GDP contraction: worse than projected?

That Covid-19 will hit hard on Pakistan economy is not news. How much will be the damage is the real question. That’s where expert forecasts come into play. Moody’s says it expects Pakistan’s economy to contract by 0.1 to 0.5 percent in FY20; the central bank expects it to contract by 1.5 percent, whereas the IMF and the World Bank also expect Pakistan’s GDP to contract in the range of 1.3 to 1.5 percent.  How much should one believe in these forecasts?

Since neither of these organizations share the details of their forecasts in terms of GDP projections by its key components of major crops, manufacturing, service sectors, and so forth, it is rather difficult to look under the hood to closely assess the reasonability of these expert forecasts.

A crude measure, therefore, and admittedly rather crude, is to look at the forecasts made by some of these organizations in the past and compare it with the actual GDP numbers. The idea isn’t naming and shaming; it’s understood that humans are not all-knowing, and that it’s okay to be wrong. The idea is to assess the degree to which forecasts are off from the actual and to gauge underlying trends thereof if any.

To this end BR Research dug up IMF's, World Bank's, and ADB's yesteryears' GDP forecasts made for Pakistan, China, India, EU, US and the World, and compared those with actual GDP growth recorded for these countries and blocs. Two types of comparisons were made: the comparison of ‘most recent’ forecast made by each organization with its earlier forecast for every given year; and comparison of actual GDP with the most recent forecast by each organization. These comparisons were run for the years 2008 to 2018, hoping it is a decent enough period for quick non-academic analysis.

The key findings: in majority of data readings (about 65% of total readings) ‘most recent’ forecasts made by each of these organizations are downward revisions of their earlier forecasts. And second, in majority of data readings (about 55%), actual GDP was lower than the ‘most recent’ forecasts made by each of these organizations. Bear in mind that ‘most recent’ forecasts are usually made at least after six months have already passed in a given fiscal year.

This general observation of overestimation or optimistic projections is also observed in the quantum of overestimation. On average the quantum of overestimation of GDP growth (calculated by comparing actual GDP growth with ‘most recent' forecasts) is higher than the size of underestimation of GDP growth for each of the countries and blocs look at for this purpose.

In the case of Pakistan, an even more interesting observation emerged that GDP growth is overestimated in bad years when economic growth is slowing for one reason or another. In other words, these organizations (as well as the central bank) are not as good in forecasting bad news, as they are in good year or business-as-usual years.

The moral of the story: if (and granted that it’s a big if) these comparisons are any guide, forecasts of GDP contraction offered by the SBP, the World Bank and the IMF are optimistic; Pakistan’s actual GDP  contraction in FY20 (and recovery, if any in FY21) may be worse than currently projected especially considering that the expected contraction stems from a once in a century event, which has left all forecasters scratching their heads.