2013-06-26 08:38:21
Jun 22nd 2013
IN DECEMBER 2004 Microsoft paid a massive $33 billion dividend to its
shareholders. The largest payment of its kind, it made up 6% of the increase in
Americans personal income that year. Examples of how big firms can have a big
impact do not come much starker. These kinds of firm-specific shocks are
typically excluded from economists models, which assume that individual
businesses ups and downs tend to cancel each other out. Yet to understand how
things like trade and GDP evolve, tracking the biggest companies is essential.
At first sight, the numbers seem to justify taking a top-down view. The
business world is huge: America has around 27m firms, Britain 4.8m. Each
country trades with hundreds of other countries across hundreds of industries,
producing thousands of country-industry trade links. The global network runs to
the millions. Because economies are built of millions of firms and trading
relationships, each seems like a speck of dust: individual companies and export
channels should not matter. This suggests that only common shocks can explain
aggregate fluctuations: a workers strike at one firm is not enough, but a
general strike is.
Yet aggregate shocks do not explain volatility very well. A 2007 Bank of Spain
paper* studies OECD countries trade balances. Common shocks (to whole
countries or global industries) explain only 45% of the variations. Hunting for
the cause of the other 55% of trade fluctuations, the authors used finer data
on 8,260 country-industry flows (59 industries and 140 trading partners) for
each OECD member. The data show that the picture of trade as millions of links
is inaccurate; in fact, flows are extremely concentrated. Most links are
unimportant. For America 99% of trade flows accounted for just 25% of trade.
But a few are vital: for the average OECD country the 25 main country-industry
flows explain two-thirds of trade, and the 100 largest 85%.
Even such detailed data mix lots of firms. The flow of cars between America and
Japan includes GM, Ford and Chrysler with Toyota, Nissan and Honda. So the
researchers dug down another level to study individual companies. In a case
study for Japan they found yet more concentration: the top five Japanese firms
accounted for 20% of exports. That suggests that trade volatility, an aggregate
statistic, could stem from just one firm s behaviour.
Some companies are certainly big enough to have that sort of effect. In America
the 2008 census showed that 981 firms with 10,000 or more staff account for a
quarter of all jobs. Of Italy s 4.5m firms, 96% are micro SMEs with fewer
than ten people; a giant like Fiat, a carmaker, sits at the other end of the
scale. Samsung alone notched up 17% of South Korea s exports in 2011. Finland
is perhaps the most extreme example, with Nokia, a telecoms titan, contributing
20% of exports and 25% of GDP growth on its own between 1998 and 2007.
This is a problem for the top-down view of the economy. A 2011 paper by Xavier
Gabaix of New York University explains how diversification works when firms are
independent and their sizes follow a regular bell-shaped distribution.
Imagine an economy where one firm produces everything: its volatility of
earnings determines volatility in GDP. But as the number of firms grows GDP
volatility shrinks, because firms shocks cancel out. With 100 firms,
volatility falls to a tenth of the level in a one-firm economy; with 1m firms,
it falls to a thousandth. Since there are more firms than this,
company-specific shocks disappear.
That s the theory, at any rate. When the distribution of firms has fat tails
(ie, there are more very small firms and very big ones) the theoretical
relationship breaks down. An economy now needs 22,000 firms for volatility to
fall to a tenth of the level in a one-firm economy (there would never be enough
firms for it to fall to a thousandth). The logic of diversification fails when
companies are sufficiently large. Firm-specific shocks do matter.
Mr Gabaix tests this new granular theory against data for the largest 100
American firms between 1951 and 2008. Stylised facts support his hypothesis.
There is a fat tail of very big firms: the 100 largest had sales equivalent to
35% of GDP in 2009, up from 30% in the mid-1980s. Their performance is
volatile: sales fluctuate by an average of 12% a year. And the correlations
between firms are low, suggesting shocks are firm-specific rather than
economywide. Next Mr Gabaix examines how well shocks involving these big firms
explain changes in GDP. Very well, it turns out. Up to 48% of the volatility of
American GDP can be traced to the performance of individual big firms.
On the rocks
The importance of the goliaths extends to trade. A widely held view is that
trade lowers volatility: exporting to more markets means greater
diversification. But in a 2012 paper Julian di Giovanni of the IMF and Andrei
Levchenko of the University of Michigan find that more foreign trade exposes
economies more to the fortunes of large firms, since they trade
disproportionately.
Central bankers should take note. Mr Gabaix shows that taking account of
firm-specific shocks can help improve economic forecasts. The models that
determine economywide decisions like those run by the Federal Reserve before
its meeting this week might be improved by looking at how big firms are doing.
There is also something to chew on for governments. The granular approach shows
that what is good for GM really is good for America. But the converse also
applies: when big firms do badly, everyone suffers. Ford, GM and Chrysler
employ close to 0.5m people. When markets tanked in late 2008, they
successfully tapped the taxpayer for billions in bail-out money. The problems
of being too big to fail stretch far beyond banking.
Sources
Trade Patterns, Trade Balances and Idiosyncratic shocks , by Claudia Canals,
Xavier Gabaix, Josep M. Vilarrubia and David Weinstein, 2007, Banco de Espa a,
Documentos de Trabajo
Country Size, International Trade, and Aggregate Fluctuations in Granular
Economies , by Julian di Giovanni and Andrei Levchenko, 2012, Journal of
Political Economy
The Granular Origins of Aggregate Fluctuations , by Xavier Gabaix,
Econometrica, May 2011
Nokia and Finland in a Sea of Change , ETLA Research Institute of the Finnish
Economy