How accurate are economic reports?

“I don't make jokes. I just watch the government and report the facts.” — Will Rogers, quoted in Saturday Review, Aug. 25, 1926 Can we believe what Washington, D.C., tells us? Forecasters have no choice but to rely on information from the nation's capital.There are several reasons for this. Government statistics are the raw material for most forecasting models, and projections and analyses by various government organizations can be key inputs shaping thinking about the future. Analyses by Washington-based public policy interest groups and lobbyists are often influential as well. The challenge for forecasters is to assess the credibility of this information. Government statistics are subject to substantial revisions. And very different answers to the same questions come from the government agencies and organized interest groups inside the Beltway. Thus, forecasters are regularly challenged to sift through the welter of statistics and analyses coming out of Washington and determine what information to trust when assembling their forecasts.

“COOKING THE BOOKS”?

On Oct. 5, 2012 — just a month before the presidential election — the financial markets and most forecasters were surprised when the Bureau of Labor Statistics reported that the civilian unemployment rate in September had dropped by 0.3 percentage points to 7.8%. For supporters of President Barack Obama's reelection campaign, it was good news. When Mr. Obama was inaugurated in January 2009, the civilian unemployment rate stood at 7.8%. Despite the $787 billion in fiscal stimulus embodied in the American Recovery and Reinvestment Act signed by the president in February 2009, unemployment kept rising to 10% in October 2009. It subsequently declined slowly, but remained well above its level when Mr. Obama assumed office — a fact that Republican challenger Mitt Romney repeated to the American electorate on an almost daily basis during the 2012 presidential election campaign. In the eyes of Mr. Obama's supporters, the sharp decline in the September unemployment rate confirmed their confidence and proved that his economic policies were producing results. However, some Republican supporters of former Massachusetts Gov. Romney cried foul. Jack Welch, the former chairman of General Electric Corp., charged on Twitter: “Unbelievable jobs numbers ... these Chicago guys will do anything ... can't debate so change numbers.” (Mr. Obama was from Chicago, and in the first of three debates with Mr. Romney, the president was widely viewed as having done a poor job.) Did someone in the Obama administration order the BLS to fake the September unemployment data to aid Mr. Obama's reelection bid? Members of the Obama administration were outraged. Labor Secretary Hilda Solis responded on CNBC soon after Mr. Welch made his allegation that: “It's really ludicrous to hear that kind of statement ... I have the highest regard for our professionals that do the calculus.” While Mr. Romney lost the election, some critics of the Obama administration persisted in asserting that something had been amiss with the September 2012 unemployment rate. Just over a year later, on Nov. 18, 2013, a headline in the New York Post exclaimed: “Census "Faked' 2012 Election Jobs Report.” The Census Bureau conducts the monthly household survey the Labor Department uses to calculate the unemployment rate. According to Post business reporter John Crudele, an unnamed source claimed that the Census Bureau either allowed or overlooked employees filling out fake questionnaires in a manner that lowered the unemployment rate. As evidence of a workplace environment that was supposedly conducive to such behavior, he cited the Census Bureau's firing in 2010 — two years before the presidential election — of an employee for turning in faked questionnaires in order to exaggerate to his superiors how many surveys he was able to complete. Mr. Crudele's un-named source claimed such be-havior on the part of other Census employees continued during the 2012 election, and that the Census Bureau failed to report the problem or the earlier firing to the Labor Department. A week after Mr. Crudele leveled his charges, they were challenged by Nelson Schwartz, a business reporter for the New York Times. He quoted former Labor Department personnel who said that the size of the monthly household survey sample — 54,000 — was far too large to be materially affected by a few dozen faked responses. Moreover, for September 2012 there was other corroborative evidence of stronger job formation from the separate monthly survey of employer payrolls.

DEGREE OF SUSPICION

I mention this episode because it illustrates the degree of suspicion surrounding the veracity of government economic data. Disappointed investors and traders and angry political partisans sometimes have complained to me that some statistic unfavorable to their trading position or political cause somehow or other has been rigged to make economic conditions appear better than they really are. I characterize such allegations as representing the “conspiracy theory” of government economic statistics. It is understandable why some observers believe that the U.S. government deliberately alters potentially unfavorable economic reports. The belief probably stems in part from investors' experiences with data released by a few foreign governments. There have been instances where government budget statistics have been very misleading in countries struggling to maintain investor interest in government debt issues. Or, in the case of Mexico during the 1994 debt crisis, there were delays in the Mexican government's usual release schedule for government budget and Treasury cash data. Some observers, too, are somewhat (or more) suspicious of Chinese economic data. Why? There may be incentives for regional government officials to overstate their region's economic activity so as to please senior Communist Party officials in Beijing and obtain more financial aid that is linked to a region's size and economic performance. With the following possible exception, I do not believe that U.S. economic data are politically man- ipulated. Government agencies are subject to ever more media scrutiny. In addition, whenever a suspiciously strong data report is issued during an election campaign, politicians are quick to encourage scrutiny by both the media and congressional investigative staffs. Another consideration is that government statistical agencies are headed by career civil servants who are generally immune from political pressures and who care about their professional reputations. That said, won't an incumbent president order stepped-up pre-election government spending both to stimulate the economy and produce better economic data? Undoubtedly, this happens in many countries. On Oct. 26, 2012 — just over a week before the November election — the Commerce Department reported an unexpectedly large 2% annualized rise in real GDP growth for the third quarter. The especially surprising component was a sharp 9.6% annualized quarterly rise in federal purchases of goods and services, which accounted for 0.7 percentage points of the overall 2% gain in real GDP. Did the Obama administration order accelerated government spending for political reasons? Critics thought the rise was suspicious. After all, the surprising gain came at a time when budget legislation dictated federal spending restraint. Leading the third-quarter rise was a 13% annualized increase in defense spending, which had declined at annualized rates of 7.1% in the first quarter and a further 0.4% in the preceding quarter. Then in the fourth quarter, such spending tumbled at a 22.2% annual pace. Critics contended that the Obama administration deliberately held back defense allocations in the previous two quarters so that such spending would stimulate the economy and boost a key economic statistic on the eve of the elections. Absent very specific evidence of communications from the administration to federal agencies either suggesting or ordering such manipulation, such charges are hard to prove. Defense spending is one of the most volatile GDP components on a quarterly basis. Also, the third quarter of a calendar year is the final quarter of the federal fiscal year, creating a strong incentive for agencies to spend the types of earlier appropriated funds that must be spent or lost before a fiscal year ends.

POLITICALLY MOTIVATED?

Although I doubt government officials politically manipulate reported government economic data, I have no doubt at all that the economic forecasts issued by government agencies are politically motivated at times. There is evidence of private-sector forecasts having a strategic element. Biases can reflect the business needs of different employers and the business strategies of independent forecasting firms. Forecasters in the public sector are only human and also can exhibit similar types of strategic behavior. Instead of perceived business needs biasing their forecasts, public-sector economists and their forecasts can be influenced by their perceptions of office politics in Washington. Cost overruns in the public sector are an all too familiar example of government forecasts going awry. Are some government officials and their staffs strategically motivated to low-ball initial estimates of a proposed program's expenses? This might well have been the case with two major government cost overruns in the administration of President George W. Bush — the start of the Iraq war in 2002 and the Medicare Prescription Drug Improvement Act of 2003. Prior to the U.S. invasion of Iraq, the Defense Department underestimated the war's costs as well as troop requirements. Both the Pentagon and Congress acknowledged the troop shortages two years later. A year after the expanded Medicare prescription drug legislation was adopted, it was revealed that its costs were initially underestimated by around $150 billion. The strategic rationales for fudging forecasts of future costs should be obvious. A government forecaster might have a personal preference for a particular policy or course of action. Also, just as in the private sector, government forecasters may consider self-preservation, in the form of promotion and job security, when contemplating a forecast at odds with an employer's specific goals.

REPUTATIONAL ISSUES

On the other hand, there are reputational issues for public-sector forecasters, as there are for their private-sector brethren who may prize their professional integrity. Civil service laws, too, help to protect government employees from undue pressure from their appointed superiors. In my years (1973-80) as an economist at the Federal Reserve Bank of New York, I never felt overt pressure from my superiors to tilt my analyses and forecasts in any specific direction. However, after reviewing a report, senior management would decide on the size of the internal distribution list. The short list limited a report's exposure and was ostensibly justified by considerations of other people's interests and their lack of time to read all of the voluminous research being churned out by the staff. However, some of my colleagues confided in me their disappointment at having a hard-researched report go only to the short list. This is an excerpt from “Inside the Crystal Ball: How to Make and Use Forecasts” by Maury Harris (Wiley, 2015).