Robert Gordon is the Stanley G Harris Professor of Economics at Northwestern University. His 2012 paper "Is US economic growth over? Faltering Innovation Confronts the Six Headwinds," sparked intense debate over the trajectory of growth in both the US and across western developed economies. In February, Gordon published a paper updating his views.
Jay Elwes: When did you first become concerned about future U.S. economic growth?
Robert Gordon: For 15 years I have been talking about how the recent inventions don’t measure up to the great old inventions of the late 19C.
JE: Why don’t they measure up?
RG: Life was utterly transformed between 1890 and 1930, and more so in the US than in the UK. The entire urban US was electrified. The house became networked from its isolation in 1870. By 1929 it was connected five ways through electricity, gas, telephone, running water and sewers, none of which existed in the 1870s. Those were extremely important transformations. The invention of the motor vehicle eliminated the horse and its droppings from urban streets. The entire panoply of communication and entertainment inventions that went from the telegraph to the telephone to the phonograph to the radio and then in right at the dawn of WW2, television, all of that happened in a very short time.
Something can’t be more than 100 percent of itself. You could only have the transition from a rural to an urban society once. You could only have the transition from 20 per cent infant mortality to near zero once. All of that was happening in those 50 years. And then in the early part of the post war years we completed the subsidiary inventions from the late 19C, with commercial air transport, air conditioning and the inter-state highway system.
So the difference is that we had, if you count them, at least five if not seven dimensions of an utter change in the standard of living. And it shows up in the productivity statistics as well. If you compare the last 40 years after 1972 and the 80 years before 1972, our productivity growth has been about 2/3 after what it was before. That’s an easy measure of the difference in the power between the second industrial revolution of the late 19th century and the third revolution of the past five decades which has centered on computers, communication, and digitalization. The third revolution has perhaps two dimensions, the computer part and the communications part that includes movement towards mobile devices and smart phones.
And if you think about the productivity effects of the computer revolution, they started way back in the 1960s, with the first computer-produced telephone bills and bank statements and went on in the 1970s with airline reservation systems. In the early 80s there was the invention of the personal computer, the ATM cash machine and barcode scanning which greatly increased productivity in retail. And so much of the impact of computers in replacing human labour had already occurred at the time the internet was introduced in the late 1990s. And actually, depending on which part of the internet you are looking at, it was introduced before then. Most of us were doing email by the early 90s. Amazon was founded in 1994, so we’re 20 years now into the age of e-commerce.
A section in my new 2014 paper is called “Stasis in the Office”. Going from about 1970, which is the date that the electronic calculator was invented, up to about 1995, you had a complete replacement of typewriters and calculators by desktop computers that could do word processing and spread sheets. Also you had the simultaneous replacement in every library in the world when they replaced their wooden boxes of book catalogues with electronic catalogues. The digitalisation of information in the academic field for instance. JSTOR for example allows us to throw out our collections of paper journals and look up anything that’s ever been published on our computer right next to us.
All that was basically all over by ten years ago. And the only thing that has happened in the last 10 years are the smart phone and tablet. So the day-to-day impact on people who are working in the business world, whose productivity we are measuring, have experienced the great transformation that is now losing steam. In the New York Times last week there was an article about the big worldwide convention of mobile telephone companies in Barcelona in which the theme of the article was everybody’s sitting round waiting for the next invention which is not happening.
This is a very different story than you are going to get from Brynjolfsson and McAfee’s book. While I can express scepticism of the future developments that they rave about, I don’t really have to in the sense that my forecast for the future is not zero innovation or zero productivity growth. Rather it’s productivity growth that’s similar to that of the last 40 years which is really optimistic I think, as it assumes we will continue to get fruitful inventions out of digitalisation comparable to what have already occurred. Much of what they are talking about with robots is happening in manufacturing and in wholesale where robots roam through the Amazon warehouse and find the books. But we’ve had more than 50 years of robots replacing people. The first industrial robot was introduced by General Motors in 1961. And indeed manufacturing in the US has continued to enjoy rapid productivity growth with no signs of a slowdown, unlike the rest of the economy where productivity growth is clearly slowing down.
JE: And on the point of the robots and the automation of industrial processes, how will new types of technology effect employment levels?
RG: We’ve had a transition in the last 50 years from manufacturing having 30 per cent of the employment in the economy down to about 8 per cent, so robots together with competition from imports and off-shoring—moving plants into other countries—all have together worked to enormously reduce the number of manufacturing jobs in the US, down maybe 8m over the last 15 years. And I expect that will continue.
While robots will continue to replace human jobs in manufacturing, there are ever-fewer jobs to be replaced. I have described this whole process as “robots and manufacturing present a beautiful ballet performed on a shrinking stage.”
So we have to ask what’s happening in the other 90 per cent of the employees. One of my hobbies is to look at every human being that I encounter in daily life and ask “will this person’s job be replaced by a robot in the foreseeable future?” So far, I cannot find a single person whose job is going to be replaced by a robot. Among those whose jobs are relatively secure from robotization are those I encountered last Friday. There were the doctors and staff at the veterinary office, the doctors, nurses, and receptionists at the dermatology clinic, and the cooks and wait staff in restaurants. There were the check-out clerks and employees restocking the shelves by hand at the local supermarket, the barber, my wife’s manicure/pedicure salon, and the professors and growing administrative staff at the large research university where I work.
With Erik and Andy, I disagree with them on a lot of things, but central to this is the percentage of total employment that potentially over a horizon of two or three decades can be expected to be replaced by robots. I think the fraction of jobs that are replaceable is much smaller than Brynjolfsson and McAfee do. Sometimes I hear them say “67 per cent,” I’d say it’s more like “15 per cent.”
JE: And emerging economies? Do they have potential to grow from new technology?
RG: Of course. Taking China in particular—it took the US from 1875, when these big inventions started, to say 2000 to achieve a twelve-fold increase in the level of productivity. (And by the way, an interesting way to package my comment on the two industrial revolutions, of that twelve-fold increase in the level of productivity, nine of the 12 multiples occurred before 1972. Only the last three have occurred since 1972.)
Well the Chinese have on their plate everything that we have invented over 120 years. And they can adopt all of these things that we’ve invented as fast as they’re willing to set aside saving in order to build all of the equipment and machinery and buildings that are necessary to implement our standard of living. Well they can’t do it entirely because they don’t have enough land. I am sure you have seen pictures of what residential areas in Chinese cities look lie. They look like the infamous Chicago public housing that was torn down within the past two decades -- mile after mile of 25 storey grey buildings hidden behind pollution. I am not concerned about the Chinese catching up to European or US standards of living. But that’s an entirely different topic. I am talking about what is happening at the frontier of developed countries and how fast that frontier is going to advance. The study of emerging markets raises a different set of questions, such as why India is having such a hard time joining the 21st Century as China has done so rapidly.
JE: You said in your most recent paper that “the common assumption that future innovation is non-forecastable is wrong.” Why did you say that?
RG: I got the idea from a book that I cite that has influenced me. It is by Jan Vijg, the geneticist, who writes very convincingly about technological slowdown and the advance of medical science and pharmaceuticals. And his book is more general than just about medicine and drugs, it’s about the slowing rate of transformation at a fundamental level in the 21st C compared to the 20th. In fact his title is The American Technological Challenge: Stagnation and Decline in the 21st Century. From him I got the quote from Jules Verne, who wrote in 1863 about what Paris would be like 100 years from then, and Verne captured many of those changes. And then I have this wonderful page from the Ladies’ Home Journal written in 1900 about what the world would be like 100 years from then, and they really hit the nail on the head in a lot of different categories. They have about 20 short paragraphs, each making a specific prediction.
JE: Can’t these just be put down to luck, these guesses?
RG: No no no. Take two of the Home Journal forecasts. One of them was that we would be able to pipe warm or cold air into our rooms just as we now turn on hot and cold running water. The warm air part was already in place. Central heating had already been invented and it was merely a matter of building new houses and transforming old houses, which costs money and takes time. So that didn’t require a leap of the imagination. Air conditioners, I don’t know when the first rudiments of electro refrigeration began. Home refrigerators began to be sold around 1920 and the first air conditioned movie theatre is dated to 1923. So to write in 1900 that these things were possible was not luck. It was working out the implications of what was already there.
So Eric and Andy and I are starting from the same set of information and trying to work out implications of what’s going to happen and we differ in the way we do that because, to go back to the big difference I mention before—they see much more of the labour force as being open to replacement by robots than I do.
JE: A riposte to your thesis of the phases of industrial and technological development might be that saying now that digital technology has nothing more to add to productivity is a bit like someone saying in 1900 that electricity that has been round since the 1870s has given us all it has to give. After all, the digital economy seems fairly robust. Whatsapp and so on. Does this not suggest that demand for digital goods is large and will continue to grow and create jobs?
RG: I would disagree that somebody at the turn of the century would have said any such thing about electricity. The fact is that in 1900 only 9 per cent of US homes were wired. It would certainly be unreasonable to suggest that the impact of electricity would stop before 100 per cent are wired and the potential for electricity in making the vertical city through elevators was already right there in front of you. The famous Woolworth building in downtown Manhattan was completed in 1911. So anybody who said we are finished with the benefits of electricity in 1900 was insane.
Nor do I say that we are finished with the benefits of digitisation. We will have productivity growth of the rate of the last 40 years and we will have inventions coming along that will use computers in new and different ways at roughly the same rate. Deep down I don’t believe that, but I don’t have to forecast any future innovation slowdown to obtain my gloomy forecast that the disposable income of the bottom 99 percent of the U. S. population will grow at only 0.2 percent over the next 25 to 40 years.
Yes we will continue to get the benefits from ever cheaper and smaller robots and the ability to manipulate huge amounts of data. We may have medical breakthroughs, although they are occurring very slowly. One danger is that we prolong physical life but don’t conquer Alzheimer’s, so that a lot of people who are able to move around don’t know where they’re going. That’s I think one of the themes of Vijg’s chapter on the future of medical care.
So where we differ is their use repeatedly of the term “point of inflection” as if technological change is speeding up. I see no evidence of that. For instance the use of electronic medical records in medicine is gradually becoming widespread. The particular medical group that we have here on the North Shore of Chicago was one of the pioneers and was completely electronic in its medical records by 2003, more than a decade ago. So one can say OK, the rest of the nation and the world is behind that. They must catch up and we will get productivity gains over the next decades as they continue to catch up with something that has already been implemented for some but not medical organizations. That’s a safe forecast and a bit like saying that in 1900 only 9 per cent of families have electricity. Eventually they all will.
The same way, eventually they will all have electronic medical records everywhere. Eventually the systems will talk to each other. So if I came down with some illness in London they would be able to communicate with computers back here. Surely that would happen within the next couple of decades. So I think a lot of these things are fairly easy to forecast.
I have just been writing about entrepreneurship in my book and this idea that there are certain basic breakthrough inventions that are then subject to a long process of incremental improvement—an idea that goes back to Schumpeter in the 1930s—we are going to continue to have these incremental improvements to the electronic computer, a great invention of the 1950s and 1960s, and its ability to manipulate information.
JE: Do technological advances tend to encourage inequality?
RG: You have to ask why now and not back between 1950 and 1975, when incomes at the bottom and at the top were growing at the same rate. So we were able to invent a lot of things without creating inequality, in fact inequality dropped markedly form the 1920s to the 1950s.
The difference now I think is the particular nature of electronic innovation that rewards education and a certain kind of skill and also some of the inequality is directly created by technology. In the example of the very high pay of sports stars, which is of course in the US we have all these major league sports, with many teams. A football quarterback or a baseball pitcher routinely earns $10m a year or more. And this is indirectly the result of cable television making possible vast increases in the revenue that is then divided up between the owners of the sports and their players.
This had already started in the 1970s with things like phonograph records and motion pictures, but the nature of innovation in the last 30 years has created a sort of “winner takes all” society. A famous economist described this a long time ago as the “Economics of superstars”. Part of the difference today is political. Where the US differs somewhat particularly from Scandinavia and from Northern Europe, is in the demise of the labour unions and the ability of large corporations to impose new and much less rewarding wage conditions, cutting benefits, reducing pensions, converting pensions from defined benefit to defined contribution. Another sort of unique US problem is forcing workers into part time work because this allows employers to avoid paying medical care costs, whereas in the rest of the world you have medical care covered by taxation which has no effect on employment decisions or arrangements.
So the causes of increased inequality combine skills that are particularly rewarded as a result of new inventions including cable television together with a sort of political increase in the power of management and decline in the power of labour. And also in the US more than in other countries we have a basic problem of high school drop outs because we have quite large percentages of our students are black and Hispanic and there is a continuing gap in test scores with whites and Asians. As the percentage of the population that is in those minority groups, particularly the Hispanics rises, then that drags down the overall average educational attainment and the wages that people can expect to earn.
Jay Elwes: When did you first become concerned about future U.S. economic growth?
Robert Gordon: For 15 years I have been talking about how the recent inventions don’t measure up to the great old inventions of the late 19C.
JE: Why don’t they measure up?
RG: Life was utterly transformed between 1890 and 1930, and more so in the US than in the UK. The entire urban US was electrified. The house became networked from its isolation in 1870. By 1929 it was connected five ways through electricity, gas, telephone, running water and sewers, none of which existed in the 1870s. Those were extremely important transformations. The invention of the motor vehicle eliminated the horse and its droppings from urban streets. The entire panoply of communication and entertainment inventions that went from the telegraph to the telephone to the phonograph to the radio and then in right at the dawn of WW2, television, all of that happened in a very short time.
Something can’t be more than 100 percent of itself. You could only have the transition from a rural to an urban society once. You could only have the transition from 20 per cent infant mortality to near zero once. All of that was happening in those 50 years. And then in the early part of the post war years we completed the subsidiary inventions from the late 19C, with commercial air transport, air conditioning and the inter-state highway system.
So the difference is that we had, if you count them, at least five if not seven dimensions of an utter change in the standard of living. And it shows up in the productivity statistics as well. If you compare the last 40 years after 1972 and the 80 years before 1972, our productivity growth has been about 2/3 after what it was before. That’s an easy measure of the difference in the power between the second industrial revolution of the late 19th century and the third revolution of the past five decades which has centered on computers, communication, and digitalization. The third revolution has perhaps two dimensions, the computer part and the communications part that includes movement towards mobile devices and smart phones.
And if you think about the productivity effects of the computer revolution, they started way back in the 1960s, with the first computer-produced telephone bills and bank statements and went on in the 1970s with airline reservation systems. In the early 80s there was the invention of the personal computer, the ATM cash machine and barcode scanning which greatly increased productivity in retail. And so much of the impact of computers in replacing human labour had already occurred at the time the internet was introduced in the late 1990s. And actually, depending on which part of the internet you are looking at, it was introduced before then. Most of us were doing email by the early 90s. Amazon was founded in 1994, so we’re 20 years now into the age of e-commerce.
A section in my new 2014 paper is called “Stasis in the Office”. Going from about 1970, which is the date that the electronic calculator was invented, up to about 1995, you had a complete replacement of typewriters and calculators by desktop computers that could do word processing and spread sheets. Also you had the simultaneous replacement in every library in the world when they replaced their wooden boxes of book catalogues with electronic catalogues. The digitalisation of information in the academic field for instance. JSTOR for example allows us to throw out our collections of paper journals and look up anything that’s ever been published on our computer right next to us.
All that was basically all over by ten years ago. And the only thing that has happened in the last 10 years are the smart phone and tablet. So the day-to-day impact on people who are working in the business world, whose productivity we are measuring, have experienced the great transformation that is now losing steam. In the New York Times last week there was an article about the big worldwide convention of mobile telephone companies in Barcelona in which the theme of the article was everybody’s sitting round waiting for the next invention which is not happening.
This is a very different story than you are going to get from Brynjolfsson and McAfee’s book. While I can express scepticism of the future developments that they rave about, I don’t really have to in the sense that my forecast for the future is not zero innovation or zero productivity growth. Rather it’s productivity growth that’s similar to that of the last 40 years which is really optimistic I think, as it assumes we will continue to get fruitful inventions out of digitalisation comparable to what have already occurred. Much of what they are talking about with robots is happening in manufacturing and in wholesale where robots roam through the Amazon warehouse and find the books. But we’ve had more than 50 years of robots replacing people. The first industrial robot was introduced by General Motors in 1961. And indeed manufacturing in the US has continued to enjoy rapid productivity growth with no signs of a slowdown, unlike the rest of the economy where productivity growth is clearly slowing down.
JE: And on the point of the robots and the automation of industrial processes, how will new types of technology effect employment levels?
RG: We’ve had a transition in the last 50 years from manufacturing having 30 per cent of the employment in the economy down to about 8 per cent, so robots together with competition from imports and off-shoring—moving plants into other countries—all have together worked to enormously reduce the number of manufacturing jobs in the US, down maybe 8m over the last 15 years. And I expect that will continue.
While robots will continue to replace human jobs in manufacturing, there are ever-fewer jobs to be replaced. I have described this whole process as “robots and manufacturing present a beautiful ballet performed on a shrinking stage.”
So we have to ask what’s happening in the other 90 per cent of the employees. One of my hobbies is to look at every human being that I encounter in daily life and ask “will this person’s job be replaced by a robot in the foreseeable future?” So far, I cannot find a single person whose job is going to be replaced by a robot. Among those whose jobs are relatively secure from robotization are those I encountered last Friday. There were the doctors and staff at the veterinary office, the doctors, nurses, and receptionists at the dermatology clinic, and the cooks and wait staff in restaurants. There were the check-out clerks and employees restocking the shelves by hand at the local supermarket, the barber, my wife’s manicure/pedicure salon, and the professors and growing administrative staff at the large research university where I work.
With Erik and Andy, I disagree with them on a lot of things, but central to this is the percentage of total employment that potentially over a horizon of two or three decades can be expected to be replaced by robots. I think the fraction of jobs that are replaceable is much smaller than Brynjolfsson and McAfee do. Sometimes I hear them say “67 per cent,” I’d say it’s more like “15 per cent.”
JE: And emerging economies? Do they have potential to grow from new technology?
RG: Of course. Taking China in particular—it took the US from 1875, when these big inventions started, to say 2000 to achieve a twelve-fold increase in the level of productivity. (And by the way, an interesting way to package my comment on the two industrial revolutions, of that twelve-fold increase in the level of productivity, nine of the 12 multiples occurred before 1972. Only the last three have occurred since 1972.)
Well the Chinese have on their plate everything that we have invented over 120 years. And they can adopt all of these things that we’ve invented as fast as they’re willing to set aside saving in order to build all of the equipment and machinery and buildings that are necessary to implement our standard of living. Well they can’t do it entirely because they don’t have enough land. I am sure you have seen pictures of what residential areas in Chinese cities look lie. They look like the infamous Chicago public housing that was torn down within the past two decades -- mile after mile of 25 storey grey buildings hidden behind pollution. I am not concerned about the Chinese catching up to European or US standards of living. But that’s an entirely different topic. I am talking about what is happening at the frontier of developed countries and how fast that frontier is going to advance. The study of emerging markets raises a different set of questions, such as why India is having such a hard time joining the 21st Century as China has done so rapidly.
JE: You said in your most recent paper that “the common assumption that future innovation is non-forecastable is wrong.” Why did you say that?
RG: I got the idea from a book that I cite that has influenced me. It is by Jan Vijg, the geneticist, who writes very convincingly about technological slowdown and the advance of medical science and pharmaceuticals. And his book is more general than just about medicine and drugs, it’s about the slowing rate of transformation at a fundamental level in the 21st C compared to the 20th. In fact his title is The American Technological Challenge: Stagnation and Decline in the 21st Century. From him I got the quote from Jules Verne, who wrote in 1863 about what Paris would be like 100 years from then, and Verne captured many of those changes. And then I have this wonderful page from the Ladies’ Home Journal written in 1900 about what the world would be like 100 years from then, and they really hit the nail on the head in a lot of different categories. They have about 20 short paragraphs, each making a specific prediction.
JE: Can’t these just be put down to luck, these guesses?
RG: No no no. Take two of the Home Journal forecasts. One of them was that we would be able to pipe warm or cold air into our rooms just as we now turn on hot and cold running water. The warm air part was already in place. Central heating had already been invented and it was merely a matter of building new houses and transforming old houses, which costs money and takes time. So that didn’t require a leap of the imagination. Air conditioners, I don’t know when the first rudiments of electro refrigeration began. Home refrigerators began to be sold around 1920 and the first air conditioned movie theatre is dated to 1923. So to write in 1900 that these things were possible was not luck. It was working out the implications of what was already there.
So Eric and Andy and I are starting from the same set of information and trying to work out implications of what’s going to happen and we differ in the way we do that because, to go back to the big difference I mention before—they see much more of the labour force as being open to replacement by robots than I do.
JE: A riposte to your thesis of the phases of industrial and technological development might be that saying now that digital technology has nothing more to add to productivity is a bit like someone saying in 1900 that electricity that has been round since the 1870s has given us all it has to give. After all, the digital economy seems fairly robust. Whatsapp and so on. Does this not suggest that demand for digital goods is large and will continue to grow and create jobs?
RG: I would disagree that somebody at the turn of the century would have said any such thing about electricity. The fact is that in 1900 only 9 per cent of US homes were wired. It would certainly be unreasonable to suggest that the impact of electricity would stop before 100 per cent are wired and the potential for electricity in making the vertical city through elevators was already right there in front of you. The famous Woolworth building in downtown Manhattan was completed in 1911. So anybody who said we are finished with the benefits of electricity in 1900 was insane.
Nor do I say that we are finished with the benefits of digitisation. We will have productivity growth of the rate of the last 40 years and we will have inventions coming along that will use computers in new and different ways at roughly the same rate. Deep down I don’t believe that, but I don’t have to forecast any future innovation slowdown to obtain my gloomy forecast that the disposable income of the bottom 99 percent of the U. S. population will grow at only 0.2 percent over the next 25 to 40 years.
Yes we will continue to get the benefits from ever cheaper and smaller robots and the ability to manipulate huge amounts of data. We may have medical breakthroughs, although they are occurring very slowly. One danger is that we prolong physical life but don’t conquer Alzheimer’s, so that a lot of people who are able to move around don’t know where they’re going. That’s I think one of the themes of Vijg’s chapter on the future of medical care.
So where we differ is their use repeatedly of the term “point of inflection” as if technological change is speeding up. I see no evidence of that. For instance the use of electronic medical records in medicine is gradually becoming widespread. The particular medical group that we have here on the North Shore of Chicago was one of the pioneers and was completely electronic in its medical records by 2003, more than a decade ago. So one can say OK, the rest of the nation and the world is behind that. They must catch up and we will get productivity gains over the next decades as they continue to catch up with something that has already been implemented for some but not medical organizations. That’s a safe forecast and a bit like saying that in 1900 only 9 per cent of families have electricity. Eventually they all will.
The same way, eventually they will all have electronic medical records everywhere. Eventually the systems will talk to each other. So if I came down with some illness in London they would be able to communicate with computers back here. Surely that would happen within the next couple of decades. So I think a lot of these things are fairly easy to forecast.
I have just been writing about entrepreneurship in my book and this idea that there are certain basic breakthrough inventions that are then subject to a long process of incremental improvement—an idea that goes back to Schumpeter in the 1930s—we are going to continue to have these incremental improvements to the electronic computer, a great invention of the 1950s and 1960s, and its ability to manipulate information.
JE: Do technological advances tend to encourage inequality?
RG: You have to ask why now and not back between 1950 and 1975, when incomes at the bottom and at the top were growing at the same rate. So we were able to invent a lot of things without creating inequality, in fact inequality dropped markedly form the 1920s to the 1950s.
The difference now I think is the particular nature of electronic innovation that rewards education and a certain kind of skill and also some of the inequality is directly created by technology. In the example of the very high pay of sports stars, which is of course in the US we have all these major league sports, with many teams. A football quarterback or a baseball pitcher routinely earns $10m a year or more. And this is indirectly the result of cable television making possible vast increases in the revenue that is then divided up between the owners of the sports and their players.
This had already started in the 1970s with things like phonograph records and motion pictures, but the nature of innovation in the last 30 years has created a sort of “winner takes all” society. A famous economist described this a long time ago as the “Economics of superstars”. Part of the difference today is political. Where the US differs somewhat particularly from Scandinavia and from Northern Europe, is in the demise of the labour unions and the ability of large corporations to impose new and much less rewarding wage conditions, cutting benefits, reducing pensions, converting pensions from defined benefit to defined contribution. Another sort of unique US problem is forcing workers into part time work because this allows employers to avoid paying medical care costs, whereas in the rest of the world you have medical care covered by taxation which has no effect on employment decisions or arrangements.
So the causes of increased inequality combine skills that are particularly rewarded as a result of new inventions including cable television together with a sort of political increase in the power of management and decline in the power of labour. And also in the US more than in other countries we have a basic problem of high school drop outs because we have quite large percentages of our students are black and Hispanic and there is a continuing gap in test scores with whites and Asians. As the percentage of the population that is in those minority groups, particularly the Hispanics rises, then that drags down the overall average educational attainment and the wages that people can expect to earn.
