Regression to the mean


-A +A

IN sports, there is a phenomenon known as the “Sports Illustrated cover jinx”, which states that a player or a team featured on the cover of the sports magazine will likely have a disappointing year in the following season, or even a disappointing performance in the following game.

This is akin to the “Manager of the Month” or “Player of the Month” jinxes in the English Premier League. Managers or players who win the awards almost always do worse in their coming games.

If one is an ardent supporter of a team or player who appears on the cover of Sports Illustrated or wins those monthly Premier League awards, those honours may indeed come across as jinxes. However, if we look more deeply into it, we can tease out a different, more potent, explanation for these scenarios.

Consider players who make the cover of Sports Illustrated. Under what circumstances would they do so? Usually, it is due to some outstanding performance in their sport. The player not only has to demonstrate excellent prowess but also is likely to be at the very top of his or her game. It may be clichéd to say it, but when you are at the top, the only way to go is down.

This phenomenon is known as “regression to the mean”. Success in anything requires skill and perseverance for sure, but it also requires a lot of “luck”. Now, luck can mean many things — it can mean being at the right place at the right time, an opposing superstar player getting injured just as he is set to face you, or a sudden flash of genius.

Thus, for someone to be successful above and beyond everyone else, that player almost certainly experienced a big bout of good luck. As we know, such good fortune does not last and so, that player will likely post scores closer to his or her “average” the following season and thus, underperform relative to his or her Sports Illustrated season.

Turning to economic growth, it is well-documented in economic research that luck plays a massive role in the rise and fall of nations. Indeed, any historian will tell you that a particular sequence of events in history that led to a particular outcome (say, World War II) occurred as a result of coincidences that may have seemed to be nothing more than blind randomness or, in other words, luck.

Similarly, the development of nations may indeed diverge due to luck. Countries that are in principle identical may diverge because small factors lead them to select different equilibria, assuming that multiple equilibria exist. For instance, it may be that a country happened to be ruled by a growth-friendly dictator, while another was stuck with a growth-destroying one.

Thus, if luck plays a part in growth, and luck (or random coincidences) is a key part of the statistical phenomenon called “regression to the mean”, then it is worth asking if growth rates do experience “regression to the mean”.

In a recent working paper entitled, “Asiaphoria meets regression to the mean”, professors Larry Summers and Lant Pritchett of Harvard asked exactly that. First, they argued that growth rate differentials between India and China and modern post-crisis growth in the US and Europe have produced “Asiaphoria”, a view that the global economy will be shaped by the two Asian giants. Indeed, this is a view that is shared by many.

However, historical evidence and data show that low persistence of growth rates has been a consistent and robust characteristic across all decades. Indeed, many forecast errors of the past century came from excessive extrapolation of performance in the recent past, whereby a country’s growth rate is viewed or assumed as a permanent characteristic rather than a transient characteristic. For instance, the view in the 1960s that the USSR economy would overtake the US economy by the 1980s and the view in the 1980s that Japan would continue to grow and out-compete the rest of the world.

Therefore, when we consider the recent high growth rates of India and China, and if we believe that luck plays a role in economic growth, then we should not expect the high growth rates of China to persist. This is especially true given the regression to the mean present in the cross-national data, where historically, the growth distribution has been an average of 2% with a standard deviation of 2%. The idea that China’s growth may persist at current rates looks even less credible when we take into account the fact that if current growth rates for China held, the 20-year gain in GDP from 2013 to 2033 for China would be US$51.1 trillion!

Indeed, we have seen some regression in Chinese growth rates in recent years. Thus, according to the authors, while there is some consensus that China will not maintain 9% to 10% growth rates, even the view that China’s growth will slow to something like 7% assumes substantial persistence.

The predicted growth over the next two decades using regressions is 3.9% and the regression standard error of estimation is 1.6%, so a continuation of even 7% is two standard deviations in the tail, and a continuation of a growth rate of 9% is three standard deviations. Therefore, by regression to the mean, we should expect much lower rates.

Malaysia should take heed of these lessons. The country’s growth has decelerated in the 2000s and 2010s relative to the 1990s. What if this deceleration was due to incredible random coincidences in the 1990s and we are therefore now observing regression to the mean?

Given that, according to Pritchett and Summers, knowing the current growth rate only “modestly improves the prediction of future growth rates” relative to simply guessing that it will be the world average and thus that past growth is not very informative about future growth, particularly over longer horizons, we should be very cautious in believing that Malaysia’s potential growth rate is akin to its historical growth rates, especially those in the 1990s.

GDP growth rates should not be the primary focus in how a country is developing. There is no doubt that luck plays a major role in how economies do year on year and so, putting too much priority on achieving some target growth rate, say, 6%, is not particularly helpful, given the fluctuation of “luck”.

However, recall the earlier example of the Sports Illustrated athlete. While “luck” certainly got that athlete on the cover, his or her success is only dependent on talent, hard work and mentality. Similarly, Malaysia’s success or failure will also be dependent on her talent (economic resources), hard work and mentality (economic, social and political environment).

In conclusion, when considering the successes of a nation in generating economic growth, sufficient humility must be employed to recognise the role of luck and random chance in those successes. Yet, it is also critical to acknowledge the importance of the things a government can do to generate success — the right institutions, the appropriate policies, the formation of a culture and so on.

As long as we do the right things for our society and repudiate the wrong things — corruption, nepotism, oppression of human and civil rights, misappropriation of public funds — we will get the process of generating growth right. Whatever the final growth figure actually turns out to be will, as always, be due to luck or random coincidences.

Nicholas Khaw is an economist-in-training at Harvard Kennedy School. Prior to this, he was an assistant vice-president of the research division of Khazanah Nasional Bhd.


This article first appeared in digitaledge Weekly, on August 10 - 16, 2015.