The tyranny of numbers: economic statistics are less certain than you think

Tristan Hanson, Head of Asset Allocation at Ashburton Investments.

True economic growth is less certain than some statistics would have us believe. Understanding this helps us make better informed decisions.

The initial estimate of US GDP in the first quarter of this year showed modest growth of 0.2% (annualised). The second estimate indicated a contraction of -0.7% after incorporating later releases of international trade, inventory and other data. Meanwhile, research from the Federal Reserve Bank of San Francisco estimates that US growth in Q1 might be underestimated by around 1.5% as a result of badly specified seasonal adjustments. So did the US economy grow in Q1 or did it contract? In truth, we don’t really know.

More controversial have been the results coming from the revised methodology to calculate GDP growth in India. Using the old method, growth in the year ending March 2014 was 4.7%; based on the new calculation the Indian economy grew by 6.9% during the same period. More recent data (using the new methodology) suggests India is now the fastest growing major economy in the world, ahead of China, another country for which GDP calculations have been routinely questioned over the years.

It is not just growth rates which have been subject to statistical revisions; in some countries there have been huge adjustments to the level of GDP. Famously, last year, Nigeria’s economy was estimated to be 89% larger after revising the method of estimating output in the economy, making it Africa’s largest economy ahead of South Africa.

Economic statistics are subject to considerable uncertainty

These statistical curiosities highlight a very important point: Gross Domestic Product, a measure of the value of final goods and services produced by a country over a given period, is a statistical construct which is subject to considerable uncertainty.

The US example above is revealing in this context. GDP figures quoted in the media and watched by investors are on a seasonally-adjusted basis. The actual level of activity fluctuates significantly from one season to the next. For example, quoting from the San Francisco Fed paper, “On average from 2000 to 2006, nominal GDP before seasonal adjustment dropped about 10% at an annual rate every first quarter and rose about 20% every second quarter”. Of course, the US GDP figures that are reported never fluctuate in such a way because they are reported after statisticians have tried to adjust the data for these seasonal swings, smoothing out quarterly growth rates. But the numbers quoted above show that the magnitude of seasonal swings are very large in relation to the typical growth rate of the economy on a seasonally-adjusted basis, i.e. how the data is normally reported. Therefore, any miscalculation of how much to adjust activity for seasonal fluctuations might have a large bearing on the GDP figure reported in a given quarter.

Moreover, GDP data can be substantially revised many years later. For example, the original estimate for the first quarter of 2008 (NB: months before Lehman Brothers collapsed) was released on 30 April 2008, indicating quarter-on-quarter annualised growth of 0.6%. This was subsequently revised higher to 1.0% growth by the time the third estimate for Q1 was released on 26 June, 2008. But subsequently, in the 2009 annual revision, it was decided that the US economy actually contracted at a rate of -0.7% in Q1, 2008. In July 2013, more than five years after the period in question, further revisions showed that the economy had been contracting at a much more alarming rate of -2.7%.

A lot of noise?

It is easy to think this is all noise and something just for ‘anorak’ economists to get excited about. After all, inadequate adjustments for seasonality will boost growth in some quarters and lower it in others, but over an entire it year should wash out.

If economic statistics were understood for what they really are, this might be true.

But, typically, this is not how economic data is reported in the media or among market commentators. Great significance is placed on GDP figures when they are released and sometimes financial markets react abruptly, only for this ‘news’ to be revised at a future date. Political careers may be made or lost depending on what (sometimes flaky) data is reported. A country’s debt sustainability is typically measured by looking at debt or deficit ratios to GDP, yet, as we have seen, in poor countries particularly, GDP levels can be very wrongly estimated. Often, GDP is mistakenly used synonymously with national welfare, something which it is clearly not (even if the two might be correlated).

Implications for investors

Did the US economy contract in Q1? Is India growing at 5% or 7%? The honest answer is: we can’t be sure. And it has important implications for investors.

For stock market investors, it is the outlook for profits or cash flows which matter, not GDP. Accounting peculiarities (or even outright fraud) might mean that company accounts and profits can be manipulated, but a share price is at least an estimate of the discounted value of future cash flows – it is certainly not a discounted value of future GDP figures, a statistical measure, nothing more.

Profits and GDP can diverge significantly and for prolonged periods of time, as has been the case in the US and Japan over recent decades, to use just two examples. Even if one could estimate and forecast GDP accurately, it would be a mistake to think profits would follow the same path or assume that the two would necessarily ‘mean revert’ over a time horizon that most typical (even patient) investors would consider useful, say perhaps a few years or so4. Sometimes mean reversion might be the case, but often not as other economic, institutional and social factors such as the relative abundance of capital and labour, the level of competition and societal norms regarding capitalism and distributive justice exert a major influence.

Similarly, for bond investors it is the average level of future interest rates which is important and uncertainty over the distribution of risks around those future interest rates. Of course, estimates of GDP and inflation will play a role in determining the market’s expectations for interest rates…but it is ultimately interest rates which matter. How else could one explain why US 5yr real bond (TIPS) yields have been persistently negative since mid-2010, a period in which the US economy has achieved real GDP growth at a rate above 2% per annum? The explanation is simple: the real Fed Funds rate has remained negative throughout the period and is expected to do so a little while longer.

When it comes to sovereign debt sustainability, ultimately it is not the Debt to GDP ratio which matters. In the short-term, it is the government’s ability to refinance its debt and, longer-term, generate revenue to cover its spending in a sustainable fashion. While government deficit or debt to GDP ratios are typically useful indicators in rich countries, they may be less helpful in poor countries. The tangible things that matter to a finance minister under pressure are the cash flows in and out of the government’s coffers and the nature of any central bank backstop or printing press, not ratios based on an ethereal GDP figure.

Conclusion

Recent debates over GDP statistics in the US, India, Nigeria and elsewhere highlight the tyranny of economic data. Our knowledge of what is going on in an economy in real time is much less precise than the media or most market commentators would have us believe.

Fortunately, stock market investors buy a share of ownership in a company’s future cashflows, not GDP. While profits growth in a country will often correlate positively with GDP growth, there can be large differences. Sometimes powerful fundamental forces drive this divergence. Other times, it might be the case that the composition of the stock market index is very different to that of the domestic economy. And while accounting profits can be manipulated, for bottom-up investors it will generally be those who do the best analysis that will be able to identify which individual companies are playing fast and loose with their reporting.

The more an investor is able to understand the nuances of macroeconomic data – and the limitations of our knowledge – the more it should translate into better informed decision-making and portfolio risk management.