Forecasting GDP: The fight over China’s GDP figure

Forecasting GDP: The fight over China’s GDP figure

Distrusting China’s official numbers but lacking credible alternatives, analysts should give less weight to GDP as an indicator of economic health and newspapers should give less weight to GDP short-term forecasts.

What is the current level of China’s GDP growth? 7% (official), 5,5% (Capital Economics), or 4% (Citi)?

In fact, nobody knows. Official data is not trusted, but credible alternatives appear hard to build. Nevertheless, GDP forecasts for the Chinese economy constantly make the headlines of newspapers, the latest buzz from Citigroup’s Willem Buiter being that China will be in recession next year, while last week dedicating an official announcement of 6.9% growth for Q3.

A political tale: Are the numbers made up?

Many regard China’s official GDP numbers with suspicion. It has been common to point to the fact that the National Bureau of Statistics (NBS) announced 7.0% for GDP growth in 2015H1, exactly meeting the 2015 official target.

Some underline that the addition of all the provincial GDP exceeds national GDP. Others reported Chinese GDP growth has been the least volatile of its kind compared other countries. Finally, one usually points to the leaked 2007 comments of Li Keqiang, now-Premier and then-Liaoning province chief, that GDP data were man made and that one would be better off looking at electric consumption, rail freight, and bank lending.

The arguments range from the fact that there exist widespread errors in GDP reporting to the traditional assumption figures twisted for political reasons in the end. The reasons for this high level of political sensitivity would be the importance given to quarterly GDP announcements in the business media and the benefits from meeting targets and maintaining economic optimism.

Many experts actually agree with the fact that quarterly GDP data is being smoothed – most likely through the change in the GDP deflator. But does this render the numbers irrelevant?

The first problem with such a conspiracy view is that one doesn’t have to go very far in the range of indicators to get concrete evidence for a significant Chinese slowdown: residential real estate construction and manufacturing data which are also very much looked into by analysts are a case in point. Why would the NBS also make up those other indicators if the goal were to display an improved view of the Chinese economy across the board?

What (credible) alternative?

The second problem is the absence of credible alternatives. Every once in a while, one comes up with a supposedly better indicator to reflect China’s economic situation. Tyler Cowen has recently pointed to beer sales whose growth has been flat last year. Why not take beer as the measure of China growth instead?

For quite some time now, researchers have used Li Keqiang remarks highlighted above to create a more precise measure of GDP growth than the official numbers released. But this approach also has several problems: the rapidly growing service sector is not taken into account in such an index.

Removing services would indeed be a major mistake. However, adjusting the index with an estimate for services would also be hard to come by. On the contrary to industrial growth, good monthly data on services don’t exist.

Citi research, which has produced the China-led recession paper, uses a Li Keqiang index, which in their words is “subjectively adjusted for the growing weight of the service sector…suggesting a likely number of 4% or less”.

Titanic attempts to create parallel data series for industrial output from micro data like the one Harry Wu from the Conference Board has built over 20 years of work doesn’t seem to take us that far as well – battered by insurmountable methodological problems.

The official industrial data that Mr Wu holds as suspect at the moment highlights a decline that is sharper than the GDP growth rate. Consequently, the problem, if any, can’t come from errors on the industrial side for the current period.

Even though the most optimistic numbers for short-term GDP may be chosen by the NBS among the range of plausible GDP outcomes, official data seems to be reliable over the long run. Several stress tests of official numbers with alternative deflators run by Carsten Holz of Hong University of Science and Technology, one of the main defenders of official statistics, have concluded that the official rate still appears as the best guess.

A paper by researchers at the Central Bank of Finland (BOFIT) has also found no systematic errors in the official rates when compared to the evolution of several other macroeconomic indicators.

Fights over specific quarterly figures actually don’t make much sense when one is reminded that GDP is computed with significant error margins and constantly subject to revisions in the following quarters and years, as more precise data comes in. This is the case for the United States, France and others, and should be even more the case for China, as a developing country with weaker accounting institutions.

Missing the bigger picture

The final problem is, therefore, the obsession on quarterly GDP numbers. As GRI pointed out in a recent article, one should look beyond the short-term slowdown.

Not only do skeptics focus too much on industrial data at a time when the service sector is significantly expanding, but they also also miss, as Nick Lardy of the Peterson Institute points out, the broader context: The transition towards a new model of economic growth that relies more on consumption and services than the investment model of the previous decades which had proved unsustainable in multiple regards.

All in all, a more accurate description of the Chinese economy would arise from a lesser importance given by the media to a single figure and to “subjective” forecasts.

Categories: Asia Pacific, Economics

About Author

Etienne Lepers

Etienne has worked as economist for the OECD, the European Systemic Risk Board at the ECB, the European Parliament, and in the country risk department of Coface. He holds master degree from LSE and Sciences Po.
The views expressed here are my own and do not represent the views of any institution I worked for.