While there may be some benefits from a warmer climate, it seems almost certain that upheaval on this scale would make the United States, and the world as a whole, poorer than it would be otherwise. How much poorer? If ours were a preindustrial, primarily agricultural society, extreme climate change would be obviously catastrophic. But we have an advanced economy, the kind that has historically shown great ability to adapt to changed circumstances. If this sounds similar to my argument that the costs of emissions limits would be tolerable, it ought to: the same flexibility that should enable us to deal with a much higher carbon prices should also help us cope with a somewhat higher average temperature.

But there are at least two reasons to take sanguine assessments of the consequences of climate change with a grain of salt. One is that, as I have just pointed out, it’s not just a matter of having warmer weather — many of the costs of climate change are likely to result from droughts, flooding and severe storms. The other is that while modern economies may be highly adaptable, the same may not be true of ecosystems. The last time the earth experienced warming at anything like the pace we now expect was during the Paleocene-Eocene Thermal Maximum, about 55 million years ago, when temperatures rose by about 11 degrees Fahrenheit over the course of around 20,000 years (which is a much slower rate than the current pace of warming). That increase was associated with mass extinctions, which, to put it mildly, probably would not be good for living standards.

So how can we put a price tag on the effects of global warming? The most widely quoted estimates, like those in the Dynamic Integrated Model of Climate and the Economy, known as DICE, used by Yale’s William Nordhaus and colleagues, depend upon educated guesswork to place a value on the negative effects of global warming in a number of crucial areas, especially agriculture and coastal protection, then try to make some allowance for other possible repercussions. Nordhaus has argued that a global temperature rise of 4.5 degrees Fahrenheit — which used to be the consensus projection for 2100 — would reduce gross world product by a bit less than 2 percent. But what would happen if, as a growing number of models suggest, the actual temperature rise is twice as great? Nobody really knows how to make that extrapolation. For what it’s worth, Nordhaus’s model puts losses from a rise of 9 degrees at about 5 percent of gross world product. Many critics have argued, however, that the cost might be much higher.

Despite the uncertainty, it’s tempting to make a direct comparison between the estimated losses and the estimates of what the mitigation policies will cost: climate change will lower gross world product by 5 percent, stopping it will cost 2 percent, so let’s go ahead. Unfortunately the reckoning is not that simple for at least four reasons.

First, substantial global warming is already “baked in,” as a result of past emissions and because even with a strong climate-change policy the amount of carbon dioxide in the atmosphere is most likely to continue rising for many years. So even if the nations of the world do manage to take on climate change, we will still have to pay for earlier inaction. As a result, Nordhaus’s loss estimates may overstate the gains from action.

Second, the economic costs from emissions limits would start as soon as the policy went into effect and under most proposals would become substantial within around 20 years. If we don’t act, meanwhile, the big costs would probably come late this century (although some things, like the transformation of the American Southwest into a dust bowl, might come much sooner). So how you compare those costs depends on how much you value costs in the distant future relative to costs that materialize much sooner.

Third, and cutting in the opposite direction, if we don’t take action, global warming won’t stop in 2100: temperatures, and losses, will continue to rise. So if you place a significant weight on the really, really distant future, the case for action is stronger than even the 2100 estimates suggest.

Paul Krugman is a Times columnist and winner of the 2008 Nobel Memorial Prize in Economic Science. His latest book is “The Return of Depression Economics and the Crisis of 2008.”