediately some interesting patterns begin to emerge, with
underwear sales appearing to mirror movements in some economies (for example the match is
near perfect for the Czech Republic (Figure 2a) and Finland (Figure 2b)). What is striking is
that for annual data a time lag is not apparent. This could mean that men’s underwear sales are
not a leading indicator (i.e. they simply follow the overall economy), or that the lag between
indicators is between one month and one year.
Plotting the data for the more dynamic economies of Asia (Hong Kong (Figure 2c), Indonesia
(Figure 2d), and Malaysia (Figure 2e)) the pattern still seems to exist. But the relationship is a
Jan
-
07
Mei
-
07
Sep
-
07
Jan
-
08
Mei
-
08
Sep
-
08
Jan
-
09
Mei
-
09
Sep
-
09
Jan
-
10
Mei
-
10
Sep
-
10
Jan
-
11
Mei
-
11
Consumption Imports
GDP
80 Do Sales of Men’s Underwear Really Predict the State of the Economy?
little more complicated; not least because the datasets are incomplete for Indonesia and
Malaysia. Certainly for Indonesia consumption appears to flatten out in 2004, whilst GDP
surges ahead (possibly a reflection of the inflation accelerating in 2004).
(a) (b)
(c) (d)
(e)
Figure 2 Men’s Underwear (a) Czech Republic; (b) Finland; (c) Hong Kong; (d) Indonesia
(e) Malaysia
Given the severe problems some of the smaller European economies are facing, it is interesting
to look at the profiles for Ireland (Figure 3a) and Portugal (Figure 3b). Again, there is a
discernable pattern, but once more the relationship is complex. Certainly, men’s underwear
sales, if a predictor, do not appear to discriminate between more dynamic and weaker
economies.
In other parts of the World there are a number of other good examples, of an apparent
relationship. For example, Armenia (Figure 3c) displays a strong relationship, whilst Mexico
(Figure 3d) appears to have a good relationship, but with a degree of complexity. Nevertheless
across all the countries discussed, a relationship does appear to exist and that if there is a time
lag between sales and the economy it appears to be less than one year.
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
GDP
Cons
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
GDP
Cons
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
GDP
Cons
2000
2001
2002
2003
2004
2005
2006
2007
GDP
Cons
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
GDP
Cons
Smith 81
Unfortunately, a simple linear correlation across the 56 countries, for which the data is
sufficiently robust, appears to disprove any relationship, as overall it was -0.204, meaning that
there is a negative relationship between sales of men’s underwear, and the economy! Even
though they initial results appear to show that the phenomenon crosses national and cultural
boundaries, the lack of a correlation may simply suggest that the relationship holds true for
some countries and not for others.
(a) (b)
(c) (d)
Figure 3 Men’s Underwear (a) Ireland; (b) Portugal; (c) Armenia; (d) Mexico
Nevertheless the differences by country appear to defy any natural logic, with the relationship
being strong for: Kuwait (0.903), Armenia (0.914), Ecuador (0.917); and to be strongly negative
for: Cyprus (-0.966), France (-0.955), Poland (-0.954), Mexico (-0.940, despite visually
appearing to have a positive relationship), Spain (-0.917), Australia (-0.901).
Even though there
are some strong positive and strong negative correlations, for most countries the relationship is
less apparent. Indeed when the correlations are plotted on a graph, they appear to be totally
random.
Table 1 Correlation of men’s underwear sales to GDP by country
Country Correlation Country Correlation
Armenia 0.914 Italy - 0.378
Australia - 0.901 Japan - 0.424
Austria - 0.463 Kazakhstan - 0.871
Azerbaijan 0.772 Kenya 0.799
Belarus 0.569 Kuwait 0.903
Belgium - 0.102 Kyrgyzstan 0.403
Bolivia 0.108 Latvia 0.819
Brazil - 0.561 Lithuania - 0.204
Bulgaria - 0.725 Malaysia - 0.774
Chile - 0.546 Mexico - 0.940
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
GDP
Cons
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
GDP
Cons
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
GDP
Cons
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
GDP
Cons
82 Do Sales of Men’s Underwear Really Predict the State of the Economy?
Table 1 Correlation of men’s underwear sales to GDP by country (cont.)
Country Correlation Country Correlation
Colombia - 0.640 Moldova 0.839
Croatia - 0.547 Mozambique - 0.522
Cuba 0.838 Myanmar 0.639
Cyprus - 0.966 Oman - 0.169
Czech Republic 0.242 Panama - 0.829
Denmark - 0.813 Poland - 0.954
Ecuador 0.917 Portugal - 0.756
Estonia - 0.371 Romania - 0.886
Finland 0.759 Russian Federation - 0.798
France - 0.955 Slovakia - 0.483
Germany - 0.565 Slovenia - 0.889
Greece 0.269 South Africa - 0.743
Hong Kong SAR - 0.692 Spain - 0.917
Hungary - 0.726 Sweden 0.269
Iceland 0.180 T. F. Yug. Rep.
Macedonia - 0.667
Indonesia - 0.556 Turkey 0.443
Iran (Islamic Rep. of) - 0.734 Ukraine 0.352
Ireland 0.346 United Kingdom 0.269
Average - 0.204
Figure 4 UN Correlation
However, these correlations relate to economies across an economic cycle, when the original
theory suggests that sales are a predictor of recession! The data relates to 1995 to 2008, a period
of growth for most economies. Nevertheless we can identify recession in Columbia, Ecuador,
Hong Kong, Iceland, Japan and Turkey, but again taken overall the data looks random.
Another way of looking at the relationship between sales and the economy is to correlate sales
to growth in GDP. The overall correlation for all countries is -0.623. This appears to suggest
that a negative relationship exists for men’s underwear sales and economic growth!
Table 2 Correlation of men’s underwear sales to GDP during a
recession by country
Country Correlation
Correlation Recess
Difference
Colombia - 0.640 0.642 1.281
Ecuador 0.917 - 0.026 - 0.944
Hong Kong SAR - 0.692 0.477 1.169
Iceland 0.180 - 0.811 - 0.991
Japan - 0.424 - 0.456 - 0.032
Turkey 0.443 0.619 0.175
Smith
Figure 5 UN correlation recession Figure 6 UN growth correlation
It is possible that men’s underwear sales are a useful economic indicator
but much rests on the quality of the data being used (including the time intervals for its rel
and its interpretation.
Nevertheless traditional econometric forecasting is usually very poor at
predicting turning (or tipping) points,
is tempting to track something like men’s underwear sales, but these should only be used with
extreme caution.
Looking at the global economy the overall correlation for all countries for s
was -0.623.
Suggesting a negative relationship between for men’s underwear sales and
economic growth! Although counter
• M
en (or their partners) buy new men’s underwear during difficult econom
cheer themselves up; or the more likely,
• M
en (or their partners) stockpile men’s underwear when it is discounted during a
recession;
• A
ware of the theory that men’s underwear sales are the first to be affected by a
recession, retailers discount
stimulating demand.
Whilst the relationship is not strong enough to really establish a new theory, it certainly appears
to disprove the original logic behind applying a Men’s Underw
Nevertheless, it is important to recognize the limitations of the data used and the methodology
that h
as been applied in this paper.
Elmer (2011) concluded that whilst a basket of
recession, he did not consider underwear to be a particularly important constituent of that
basket.
4. CONCLUSION
Overall the investigation appears to suggest that men’s underwear sales are unrelated to the
st
ate of the economy. Indeed far from being a highly price elastic good, their sales are more
comparable
to those of a basic commodity.
be somewhat unaffected by the trade cycle, in other their movements, al
simply follow the economy.
Therefore, for most countries it would be unwise (if not foolhardy)
to construct a Men’s Underwear Index to predict movements in the economy.
If data on men’s underwear sales is available before data on the
to gain some insight into economic movements. In the main, one could only advocate using a
MUI to predict recession and only after some more robust analysis has been done to understand
the relationship (strength of the c
lags exist. Countries which might w
and the USA.
Although a much more rigorous analysis than has been applied in
would be required.
All other countries should treat the MUI as an interesting apocryphal and
Figure 5 UN correlation recession Figure 6 UN growth correlation
It is possible that men’s underwear sales are a useful economic indicator
but much rests on the quality of the data being used (including the time intervals for its rel
Nevertheless traditional econometric forecasting is usually very poor at
predicting turning (or tipping) points,
going into recession and coming out of one. Therefore, it
is tempting to track something like men’s underwear sales, but these should only be used with
Looking at the global economy the overall correlation for all countries for s
Suggesting a negative relationship between for men’s underwear sales and
economic growth! Although counter
-
intuitive there are a number of possible explanations:
en (or their partners) buy new men’s underwear during difficult econom
cheer themselves up; or the more likely,
en (or their partners) stockpile men’s underwear when it is discounted during a
ware of the theory that men’s underwear sales are the first to be affected by a
recession, retailers discount
men’s underwear at the first sign of a recession, thereby
Whilst the relationship is not strong enough to really establish a new theory, it certainly appears
to disprove the original logic behind applying a Men’s Underw
ear Index to
Nevertheless, it is important to recognize the limitations of the data used and the methodology
as been applied in this paper.
Certainly in a recent study of the Swiss economy Simon
Elmer (2011) concluded that whilst a basket of
indicators may provide a useful predictor of
recession, he did not consider underwear to be a particularly important constituent of that
Overall the investigation appears to suggest that men’s underwear sales are unrelated to the
ate of the economy. Indeed far from being a highly price elastic good, their sales are more
to those of a basic commodity.
For many countries men’s underwear sales appear to
be somewhat unaffected by the trade cycle, in other their movements, al
Therefore, for most countries it would be unwise (if not foolhardy)
to construct a Men’s Underwear Index to predict movements in the economy.
If data on men’s underwear sales is available before data on the
economy, it is tempting to use it
to gain some insight into economic movements. In the main, one could only advocate using a
MUI to predict recession and only after some more robust analysis has been done to understand
the relationship (strength of the c
orrelation and the causal relationship) and what, if any time
lags exist. Countries which might w
ant to construct a MUI include:
Armenia, Ecuador, Kuwait,
Although a much more rigorous analysis than has been applied in
All other countries should treat the MUI as an interesting apocryphal and
83
Figure 5 UN correlation recession Figure 6 UN growth correlation
It is possible that men’s underwear sales are a useful economic indicator
of the US economy,
but much rests on the quality of the data being used (including the time intervals for its rel
ease)
Nevertheless traditional econometric forecasting is usually very poor at
going into recession and coming out of one. Therefore, it
is tempting to track something like men’s underwear sales, but these should only be used with
Looking at the global economy the overall correlation for all countries for s
ales to GDP growth
Suggesting a negative relationship between for men’s underwear sales and
intuitive there are a number of possible explanations:
en (or their partners) buy new men’s underwear during difficult econom
ic times to
en (or their partners) stockpile men’s underwear when it is discounted during a
ware of the theory that men’s underwear sales are the first to be affected by a
men’s underwear at the first sign of a recession, thereby
Whilst the relationship is not strong enough to really establish a new theory, it certainly appears
ear Index to
global economies.
Nevertheless, it is important to recognize the limitations of the data used and the methodology
Certainly in a recent study of the Swiss economy Simon
indicators may provide a useful predictor of
recession, he did not consider underwear to be a particularly important constituent of that
Overall the investigation appears to suggest that men’s underwear sales are unrelated to the
ate of the economy. Indeed far from being a highly price elastic good, their sales are more
For many countries men’s underwear sales appear to
be somewhat unaffected by the trade cycle, in other their movements, al
ong with other goods,
Therefore, for most countries it would be unwise (if not foolhardy)
to construct a Men’s Underwear Index to predict movements in the economy.
economy, it is tempting to use it
to gain some insight into economic movements. In the main, one could only advocate using a
MUI to predict recession and only after some more robust analysis has been done to understand
orrelation and the causal relationship) and what, if any time
Armenia, Ecuador, Kuwait,
Although a much more rigorous analysis than has been applied in
this paper
All other countries should treat the MUI as an interesting apocryphal and
84 Do Sales of Men’s Underwear Really Predict the State of the Economy?
stick to more traditional forms of economic forecasting. These findings are consistent with
men’s underwear being treated as a basic commodity in most economies, but in a few countries
(especially richer countries like the US) they have become disposable fashion items.
5. REFERENCES
Bai, J., Ng, S., 2008. Forecasting Economic Time Series Using Targeted Predictors. Journal of
Econometrics, Volume 146, pp. 304–317.
Box, G., Jenkins, G, 1970. Time series analysis: Forecasting and control. San Francisco: Holden-Day.
Davison, A., Hinkley, D., 2003. Bootstrap methods and their application. Cambridge: Cambridge
University Press.
Diebold, F., Rudebusch, G., 1991. Forecasting Output with the Composite Leading Index: An Ex Ante
Analysis. Journal of the American Statistical Association, Volume 86, pp. 603–610.
Elmer, S., 2011. Modern Statistical Methods Applied to Economic Time Series. ETH Zurich: KOF
Swiss Economic Institute, Dissertation Series No 6.
Emerson, R., Hendry, D., 1994. An Evaluation of Forecasting Using Leading Indicators. Mimeo.
Oxford: Nuffield College.
Greenspan, A., 2007. The Age of Turbulence: Adventures in a New World. New York: Penguin Press.
Harrod, R., 1942. Toward a Dynamic Economics: Some Recent Developments of Economic Theory and
their Application to Policy. London: Macmillan.
McGuckin, R., Ataman, O., Zarnowitz, V., 2001. The Composite Index of Leading Economic Indicators:
How to Make It More Timely. Cambridge, MA: National Bureau of Economic Research, Working
Paper 8430.
Mintel, 2008. Men's Underwear - US - December 2008. Chicago: Mintel.
National Public Radio. http://www.npr.org/templates/story/story.php?storyId=17759341 (accessed
16/8/2011).
OECD, 2008. OECD System of Composite Leading Indicators. Paris: OECD.
Rogers, R., 1998. Handbook of key economic indicators. New York: McGraw-Hill Professional.
Smith, P., 2011. The UK Apparel Market: Profiles of Ten Leading Apparel Retailers. Global Apparel
Markets, Volume 13, pp. 11-57.
Stock, J., Watson, M. 1992. A procedure for predicting recessions with leading indicators: Econometric
issues and recent experience. Cambridge, MA: National Bureau of Economic Research, Working
paper 4014.
The Economist (30/7/2011).
The Jakarta Globe (31/8/2009).
United Nations Statistics Division. http://unstats.un.org/unsd/default.htm (accessed 16/8/2011).
US Census Bureau. http://www.census.gov/ (accessed 16/8/2011).
Washington Post (31/8/2009).
Yerex, R., 2011. The Consumer-Driven Economy at a Crossroads. Business Economics, Volume 46, pp.
32-42.
