Must-Read: Alisdair McKay and Ricardo Reis: Designing effective automatic stabilisers of the business cycle

Posted on by Brad DeLong

Must_Read: Alisdair McKay and Ricardo Reis: Designing effective automatic stabilisers of the business cycle:

Brexit has raised the possibility of a recession on both sides of the Atlantic….

Unable to use traditional remedies like monetary or fiscal policy stimulus, policymakers may consider automatic fiscal stabilisers…. The social insurance system stands at the centre of automatic stabilisers, yet we still know too little about how it affects the macroeconomy (Blanchard et al. 2010)…. We focus on two key programmes: a progressive tax system, and unemployment insurance.  We find that unemployment insurance has a substantial stabilising effect on the business cycle, and as a result the optimal policy is to have a more generous unemployment benefit than would be optimal in a world without macroeconomic fluctuations.  On the other hand, the progressivity of the tax appears to have little effect on the business cycle….

[The] redistribution channel based on marginal propensities to consume is the classic explanation of how unemployment benefits function as an automatic stabiliser, we find that benefits have an even more important effect on the economy. Households face considerable uncertainty over employment and over the wages they earn when employed. The fear of losing their income because of an unemployment spell in the future leads workers to want to save, and so reduce their consumption.  That is, unemployment reduces aggregate consumption demand not just by reducing the current income of unemployed workers but also by inducing precautionary savings by all workers who fear the possibility of unemployment.  By making unemployment less painful, unemployment benefits reduce these concerns and give workers confidence to spend….

When we quantify our model, we find that business cycle stabilisation considerations increase the optimal unemployment benefit replacement quite substantially.  We find that in the optimal unemployment insurance replacement rate is 49%, while it would be 36% in the absence of business cycles…. The unemployment insurance system is very effective at stabilising the business cycle by dampening the cyclical swings in household precautionary savings motives…. Recessions have large welfare costs due to the increase in uninsurable risks that households face.

While similar arguments can be put forward for the benefit of more progressive income taxes, both in terms of redistributing resources in a recession and by lowering precautionary savings, progressive income taxes have a stronger negative effect on average economic activity and a more limited effect on volatility, because of the comparatively small variation in income of those continuously employed during the business cycle. Therefore, we find that considering automatic stabilisation has a negligible effect on the desired level of progressivity.

Must-Read: John Taylor (2000): Reassessing Discretionary Fiscal Policy

Posted on by Brad DeLong

Must-Read: This looked very good at the time. (I should know: I commissioned it, and strongly argued through the editorial process that it was an essential thing to publish–and Alan Krueger, Jim Hines, and Tim Taylor were more than eager to concur.) But today it looks a lot less smart:

John Taylor (2000): Reassessing Discretionary Fiscal Policy:

Recent changes in policy research and in policy-making call for a reassessment of countercyclical fiscal policy…

Countercyclical fiscal policy should focus on automatic stabilizers rather than discretionary actions. Monetary policy has been reacting more systematically to output and inflation; long expansions in the 1980s and 1990s demonstrate policy effectiveness. It is unlikely that discretionary countercyclical fiscal policy could improve things, even with less uncertainty about fiscal impacts. A discretionary countercyclical fiscal policy could make monetary policy making more difficult. Discretionary fiscal policy should focus on long-run issues, such as tax reform and social security reform.

Taylor, John B.. 2000. ‘Reassessing Discretionary Fiscal Policy.’ Journal of Economic Perspectives, 14(3): 21-36.
DOI: 10.1257/jep.14.3.21

Question: if you want to assign–and we may well want to–responsibility for stabilization policy to central banks, what passive support do they need more than our current automatic stabilizers and what additional active tools do they need in order to make it work well?

Posted in Uncategorized

Must-Read: Jesse Rothstein: The Economic Consequences of Denying Teachers Tenure

Posted on by Brad DeLong

Must-Read: Jesse Rothstein: The Economic Consequences of Denying Teachers Tenure:

There’s just not actually a long list of people lining up to take the jobs…

If you deny tenure to someone, that creates a new job opening. But if you’re not confident you’ll be able to fill it with someone else, that doesn’t make you any better off. Lots of schools recognize it makes more sense to keep the teacher employed, and incentivize them with tenure. It’s all about tradeoffs. If you get rid of tenure and start laying off lots of teachers, and you don’t do something else to make the job more attractive, then you won’t get enough teachers… basically economics 101…. You get more people interested in teaching when the job is better, and there is evidence that firing teachers reduces the attractiveness of the job….

During the Vergara trial there was a debate over what would happen if we lengthened the probationary period, and whether principals would wait until the end of the clock to fire a bad teacher, or if they would they fire a lot of bad teachers at the end of their first year. I argued that it was unreasonable to expect the principals to make the firing decisions before they had to. It’s a lot harder to say, “I’m confident this teacher wouldn’t work out” than it is to say, “Well let’s just give it a little more time.”

Must-Read: Nick Bunker: Why Slightly higher Inflation Might Be a Benefit

Posted on by Brad DeLong

Must-Read: A somewhat puzzling finding by Nakamura et al. that higher inflation–at least in the sub-10%/year environment of the post-WWII United States–does not degrade the functioning of the price system by increasing the spread of real prices. If this holds up, then the only remaining cost of steady, moderate inflation is having to change your prices more frequently. And even though are models tend to say that should be a heavy burden, I find that very unlikely: whatever the reason for fixed, posted prices is, I cannot see it having the implication that it is costly to change them to compensate for ongoing inflation once a year rather than once every three to five years.

Nick Bunker: Why Slightly higher Inflation Might Be a Benefit:

The U.S. Federal Reserve Board hasn’t hit its stated inflation target of two percent in more than four years….

With the central bank’s benchmark Federal Funds interest rate at close to zero, perhaps it’s time to rethink that target rate. Why not aim for a higher inflation rate of, say, 4 percent, which would allow inflation-adjusted interest rates to go even further below zero  to help boost economic growth during a possible future downturn?… Emi Nakamura [et al.]… built a brand new dataset on inflation before tackling a very important question on the costs of inflation….

Nakamura [et al.]… find… that the absolute price changes don’t vary that much regardless of the inflation rate. This means it doesn’t look like higher rates of inflation cause more price dispersion…. [Instead,] price hikes happen more frequently when inflation is higher, which gives more credence to “menu-cost” models of price setting…

Credible research designs for minimum wage studies

Posted on by David Evans

The employment consequences of increasing the minimum wage in the United States continue to be a major subject of debate, but how researchers choose to estimate the effects of raising the minimum wage can substantially affect the results of their work. In “Credible research designs for minimum wage studies”,1 my coauthors and I examine one group of low-wage workers—teenagers—whose hourly wages are significantly raised by minimum wage increases.

Download File
Issue Brief: Credible research designs for minimum wage studies

A common objection to raising minimum wages is that doing so will reduce the employment opportunities of low-skilled workers such as teenagers. We show, however, that some studies find negative effects of the minimum wage on teen employment because they fail to control for other economic factors that independently reduced employment around the time of a minimum wage increase. After controlling for these factors, we demonstrate that the large, negative effect on teen employment disappears.

State minimum wage increases are heavily concentrated in different regions of the country

After the frequency of federal minimum wage increases slowed during the 1980s and 1990s, many states began experimenting with setting their own wage floors above the federal level. By last year, more than half of all states had a minimum wage higher than the federal floor. This extensive history of state-level policies offers an attractive set of experiences for researchers to assess the effects of minimum wages.

Economists have developed a large body of research comparing the labor market outcomes in states that raise their minimum wage versus those that don’t. Yet a naive comparison of these two groups of states can lead to misleading conclusions because the variation of state-level minimum wage policies is not random (which is ideal for assessing the impact of government policies) and is instead geographically concentrated. (See Figure 1.)

Figure 1

My coauthors and I explain in our paper that the map divides states into two groups: states with high average minimum wages and states with low average minimum wages during the 1979-2014 period. States that have high minimum wages were more likely to have been raising their respective wage floors above the federal floor. States with low minimum wages typically followed federal policy. This difference is clearly region-specific.

This clustering of minimum wage policies within regions of the country is an obstacle for credible research on the minimum wage because comparing the employment of minimum wage-raising and non-raising states effectively compares regions such as the Northeast versus the South. Employment patterns differ in these regions because of a host of economic and political reasons not affected by the minimum wage. High minimum wage states, for example, also boast higher unionization rates and experienced smaller declines in unionization over the past three decades. They were much more likely to vote Democratic in the 2012 presidential election. In low minimum wage states, over the past 20 years, the foreign-born share of the population grew much faster.

In summary, there are reasons to be concerned that states that tend to raise their minimum wages have different employment trends than states that do not, irrespective of the minimum wage. Simply comparing minimum wage-raising and non-raising states can therefore give misleading estimates of the effects of the policy.

States where the minimum wage is high were experiencing employment problems even before minimum wages went into effect

Some minimum wage research does not adequately address the problems caused by the non-random pattern of minimum wage increases. In our paper my co-authors and I re-examine a key 2014 study, “More on Recent Evidence on the Effects of the Minimum Wage in the United States,” by David Neumark at the University of California-Irvine, Ian Salas at Johns Hopkins University, and William Wascher at the Federal Reserve Board.2 This study finds large, negative employment effects of the minimum wage on teenagers, a demographic group with a large share of minimum wage workers.

The methodology behind this study, however, also generates an implausible conclusion—that teen employment in high minimum wage states was falling in the years before the minimum wage was increased. The mistaken results of this study are a consequence of not controlling for the striking spatial pattern of minimum wage increases in different regions of the country.

One simple way to assess directly whether minimum wage-raising and non-raising states are comparable is to look at labor market outcomes before a minimum wage increase actually occurs. Borrowing from the language of randomized control trials in medicine, this can be thought of as a placebo or falsification test. We should not see the effects of drug before a drug is administered in a well-designed experiment. If we observe in a drug trial that the health of the treated group was improving relative to the control group even prior to taking the drug, then we should be hesitant to ascribe subsequent improvements in health to the causal effect of the drug itself.

Similarly, my co-authors and I examine the estimated effects of the minimum wage on teen employment in the years before and after minimum wage increases. We compare the results from the approach favored by Neumark, Salas, and Wascher to the approach we favor. First, we estimate the effects of the minimum wage using the preferred model of Neumark, Salas, and Wascher, in which any state is on average a good control for another state in the absence of minimum wage increases. Second, we consider the model that we prefer, one that controls for the non-random geographic pattern of minimum wage increases, allowing states to have different teen employment trends and restricting comparisons to nearby states.3

With these two models in hand, my co-authors and I calculate the elasticity of teen employment with respect to the minimum wage—specifically the percent change in teen employment in response to a 1 percent change in the minimum wage. Our findings show that the minimum wage did not lower teen employment in our preferred model (with rich spatial controls) which passes the falsification test. (See Figure 2.)

Figure 2

The model preferred by Neumark, Salas, and Wascher shows large, negative effects of the minimum wage at the time of the minimum wage increase and afterwards (years 0 through year 3, in the left panel of Figure 2). At the same time, the model decisively fails the falsification test. Teen employment in minimum wage-raising states is already low in all three years prior to the minimum wage increase (years -3 through -1). For every year prior to the minimum wage increase, their model produces a statistically significant negative employment effect—even though no minimum wage increase was enacted in those years.

This violation of the “parallel trends” assumption—that states on average have similar teen employment trends in the absence of minimum wage differences—is a clear sign not to trust the estimated employment effects of this model.

In contrast, my co-authors and I successfully control for the non-random pattern of minimum wage increases by using a model with a richer set of geographic controls (see the right panel of Figure 2). As is expected by a credible research design, there is not much change in teen employment in the years before the minimum wage increase. Teen employment elasticities during the pre-treatment period are generally small and are all statistically insignificant. In addition, there are no indications of negative employment effects in the years after the minimum wage increase. None of the point estimates are economically large or statistically significant by conventional standards.

Although these geographic controls are natural choices to account for the non-random spatial pattern of minimum wage increases presented in Figure 1, there are other research designs one could consider. In our paper, we present additional evidence from research designs that incorporate other controls for the non-random nature of minimum wage policies, such as limiting comparisons to nearby counties, or using estimators based on techniques that select alternative control groups. Many of these estimates also suggest small-to-no employment effects for teens.

As my coauthors and I explain, teenagers are a small share of the total workforce, but the studies of this group are nevertheless informative for understanding the overall employment effects of the minimum wage. If there were widespread employment losses due to the minimum wage, then we would likely see them among low-skilled groups such as teenagers. Yet the evidence we present in our paper suggests that teenagers did not suffer job losses for the kinds of minimum wage increases typically experienced in the United States over the past 35 years.

(Photo by Adam Croot, via Flickr)

Must-Reads: August 17, 2016

Posted on by Brad DeLong

Should Reads:

Would cutting the U.S. corporate income tax boost wages significantly?

Posted on by Ed Paisley

In a column for the The Wall Street Journal earlier this week, economists Kevin Hassett and Aparna Mathur of the American Enterprise Institute offered a solution to wage stagnation in the United States: a cut to the U.S. corporate income tax rate. The crux of the argument made by Hassett and Mathur is that the incidence of the corporate income tax falls mostly on labor (employees) and less on capital (shareholders). But that’s not necessarily the view of many other studies on the subject. It’s not clear how much of a boost in wages most workers could get from a cut in the corporate rate.

Hassett and Mathur argue that the more the incidence of the corporate income tax falls on labor, the more of a boost in wages workers would get from a tax cut. They point to their own research as well as other studies showing that, despite its intentions, the corporate income tax gets paid by workers. Other economists disagree. Reed College’s Kimberly A. Clausing, in a 2012 paper, raises some questions about several studies finding that the incidence falls mostly on workers, noting that the theoretical work in the area has “failed to reach a clear consensus,” and that the empirical work “is sparse and suffers from essential limitations.” Her conclusion is that while there are some indications that corporate taxation reduces wages, the preponderance of evidence finds no effect at all.

In further thinking about who actually pays the corporate income tax, it’s worth looking at two big reasons why some economists think capital can skirt it. First, capital is far more mobile that labor. The thinking goes that if a country or state increases its corporate income tax, then capital can easily move to a new, lower-tax location and get a higher return there. This ability to leave ends up pushing the incidence onto workers who are far less likely and able to move locations.

Tax rates, however, are far from the only reason why corporations locate somewhere. California has a very high corporate tax rate for a state, but one out of nine establishments in the United States is located there. The reason is that firms often need or want to be close to each other. In the case of California, think of Silicon Valley. A company might find it worthwhile to stay in a location even if the tax rate gets hiked a bit. So firms are less mobile than some research papers and economist might think due to the benefits of location. The result is that capital would bear more of the tax burden.

At the same time, models of tax incidence often assume that the return on capital earned by shareholders are the “normal” return you’d get from a risk-free investment. But most of the profits that corporations earn and on which they are taxed are not from risk-free investments but rather from “excess returns,” or profits derived from riskier investments or companies’ market power in their respective marketplaces. (The discussion of the incidence of the U.S. corporate income tax on page 17 of this Congressional Budget Office report makes this point well.) A tax on these returns would be borne even more by capital, which means a tax cut wouldn’t lead to significant wage gains.

So a broad look at the evidence doesn’t deny that some portion of the U.S. corporate income tax is borne by workers. It just appears that it’s not a significant portion and therefore unlikely that a reduction in the tax rate would do very much to boost wages for U.S. workers.

(Photo by Susan Walsh, Associate Press)

Why slightly higher inflation might benefit the U.S. economy

Posted on by Ed Paisley

The U.S. Federal Reserve Board hasn’t hit its stated inflation target of two percent in more than four years, at least according to its preferred measure. One reason for this sluggish pace is the extraordinary decline in oil prices, but the Fed still isn’t seeing much process toward hitting its goal. So with the central bank’s benchmark Federal Funds interest rate at close to zero, perhaps it’s time to rethink that target rate. Why not aim for a higher inflation rate of, say, 4 percent, which would allow inflation-adjusted interest rates to go even further below zero  to help boost economic growth during a possible future downturn.

One reason this could be a problem is that many models of the U.S. economy say that higher inflation would deliver big distortionary problems. One of the large potential costs, price dispersion, happens because, as Noah Smith of Bloomberg View puts it, “when companies want to change their prices but for some reason can’t, inflation distorts prices from what they should be, which decreases economic efficiency.” A new working paper, however, looks at past experiences with these potential inflation-induced inefficiencies during a period of high inflation and find evidence that maybe there isn’t that much to fear.

Economists Emi Nakamura, Jon Steinsson, and Daniel Villar of Columbia University and Patrick Sun of the Federal Communications Commission built a brand new dataset on inflation before tackling a very important question on the costs of inflation. The economists painstakingly created a data set on individual price changes using old U.S. Bureau of Labor Statistics data. Creating these data allow the researchers to look at how prices changed during the late 1970s and early 1980s, an era of high inflation.

To look at how inflation could affect price setting and increase price dispersion, the four economists look at the absolute size of the price increase and its relationship to the pace of inflation. If prices are raised by a larger amount then price dispersion increases with inflation and therefore higher inflation would reduce efficiency. But what Nakamura, Steinsson, Sun, and Villar find is that the absolute price changes don’t vary that much regardless of the inflation rate. This means it doesn’t look like higher rates of inflation cause more price dispersion.

What the four economists do find is that price hikes happen more frequently when inflation is higher, which gives more credence to “menu-cost” models of price setting, or models where companies can’t easily change prices.  So if a major source of the cost of inflation in leading models doesn’t seem to show up in the data, perhaps economists and policymakers should rethink their devotion to near-zero inflation rates. Ten percent inflation is certainly too high, but a rate of 4 percent seems to have been consistent with steady economic performance in the past, such as in the 1980s, as well as provide more monetary policy firepower in an era of low interest rates.

(Photo by Charlie Neibergall, Associate Press)

Must-Read: Justin Fox: High-Tech Manufacturing Isn’t Worth Much

Posted on by Brad DeLong

Must-Read: Justin Fox: High-Tech Manufacturing Isn’t Worth Much:

These are the world’s five largest technology companies, ranked by revenue:

High Tech Manufacturing Isn t Worth Much Bloomberg View

How about that Hon Hai!?! The full name is Hon Hai Precision Industry Co., but you probably know it as Foxconn…. Foxconn is really big, and it’s about to get bigger after completing the acquisition Friday of once-great Japanese electronics maker Sharp. But before you get all excited (or worried) about tables turning and China rising, here’s how profitable those same five companies are:

High Tech Manufacturing Isn t Worth Much Bloomberg View

You know the story about Amazon.com… it’s really a consumer discretionary company, with peers such as Macy’s and McDonald’s… [and] Amazon has been investing more money in research and development than all but one or two other companies on earth–which is why I’ve included it here among the world’s tech giants, and also another big reason its margins are so low…. It just so happens that contract manufacturing of electronics is a really low-margin business. Foxconn’s profit margin is 3.1 percent; at the No. 2 contract manufacturer, California-based Flex (formerly Flextronics) it’s 2 percent. Foxconn’s acquisition of Sharp is an attempt to boost those margins…. Still, it’s fair to say that investors around the world aren’t expecting a whole lot from these efforts:

High Tech Manufacturing Isn t Worth Much Bloomberg View

Part of the difference in market caps may be… overvalued U.S. equity… that U.S. corporate executives care more about keeping investors happy…. But the main divide I see here is that Samsung and Hon Hai manufacture things… while Apple, Amazon and Microsoft all sell products with their brand names on them, they contract out all the manufacturing…. In 2009, Harvard Business School professors Gary Pisano and Willy Shih argued in the pages of the Harvard Business Review that technology companies that outsourced production were making a big mistake, and that the U.S. was being permanently harmed as high-tech manufacturing know-how moved to Asia. They may still turn out to be right about the geographical impact…. But for individual tech companies, getting out (or staying out) of the low-margin, high-investment business of manufacturing just keeps paying off.

Must-Read: Gavyn Davies: What Caused the Fed’s Dovish Turn?

Posted on by Brad DeLong

Gavyn Davies: What Caused the Fed’s Dovish Turn?:

The change in the Fed’s guidance about “normalisation”… came… [as] the dollar peaked (and equities collapsed) in February/March…

…a change in their expected mix between the exchange rate and domestic interest rates in delivering the tightening in financial conditions that they desired at the time. A higher dollar essentially forced them to accept lower interest rates in order to deliver roughly the same path…. Furthermore, this was not expected to be just a temporary shift…. The best analysis of this issue from inside the Fed has come from Lael Brainard… the importance of global deflation risks, especially in China. These risks will lead to a very long period of aggressively easy monetary policy outside the US, which in turn will make the dollar far more sensitive to US rate hikes than has been the case in some earlier periods. The Fed should not, she argues, ignore this when setting US rates: the equilibrium interest rate in America has been reduced by events overseas. This analysis lies outside the Fed’s normal comfort zone, and contradicts the standard analysis produced by Stanley Fischer among others last year. It was overlooked entirely by Ben Bernanke last week. But it does seem to have determined the behaviour of the FOMC for much of this year.