View The Redistribution Recession now on amazon.com

You can read the first chapter, and search the entire book, for free.  When you're ready to read the entire book, it's only $30 including shipping!

Get The Redistribution Recession extras on its www page.

The iPhone5 and Consumer Spending

Copyright, The New York Times Company

That the introduction of iPhone 5 increases consumer spending is no triumph for Keynesian economics, but merely an advertisement for plans by the Bureau of Economic Analysis to improve national accounting.

Suppose there were an island economy with 100 able-bodied adults, all of whom work as fisherman, each catching a fish a day. The 100 fish are eaten by the island people, which makes inflation-adjusted daily consumer spending in the economy – or consumption, as economists call it – equal to 100.

(In order to clarify concepts and measurement practices, I have deliberately kept this model economy simple, with no unemployment, liquidity traps and the like.)

Now suppose that 10 of the fisherman decide to quit fishing to build, nurture and tend to an apple orchard. In the beginning, the orchard produces no apples, so consumer spending drops to 90 fish a day, because only 90 adults are out fishing while the other 10 are developing an orchard.

Time passes, and a bountiful 100-apple harvest occurs. Because the apples taste good, the island fishermen obtain all 100 apples from 10 orchard owners in exchange for fish. The apple harvest has by itself boosted consumption in the economy, which suddenly jumped from 90 fish a day to 90 fish a day plus 100 apples.

It would be ridiculous to proclaim that the apple harvest and its effect on consumer spending prove that the economy is plagued by a liquidity trap, or to insist that the immediate effect of the harvest on consumer spending is independent of the quality of the apples. By construction, this model economy has no Keynesian features, yet nonetheless a harvest has a large effect on measured consumption.

Consider now the American economy, in which Apple has just released its iPhone 5, and some economists say they believe that the release itself will noticeably increase aggregate consumer spending. Paul Krugman has gone even further, to assert that the increase is a proof for Keynesian economics and that the “short-run benefits from the new phone have almost nothing to do with how good it is.”

Even if Keynesian economics were completely wrong, economists would expect the iPhone 5 release to cause consumer spending and gross domestic product to be greater than it was before the release, to a degree related to the phone’s overall value to consumers. Note that, as in my island economy, the development of Apple products reduces consumption before the release, because the people working on the coming products are not available to produce consumer goods during that time.

Much of the development work on the iPhone 5 did not, before its release, count as investment or G.D.P. (G.D.P. is the sum of public and private consumption and public and private investment). The national accounts treat research and development activities as intermediate inputs, which means that they are subtracted from revenue for the purpose of determining a corporation’s contribution to national production.

This same is true for, say, Apple’s legal expenses in developing patents (many of which are discussed on the Mactech Web site) and license terms for their new product.

These development activities appear as G.D.P. only when the product is completed and sold. If the product is not valuable, it will not sell and will not count for much, although national consumption could still rise if upon project completion the developers move out of development and into the production of consumer goods.

(Research and development employees usually receive wages during the development phase, but their prerelease compensation comes out of corporate profits).

For the purposes of understanding the timing of economic activity, the national accounts’ treatment of research and development is a weakness, because it recognizes the developers’ activity at the wrong time – only after the product is released. Economists at the Bureau of Economic Analysis (the agency producing our nation’s accounts) are aware of this weakness, how it has become acute with the rise of the technology sector and steps they might take to improve the accounts. They are doing the best they can with the data and economic research results that are available.

Measured consumption rises in the quarter when people buy their new iPhones, not when they actually use them (in my island model, the consumption would be measured when the apples are bought, not when they are eaten). IPhones are durable goods that are used for years after Apple sells them as new, including by second and third owners.

Curiously, the timing of measured iPhone consumption would be markedly different if Apple or cellular carriers had chosen to rent the phones rather than sell them, because the measured consumption would occur each month when users paid their rental fees (the Bureau of Economic Analysis is aware of this weakness and rectifies it in sectors where it is most acute, like the housing sector).

The iPhone 5 release is no occasion to cheer for wasteful government spending, but perhaps does help make the case for a larger budget at the Bureau of Economic Analysis, so that it can continue its progress on measuring the amount and timing of economic activity.

Catching Up with Blog Archives

A bunch of posts dated from June, July, and August just appeared because I am archiving my economix entries.

Meanwhile, Supply & Demand appears in this cool video

21st Century Inequality Trends

Copyright, The New York Times Company

A couple of important measures of labor-market inequality have played out since 2008 much the way they did previously.

One measure of changing inequality in the labor market, commonly used by economists, is the annualized 90-10 change: the annualized growth rate of wages at the 90th percentile of the distribution of wages for men working full time minus the growth rate of wages at the 10th percentile of the same distribution.

Because the 90th percentile wage is the wage below which 90 percent of working men earn, the annualized 90-10 change can be interpreted as the degree to which the wages of high-earning men grow more, or fall less, than the wages of low-earning men. The measure can in principle be negative, in which case the wages of less-skilled people would be partly catching up with the wages of skilled people.

I measured these changes using the Census Bureau’s Current Population Survey Merged Outgoing Rotation Group sample. (Analysis of wage patterns is often performed with the Census Bureau surveys, although most of them lack information on employee fringe benefits over and above wages and salaries.) The left part of the chart below shows the annualized 90-10 change over three time periods.

The first time period is 2000-8, the years of George W. Bush presidency, and the years before the 2009 depths of the “Great Recession.” Wages grew faster for high-skill people during these years: about 0.5 percentage points a year extra (roughly a cumulative four extra percentage points for the entire period).

Wages also grew faster for high-skill people between 2008 and 2010 (the last year for which I have data). The average growth differential between them and low-skill people was a whopping 2.6 percentage points a year.

One interpretation of these results is that the rewards for accumulating skill have been increasing over time, especially in the last two years. The more common and less euphemistic interpretation is that the rich have been getting richer, especially recently.

Perhaps the Great Recession of 2008-9 and slow recovery is to blame for all of this. For this reason, I looked separately at the recession period 2000-2. That recession (0.6 extra percentage points a year) looks a lot like the longer 2000-8 period (0.5 extra percentage points a year).

Another indicator of inequality is wage equality between the genders. The right part of chart shows changes in gender wage equality. The gender measures are positive if and when wages grow faster for women than for men and negative if and when male wages grow faster. The former is usually viewed as progress, because for centuries women have been earning less than men.

During the George W. Bush presidency (and a number of the presidencies before him), wages grew more for women than for men, about seven-tenths of one percentage a year more on an annual basis. During the first two years of the Obama presidency (or, if you want, during the Great Recession), the gender wage gap closed even more rapidly (1.3 percentage points a year), though less rapidly than it did in during the previous recession (1.6 percentage points a year).

President Obama is sometimes likened to Lyndon B. Johnson or Jimmy Carter, but theirs were not years when high-skill people were gaining ground at a faster pace than low-skill people and not years when women’s wages were catching up to men’s. When it comes to measures of labor-market inequality, the last few years so far look at lot like the years before 2008.

Perhaps that shows how Presidents Bush and Obama are not so different in their economic policies, or that presidents have little impact on important economic trends.

Social Insurance and Layoffs

Copyright, The New York Times Company

Unemployment insurance and other types of social insurance subsidize job separations and thereby result in too many layoffs and too few people employed.

A variety of programs help workers after they leave a job and do not start a new one, depending on the circumstances of the job separation.

Unemployment insurance is often available when the worker was laid off and continues to look for work. Disability insurance is available when a worker’s health makes it too difficult to remain on the job. Social Security’s old-age insurance program provides income for elderly people after they leave their jobs.

Layoffs, disability events and retirements have some differences, of course, which is why each type of job separation has a separate insurance program. But in each case, a working relationship between an employer and an employee has been terminated, and the worker has not started a new one with, say, different working conditions or a different rate of pay.

In their analyses of disability and old-age insurance, economists have found that insurance reduces the cost of job separations and thereby increases their numbers, because the insurance helps replace the income and production that is lost when the worker stops working.

The prospect of the insurance payments gives employers less reason to change the nature of a job to encourage a disabled or elderly employee to remain at work and gives employees less reason to accept changes in working conditions or pay that would make it easier for employers to retain them.

David Autor of the Massachusetts Institute of Technology has studied the United States federal disability-insurance program and finds that it “provides no incentive to employers to implement cost-effective accommodations that would enable disabled employees to remain on the job.”

The Congressional Budget Office explains further about disability insurance that “because the D.I. program is funded through a flat-rate payroll tax on employers and employees, employers do not bear the costs associated with a disabled worker who stops working and becomes a beneficiary in the D.I. program.”

Note that disability benefits are paid to employees, not employers. Nevertheless Professor Autor and other economists conclude that the benefits affect employer behavior because the employment relationship is exactly that: a relationship between employer and employee.

If an employee has better, or less bad, options outside the relationship, then the employer will find the employee more expensive to keep. Disability and other forms of social insurance increase the income employees can receive outside the job and thereby make employees more expensive from an employer’s point of view.

The economists Jonathan Gruber and David Wise have found that Social Security provisions “provide enormous incentive to leave the labor force early.” Just like disability insurance, Social Security provisions can shift some of the burden of job separations from the private sector to the public insurance programs and thereby give the private sector too little incentive to prevent or delay the separations.

Employers sometimes experience reductions in demand from their customers, as auto manufacturers and home builders did early in the recent recession. One way they react is to lay off part of their work forces. But they could also adapt to less demand by work-sharing, reducing prices charged their customers (or increasing those prices less than the general rate of inflation) or reducing wages.

Smart employers recognize that one of these adjustments — layoffs — brings forth help from the government through its safety-net programs (on behalf of employees); the other adjustments do not. If the safety net were less generous, there would be fewer layoffs during a recession, because employers would adjust less with layoffs and more in other ways.

(State unemployment insurance programs are, and have been, “experience rated” in the sense that employers sometimes find their payroll taxes increased for each employee they dismiss. However, the experience rating is imperfect; some employers are already at the maximum tax rate and further layoffs would not increase it. More important, the effect of experience rating on employer costs of layoffs was present even before the recession. What’s new since 2008 are the federal extended and emergency unemployment programs that are not experience-rated, thereby adding to the benefits an unemployed person can expect to receive without adding to the taxes levied on his former employer).

Thus, even if it were true that the unemployed completely ignored the safety net’s generosity in their decisions to seek and accept jobs, the safety net would still increase unemployment during a recession by increasing layoffs.

Yet economists have recently forgotten this important connection between unemployment insurance and the number of people employed. The C.B.O. has looked at the economic effects of unemployment insurance and noted that extending unemployment benefits would “reduce the intensity of some workers’ efforts to search for a new job because the higher benefits would lessen the hardship of being unemployed.” But the C.B.O. concluded that “the net impact on the unemployment rate from some workers’ reduced efforts to find a job would be slight.”

Although it looks at incentives to avoid job separations in its consideration of disability insurance, the C.B.O. makes no mention of the same incentives in its analysis of unemployment insurance and consequently is premature in its rejection of the basic economic proposition that paying people for not working will reduce the number of people who work.

Is the Fiscal Cliff a Big Deal?

Copyright, The New York Times Company

With their Keynesian analysis, the Congressional Budget Office and others have exaggerated the effects of the “fiscal cliff” on the labor market and the economy.

Come January, current law provides for significant cuts in federal spending and for tax increases – and thereby significant federal budget-deficit reduction. These provisions have been collectively described as the “fiscal cliff,” which emerged when Democratic and Republican leaders could not agree on plans on spending and taxes.

The Congressional Budget Office has warned that the fiscal cliff will cause a double-dip recession, but its analysis for 2013 is based on the Keynesian proposition that anything that shrinks the federal budget deficit shrinks the economy, and the more the deficit is reduced the more the economy is reduced.

In many circumstances, the Keynesian proposition reaches the wrong conclusions about economic activity, because deficits do not necessarily expand the economy or prevent it from shrinking. For example, reducing the deficit by cutting unemployment insurance – it’s one of the programs that would be cut in January – would shrink the economy in the C.B.O.’s view.

But in reality, cutting unemployment insurance would increase employment, as it would end payments for people who fail to find work and would reduce the cushion provided after layoffs.

Helping people who are out of work may be intrinsically valuable because it’s the right thing to do, but the Congressional Budget Office is incorrect to conclude that it also grows the economy or prevents it from shrinking. Paying people for not working is no way to put them to work.

The Keynesian proposition about budget deficits ignores incentives of all kinds, so its incorrect conclusions about the fiscal cliff are not limited to unemployment insurance. Another example: the fiscal cliff would put millions of Americans on the alternative minimum tax, which Keynesian analysis said would shrink the economy solely because it collected more revenue.

Yet economists who have studied the alternative minimum tax have found that its effects on incentives to work and produce are essentially neutral, compared with the ordinary federal personal income tax.

(The Congressional Budget Office does not use pure Keynesian analysis for its long-term projections, which include labor-supply incentive effects of tax rates, but apparently has decided that incentives’ effects can be safely neglected in the short term.)

None of this implies that the fiscal cliff will expand the economy, because some of its provisions will increase the penalties for working and producing.

The fiscal cliff would cut Medicare payments to doctors by 2 percent, which reduces doctors’ reward for treating Medicare patients. This may cause doctors to work less (or to work more for non-Medicare patients). The fiscal cliff would end the “Bush tax cuts” provisions, some of which have been enhancing the incentive to work (but beware – not all laws labeled “tax cuts” enhance incentives).

Perhaps the incentive-reducing provisions of the fiscal cliff outweigh its incentive-enhancing provisions, in which case the Congressional Budget Office has arrived at approximately the right answer for the wrong reasons.

But even in that lucky case, the C.B.O.’s quantitative estimates of the fiscal cliff’s economic effects are not reliable until they fully incorporate economic incentives.

Is Deficit Spending the Answer?

Copyright, The New York Times Company

Deficit spending comes in several different flavors, each of which varies in terms of its effect on the labor market and the economy.

Deficit spending occurs when government spending exceeds government revenue. By official estimates, the federal government budget deficit has been $1.3 trillion during each of the last three fiscal years and even larger the year before that, when the financial crisis and bailouts were at their peaks. Previously, the federal deficit had never reached $0.5 trillion.

The alternatives to deficit spending are a balanced budget or a surplus budget, when government revenue is at least as much as its spending.

Economists do not fully agree about the macroeconomic effects of deficit spending, compared with the balanced-budget alternative, but they do agree that not all deficit spending is the same. Deficit spending that is the result of extra government spending is different from deficits that come from tax cuts. Moreover, the forms of the extra spending matter, as do the forms of the tax cuts and how the debt will be repaid in the future.



One form of deficit spending is extra government employment (civilian or military), as during wartime, paid for with extra taxes after the war is over. This type of spending probably increases aggregate employment during the war because the government is paying people to work and, while the deficit spending lasts, not yet taxing them extra for working.

This type of deficit spending is relevant today, because America continues to fight wars in the Middle East and to fight the war on drugs in our hemisphere. However, this type is not much different during the last four years of trillion-plus deficits than it was before.

The more important source of enlarged federal deficits is increased spending on transfers, like food stamps and unemployment insurance, and in-kind subsidies for the poor, like Medicaid. Transfer spending helps poor people, but paying people for low incomes or for unemployment has the effect of reducing the reward to work, rather than increasing it as government employment programs might.

By considering work incentives, I conclude that the contribution of transfer spending to the deficits of the last four years have reduced employment, rather than increasing it as wartime deficits might.

The temporary payroll tax cut has also added to the government deficit over the last two years. The payroll tax is levied on people who work and not on people who are out of work, so the cut had the effect of reducing the tax penalty on work. This helped offset the employment-depressing effect of transfers, although my estimates suggest that the offset was less than 100 percent (more on those in future blog entries).

Deficit spending adds to the government debt, because the government has to borrow to obtain the funds it does not have from taxes. It is sometimes argued that deficit spending reduces employment because of fears over the future repayment of the debt. But future fears can also encourage people to work harder to save more for the bad economic situation that is anticipated in the future and to work harder to take advantage of today’s tax rates, which might seem low compared with what lies ahead.

Moreover, the bond market pays dearly for United States government bonds: they may be the most expensive bonds (that is, the bonds with the lowest yields) in the world. If the market continues to value United States government bonds so dearly, much of the United States debt may never need to be paid off.

This may seem like a free lunch, but economists understand it as a “liquidity service,” or feeling of safety that the government supplies to the marketplace for which the government is compensated (Milton Friedman’s classic argument said low yields on government securities indicate that more of the securities should be supplied to the market).

The secret to understanding the effects of deficit spending on the labor market and the economy is to examine the incentives created by the additional spending and by tax cuts.

Trickle-down Fairy Dust

Copyright, The New York Times Company

Generally, transfers do not expand the economy, regardless of whether the recipients are rich or poor. Rather, they confer benefits that are limited to the direct recipients of those transfers.

President Obama recently denigrated the Romney-Ryan economic plan by saying the Republican candidates believe that if Congress gave “more tax breaks to the wealthiest Americans, it will lead to jobs and prosperity for everyone else.” He called that approach “trickle-down fairy dust.”

The trickle-down theory says that a policy benefiting a specific group, like the wealthiest Americans, will somehow confer benefits on everyone else. One common version of this view is that the benefiting group increases its spending, which in turn benefits the industries whose products are purchased with such additional spending, perhaps leading to the hiring of more workers to meet the increased demand.

The Obama administration has its own version of this — what we might call trickle-up — which says that policies benefiting unemployed people even help people who are not unemployed, because the unemployed react to their government benefits by spending more.

Both of these views are flawed for two reasons. First, the transfer of resources from one group to another probably does not increase aggregate spending (if incentives are held constant), because we have to consider both the spending of the receiving group and the spending of the group that finances the benefits through tax payments or loans to the government.

Transfers may change the composition of spending to the extent that the benefiting group spends its resources differently than the financing group. Cutting taxes on the rich and raising them on the poor would probably, holding incentives constant, increase spending on fancy restaurants, investment goods and yachts, and decrease spending on, say, groceries and cellphones.

The altered composition of spending can have some interesting, but probably small, effects on the labor market, depending on the labor intensity of the various industries involved (as I noted in a previous post). But changing the composition of spending is not the same as changing the total.

Few transfer programs hold incentives constant. Unemployment insurance reduces the incentive to find work and (especially when it is federally financed) reduces the incentive for businesses to avoid layoffs, and in this way reduces aggregate spending. Sometimes so-called tax cuts and tax credits can reduce spending by discouraging work and redirecting economic activity to less-productive uses.

Even if a transfer from one group to another did increase aggregate spending and output, the second flaw of trickle-down theory is that the additional spending confers benefits beyond the direct beneficiaries of a transfer.

You might think that more spending on, say, groceries would benefit grocers and the farmers who supply them. It’s true that a grocer receives funds when a new customer comes to his register. But he also has to provide more groceries or have one of his other customers get by with fewer groceries, and groceries are not free to supply. The funds a seller receives from a new customer may just offset the total costs of the goods provided to the customer, so the seller is hardly better off.

A more nuanced analysis might assume that customers are normally charged more than cost (broadly defined) for the goods they receive, so that sellers are a little better off when aggregate spending increases. It might also consider that some industries are subject to economies of scale, so that other customers are better off when a new customer enters the market. Industries (and the workers they employ) sometimes pay more taxes when they expand, so the public treasury benefits when there’s more aggregate spending (the so-called Laffer curve is a special case).

None of this implies that a good public policy cannot transfer from one group to another. Transfers may intrinsically be good: we may enjoy helping the poor or value the freedom associated with limited taxation. Or the direct beneficiaries from redistribution may gain much more than the rest of us lose.

But until we know more about the magnitude of the second- and third-round effects of transfers on people who do not receive them, redistribution cannot be justified as helping those who finance it.

Quiet Time at Supply and Demand

The blog has been quiet for the last couple of months -- not even copies of my weekly economix posts. Loyal readers, please accept my apologies. I have been very busy finishing my book.


I hope and expect that the book will offer you more enjoyable and user-friendly reading about supply and demand than the blog posts it displaced. Once the book is out I can return to more frequent blogging and solicit your opinions on the next book!

Food Stamps: Married People Need Not Apply

Copyright, The New York Times Company

A fundamental difference between unemployment compensation and the food stamp program is their treatment of household income. The result is that food stamp participation among married people is relatively rare.

Traditionally, the food stamp program (now called SNAP) was for the poor: participants had to demonstrate that their incomes and assets were both low, and the program was financed out of general revenue. Unemployment compensation was an insurance program, financed from contributions by participating workers and awarded regardless of one’s assets or the earnings of other household members.

Today, the programs have converged in many ways. Much unemployment compensation (namely, the extended and emergency payments made to people whose unemployment has lasted more than 26 weeks) is financed by the federal government. Most states admit participants into SNAP using “broad-based categorical eligibility,” which (with the exception of three states) means that assets are not checked. For a few years unemployment benefits were available even for those who had been collecting for almost two years; food stamp participation can continue indefinitely.

But an important difference between SNAP and unemployment compensation has remained throughout the recession: the SNAP program checks the income of all household members and bases eligibility on the entire household’s income, not just the situation of the household head.

Naturally, children do not earn much income, so the practical result of the household income rule is that participation in SNAP among the unemployed depends very much on marital status. An unmarried, unemployed household head may have children in the family, but by definition has no spouse, and thereby is unlikely to have much additional family income.

In contrast, an unemployed person with a spouse earning, say, $30,000 a year, has little chance of participating in SNAP because the spouse’s income alone would most likely disqualify the entire household.

The chart below shows program participation rates among unmarried household heads. Using Census Bureau data, I identified household heads and spouses who experienced some unemployment during calendar year 2010 and classified them by their wages and salaries for the year. SNAP participation depends on income relative to household size – larger households need more food and therefore can participate in the program at somewhat higher incomes – so I expressed each person’s wages and salaries relative to the federal poverty guideline for households their size. For example, a value of 1 on the chart’s horizontal axis represents people whose wages and salaries were equal to the federal poverty guideline for their household, which was $22,050 for a family of four.

The red line in the chart is SNAP participation rates for unmarried people relative to married people. The blue line is unemployment insurance receipt for unmarried people relative to married people. For example, for household heads and spouses experiencing unemployment and earning no wage or salary income in 2010, the rate of receiving unemployment insurance was about the same for married and unmarried people: a participation rates ratio of about 0.9. But the SNAP participation rate was twice as great among the unmarried.

For the other earnings categories, unmarried SNAP participation rates range from 1.5 to 2.5 times what they are for married people.

The bottom line is that SNAP is largely a program with unmarried participants. SNAP may be an effective way of feeding people, but its low participation among married people may make it politically less popular than unemployment compensation.