Of That

Brandt Redd on Education, Technology, Energy, and Trust

31 July 2017

Why Assessment?

Folio Corporation LogoMost departments or ministries of education state the purposes of assessment. I'm particularly fond of New Zealand's statement:

The primary purpose of assessment is to improve students’ learning and teachers’ teaching as both respond to the information it provides. Assessment for learning is an ongoing process that arises out of the interaction between teaching and learning.

I like this because it captures the feedback process and acknowledges that both students and educators should respond to that feedback. It also encompasses the various goals of assessment — to inform individual student learning, to measure the effectiveness of the learning system, and to serve as evidence of student skills.

Today I'm writing about the purposes of assessment and the value of standardized assessment.

Inform Individual Student Learning

The most important use of assessment is to improve individual learning. When used properly, assessments improve individual learning in three ways. 1) Exercising and demonstrating skills reinforces student understanding and helps retention. 2) Student attention to the assessment results can increase motivation and direct their choice of learning activities. 3) Educator attention to assessment results can direct their assignment of learning activities or inform interventions.

All of these involve educational feedback loops. However these impacts are only achieved if the right assessment is used for the right purpose. For example, many high-stakes assessments were developed primarily to comply with regulations, such as The No Child Left Behind Act (replaced by ESSA). The reports required by these regulations focus on the percentage of students at each institution that achieve standards for grade-level competency. A test focused on that type of report centers on the threshold of competency. It can indicate with great reliability whether a student is above or below the threshold but may not be reliable for other insight. Such a threshold test is a poor choice for informing learning activities, diagnosing areas of weakness, or measuring growth.

More advanced tests include questions at a variety of skill levels centered on the expected competency level. These tests indicate the student's competency on a continuous scale. Accordingly, they can indicate how far ahead or behind the student is, not just whether they are above or below a certain threshold. By comparing scores from successive tests, you can measure student growth over a period of time. Advanced tests include questions designed to get measure greater depths of knowledge. Such tests offer more reliable detail about student skills in specific areas.

One objection to advanced tests is that it takes more questions and more time to measure skills at this level of detail. The use of computer adaptive testing can shorten the test while maintaining reliability.

Measure the Effectiveness of the Learning System

When standardized assessments are used to measure the effectiveness of the learning system, individual student results are aggregated to indicate the fraction that are achieving competency levels. Typically results are compared with previous years to see if schools are improving. This Delaware Press Release is a typical example of the public statements made each year.

Assessments like these are based on the premise that if you measure performance and report on it, then performance will improve. Unfortunately, education has proven to be a stubborn counter example of this premise. Sixteen years after the No Child Left Behind act mandated standardized testing and established remedies for underperforming schools there has been limited progress. This leads some to call for abandonment of standardized assessment altogether. But if we don't measure performance, we will never know if we are succeeding.

These are our contemporary challenges: discovering the factors under that contribute to better learning and investing the resources needed to improve those factors. Continuing to measure performance will support gathering evidence of principles of effective teaching and learning.

Less frequently applied but equal in importance is using assessments to evaluate parts of the learning system. For example, assessment data can be used to compare different curricula or textbooks, for continuous improvement of online learning systems, and to evaluate the effectiveness of professional development programs.

Provide Evidence of Student Achievement

Since 1905, the primary measure of student achievement in the U.S. has been the Carnegie Unit. This measure uses the time a student spends in the classroom as proxy for how much they have learned. A century later, in 2005, New Hampshire began converting to a competency-based system where student skill is measured directly rather than by proxy. Other states have programs that allow competency measures as an alternative to seat time. Such measures depend on high quality assessments that are aligned to specific and relevant standards of achievement.

Standardized and High Quality Assessment

It has become common, in recent years, to complain about achievement standards and the associated standardized assessments. A typical protest might be, "My child is not standardized." To be sure, our goal should not be to achieve sameness among children and this is not the purpose of achievement standards. Rather, we recognize that people need to achieve a basic competency level in language arts and mathematics in order to function and achieve in our society. The standards are intended to reflect that basic competency with the hope that students and educators will build a wide variety of skill and achievement on that core foundation.

All of these uses — informing learning activities, measuring program effectiveness, and providing evidence of achievement — depend on the quality of the assessment. An assessment will provide poor guidance if it is sensitive to the wrong factors, is unreliable, or is tuned to the wrong skill level. I've written before that personalized learning is currently the most promising approach to improving learning. Choosing high quality assessments to inform personalization is essential to the success of such programs.

That should be our demand — that states, districts, and schools give us evidence of the quality of the assessments they use.

30 June 2017

Reducing Income Inequality when Productivity Exceeds Consumption

Prototype apple harvester by Abundant Robotics.Among the last bastions of labor demand is Agricultural Harvesting. Every fall, groups of migrant workers follow the maturation of fruit and vegetables across the different climates. But even those jobs are going away. In one case, Abundant Robotics is developing robots that use image recognition to locate ripe apples and delicate manipulators to harvest fruit without bruising it.

In my last post I described how creatively destructive innovations like these have increased productivity in the United States until it exceeds basic needs by nearly four times. If it weren't for distribution problems, we could eliminate poverty altogether.

The problem is that when production exceeds consumption, prices fall and the economy goes into recession. As I wrote previously the U.S. and most of the word economy have relied on a combination of advertising, increased standards of living, and fiscal policy to stimulate demand sufficiently to keep up with productivity. But these stimulus methods have a side effect of increasing wage disparity.

The impact is mostly on the unskilled labor force as these jobs are the easiest to automate or export. Even though the income pie keeps growing, the slice owned by the lowest quintile shrinks. Free trade exacerbates the spread between skilled and unskilled labor as do attempts to stimulate consumption through government spending, low interest rates, and tax cuts. (Please see that previous post for details on all of this.)

Conventional Solutions and Their Limits

This is a gnarly problem. Potential remedies tend to have side effects that can make the situation worse, or at least limit the positive impact. In this post I'll name some of the conventional solutions and then attempt to frame up the kind of innovation that will be required to properly solve the problem.

Infrastructure Investment

Investments in infrastructure are a good short-term stimulus. Constructing or upgrading roads, bridges, energy, and communications infrastructure employs a lot of people of all skill levels. As most infrastructure is government-funded, it's a convenient fiscal stimulus. Trouble is that in the long run these infrastructure improvements result in greater overall productivity thereby reducing the labor demand.

Progressive Taxes

A common method of wealth redistribution is a progressive tax. Presumably the wealthy can afford to pay a larger fraction of their income than those with lower incomes. Progressive taxes are effective tools but, in the U.S., we have pretty much reached the limit of how much a progressive system can help low income households. Those earning less than the income tax threshold already pay no taxes. For a single parent with one child, the marginal tax threshold for 2016 was approximately $21,000 before accounting for federal and state health care benefits which amount to approximately $9,400 more.

Since the lowest tax bracket is already zero we can consider increasing the tax rate in upper tax brackets. While that may increase the amount paid by the wealthy it doesn't directly benefit the poor. Indeed, the resulting economic slowdown may worsen the situation.

Refundable Tax Credits

Through tax credits you can reduce the lowest effective income tax rate to negative values — paying money rather than collecting it. In the U.S. the Earned Income Tax Credit (ETC) serves this role with progressive benefits for the number of children in the household. Because the ETC is designed as a percentage of income, the benefit increases as the individual earns more money before being phased out at higher income levels. This is intended to incentivize workers to find and improve their employment even while drawing government benefits.

The downside of tax credits is that they are tied in with the very complicated process of filing income tax returns. The Government Accounting Office and IRS indicate that between 15% and 25% of eligible households don't collect credits to which they are entitled. Nevertheless, this is a promising approach because tax credits are unrelated to productivity and they have a direct impact on income inequality.

Increased Minimum Wage

Recently there have been increased calls for a $15 minimum wage. Many cities and some states have already passed laws to increase wages to that level. Critics of the minimum wage point out that increased wages can lead to increased prices — thereby reducing the benefit to low-income workers. And indexing minimum wage to the cost of living, as certain states and municipalities have done, may eliminate entry level jobs and accelerate inflation.

Many are watching to see how experiments in Seattle, New York City, and San Francisco will work out. Two recent studies of Seattle's minimum wage offer early indicators. One, from the University of Washington indicates an increase of 3% in total wages in low-paying jobs (the desired outcome) but a reduction, by 9%, of total hours worked (not so desirable). A study by UC Berkeley confirms the increase in total wages and indicates no reduction in overall employment though it does show a decline in employment of workers by limited-service restaurants (fast food). Hours worked was not among the data in the Berkeley study.

These are preliminary results but they tend to confirm the expected pattern. An increased minimum wage incentivizes greater automation and, thereby reduces the total labor demand. For example, fast food restaurants are deploying self-serve kiosks in place of human cashiers. So, while the employed benefit from higher wages, many jobs may be eliminated.

Idling Portions of the Workforce

A recession naturally reduces the number of workers until production is a closer match to demand. Unfortunately, those who lose their jobs are predominantly low-income workers, people who are least able to tolerate job loss. Better options preserve household income while reducing hours worked. These include mandatory vacation time, a shortened work week, and more holidays. When couples elect to have only one spouse be employed that also reduces the workforce.


Bar graph showing decline in jobs requiring high school or less and increas in jobs requiring a postsecondary degree Increasing educational achievement is the intervention most dear to my heart. Not only does a better education enable better wages, but it also results in better health, greater happiness, more active citizenship, and reduced violence. Those with higher educational attainment make better financial decisions. All of this results in a better quality of life.

As more and more routine jobs are automated, the opportunities for those without a postsecondary degree or certificate will continue to diminish. A recent study at Georgetown University indicates that, by 2020, 65% of jobs in the U.S. will require postsecondary education. That's up from only 28% in 1973.

Certainly greater educational achievement is a part of the solution as it moves more people into higher-wage jobs. But those jobs pay higher wages precisely because they are more productive. So, as more people achieve higher levels of education, productivity will continue to increase.

Working With Market Forces

Nobel Laureate Milton Friedman observed that the communication function of markets is as important as the commerce function. When demand rises then prices rise. Rising prices signal suppliers to produce more goods. On the other hand, excess goods result in falling prices signaling manufacturers to reduce production. No command economy has achieved the signaling efficiency of the market.

The same signaling occurs in labor markets. Higher wages for more-skilled jobs will encourage people to seek the necessary education and training to qualify for those jobs. But the counter-point is our contemporary concern, when there is excess labor available, especially for low-skill jobs, then wages may fall below the point where workers can earn an adequate living. Some workers will retrain, and many programs will help pay for that. But retraining takes time, interest, and an affinity for the new field.

With worker productivity in the U.S. soon to crest 4 times basic needs the natural market signal is for less production – but that would mean unemployment. To date, we have interfered with that signal by artificially propping up demand. Massive advertising, high consumer debt, low interest rates, the housing bubble, planned obsolescence; all are symptoms of the interference. Besides, the overconsumption caused by this stimulus is also damaging to the environment.

As productivity continues to increase we will have to allow the signal to get through – it will inevitably anyway. The challenge is finding ways to match production to demand while sustaining employment and wages especially for the most vulnerable.

Framing the Problem

The innovation we need is this:

A way to distribute abundant resources more equitably; while preserving incentives to learn, work, and make a difference; and allowing market signals to balance production and consumption.

We can't look to the past because the challenge of abundance is different from any faced by previous generations. It's going to require a generous sharing of ideas, some experimentation, and development of greater trust between parties.

I'm optimistic that there is a solution because never before in history has society had such plentiful resources as today.

31 March 2017

Cut Taxes or Increase Spending - Is the Debate Obsolete?

Photo of the U.S. Capital with full moon overhead.

As the Trump administration turns its attention to a tax reform plan debate swirls about the best way to stimulate the economy. Traditionally, Democrats have advocated for increased government spending while Republicans have fought for reduced taxes. Both methods succeed in stimulating and both have their roots in the theories of prominent economists. But it may be that both strategies, and the theories that support them, are obsolete in a day when production is many times basic needs.

Government spending advocates cite the work of John Maynard Keynes. Prior to Keynes, neoclassical economists theorized that free markets should naturally balance the economy toward full employment. Keynes observed that economies tended to swing between boom and bust cycles and advocated government intervention through fiscal policy (government taxing and spending) and monetary policy (central bank regulation of the money supply) to moderate the swings. Keynesian theory was influential in addressing the Great Depression and remained dominant following World War II into the 1970s.

Among the expectations of early Keynesian economics was that high inflation and high unemployment should not co-exist. Economist Milton Friedman challenged that notion and was proven right when "stagflation" emerged in the 1970s. Friedman theorized that stagflation and related poor economic conditions result from excessive or malinformed government intervention. The solution, he said, was to free the market through reduced regulation and lower taxes. This school of thought is generally known as "supply-side" or "monetarist". President Reagan successfully employed that approach early in his presidency launching a sustained period of economic growth that continued through the Bush and Clinton administrations.

Today, Keynesian economics is associated with greater regulation, increased government spending, and with an overall trust in government interventions. Meanwhile, monetarist economics is associated with free markets, reduced taxes, and with an overall trust in the market's ability to self-balance. In fact, both schools of thought are much more nuanced than these broad strokes. On the Keynesian side, it matters a lot where the government spends its money. On the monetarist side, it matters a great deal which taxes are reduced and how regulations are tuned. Earnest theorists on both sides have a healthy respect for the other theory.

But the nuance is quickly lost in the morass of political debate. Indeed, I fear that most political Keynesians choose that theory because it justifies their existing desire to increase government spending. And most monetarists choose supply-side theory for it's justification of reduced taxes and regulation. In each case I think they first choose their preferred intervention and then select a theory to justify it.

Through the latter half of the 20th century U.S. government economic focus was pretty much what Keynes described - moderating the boom and bust cycle toward more stable continuous growth. During slow cycles this meant adding economic stimulus, through increased spending and reduced taxes. When inflation started to get out of hand, government would slow things by increasing taxes, reducing spending, and reduced interest rates. Reagan met the stagflation challenge (high inflation and high unemployment) with an unusual combination of reduced taxes (to stimulate hiring) and increased interest rates (to slow inflation). Nevertheless, Reagonomics still used the same tools, just in different ways.

Our contemporary challenge is a new one. Since roughly 2001 the economy has required continuous stimulation to maintain growth. Radical new stimuli such as Quantitative Easing and zero interest rates have been used. Previously experts avoided those stimuli because of their potential to provoke high inflation yet inflation remains at historically low levels and it seems that, without continuous stimulation, the economy will slow to a crawl.

Production compared to Basic Needs

Output per hour of all persons 1947 to 2010

The new economic challenge is due principally to the rapid increase in workforce productivity. According to the U.S. Bureau of Labor Statistics individual worker productivity has more than quadrupled since World War II. Overall productivity per person in 2012 was 412% that of 1947.

Productivity growth becomes even more striking when compared with basic needs. In 2014 U.S. per capita GDP was $54,539. The basic needs per capita that same year was approximately $13,908. So, per-capita production exceeds basic needs by nearly four times. And while the basic needs side of this equation includes the whole population, the productivity side only accounts for those employed, it doesn't include unemployed workers and people choosing not to seek paid employment. So productive capacity compared with basic needs would be even higher.

If it weren't for problems of distribution this would be a great thing! For the first time in history, society has sufficient capacity to provide comfortable housing, plenty of food, health care, entertainment, and leisure time for all. The challenge is that, in a market economy, productivity increases disproportionally benefit those who are already at the higher end of the wage scale.

Disproportionate Benefits

Creative destruction is the term economists us to describe the transformation of an industry by innovation. It is usually associated with the elimination of jobs due to new technology but any innovation that increases individual productivity qualifies. Some examples: The backhoe replaces the jobs of several ditch diggers with that of a more-skilled heavy equipment operator. Computerized catalogs reduce the demand on librarians. Industrial robots replace factory workers. The common feature of such innovations is that they substantially increase the productivity of individual workers. Frequently, these innovations also result in jobs moving upscale — requiring more skill or training and with correspondingly higher pay.

Creatively destructive innovations have led to enormous productivity increases in recent decades thereby reducing the demand for labor. As with any market, when supply increases or demand declines the value also declines. In this case the value of routine jobs has declined dramatically. Here's how economist Dr. David Autor describes it.

And so the things that are most susceptible to computerization or to automation with computers are things where we have explicit procedures for accomplishing them. Right? They’re what my colleagues and I often call “routine tasks.” I don't mean routine in the sense of mundane. I mean routine in the sense of being codifiable. And so the things that were first automated with computers were military applications like encryption. And then banking and census-taking and insurance, and then things like word processing and office clerical operations. But what you didn’t see computers doing a lot of — and still don't, in fact — are tasks that demand flexibility and don't follow well-understood procedures. I don’t know how to tell someone how do you write a persuasive essay, or come up with a great new hypothesis, or develop an exciting product that no one has seen before. ... What we’ve been very good at doing with computers is substituting them for routine, codifiable tasks. The tasks done by workers on production lines, the tasks done by clerical workers, the tasks done by librarians, the tasks done by kind of para-professionals, like legal assistants who go into the stacks for you. And so we see a big decline in clerical workers. We see a decline in production workers. We see a decline even in lower-level management positions because they’re all kind of information processing tasks that have been codified.

Recent creative destruction has predominantly affected lower-middle-class jobs and manufacturing jobs. While increased productivity has made our nation more wealthy as a whole, large sectors of the labor force have been left behind. This may be the biggest factor behind the slow recovery from the 2008 recession. Automation substituted for jobs that were eliminated during the recession; those jobs are not coming back.

The decline in U.S. manufacturing employment has been balanced, in part, by growth in the service sector. This makes sense; growth in productivity has resulted in greater overall wealth. On average, people in the U.S. have more money to spend on eating out, recreation, vacations, and health care. But again, the benefits are not evenly distributed. As workers displaced from manufacturing have moved into the service sector, wages in that area have stagnated.

Disproportionate Impact of Globalization

Economists have consistently advocated for free trade. The math is incontrovertible; when regions or countries with different costs of production trade goods and services, all communities benefit as each is able to specialize and all benefit from the overall productivity increase.

Only recently have economists begun to study how free trade impacts sectors of the economy rather than the economy as a whole. Unsurprisingly, the impact to the U.S. has disproportionally affected manufacturing and routine labor. Here's another quote from Dr. Autor:

When we import those labor-intensive goods, we’re going to reduce demand for blue-collar workers, who are not doing skill-intensive production.  Now we benefit because we get lower prices on the goods we consume and we sell the things that we're good at making at a higher price to the world. So that raises GDP but simultaneously it tends to make high-skilled and highly educated labor better off, raise their wages, and it tends to make low-skilled manually intensive laborers worse off because there is less demand for their services — so there's going to be fewer of them employed or they're going to be employed at lower wages. So the net effect you can show analytically is going to be positive. But the redistributional consequences are, many of us would view that as adverse because we would rather redistribute from rich to poor than poor to rich. And trade is kind of working in the redistributing from poor to rich direction in the United States. The scale of benefits and harms are rather incommensurate. ...

We would conservatively estimate that more than a million manufacturing jobs in the U.S. were directly eliminated between 2000 and 2007 as a result of China's accelerating trade penetration in the United States. Now that doesn't mean a million jobs total. Maybe some of those workers moved into other sectors. But we've looked at that and as best we can find in that period, you do not see that kind of reallocation. So we estimate that as much as 40 percent of the drop in U.S. manufacturing between 2000 and 2007 is attributable to the trade shock that occurred in that period, which is really following China's ascension to the WTO in 2001.

Manufacturing Output Versus Employment

During the campaign, Donald Trump and Bernie Sanders both advocated rethinking free trade. Perhaps we can use tariffs or government incentives to return manufacturing back to the U.S. As it turns out, that's already happening even without incentives. As labor costs increase in Asia the offshoring advantage is diluted. Many manufacturers are, indeed, opening new U.S. plants. The trouble is that returning manufacturing doesn't result in substantial job or wage growth. These are highly automated plants, employing a fraction of the workers whose jobs were eliminated when manufacturing went overseas. For example, Standard Textile just opened a new plant in Union, SC to make towels for Marriott International. Due to automation, the plant only created 150 new jobs. A generation ago the same plant would have employed more than 1000 people. And many of the new jobs are more highly skilled — designing, operating, and maintaining automated machinery.

Creative destruction and globalization are working together here. Both increase overall GDP, both increase individual worker productivity, both increase total wealth, and both disproportionately benefit skilled upper-middle-class workers over blue collar and middle-management workers. Any benefit from manufacturing returning to the U.S. will be blunted by the increase in automation reducing labor needs and shifting what remains to more skilled jobs.

Demand-Side Economics

So far, we have looked at the supply side of labor. The massive increase in productivity over the last six decades has been driven by innovative technology with global trade as an accelerant. As noted before, when the supply of labor exceeds demand then the value decreases. When supply exceeds demand across the economy as a whole then you get a recession.

From the end of World War II through the rest of the 20th century we succeeded in driving demand to keep up with supply. Advertising grew tremendously as an important demand driver. Television programs established new norms: two cars per family, a large home in the suburbs, annual luxury vacations, and designer clothing labels to name a few. Home appliances like air conditioning and a dishwasher went from luxury to necessity.

Government has participated in driving demand. Housing programs made home ownership much more accessible. So much so that it resulted in the 2007 real estate bubble. Likewise, the Federal Reserve has kept interest rates down ensuring that consumer credit remains accessible and people can buy ahead of income.

In the 21st Century we seem to have reached the limits of demand stimuli to compensate for ever increasing productivity. Smaller cars like the Mini Cooper or Fiat 500 have become stylish. Even the wealthy are choosing to reduce consumption — buying smaller homes or moving into the city. The result is that it takes increasingly strong stimuli to keep the economy moving. For the recession of 2008 the government spent unprecedented amounts of money borrowing directly from the Federal Reserve to do so. Despite this pressure, interest and inflation rates remain at historically low values.

Increase Spending or Cut Taxes?

And so we return to the contemporary debate: Should government increase spending or cut taxes to stimulate the economy? When government cuts taxes, individuals and companies have more disposable income. Presumably they will spend some of that income and save part. When government increases spending, it chooses directly where that money will be spent. Both theories depend on "trickle-down" effects even though that has traditionally been associated with tax cuts. In each case, the direct beneficiary of government policy employs more people and purchases more goods and services; those employees and suppliers also do more business and the impact "trickles" through the economy. The primary differentiator is whether you have greater trust in government (increase spending) or the market (cut taxes) to determine who is at the top of the trickle-down pyramid.

The question is really obsolete. Regardless of which stimulus you choose, demand stimuli are increasingly unable to keep up with increased productive capacity. As a country, we already produce nearly four times basic needs and the multiplier will continue to grow. Meanwhile, the twin pressures of Creative Destruction and Globalization will continue to drive the greater benefit of demand stimulus to those who already earn higher wages. Under either strategy, wage disparity will continue to worsen despite attempts by policymakers to direct tax breaks or government spending toward lower income households.

It seems that we will need a greater economic innovation than either of these 20th century solutions. In my next blog post I will write about some promising ideas. More effective education for all students is, of course, an essential component but insufficient by itself.

Estimating Basic Needs Per Capita: The Self-Sufficiency Standard is an measure of the income necessary to meet basic needs without assistance. Values are in terms of household. National averages aren't published so we have to make an approximation starting with samples of two cities. The cost of living index for Milwaukee, WI is 101.9% of the national average. Rochester, NY is exactly 100.0%. The 2014 Average household size in the U.S. was 2.54. We round up to 3 - two adults and one child. For Milwaukee the 2016 Self-Sufficiency Standard for that household is $43,112 annually. For Rochester, the 2010 Self-Sufficiency Standard for the same family is $40,334. Per-capita values are $14,371 and $13,445 respectively. Averaging the values comes to $13,908 approximate U.S. basic needs per capita in 2014. To be sure, there's a lot of variability across region, household size, medical needs, and so forth. I also mixed figures across 2010-2016. Nevertheless, this is a good enough working figure for comparing to per capita production in the same timeframe.