Summers spoke at The World Bank on March 30, 2016 at a seminar on the Future of Price Statistics: Innovation in Data Technology and Methods calling data the “ultimate public good.” Summers said, “The one thing I can promise you is that vastly better data collection will produce benefits that we can foresee, but even greater benefits that no one can imagine.The World Bank, Washington, D.C.
March 30, 2016
Sri [Indrawati], thank you very much for that very generous introduction. It reminded me of the old line that Lyndon Johnson used to use—’I wish my parents had been here for that. My father would’ve appreciated it, and my mother would have believed it.’
It is good to come to a nerdy place like the World Bank because you don’t get introduced with the economist jokes that I am too often introduced with. It wasn’t so long that a guy said, “‘Larry, do you know what it takes to succeed as an economist?’ I said no. He said, ‘An economist is someone who’s pretty good with figures but does not quite have the personality to be an accountant.’” That was in Moscow and no one got the joke.
I am really glad to be here. I am going to dedicate my remarks today to my late father who devoted much of his career to the development of price statistics through the International Comparison Project when I was a boy.
That was a project that was taking a one-off snapshot laboriously of ten countries. Over time, 10 countries became 16 countries, 16 countries became 40 countries. The snapshot became the continuous Penn World Tables. And ultimately the work was institutionalized at the World Bank, and today is made use of in the world’s economic statistics. That would not have happened without the World Bank’s recognition of the importance of data.
So everyone should understand that the World Bank has been making a major contribution to the global statistical effort for decades. Sri, I thank you, I thank Jim Kim, I thank the previous presidents of the World Bank for their dedication to what I regard as a profoundly important effort.
I want to assert four propositions this morning for your consideration. First, the standard history of science as you have learned it is mostly wrong. The proposition that you all learned in some version of your high school classes is what Freeman Dyson has called the Kuhn view of the history of science.
The idea is that there is a theory and then somebody thinks hard about the theory, and realizes predictions that the theory makes, and then tests those predictions. And, sometimes the predictions are confirmed but eventually some prediction is found and the experiments do not confirm the prediction. Then a new hypothesis is developed. It is tested and it is improved, and ultimately science progresses through hypothesis testing and experimenting.
We all learn that science drives large parts of human progress and hypothesis-testing and experiment drives science. And there are obviously areas in which that is true. When the bending of light was observed in the 1919 eclipse, it was a great victory for Einstein’s general theory of relativity.
But what recent scholarship has demonstrated, led by my Harvard colleague Peter Galison, is that a very large part of science actually progresses just by looking. Better information and simple observation drive science.
What led the world after 1500 years of believing in Ptolemaic astronomy, with the earth as the center of the universe, to come to understand that planets orbited the sun and moons orbited planets? It was the invention of the telescope and the observation that moons were orbiting Jupiter, which you can see very simply with your own eyes. And then there was not any other argument to be had anymore.
The most important discovery in the history of biology was the cell. How was it made? Antonie van Leeuwenhoek developed the microscope. People looked in the microscope at water and they saw bacteria cells, and then they knew there were cells. I could continue with more examples. If you think about the structure of the atom—which underlies everything about radioactivity, everything about nuclear power—it is a consequence of particle accelerators and simple inference from those particle accelerators.
So what actually drives progress as much as all this philosophical stuff about hypothesis testing, is the ability to observe better. And with the ability to observe better, things which were formerly difficult to see or impossible to understand became self-evident and progress is made.
That drives me to my second observation. If mathematics is, as is often said, the queen of the sciences, statistics is the queen of the social sciences. Let me give you a few examples of the power of statistics to make obvious what was previously almost unimaginable.
If you read a health book giving people advice about healthy habits written before 1946, that book will tell you not under any circumstances over-exercise. It will be explained that if you over-exercise you might put strain on your heart and cause yourself to have a heart attack. It will further be explained that when you exercise you get hungry and you eat more, and so exercise makes you fat.
How did the world learn that that was wrong? Not, I promise you, because some cardiologist had a theory about strengthening the heart muscles. Not, I promise you, because someone had a theory which they tested about exercise promoting flow of blood to the brain. No. As part of a broad statistical understanding at improving health in the United Kingdom, they did a broad survey of many workers, and particularly public employees, and asked them all kinds of questions about their lives and their subsequent health status.
And they observed—because it leapt out of them as a huge fact—that bus conductors on double decker buses had two-thirds the heart attack rate that bus drivers did. And if you think about that observation for only 30 seconds, there can only be one explanation. Bus conductors were, if anything, of lower socio-economic status than bus drivers; it’s just that on double decker buses, bus conductors were climbing up and down the stairs all day and they had a much lower heart attack rate. And we went from having no idea to knowing the answer just because there was a lot of data around and somebody looked at that shrewdly.
I’ll give you another example from the health field. It was—you could argue about this—somewhere between 15 and 40 years from the time when statisticians noticed that people who smoked more, died more until the time that anyone provided a theory of what exactly it was in nicotine or tar that biologically did damage to people’s lungs. Again, data and staring told us hugely important things for human welfare that we did not previously know.
My last example is two orders of magnitude more trivial than the two I have just mentioned but I think illustrates the pervasive application of the point. If one had had to imagine a sphere in which analytics, statistics and statistical research would be useless, I think it would be hard to imagine a sphere more plausible for that conclusion than baseball. And yet I would submit for your consideration that Moneyball is a test for our times. What it showed was that all kinds of things that baseball “knew” from experience were not true. For example, it “knew” that sacrifice bunts were a good idea, which the data show is not generally the case.
This is a very international audience so I realized not everyone might get this reference. If I were more adroit than I am, I would be using a soccer analogy but I promise baseball will pass soon.
The data showed that some short guys who are hard to pitch to got a lot of walks. And it was found that a walk is 85% as good as a single. And all this changed who was deemed a good baseball player, and who was not deemed a good baseball player. Statistics can tell you who will play baseball well—better than making people run the 40-yard dash or measuring their reflex time. So think about how many other areas statistics has the potential to transform.
It is ultimately observation and common sense rather than genius that drives scientific progress. And it is ultimately statistics that enable us to look and see phenomena and observe what is happening. That is what makes statistics so central to progress in modern life.
My third observation. Technology is transforming all of this. My father wrote his PhD thesis when I was a little boy. He actually would go to the computer and would go behind the computer and when he wanted to add numbers, he would put one set of wires into their sockets. When he wanted to subtract numbers, he would put a different set of wires into their sockets.
About 15 years after that, the Apollo program sent a man to the moon with orbits that were all calculated on sophisticated computers and telecommunications systems that were based on what were then modern communication technology.
At that time, my family was living in London. And because it was so expensive, only twice during the year did we call my grandmother in New York. My mother rehearsed her children in giving their messages to our grandmother so we could give them quickly, so that the phone call would not last too long. That was 15 years after we were plugging wires in.
[Holds up smartphone.] This device has about 15 times the computing power that the entire Apollo program had, this device. And today, there are two billion of them in human circulation. And it is a prediction that one can make confidently that within a decade there will be more of these than there will be adults on planet earth.
Now, admittedly, there will be some people in Hong Kong who will own four of them. So the fact that there will be more of these than there are adults on planet earth does not quite mean that everyone will have one. But it does mean that a very, very large fraction of all the people who are on planet earth will have these devices.
In fact, they will never in most cases have personal computer. They will never make a phone call that is transmitted over a phone wire. They will leapfrog the technologies that have defined our adult lives.
Sri, when I think about this, I think about what was an enormously meaningful experience for me in 1979, I had my first experience visiting a developing country. At the time, the Harvard Institute for International Development had a major project advising your country. I was a very junior member of a team that advised Mr. Wijoyo and in particular Ali Wardhana, who was one of your predecessors as Finance Minister. I went to Indonesia; it was a hugely meaningful experience in all sorts of ways for me.
But I’ll just comment on an unimportant aspect. I was, at the time, a passionate Boston Red Sox fan. That is baseball, for those of you who don’t know. [Laughs]. A passionate Red Sox fan. And when the Red Sox played on Monday night, I wanted to know what happened. And when did I learn what happened? I would learn on Thursday. Why would I learn on Thursday? Because the game would be completed too late to get into the Tuesday International Herald Tribune. It would get into the Wednesday edition. That paper would get to the door of the Borobudur Hotel near the Finance Ministry at four in the afternoon on Thursday.
And there was no CNN. There was no cable TV. There was no TV in English in my hotel room. It cost one day’s salary to make a four-minute phone call to the United States and I did not want to know about the Red Sox that much. [Laughs.] And so it was Monday until Thursday until I knew what happened to the Red Sox. And yet anyone backpacking in Sumatra today would take it for granted that they could be instantly get that type of information.
And so we live in an entirely different world of technology. And therefore we live in an entirely different world in terms of the capacity to produce data and information than we have ever lived in before. Some of it—the least part of it—is simply we can use these tools to collect data in ways that are far more efficient and far more rapid than we previously envisioned.
That’s why something like the Billion Prices Project at MIT, which can provide daily price information, is so important. That’s why I am excited to be a Director and involved with Premise in its capacity to mobilize these technologies as widely as possible. That’s why Planet Labs, with its capacity to scan and monitor environmental conditions, represents such a profound innovation.
But I think there is actually something even deeper going on, which is that the processes of data collection and the processes of carrying out the phenomena about which the data is being collected are becoming ever more organically related.
An example: one of the more important apps on my iPhone is Waze. What is Waze? Waze is basically a system that scans where I am as I drive around, sees how rapidly I’m moving, makes a judgment about traffic congestion, integrates that judgment about how dense it is where I am with the density where thousands of other people are, and then recommends routes to all of us so that we all get where we are going more rapidly. The more people who use it, the better the data gets. The better the data gets, the better the advice gets. The better the advice gets, the more people who use it, and it is an enormously positive feedback loop.
And increasingly we are seeing that in more and more areas. New technology, new forms of partnership, new relations between the measurement process and the actual process that is being measured, all of this has created a revolution in technology.
My fourth observation. This will change the world vastly for the better because data is the ultimate public good.
I remember a moment when I was fairly young that was an important part of why I decided to become a macroeconomist. I understood there were all these people who argued about how the economy worked. And that when they argued about how the economy worked, it affected the policies that countries pursued. And then those policies affected the economic outcomes. And I realized that if something I was able to do at some point in my life led to the unemployment rate in the United States being one-tenth of one percent lower for one month, that would mean 160,000 fewer people out of work. That would mean several hundred thousand children who did not see their father or mother be discouraged by losing a job. That would mean nearly a billion dollars of income more for families trying to meet their budgets.
I thought about the impact that one-tenth of one percent of unemployment for one month would have on hundreds of thousands of people, and I thought about the impact that a teacher or a doctor or a marketer could have. It seemed to me that just because the causal chain might be longer between thinking about economics and data and outcomes and I would not ever see the particular person I might have helped and certainly no one would say thank you, that it was no less morally important.
And I would suggest to you that the projects in which you are all engaged in better measurement have the potential to be worth far, far, far more than one-tenth of one percent on the unemployment rate for one month.
Consider this. The central planetary goals that have been set by the world’s leaders of every major country are the Sustainable Development Goals. Most people here understand them better than I. There are more of them, there’s not quite one of them for every citizen in the developing world (laughs) but we’re in that direction—there are people who understand them more. But I do not think I paint with too broad a brush when I say that global poverty is central to them. And yet, the most recent accurate data we have for many crucial countries is 2011.
Now, you could say there are long lags. You could say that it is really hard to collect data. You could say that it is really very, very difficult, and very, very complicated, and that the data is from 2011. It is very hard, it is very complicated, it is very long.
Here is what I’m here to tell you: World War II was very long and very complicated and involved vast mobilization. It took three-and-a-half years from the time the United States entered World War II to the time that the United States won World War II. And again, the most recent data we have available on what we have identified as an absolutely central indicator is five years old.
How many ministers, Sri, stay in office as Finance Minister for five years? Not very many. And you know what, I think you will agree with me as a former Finance Minister that as altruistic as you are, as altruistic as I was when I was a Finance Minister, I cared more about what happened when I was Finance Minister than I did when my predecessor was Finance Minister, or that I cared about what happened when my successor was Finance Minister.
So, making data available in real-time is central to having that data be acted on, it is central to having that data galvanize political action, it is central to having that data make a difference. That might once have been true but impossible to act on. It is possible to act on today.
Today, when I am in Jakarta, I can know what happened in the Red Sox game about three nanoseconds after it happens. And there is no reason why we cannot in continuous time monitor what is happening across this planet. That carries with it many, many benefits. It carries with it that we get the information instantaneously. As the adage says ‘what you count, counts.’
If I could give you one more example. The world is doing today—again, this is something people can argue about and I’m painting with a broad brush—the world is doing substantially better with respect to global climate change today than one might have reasonably expected two years ago. Not well enough, but it is doing better.
In large part, that is driven by an enhanced Chinese commitment to addressing these challenges. Of course, that is primarily driven by what is happening, or not happening, in the Chinese economy. But I would submit that the fact that one of the most common app on iPhones in Beijing is AQI—air quality indicators—has a great deal to do with the energy China has devoted to this problem.
Measurement matters. Observation counts. And it drives action. And that’s why data is so important.
Two final observations as part of making the case for data, if I might.
Knowing how much is spent, without knowing what prices are—is meaningless. If you told me that my height was 50, that would be an entirely uninformative statement unless you told me that I was 50 inches, or 50 yards, or 50 meters, or 50 parsec, or 50 something. Telling me my height was 50 would be utterly without meaning.
Likewise, saying someone spent 60 without saying what 60 was, or simply identifying 60 as 60 of certain piece of paper—the rupee—is meaningless as well. No economic data are of any real meaning at all without prices. Just as in the same way that no height data has any meaning at all without units. So when you work on price statistics, you are at one level working on one piece of the broad puzzle of economic data. But you are at another level, opening the key to understanding — using the key — to unlock all of what is going on in the economy.
There are no statistics more fundamental than price statistics.
A final thought: Thomas Watson was a visionary at IBM. He estimated in the 1940s that the global demand for computers would be five to 10. Not five to 10 billion. Not five to 10 million. Not five to 10 hundred thousand. Not five to 10 thousand. Five to 10.
In the late 1970s, AT&T had conceptualized what we today call the cell phone. They could not really see why with pay phones around anybody would have any great desire to carry around a cell phone except for maybe a few super important people. And so they thought the global demand for cell phones would be about 10,000. My favorite cartoon is of a little boy being asked what he wants to do when he grows up. And he answers, ‘a job that has not been invented yet.’
As with Moneyball, as with those British conductors, the one thing I can promise you is that vastly better data collection will produce benefits that we can foresee, but even greater benefits that no one in this room can imagine. That is why data is the ultimate public good, why meetings like this are so important, why the work you are doing is so important. And why, if I might, Madame Vice President, the work of the World Bank in supporting data, in supporting the ultimate public good, is as important for the long run as any other work the World Bank does.
Thank you very much.