One reason economists sometimes compare themselves to astrologers is that their forecasts are often equally accurate. But when it comes to analyzing government policy, economists have trouble even figuring out what happened in the past.

Case in point: The most recent Congressional Budget Office (CBO) report of President Obama’s $831 billion “stimulus” bill—also known as the American Recovery and Reinvestment Act (ARRA).

The Administration vowed that the bill would “save or create” (a squirrelly measure in itself) 3 million to 4 million jobs. Last week, CBO’s assessment—the 10th quarterly analysis required under the law—estimated that the ARRA had increased the number of people employed by somewhere between 300,000 and 2 million—far fewer than the Administration promised. The gap is similar to the difference between flying in a commercial airliner and orbiting the earth in a space shuttle.

Except for this inconvenient fact: 400,000 fewer Americans were working last month than had jobs in February 2009, when the ARRA was enacted (132.8 million then, 132.4 million in January 2012). In other words, far from saving or creating jobs, the stimulus has lost nearly a half million of them. The unemployment rate is higher than the Administration predicted, too.

The discrepancy is not the result of CBO incompetence. CBO’s analysts are in fact serious, skilled professionals. The basis of the problem is that CBO can only estimate the results of the stimulus, a fact that CBO doesn’t hide. Note, for example, how many times the word estimate turns up in the following passage from the CBO report (reading it out loud may help):

Multiplying estimates of those per-dollar effects [from stimulus spending] by the dollar amounts of each element of ARRA yields an estimate of the law’s total impact on output. To produce estimates of ARRA’s total impact on employment, CBO combined the estimate with estimates of how changes in output affect the unemployment rate and participation in the labor force. [Italics added.]

Given that the law requires CBO analyze the impact of the stimulus, it has to estimate its impact because it is next to impossible to measure the actual real-world results of the stimulus. But that doesn’t mean CBO has to rely on estimation techniques that are a proven failure.

CBO uses the same theories, models, and calculations used to predict how the stimulus would work in the first place to now estimate its supposed results. Thus the conclusions are self-fulfilled prophecies, about as sound as a horoscope.

CBO’s estimates before the stimulus became law, and its analysis of the impact since, are wrong because they swallow the same Keynesian Kool-Aid that saturates the whole stimulus concept. CBO, and everyone else that fell for the logic that government spending could create jobs, forgot that before Congress can spend a dollar, it first must take that dollar out of the economy by taxing it away from the families and businesses that earned it or borrowing it from credit markets.

When the government engages in Keynesian spending programs such as the stimulus, it simply shifts resources from where the market allocates them to where Congress desires they go. It’s like moving water from one end of the pool to the other and expecting the water level to rise.

Overall, CBO concedes:

Although CBO has examined data on output and employment during the period since ARRA’s enactment, those data are not as helpful in determining ARRA’s economic effects as might be supposed because isolating the effects would require knowing what path the economy would have taken in the absence of the law. Because that path cannot be observed, the new data add only limited information about ARRA’s impact.

Until CBO, and other economists, refine their estimating techniques to include the jobs that the economy foregoes because Congress has reshuffled resources, they will continue to vastly overestimate the benefits of the stimulus. Meanwhile, they would be as well off consulting the zodiac.