World Service: Africa GDP & Royal twins

Summary

---

More or Less: Africa GDP & Royal Twins

Host: Tim Harford
Date: December 10, 2012
Podcast: BBC Radio 4 – World Service
Episode Theme:
Debunking and illuminating the numbers behind news stories: this episode tackles recent dramatic revisions in African GDP statistics (focusing on Ghana and Nigeria), and explores the statistical claims around royal pregnancies and the likelihood of twins.

---

1. Ghana’s Dramatic GDP Growth: Why the Numbers Suddenly Change

[00:13–03:04]

Background

• In late November, the UK saw modest economic growth of 1% in a quarter, but Ghana’s GDP “grew by 60% overnight” in 2010 after a statistical revision.
• This huge jump wasn’t due to real immediate economic growth, but because “the Ghanaian statistics hadn’t kept up.” Nigeria is also planning a similar update.

How GDP Measurement Works in Practice

• Tim Harford consults Morten Jerven (Simon Fraser University, author of Poor Numbers):
  • Ghana’s last “base year” economic estimate was from the 1990s. Over time, as the economy changes (e.g., appearance of mobile phones), older stats don’t capture new sectors.

“In the 90s Ghana just makes hot dogs and hot dog buns, but by 2011 it’s also making iPhones… what is the statistical process by which these GDP numbers get out of whack?” — Tim Harford [01:37]

“You make a base year estimation where you have more complete picture...when growth is estimated you rely on certain proxies...if the structure of the economy changes radically, a lot of growth may be missed.” — Morten Jerven [01:54]

• Implication: Many African countries’ GDP numbers can be outdated, sometimes missing significant technological and market changes.

---

2. Can We Trust Ghana’s Current GDP Figures?

[03:04–04:21]

• Sydney Casely-Hayford (Ghana business/financial analyst) argues that estimates are still unreliable due to the informal economy:

“It is uncertain exactly what degree of variation we have in our GDP figures. I would hazard a guess and say we’re probably out by about plus 10, 15, maybe 20% yet...” — Sydney Casely-Hayford [03:14]

• Tim Harford asks: “How much does it matter if GDP estimates aren’t always entirely accurate?” He points out that while Ghana is getting wealthier and statistics better, some like Jerven are concerned about the implications.

---

3. Why Accurate Numbers Matter

[03:55–04:47]

• Morten Jerven: Reliable stats are basic for governments and international organizations.

  • E.g., Ghana's statistical reclassification means it's “not eligible for concessional lending from the World Bank.”

• Todd Moss (Center for Global Development, DC): Less concerned, sees value in comparing year-to-year changes rather than across countries:

“If you’re measuring more or less the same thing year to year, you’re still getting a picture into growth or shrinkage of parts of the economy...” — Todd Moss [04:29]

---

4. Who’s Responsible for Statistical Integrity?

[04:47–05:55]

• Jerven highlights three groups needing vigilance:
  1. Data Users: “Question your evidence and...check your sources.”
  2. Data Disseminators (World Bank, UN, Eurostat): Must label and clarify their numbers and note outdated base years.
  3. Data Producers: Ultimately responsible for the quality of statistics.

---

5. The Real Problem: Neglect of Statistical Offices

[05:43–06:26]

• Shanta Devarajan (World Bank Chief Economist, Africa): Argues statistical offices are often under-resourced and ignored.

“In many countries, the Statistical office is like an orphan...the minister is not even aware that the statistical office is under his ministry.” — Shanta Devarajan [05:55]

• Poor funding/capacity means poverty or household surveys only once every five or ten years: “countries that are too poor to figure out how poor they are.”

---

6. Royal Pregnancy & Twins: Misuse of Statistical Claims

[06:26–08:23]

Media Hype & Probabilities

• In light of the Duchess of Cambridge’s severe morning sickness, claims spread about her likelihood of having twins. Tim Harford calls on Yan Wong (evolutionary biologist) to analyze the actual chances.

What the Data Shows

“The number of people who have twins is very small. It's about 1.5% in the UK. The number of people who suffer from morning sickness...is also, fortunately, very, very low. It's about 1%.” — Yan Wong [07:11]

• Large datasets (Norway, Denmark, Sweden) show that morning sickness does increase the chance of twins, but only marginally:
  • 50–100% increase in relative risk — i.e., from 1.5% to about 2.5%.

“So it’s still pretty low, actually, the probability of having twins.” — Yan Wong [08:10]

Takeaway: Don’t Misread Relative Risk

• Tim Harford warns against being misled by relative risk headlines:

“Double what chance?... Without knowing the underlying probabilities, we’re not really learning a lot.” — Tim Harford [08:23]

---

Memorable Quotes & Key Moments

• “[Ghana’s GDP grew by 60% overnight...but it didn’t reflect any overnight change in the economy itself...]” — Tim Harford [00:34]
• “The reason why we don’t have more frequent poverty statistics is that the statistical office is very weak and underfunded...” — Shanta Devarajan [05:55]
• “Don’t rush to the bookies just yet.” — Tim Harford [08:17]

---

Important Timestamps

• 00:13–03:04: Ghana/Nigeria GDP revisions explained
• 03:14: Casely-Hayford on Ghana’s informal economy
• 03:55: Jerven & Moss debate why the stats matter
• 04:54: Three groups responsible for reliable data
• 05:55: Devarajan on neglected statistical offices
• 06:26: Transition to royal pregnancy statistic discussion
• 07:11: Yan Wong on twin/morning sickness stats
• 08:23: The problem with relative probability in the media

---

Summary Takeaways

• Massive jumps in African GDP are sometimes statistical mirages caused by out-of-date base years — not genuine overnight growth.
• Economist guests urge caution in trusting such numbers, highlight critical resource and incentive problems in African statistical agencies, and warn that data’s political and international uses make precision important.
• On royal twin speculation: media claims often misuse ‘relative risk,’ vastly overstating small probabilities.
• Always check the underlying numbers behind impressive-sounding statistics.

---

Transcript

[A]

Thank you for downloading from the BBC. For details of our complete range of podcasts and our terms of use, go to bbcworldservice.com podcasts

[B]

hello and welcome to More or Less on the BBC World Service with me, Tim Harford. We're a numerical guide to Life, Liberty and the Chants of Royal Twins. But first, in late November, official figures confirmed that the UK economy grew by 1% between July and September. That was a cheery bit of news, but it's not a patch on Ghana, whose gross domestic product, or GDP, grew by 60% overnight back in 2010. The country was immediately upgraded from low income to lower middle income. Of course, the growth didn't reflect any overnight change in the economy itself, just the fact that the economy had been growing for years and the Ghanaian statistics hadn't kept up. Now Nigeria is planning a similar statistical revision and again, its measured GDP could increase dramatically. So what's going on? I spoke to Morton Jervan, an assistant professor at Simon Fraser University in Canada and the author of Poor How We Are Misled by African Development Statistics and what We Do About It. He explained that it all had to do with the way GDP was calculated with reference to a base year, and the calculations had simply got out of date.

[C]

The GDP statistics in Ghana had not been. In fact, the last time they made a base year estimate for the Ghanaian economy was in the 90s.

[B]

So if we imagine, for instance, that in the 1990s Ghana just makes hot dogs and hot dog buns, but by 2011 it's also making iPhones. I know it doesn't really make iPhones or hot dogs, but just imagine what is the statistical process by which these GDP numbers get out of whack.

[C]

The best way of explaining it is to perhaps to compare it with how GDP statistics is compiled in the uk, where you would think you add up all, as you said, production of hot dogs and iPhones and all other economic activities in every year and then the next year you do the same thing and then you compare how is the economy doing in a country like Ghana? You don't have availability of data on all production of all types of food crops, small scale services and so forth. So therefore you make a base year estimation where you have more complete picture of the economy. And then in the following years when growth is estimating you rely on certain proxies. You assume that large part of the economy just grows in line with population. And finally, what happens if the structure of the economy changes radically? That means that a lot of growth may be missed. There is large technological change between 1993 and 2006, the mobile phone which did not exist in the 90s. And now we have millions and millions of mobile phone users in Ghana so that these kind of sectoral changes were missed as time passed.

[B]

So Ghana's GDP figures weren't right in the past, but are they right now? Sidney Caseleigh Hayford is a business and financial analyst from Ghana and he doesn't think so.

[D]

Until we are able to actually go in and do a proper quantification of the informal economy in this country, it is uncertain exactly what degree of variation we have in our GDP figures. I would hazard a guess and say that we probably out by about plus 10, 15, maybe 20% yet that we should actually include in the number.

[B]

Sydney Caseleigh Hayford but how much does it matter if GDP estimates aren't always entirely accurate? The truth is that Ghana is getting richer and its statistics are getting better. So what's to dislike? But Morden Gervin is worried.

[C]

Reliable economic statistics are basic to the operation of governments also in developing countries. These kind of statistics are also vital to international organizations, to non governmental organizations that for instance provide aid to Ghana. So that for instance, now that Ghana is a middle income country, it is, according to that statistic, not eligible for concessional lending from the World bank, for instance.

[B]

Todd Moss is a senior fellow at the center for Global development in Washington D.C. and he thinks that the statistics being out isn't such a bad problem.

[E]

I think it makes cross country comparisons even less valuable than we thought. But I still think if you're measuring more or less the same thing year to year, you're still getting a picture into growth or shrinkage of parts of the economy and that's what you're really trying to do.

[B]

So maybe it's a big problem and maybe it isn't. But practically speaking, what should be done? Here's Morten There is at least three

[C]

groups that need to really rethink what they're doing. One of them are data users. You need to question your evidence and try to do as historians do, check your sources. Now when it comes to data data disseminators, which is what the World bank is and other United nations and Eurostat and so forth, they need to really start labeling their product correctly. What happens if you download the statistics now from the World bank today? Well, you will find that the estimate on Ghana is updated, but for instance, estimate for Nigeria is still done according to a base year from 1990 and it's hugely misleading. Finally, to sum up with the third group, data producers, in the end, they are the ones that need to take responsibility for the data they put out there.

[B]

And this question of responsibility is central. Shanta Devarajan is the World Bank's chief economist for Africa. And he thinks that too often African economic statistics just fall between the cracks.

[F]

You know, in many countries, the Statistical office is like an orphan. I mean, I've encountered cases where the minister is not even aware that the statistical office is under his ministry. And that's the real problem. The reason why we don't have more frequent poverty statistics is that the statistical office is very weak and they're underfunded and don't have the capacity to run poverty surveys or household surveys more than once every five years or 10 years. I think we're involved in something like 30 African countries in trying to build statistical capacity.

[B]

Shanta Devarajan of the World bank talking about the problem of countries that are too poor to figure out how poor they are. Now, in light of the royal pregnancy, there's been a lot of speculation. The Duchess of Cambridge is suffering from very severe morning sickness. And we are told by opportunistic bookmakers and a giddy media that means twins. This seems incredibly premature so early in a pregnancy. But the way the numbers have been used or abused by makes a broader statistical point. We thought that the only way to cover the experience of pregnancy and morning sickness sensibly is to have two men discuss the issue. So I'm joined by somebody who isn't likely to suffer from either condition. Evolutionary biologist and television presenter Yan Wong.

[G]

It certainly is true that there is this association. The problem is that we are talking very small numbers of people. The number of people who have twins is very small. It's about 1.5% in the UK. And the number of people who suffer from morning sickness, acute enough to require hospital admissions, is also fortunately, very, very low. It's about 1%. And so to try and spot whether or not there's a correlation here, you need huge data sets.

[B]

Do we have huge data sets?

[G]

And we do.

[B]

Hooray.

[G]

There are data sets from Norway of almost a million births between 1967 and 2005. There's ones from Denmark. There are also ones from Sweden.

[D]

Very good.

[B]

What do we learn? Should we believe the hype? It's twins. It's going to be twins.

[G]

It looks like our best guess is somewhere between maybe 50% and 100% increase in the probability of having twins, but

[B]

that's in the relative risk. So we're going up from about 1.5%

[G]

to about 2.5 to about 2.5%, something like that. So it's still pretty low, actually, the probability of having twins.

[B]

Very good. So don't rush to the bookies just yet. Yanwan.

[G]

No, indeed.

[B]

Thank you very much.

[G]

Thank you.

[B]

Yan Wong and this raises a very important point. We're often told frightening sounding stuff, that drinking alcohol will double our chance of mouth cancer or watching television will treble the chance of going blind or whatever. This sort of thing just isn't very informative. Double what chance? Treble what chance? Without knowing the underlying probabilities, we're not really learning a lot. You are not very likely to go blind while watching television and the Duchess of Cambridge is not very likely to have twins. Well, that's enough speculation for this week. Do please send your comments and your questions to more or lessbc.co.uk and you can download more editions of this program at our website, bbcworldservice.com moreorless goodbye.

[A]

There are dozens of different podcasts now available from the BBC, including news, documentaries, science, business, arts and sport. The details of them all go to bbcworldservice.com podcasts.