Survey Respondents: Mercenaries of Slaves?

Copyright, The New York Times Company

Economic data is valuable, which is why survey respondents should be paid for their efforts.

Several members of Congress have proposed to eliminate the American Community Survey (a survey of households with special attention to geography and housing), or at least change how the survey is conducted, because it is too intrusive. In particular, the American Community Survey mandates that, with little or no compensation, randomly selected individuals assist the Census Bureau in their data collection by answering a number of questions about how they live.

Economists and business people have recently spoken out on the significant value of the American Community Survey data for their research. The economists Emi Nakamura, Jon Steinsson and Nicolas Vincent explained how the Census Bureau data “provide researchers and the public with a trove of information on everything from the size of families’ mortgage payments in Boise, Idaho, to the nation’s median annual income.”

I agree that economic survey data are valuable, and have used them many times in my own research. But in documenting the value of those data, economists have so far failed to address legislator concerns about survey burdens.

The survey for households consists of 30 pages of forms and instructions. The Census Bureau estimates that the survey takes an average of 38 minutes for each household to complete (including time for reading instructions), which is a total of 1.31 million hours per year for the 2.06 million survey participants in 2010.

But a few economists insist that the survey burdens are absolutely necessary: “Making census-type surveys voluntary … can actually increase costs because the sample size would need to be increased to offset biased response rates.” Processing data from a voluntary collection scheme, they believe, would be too difficult and would take away from some of the value provided by those surveys.

This episode reminds me of debates about military manpower during the Vietnam War. Before 1973, our government would require randomly selected individuals – draftees with low draft lottery numbers – to assist the Department of Defense by acquiring military skills and joining troops in the battlefield. President Nixon appointed the Gates Commission in 1969 to consider whether military manpower could instead be acquired in a normal labor market – that is, on an all-volunteer basis.

At first, military leaders were opposed to an all-volunteer system, referring to volunteers as “mercenaries” and insisting that an all-volunteer system would be less effective. (Gary S. Becker also tells the story that his paper “The Case Against the Draft” was rejected by the Air Force-supported Rand Corporation because Air Force leaders were convinced that the draft was beneficial.)

It was also believed that a draft is cheaper because it allows the military to recruit manpower without paying market wages. But as Milton Friedman persuasively explained to fellow members of the Gates Commission, the cost of the military is not limited to the Department of Defense payroll but also includes the burdens imposed on those drafted and their families. By paying market wages, the all-volunteer system helps shifts some of that burden from the families with unlucky draft numbers to taxpayers generally.

For the same reason, I don’t believe that the randomly chosen American Community Survey respondents should be asked to bear the burden of the survey collection. Rather, taxpayers should bear it by compensating survey respondents for their voluntary participation.

Although the official survey budget does not include an “expenditure” for the time of survey participants, survey participants’ time amounts to a social cost of about $54 million per year (valuing hours answering survey questions at the national average labor compensation per hour of about $41.50).

As the University of Chicago economist Tomas Philipson’s research suggests, using incentive-based compensation for survey participants — similar to that used in many other labor markets — could make government data not only more plentiful but also of better quality, compared to compelling people and businesses to take part at no pay.

Economic data would be of better quality if supplied, as Milton Friedman put it, by mercenaries rather than by slaves.

What Happened to Microsoft's Monopoly?

Copyright, The New York Times Company

As Apple introduces a range of new products, it is worth remembering that there is no such thing as a monopoly in the computer industry.

Fourteen years ago, the Justice Department accused Microsoft of maintaining a monopoly in personal-computer operating systems and expressed concern that Microsoft would extend that monopoly.

At the time, 90 to 95 percent of personal computers powered by Intel processors were running a Microsoft operating system. Some analysts said that this left PC owners little choice, and that Microsoft was preventing the marketplace from functioning well. Others suggested that the market was functioning quite well and that so many PCs ran a Microsoft operating system because of the combination of low price, desirable features and networking advantages.

Two years later, a federal district judge agreed with Department of Justice and ordered Microsoft to be broken into several technology companies – a ruling later vacated by a higher court.



Watching Apple introduce some amazing products last week, including laptops (which have been running Intel processors for several years now) and operating systems for mobile and desktop devices, it is difficult to imagine that the Justice Department once thought that, without government intervention, consumer choices would be significantly limited by Microsoft.

Interestingly, on Tuesday, Microsoft announced that it was introducing a tablet, the Surface, that will compete in the strong and growing market dominated by Apple’s iPad. Market forces are clearly working to give consumers choices in the market.

The first chart below shows Apple’s share price as a ratio of Microsoft’s, both on a monthly basis and adjusted for dividends and splits. The ratio is normalized to one in June 1998, when Justice filed its antitrust suit. I have indicated some of the interesting market events that began a significant movement of customers away from Microsoft and toward Apple: the introduction of a Windows-compatible iPod, the opening of the iTunes store and the introduction of the iPhone (both a phone and a pocket-sized personal computer, running an operating system written by Apple).

finance.yahoo.com

Now, relative to Microsoft’s share value, Apple share values have increased almost 60 times, largely because of Apple’s new products and the market’s anticipation of its future products. This next chart from Asymco.com shows some of that history by quantifying the number of number of personal computers shipped in each year since 1975.

The Rise and Fall of Personal Computing

Asymco.comGrowth of computing platforms, in thousands of units shipped per year. Scale at left is logarithmic.

The chart shows PCs (which here means I.B.M.-compatible lineage, primarily running a Microsoft operating system) overtaking other personal-computing machines in the mid-1980s. However, in the mid-2000s Macintosh and iPhone shipments took off, soon followed by iPads and Androids. By 2011, PCs were less than half of all shipments.

In the computer industry at least, current market shares are no guarantee of future success. What might appear to be monopolies are frequently and decisively broken by innovative competitors rather than antitrust regulators.

Leaks and Press Freedoms

Copyright, The New York Times Company

Government leaks and a free press are not always compatible.

Democracy is said to depend on freedom of the press and other news media, especially the freedom to publish information and opinions without approval or censure from government officials. Empirically, there is a strong correlation between press freedoms and the fairness of elections, absence of autocratic leaders and other hallmarks of democracy.

Full and complete freedoms of the press and other news media are not an automatic consequence of economic development, education and other factors thought to foster democracies. For example, a 2001 World Bank study of 97 countries found that governments commonly owned their nation’s television stations.

Even in Western Europe, countries otherwise known for their political freedoms, government television stations made up more than half of the television market.

In the United States, most newspapers, television stations and cable television channels are owned and operated by private or publicly owned companies, and the government has little to say about what is published in or broadcast over them, with the exception of obscene, indecent or profane programming transmitted by conventional broadcasting systems, which are subject to regulation by the Federal Communications Commission.

One suspects that Americans and news organizations would loudly object if the government offered cash subsidies to newspapers or television stations as a reward for printing or broadcasting favorable coverage of elected officials or levied special taxes on news organizations as a punishment for unfavorable coverage.

Such practices would not explicitly prohibit unfavorable coverage but would make it more difficult for outlets that displeased the government to survive in a competitive marketplace for readers and viewers.

But candidates and officials can and do reward specific news organizations with personal appearances and interviews that the news organizations and their owners find valuable — or withhold interviews from those they find hostile, as Barack Obama did with Fox News for much of his first presidential campaign.

The private sector does not have the moral high ground, and there are times that private media outlets are unfair to government officials. Nevertheless, the American approach has been to prohibit the government from levying cash penalties on or offering cash rewards to media outlets on the basis of the fairness of their coverage.

Sometimes, however, government officials possess information that would be of interest to the public, and thereby valuable for news organizations. In principle, officials could reward favorable coverage by distributing or “leaking” more information to sympathetic outlets than to unfriendly ones.

We don’t yet have a data set or systematic study of the transactional characteristics of leaks — such as which outlets receive the most leaks and whether those outlets are more favorable in their coverage to the agencies doing the leaking. But government leaks can potentially be a noncash currency that can be traded between the government and news organizations.

Democracy is not jeopardized by a few leaks, and news media deal in a whole array of sports, business and other information to which government officials have no special access. But punishing government leakers is not inconsistent with democratic values. By preventing leaks from being used by the government as a currency for rewarding or penalizing media outlets, it could help keep the press truly independent.

Active Government Means More Lobbying

Copyright, The New York Times Company

As our government enters into new policy areas and revamps old ones, expect lobbyists and others to be more politically active than ever.

Government has undertaken a whole new range of activities in the last several years, from running a major automobile manufacturer to buying stock in banks to providing health insurance for able-bodied middle-aged adults. Other areas of government involvement, like education, military spending, Medicare and Medicaid have become increasingly costly over the last decade.

Some people think government activities in these areas are long overdue. Others say government expansion was unnecessary. Either way, both sides can agree that businesses, institutions and families have more at stake in their government.

It’s only natural, then, that businesses, institutions and families would spend more resources trying to influence public-sector decisions. Just as in many other types of behavior, political behavior is affected by costs and benefits. And the costs and benefits of influencing the government have continued to grow.

State and local governments have been cutting back some employees and reducing some activities. But in the short run, businesses and people have a lot at stake in state and local government activities, too. If nothing else, they can try to influence the composition of state and local government cuts.

Their interest in public policy is an important reason that businesses, organizations and individuals make campaign donations, and why these various parties protest against government actions in a variety of ways.

Some people have expressed dismay at the unprecedented amounts spent on 2012 political campaigns. But heightened political spending and other forms of political participation are an expected consequence of our more active government.

As explained by a pioneer in political economy research, Gordon Tullock, the real surprise about spending on campaigns and lobbying is how little it is compared with the amount of resources controlled by governments.

The federal government spends about $4 trillion every year, and state and local governments another $2 trillion, not to mention the resources these governments control through regulatory activities.

At the same time, estimates of aggregate campaign and lobbying spending are well below than 1 percent of total government spending. For example, analysis of filings under the Lobbying Disclosure Act finds that $3.5 billion was spent on lobbying in the year 2010.

Although economists have trouble explaining why observed lobbying spending is so little in total, economic analysis has been successful at explaining why there is more lobbying in California than, say, Vermont and why lobbying expenditure often peaks at the height of legislative activity. More is at stake.

For better or for worse, an active government begets lobbying activity.

Price Discrimination -- An Example from the Cell Phone Industry

The major cell phone carriers charge iphone users approx $90 per month for cellular and data services on an iphone. At least one pre-paid carrier offers cellular and data services for iphone owners at $45 per month ($50 with taxes, or about $42 with taxes if prepaying one year).

That pre-paid carrier is Straight Talk, which appears to be a partnership between Walmart (retail), Net10 (an experienced pre-paid carrier), and AT&T (cellular network), offering pre-paid services on a number of smartphones and other cell phones. (I am no expert on the partnerships, but my son has his iphone4 on straightalk and I see obvious fingerprints of all three partners).

Despite the 100% price difference, both marketing and setup costs are preventing most iphone users from switching away from the major carriers. On the marketing side, Walmart and Straight Talk are pretty quiet about the fact that iphone users can use their $45/mo service (did their AT&T partner pay them to be quiet?). Even cnn.com had a story claiming (incorrectly) that, as of today, none of the pre-paid carriers are available for iphone users.

One the setup cost side, iphone owners need to take four separate steps to join the service:

1. Obtain an AT&T compatible iphone (new or used).  I don't think you can buy one at Walmart without signing up with AT&T or Verizon, but many other places cell them contract free.
2. Purchase a $15 AT&T compatible micro-sim card from Straight Talk and insert it into iphone.
3. Activate the micro-sim with Straight Talk (if you had a phone number from another carrier that you previously were using with iphone or any other cell phone, at this stage you can assign it to the micro-sim).  It may take up to two days from the time that you activate and the time that your iphone can begin to place and receive calls and/or text messages.  At this stage you pay your $45.
4. Set the APN numbers on the iphone to work with Straight Talk (this step is necessary to use data services).  One way to do this is to connect iphone to wifi and then go to go to http://www.unlockit.co.nz, and the iphone will automatically adopt the proper APN numbers.  Update Oct-2013: follow these instructions to get your iphone to automatically adopt the proper APN numbers.

I suspect that knowledge of these steps will diffuse pretty quickly, and in less than a year major carriers will have to reduce their monthly cellular charges to be much closer to Straight Talk.

Update Oct-2013: I did not correctly predict AT&T's pricing.  It appears that AT&T is actually charging a bit more (eg., no more 200-MMS packages) than they used to.

Update Jan-2014: AT&T has finally moved prices a lot closer to Walmart.

The Asymmetric Recovery

Copyright, The New York Times Company

By one measure, the labor market has not recovered at all. By another, the recovery is complete.

The red line in the chart below is a monthly index of the employment-to-population ratio, normalized to a value of 100 in December 2007, when the recession began. In this series, each employed person counts the same, regardless of how many hours she or he works.

Federal Reserve Bank of St. LouisAverage weekly hours of private employees (blue line) have returned to the level last seen before the recession of 2008-9, shown as gray area. But the percentage of Americans with jobs (red line) plummeted in the two years after the recession began and has remained steady since then.

The employment-to-population ratio fell more than 5 percent during the official recession period and fell almost an additional 2 percent in the second half of 2009. The ratio has been essentially constant since then; there has been no meaningful increase in the fraction of Americans who are employed.

The blue line shows the average number of hours worked by people with jobs in the private sector. In this series, only people with jobs are included in the calculation.

The hours series reflects a large contraction of more than 2 percent during the recession years. In other words, a number of employees found that their hours were cut during the recession and that long-hours jobs like construction were lost to a greater degree than short-hours jobs.

Unlike the employment-to-population ratio, average work hours have largely recovered since 2009. Earlier this year, the average hours series reached 100, which was its value for much of 2007.

To put it another way, 93 percent of the people who would have been employed in an economy such as we had in 2007 are employed today and are likely to feel that they are as busy with work today as they were then. The other 7 percent are not working at all.

There are several theories as to why the recovery has been so asymmetric in terms of hours and employment.

One is that the unemployment insurance program, eligibility for which has expanded significantly since 2007, pays people to be without a job but usually does not pay them for having reduced hours. Thus, unemployment insurance does more to prolong joblessness than it does to prolong working at reduced hours.

Another theory is that actual or anticipated health insurance costs discourage employers from having too many employees; instead, they accomplish their tasks by having existing employees work longer hours.

Because health insurance costs accrue on a per-employee basis, an employer with, say, 38 employees working 40 hours each has the same total number of work hours as an employer with 40 employers working 38 hours each, but the former employer’s health insurance costs are 5 percent lower.

Yet another theory is that employers are concerned that their economic conditions may change quickly and find it easier to adjust work schedules than to adjust the number of employees.

And still another theory notes that the employment-to-population ratio would have fallen even without a recession as baby boomers reached retirement age. At the same time, those retirements might not reduce average work hours among those who had not yet reached retirement and might even increase them.

In any case, it is worth monitoring the separate recoveries for employment and for hours, as they can shed light on supply-and-demand factors that caused the recession and have prevented a full recovery by all measures.

Suddenly Starting a New Job after 79 Weeks of Unemployment

On the 12th of this month, unemployment benefits were terminated for persons living in Illinois who had so far been unemployed and collecting benefits for 79-99 weeks.

The New York Times interviewed one of the persons affected, Candace Falkner, 50. Persons, like Ms. Falkner, who are unemployed for 79 weeks or more are often described as being unable to find a job. They would never turn down a job just because they were receiving unemployment benefits.

Yet, according to the New York Times, Ms. Falkner is working now, only a week or two after her unemployment benefits ran out. With no jobs to be found for the prior 79 weeks, did Ms. Falkner's job coicidentally appear?

The story describes Ms. Falkner's new job as "a commission-only, door-to-door sales job," and describes Ms. Falkner as a person with a master's degree, so I think we are to understand that Ms. Falkner is qualified for a better job than that. If so, it seems the 79+ weeks that she was unemployed cannot be described as a time when there were no jobs to be found, but rather a time when there were no "good" jobs to be found.

That's exactly the story I have been telling on this blog for several years: unemployment insurance makes unemployment last longer, even while it goes to people in unenviable situations. It also raises wages, as the jobs that are not good enough for the unemployed to accept either go unfilled or are not created in the first place.

Perhaps these consequences are desirable or at least tolerable for the intrinsic benefit of knowing that taxpayers are helping people when they experience tough times. But they are consequences nonetheless: the fact is that some of the reason that the unemployment rate is 8 percent rather than 5 is that unemployment is subsidized.

Do Tax Credits Encourage Work?

Copyright, The New York Times Company

The earned-income tax credit is often said to encourage work, but it may do just the opposite.

This refundable federal income tax credit of a few thousand dollars a year is paid to families with positive but low earned income (that is, wages or salaries) for a calendar year.

The chart below shows the credit’s schedules for the 2011 tax year as a function of annual earned income for a given family situation (other family situations have the same basic shape). The schedule shown illustrates the mountain-plateau pattern described above: an increasing portion for the lowest incomes, a flat portion, a decreasing portion and then finally a flat portion of zero.

Source: Internal Revenue Service

Along the increasing portion, the credit adds to the reward from working because a few more weeks or hours worked during the year tends to add marginally to the income from work, which adds to the amount of the credit.

Because of this pattern among the lowest incomes, the earned-income tax credit is often said to encourage work among low-income people. Indeed, studies have shown that the credit leads low-skill single mothers to work more and may be the most effective program at raising households out of poverty (the poverty line tends to be near the middle of the plateau portion of the earned-income tax credit schedule).

For the same reasons that the credit encourages more work among people who might otherwise earn close to zero during a year, it can also influence some people to work less — those with earnings at or slightly above the downward-sloping or “phase-out” portion of the schedule, where people lose about 20 cents of their credit for every additional dollar earned during a year.

In other words, for households on the downward-sloping portion of the earned-income tax credit schedule, the credit acts as an extra 20 percent tax on the income they earn in that range. The work-encouraging potential of the credit occurs only on the upward-sloping portion.

At first glance, it might appear that a lot of people might earn no income and might therefore be among those encouraged to work by the credit. After all, many people spend some time unemployed and earning no income.

But the credit applies to a household for a calendar year. It is uncommon for nonelderly heads of household to be out of work for 12 consecutive months and, even when they are, for those 12 months to coincide with a calendar year.

Someone unemployed from July 2010 to June 2011 but otherwise employed, for example, would have positive income in both calendar years 2010 and 2011. Even when someone is out of work for an entire calendar year, he or she may have a spouse who does work at least part of the year.

For these reasons, it is more common for families to be on the part of the earned-income tax credit where it acts as a tax, rather than a reward to additional work.

When a Spending Cut Can Be Like a Tax Increase

Copyright, The New York Times Company

Sometimes Republicans and Democrats are like Coke and Pepsi: a lot more different in marketing than they are in substance.

Aging, changes in the health sector and the recession all serve to increase what the government spends on social welfare programs. Republicans and Democrats both acknowledge that some kind of adjustment is needed to bring the amount the government spends in line with the amount that it taxes.

Republicans are more apt to say that spending needs to be cut in order to match government revenues that are currently less than spending. Indeed, many of them have taken a “no new taxes” pledge. Democrats are more apt to consider raising revenue to help pay for the cost of social spending.

To the degree that Republicans propose to cut spending by means-testing programs that were formerly provided to all income groups, there may be little or no economic difference between Democratic and Republican proposals.

Take, as an example, Medicare, the health-insurance program administered by the federal government primarily for people 65 and older. It is a universal program, rather than an antipoverty program: all citizens can receive benefits from the program when they become old, regardless of how rich or poor they are.

To simplify the arithmetic, suppose the medical goods and services provided cost an average of $11,000 a year per beneficiary. The government could provide Medicare free of charge, in which case each beneficiary effectively receives $11,000 in premium assistance from the public treasury in order to pay for the cost of the program.

The premium assistance by itself tends to increase the government deficit in proportion to the size of the older population, because the government has to pay for the medical goods and services. One idea, close to some Republican thinking, is to reduce the deficit by cutting government spending by “means-testing”: limiting the assistance to people who cannot afford to pay the premium themselves.

For example, the premium assistance might be reduced 10 cents for every dollar of a beneficiary’s income. (Medicare Part A is currently is a universal program. Medicare Parts B and D currently subsidize premiums for all beneficiaries, but to a greater degree for those with low incomes; in this regard Medicare is a means-tested premium assistance program.)

The universal and means-tested approaches are shown in blue and red, respectively. The horizontal axis shows annual income ranging from zero to $120,000 a year. The universal premium assistance approach is shown as a horizontal line, because all beneficiaries receive the same assistance regardless of income. The green means-tested schedule slopes down, because high income people receive less assistance than low-income people.

An alternative approach would be to keep the universal assistance but help pay for it with an additional 10 percent income tax on people 65 and older. A 10 percent income tax on incomes below $110,000 a year is shown in red in the chart. Democrats have proposed raising taxes on wealthier Americans to avoid deeper spending cuts, though none have called for a tax on the elderly, as in this hypothetical example.

Under the income tax version, the net benefit to beneficiaries is the difference between the blue and the red lines in the chart, which coincides with the green means-tested proposal.

Although the new tax and cut-by-means-test proposals sound different – the former proposal raises revenue without cutting spending while the latter proposal cuts spending without raising taxes – they are economically equivalent in these examples.

Both proposals cut net benefits (or raise net taxes) more for higher-income beneficiaries than for low-income beneficiaries. Putting more of the burden on higher-income beneficiaries is sometimes described as fair or equitable, but it also adds to the penalty for having a high income and, equivalently, subtracts from the burden of having a low income. That results in more people with low incomes, because incomes are determined in part by effort, which is affected by costs and benefits.

My point here is to advocate equity rather than incentives and to point out that both proposals — tax increase and spending cut — can increase equity and reduce incentives, in much the same ways.

Despite the economic similarities between the two approaches, I expect Republicans and Democrats to continue to differentiate their policies excessively, with Republicans arousing taxpayers over Democratic plans to raise taxes and Democrats alarming beneficiaries about imminent cuts in important social programs.

Food Stamps and Unemployment Insurance

Copyright, The New York Times Company

The Department of Agriculture’s food-stamp program, now known as the Supplemental Nutrition Assistance Program, or SNAP, was originally intended as a program for the poor. But it has transformed itself into an important source of support for the unemployed.

Traditionally, food-stamp program participants were subject to an asset test – households with a total of more than a few thousand dollars of assets in their bank accounts, automobiles and other assets were not eligible even if income was zero. In this way, food stamps had a lot in common with Medicaid and other antipoverty programs.

The welfare reform of the 1990s also required able-bodied adult food-stamp recipients, without children, to be working or enrolled in a job-training program, or their eligibility would be limited to a few months.

For these two reasons, the food-stamp program had little in common with unemployment insurance, which offers weekly cash benefits to people who have lost their jobs and have been unable to find and start a new job.

Unemployment insurance has no asset test; even people with money in the bank can receive benefits from the program, as long as they have been laid off from a job and continue to look for a new one. Unemployment insurance also has no work requirement. Indeed, it has just the opposite – a person going back to work has his or her unemployment benefits terminated.

As a result of the very different eligibility rules for the two programs, a vast majority of people receiving unemployment insurance were not receiving food stamps. By my estimates, only about one-quarter of households with a head or spouse unemployed were receiving food stamps in the early 2000s.

But the food-stamp eligibility rules have changed markedly in the last several years, bringing the program closer to unemployment insurance. Food stamps effectively no longer have an asset test. States have also received waivers from work requirements during the recession (for a while, the requirements were waived nationwide by the 2009 stimulus law).

As a result, food-stamp participation is now more common among the unemployed. I constructed the chart below from Department of Agriculture quality-control data on the activities of non-elderly food-stamp recipients back to 2000 and from Census Bureau data on the prevalence of unemployment among the non-elderly population.

Sources: Department of Agriculture, Census Bureau

The former source gives me the number of non-elderly household heads and spouses who are both unemployed and receiving food stamps for their households, and the chart shows their number as a share of the total unemployment among non-elderly household heads and spouses.

The ratio was about one to four in the early 2000s and actually fell during the 2001 recession as unemployment increased more than food stamp participation did. The ratio rose in the mid-2000s as states relaxed their eligibility rules and again in 2008 as the federal government relaxed its rules.

By the 2010 fiscal year, half of non-elderly households with an unemployed head or spouse were receiving food stamps.

Meanwhile, the eligibility rules for unemployment insurance are scheduled to tighten this year, as the federal government puts shorter limits on the length of time that people can receive benefits. Thanks to the waivers cited above, food-stamp participation effectively has no time limit.

I expect that soon the food-stamp program will support more unemployed households than unemployment insurance does.