Whenever you talk about counterfactual versions of the 2009 stimulus bill, you end up with some version of the same response: The stimulus failed — after all, look how high unemployment is! — and now you’re telling me it should’ve been bigger? What’s wrong with you?

Most authorities don’t think the stimulus failed. The nonpartisan Congressional Budget Office, for instance, says it created between 1.2 million and 4.6 million jobs “compared to what would’ve happened otherwise.” IHS Global Insight, Macroeconomic Advisers and Moody’s Economy.com all estimate that the laws ultimate impact will be roughly 2.5 million jobs. Economists Mark Zandi and Alan Blinder put it at 2.7 million jobs.

Those estimates rely on sophisticated macroeconomic models as opposed to experimental data. There’s also an effort in the economic profession to study the stimulus’s effects in more direct ways — usually by focusing on a specific policy rather than the whole bill -- but I haven’t done a systematic analysis of this literature yet. Dylan Matthews and I are working on a more comprehensive look at those papers now, so stay tuned.

But though questions of how well the stimulus worked may be complicated, simply pointing out that unemployment remains high, or that the recession was worse than anyone realized back in January of 2009 and so projections made in January of 2009 haven’t held up very well, is neither here nor there. The fact that a starving man is still hungry after eating a burger doesn’t mean the burger did a bad job.

Any serious assessment of the stimulus needs to include some variation of the six words that appear in the CBO’s study: “compared to what would’ve happened otherwise.” If the stimulus created three million jobs even as the economy lost seven million jobs, the stimulus worked. If it created only one million jobs, it performed far worse than the administration promised.

Either way, the stimulus was swamped by an economic crisis it was never large enough to neutralize. But that doesn’t make it unsuccessful. That just makes its success or failure difficult to discern. The same thing would’ve happened if the recession had been milder than expected and economy had only lost a million jobs. In that world, a $787 billion stimulus that only created 500,000 jobs might look better, but it would’ve been a huge failure. If your argument can’t survive that cointerfactual — and “look at the unemployment rate!” very much can’t — then your argument is wrong.