(Bloomberg Opinion) -- Just how serious will the economic impact of the coronavirus be? Amid vast uncertainty, some very large numbers are flying around, and there’s a lot of confusion over what they mean. Peering through this fog, it’s worth noting: Authoritative official forecasters are far more pessimistic in the short term than most private-sector analysts.
My colleague Justin Fox recently drew attention to one persistent source of muddle about the numbers — namely, the U.S. habit of annualizing quarterly rates of growth. When Goldman Sachs, for instance, recently said output would fall by 34% in the second quarter, it meant that it would actually fall by around 10%. (The bigger number is what the annual decline would be if that quarterly rate persisted for a whole year.)
The usual practice of annualizing quarterly rates, questionable in the best of times, is downright absurd under current circumstances. This isn’t just a matter of understanding that there are two equally valid ways of presenting the numbers and needing to know which one is being used. Extrapolating a quarterly change to an annual rate, when the change in the quarter is due to a sudden and extraordinary economic lockdown, is plain wrong.
Imagine, for instance, that output will fall as Goldman expects, by 10% in the second quarter. This rate of decline won’t persist. Goldman doesn’t expect it to persist. Even if the recovery is less strong than hoped — even if it stalls entirely — there are no grounds to think that output will fall by another 10% in the third quarter, and then another 10% in the fourth, and then another 10% in the first quarter of 2021. Three more quarters of no recovery at all would still leave output just 10% below its level in the first quarter. Annualizing the quarterly 10% alludes to an entirely imaginary extended decline. To give a sense of what’s actually being predicted, short-term drops in output should be expressed as drops in levels — using quarterly, not annualized, rates.
So does this mean that the scale of the problem, bad as things might be, is being exaggerated? Unfortunately, it does not. Once you set aside the confusion due to annualizing rates, you notice something else — a very big gap between private estimates of the initial impact and some of the still-emerging official estimates.
Note that these official projections aren’t generally being called “forecasts.” Given the current uncertainty, it’s right to avoid that term, since standard forecasting models can’t be used in the normal way — what’s happening lies too far outside the experience captured in historical data. Whatever they’re called, the short-term outlooks that official analysts are weighing are extremely severe.
Economists at the OECD estimate the “potential initial impact of partial or complete shutdowns” to be 25% of GDP for the U.S., and more than that in Japan, Germany, France and the U.K. To be clear, this is the projected initial drop in the level of output due to the shutdown. If the shutdowns last for all of the second quarter, that means a second-quarter drop of 25% — not annualized. This week, Britain’s Office of Budget Responsibility released a “coronavirus reference scenario” that posits a 35% drop in GDP in the second quarter — again, not annualized. And in a recent note for the Federal Reserve Bank of St. Louis, President James Bullard described a “rough initial estimate” of a GDP drop of 50% if full lockdowns are in place. Once more, that’s a drop in the level of GDP, not an annualized rate of decline.
Incidentally, in an indication of the confusion that annualizing quarterly rates can induce, Bullard objected to annualizing rates by noting that a drop of 50% in the quarter would be a drop of roughly 200% at an annual rate — and output, of course, can’t drop by more than 100%. In fact, his arithmetic is adrift: When you extrapolate a decline, the multiplying factor is applied to a diminishing base. A 50% quarterly drop is a decline of 94% at an annual rate. Never mind. His main point stands: Annualizing quarterly changes under current circumstances makes no sense.
All these forecasts, reference scenarios, estimated impacts and rough initial estimates should be judged in the context of unparalleled uncertainty surrounding the economic outlook. If the lockdowns are eased before the end of the quarter (as seems likely), and output comes back quickly (which is debatable), the “initial impact” will be bigger than the drop in output over the quarter as a whole. That would narrow the range of projections to something in between the highest and lowest current estimates.
Nonetheless, right or wrong, most private forecasters seem much less pessimistic at the moment than the emerging official judgments about where things stand.
This column does not necessarily reflect the opinion of Bloomberg LP and its owners.
Clive Crook is a Bloomberg Opinion columnist and writes editorials on economics, finance and politics. He was chief Washington commentator for the Financial Times, a correspondent and editor for the Economist and a senior editor at the Atlantic.
©2020 Bloomberg L.P.