## Tuesday, August 29, 2017

### The Migration Question: A Highly Politicized Issue threatening to Split the European Union [feedly]

The Migration Question: A Highly Politicized Issue threatening to Split the European Union
http://www.globalpolicyjournal.com/blog/29/08/2017/migration-question-highly-politicized-issue-threatening-split-european-union

This is the second of a four-part series of posts on Poland's, Hungary's and the Czech Republic's perilous play with the EU's refugee relocation agreement.

This is not the first time the European Commission reverted to the instrument of the law infringement procedure in the field of migration policy. In 2008, the EU initiated a similar procedure against Greece. Athens was criticized for the lack of sufficient legal guarantees regarding the examination of asylum applications, which became a problem in particular after the transfer of asylum seekers from another member state to Greece under the Dublin regulation. The Dublin regulation stipulates that in principle asylum-seekers will be returned to the EU country in which they entered the EU territory for the processing their asylum application. Also, there was a similar procedure against Germany in 2012, as Berlin was dragging its feet with the implementation of the so-called "Blue-Card" directive, regulating legal third country migration into the EU. Both procedures were successful, as Athens and Berlin adopted the necessary legal adjustments and never questioned the legality of the procedure.

In contrast, the current infringement procedure is quite different, as it is of highly politicized nature. The defiant countries stress the oppressive nature of the relocation scheme on the one hand, and oppose the "EU dictating" and interfering with the realm of national sovereignty in principle. On the other hand, the procedure reflects the annoyance of Brussels over the slow response to the relocation scheme, not just in the Central Eastern European (CEE) area but in all EU countries. While the initial aim of 120,000 refugees to be relocated was not very ambitious, only 20,000 asylum seekers have been moved so far. Positions on the measure also differ among Western EU member states. As of 21 August 2017 Austria has relocated no refugees at all, France has still to take in 15,614 refugees (out of the agreed 19,714), Germany 20,146 (out of the agreed 27,536), Sweden 2,201 (out of the agreed 3,766) and the Netherlands 3,744 (out of the agreed 5,947). The reason for the lackluster response in all the EU has been attributed mainly to the presidential elections in Austria in October/December 2016 and its upcoming general elections in November 2017, the elections in the Netherlands in April 2017, in France in April/May 2017 and in Germany in October 2017, with the migration-critical parties in all these countries threatening to take over central political institutions such as the presidency (in Austria and France) or challenge the established parties (such as in the Netherlands and in Germany). Against this backdrop, the EU is well aware of the highly politicized nature of the relocation scheme and how Eurosceptic parties may capitalize on it in order to shift political balances to the right, i.e. towards nationalist and anti-EU positions in most EU member nations.

Nevertheless, the EU decided to make an example of Poland, Hungary and the Czech Republic which object the scheme completely, in order to "motivate" other countries to take the plan more seriously, in particular given that the scheme expires in September 2017. Not taking any steps would be acknowledging a serious political failure of the bloc. The EU has already a problem of credibility among its member states, as it cannot effectively move against Hungary and Poland concerning the rule of law violation in both countries, as for this the bloc would need an agreement of all member states (minus the countries in question) to impose concrete sanctions. Such an agreement is highly improbable.

Despite the infringement procedure, the CEE governments remain critical of the relocation scheme, stressing that it was never working due to its ill-conceived nature, and that it constitutes a violation of basic EU law. Firstly, according to Warsaw, Budapest and Prague, the scheme was politically "nonsensical" aggravating the refugee crisis, rather than solving it, since the EU's relocation plan would actually represent a pull factor, encouraging more migrants to come to Europe thus contributing to a probable collapse of the entire Schengen zone, i.e. the principle of border-free movement within the EU, since many countries would start to re-introduce border controls to prevent illegal mass-migration which continues to get 100,000s per year mainly to the shores of Italy across the Mediterranean. This argument highlights that the majority of the asylum-seekers from the MENA region coming to Europe in 2015 and 2016 were actually economic migrants targeting the wealthy welfare states of the EU such as Germany and Sweden, rather than the poorer ones such as Poland and Hungary. This analysis is sustained by the fact that less than a third of the 300,000-400,000 illegal migrants arriving in Italy per year submitted an asylum request at all since 2014. The migrants arrive with the help of a massive fleet of NGO ships, i.e. of private rescue organizations not controlled democratically nor by the European member nations nor the EU who are able to professionally transfer around 1000 illegal migrants per day across the Mediterranean, i.e. as a gigantic illegal "transfer industry" which according to the Italian authorities has been partly cooperating with human traffickers and is making a business out of mass-migration to Europe worth around 35 billion US$a year according to official estimates by the International Organization for Migration (IOM). Secondly, the three defiant countries point out that the EU decision of September 2015 to establish forced relocation to be imposed to its member states was illegal, as, among others, the Council applied a majority decision instead of unanimity and did not consult the European Parliament sufficiently. For years, the EU has applied a rule according to which with regard to highly controversial issues the member states would seek unanimity even though formally a majority decision might be possible. If, however, a controversial majority decision is enforced on others, then according to the V4 states it would equal "a tyranny of majority". Thirdly, Poland, Hungary and the Czech Republic argue that since the bulk of the refugees prefer the welfare state countries such as Germany, Italy and Sweden who give them immediate and unrestricted full access to all services and features without any previous contributions, they would prefer to leave the poorer ones such as Poland, Hungary and Romania after the resettlement anyway, and these countries would need to stop them against their will and thus violate the Geneva Convention. Fourthly, according to the V4 there is an increased threat of terrorism as a result of the relocation scheme, as, for instance, IS terrorists are known to infiltrate the migration routes particularly from Libya, Tunisia and Egypt across the Mediterranean to Italy and recruit young male refugees, many of whom come to Europe without any parental company. As the Italian authorities analyzed in July 2017, there were connections between terrorist organizations in Northern Africa and the Mafia criminal organization in Sicily and the Italian South who founded appropriate NGO's to host the migrants in its own structures paid for by the state, i.e. making a business out of the Italian law that guarantees every migrant 35 Euro per day, i.e. 1000 Euro per month which is more than many Southern Italian pensioners get after 40 years of work. According to the International Organization for Migration, illegal mass migration has become the third most profitable, but much less dangerous business for the Mafia after the smuggling of drugs and weapons. Italian conservative commentators have long speculated that while most bigger other EU nations were hit and although Italy accepts the largest numbers of migrants by keeping its sea borders open and as the only country which offers its ports to the NGO ships carrying mass migrant waves, there were no terror attacks in Italy exactly because due to this policy Italy serves as an easy entry and exit area from and into the EU for fundamentalists not interested in putting the status quo at risk by attacking the Italian peninsula. That is why the relocation plan, according to Warsaw, Budapest and Prague, is dangerous and could endanger the national security also of their countries. Ireneusz Pawel Karolewski, Dr., is Chair of Political Science at the Willy Brandt Centre for European Studies of the University of Wroclaw, and Associate Professor of Political Science at the University of Potsdam. Author of "European Identity Revisited: New approaches and recent empirical evidence" (Routledge 2016). Contact: karole@uni-potsdam.de. Roland Benedikter (corresponding author), Dr. Dr. Dr., is Research Professor of Political Analysis in residence at the Willy Brandt Centre for European Studies of the University of Wroclaw, and Global Futures Scholar at Eurac Research Bolzano-Bozen, Autonomous Province of South Tyrol, Italy. Contact: rolandbenedikter@yahoo.de. Part 1 is available here -- via my feedly newsfeed ### Jared Bernstein: #Fiscal_Friday: Three fiscal points to absorb before you start your weekend [feedly] #Fiscal_Friday: Three fiscal points to absorb before you start your weekend http://jaredbernsteinblog.com/fiscal_friday-three-fiscal-points-to-absorb-before-you-start-your-weekend/ –Deficit neutral vs. revenue neutral: This is really important and while the distinction is straightforward, I'm seeing some confusion out there. The R's get this and they're clearly, though subtly, shifting their rhetoric from rev neutrality to deficit neutrality. To pass their tax cuts under budget rules that allow them to evade a Senate filibuster, the cuts cannot raise the deficit outside the 10 year budget window. There are two ways to do so. One way, revenue neutrality, is to raise other revenues through base broadening to offset the cost of the rate cuts. The other way to cut taxes and hold the deficit constant is to cut spending. That's deficit neutrality, and since federal taxes and spending are both progressive, cutting spending to pay for cutting taxes is doubly regressive, as I argue here. I've also argued that, given our demography, climate change, inequality, geopolitics, infrastructure needs, etc., we're going to need more, not less (and not even the same amount of) revenues going forward – at least, as long as we boomers remain in the system. So revenue neutrality is, in my view, already too low a bar. But deficit neutrality is no bar at all. –A quick point about the debt ceiling: If you read this blog, then you surely know this already. But just in case, here it is: Congress must raise the debt ceiling to pay for spending they've already approved. As the date that the ceiling must be raised approaches, you'll hear some Republicans argue that raising the debt ceiling is fiscally irresponsible and just encourages more spending. That argument is perfectly analogous to this one: paying your restaurant bill after you've eaten is irresponsible and just encourages more dining out. –This one is particularly technical, but there's a budget scheme R's are flirting with that could artificially make their tax cuts look$440 billion cheaper than they really are. The jargon here is that in scoring the cost of their cuts, they're talking about using a "current policy baseline" versus a "current law baseline." The reason they'd do so is so they don't have to come up with a bunch more revenue to pay for expiring tax breaks that they plan to keep in place. Simply assume, contrary to the law, that the cuts remain in place and you don't have to raise more taxes to pay for them. CBO won't buy it, but if they try to pull this off in their own scores, we'll be blowing the whistle.

Here's the explanation from my CBPP colleagues:

That \$439 billion figure represents the cost of extending, through the next decade, dozens of corporate and individual tax provisions that would otherwise expire or have already expired under current law.  A current law baseline reflects their scheduled expiration, so proposals to extend them would cost money.  A current policy baseline, by contrast, assumes that they will remain in effect indefinitely, so proposals to extend them would not lose revenues.  That could help a Republican tax bill appear to be less costly than if it were assessed relative to current law — and could be crucial for hitting a desired revenue target, particularly one that facilitates the use of a privileged, fast-track status in the Senate that would allow the bill to pass without any Democratic support.

-- via my feedly newsfeed

### IMF Blog: Off the Charts: Your Favorite 5 Charts [feedly]

Off the Charts: Your Favorite 5 Charts
https://blogs.imf.org/2017/08/28/off-the-charts-your-favorite-5-charts/

-- via my feedly newsfeed

By IMFBlog

August 28, 2017

(photo: iStock by Getty Images).

Much as sailors use nautical charts to determine their location at sea, economists use charts to show who we are, where we are, and where we might be going.

In the Spring, we began our Chart of the Week feature on the blog: snapshots in time and over time of how economies work to help illuminate the uncharted waters ahead for the global economy.

Here are our top five charts of the week, based on readership:

### Rising Markups and Falling Productivity

A bit wonkish -- but this may be a key part of the slow productivity/ slow growth puzzle. "Rising Markups" == monopoly rents!

Rising Markups and Falling Productivity

In between checking weather updates here in Houston, I finally got around to reading the new working paper by Jan De Loecker and Jan Eeckhout (DE), on "The Rise of Market Power and the Macroeconomic Implications". The authors find that the markup of price over marginal cost charged by public US firms has been rising steadily since 1960, and in particular after 1980. It has generated a lot of commentary already, as in this post by Noah Smith, this post by Karl Smith, and this from Tyler Cowen. A lot of the discussion so far has been about how to interpret their results, and whether it indicates a rise in "market power", and that devolved somewhat into a discussion of how to define "market power" in the first place.

I want to jump in with some of my own thoughts on the paper, but I'm going to avoid getting into this "market power" interpretation thing. I'm more interested in what DE's results have to say about the productivity slowdown. Let's start with how DE get these numbers in the first place.

## Estimating markups

Here's the thing; you can't observe these markups. While the price of a specific good might be knowable, the marginal cost of that good is not. The average cost of producing that good - total costs divided by number of goods sold - is not the same thing as the marginal cost. I'd be willing to bet that a lot of firms would be hard-pressed themselves to tell you the marginal cost of an additional good. And that doesn't even get into the issue with multi-product firms, where their aggregated financials cannot tell you about the specific prices and costs associated with specific products.

So when DE are talking about estimating markups, they are extracting the markup of a firms "price" over the firms "marginal cost", where the price is some agglomeration of the prices of their various products, and the cost is some agglomeration of the cost of their various inputs. That's not a knock on the approach, but it should be kept in mind.

How do you extract this markup if you don't have the exact costs, or even prices? DE assume that firms are cost-minimizers, a less forceful assumption than strict profit maximization. Whether firms are pricing their goods to maximize profits or not is immaterial to DE, but they assume that firms are trying to produce their goods or services using the lowest-cost combination of inputs.

You can see the logic involved in the firm's decision pretty easily. If a firm's markup is the price relative to marginal cost, then for the marginal good it produces, it should be that

Rev=Markup×MC×InputsRev=Markup×MC×Inputs

where RevRev is the revenue from that marginal good, MarkupMarkup is the … markup, MCMC is the cost of an additional bundle of inputs, and InputsInputs are the bundles of inputs it takes to produce that marginal good. Simply rearrange this to

Markup=RevMC×InputsMarkup=RevMC×Inputs

and you've got it. DE have data on firm-level revenues, and they have data on firm-level costs for inputs. There is an extra step, which is that this kind of condition should hold for each separate input (i.e. labor and capital) as opposed to "Inputs" the way I've described it. All this does is insert a parameter into this expression that weights revenues by how important that specific input is to production. If we were working with labor, for instance, we might say that the revenue attributable to laboris only 60% of total revenues, and so divide 0.6 times revenues by the cost of labor to arrive at the markup. For the purposes of this post, though, I think you can easily stick with the idea that markups are obtained from ratios of revenues to costs.

DE do this for all the public firms they have data for, for all the years running from 1960 to recently. The figure below is the central result, which is that markups rose steadily from 1960's - when they were about 1.18 on average - to today, when they are about 1.67 on average.

This is one of those papers where it shocks you that no one has done this before. This paper takes some relatively off-the-shelf techniques, and uses them to generate an incredibly interesting set of estimates. Granted, the authors are some of the people who put those techniques up on the shelf in the first place, but the innovation in this paper is the application to the long panel of US public firms, not the techniques. The results in this figure are what have generated a lot of commentary.

Let's start with the big empirical caveats. If you want to disbelieve DE's results, then you'd have to think hard about the effect of these:

1. The sample is only public firms. Probably the biggest question mark. The average markup charged by all firms in the economy may not have changed much over time, but if there was a distinct change in the choice to either go public, or stay public, starting around 1980, and that change was related to markup sizes, then this could just be sample selection. For example, you could argue that as investors became laser-focused on quarterly earnings, this meant that the only firms willing to go public or stay public were those with large markups that could generate the expected earnings growth. All the low markup firms are private, and so don't show up in the DE dataset.

2. Industry production functions are assumed to be identical for all firms and over time. I mentioned above that DE actually use specific inputs in the calculation, and for that they need these weights (like the 60% for labor I mentioned). To get those, they estimate production functions for each industry using their data. And that estimation assumes that the weights are identical for all firms in an industry (i.e. American and United each have a weight of 44% on labor) and over time (i.e. United's weight on labor is 44% every year from 1960 forward). DE have an appendix where they relax this assumption, and show that the growth in markups over time (but not their level) is unchanged, so this is unlikely to be driving the results.

3. Cost minimization may not hold. This doesn't seem like a strong assumption, and it is certainly weaker than assuming full profit maximization. But it is an assumption. If the firms in the DE dataset are not operating to minimize costs (perhaps they just care about maximizing market share or executive compensation?) then the first order condition DE use would be picking up something other than the true markup. I don't have a great sense of how the bias would work here, or a great sense of whether this could be a big or a small problem. But it could be a problem.

If you are willing to take the DE results as indicative of what is happening in the US as a whole, then the second half of their paper is for you. Here, they break down several secular trends in the economy (e.g. declining labor shares, declining productivity growth) and show how the rise in markups can account for them. As the association of markups with productivity is something I've been banging around on for a few years - you could start with this post and work backwards - let me focus on that.

## Productivity Accounting

The presence of markups - and changes in markups - have implications for measured productivity growth. And DE corroborate that, while offering the data to be more precise about the effects. But they do it using a slightly different line of attack than I have in the past, so let me digress for a moment to try and link this all together.

There are two ways we could describe aggregate output. The first is using a production function, like

Y=TXY=TX

where TT is technology and XX is an index of inputs. Technology, T, is the for-realisies measure of how good we are at transforming inputs into useful output. It is what we really would like to measure. Based on this production function, we could back out technology as

T=YX.T=YX.

Leave that aside for the moment, we'll come back to it.

We can also describe aggregate output using incomes, meaning we can write

Y=Π+wXY=Π+wX

where ΠΠ are economic profits, and ww is the "wage" of inputs. For those wages, think of something like a weighted average of actual wages (to labor) and payments to capital and/or natural resource.

What are profits? Well, let's assume that there is a markup, μμ, which recall measures the ratio of price to marginal cost. If μ=1.5μ=1.5, then firms are charging 50% over cost. So profits are equal to 50% of costs. Another way of saying this is that profits are μ1μ1 times costs,

Y=(μ1)wX+wX=μwXY=(μ1)wX+wX=μwX

or total output can be expressed as the markup times the wage times the inputs used. We're still the world of pure accounting, by the way. No behavioral assumptions have been made.

Now, go back and compare the original production function version of output, Y=TXY=TX, to this markup version, Y=μwXY=μwX. From these, it shows us that we could write

T=μw.T=μw.

Or technology is equivalent to the markup times the wage. I like to think of this expression this way: technology (or an increase in technology) can be used for one of two things (a) raising the markup or (b) raising the wage paid to inputs. One quibble with DE is that they say something along the line of "if markups go up, this raises technology", which should more accurately be phrased "if markups go up, this implies that tecnology was higher", which eliminates the implicit causal statement.

So we've got two ways to backing out technology, T, from the data. Either use T=Y/XT=Y/X, or use T=μwT=μw. In terms of growth rates, the first says that the growth rate of technology (using little hats to denote growth rates) is ^T=^Y^XT^=Y^X^, and the second says that ^T=^μ+^wT^=μ^+w^.

That's great, but when we go to the data is it quite plausible that we make mistakes. That is, we go to the data and calculate residual productivity, P. And unless we are exactly right about how the world works, our measure of P may not be equal to T. Here's what I mean. Take the production function version of things, which I've talked about a lot before.

^P=^Y^Xwrong=^Y^X+(^X^Xwrong)=^T+(^X^Xwrong)P^=Y^X^wrong=Y^X^+(X^X^wrong)=T^+(X^X^wrong)

The growth rate of productivity is the growth rate of output minus the growth rate of a possibly mismeasured index of input use. This mismeasurement may be because we really don't know how to count units of capital or human capital correctly, or because we are weighting those counts incorrectly. The point I've made in a few posts in the past is that by ignoring markups, you could may be making a weighting mistake, and this would lead to ^Xwrong>^XX^wrong>X^, so productivity growth understates technological growth.

If you instead were going to try and measure productivity growth using the real wage, then you could get something like

^P=^μwrong+^X=(^μwrong^μ)+^μ+^X=^T+(^μwrong^μ)P^=μ^wrong+X^=(μ^wrongμ^)+μ^+X^=T^+(μ^wrongμ^)

where you might be making an error about how fast markups are growing. If you do, and in particular if you assume ^μwrong<^μμ^wrong<μ^, which you might do if you assumed markups were unchanging over time, then your productivity growth term understates technological growth.

And this is what DE essentially do in the very last part of the paper. They say that by falsely assuming markups are constant, productivity growth measured using changes in real wages is understating true technological growth. When they make an adjustment to allow for the real growth in markups they observe, they find that true technological growth was around 3-4% per year over the last few year, as opposed to almost 0% growth in measured productivity.

This is a massive difference, and suggests that (a) technological progress is clipping along very nicely, thank you, and (b) that almost all of the gains from this are being leveraged into higher markups rather than higher real wages to inputs like labor. Entirely consistent with stories of increased market power and concentration.

There isn't a glaring issue with the analysis that DE do here, but they could go further with it. The papers that establish the discrepancies between growth in productivity, P, and technology, T, point out that much of the problem stems from trying to do this at the aggregate level alone. If you had individual firm-level (or plant-level) data on technological growth, then you could add those up directly (suitably weighted by their size), to arrive at TT directly. No need to back it out from aggregate information on inputs used, or wages.

And DE have all the data to do this. When they estimate the production functions that allow them to recover the markups, they also recover an estimate of technology for each firm. It would be straightforward to find technological growth from the "bottom-up", and compare it to what we get using aggregate methods. This would give us (a) a true(r) measure of technological growth and (b) help identify what precisely we are doing wrong with the aggregate measures of productivity growth. I'll let DE add that to the near-infinite list of things that people have suggested they do with this data.

## Why are markups higher?

From the perspective of productivity, and for most of the other secular trends DE talk about, the source of the markup is irrelevant. Arguing about whether markups represent market power or not, or whether monopolistic competition is equivalent to market power, adds nothing to the discussion of the implications. They would matter if we were talking about how to change markups. But at this stage DE are simply documenting the growth in markups and showing that these can be associated with a number of phenomena.

As much of the discussion about the markups has circled around market power, let me point out two additional possible sources of markups, and growth in markups. First, scale. If there are increasing returns to scale, then the markup will be above one. Think of a firm with big fixed costs. They have to charge above marginal cost in order to recover those fixed costs. You could imagine that what has changed over time is that the firms in the DE dataset invested in some fixed costs that gave them increasing returns to scale, and the markups are covering those costs. The problem with this explanation is that you wouldn't expect labor's and capital's share of output to be declining over time in this case, which they are. So perhaps it doesn't matter much.

Second, it could be us. That is, the markup could be rising because consumers are getting more particular about their bundle of purchases. Whether these firms are true monopolists, or involved in monopolistic competition, or compete in some kind of strategic game with one another, the more inelastic our demand gets for their product, the bigger the markups they can charge. I think of this as a caveat of structural change. As our basic material needs get met more easily over time, we get pickier about the goods and services we consume, making our demand more inelastic. This could explain some of the rise in markups.

If you take seriously the DE findings, then decomposing this rise in markups into different sources is the natural next step. The other step is to find ways to confirm that this rise in markups is robust to non-public, and possibly non-US, firms. Tracking similar changes across countries would help eliminate or confirm some explanations (US-specific anti-trust policies). Nevertheless, I found the DE paper an intriguing (and very large) step forwards.

-- via my feedly newsfeed