Wednesday, April 11, 2018

Graeme Wheeler and the RBNZ: Policy Errors or Bad Luck?

Many people criticized the RBNZ ex-governor Graeme Wheeler for not bringing the CPI inflation “up” to hit the target. In other words reducing the OCR. Some went further to call for a negative OCR! 

Two things influence moving the CPI inflation to meet the target, policy and luck. By luck I mean the randomness of the shocks. Shocks are very difficult to identify ex-ante. I don’t know Graeme Wheeler, but examining the data indicate that perhaps he was mostly unlucky. 

The main job of the central bankers is to analyze the data, as they arrive, and try to identify the nature and the permanency of the shocks in order to make forward looking policies to offset their effects, which is a very difficult job. Failing to identify the shocks gives rise to ‘gut feelings’ and backward looking policies. Clearly luck plays a big role.

Soon after signing the first Policy Target Agreement (PTA) in 1990, inflation came down to hit the target much sooner that it was anticipated by anyone (back then the RBNZ experimented with instruments other than the setting the Official Cash Rate - OCR). One arguments could be that the inflation expectations adjusted downward quickly. Perhaps it was believed that the RBNZ was committed to achieving the job.[1] Once inflation expectations were anchored and the RBNZ held inflation around the target successfully it gained reputation.[2] These are valid arguments. Don Brash was a credible inflation fighter. He was fortunate too. No significant shocks occurred until 1997.[3] Even Milton Friedman told him that he is a fortunate man.

Regardless of its reputation and known commitment to keeping the CPI inflation on target, the RBNZ messed up a big deal in 1997, the Asian Financial Crisis. Of course no one saw the shock coming; the RBNZ reacted to it; and made a costly policy error (around 3 billion dollars of losses in GDP initially if I recall correctly, but I could be wrong). The year 1997 was both bad luck and bad policy.

In 1999 the RBNZ began setting the short-term interest rate (OCR) as a policy instrument. The next big shock was the 2007-2008 Global Financial Crisis (GFC). Monetary policy, which uses the interest rate (price) rather than money (quantity) has different dynamics. Quantitative Easing as a policy response in the US, the EU and elsewhere did not generate inflation. A successful anchoring of inflation expectations removes the positive correlation between CPI inflation and the growth rate of money. I illustrated that in an RBNZ discussion paper (Money in the Era of Inflation Targeting, 2001).[4] 

Simply modeled, the CPI is a weighted average of the domestic and foreign prices. Let the d Log CPI = a d Log P + (1-a) (d Log P*+d Log S), where (d) is the annual difference, P is the GDP implicit price deflator, P* is the US CPI, S is the exchange rate and (a) is a weight.[5] The RBNZ has no control over the exchange rate. Macroeconomic fundamentals cannot forecast the exchange rate because its fluctuations are dominated by randomness rather.[6] Therefore, any governor would find it difficult to push the CPI inflation up/down during times of extraordinary appreciation/depreciation of the currency.

Let’s look at the data. Unfortunately I do not have real-time data as Athanasios Orphanides taught us to do, but I do not think the exchange rate is subject to revisions anyways (price might be revised).

Figure (1) plots the actual d Log CPI (inflation) on the vertical axis and d Log P + (d Log P*+d Log S) rates of change on the horizontal axis.  The weight could be estimated, but it does not matter in this case.

The association between the CPI inflation, and the domestic and foreign prices is evident in the data until the end of 2009, then a clean break in the correlation occurred and remained until the end of the ex-Governor’s tenure.

I separate the variables to shed more light on the components.

Figure (2) plot the CPI inflation and NZ price of foreign exchange rate depreciation (P*+S). The picture is similar to figure (1). There is a break in the correlation in 2009. 

Figure (3) plots the CPI inflation and P, the domestic price inflation. There is a spike  in domestic price inflation in 2014. 

Figure (4) plots the CPI inflation rate and the US CPI inflation rate, and they seem to be correlated except for the period of the Great Recession in the US.

So it is the exchange rate! 

Given these large, unpredictable, and persistent foreign exchange rate shocks, I imagine that it was very hard to hit the CPI inflation target during that period even if the OCR was set to zero or even negative.

CPI inflation reached 1.9 in September and fell to 1.6 percent in December 2017 and these fluctuations are associated with highly unpredictable random changes in the exchange rate.

Some may argue that Graeme Wheeler could have done something about domestic price shocks, but I find it difficult to believe that he or anyone else could do anything about the exchange rate appreciation, which makes it very difficult to inflate and hit the CPI target. He (and the RBNZ) was more unlucky than anything else.

[1] There is a large literature on commitment.

[2] There is a large literature on reputation.

[3] Senior colleagues at the RBNZ told me that they (the RB) were targeting lower inflation a long time before the signing of the PTA in 1990.
[5] The variables are logs. I use the resorcinol of the exchange rate posted on the RBNZ webpage.

[6] The variance of the exchange rate depreciation rate is several folds larger than the variance of macroeconomic fundamentals.

Tuesday, March 6, 2018

A W H Phillips

A W H Phillips

This is my first blog for 2018, and it is about a New Zealander called Alban William Phillips, the famous author of the celebrated Phillips curve (the relationship between inflation and unemployment). I want to tell you a little story. It is not about the Phillips curve per se, but about the 'H' in the A W H Phillips Memorial Lecture at the New Zealand Association of Economists (NZAE). 

As the president of the NZAE in 2003, I wanted to establish an annual Phillips lecture. The idea was discussed and voted on at an NZAE meeting. The council decided to get the permission of Phillips family. My good friend and vice president then, Grant Scobie, suggested that we solicit the help of Professor Brian Silverstone of Waikato University to locate the family. Brian, so smart, found the Phillips sisters address from voting rolls and we wrote to them. From my distant memory, a short letter of two small-cut white papers arrived in the mail shortly, and I recall Grant Scobie reading the letter to the council in the following meeting.

The letter said that the sisters and Phillips daughter held a meeting and decided to accept the NZAE idea of establishing a memorial lecture in his name on one condition. The condition was that to add his middle initial 'H' to his name. No one knew about his middle initial then. We learned that the H stands for Housego, hence Alban William Housego Phillips. Professor Patrick Minford gave the NZAE first memorial lecture, and could be found online. Thanks to all those involved in setting up the lecture. I would also like to thank Professor Mark Holmes of Waikato University for correcting the record on the NZAE website.

Saturday, December 16, 2017

My Last Blog on Housing Prices in New Zealand, Australia and the US

Housing price increase and monetary impulse transmission mechanism explain the demand for housing and rising prices.

Economists in New Zealand recognize that the rising housing price is related to short supply of lands, zoning, council restrictions, etc. Thus the supply curve of housing is rather very steep (maybe vertical).

The demand curve for housing is downward slopping. When demand shifts up for any reason, the price of housing increases substantially because of the steepness of the supply curve.

Some people call the significant house price increase, a bubble, but it is just a natural result stemming from the steepness of the supply curve.

Others think that the increase in population growth, particularly because of immigration, increases the demand for housing, hence higher price. Not true, because the variations in total population growth are too small compared with the variations in housing price growth. Housing price variation is 14 times larger than population’s.[1]

Population growth cannot explain housing price.

The same is true in Australia; variations of housing prices 20 times bigger than population’s.[2]In the US, the variations is 24.75 times bigger than populations.[3]

The most likely shifter of the demand for housing (and demand for all assets), however, is an increase in the demand for money. The monetary policy regimes in New Zealand and Australia, among many other developed nations, ignore money totally. The Reserve Banks control the short-term interest rate so when the demand for money increases the central banks supply the required amounts at a particular interest. The story ends here for them.

In fact, the story does not end there in the market. The increase in the demand for money demand affects relative assets prices because it changes the marginal utility of money relative to the marginal utility of all assets, thus relative prices must change. This is the standard Portfolio Theory transmission mechanism, which has been totally ignored by central banks. Once relative prices changed people adjust their holdings of money, bonds, and other assets. This is a continuous dynamic process to establish equilibrium in the money and asset markets, and eventually money has to be willingly held.

With inflation and interest rates low, the cost of holding money is very small. Money demanded and supplied must be spent on the purchases of assets and goods, hence high prices.

The variation in money matches the variation in housing prices in New Zealand and in Australia and the US.[4] The scatter plots below show the relationships between money growth (Red), population growth (Blue) and house price growth. The variations along the 45 degree line clearly favor money as a likely driver of high demand for housing rather than population.

The same is true for Australia and the US.

In 2013 I wrote about Macro Prudential policy. One of my arguments to dealing with rising house price was to stick to monetary policy rather than venture into regulations. To do so I suggested a new price index, which includes house prices. It could also include other asset prices. The central bank could target that index because asset prices respond to monetary policy. Interest rate could reduce demand for all goods, services, and assets. The current inflation targeting regime does not recognize that a random increase in the demand for money increases demand for assets such as housing.

My paper (with Moosa) -Monetary Policy, Corporate Profit and House Prices - is  forthcoming in Applied Economics.

[1] Using quarterly data from Mar 1991 to Jun 2017, the standard deviation of population growth is 0.001385 while that of housing price is 0.019592.

[2]  For data from Dec 2003 to Jun 2017, the standard deviation of housing price growth is 0.018 compared with 0.0009 for population growth.

[3] For data from Mar 1975 to June 2017, the standard deviation of housing price growth is 0.0124 compared with 0.0005 for population growth.

[4] The standard deviation of M1 growth and housing price growth in New Zealand are almost equal, 0.024 and 0.019 respectively. For Australia, they are 0.02 and 0.018 respectively. For the US, they are 0.0143 and 0.0124 respectively

Friday, October 20, 2017

Immigration and productivity in New Zealand

Productivity growth is a complicated process.

Differences in the accumulation of the factors of production (capital and labor) in different developed countries cannot explain the differences in output per person growth rates. We have no evidence to the contrary. The same is true for intangible capital such as human capital or other forms of capital.[1]

So if economists are unsure about the explanation how would politicians make policy about “closing the productivity gap”?

Generally speaking, nations are rich and prosperous –countries endowed with natural resources notwithstanding – if people produce more goods per hour-worked.
Producing the goods is only half the problem though. People have to sell the goods. New Zealand is not a big domestic market so we have to export our products to other markets.

To compete with other countries in the global market we not only have to sell more goods, we have to sell new variety and quality goods.

We don't know how to measure improvement in quality very well. With respect to variety, however (a very few exceptions notwithstanding) we have not really produced a lot. We produce and export unskilled-intensive commodity-dominated goods. Milk remains milk, timber is timber and lamb is lamb, and neither is Samsung or an IPhone…or seedless melon etc. you get the point.[2]

In addition to the quality and variety of new goods, growth also requires cost reduction. All modern growth models, which can generate growth require increasing returns to scale in the factors of production and technological progress.

Production of new, quality and variety goods at lower cost is simply what Total Factor Productivity (TFP) is all about. It is the key to explaining income per person and we do not really know a lot about TFP, do we? The neoclassical growth model says nothing (The Solow residuals).

The strongest available evidence is for investments in R&D as the driver of endogenous growth. TFP growth requires a big monopolistic R&D sector. We don’t have such sector. The New Zealand economy is made of small firms.

One would think that in the absence of an R&D market the government could play a role. In my research with the late Robin Johnson and Steve Stillman we studied R&D stock in New Zealand over a period of 40 years and found no evidence of positive spillover from public R&D investments.[3] Successive governments over four decades produced no effect. Why should we believe that the government could do anything differently?  

Sixty years ago, the late Nobel Laureate Simone Kuznets argued that, under the assumption of increasing returns to scale, per capita output growth is tied to population growth in the long run.[4] He cited different effects arising from natural population growth, immigration, and decreasing death rates. His idea, although mathematically valid and easy to prove, was controversial then, but it is less so today because we have endogenous growth models with scale, and increasing return to scale production function, which have good empirical validity.

Kuznets calls one of the connections between population growth and productivity growth: useable knowledge. To produce these new goods (e.g., the IPhone), a country needs to exploit and use the knowledge created in the advanced world.

This knowledge, “the effective advanced-world research efforts,” measures the time and efforts scientists, inventors, innovators in these countries etc. put into producing it.

The larger the population (in advanced and developed countries) that higher is the probability that more people will be working in the production of knowledge.

So TFP growth in New Zealand or any other developed country is a function of the growth rate of effective world research efforts, which essentially explains output per capita growth.

In the long-run, output per capita growth in any developed country is a function of the population growth in the advanced world. Generally speaking, immigration is just one way to increase population, the labor force and the probability of more R&D.  

That said, the advanced countries today have declining working age populations, stagnant or declining fertility rates, declining youth population, and more aging population. These trends seem to be associated with long-run declining productivity growth, which we observe in the data currently.

Along the transitional path between now and the long run, productivity growth in the developed countries is slowing because TFP growth is slowing, and TFP growth is a function of a slowing growth of effective world research efforts. These declining efforts imply that scientists, innovators, researchers etc. allocate less time to the production of knowledge for the same reasons workers decide to work less hours. The supply of labor depends on taxes, the consumption-output ratio, the share of capital in production, the relative value of leisure, and demographic among many other variables.

Here is the trend in the growth of research efforts.[5]

The conclusion of this story is that New Zealand productivity growth is bound to be relatively low if NZ does not exploit/use the effective world research efforts or/and if the knowledge diffuses at a slower rate, or/and if the productivity growth in the developed countries is declining.

But why should we exploit more usable knowledge or have the frontier technology and best practices diffuse faster? We don’t produce and export new varieties of high quality skill-intensive goods at low cost, do we?  

According to this model, New Zealand productivity will not improve from having more immigrants, no matter how skilled they are, if we continue to produce and sell unskilled-intensive goods (milk, timber, meat, etc.). There are difficult structural issues pertinent to our production that constrain our productivity growth, which need to be resolved first.

[1] Prescott, E. C., Needed: A theory of Total factor Productivity, International economic Review, Vol. 39, Issue 3, (August 1998), 525-551.

[2] Before joining the EU, the U.K. bought all New Zealand’s farm products so competition was not an issue.  And I say seedless melon because I saw quality Australian seedless melon selling for about 80 New Zealand dollars a melon in the Middle East.

[3] Johnson, R., W. A. Razzak, and S. Stillman, Has New Zealand benefited from its investments in research and development, Applied Economics, 2007, 39, 2425-2440.  

[4] Simon Kuznets, Population Change and Aggregate Output in Demographic and Economic Change in Developed Countries, Universities – National Bureau, Columbia University, 0-87014-302-6, 1960, 324-351.
[5] Research effort is the sum of the product of (weighted) stock of human capital and the number of researcher.

Friday, October 6, 2017

Do we need more or less immigrants in New Zealand?

Here are four self-explanatory graphs.

Working age population (15-64 years old) levelled then began falling. I think it will be catching up with fall observed in OECD countries, which began earlier.

Falling in the number of young people (less than 15 years ole).

Falling, followed by a flat fertility rate (child per woman).

Increase in the number of old people (older than 64).

The answer should be clear.

Wednesday, August 16, 2017

The New Zealand Election and Monetary Policy

The Labour Party wants the RBNZ to target unemployment in addition to inflation. No one likes unemployment, but this does not make it a good policy change. It has been used before and failed everywhere.

Most probably, the policy meant to target the unemployment rate in the short-run (over the business cycle or maybe over the election cycle).[1] A reduction of the short-run unemployment rate below the Natural Rate of Unemployment (NRU) requires the RBNZ to generate higher inflation. According to the Phillips curve, there is a negative relationship between wage inflation and unemployment. The celebrated Phillips curve was first introduced by a Kiwi, Bill Phillips in 1958

Politicians liked that relationship because it meant that they could trade-off more inflation for lower unemployment. Unfortunately, it is not that simple. It is a textbook economics that the Phillips curve shifts up and down every time the expectations about the inflation rate change. And, what matters for work is the real wage rate – the wage adjusted for inflation – not the nominal wage; in other words, the purchasing power of the money wage.   

The unemployment rate changes in the short run: (1) because firms hire labour as long as the real wage paid to workers is below labour productivity (the marginal productivity of labour); stop hiring when the real wage rate is  equal to labour productivity; and layoff labour when real wages increase above labour productivity. And, (2) because workers accept job offers when the real wage rate is higher than the reservation wage, and don’t when it is lower (here). The reservation wage depends on the generosity of unemployment and other benefits.  

The RBNZ comes into the picture because it can affect real wages via expected inflation. Workers and employers take this expected inflation rate to calculate the purchasing power of the offer (i.e., the real wage).

This policy will not work according to the traditional Phillips curve because the relationship between inflation (money wages) and unemployment is unstable, especially in New Zealand because the RBNZ has successfully removed the trend from inflation (stationary).  The plot (RBNZ data) shows no negative relationship to be exploited.

The relationship between inflation and unemployment changes overtime. It has been observed to be positive, or even zero. Therefore, there is no guarantee that the RBNZ can reduce unemployment in the short run by increasing inflation; quite the opposite could happen. The U.S. and most of the industrial nations followed such policy before and those of who remember the 1970s and the 1980s must remember the double digit inflation rate and the devastation that followed.

So how will Labour design the new PTA policy?
There will be no inflation target as we know it today because they have to inflate the economy to reduce unemployment.

First, how? They either reduce the interest rate, perhaps making it negative as other central banks have done. Or, use quantitative easing, i.e., pump more money into the system by buying securities. Here is the recent IMF view on the effect of negative interest rate on the banking system and financial stability.

I would add that inflating the economy would impact the housing market quite considerably. More money, more inflation, more spending and more demand for housing. That would make the job of the new governor very difficult because the RBNZ also regulates the financial system and formally cares about stabilizing housing prices.

Second, by how much they have to increase inflation, and could they? It will have to be a lot given that monetary conditions defined by the current exchange rate and credit restrictions seem tight.

That said, it might be worse if Labour is thinking about the long-run unemployment. In this case they have an old and a dead idea. The government must come up with an estimate of something called the Non-Accelerating Inflation Rate of Unemployment (NAIRU) to replace the current inflation target. It means a particular inflation rate (not a range and not any number) that does not accelerate and consistent with the long-run unemployment rate. This is not economics, but rather alchemy. If there is no trade-off between inflation and unemployment in the long run, i.e., a vertical Phillips curve, then there is no such an inflation rate because the vertical Phillips curve implies that any inflation rate is consistent with long-run unemployment. The NRU is not the same thing as the NAIRU. The former is determined in the labour market, not by monetary conditions while the latter is about an unstable relationship between inflation and the unemployment. Most people agree that there is no trade-off in the long run. If the policy is indeed about the long-run, not about the short run then we know it won’t work.

The NRU is unknown and the estimates are highly uncertain. So whatever you want to call the long-run unemployment rate, what is that number? A few weeks ago the chief economist of the RBNZ John McDermott gave a speech, where he said that they are looking for the starts, among them I assume is U* - the long-run unemployment rate. No one knows for sure what that number is.

Also, the economic motivation behind this policy is unclear.  The Labour Party must have been advised that the relationship between inflation and unemployment is unstable, which is especially true in New Zealand. Also history shows that it never worked, because it generated persistent inflation with no reduction in unemployment, and took considerable efforts and costs to undo. In fact our labour indicators are reasonable OECD.   

Because the NRU is determined in the labour market and not by the RBNZ. Politicians must look at labor market policies that encourage work, reduce benefits, increase training and up-skilling, increase matching efficiency, help small businesses, and increase foreign direct investments etc. because monetary policy cannot deliver lower unemployment. In fact the proposed policy may increase volatility and generate higher inflation only.

[1] Economists know that monetary policy cannot change the Natural Rate of Unemployment. Milton Friedman explained 50 years ago. No new evidence suggests otherwise.

Tuesday, July 25, 2017

Politics and Public Finance in New Zealand’s Election Year

Surely no one likes poverty. Politicians of different parties might share the same preferences, but they have different budget constraints. The recent Green’s stump speeches call for ending poverty by significantly increasing social benefits paid for by higher taxes on the “rich.” Rhetoric aside, a tax increase is not a good economic idea especially when our productivity is already sluggish. Social benefits do not end poverty.

For example, over the period 2001 to 2015, the New Zealand government’s average tax revenue as a percent of GDP (29 percent on average) is higher than Australia (23 percent on average); it is significantly higher than Singapore (13 percent on average), Germany (11 percent on average), and Switzerland (9.5 percent).[1] For New Zealand, the tax revenues are mostly from taxing consumption, labor income and company profits. Australia’s average tax revenue is less than New Zealand’s even though it taxes capital gains and New Zealand doesn’t.[2] The figure below plots the World Bank Development Indicators data (July 2017).

So why does New Zealand, which is the least populated among all these countries, has such high tax revenue/GDP ratio? More spending! Typically, government’s major expenditures are on health, education, social services, and the military. Our military spending is trivial, about 1 percent of GDP so I will ignore it.[3] However, on average over the period from 2001 to 2015, New Zealand spent 9.45 percent of GDP on health (about 26 billion dollars), more than Australia (8.79 percent), only slightly less than Germany (10.7 percent) and Switzerland (10.9 percent) and significantly less than Singapore (3.75 percent).[4] See the figure below.

On education, New Zealand’s spending is the highest, nearly 6.5 percent of GDP (about 16 billion dollars). Australia (5 percent), Germany (4.7 percent) and Switzerland (5 percent) spend less than New Zealand. Singapore spends the least, 3.25 percent only.

The Labour party says (here) that it will “address chronic under-funding of health, education…” even though the data suggest that there is no such under-funding! New Zealand spends more than Australia on health, just as much as Germany and Switzerland, and three times as much as Singapore. On education, spending surpassed all of them.

More public spending on services does not necessarily mean better outcomes. There is a strong empirical evidence for that. In education, our students do not do as well as Singaporeans. Performances in standardized tests such as PISA, for example, indicate that Singapore, which spends much less public money on education, is always on the top of the world.[5]

Also, Singapore’s average annual labor quality growth, although small, 0.9 percent (see the Conference Board data), still the highest comparably.[6] Germany and Switzerland have an average annual growth of labor quality of 0.1 percent only. New Zealand’s average annual labor quality growth is 0.6 percent, still slower than Singapore, but much higher than both Germany and Switzerland. The average annual growth rate of labor quality in Australia is half that of New Zealand (source: Conference Board).

The same is true for public health. Although it is difficult to measure output, there is no credible empirical evidence that more spending on national public health systems improves outcomes.

Government spending is a function of the size and the scope of market failure. The question is how much market failure is there in education and health to justify more public spending. I do not think we answered this question in New Zealand.
Here is the big spending item. OECD data show that average net total social expenditure as a percent of GDP for New Zealand is 16.5 percent (43 billion dollars). New Zealand is not alone: it is 19 percent in Australia, 25.4 percent in Germany, and nearly 22 percent in Switzerland.[7] And here is the difference: Singapore spends only 3.5 percent of its GDP on social programs!  

Nevertheless, some expenditures on social benefits could be justified. However, benefits raise the reservation wage and increase unemployment. Singapore, which spends the least on social benefits has the lowest unemployment rate, 1.7 percent in 2015.[8]    

Before getting excited about increasing social spending, note that taxes reduce labor productivity.[9] The plot below shows that Switzerland, Germany, and Singapore, which tax the least, have higher GDP per capita than us.

My advice to the politicians who advocate more taxes is to consider alternative policies to help the poor without taxing potential productivity of the whole country.  
New Zealand already has the highest tax revenue as a percent of GDP. They could reallocate expenditures, e.g., increase X and reduce Y. There must some waste in the public sector; cut it. Better, think about firm productivity-indexed wage subsidy (See the Nobel Laureate Edmund Phelps).      

[1] I chose Singapore because the data are available. Korea and Hong Kong would be just as good examples to use for comparisons regarding tax and spending issues. I chose Switzerland because there has been some public interest in this country as a model that New Zealand should emulate. 

[2] Since 2009 Germany levies a flat rate tax on private income from capital and capital gains. The tax rate is 25 percent plus 5.5 percent solidarity surcharge. The tax is levied at German sources as capital yields tax. There is a tax refund for personal income tax rate below percent.

[3] Australia spends about 1.8 percent, Germany 1.2, and tiny Singapore spends 3.5 percent of its GDP on defense.

[4] The IMF World Economic Outlook data (2017) estimated GDP at current prices in 2016 to be 261 billion New Zealand dollar.

[6] “Measure of the changes in the composition of the workforce. This indicator is based on underlying data on employment and wages by educational attainment, which are estimated econometrically in some cases.”

[7] Data for Switzerland correspond to 2013 which the last data published on the OECD stats.

[8] The average unemployment rate for the period 1970-2015 (a proxy for the natural rate) is 3.78 percent. Unemployment in Germany, Switzerland and Australia are similar to New Zealand on average.

[9] For international evidence see for example, Razzak and Belkacem (2016), Taxes, Natural Resource Endowments, and the Supply of Labor: New Evidence, in Handbook of Research on Public Finance in Europe and the MENA Region, IGI Global Research Publishing, USA, (eds.,) M. Mustafa Erdoğdu and Bryan Christiansen, Chapter 23, PP 520-544, May 2016.