This column ran in the Financial Times on 9/26/200 and is used with permission.

Click here to see the Financial Times story

Back to the September Statistics Report

Back to the Statistics Center

The Value of Good Data

"The US statistical agencies need better funding if they are to do their job properly, argues Diane Swonk."

For years, the statistical agencies in Washington have  suffered from a combination of budget cuts and outright  neglect. The situation has now reached a state of crisis.

This year the release of second-quarter figures for US  gross domestic product nearly caused the Bureau of Economic  Analysis entire computer system to crash, an event that  threatened to compromise all the GDP accounts for a nation  known for the integrity of its official data.

If the BEA and the Census bureau do not receive higher  funding than currently being proposed by the House of  Representatives, the very relevance of the GDP data will be  called into question. Improvements to incorporate the impact  of electronic commerce and the interest, in particular, will  not be completed. Revisions to the GDP and trade releases will  have to be made quarterly, instead of monthly, further  compromising the timeliness of the data.

Business leaders and financial analysts have begun to  understand the magnitude of the problem. Statistics on the  macroeconomy shape everything from business strategy to  portfolio management. Even a rumour of a surprise in one of  these critical figures can move billions around the world in  an instant.

Policymakers are also acutely aware of the dangers of  faulty data. Recent studies suggest that the 1990 recession  might have been avoided had accurate information on the US  economy been available. Data at the time showed the US was  still in an expansion as late as October; but on revision, the  data showed that the country had slipped into recession two  months earlier, in August.  

The 1990 recession caught many, including those at the  Federal Reserve by surprise, which is understandable given the  information available at the time. If the data had been  accurate, there is a chance that the Fed would have acted  sooner to help the economy.  

More recent examples of the importance of good data  and  the risks of bad data  include the emerging market crises of  1997 and 1998. Nobody knew the severity of the situation until  it was too late to act. The result was widespread capital  flight, first from emerging Asia, and later from Latin  America, deep recessions, and broad-based financial market  turmoil.  

For years, the US statistical agencies have struggled to  get by, yet succeeded in improving the way data are  calculated. They pushed for better ways to capture inflation  when it finally hit Congressional radar screens in the 1990s,  and more recently, developed better measurements of the  contribution being made by the often intangible information  and technology sectors.

Until recently, however, complaints of the compromises that  these agencies were having to make with antiquated equipment,  uncompetitive pay packages, and the elimination of less  important (but still valuable) data series, were falling on  deaf ears. It was apparently easier for Washington to  subsidise the US mohair industry, which costs more than the  additional funding being requested by the statistical  agencies, than to ensure good economic data.

This issue, however, will be harder to ignore in future.  The National Association for Business Economics, the largest  association of economists, policymakers and strategists of its  sort in the world, has stepped up demands for better quality  and more timely data. More than 80 per cent of our members  believe that the higher funding levels being requested by the  White House and the Senate should not only be met, but  exceeded.  

Alan Greenspan, a former president of the association, said  it best to a Senate panel earlier this year: I am  extraordinarily reluctant to advocate any increase in  spending. So its got to be either a very small amount or a  very formidable argument that is involved. And I find, in this  case, that both conditions are met.

Moreover, we have found allies in almost every industry and  association we have approached. Indeed, finding enemies on the  debate is difficult. Most people on Capitol Hill even seem to  support the cause intellectually. They have been sidetracked,  however, by the need to make a flurry of promises to a host of  constituents in an election year, which do not include the  statistical agencies. The word data appears to be among the  most uninteresting four-letter words in the English language.  The result is neglect of our statistical infrastructure.

We are asking the House to agree $48.9m funding for the  BEA, $6.7m above its 1995 budget, and 173.8m for the Census  bureau.

I fear that good times breed complacency, and that  election-year politics will overshadow sound policymaking.  Support for the statistical agencies may allow us all to enjoy  the prosperity we have come to know a little longer, no matter  the outcome of the election. The return on such a small  investment will be felt worldwide.

The writer is chief economist for Bank One Corporation and the  immediate past-president of the National Association for  Business Economics.